CheemaMJM_PhD_Thesis.

CheemaMJM_PhD_Thesis.
Understanding water resources conditions in data scarce
river basins using intelligent pixel information
Case: Transboundary Indus Basin
M.J.M. Cheema
Understanding water resources conditions in data scarce
river basins using intelligent pixel information
Case: Transboundary Indus Basin
Proefschrift
ter verkrijging van de graad van doctor
aan de Technische Universiteit Delft,
op gezag van de Rector Magnificus prof. ir. K.C.A.M. Luyben,
voorzitter van het College voor Promoties,
in het openbaar te verdedigen
op dinsdaag 29 mei 2012 om 15:00 uur
door
Muhammad Jehanzeb Masud CHEEMA
Master of Science
University of Agriculture Faisalabad
geboren te Sargodha, Pakistan
Dit proefschrift is goedgekeurd door de promotor:
Prof. dr. W.G.M. Bastiaanssen
Samenstelling promotiecommissie:
Rector Magnificus
Prof.dr. W.G.M. Bastiaanssen,
Prof.dr.ir. N.C. van de Giesen,
Prof.dr. S. Uhlenbrook,
Prof.dr.ir. P. van der Zaag,
Prof.dr.ir. H.H.G. Savenije,
Dr. F. van Steenbergen,
Dr. W.W. Immerzeel,
voorzitter
Technische Universiteit Delft, promotor
Technische Universiteit Delft
Technische Universiteit Delft en UNESCO-IHE
Technische Universiteit Delft en UNESCO-IHE
Technische Universiteit Delft
Meta Meta
Universiteit Utrecht
The research described in this dissertation was performed at the Water Resources Section,
Faculty of Civil Engineering and Geosciences, Delft University of Technology, Delft, the
Netherlands. The Higher Education Commission (HEC), Pakistan is thanked for providing
funds to carry out this research. The International Water Management Institute, Pakistan is
also thanked for providing financial support for additional months.
Copyright by M.J.M. Cheema, 2012 ([email protected])
All rights reserved. No part of this publication may be reproduced or utilized in any form
or by any means, electronic or mechanical, including photocopying, recording or by any
information storage and retrieval system, without the prior written permission of the
author.
ISBN: 90-6562-299-3
Published by . VSSD, Delft, the Netherlands
Keywords: Indus Basin, land use, surface soil moisture, ETLook, evaporation,
transpiration, groundwater depletion
To my family
Contents
Acknowledgements………………………………………………………..…….ix
Symbols and Abbreviations…………………………………………………….xi
1
Introduction ............................................................................................................... 1
1.1 Transboundary river basins ................................................................................. 1
1.2 Water conflicts and treaties ................................................................................. 2
1.3 Indus water treaty................................................................................................ 3
1.4 Transboundary aquifer ........................................................................................ 9
1.5 Data availability and sharing issues .................................................................. 11
1.6 Redefinition of water resources management ................................................... 12
1.7 Remote sensing in hydrology and water management ...................................... 12
1.8 The research justification .................................................................................. 14
2
Study area ................................................................................................................ 17
2.1 Geographical description .................................................................................. 17
2.2 Hydro-climatology ............................................................................................ 18
2.3 Indus river, major tributaries and doabs ............................................................ 19
2.4 Groundwater ..................................................................................................... 20
2.5 Agriculture and cropping pattern ...................................................................... 22
3
Land use and land cover classification in the irrigated Indus Basin using growth
phenology information from satellite data to support water management
analysis ..................................................................................................................... 25
3.1 Introduction ....................................................................................................... 25
3.2 Study area ......................................................................................................... 26
3.3 Methodology ..................................................................................................... 28
3.4 Results and discussion ...................................................................................... 30
3.4.1
Phenology ........................................................................................... 30
3.4.2
Effect of physical condition on LULC ................................................ 34
3.4.3
Accuracy assessment .......................................................................... 38
3.5 Conclusions ....................................................................................................... 46
4
Local calibration of remotely sensed rainfall from the TRMM satellite for
different periods and spatial scales in the Indus Basin ........................................ 48
4.1 Introduction ....................................................................................................... 48
4.2 Materials and methods ...................................................................................... 50
4.2.1
Study area ........................................................................................... 50
4.2.2
Rainfall systems over the Indus basin ................................................. 51
4.2.3
TRMM retrieval algorithm ................................................................. 51
4.2.4
Data availability .................................................................................. 54
4.2.5
Methodology ....................................................................................... 55
4.3 Results and discussion ...................................................................................... 57
4.3.1
Technique -1 ....................................................................................... 57
4.3.2
Technique -2 ....................................................................................... 59
4.3.3
Validation ........................................................................................... 62
4.3.4
Temporal and spatial deviation analysis ............................................. 63
4.3.5
Agricultural landuse – rainfall relationship......................................... 65
4.4 Conclusions ....................................................................................................... 67
5
Validation of surface soil moisture from AMSR-E using auxiliary spatial data in
the transboundary Indus Basin .............................................................................. 70
5.1 Introduction ....................................................................................................... 70
5.2 Materials and methods ...................................................................................... 71
5.2.1
Study area and landuse patterns .......................................................... 71
5.2.2
Remote sensing data ........................................................................... 72
5.2.3
Methodology ....................................................................................... 72
5.3 Results and discussion ...................................................................................... 76
5.4 Summary and conclusions ................................................................................ 87
5.5 Appendix: Soil moisture retrieval algorithm ..................................................... 88
6
The surface energy balance and actual evapotranspiration of the
Transboundary Indus Basin estimated from satellite measurements and the
ETLook model ......................................................................................................... 91
6.1 Introduction ....................................................................................................... 91
6.2 Study area ......................................................................................................... 93
6.3 Material and methods ........................................................................................ 94
6.3.1
vi
Satellite data and pre-processing......................................................... 94
6.3.2
Meteorological data ............................................................................ 96
6.3.3
Theoretical background of ETLook .................................................... 97
6.3.4
Calibration and validation approaches .............................................. 101
6.3.5
Sensitivity and uncertainty analysis .................................................. 102
6.4 Results and discussion .................................................................................... 103
6.4.1
Surface energy balance ..................................................................... 103
6.4.2
Actual evapotranspiration estimates ................................................. 105
6.4.3
Validation ......................................................................................... 108
6.5 Summary and conclusions .............................................................................. 113
7
Spatial quantification of groundwater abstraction for irrigation in the Indus
Basin using pixel information, GIS and the SWAT model ................................ 115
7.1 Introduction ..................................................................................................... 115
7.2 Material and methods ...................................................................................... 116
7.2.1
Study area ......................................................................................... 116
7.2.2
Soil and Water Assessment Tool ...................................................... 117
7.2.3
Data ................................................................................................... 119
7.2.4
ETLook ............................................................................................. 121
7.2.5
Model calibration procedure ............................................................. 122
7.2.6
Pixel based groundwater abstraction data ......................................... 123
7.3 Results and discussion .................................................................................... 124
7.3.1
Model calibration .............................................................................. 124
7.3.2
Spatial patterns of water supply and consumption ............................ 127
7.3.3
Accuracy assessment ........................................................................ 132
7.3.4
Water balance ................................................................................... 134
7.4 Conclusions ..................................................................................................... 135
8
Summary and conclusions .................................................................................... 137
8.1 Rationale ......................................................................................................... 137
8.2 Pixel land use .................................................................................................. 138
8.3 Pixel rainfall .................................................................................................... 139
8.4 Pixel surface soil moisture .............................................................................. 139
8.5 Pixel evapotranspiration .................................................................................. 140
8.6 Pixel groundwater abstraction ......................................................................... 142
8.7 New data sources ............................................................................................ 144
vii
8.8 Development of applications .......................................................................... 146
8.9 Conclusions ..................................................................................................... 147
9
Samenvatting ......................................................................................................... 151
9.1 Motivatie ......................................................................................................... 151
9.2 Pixel landgebruik ............................................................................................ 152
9.3 Pixel neerslag .................................................................................................. 153
9.4 Pixel oppervlak bodemvocht ........................................................................... 154
9.5 Pixel verdamping ............................................................................................ 155
9.6 Pixel grondwateronttrekking ........................................................................... 156
9.7 Nieuwe gegevensbronnen ............................................................................... 158
9.8 Toepassingsontwikkeling ................................................................................ 160
10 References .............................................................................................................. 163
Curriculum vitae………………………………………………………………………..185
Publications……………………………………………………………………………...186
viii
Acknowledgements
I give honor and thanks to Almighty Allah, the source of knowledge and wisdom, who
endowed me with the abilities for successful execution of this PhD research. I have been
fortunate to work under dynamic supervision of Prof. Dr. Wim Bastiaanssen. His
intellectual inspiration, valuable guidance, encouragement and sparing time from his busy
schedules for lengthy stimulating discussions have been invaluable to me. I have learned a
lot from him professionally as well as personally, which has significantly improved my
professional capabilities. Thank you for all this and especially for arranging my difficult
administrative requests. I am also extremely grateful to Dr. Walter Immerzeel for his
helpful discussions on the SWAT model application. Thanks for significantly contributing
to this research and for consistent encouragement and pushing me to wrap things up.
Funds for this research were generously provided by Higher Education Commission (HEC),
Pakistan and I am greatly indebted. Additional funds were made available by IWMIPakistan to support me for a few months of additional stay at TUDelft to complete the PhD
conveniently. I also thank University of Agriculture Faisalabad (UAF) for granting me
leave to enable me pursues this research. These funding institutes and their donors are
gratefully acknowledged. Special thanks go to Rao Azhar (HEC, Pakistan), Loes Minkman
(NUFFIC) and Franca Post (CICAT, TUDelft) for making all administrative and logistic
work in the Netherlands possible. I am thankful to Dr Vladmir Smakhtin and Dr Asad
Sarwar for their gentle and highly professional attitude, which greatly facilitated to
successfully complete this study.
For this study, secondary information was collected from various government agencies in
Pakistan, including the Pakistan Meteorological Department (PMD), the Punjab Irrigation
Department (PID), the SCARP Monitoring Organization (SMO) and the Indus Water
Commission (IWC). Here I would like to thanks Engr.Sheraz Jamil Memon and Engr. Faris
Kazi of IWC, Habib Ullah Bodla of PID and Dr Muhammad Arshad of UAF for their
positive attitude and making it possible to get precious databases.
Many thanks go to the colleagues in the section of Water Resources at TUDelft for the great
assistance I received from them. Although I am grateful to everybody for the pleasant time,
I would like to mention some colleagues in specific. Hanneke de Jong and Betty Rothfusz,
thank you both for all administrative assistance you provided. Martine Rutten, Ilyas Masih,
Saket Pande and Zheng Duan thank you all for good discussions. Reeza, Jacqueline and
Congli for providing a friendly environment in the office. Thanks Miriam for your friendly
and caring attitude and also for being my paranimf. Special thanks to Atiq, Naveed and
Faisal for providing an atmosphere that always give me a feeling as I am in my homeland. I
am also extremely thankful to Annemarie Klaasse and Henk Pelgrum of Water Watch for
providing necessary support in collecting satellite data and understanding ETLook
algorithm.
I want to thank my friends who have made sure that not my whole life consisted of doing a
PhD. In particular, I want to mention Bilal Ahmad, Faisal Nadeem, Fakhir, Seyab, Shah
Muhammad, Atif, Laiq, Malik Aleem and Iftikhar Faraz who were always ready to play
cricket and arrange dinners. Sarfaraz Munir and Syed Iftikhar Kazmi are specially thanked
for the nice company which provided me an excellent opportunity to share my feelings and
ix
concerns more openly with someone from my own country, Pakistan. How can I forget the
pleasant gupshup with Zahid Shabbir and fight with high velocity opposing winds while
riding bikes from Rotterdam to TUDelft and back. It was a great adventure of my life,
which I will not forget. Of course, this list is not complete and I want to thank all my
friends but I would prefer to rather do this in person than in the form of an exhaustive list.
I wish to express my gratitude to my family for their love, good wishes, inspirations and
unceasing prayers for me, without which the present destination would have been mere a
dream. The dream of my father, Masud Ata Cheema, to see me a doctor comes true. Today,
I am missing my loving mother, but I am sure she will be happy in heaven. I would like to
thank my uncles Dr Zahid Ata Cheema and Mr. Muhammad Aftab Mehmud who motivated
me to start my PhD study. I also want to thank my aunts, brothers (Jehangir Masud Cheema
and Mughees Aftab) and sisters (Kshif and Adeela) for their prayers and well wishes.
Finally, I thank my wife and children for their patience and perseverance during long period
of our separation and care and support while our stay in the Netherlands. Raheela, without
you I would not have been able to finish this thesis as you always ask on which paper I am
working, how many are submitted and how many are published? This kept me focused on
the final goal. Final thanks to my little fairies, Shaiza and Hamima for their prayers and
love. The sweet company of you made this tough journey a very pleasant and memorable
experience of my life and I will never forget these moments.
x
Symbols and Abbreviations
List of symbols
α
αo
βw
cp
Cr
DEPgw
Δe
ΔSus
E
ET
ETo
ETSWAT
ETETLook
ε
εs
εfw
G
H
I
IRRcw
IRRgw
IRRRS
IRRSWAT
Ksf
λE
Ln
LOSScw
Φ
Ψ
Qgw
Qsurf
Qlat
Qperc
ρ
R
R↓
r
R2
ra,soil
ra,canopy
rcanopy
Shape factor
Surface albedo
Water use distribution parameter
Specific heat of dry air
Capillary rise in the unsaturated zone
Net groundwater depletion
Vapor pressure deficit
Change in storage of the unsaturated zone
Evaporation
Evapotranspiration
Reference crop evapotranspiration
Actual evapotranspiration modeled by SWAT
Actual evapotranspiration estimated by ETLook
Dielectric constant
Dielectric constant of soil solids
Dielectric constant of free water
Soil heat flux
Sensible heat flux
Interception
Canal water supplied at farm gate
Gross groundwater abstraction
Total irrigation estimated by remote sensing
Total irrigation applied in SWAT
Ability of plant to extract soil moisture
Latent heat flux
Net longwave radiation
Canal water losses
Available water capacity of soil
Soil evaporation compensation factor
Return flow from shallow aquifer
Surface runoff
Lateral flow through unsaturated zone
Percolation to saturated zone
Air density
Rainfall
Incoming shortwave radiation
Pearson’s product moment correlation
Coefficient of determination
Aerodynamic resistance for soil
Aerodynamic resistance for canopy
Canopy resistance
xi
Rn
Rn,soil
Rn,canopy
rop
Rr
rs
rs,min
rsoil
rsp
RSWAT
Rtoa
RTRMM
SeFC
Sesub
Setop
Sm
Sr
St
Sv
T
Tair
Tb
Tp
τa
τc
τo
τMODIS
τr
τsw
U2
θsat
θo
θAMSRE
θsat,xy
θres,xy
wup,z
ω
Λ
Z
z
zd
zroot
xii
Net radiation
Net radiations at soil surface
Net radiations at canopy
Reflectivity from smooth soil surface
Rainfall rate
Spearman’s rank correlation coefficient
Minimum stomatal resistance
Soil resistance
Reflectivity from rough soil surface
Rainfall from SWAT
Top of atmosphere radiation
Satellite rainfall
Effective saturation at field capacity
Subsoil effective saturation
Topsoil effective saturation
Soil moisture stress
Radiation stress
Temperature stress
Vapor pressure stress
Transpiration
Air temperature
Brightness temperature
Potential plant transpiration
Atmospheric optical thickness
Vegetation optical thickness
Oxygen opacity at nadir
Short wave transmissivity from MODIS
Precipitation optical thickness
Shortwave transmissivity
Wind speed
Saturated soil moisture content
Volumetric water content
AMSRE surface soil moisture
Saturated moisture content at 1 km pixel (x,y)
Residual moisture content at 1 km pixel (x,y)
Plant water uptake factor
Single scattering albedo
Evaporative fraction
Radar reflectivity factor
Depth from soil surface
Damping depth
Depth of root development in the soil
List of Abbreviations
ALOS
amsl
AMSR-E
APHRODITE
AVHRR
AWR
CAMS
CCA
CERES
CMAP
CRU
CWR
DAAC
DEM
DOY
DN
EROS
ESA
ETLook
FAO
FY-2
GDA
GHz
GIS
GOP
GPCC
GPCP
GRACE
GLC
ha
hr
HRU
IB
IB-IN
IB-PK
IBIS
IBSP
ICID
ICIMOD
IDW
IGBP
IN
IRSA
ISODATA
Advanced Land Observing Satellite
Above Mean Sea Level
Advanced Microwave Scanning Radiometer – EOS
Asian Precipitation Highly Resolved Observational Data Integration
Advanced Very High Resolution Radiometer
Australian Water Resources
Climate Assessment and Monitoring System
Canal Command Area
Clouds and Earth’s Radiant Energy System
Climate Prediction Center’s merged Analysis of Precipitation
Climatic Research Unit
Crop Water Requirement
Data Active Archive Centers
Digital Elevation Model
Day of Year
Digital Numbers
Earth Resources Observation and Science
European Space Agency
Evapotranspiration Look
Food and Agriculture Organization
Feng Yun 2 (Earth Observation System)
Geographical Differential Analysis
Giga Hertz
Geographic Information System
Government of Pakistan
Global Precipitation Climatology Centre
Global Precipitation Climatology Project
Gravity Recovery and Climate Experiment
Global Land Cover
Hectare
Hour
Hydrological Response Unit
Indus Basin
Indus Basin Indian part
Indus Basin Pakistani part
Indus Basin Irrigation System
Indus Basin Settlement Plan
International Commission on Irrigation and Drainage
International Centre for Integrated Mountain Development
Inverse Distance Weighted
International Geosphere-Biosphere Program
India
Indus River System Authority
Iterative Self Organizing Data Analysis Technique
xiii
IWASRI
IWC
IWT
IWMI
JAXA
Km3
KPK
LAI
LAIeff
LIS
LT-1
LULC
MERIS
mha
MINFAL
MODIS
NASA
NCDC
NDVI
NGU
NIR
nm
NOAA
NSE
NSIDC
PARC
PID
PK
PMD
PR
RA
RE
RFI
RH
RMSE
ROI
RS
SC
SEBAL
SEE
SI
SMMR
SPOT
SRTM
SSM/I
SWAT
xiv
International Water Logging and Salinity Research Institute
Indus Water Commission
Indus Water Treaty
International Water Management Institute
Japanese Space Agency
Cubic kilometer
Khyber Pakhtunkhwa
Leaf Area Index
Effective leaf area index
Lightning Imaging Sensor
Length per Time
Land Use and Land Cover
Medium-spectral Resolution Imaging Spectrometer
Million Hectares
Ministry of Food, Agriculture and Livestock
Moderate Resolution Imaging Spectro-radiometer
National Aeronautics and Space Administration
National Climatic Data Center
Normalized Difference Vegetation Index
Net Groundwater Use
Near Infrared
Nano Meter
National Oceanic and Atmospheric Administration
Nash-Sutcliffe Efficiency
National Snow and Ice Data Center
Pakistan Agricultural Research Council
Provincial Irrigation Department
Pakistan
Pakistan Metrological Department
Precipitation Radar
Regression Analysis
Relative Error
Radio Frequency Interference
Relative Humidity
Root Mean Square Error
Regions of Interest
Remote Sensing
Sensitivity Coefficient
Surface Energy Balance Algorithm for Land
Standard Error of Estimates
Scattering Index
Scanning Multi-channel Microwave Radiometer
Satellite Probatoire d’Observation dela Terre
Shuttle Radar Topography Mission
Special Sensor Microwave/Imager
Soil and Water Assessment Tool
SWIR
TRMM
TMI
UN
USGS
VC
VIRS
VWC
WAPDA
WMO
yr
Shortwave Infrared
Tropical Rainfall Measurement Mission
TRMM Microwave Imager
United Nations
United States Geological Survey
Vegetation Cover
Visible-Infrared Radiometer Scanner
Vegetation Water Content
Water and Power Development Authority
World Meteorological Organization
Year
xv
1 Introduction
The most precious resource on earth, vital for human sustainability is water. Exponential
increase in global population and unconstrained water resource utilization threatens the
spatial and temporal availability of the world’s freshwater resources. The threat is more
severe in developing countries where the majority of the population practices agriculture.
Agriculture accounts for instance for 30% of the economy in a country such as Pakistan.
Surface water and groundwater (separately or in combination) are used to fulfill the crop
water requirements. The declining water resources need to be managed in an integrated way
at basin scale. In fact, water management has to undergo significant improvements in terms
of vision, targets, decision-making, and accounting. Hydrological simulation models can be
used as analytical tools for determining the water flow paths (e.g. Andersson et al., 2006)
and the impact of water management measures on irrigation systems (e.g. Droogers et al.,
2000). The transboundary nature of river basins and the limited availability of data is
however a hindrance for good modeling. Spatial information of topography, land use,
rainfall, soil moisture, evapotranspiration, and leaf area index, derived from remote sensing
can be used for, and will enhance spatially distributed modeling. Such data can also be used
to validate and calibrate hydrological models. This thesis aims to improve the knowledge
base of river basins by using satellite measurements and advanced hydrological models in
data scarce environments to support short term and long term planning and water allocation
processes. The transboundary Indus Basin is used as a case study.
1.1
Transboundary river basins
The river basin is the basic geographic unit which collects and provides water for the basin
ecosystems itself, but also for agriculture, industry, and socio-economic development
within the basin and downstream. The water of a basin flows across and underneath
international boundaries to sustain agro ecosystems, whose boundaries do not coincide with
the political boundaries. Such situations complicate the study of water flows and resource
management. Nearly half the world is situated in 263 international river basins bearing 40%
of the world’s population(Wolf et al., 1999). These 263 international rivers generate 60% of
global fresh water. Most of these rivers are situated in Europe (69), followed by Africa(59),
Asia (57), North America (40) and South America (38) as illustrated in Figure 1-1.
Figure 1-1 International river basins situated in the continents. (Source: Oregon State Uni)
The transboundary nature of the river basins has resulted in acrimonious disputes over
water. Any change in upstream water use can severely affect the downstream users,
although they may be thousands of kilometers apart from each other. The effects of land
and water use planning in one part of the basin is vital for the users in another part of the
basin (e.g. Molden et al., 2001). One example of such a river basin is the Indus Basin where
the riparian countries have strained relationships over water flows. The Indus is a
transboundary basin encompassing Pakistan, India, China and Afghanistan (Figure 1-2).
Therefore, flow commitments by means of water treaties between the co-basin states are
necessary for sharing and better utilizing of the resources.
Figure 1-2 Location of the Indus Basin
1.2
Water conflicts and treaties
All water users are hydrologically connected in a river basin. Upstream water use has a
direct effect on the downstream users even thousands of kilometers away, or in another
country. By promoting water supplies, upstream water users can cause dramatic
consequences for downstream water users and their environments. Upstream riparian
proprietors should not deprive downstream water users access in terms of quantity and
quality. Even environmentally endorsed and acceptable practices to improve biodiversity
and reduce soil loss upstream can lead to extermination of flora and fauna in downstream
flood plains and estuaries. During wars, manipulation of the river waters can also be used as
an offensive or defensive military weapon (Gleick, 2008).
Water flowing across political boundaries has resulted in various conflicts and agreements
of cooperation (treaties) among the riparian countries. During the last 60 years, 37 incidents
of conflicts among the riparian countries over water are reported by Wolf (1998). For
example, partitioning of India and Pakistan in 1947 divided the Indus Basin, which caused a
continuous threat of war over water flows. India, occupying the upstream portion of the
basin, had control of the barrages and diverted water to its own lands. It caused a serious
2
environmental threat to the lower riparian areas in Pakistan. The construction of new
storage and hydropower facilities during 1947 to 1960 made the situation even worse.
The Nile Basin is shared by ten countries. For many years, there was tension among the
countries over the use of the Nile. For example, Tanzania vowed to use water from Lake
Victoria (that feeds the Nile) for its domestic use; thus causing tension with lower riparian
country Egypt. The tension between Egypt and Sudan over the water rights of the Nile
increased in 1958 (Mandel, 1992). The plan by Egypt to divert water for use in the Sinai
desert was strongly opposed by Ethiopia and the two countries were at the verge of war in
1980. The conflict between Syria and Iraq over the water of the Euphrates River stems from
1975 and is basically an un-solved issue. Construction of the Ataturk dam on the Euphrates
river by Turkey has substantially reduced the flows to Syria (Zawahri, 2006). Crossfire
between Israel and Syria over the water rights in the Huleh Basin occurred during 1951-53.
Conflict arose between India and Bangladesh over the use of the Ganges River water and
the conflict intensified in 1975 when India started to construct the Farakka barrage to
unilaterally control the flow of the Ganges River. In the early 1990s Hungary and Slovakia
started with the Gabukovo-Nagymaros barrage system along the Danube River. Conflict
arose between the two countries and Hungary deployed troops to keep the system
inoperative(Fuyane and Madai, 2001).
To resolve such conflicts, the riparian countries have to come up with treaties defining
water rights. The Food and Agriculture Organization (FAO) of United Nations reported
that 3600 treaties on the use of international waters have been formed between 805 A.D. to
1984 A.D. Historically, the treaties to resolve water conflicts date back to 2500 B.C. when
the two states of Lagash and Umma signed an agreement to end conflict along the Tigris
River.
A water treaty was signed between Mexico and United States in 1944 which defined the
water rights and delivery responsibilities associated with the Colorado and the Rio
Grande/Rio Bravo basins (Gastélum et al., 2010). Similarly, the Mekong River
Commission was established in 1995 to efficiently utilize the resources of the Mekong
River. In 1959, Egypt and Sudan signed an agreement to fully utilize the Nile water and
established a Permanent Joint Technical Commission (Mandel, 1992). In 1996, India and
Bangladesh signed a 30-year-treaty to share the flows of the Ganges which ended the
dispute over Indian unilateral water diversions from the Farakka barrage. A similar effort
was made between India and Pakistan in 1960, to resolve their water conflicts in the Indus
Basin.
1.3
Indus water treaty
Development of irrigation systems in the Indus Basin dates back to the Harrapan
civilization 2300 B.C. to 1500 B.C. (Fahlbusch et al., 2004). During the 2nd millennium,
various Mughal emperors constructed limited canal systems to irrigate dry lands along the
Ravi, Chenab and Sutlej rivers (Thatte, 2008). The systematic development of irrigation
canals with weir-controlled structures started during British rule in 1850, when the 395 km
long Upper Bari Doab canal (UBDC) was constructed. The headwork was constructed on
the Ravi River at Madhopur in 1873.The next large project was the development of the
Sirhind canal from the Sutlej River at Ropar to irrigate the districts of Ludhiana, Ferozpur
3
and Hissar, etc. It became operational in 1882. Afterwards, a network of canals was
developed all over the Indus Basin, the Sidhnai canal taking off from the Ravi River was
constructed in 1886. The Lower Chenab Canal (LCC) from Khanki headwork in the
Chenab River became operational in year 1900. The Lower Jhelum Canal (LJC)
commenced in 1901 from the left bank of the Jhelum River at Rasul barrage. A schematic
diagram of irrigation system is shown in Figure 1-3 to give an idea of the location of
various barrages and link canals in the basin.
During the late 19th century, severe famine occurred that resulted in the establishment of the
1st Irrigation Commission of India in 1901. It came up with a proposal to transfer west
flowing rivers eastwards to cope with the severe famine in the eastern parts. This proposal
seems to be the base of the Indus Water Treaty (IWT) (Thatte, 2008).
Development of The Triple Canal Project (Upper Jhelum canal: UJC, Upper Chenab canal:
UCC, and Lower Bari Doab canal: LBDC) was proposed by the commission in 1905. It
comprised a system of linked canals, including irrigation systems, starting from the Jhelum,
through Upper Jhelum canal, to the Chenab River, and then to the Ravi River through the
Upper Chenab canal. The project was completed in 1915.The gigantic Sutlej Valley Project
(1921) was designed to replace the old-shutter type weirs with gate-controlled barrages.
Four weirs at Ferozpur, Sulemanki, Islam and Punjnad were constructed. The former three
were completed in 1927 and the latter one in 1933. Four canals, the Pakpattan, Dipalpur,
Eastern and Mailsi canals were constructed in 1933 as part of this project. The Haveli and
Rangpur canals were then completed in 1939, taking off from the Trimmu headworks,
downstream of the confluence of the Jhelum and Chenab.
In 1947, independence from the British rule resulted in the partitioning of the two riparian
countries (Pakistan and India, sharing the major portion of the basin). Two major
headworks, one at Madhopur on the Ravi and the other at Ferozpur on the Sutlej(rivers
flowing eastward from India to Pakistan) went under Indian control. Irrigation in the
Pakistani part of the Punjab province was dependent on these headworks. The Indian
possession of the headworks resulted in administrative problems to regulate and supply
water.
To cope with the foreseen water crisis due to the stopping of east flowing rivers, Pakistan
started to construct various barrages and linked canals to divert west flowing rivers
eastward. For example, the Balloki-Sulemanki link (BSL:1954), the Marala-Ravi link
(MRL:1956) and the Bombanwala-Ravi-Badian-Dipalpur link (BRBD: 1956) canals were
constructed for this purpose. The Taunsa barrage was constructed in 1958. The Abbasia
canal was extended, and the Thal canal project was undertaken.
Meanwhile, India started construction of the Ferozpur and Rajasthan feeders in 1947. The
Bhakra Nangal project started in 1948. The Harike barrage was completed in 1952. The
Madhopur-Beas link canal was constructed in 1955 to divert waters of the Ravi to the Beas
(Thatte, 2008). The Bhakra canal (remodeling of the Ropar headworks) and the Sirhind
canal system were completed in 1955. The Rajisthan canal project was initiated in 1958.
4
Figure 1-3 Schematic diagram of the Indus Basin Irrigation Systems. C denotes canals, F
feeders and L linking canals.
5
Due to disagreement on water use between the two countries, India diverted all flows from
the east flowing rivers (the Ravi, the Beas and the Sutlej). It created water scarcity and an
environmental threat in the eastern part of the Indus Basin located in Pakistan. Various
conflicts arose between the two countries on water distribution of the rivers in the Indus
Basin.
To resolve these issues, water rights were defined under World Bank and United Nations
auspices in 1960, by the signature of the famous Indus Water Treaty (IWT) between India
and Pakistan. Nine articles with seven annexure were defined in the treaty. According to the
IWT, the flows of three main west flowing rivers (the Indus, the Jhelum, and the Chenab)
were available to Pakistan, while India had exclusive rights to waters of rivers flowing east.
The treaty prohibited both countries from undertaking any structures that may change the
volume of daily flows (Article II). Article III restricted India from constructing storage
facilities on west flowing rivers. However, India was allowed to construct incidental limited
storage on the western rivers. This was allowed only if the design was communicated to
Pakistan six months in advance. The design needed to be approved by Pakistan and storage
of structures should not exceed the defined volume.
A permanent Indus Water Commission (IWC) was established under the IWT article VIII
for smooth implementation of the treaty. The commission was to meet once in the year
alternately in Pakistan and India. The functions of IWC were to establish and maintain
cooperative agreements for IWT implementation, provide a report at the end of each year,
inspection of rivers once in five years, and to settle disputes. The commission was also
responsible to share data on agricultural use, hydro-electric power generation, water
storage, and flows in the rivers. Under the Article VI, both countries were supposed to share
daily gauge and discharge data, reservoir extractions, canal withdrawals and escapes.
The IWT was successfully implemented in the first few decades and a number of reservoirs
and a network of inter-river linking canals were constructed in the Indus Basin under the
Indus Basin Settlement Plan (IBSP). The details of the linking canals along with their year
of construction are provided in the Table 1.1.
Table 1.1 Linking canals constructed in the Indus Basin before and after IWT
S.
No
1
2
3
4
5
6
7
8
9
10
11
12
13
6
Linking canal
Upper Chenab
Upper Jhelum
Balloki-Sulemanki
Marala-Ravi
BRBD
Madhopur-Beas
Trimmu-Sidhnai
Sidhnai-Mailsi
Mailsi-Bhawal
Rasul-Qadirabad
Qadirabad-Balloki
Chashma-Jhelum
Taunsa-Punjnad
Off taking Linked rivers Construction Country
Barrage
year
Marala
Chenab-Ravi
1912
Pakistan
Mangla Jhelum-Chenab
1915
Pakistan
Balloki
Ravi-Sutlej
1954
Pakistan
Marala
Chenab-Ravi
1956
Pakistan
Marala
Chenab-Ravi
1956
Pakistan
Madhopur
Ravi-Beas
1955
Pakistan
Trimmu
Chenab-Ravi
1965
Pakistan
Sidhnai
Ravi-Sutlej
1965
Pakistan
Sidhnai
Ravi-Sutlej
1965
Pakistan
Rasul
Jhelum-Chenab
1967
Pakistan
Qadirabad Chenab-Ravi
1967
Pakistan
Chashma
Indus-Jhelum
1970
Pakistan
Taunsa
Indus-Chenab
1970
Pakistan
Length
(km)
142
142
63
101
175
20
71
132
16
48
129
101
61
14
15
16
Beas-Sutlej
Sutlej-Yamuna
SutlejHaryanaAlwar
Pandoh
Nangal
Ferozpur
Beas-Sutlej
Sutlej-yamuna
1977
U.C
P
India
India
India
37
214
U.C: Under construction; P:Proposed; Sources:(Thatte, 2008; Wilson, 2011)
After signing the IWT, the government of Pakistan started some mega projects. These
included construction of two large dams, the Mangla dam (1966) on the Jhelum River and
the Tarbela dam (1976) on the Indus River, construction of eight large capacity linking
canals, six barrages and remodeling of three of the existing inter-river linking canals. There
was no big irrigation canal project implemented after these developments. However,
construction of three new irrigation canals: the Raini canal, the Greater Thal canal, and the
Kachhi canal was approved in 2002. Amongst these, the former two are under construction.
Construction of the large capacity multi-purpose Diamer-Basha dam on the Indus about 315
km upstream of Tarbela dam was initiated and is expected to be completed in 2018. The
Kurramtangi dam on the Kurram River and the Munda dam on the Swat River are also
proposed for construction. Detail of the major reservoirs constructed in the Indus Basin is
provided in the Table 1.2.
Table 1.2 Major reservoirs constructed in the Indus Basin
S.No
1
2
3
4
5
6
7
8
9
10
11
12
13
Reservoir
Tarbela
Mangla
Chashma
Diamer-Basha
Kurramtangi
Munda
Bhakra
Pong
Pandoh
Thein
Salal
Baglihar
River
Indus
Jhelum
Indus
Indus
Kurram
Swat
Sutlej
Beas
Beas
Ravi
Chenab
Chenab
Indus
Country
Pakistan
Pakistan
Pakistan
Pakistan
Pakistan
Pakistan
India
India
India
India
India
India
India
Construction year
1976
1966
Under construction
Under construction
Under construction
1963
1974
1977
1995
2004
Under construction
In India, the Bhakra dam was completed in 1963 while the Rajasthan feeder canal was
finished in 1964. The Pong dam (1974) and the Pandoh dam (1977) were constructed on the
Beas River. The Beas-Sutlej link canal was constructed in 1977 to divert water from the
Beas to the Sutlej. In 1985 a lift irrigation scheme was completed in the Haryana district.
The Indira Gandhi Nahar Phase I was constructed in 1999. The dam on the Ravi was
completed in 2001. Phase II of the Indira Gandhi Nahar project was completed in 2006. The
Wullar barrage/Tulbul navigational project in the states of Jammu and Kashmir was
proposed by India in 1984. The Salal dam project on the Chenab in the Jammu and Kashmir
states was started in 1970 and step-wise completed in 1995. Another mega project
downstream of Salal dam is the Baglihar dam. The construction began in 1999 and the first
phase was completed in 2004. The locations of the various structures constructed after IWT
are shown in Figure 1-4.
7
Figure 1-4 Location of reservoirs and barrages constructed on the Indus River and its
tributaries.
During the last decade, several issues have arisen, on which IWC is working to resolve
within its mandate. India has for example started construction of storage structures on the
tributaries of the Indus, whose rights were given to Pakistan. The Wullar barrage/Tulbul
hydropower projects on the Jhelum, and the Kishan Ganga hydropower project on the
Kishan Ganga river, a tributary of Jhelum, are few examples (Zawahri, 2009). Construction
of the Baglihar dam on the Chenab with storage capacity of 37 million cubic meter (MCM)
is considered a violation of IWT. Afghanistan is also planning to control the water of the
Kabul River (Lashkaripou and Hussaini, 2007) with financial and technical support from
India. Similar structures are proposed upstream of the Jhelum and Chenab (Khan, 2009). It
is argued that, although there is a provision in IWT to construct hydro-power generation
projects, storage structures must not exceed 12.35 MCM capacities. The construction of
these storage structures on western rivers will have catastrophic consequences for Pakistan
8
as reduced flows resulting from filling of these dams during low flow season could destroy
the rabi seaon crops in Pakistan (PILDAT, 2010).
The average annual flows of major rivers in the basin are provided in Table 1.3. These
flows represent the pre-treaty (1922-61) and post-treaty (1985-2002 and 2007-2010)
situations. The flows show decreasing trends for both west and east flowing rivers. The
average flow of eastern rivers into Pakistan was reduced by 75% to 92% during the years
1985-2002 and 2007-2010, respectively. Pakistan can utilize only residual flows from these
east flowing rivers. However, these flows are variable and available only during the
monsoon season. About 17% reduction in the average flow of the west flowing rivers is
also observed. Climate change and its variability may cause reduction in flow of the west
flowing rivers (Ahmad, 2009). However, the upstream interventions could also be the cause
of reduced flows.
Table 1.3 Average flows in major rivers of the Indus basin before and after IWT
River
Rim station Average Annual Average Annual
Flow
Flow
(1922-61)
(1985-2002)
(km3)
(km3)
Kalabagh
114.4
94.1
West Indus
28.3
23.7
flowing Jhelum Mangla
rivers Chenab Marala
31.9
24.5
Ravi
Below
8.6
4.0
East
Madhopur
flowing
Below
17.2
2.2
rivers Sutlej
Ferozepur
Total
200.4
148.5
Average Annual
Flow
(2007-10)
(km3)
101.9
19.3
23.9
1.1
0.8
147.0
Source: (Khan, 1999; GOP, 2011; IUCN, 2011)
An integrated approach to manage transboundary water resources can lead to development
and revision of water treaties between states, and prevent potential conflicts and resolve
disagreements. Provision of objective information to facilitate negotiations between various
fellow states requires tools that can monitor spatial and temporal changes in water demand
and water use over vast areas.
1.4
Transboundary aquifer
Continues population growth in the Indus Basin resulted in mounting pressure on increased
food production. It is estimated that to feed the increasing population, 40% more food will
be required by the year 2025. The surface water resources in the basin (especially in
Pakistan) are limited and variable. The upstream interventions by India have also threatened
the timely availability of surface water downstream. Reduction of storage facilities in
Pakistan can result in up to 50% shortfall of crop water requirements by the year
2025(Alam and Bhutta, 1996).
9
The Indus Basin aquifer has large groundwater reserves. The development of this reserve
started 30 years ago (Sarwar, 2000). Inadequate and variable surface supplies forced
farmers to start irrigating with groundwater. Local and readily available groundwater makes
irrigation more productive compared to surface water irrigation. Currently 40–50 % of
agricultural water needs in the basin are met through groundwater (Sarwar and Eggers,
2006). Both in Pakistan and India large numbers of irrigation wells have been added every
year, which resulted in 20-30% increase in groundwater abstractions during the last 20
years (Qureshi et al., 2010b). The groundwater withdrawal exceeds annual recharge causing
imbalance in groundwater reserves. This process is accelerating in the province of Pakistani
and Indian Punjab, Haryana and Rajasthan states(Shah et al., 2000).
Groundwater is also used in conjunction with surface water. Conjunctive use is in practice
on more than 70% of the irrigated areas within the Indus Basin(Qureshi et al., 2010b).
Figure 1-5 shows that the 29% more area came under conjunctive irrigation during the last
20 years. These values represent the irrigated area in Punjab, Sindh and Khyber
Pakhtunkhwa (KPK) provinces of Pakistan only. The situation in the Indian part of the
basin is not different. Sustainability of major crops in the basin is now heavily dependent on
groundwater.
Uncontrolled and unregulated use (over exploitation) of groundwater in many areas of the
Indus Basin resulted in saltwater intrusion into the aquifer (Kijne, 1999). Water yield of
wells is declining and pumping cost is increasing due to deepening of the water table.
Salinization associated with the use of poor-quality groundwater for irrigation has raised the
severity of the problem (Qureshi et al., 2010a).
Figure 1-5 Annual trends of surface, ground and conjunctive water use in the Pakistani
part of the Indus Basin
10
1.5
Data availability and sharing issues
A major obstacle in transboundary river basin water resources management is that the
fundamental information on water flows, sources of water, and water demand is either
missing or not accessible. The downstream riparian countries depend on their upstream
neighbors for data collection and sharing. If this does not happen the downstream countries
cannot prepare themselves to cope with floods and droughts or generate hydropower
(Zawahri, 2008). This problem is more severe in basins in developing countries and the
Indus Basin is an example. The vastness of the basin, budget constraints, political distrust,
and its transboundary nature is a hindrance in establishing a comprehensive measurement
network.
Rainfall is an important component of the hydrological cycle but cannot be used in water
management studies if measurement stations are scarce. In the case of the Indus Basin, less
than four rain gauge stations are available for an area of 10,000 km2, which is insufficient
for basin scale studies.
The situation is even worse for in-situ soil moisture and evapotranspiration measurements.
There is no flux station (to the author’s knowledge) available in the whole basin that
provides continuous information on soil moisture status and actual evapotranspiration. The
same applies for land use and crops grown in the entire basin. Some spatial databases are
available describing the land uses of the basin but these databases are outdated or/and
coarse with little detail on cropping patterns.
River flows are monitored by the Water and Power Development Authority (WAPDA) of
Pakistan. The discharge data is collected by a network of manual and automated
observation stations installed at various points along the rivers especially in upstream areas.
It is the only data available.
Apart from collection, the accessibility of the data is also not straightforward. Acquisition
of long term data series is difficult and involves a series of bureaucratic permissions.
Accessibility is also hampered by the fragmented structure of governmental institutions
designated with various water management roles and tasks. There is seldom any
coordination among the departments involved in data collection and system planning. Due
to lack of coordination and institutional problems, the data collected by these departments is
of little use to decision makers and water resources planners in order to manage water flows
effectively. Recently, the Provincial Irrigation Departments (PIDs) and the WAPDA
initiated several projects to integrate databases. The successful completion of these projects
will be a big step forward in achieving a comprehensive hydrological database.
Moreover, the continued political turmoil and distrust between the two countries make it
difficult to carry out basin scale integrated water resources management. There is also no
trust in the quality of data shared between the two countries because it cannot be verified,
and is politically biased. Regional cooperation on water issues and comparisons between
fellow states in a basin requires a standardized description of the water flows and the
emerging processes. One possible way of promoting a climate of confidence and favorable
political will is by building adequate databases for water accounting on basin scale. It must
be admitted that nobody has reliable data related to water resources conditions, as data
11
gathering and definition procedures greatly differ. Satellite data can resolve this dilemma
(Bastiaanssen, 2000).
1.6
Redefinition of water resources management
Water scarcity is not only due to the physical shortage of water but also to poor
management; or in the words of the World Water Council: Today’s water crisis is not about
having too little water to satisfy our needs. It is a crisis of managing water so badly that
billions of people and the environment suffer badly (Cosgrove and Rijsberman, 2000).
Conventional water resource planning and management is mainly focused on blue water
(water in streams, rivers, aquifers, lakes and reservoirs).There is a need to incorporate
rainfall, especially in arid and semi arid basins, that infiltrates naturally into the soil and on
its way back to the atmosphere in the form of evapotranspiration (green water)(Falkenmark
and Rockström, 2006).Managing non-beneficial evaporation will result in a significant
reduction in water use that can be re-allocated to other users.
Planning and management of surface water resources is important. However, under the
current situation where groundwater utilization is upto 50% of total irrigation supplies,
there is a need to plan and manage groundwater resources to maximize basin level
efficiency. Groundwater can be a primary buffer against drought, as its response to short
term climate variability is slower than surface water systems. The mismanagement of this
buffering system can lead to serious impacts on the environment and ultimately on food
security (Ahmad, 2002). Sustainable management of groundwater is considered a more
serious challenge than development (Shah et al., 2000). The challenge is complex and
management is not straightforward. The absence of a robust knowledge base is a major
hindrance to sound management. In general, the integrated system, correctly managed, will
yield more water at more economic rates than separately managed surface and groundwater
systems.
1.7
Remote sensing in hydrology and water management
Transboundary river basin water resources management gains trust and faith if rainfall,
diverted water, soil moisture, crop evapotranspiration and vegetation growth data is (i)
collected at a range of scales, (ii) adequate, and (iii) available and accessible throughout the
basin. Hydrologists cannot (in a relatively short time span) diagnose the water flow path at
the regional scale if hydrological data is poor or incomplete. It requires considerable time to
thoroughly quantify or model the hydrological processes and cycles in a river basin using
other parties’ data.
Satellite data is an attractive alternative for data required by hydrological models and to
provide spatial information to decision makers. Satellites provide objective data for
database building (for various applications, see Table 1.4), which is politically neutral and
cannot be manipulated. Satellite measurements reflect the land surface features and the
observable landscape patterns resulting from socio-economic development, prevailing
jurisdiction, agricultural practices, hydrological processes, and irrigation management.
Because they are direct measurements, satellite observations are often more reliable than
secondary data. For instance, the irrigated area in the Gediz River Basin in Western Turkey
appeared from the satellite images to be 60% larger than from the secondary data obtained
12
from governmental statistics (Bastiaanssen and Prathapar, 2000). It is obvious that if such
types of secondary data are used to establish intra-basin water cooperation, disputes and
conflicts can potentially worsen and trust will fade away.
Table 1.4Satellite measurements for possible applications in transboundary river
basins.(Source: Bastiaanssen and Prathapar, 2000)
Discipline
Application
Hydrology
Snow cover, rainfall, soil moisture, evapotranspiration
Agriculture
Irrigated areas, rainfed areas, crop identification, biomass growth, crop yield,
irrigation performance
Environment Forest area, wetlands, rangelands, water logging, salinization, water quality
Geography
Digital elevation, land slope, land aspects, land cover, land use
There are large numbers of satellites in the earth’s orbit which are being used to acquire
information on hydrological and biophysical parameters. Pixel size varies from few metres
to kilometres and temporal resolution varies from 3-hours to months. For example, the
Tropical Rainfall Measuring Mission (TRMM) provides 3-hour rainfall rate estimates at 25
km pixel resolution since 1997. The Advanced Microwave Scanning Radiometer-Earth
Observing System (AMSR-E) observes atmospheric, land and oceanic parameters. Daily
soil moisture estimates at 25 km pixel resolution are available through AMSR-E. Daily
evapo transpiration can be estimated using AMSR-E and MODIS satellites at 1 km grids.
NDVI, LAI, land use, albedo, biomass at 1 km resolution can also be estimated from
MODIS, SPOT vegetation etc. Ground water levels can be estimated using the GRACE
satellite that provides monthly changes in storage change at 400 km grids.
Spatially distributed hydrological models are in use to compute rainfall-runoff processes,
river flow, erosion and sediment transport, land-atmospheric interaction, water allocation
planning, irrigation supply, groundwater recharge and ecological responses to land and
water resources management. Beven and Fisher (1996) recognized that remotely sensed soil
moisture, ET and snow cover estimates are necessary for scaling the hydrological processes
in basin scale hydrological models.
Many researchers have used satellite data in hydrological models in un-gauged or data
scarce regions (Droogers and Bastiaanssen, 2002; Immerzeel et al., 2008b; Winsemius et
al., 2008; Wipfler et al., 2011). Calibration and validation of these models need long term
data series obtained from dense measurement networks. However, in the basins like Indus,
such data is meager, thus causing a high level of uncertainty in the model results. Spatially
variable information describing topography, crop types, land use, climatic data, and leaf
area index, derived from remote sensing can be used for modeling across basins. This
presents a new way to study the hydrological processes, water resources depletion, food
security and environmental development in international river basins. It has opened a new
protocol where central governmental bodies and internationally controlled agencies get
uniform information. Remotely sensed information has a public domain status, and
everybody can have access to raw satellite data due to the Earth Observing System with
unlimited access to Data Active Archive Centers (DAACs). Federal Governments and the
13
UN can inspect land and water resources management issues, either by hiring their own
experts or by involving commercial consultants.
1.8
The research justification
An integrated holistic approach to international river basin management is needed, in which
the basin is accepted as the logical unit of operation. A multi-sectoral, integrated system,
complemented by information sharing, transparency and wide participation is therefore best
suited to encompass all these elements. Such an integrated system approach to evaluate the
interaction between the hydrological processes in the mountains, river flow generation,
water retention in reservoirs, groundwater pumping and agricultural water use in the Indus
Basin is largely lacking. In the past, most scientific modeling research concentrated on the
parts with well-established databases (e.g. Sarwar, 2000; Ahmad, 2002; Arshad, 2004;
Habib, 2004; Hussain, 2011). These studies are valuable to test hypothesis and to construct
local scale hydrological knowledge. However, a complete understanding of the
hydrological processes can only be obtained if the research focus is to establish a solid basis
for solving real life problems on the entire basin.
A huge number of hydrological models are available to use in exploration of different
hydrological processes. These models need input data that is limited or have inaccuracies.
They must be estimated either by some relationship with physical characteristics or by
tuning the parameters in order to have responses close to observed ones, a process known as
calibration.
Calibration of physically based, distributed models is complex given the limitations of data,
the complexity of the mathematical representation of hydrological processes, and the
incomplete knowledge of basin characteristics. Model calibration is usually based on the
comparison between modeled and observed values from a few gauging stations.
The problem of parameterization and lack of data for sound validation of modeling of large
basins can be overcome by hydro-meteorological information from earth observation
satellites. Land use, rainfall, soil moisture, water levels, total water storage changes,
evapotranspiration, etc. are examples of data that can be obtained via satellites. These
spatially distributed parameters can be used for distributed hydrological modeling and
validation.
There is no satellite dedicated to water management application. Various vegetation and
water parameters are derived from different space borne spectral radiometers. Complex
algorithms are used to transform original satellite measurements into spatially and
quantified pixel information. Pixels need to be trained and made intelligent by scientists
because spectral radiance (W m-2 sr-1 m-1) is a signal only. Uncertainty also exists in this
conversion process.
The overall objective of this thesis is “The development of methodologies to efficiently
utilize satellite measurements in hydrology and to model the conjunctive water use for data
scarce river basins”.
This study is unique in that it combines ET and rainfall with water available from reservoirs
to determine water balances and determine water flows with complex patterns of
conjunctive use. The following innovations will result from this study:
14
(1) Simple calibration and validation techniques for spatial data in data scarce conditions
will be developed. (2) A distributed pixel knowledge base on water flow paths and
groundwater interactions for the entire basin will be constructed. (3) A hydrological model
suitable for providing near real time data and capable of testing alternative solutions to
combat over-exploitation and verify IWT agreements will be designed.
This PhD study will prove that intelligent pixels combined with hydrological models will
generate reliable data to effectively deal with water allocation issues such as (i) tempered
groundwater exploitation, (ii) definition of volumetric water rights, including compulsory
return flows, (iii) efficient irrigation systems, and (iv) vulnerability to climate changes.
While climate change is on the international radar screen, the real challenge is to improve
current manage of water resources, and to control conjunctive use in a sustainable manner.
15
16
2 Study area
2.1
Geographical description
The study area selected for this study is the Indus Basin which lies between latitude 24°38′
to 37°03′ N and longitude 66°18′ to 82°28′ E. The Indus Basin is located in four countries
(Figure 2-1). The lifeline of the Indus Basin is the Indus River that traverses China
(upstream), Afghanistan, India and Pakistan (downstream). The total size of the basin is
1.162 million km2. The largest area of the basin is in Pakistan (53% of total). The area in
India is 33% followed by China and Afghanistan with 8% and 6%, respectively. Elevations
range from 0 to 8600 m above mean sea level (a.m.s.l). The Basin has complex
hydrological processes due to variability in topography, rainfall, land use, and water use.
Figure 2-1 Location of the Indus Basin showing main tributaries and provinces/states of
different countries in the basin.
17
2.2
Hydro-climatology
The climate of the basin varies spatially and is characterized by large seasonal fluctuations
in temperature and rainfall. The major part of the basin is dry and located in arid to semi
arid climatic zones. The upper (northern and north-eastern) parts have harsh winters with
significant snowfall while the middle and lower parts have comparatively mild winters but
hot summers. The average annual rainfall varies from less than 200 mm in the desert area to
more than 1500 mm in the north and north-east parts of the basin. The 30 years (1961-90)
average reference crop evapotranspiration (ETo) varies between 650 mm and 2000 mm in
the northern parts and southern desert areas of the basin, respectively. These values were
obtained from the International Water Management Institute (IWMI) world water and
climate atlas (http://www.iwmi.cgiar.org/WAtlas/Default.aspx).
The temporal variation of rainfall and ETo within the year also varies markedly (Figure
2-2). The ETo is higher during the months of May and June, corresponding with the prerainy season. Most of the rainfall occurs during the months of July, August, and September.
Figure 2-2 Monthly variation of average rainfall and reference evapotranspiration rate
(ETo) in the Indus Basin.
There are two sources of rainfall in the Indus Basin: the Monsoon and the Western
Disturbances. The former takes place from June to September and the latter from December
to March (Lang and Barros, 2004; Bookhagen and Burbank, 2006).
The Monsoon season is caused by moist air currents from the Arabian Sea and Bay of
Bengal. Monsoon rainfall occurs mainly due to heat difference of the land and sea. The heat
difference creates pressure gradients causing wind fluxes from ocean to land (Muslehuddin
et al., 2005). The moist air from the ocean moves towards the north, passing through the hot
basin plain (Houze et al., 2007). Most of the rainfall in summer is due to this phenomenon,
causing intensive convective rainfall (Singh and Kumar, 1997). It is intensive in the months
of June, July and August.
18
The weather systems responsible for winter rainfall are mid latitude Western Disturbances
(Thayyen and Gergan, 2009). They originate over the Caspian Sea and move from the west
to east (Singh and Kumar, 1997).These are formed due to large scale interaction between
the mid latitude and the tropical air masses. The interaction process results in the formation
of westerly troposphere synoptic scale waves. These disturbances cause stratiform rainfall.
The orographic effect may cause intensification, resulting in extensive cloudiness, heavy
precipitation and strong winds. However, sometimes their movement slows down causing
local heavy snowfall over the hilly areas (Dimri, 2006).
2.3
Indus river, major tributaries and doabs
The Indus River originates in Mount Kailash in Tibet (China) on the north side of the
Himalayas at an altitude of 5,486 m (Jain et al., 2007). The Indus is fed by 24 tributaries
with eight as major tributaries. The Jhelum, the Chenab, the Ravi, the Sutlej and the Beas
Rivers are east flowing tributaries, while the Kabul, the Gomal and the Gilgit Rivers flow
west and north, respectively.
The Jhelum River originates in the upper end of Kashmir valley and joins the Chenab River
near Trimmu barrage in Pakistan. The origin of the Chenab is in the Himalayas and flows
into the Himachal Pradesh (India) and Jammu and Kashmir states. Further down, the
Chenab enters Pakistan upstream of the Marala barrage. The Ravi River originates near the
Kangra district of Himachal Pradesh and joins the Chenab in Pakistan. The Sutlej River
arises from the lakes of Mansarover and Rakastal in the Tibetan Plateau at an elevation of
about 4,570 m. The Sutlej joins the Chenab at Panjand (Pakistan). The Beas River
originates in the Rohtang Pass in Himalayas at an elevation of 3,960 m and joins the Sutlej
above Harike in India before entering into Pakistan. The Chenab then flows into the Indus
above Guddu barrage (Pakistan). The Gilgit River arises in the northern areas of Pakistan
with upper reaches mostly glaciated and covered with permanent snow. The Kabul River
originates in the south-eastern slopes of the Hindu Kush range in northern Pakistan. It flows
through the Chitral valley of Pakistan and then enters Afghanistan to meet the Indus further
down, above the Kalabagh barrage near Attock in Pakistan. All these tributaries of the
Indus are generally fed by snowmelt and monsoon rains in the summer (85%) and partially
by rains in winter (15%). The average seasonal flows of the major rivers in the Indus Basin
with their source of origin are given in the Table 2.1.
Table 2.1 Average seasonal flows of the Indus River and its tributaries
Rivers
Indus
Jhelum
Chenab
Ravi
Origin
Mount kalash,
Tibet(China)
Jammu &
Kashmir state
Himachal
Pardesh
Himachal
Pardesh
Length Catchment area
(km)
(km2)
Average flow
3 -1
(km yr )
3,180
288,000
83.15
816
39,200
28.7 (Mangla)
1,232
41,760
29.0 (Marala)
880
24,960
4.46 (Madhopur)
Major
Reservoirs
(Tarbela) Tarbela
Wular,
Mangla
Salal,
Baglihar
Thein
19
Beas
Himachal
Pardesh
Mount kalash,
Tibet(China)
Hindukush
range
Lake Shandur,
Pakistan
Sutlej
Kabul
Gilgit
464
9,920
16.0 (Mandi)
1,536
75,369
18.0 (Ropar)
700
12,888
21.4 (Warsak)
Pong,
Pandoh
Bhakra,
Nangal
Sources: (Thatte, 2008; FFC, 2009; ICID, 2009)
The Indus plain consists of relatively flat zones between the Indus River and its major
tributaries i.e. Jhelum, Chenab, Ravi, Beas and Sutlej. Each flat zone is called a doab,
meaning a land bounded by two rivers (Thatte, 2008). There are five doabs in the Indus
Basin namely the Thal doab (land between the Indus and Jhelum rivers), the Chaj doab
(between the Jhelum and Chenab rivers), the Rechna doab (between the Chenab and Ravi
rivers), the Bari doab (between the Ravi and Beas rivers) and the Bist doab (between the
Beas and Sutlej rivers). These plains produce little runoff compared to the hilly areas which
contribute the major portion of the runoff. Table 2.2 summarizes the total and irrigated
areas in each doab in the basin as well as Pakistan’s part of the basin.
Table 2.2 The doabs in the Indus Basin and area under irrigation in each doab
No
Basin’s Area
Total Irrigated
area
area
(mha)
(mha)
1
Thal
Indus, Jhelum
3.2
1.25
2
Chaj
Jhelum, Chenab 1.05
0.85
3
Rechna Chenab, Ravi
3.12
2.80
4
Bari
Ravi, Beas
3.87
3.50
5
Bist
Beas, Sutlej
1.02
0.83
Sources: (Kureshy, 1977; Ullah et al., 2001; Qureshi et
Bastiaanssen, 2010)
2.4
Doab
Encompassing
rivers
Pakistan’s Area
Total Irrigated
area
area
(mha)
(mha)
3.2
1.25
1.05
0.85
2.97
2.29
3.01
2.73
−
−
al., 2002; Cheema and
Groundwater
A basin level study conducted by WAPDA Pakistan in 1965 described the nature of the
aquifers in the basin. According to WAPDA (1965), “the Indus plain is underlain by deep,
mostly over 300 m deposit of unconsolidated, highly permeable alluvium consisting
primarily of fine to medium sand, silt, and clay. Fine-grained deposits of low permeability
generally are discontinuous so that sands, making up to 65 to 75 percent of the alluvium,
serve as a unified, highly transmissive aquifer”. The use of groundwater for irrigation and
low levels of replenishment of the aquifers resulted in high levels of depletion.
The groundwater within the basin varies spatially in terms of its water table and water
quality, depending on usage (agricultural and domestic). Before inception of irrigation
systems in the basin, the groundwater table varied between 20-30 m. Recharge from earthen
canals and irrigated fields resulted in a significant rising of the water table in certain
20
locations, while the conjunctive use of ground water with surface water has resulted in
lowering of the water table at other areas. The seasonal fluctuations of water table before
and after monsoon for the year 2002 are provided in the Figure 2-3. These water table maps
were provided by the International Water Logging and Salinity Research Institute
(IWASRI).
Figure 2-3 Pre and post monsoon depth to water table in irrigated areas of the Pakistani
part of the Indus Basin for the year 2002.
The water table depth in the irrigated areas of the Punjab and Sindh provinces is more than
30 m before monsoon except in a few pockets. Some areas of the aquifer have depths even
more than 120 m. A rise in the water table is observed after the monsoon season especially
in the Sindh province. These areas are under serious threat of water logging because water
rises significantly after rains. In general, a continuous trend of water table decline is
observed, especially in the Punjab province, which points to a serious imbalance between
abstractions and recharge. Figure 2-4. for example, shows how the areas (with groundwater
table depth of 30 m or more) increased between 1982 and 2002 in different canal
commands in the Pakistani part of the basin. The canal command in the Punjab province
showed a significant increase in the area with water table depths of 30 m or more over a 20
year period (1982-2002). The canal commands in the Sindh province showed a reduction in
areas with a 30 m depth to water table.
21
Figure 2-4 Change in area with groundwater table depth 30 m and more between 1982 and
2002 in different canal commands of the Indus Basin.
The era of unlimited dugwell and tubewell installations has encouraged farmers to
augment shortages in surface water with groundwater (Shah et al., 2000). Excessive
pumping resulted in deteriorating groundwater quality and diminishing phreatic surfaces
across the Indus Basin. Poor quality groundwater and water logged soils occur in the
downstream areas. A persistent flow of about 12.3 km3 is required below Kotri barrage
(the last gauged structure on the Indus) to meet environmental flow requirements of the
river, reduce salinity, and control sea water intrusion (PILDAT, 2003). Flow can be as low
as 0.36 km3in drought years, to as high as 113 km3 in wet years.
2.5
Agriculture and cropping pattern
The Basin provides food for 200 million inhabitants. Irrigated agriculture is practiced in
large parts of the basin (~22.6% of total area) to meet food requirements. Rainfall is not
sufficient to meet the crop water requirements. Monthly crop water requirement (CWR:
difference between ETo and effective rainfall) for two selected stations (Lahore and
Hyderabad) is provided in
Figure 2-5. Hyderabad is located in a relatively drier part of the basin with low rainfall and
higher ETo values resulting in higher CWR as compared to Lahore. Lahore receives
sufficient rainfall especially in the monsoon; thus limiting CWR in the months of July and
August.
22
Figure 2-5 Monthly variation in crop water requirement at two selected location
(Hyderabad and Lahore) representing the lower and middle part of the Indus Basin.
There are normally two agricultural growing seasons: the rabi covering November,
December, January, February, March, April; and the kharif covering May, June, July,
August, September and October.
Rainfed agriculture is practiced in upstream parts of the Indus Basin. “Savanna deciduous”
(11.1%), “pastures deciduous alpine” (6.7%), “pastures deciduous” (6.5%), and “bare soil”
(6.3%) are other dominant land use classes in the basin(Cheema and Bastiaanssen, 2010).
Cropping pattern is defined as the sequence in which crops are grown in a given area over a
period. A specific cropping pattern is in practice in the basin and farmers rarely change it.
The growing season is sufficiently long for two crops and double cropping is widely
practiced. There are varieties of crops grown, but wheat is the dominant crop in rabi and
rice and cotton in kharif. There are also tracts of sugarcane that is a full year crop. Orchards
are also grown on 3.6% of the basin area mixed with other crops (Figure 2-6 ). Seasonal
fodder crops are also grown to meet the needs of livestock. Historical data show no
significant change in cropping patterns.
Figure 2-6 Rice and Orchard fields in the irrigated areas of the Indus Basin
23
24
3 Land use and land cover classification in the irrigated
Indus Basin using growth phenology information from
satellite data to support water management analysis
Chapter based on: Cheema, M.J.M. and Bastiaanssen, W.G.M., 2010. Land use and land cover
classification in the irrigated Indus Basin using growth phenology information from satellite
data to support water management analysis. Agricultural Water Management 97, 1541-1552.
3.1
Introduction
Advanced modeling of hydrological processes in vast river basins and continental scale
land-atmosphere interactions requires land cover and land use information. Note that land
use and land cover are not interchangeable. The term land cover defines the physical
surface conditions. Examples of land cover classes are water, bare soil, grass, crops, forests
etc. Land use reveals the type of application that humans have created to their own benefits.
Examples of land use variation for the same land cover class are industrial settlements vs.
residential areas (land cover: build up), irrigated vs. rainfed cropland (land cover:
cropland), reservoirs vs. natural lakes (land cover: water), recreation vs. pastures (land
cover: grass), timbering vs. environment (land cover forests). In case of a mixture of land
cover and land use classes on the same map, it is common to refer to Land Use and Land
Cover (LULC) classes.
LULC information is often used to define the physical land surface properties such as curve
numbers (for runoff), surface roughness (for evapotranspiration ET), albedo (for ET),
vegetation cover (for biodiversity), rooting depth (for available soil moisture), drought
resistance, etc. Hence, modeling of the water balance and preparation of water accounting
(Molden, 1997) requires LULC to be known. The process of water accounting is generally
based upon the amount of water used by different land uses. A number of frameworks have
been developed for water accounting (Niblack and Sanchez, 2008; Turner et al., 2008).
These frameworks distinguish beneficial and non-beneficial water use by land use classes.
This is only possible with accurate LULC classification.
Several global scale land cover databases have been developed since the early nineties. e.g.
Global Land Cover Characteristics (GLCC) database carried out under the flag of the
International Geosphere-Biosphere Program (IGBP) ,International Water Management
Institute (IWMI) land cover database, Global Land Cover (GLC) 2000 and Global land
Cover (Glob Cover) developed by European Space Agency (ESA). GLCC database was
developed using 1 km resolution Advanced Very High Resolution Radiometer (AVHRR)
data of monthly Normalized Difference Vegetation Index (NDVI) using unsupervised
classification and subsequent refinements were carried out to develop land cover database
for the year 1992-93 (Loveland et al., 2000). IWMI database was based on the year 2001-02
and used Moderate Resolution Imaging Spectroradiometer (MODIS) 500 m NDVI datasets.
It was developed using unsupervised classification and refined by ground truth data.
GLC2000 was global land cover database developed using 9 months Satellite Probatoire
d’Observation delaTerre (SPOT) vegetation data for the year 1999-2000. Glob Cover was
25
developed recently using the 300 m resolution Medium Resolution Imaging Spectrometer
(MERIS) satellite data for the year 2005-06 (Arino et al., 2008).
The mapping accuracies of these global products are reasonable (i.e.63% – 83%) but these
were developed for use in global climate studies. All these data sets are most useful for
analyzing general land cover patterns at a continental or large scale. These global datasets
are, however, inappropriate to support basin water management analysis and applications
because they lack details on land use information. Such global databases cannot distinguish
specific crops and only detects dominant land covers leading to a large percentage of mixed
classes with natural vegetation (Portmann et al., 2010). Since within irrigated land uses,
different crops are grown which have specific crop water requirements. Therefore, more
specific crop based classifications are needed to implement proper water allocation plans.
General land cover information by means of a few classes, without the land use functioning
is thus insufficient and novel techniques to determine land use classes need to be developed
(Meyer and Turner, 1994; Schwarz and Zimmermann, 2005).
Therefore, in this paper a methodology has been presented to derive LULC for the vast
Indus Basin. This is done by integrating satellite derived information on NDVI time series
with ground information and expert knowledge on the growing patterns (phenology) of the
crops. This information is further applied to identify different crop rotations in growing
seasons in order to get real pattern of water use. The goal is to develop an up to date,
regionally consistent and detailed LULC map with affordable efforts that could also be
applied to obtain LULC information for other basins with water resources problems in the
world. This database will be used in further hydrological studies on the Indus Basin.
The Indus Basin was selected for this study because the water resources of this international
river basin are under pressure. There is not enough water to supply all water use sectors
with sufficient quantity. Although good efforts have been undertaken (Habib, 2004), there
is a genuine interest to better understand the hydrology and water management issues of the
Indus Basin.
3.2
Study area
The international Indus Basin is located in four countries (Figure 3-1). The basin lies in
between24° 38′ to 37° 03′ N latitude and 66° 18′ to 82° 28′ E longitudes. The life line of the
Indus Basin is the Indus River that traverses China, Afghanistan, India and Pakistan, when
moving from upstream to the downstream end of the basin. The total size of the basin is
116.2 million hectares (mha). The vast area of the basin is located in Pakistan (53% of
total). The area in India is 33% followed by China and Afghanistan with 8% and 6%,
respectively. The basin is bounded by the Karakoram and the Hindukush ranges on the
north, the Himalayas on the north-east, the Sulaiman and the Kirther ranges on the west and
the Arabian Sea on the south.
26
Figure 3-1 Location of Indus Basin and provinces of different countries in the basin with
information on main tributaries of the Indus river.PK stands for Pakistan and IN for India.
The Indus river originates from Mount Kailash in Tibet (China) on the north side of the
Himalayas at an altitude of 5,486 m (Jain et al., 2007). The Indus river is fed from 7 major
tributaries. The Jhelum, the Chenab, the Ravi, the Sutlej and the Beas rivers are the eastern
tributaries, while the Kabul and the Gilgit rivers being western and northern tributaries,
respectively (Figure 3-1).
The Jhelum River originates from the upper end of Kashmir valley and joins the Chenab
River near Trimmu barrage in Pakistan. Origin of the Chenab River is in the Himalayas and
flows into Himachal Pradesh (India) and Jammu and Kashmir state. Afterwards, the Chenab
River enters in Pakistan at upstream of Marala barrage. The Ravi River rises near the
Kangra district of Himachal Pradesh and joins the Chenab River in Pakistan. The Sutlej
River rises from the lakes of Mansarover and Rakastal in the Tibetan Plateau at an elevation
of about 4,570 m. The Sutlej River joins the Chenab River at Panjand (Pakistan). The Beas
River originates from Rohtang Pass in Himalayas at an elevation of 3,960 m and joins the
Sutlej River in India before entering into Pakistan. The Chenab River then flows into the
Indus River near Guddu Barrage (Pakistan). The Gilgit River arises in the Northern areas of
Pakistan with upper reaches mostly glaciated and covered with permanent snow. All these
tributaries of the Indus River are generally fed with snow melt and monsoon rains in the
summer (85%) and partially with rains in winter (15%).
27
The climate of the basin varies spatially and is characterized by large seasonal fluctuations
in temperature and rainfall. The upper (northern and north-eastern) parts experiences harsh
winter with significant snowfall while the middle and lower part has comparatively mild
winters but hot summers. The average annual rainfall varies from less than 200 mm in the
desert area to more than 1500 mm in the north and north-east of the basin. Two growing
seasons (rabi and kharif) normally prevails in the basin. The rabi season comprises months
of November, December, January, February, March and April while May, June, July,
August, September and October represents kharif season.
3.3
Methodology
Every LULC class has particular phenological variations of vegetation cover throughout the
year. Vegetation cover describes the foliage covered fraction of a natural land surface. The
duration of various cropping seasons ranges from 90 days to 150 days and multi-temporal
images of vegetation cover capture these dynamics. Vegetation indices are based on
differential absorption, transmittance, and reflectance of spectral radiance by the vegetation
in the red and near-infrared electromagnetic radiation. Green leaves absorb most of the
radiation in the visible part and reflect in the near-infrared spectrum range. The green leaf
density increases the photosynthetic activity. The unique behavior of green leaf
development in terms of duration and peak of phenological stages in various agroecosystems makes these ecosystems distinguishable using time series analysis.
The most common vegetation index is the NDVI derived from the visible and near infrared
channel reflectance (Tucker, 1979). The NDVI time series can be helpful in vegetation
detection (Hansen et al., 2000; Wardlow and Egbert, 2008). For the current study, multitemporal NDVI images were obtained from the Vegetation (VGT) sensor, which is on
board of the SPOT satellite. The spatial resolution of the sensor is 1 km while its 2250 km
swath width enables the sensor to acquire data on a daily basis. The daily global coverage
and spatial resolution, makes it a very suitable sensor for large basin vegetation mapping
(Mucher and de Badts, 2002).
The SPOT VGT sensor has four spectral bands: blue (0.43-0.47 μm), red (0.61-0.68 μm),
near-infrared (NIR: 0.78-0.89 μm) and shortwave infrared (SWIR: 1.58-1.75 μm). The red
and NIR bands are used to characterize vegetation. The raw NDVI data set covering the
Indus Basin for the year 2007 were downloaded from the Processing and Archiving Image
Center, hosted by VITO, the Flemish Institute for Technological Research, Belgium
(http://free.vgt.vito.be). This site provides geometrically and radiometrically corrected 10day synthesis SPOT VGT NDVI (V2KRNS10) products. Thirty six NDVI images from
January 2007 to December 2007 were obtained in order to include full phenological
information of one complete annual cycle. The SPOT vegetation data were originally
available in Digital Numbers (DN) that were then converted into NDVI using the following
equation.
3.1
Since no ground observations were available initially, an unsupervised classification (e.g.
Giri and Jenkins, 2005) was performed on the stacked image. Unsupervised classification
identifies clusters by their spectral similarities (in this case 36 spectra; one for each 10
days) and allows the feature space to segment into similar spectral clusters (Rashid, 2007).
28
Different clustering methods are available for unsupervised classification like k-mean
method and Iterative Self Organizing Data Analysis Technique (ISODATA). ISODATA is
based on Euclidean distance, in which spectral distances between candidate pixels are
compared to each cluster mean. A pixel is assigned to the cluster whose mean is closest to
the candidate pixel. New cluster centers are computed by averaging the locations of all the
pixels assigned to that cluster (Campbell, 2002). An ISODATA technique was used with
95% convergence threshold for land use classification.
The classification was refined by applying expert knowledge of the cropping patterns
adopted in the basin (Figure 3-2). Multi-cropping systems can be observed in irrigated
areas. The knowledge of crop phenology helped to identify crops with specific NDVI
temporal profiles. The beginning of a growing season for a particular crop was considered
when there was a significant increase in NDVI. The onset of the NDVI is a result of
increased photosynthetic activity. Similarly, a decrease in the NDVI reflects the end of
growing period. Hence, NDVI time profiles can be used to identify multi-crops during the
growing season.
Figure 3-2 Cropping calendar adopted in the Indus Basin. Rabi expresses the winter crop
and kharif the summer crop.
The possible effects of physical conditions (e.g. temperature, elevation and slope) on land
use classes were also studied. For this, a 90 m resolution Digital Elevation Model (DEM)
was obtained from the Shuttle Radar Topographic Mission (SRTM) database (Jarvis et al.,
2008). The average temperatures were attained from temperature datasets (period 1961-90)
developed by the Climatic Research Unit, University of East Anglia, United Kingdom
(New et al., 2002). The mean temperature, elevation and slope were extracted for individual
class to examine the spatial variability of land use due to physical conditions.
Four types of accuracy assessment studies were performed in order to assess the quality of
the generated LULC map. First is the classical error matrix approach, which uses
independent classification and reference data to have a precise knowledge of the ground
situation (Latifovic and Olthof, 2004). A ground truthing survey was conducted during
September-October to capture peak kharif cropping season and in January to coincide with
the rabi cropping season conditions. Due to the vast dimension of the Indus Basin, ground
truthing was focused on the middle and lower reaches that have different agro-ecological
regions. The class with 70% dominance of certain land use was selected. (e.g. if 70% of the
observed area was wheat, then the area was classified as wheat).
Existing global and regional (IGBP, IWMI,GLC2000) land cover maps were compared
with the newly developed map as a second accuracy assessment test. One map was
produced by Boston University, Department of Geography using MODIS land cover
product MOD12Q1 V004 based on 17 classes IGBP scheme in 2004 (MODIS, 2004). The
29
map was accessed from http://duckwater.bu.edu/lc/mod12q1.html. The other map was
developed by (Thenkabail et al., 2005) using MODIS 500 m NDVI time series for the year
2001, and GLC2000 (Agrawal et al., 2003) used 1 km SPOT-Vegetation NDVI time series.
A subset covering the Indus Basin was extracted from those maps. The subset maps were
compared after rescaling to a single 1 km resolution. For the sake of comparison, some
classes were merged to make analysis easier.
The third accuracy test was carried out by using local studies. These local studies (Hussain,
1998; Bastiaanssen et al., 2003; Singh et al., 2006), focus on the southern and eastern parts
of the Indus Basin. Furthermore, ancillary data was used for refinement as well as fourth
accuracy test. Ancillary data included agro-ecological regions and cropping pattern maps
developed by Pakistan Agricultural Research Council (PARC), Islamabad, district based
Agricultural Statistics of Pakistan compiled by Ministry of Food, Agriculture and Livestock
(MINFAL), Government of Pakistan, Food and Agriculture Organization (FAO) and
International Centre for Integrated Mountain Development (ICIMOD) publications. The
limitation in data collection from Indian side of the basin restricted to use data from
Pakistan only. The area fractions of different crops grown in different administrative units
reported by MINFAL and estimated by remote sensing were compared. The coefficient of
determination (R2) was used to check the reliability of the estimates.
3.4
3.4.1
Results and discussion
Phenology
Initially clustering of five classes was applied for the separation of the total Indus Basin
into (i) water/ice, (ii) barren, (iii) shrubland /grassland, (iv) natural vegetation (forests) and
(v) cropland. The division of the basin into these five classes provided a first understanding
of spatial variability in land cover (Table 3.1). The class “Barren” appears to be a major
class for all countries. The areal extent is followed by the class “croplands” for Pakistan and
India. Afghanistan has a small area under “croplands”, while it is negligible in China. The
area in the Chinese part of Indus Basin is located at a higher altitude where no agriculture is
possible. The Chinese part is mostly comprises of “water/ice” and “barren” land covers. By
absence of vegetation and the dynamics of snow and snowmelt processes, it is not
straightforward to differentiate between “water/ice” and “barren” with the information
used.
Table 3.1 Major LULC classes and their spatial distribution among the countries of Indus
Basin.
LULC type
Water/ice
Barren
Shrubland/grassland
Natural vegetation
Croplands
Total
30
Pakistan
Area (mha)
6.33
21.46
10.84
9.77
13.13
61.53
India
Area(mha)
7.68
10.51
4.26
3.95
12.52
38.92
China
Area(mha)
4.66
4.22
0.01
−
−
8.89
Afghanistan
Area (mha)
0.82
4.51
1.30
0.51
0.04
7.18
The number of classes was then increased to identify the land use. This resulted into a first
round of 45 classes which provided the basis for further refinement and analysis. This
classification was made by taking into consideration cropping calendar and dominance of a
particular crop in the area. The class cropland was partitioned using NDVI temporal
profiles and expert knowledge of cropping patterns. Some classes were merged on the basis
of information obtained during ground truthing. NDVI profile similarities were also
considered during this merging.
This procedure reduced the number of classes from 45 to 27. The resulting mean NDVI
time profiles of the final agricultural classes are shown in Figure 3-3. Figure 3-4 shows
only natural land cover classes. These NDVI time profiles reflect the mean values for each
individual class. A three period moving average filter is used to smooth the profile as
described by Reed et al.(1994). It can introduce time lag which is avoided by using one
period before and one period next from the time of analysis.
Figure 3-3 Mean NDVI temporal curves for irrigated and rainfed crops in Indus Basin for
the year 2007.A moving average of 3 periods has been used to smooth the lines.
31
Figure 3-4 Mean NDVI temporal curves for forests, pastures and savannas in the Indus
Basin for the year 2007.
Figure 3-3 and Figure 3-4 confirm that these 27 clusters have a high degree of
seperatability: none of the final classes has a similar phenology. The ISODATA clustering
technique met the expected goal to separate phenological differences such as the start of
growing season, the end of the growing season and growing length of a particular crop. The
two peaks in one annual cycle (Figure 3-3) represent multi cropping systems with intensive
irrigation practices. The start and end of these peaks distinguish major crop types grown in
the study area.
The class “irrigated rice, wheat rotation” has two distinct peaks, one at the end of February
and other at the end of August. This matches with the cropping period of wheat from
November to April (i.e. rabi) and rice from June to October (i.e. kharif). Moreover, the area
in this class was completely cultivated with only a very small percentage of fallow land
thus showing a higher NDVI from all other crop types.
The rainfed crops have lower NDVI values and their timing of peak may differ due to
pattern difference in natural climatic conditions as compared with irrigated area. The
recession of NDVI curves for rainfed crops like “rainfed crops mixed cotton, wheat
rotation/fodder” and “rainfed crops general” show similar trends as the one’s of irrigated
crops but with lower NDVI values. The crop labeled as “rainfed crops wheat/grams” shows
a different trend than other rainfed crops. Under this class, only rabi crops (wheat and
grams) were grown. NDVI values for the rest of the year is flat showing no distinct
vegetation.
32
Similarly, Figure 3-4 clearly differentiates between forests, pastures, savannas, snow and
desert. The evergreen forests (both needleleaf and broadleaf) in the northern region of the
Indus Basin have temporal stable NDVI values throughout the year. However, the class
“forests deciduous alpine” shows a sudden decline in NDVI during the months of
December until March. Presence of snow, during winter, has resulted in suppressed foliage
cover. Emergence of new leaves in April, improves greenness and higher NDVI values are
observed in the remaining months.
The NDVI time series is effective to separate seasonal and permanent snow cover in the
study area. Usually snow has very low NDVI values (< 0) because the near-infrared
reflectance of water is smaller than for red reflectance. For permanent snow, the NDVI
values are lower than zero for the whole year while the class “snow and ice temporary” has
NDVI values lower than zero during the winter season only. The resulted LULC map is
presented in Figure 3-5.
Figure 3-5 Land use and land cover map of the Indus Basin developed from 36 SPOTVegetation based NDVI values covering the annual vegetation phenological cycle of 2007.
The international boundaries are superimposed on the map.
The areal extent of the LULC classes is summarized in Table 3.2. Deciduous savanna with
12.94 mha is the most dominant land use type, followed by rainfed crops in general (11.71
mha). Deciduous savanna appears in the Kabul river sub basin and in the Himalaya
mountain ranges. The dominant class in the Indus Basin is irrigated agriculture (23%),
followed by pastures (16.2%), rainfed crops (16.1%), savannas (15.9%) and forests (10%).
Irrigated rice, wheat rotation appears to be the most common irrigated crop (9.69 mha). An
area of 3.25 mha of irrigated rice, wheat rotation class is located in Pakistan. This is 8%
33
higher than reported by MINFAL (2007). This can be explained by the fact that the gross
area of one pixel is 100 ha (1 km×1 km) and not all land within on pixel consists of
agricultural fields.
Table 3.2 LULC types and their areal distribution in the Indus Basin
3.4.2
Effect of physical condition on LULC
The physical condition such as elevation, temperature and slope can affect the spatial
distribution of land use classes as indicated by Mahajan et al. (2001), Wang et al. (2003)
and Fang et al. (2005). They conducted their studies in an Indian watershed, USA and
North China, respectively. Such effects of physical conditions are also significant in the
Indus Basin. The elevation traverses longitudinally across the entire basin with maximum in
the north and minimum towards south. The temperature is inversely related with elevation.
Higher temperatures are usually observed in the plain of Indus Basin with very mild slope
and average elevation 100 m to 300 m above Mean Sea Level (a.m.s.l). The different LULC
classes with respect to orography are shown in Table 3.3.
34
Table 3.3 LULC classes distribution in Indus Basin with respect to (i) elevation a.m.s.l (ii)
slope and (iii) temperature.
No
Class name
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Snow and ice permanent
Snow and ice temporary
Bare soil
Very sparse vegetation
Pastures deciduous
Pastures evergreen lowland
Pastures deciduous alpine
Savanna evergreen open
Savanna evergreen closed
Savanna deciduous
Forests evergreen needleleaf
Forests evergreen broadleaf
Forests deciduous alpine
Forests/cropland alpine
Irrigated mixed cotton,wheat
rotation/orchards
Irrigated mixed cotton,wheat
rotation/sugarcane
Irrigated rice,wheat rotation
Irrigated mixed rice,wheat
rotation/cotton
Irrigated wheat,fodder rotation
Irrigated rice,fodder rotation
Irrigated mixed rice,wheat
rotation/sugarcane
Rainfed crops wheat/grams
Rainfed crops mixed
cotton,wheat rotation/fodder
Rainfed crops general
Rainfed crops and woods
Urban and industrial settlements
Water bodies
16
17
18
19
20
21
22
23
24
25
26
27
Average
elevation
(m)
5220
4900
690
2000
450
1250
4450
1210
1250
3540
1085
1715
3370
2175
200
Average
slope
(Degree)
23.0
23.8
2.7
8.2
3.5
8.8
14.3
11.4
14.6
20.3
15.3
20.3
26.8
22.6
0.9
Average
annual 24 h T
(°C)
-4.2
-2.7
24.6
13.6
24.2
19.1
2.2
19.8
16.9
4.5
16.6
13.8
4.0
9.2
25.7
225
0.8
24.0
210
165
0.6
0.7
26.0
25.6
175
100
150
0.6
1.2
2.4
25.6
26.5
25.3
185
200
0.6
1.2
26.2
27.2
375
840
350
430
3.5
7.9
2.4
1.6
24.7
21.0
23.5
24.3
Low air temperatures are observed at the higher altitudes in the basin. This region is
dominated by snow which suppresses vegetation. Majority of LULC falls under the classes
“snow and ice permanent” and “snow and ice temporary”. The class “snow and ice
permanent” occurs at higher elevations (5220 m) and 23 degree average slope. This class
covers most of the glaciated part of the basin. The low temperature allows snow to sustain
permanently in the form of glaciers. An increase in temperature during summer is not
sufficient to cause the snow to melt. The second class in this region is “snow and ice
temporary”, that attains an average elevation of 4900 m with steep slopes (~24 degrees). At
35
this elevation, the temperature rises over the freezing point during summer. This causes the
snow to melt, thus allowing vegetation to grow on the moist soils. Moreover, the slope for
the second class is relatively higher which enhances snow melt. It provides a sound reason
to divide snow and ice into two classes. Figure 3-6(a) & (b) better interprets the effect of
temperature on NDVI values for these two classes.
Snow and ice temporary
Snow and ice permanent
15
15
1.0
Temperature
1.0
Temperature
NDVI
NDVI
0.9
0.9
10
10
0.8
0.8
0.7
5
0.7
5
0.6
0
r
r
em
be
0.4
D
ec
ob
er
ct
N
ov
O
te
-5
em
be
t
r
m
be
ly
us
Au
g
Se
p
ne
Ju
Ju
il
ay
M
ar
c
Ap
r
h
0.5
M
r
r
em
be
D
ec
ob
er
ct
em
be
O
N
ov
t
r
m
be
te
ly
us
Au
g
Se
p
ne
Ju
Ju
il
ay
M
h
ar
c
Ap
r
M
0.4
NDVI
0.5
-5
Temperature C
Ja
nu
ar
y
Fe
br
ua
ry
0
NDVI
Temperature C
Ja
nu
ar
y
Fe
br
ua
ry
0.6
0.3
0.3
-10
-10
0.2
0.2
0.1
0.1
-15
-15
0.0
0.0
-20
-0.1
-20
-0.1
Months
Months
(a)
(b)
Savanna deciduous
Pastures deciduous alpine
20
20
1.0
Temperature
1.0
Temperature
NDVI
NDVI
0.9
0.9
15
15
0.8
0.8
10
0.5
0
r
r
D
ec
em
be
N
ov
em
be
r
ob
er
O
ct
m
be
te
ly
us
t
Ju
e
Ju
n
ay
M
il
Ap
r
Au
g
Se
p
-5
h
ua
r
Ja
n
0.3
ar
c
y
0.4
r
r
em
be
D
ec
em
be
ct
O
Se
p
N
ov
m
be
te
ob
er
r
t
ly
us
Au
g
ne
Ju
Ju
il
ay
M
Ap
r
h
ar
c
ua
ry
M
nu
ar
Ja
-5
Fe
br
y
0.4
ua
ry
0
0.6
5
0.2
-10
NDVI
0.5
Fe
br
5
0.7
M
0.6
Temperature C
0.7
NDVI
Temperature C
10
0.3
0.2
-10
0.1
-15
0.0
0.1
-15
0.0
Months
Months
(c)
(d)
Forests deciduous alpine
Forests evergreen needleleaf
30
1.0
Temperature
30
1.0
NDVI
Temperature
NDVI
0.9
25
0.9
25
0.8
0.8
20
20
0.7
0.7
0.5
0.4
5
15
0.6
NDVI
10
NDVI
Temperature C
0.6
Temperature C
15
10
0.5
0.4
5
0.3
0
0.3
Months
(e)
36
r
r
em
be
D
ec
em
be
r
ob
er
m
be
ct
O
N
ov
t
ly
us
Ju
ne
Ju
ay
M
il
Ap
r
h
ar
c
ua
ry
0.2
0.1
0.0
Months
(f)
te
-10
Se
p
0.0
Au
g
-5
M
0.1
Fe
br
y
nu
ar
Ja
r
r
em
be
D
ec
em
be
ob
er
ct
O
m
be
te
N
ov
r
t
us
Au
g
ly
-10
Se
p
ne
Ju
Ju
ay
M
il
Ap
r
h
ar
c
ua
ry
M
nu
ar
Ja
-5
Fe
br
y
0
0.2
Irrigated mixed cotton,wheat rotation/orchards
Irrigated rice,wheat rotation
40
1.0
Temperature
1.0
40
NDVI
Temperature
0.9
35
NDVI
0.9
35
0.8
0.8
30
30
0.7
0.6
0.5
0.4
15
25
0.6
0.5
20
0.4
15
0.3
0.3
10
10
0.2
0.2
5
0.1
0.0
Months
r
r
em
be
O
D
ec
ob
er
em
be
ct
m
be
te
Se
p
N
ov
t
r
ly
Au
g
us
ne
Ju
Ju
il
ay
M
h
ar
c
Ap
r
nu
ar
Ja
ua
ry
y
r
em
be
r
0.1
0
D
ec
ob
er
ct
O
em
be
m
be
te
N
ov
t
r
ly
us
Au
g
Se
p
ne
Ju
Ju
il
ay
M
h
Ap
r
ar
c
ua
ry
M
Fe
br
Ja
nu
ar
y
0.0
M
0
Fe
br
5
Months
(g)
(h)
Rainfed crops wheat/grams
Rainfed crops mixed cotton,wheat rotation/fodder
40
1.0
Temperature
40
1.0
NDVI
Temperature
0.9
35
NDVI
0.9
35
0.8
0.8
30
30
0.7
0.6
0.5
0.4
15
25
0.6
20
0.5
0.4
15
0.3
10
0.3
10
0.2
0.1
Months
r
De
c
em
be
r
em
be
No
v
r
ob
er
m
be
ct
te
O
t
us
ly
ne
Ju
Ju
ay
M
il
Ap
r
h
ar
c
Au
g
Se
p
Ja
nu
ar
y
0.0
ua
ry
0
r
em
be
r
De
c
ob
er
em
be
No
v
r
m
be
ct
O
t
us
te
Au
g
Se
p
ly
ne
Ju
Ju
ay
M
il
h
Ap
r
ar
c
M
Fe
br
nu
ar
ua
ry
0.0
y
0
0.2
M
0.1
5
Fe
br
5
Ja
NDVI
20
Temperature C
25
NDVI
Temperature C
0.7
(i)
NDVI
20
Temperature C
25
NDVI
Temperature C
0.7
Months
(j)
Figure 3-6 Average temperature and NDVI relationship for (a) snow and ice permanent
(b) snow and ice temporary (c) pastures deciduous alpine (d) savanna deciduous (e)
forests deciduous alpine (f) forests evergreen needleleaf (g) irrigated mixed cotton, wheat
rotation/orchards (h) irrigated mixed rice, wheat rotation (i) rainfed crops wheat/grams
(j) rainfed crops mixed cotton, wheat rotation/fodder.
The class “pastures deciduous alpine” comes next at an average elevation of 4450 m a.m.s.l
and 14 degrees average slope, followed by “savanna deciduous” (3540 m) but with more
slope (20 degrees). Higher altitudes and severe temperatures in winter suppress vegetation.
Therefore, Figure 3-6(c) & (d) show lower NDVI value in winter for these classes. NDVI
gradually increases in summer due to rise in temperature, favorable for vegetation growth.
“Forests deciduous alpine” and “forests evergreen needleleaf” are usually found at an
altitude of 3370 m and 1085 m with average slopes of 27 degrees and 15 degrees,
respectively. Low temperatures in the winter cause defoliation. Thus, NDVI for “forests
deciduous alpine” in Figure 3-6(e) is lower in the winters, and improves significantly, in
summer due to favorable temperature. The “forests evergreen needleleaf” grows at a lower
altitude, and don’t become dormant. Temperature in these altitudes is mild and favorable
for vegetation. Usually evergreen plantation is observed in these areas. The NDVI values in
Figure 3-6(f) shows constant trend across the whole year.
Irrigated agriculture is mostly concentrated in the plain areas of the Indus Basin with an
average slope between 0.6 to 2.4 degrees. The elevation ranges from 100 m in lower Indus
Basin to 225 m in the middle parts of the basin. Extensive food requirements of about one
37
billion inhabitants are met from this irrigated plain. The common food and cash crops are
wheat, rice, sugarcane, cotton, orchards. The mild temperatures during the winter and
higher temperatures in summer, allows two growing seasons. Different growth stages of
crops need specific temperature. For example, wheat needs mild temperature during early
growth stages while requires warmer climate at ripening. Figure 3-6(g) & (h) clearly depict
these phenological variations of crops with temperature trends.
Rainfed crops also help to fulfill the food requirements. The rainfed crops are usually
grown at a higher altitude as compared to irrigated crops (180 to 840 m, a.m.s.l). The
average slope ranges from 0.6 degree (flat areas) to 7.9 degrees. The rainfall varies spatially
for different altitudes. The NDVI curves for “rainfed crops wheat/grams” and “rainfed
crops mixed cotton, wheat rotation/fodder” are shown in Figure 3-6(i) & (j), respectively.
The temporal changes in NDVI values with temperature for different growth stages of crops
reinforce our argument to use phenological cycle to identify land use classes.
3.4.3
Accuracy assessment
The usability of the LULC classification for water management analysis depends on the
reliability of the developed LULC map. A number of accuracy assessment approaches has
been developed (e.g. Congalton, 1991; Friedl et al., 2000; Cihlar et al., 2003). In this study,
a 4-step accuracy assessment was carried out.
3.4.3.1 Ground truthing
As a first step, the error matrix approach described by Campbell (2002) is used to express
the accuracy. An error matrix constructed by plotting LULC classes from the unsupervised
classification algorithm against the LULC information gathered from ground truth data is
shown in Table 3.4. Only classes with ground inspection data are included in the analysis.
38
Irrigated mixed
cotton,wheat
rotation/sugarcane
Irrigated rice,wheat
rotation
Irrigated mixed
rice,wheat
rotation/cotton
Irrigated wheat,fodder
rotation
Rainfed crops
wheat/grams
Rainfed crops mixed
cotton,wheat
rotation/fodder
Rainfed crops general
Rainfed crops and
woods
Urban and industrial
settlements
Water bodies
21
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
0
22
0
0
0
0
0
0
0
0
0
0
0
0
0
0
10
0
0
0
0
0
23
0
0
0
0
0
0
0
1
0
1
0
0
0
0
0
8
0
0
0
0
24
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
2
35
0
2
1
25
0
0
0
0
0
0
0
0
0
0
1
0
2
0
0
0
0
15
0
1
26
0
1
0
1
1
0
0
5
4
2
0
1
1
1
2
1
1
2
43
2
27
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
Tot
3
10
3
4
15
3
8
88
107
42
13
10
19
2
13
11
47
22
50
14
4
75
13
62
89
75
110
69
42
88
12
83
6
83
16
94
1
100
10
100
10
80
41
85
19
79
68
63
9
100
484
3
5
6
8
9
10
11
15
16
17
18
19
20
21
22
23
24
25
26
27
3
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
5
0
9
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
0
6
0
0
3
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
8
0
0
0
3
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
9
0
0
0
0
11
0
0
0
1
0
0
0
0
0
0
0
0
1
0
0
10
0
0
0
0
0
3
0
0
Tot
U.A.
4
75
10
90
3
100
4
75
13
85
P.A = Producer’s accuracy; U.A. = User’s accuracy
Irrigated rice,fodder
rotation
Irrigated mixed
rice,wheat
rotation/sugarcane
Irrigated mixed
cotton,wheat
rotation/orchards
20
0
0
0
0
0
0
0
0
0
0
0
0
15
0
0
0
0
1
0
0
Savanna deciduous
19
0
0
0
0
0
0
0
1
0
0
0
5
0
0
0
0
0
0
0
0
Savanna evergreen
closed
18
0
0
0
0
0
0
0
1
0
0
10
0
0
0
0
0
1
0
0
0
Savanna evergreen
open
17
0
0
0
0
0
0
0
1
3
37
0
1
0
0
0
0
0
0
0
0
Pastures evergreen
lowland
16
0
0
0
0
1
0
0
10
76
2
2
3
1
0
0
0
8
2
4
1
Pastures deciduous
15
0
0
0
0
0
0
0
67
21
0
0
0
0
0
0
0
0
1
0
0
Bare soil
0
0
0
0
0
0
0
1
0
0
0
11
0
0
0
0
1
0
8
2
2
0
0
0
0
0
0
0
0
0
0
0
Ground
classes
Map
Classes
Forests evergreen
needleleaf
Table 3.4 Error matrix for the accuracy assessment of LULC classification.
P A.
100
90
100
75
73
100
100
76
71
88
77
50
79
50
77
73
74
68
86
64
40
A total of 484 points were collected during the ground truthing survey and used in the error
matrix analysis. Twenty LULC classes were visited and has minimally three observation
points. The error matrix in Table 4 shows an overall accuracy of the classified map is 77%.
The average producer’s and user’s accuracy is 78% and 83%, respectively. Hence, 83% of
all land use classes identified on the map agree with the field observation. Considering that
the satellite resolution is coarse (100 hectares) and the field sizes (0.4 hectares) are small,
this accuracy is rather satisfactory and in agreement with accuracy levels achieved in
different land use and crop identification studies (e.g. Bastiaanssen, 1998). He concluded
that with extensive field work, crops can be identified with an average accuracy of 86%.
However, this accuracy level fluctuates from 49% to 96%, depending upon the spatial
coverage of the satellites and field size.
Thunnissen and Noordman (1997) suggested 70% minimum classification accuracy should
be attained at regional scale. Giri and Jenkins (2005) and Wardlow and Egbert (2008)
attained 77.3% and 84% accuracies from MODIS 500 m and MODIS 250 m imagery,
respectively. An accuracy of 73% for heterogeneous and 89% for homogeneous pixels were
attained by Knight et al. (2006) using MODIS 250 m NDVI data. Keeping in view the
heterogeneity of different classes in the present study, the classification accuracy attained is
in fair agreement with the previous work.
The producer’s accuracy of two classes namely “irrigated wheat, fodder rotation” and
“irrigated mixed rice, wheat rotation/sugarcane” is 50% while their user’s accuracy is 83
and 100%, respectively. The apparent reason for this lower producer’s accuracy is the
mixed agricultural cultivation processes that occur at the 100 ha scale. In addition to that, a
smaller number of ground truth points are available for the class “irrigated mixed rice,
wheat rotation/sugarcane” resulting in lower (50%) producer’s accuracy and 100% user’s
accuracy. The three irrigated classes “irrigated mixed cotton, wheat rotation/orchards”,
“irrigated mixed cotton, wheat rotation/sugarcane” and “irrigated rice, wheat rotation”
constitute the major irrigated cropland systems of the Indus Basin. These are distinct with
user’s accuracy of 75%, 69% and 88% while producer’s accuracy 76%, 71% and 88%,
respectively.
The user’s accuracy of three rainfed crop classes, “rainfed crops mixed cotton, wheat
rotation/fodder”, “rainfed crops general” and “rainfed crops and woods” is 80%, 85% and
79%, respectively. The producer’s accuracy of these classes is 73%, 74% and 68%,
respectively. However, the class “rainfed crops wheat/gram” shows 100% user’s accuracy
and 77% producer’s accuracy while remaining 23% of this class is falling in “rainfed crop
general” and “urban and industrial settlements”. The user’s and producer’s accuracy of
irrigated and rainfed crops is more than 70%, in general. These are key classes for water
management analysis. Therefore, it is important that LULC classes with a large acreage
have an acceptable accuracy.
The accuracy assessment using error matrix is normally dependent on sample points.
Limited sample points can lead to assigning the correct class by chance(Foody, 2002). Such
chance agreement effects can be assessed using Kappa (k) coefficient suggested by
Cohen(1960).The Kappa coefficient incorporates the off-diagonal elements of the error
matrices. It represents agreement obtained after removing the proportion of agreement that
could be expected to occur by chance (Congalton, 1991). Kappa coefficient is well suited
for accuracy assessment of LULC maps (Vliet, 2009). Jonathan et al. (2006) attained 0.72
Kappa coefficient in a study conducted in Brazil using MODIS 250 m data. Wang and
Tenhunen (2004) reported lower values of Kappa coefficient (0.14 and 0.20) for IGBP-DIS
released land cover map and map developed by unsupervised classification using AVHRR
NDVI temporal profiles, respectively. Melesse and Jordan (2003) achieved 79% (Kappa =
0.79) agreement between on ground and Landsat derived land use map of Florida using
unsupervised classification. Thapa and Murayama (2009) tested different classification
approaches using ALOS (Advanced Land Observing Satellite) in Japan and reported Kappa
coefficient ranging from 0.71 to 0.87. Kappa coefficient attained in this study was 0.73,
which is in moderate agreement range as suggested by Congalton and Green (1999).
3.4.3.2 Existing databases
The second check was performed against existing LULC maps. These maps encompass
continental or global scale and have single legend classes. For the sake of comparison,
some classes were merged to make analysis easier. A summary of major classes obtained
from different studies is shown in Table 3.5.
Table 3.5 Comparison of major classes derived from different sources.
Class
Irrigated
Rainfed
Forest
Bare soil
Pastures
Snow and ice Perm./temp
Total
Present study,2007
(mha)
29.10
18.61
8.73
7.37
37.40
9.77
110.98
IGBP,2004
(mha)
34.36
0.23
3.11
28.65
44.48
2.03
112.86
GLC,2000
(mha)
43.32
0.88
9.34
6.83
37.85
12.58
110.80
The statistics show some marked differences in multi cropping, rainfed and desert land use
classifications. The barren class defined in IGBP map is re-expressed into bare soil and
rainfed classes. In some desert areas (e.g. Thal desert), there is an increasing trend of NDVI
during rabi indicating increase in vegetation, which must be related to “rainfed crops
wheat/grams” in the area. The Eastern side of the Indus Basin is found similar to our map.
Urban areas were difficult to demark using NDVI data. MODIS data has the ability to better
discriminate urban areas (Giri and Jenkins, 2005). Therefore the urban area class was taken
from the IGBP urban area mask.
The irrigated classes in GLC2000 map is classified as “intensive irrigated agriculture” and
“irrigated agriculture”. No information can be found on dominant cropping patterns in the
area. The class “pastures/savanna” in the present study is in accordance with GLC2000.
However, GLC2000 has reported more area under the classes falling in general “forest
class” of the two maps.
The LULC map developed by IWMI (Thenkabail et al., 2005) focused more on the Ganges
Basin, as well as the lower and eastern Indus Basin. The land use classification developed
in the present study is in accordance to that map. It facilitated to improve the quality of the
42
produced map for the eastern side of the basin, as in present study, no ground truthing was
carried out in this part. The irrigated area in the canal commands of the lower Indus Basin is
classified as a general class named “irrigated water logged crops (Indus) rice, shrubs”
which cannot properly discriminate different crops grown in the area. The IWMI map
appears to be more generalized in the Indus Basin.
3.4.3.3 Localized studies
Some small scale local studies were also analyzed to check the accuracy of classes falling
in southern and eastern parts of the Indus Basin. Bastiaanssen et al. (2003) prepared a
LULC classification in the Sirsa region of Haryana state, India representing eastern parts of
Indus Basin. They used two Landsat-7 images, one for kharif and other for rabi season, for
the year 2002. Ground truthing was also performed in India for accuracy assessment. These
images were originally in 30m resolution but were re-scaled to 1 km to match with the
current map resolution. Wheat being major rabi crop, is grown at 47% of the area while
42% area is classified as bare soil. In reality, when using 1 km resolution images, wheat and
soil could be classified as mixed. By putting wheat and soil together, the difference between
their maps and the LULC map presented in this paper for the Sirsa area was 2.3% only.
Similarly, kharif crops are grown at 54% of total study area while about 33% area is
classified as desert. Clouds and shadow are also present in the kharif image. The present
study classified desert into rainfed crop classes. Considering kharif crops and desert mixed
for 1 km resolution, both maps has 12.6% difference.
3.4.3.4 Ancillary databases
Ancillary data can be useful to further improve pixel based recognition of land surface
objects. Therefore, agricultural census of this area can provide a good estimate of the
accuracy on mapping agricultural land use classes. The crop area statistics are used to
compute cropped area fractions of crops in different administrative units in the Indus Basin.
The cropped area fraction of wheat, rice, cotton and grams representing irrigated and
rainfed crops is given in Figure 3-7(a),(b),(c) and (d), respectively. The results in Figure
3-7(a) show that the wheat area (comprising major portion of rabi cropping) reported by
MINFAL(2008) is in close agreement with the area estimated in present study. Fairly high
coefficient of determination (R2 = 0.87) between the area fractions reported and estimated is
observed. However, this is also a fact that other rabi crops especially mustard are grown in
the area.
43
b
a
c
d
Figure 3-7 District-wise (a) wheat area fraction (b) rice area fraction (c) cotton area
fraction and (d) gram area fraction for rabi 2006-07, kharif 2007,kharif 2007 and rabi
2006-07,respectively.
The results shown in Figure 3-7(b) &(c) are for rice and cotton areas, respectively. These
are the major kharif crops grown in the region. The R2 values between the reported and
estimated area fractions are 0.82 and 0.63 for rice and cotton crops, respectively. In
different areas large tracts of millet, sorghum, maize, groundnut and tobacco have also been
grown during kharif. The cropped area of these scattered crops is hard to estimate because
of similarity in the growing season. This creates mixed pixels and thus directly affects the
results. This is a drawback of using coarse resolution data. Cropped area fraction of one
rainfed crop i.e. grams is shown in Figure 3-7(d). The results of this class are in good
agreement (R2 = 0.91) with the distribution of grams from the 1 km NDVI data. Despite
these complex cropping patterns, the present study provided an overall general picture that
agrees with the cropping pattern of major crops.
44
The LULC results were also verified against non-agricultural land use information
(MINFAL, 2007). FAO and ICIMOD were also consulted for forests, snow and ice classes.
Total cultivated area calculated in the present study is 13% more than that reported by
MINFAL (2007). This may be due to the fact that rainfed crops are not reported separately
in MINFAL report. The irrigated area reported by FAO (2009) and Habib (2004) is 18.84
mha and 14.80 mha for the year 2005 and 2004, respectively. Present study estimates
irrigated area as15.73 mha. These deviations are certainly related to the accuracy of the
LULC mapping procedure proposed, they should also be ascribed to the field work and
interpretation procedures for such vast regions. The typical variations of 5 to 10% are,
however, acceptable.
The forested area classified in the present study is 12.7% more than MINFAL statistics. The
area reported by present study under food and cash crops (wheat, cotton, rice, sugarcane,
orchards) is similar to that reported by MINFAL (2007). The area of class “snow and ice
permanent” is compared with ICIMOD (Jianchu et al., 2007) for the whole Indus Basin.
The results are 20.9% lower than ICIMOD. In fact, the snow and ice class in present LULC
classification is divided into “snow and ice permanent” and “snow and ice temporary”. This
difference in reported area can be due to the fact that ICIMOD has taken temporary snow as
permanent. Detailed comparison of areal extent of different classification statistics are
shown in Table 3.6.
Table 3.6 Classification statistics of LULC in Indus Basin (Pakistan) compared with
different sources.
S.No
1
2
Total cultivated area
Irrigated area
3
4
5
Rainfed crops
Forests
Food and cash crops
(Wheat, rice,
sugarcane, cotton,
orchard)
Rainfed wheat
Rainfed grams
Fodder
Snow and ice
permanent*
6
7
8
9
*
Classification
Statistics
Present
Study
Area
(mha)
26.89
15.73
MINFAL
Others
Source
Area
(mha)
23.39
17.37
11.16
4.80
4.19
18.23
18.22
1.45
1.33
2.74
4.18
1.24
1.05
2.50
-
Area
(mha)
27.07
18.84
14.80
7.63
FAO,2009
FAO,2009
Habib,2004
Habib,2004+FAO,2009
5.28
ICIMOD
For whole Indus Basin
45
3.5
Conclusions
Spatial and updated land use and land cover information is essential to support basin scale
water management. Due to their lack of details, existing global land cover data sets are
inadequate to fulfill this role. This research has demonstrated that it is possible to discern
LULC classes with managed land use and water use. Temporal profiles of NDVI from
SPOT-Vegetation were used to capture the seasonal phenological variations of 27 classes.
The ground information and expert knowledge of cropping patterns can efficiently be used
to distinguish different crop classes. The overall accuracy attained in this research is 77%
with user’s and producer’s accuracy being 83% and 78%, respectively. The Kappa
coefficient of 0.73 is in moderate agreement range. This is reasonably good and similar to
the accuracy reported by different global scale studies. The vegetation phenology appears to
be strongly coupled to elevation, slope and temperature regimes. The overall crop statistics
of food and cash crops show good agreement with the statistics reported by governmental
organizations. Coefficient of determination for the cropped area fractions reported and
estimated for wheat, rice, cotton and gram crops is 0.87, 0.82, 0.63 and 0.91, respectively.
The mixed pixels especially for rice and cotton crops is a cause of lower R2 for these two
crops as other crops are also grown within the same season. The accuracy is less
satisfactory for agricultural plots with a small size and major focus on food and cash crops.
The comparative advantage is that the classification is more suitable for hydrological, water
resources, agricultural, forestry and environmental studies. The knowledge of dominant
crop rotation schemes can play an essential role in the planning of food security and rural
development. The crop water requirements vary for specific crops. Therefore, crop based
classification helps water policy analysts and managers to formulate better plans with an
improved knowledge base on the extent of the natural resources, certainly when the basin
has an international character. The technique of classifying different water consumers in the
Indus Basin offers a robust method to categorize beneficial and non-beneficial water use,
being one of the key ingredients of water accounting procedures.
46
47
4 Local calibration of remotely sensed rainfall from the
TRMM satellite for different periods and spatial scales in
the Indus Basin
Chapter based on: Cheema, M.J.M. and Bastiaanssen, W.G.M., 2012. Local calibration of
remotely sensed rainfall from the TRMM satellite for different periods and spatial scales in the
Indus Basin. International Journal of Remote Sensing 33, 2603-2627.
4.1
Introduction
Rainfall is an important component of the water cycle and for food production. Its accurate
estimate is therefore vital for hydrological, water accounting and water withdrawal studies
in the catchments and international shared river basins (Verdin and Klaver, 2002; Tobin
and Bennett, 2010). The majority of the world’s cropland depends on rainfall as the major
source of water. Historically, rainfall is measured by rain gauges, which provided
reasonably accurate measurements at one point or field plot only. Estimating rainfall at
basin level at appropriate spatial and temporal scales is still a challenge (Sawunyama and
Hughes, 2008).
Point measurements are not a true representation of areal average rainfall values (Draper et
al., 2009). Rain gauges are subject to various systematic and random errors. Systematic
errors are commonly a result of wind, wetting losses, evaporation from containers,
splashing of water, and blowing or drifting of snow (Nespor and Sevruk, 1999). The
magnitude of these errors can range from 2-50%, 2-10%, 0-4%, 1-2% and 10-50%,
respectively (Rubel and Hantel, 1999). Observational and instrumental errors are major
random errors in rain gauge measurements. These (systematic and random errors) may
result in up to 30% difference between measured and actual rainfall (WMO, 2006).
Moreover, a sparse rain gauge network cannot reflect rainfall variability caused by
topography and orography, and will result in erroneous estimates of areal rainfall
(Andréassian et al., 2001).
There are global rainfall databases with data merged from rain gauges, radar observations,
numerical weather prediction models, and satellite estimates. Some examples are: the
Global Precipitation Climatology Project (GPCP), Global Precipitation Climatology
Centre (GPCC), Climatic Research Unit (CRU) precipitation database, Climate Prediction
Center’s merged Analysis of Precipitation (CMAP), Asian Precipitation Highly Resolved
Observational Data Integration towards the Evaluation of Water Resources
(APHRODITE) and the ERA re-analysis data set. But their spatial scales (e.g. 100
km×100 km, 250 km×250 km, and 250 km×375 km) are too coarse to carry out
hydrological and water management studies. The temporal coverage is also insufficient.
Such global datasets of rainfall are often based on point measurements that are spatially
interpolated with geo-statistical procedures. The geo-statistical interpolations can only be
useful if a very dense rain gauge network is available. In absence of such a network, the
resulted rainfall maps are very general and do not reflect orographical, land surface and
atmospheric processes. The World Metrological Organization (WMO) is connected for
instance with about 40 observatories in the Indus Basin. Such a rain gauge network in the
48
Indus Basin is insufficient for supporting studies and applications. It is also insufficient for
flood warnings, as has been witnessed during July and August 2010.
The need for more accurate spatially distributed rainfall estimates can be met by satellite
based sensors (Huffman et al., 2001). Advancements in remote sensing make it practically
possible to adopt rainfall estimates from satellites as an alternative source of information
(Din et al., 2008) as long as proper ground truthing occurs. Spaceborne sensors provide
continuous monitoring of rainfall both spatially and temporally. Such data are generally
readily available over longer periods and cover large areas (Immerzeel et al., 2009).
The Tropical Rainfall Measuring Mission (TRMM) provides regional coverage at higher
temporal resolution as compared to other gridded products, but at the cost of a low spatial
resolution. The indirect measurement of precipitation by onboard sensors also has
uncertainties (Hong et al., 2006; Hossain et al., 2006). These uncertainties are associated
with lack of rainfall detection as well, false detection and bias (Tobin and Bennett, 2010).
Both temporal errors (± 8 to ±12% per month) and sampling errors (~ 30%) can be
expected in TRMM rainfall estimates (Franchito et al., 2009). Such errors can result in
erroneous applications if applied without calibration (AghaKouchak et al., 2009;
Gebremichael et al., 2010). Therefore, TRMM satellite estimates need area specific
calibration to reduce such errors.
The correction of satellite products are normally carried out by comparing satellite data
with in-situ rainfall measurements. Many efforts have been made in this regard (e.g. Ji and
Stocker, 2003; Chokngamwong et al., 2005; Dinku et al., 2007). These studies were done
at global and regional scales (Thailand and Africa respectively). TRMM product has
shown varying accuracies in different regions and for different adopted methods. Ji and
Stocker (2003) and Chongamwong et al. (2005) observed correlation of 0.56 and 0.86
between satellite and rain gauge measurements, respectively. Dinku et al. (2007) observed
Nash-Sutcliffe efficiency of 0.81 and 25% root mean square error between satellite and
rain gauge data averaged over 2.5° grid boxes. Villarini and Krajewski (2007) tested
accuracy for a single 25 km × 25 km pixel containing 23 rain gauges in Oklahoma and
found 0.55 correlation between the satellite and rain gauges values.
The mismatch between spatial rainfall estimates from satellites (Grimes et al., 1999) and
point measurements by rain gauges unavoidably led to distrust in both data sets (Xie and
Arkin, 1996; Ciach et al., 2000; Omotosho and Oluwafemi, 2009). All these studies
reported area specific bias in the satellite estimates. Therefore, the uncertainty in TRMMbased rainfall in the transboundary Indus Basin needs to be inspected carefully prior to its
use in hydrological and agricultural water management studies. Rainfall controls the
renewable water resources, affects safe withdrawals for irrigation and provides moisture
directly to thirsty crops during the rainfall seasons. This paper is novel because it
integrates scarce measurements from rain gauges with time series of TRMM data to
generate advanced information systems that is otherwise not available in vast irrigated
basins such as the Indus.
The aim of this paper is to develop a calibration protocol for TRMM rainfall data at
different spatial and temporal scales. Two research questions were addressed to investigate
the adoptability of satellite derived rainfall:
49
12-
4.2
4.2.1
What is the deviation between satellite rainfall estimates and rain gauge
measurements at different temporal and spatial scales?
How could geographical influence on satellite-based rainfall be described and
correct the TRMM data?
Materials and methods
Study area
The study area is Indus Basin, with large variation in climate and topography. The Indus
Basin lies between 24° 38′ to 37° 03′ N latitude and 66° 18′ to 82° 28′ E longitude. It
covers Pakistan, India, China (Tibet) and Afghanistan (see Figure 4-1). The basin has a
surface area of 116.2 million ha (mha), with elevations ranging from 0–8600 m above
mean sea level (a.m.s.l). The basin is bordered by the mighty Himalayas in the north-east;
the Karakoram and the Hindukush mountain ranges in the north; the Sulaiman and the
Kirther ranges in the west and the Arabian Sea in the south.
Figure 4-1 Location of the Indus Basin (marked in green) showing elevation and location
of rain gauge stations used in this study.
50
4.2.2
Rainfall systems over the Indus basin
The major part of the basin is dry and located in arid to semi arid climatic zones. The
annual rainfall is higher in the north and the north eastern tracts (>1200 mm yr-1) and
lower toward the middle and the southeast (<300 mm yr-1). There are two sources of
rainfall in the Indus Basin: the Monsoon and the Western Disturbances. The former takes
place from June to September and the latter from December to March (Lang and Barros,
2004; Bookhagen and Burbank, 2006).
The Monsoon season is caused by moist air currents from the Arabian Sea and Bay of
Bengal. Monsoon rainfall occurs mainly due to heat difference of the land and sea. The
heat difference creates pressure gradient causing wind fluxes from ocean to land
(Muslehuddin et al., 2005). The moist air from the ocean moves towards the north passing
through the hot basin plain (Houze et al., 2007). Most of the rainfall in summer is due to
this system resulting in intensive convective rainfall (Singh and Kumar, 1997). It is
intensive in the months of June, July and August.
The weather systems responsible for winter rainfall are mid latitude Western Disturbances
(Thayyen and Gergan, 2009). They originate over the Caspian Sea and moves from the
west to east (Singh and Kumar, 1997).These are formed due to large scale interaction
between the mid latitude and the tropical air masses. The interaction process results in the
formation of westerly troposphere synoptic scale waves. These disturbances results in
stratiform rainfall. The orographic effect may cause intensification, resulting in extensive
cloudiness, heavy precipitation and strong winds. However, sometimes their movement
slows down causing local heavy snowfall over the hilly areas (Dimri, 2006).
4.2.3
TRMM retrieval algorithm
The TRMM satellite has both passive and active sensors on board, which measured
rainfall since 1997 (Kummerow et al., 2000). It is a low-latitude satellite that includes the
Precipitation Radar (PR), along with a multi-channel passive TRMM Microwave Imager
(TMI). TMI compliments the PR by providing total hydrometeor (liquid and ice) content
within precipitation systems. A Visible-Infrared Radiometer (VIRS), a Lightning Imaging
Sensor (LIS), and Clouds and Earth’s Radiant Energy System (CERES) are also onboard
TRMM. The parameters of the sensors on board TRMM are given in Table 4.1.
Table 4.1 System parameters of sensors on board TRMM satellite
Item
Frequency
(GHz)/
Wavelength
(nm)
Swath (km)
Foot print/
Res.(km)
Product
PR
13.8
(GHz)
TMI
10.7,19.3,21.3,
37,85.5 (GHz)
VIRS
0.63,1.61,3.75,
10.8,12(nm)
LIS
0.778
(nm)
CERES
0.3–50 (nm)
250
5 km
880
6–50 km
830
2.5 km
600
4 km
±82°
10 km
Radiance
Rr
Lightnin
g
Radiance
fluxes
Rr§
Rr
Rs¶
Rs
§
rain rate (mm hr-1) ¶ rain structure
51
PR is the first and to date only rain radar in space. It can detect minimum rainfall rate upto
0.7 mm hr-1(Kawanishi et al., 2000). The intensity of the back scattering is converted to
the radar reflectivity factor Z using the Rayleigh scattering approximation (Probert-Jones,
1962):
Z = ∑i D i
6
4.1
6
–3
where Z is radar reflectivity factor in mm m . Diis the diameter of raindrop “i” for over
all drops per unit volume. The radar reflectivity factor can be estimated from the sixth
moment of the drop size distribution:
Z = C ∫ D 6 N ( D )dD = 720 N ο Λ7 = 296 Rr
1.47
4.2
-1
where Rr is the rain rate in mm hr . The constant C, which depends on the refractive index,
differs between water and ice particles. For water, C has a value of 1.0, while C is 0.93 for
ice (Prabhakara et al., 2002). Drop size distribution N(D) in a given size range dD is
proposed by Marshall and Palmer (1948). No=8×103 mm-1 m-3 and Λ = 4.1 Rr-0.21 mm-1. Z–
Rr power laws depend on climatic and actual meteorological circumstances (e.g.,
stratiform vs. convective precipitation). Therefore, different Z–Rr relations may be in use
based on a particular situation (Schumacher and Houze, 2003). An appropriate Z–Rr
relation is used to provide radar reflectivity product 1C21 (Iguchi et al., 2000).
TMI is passive microwave radiometer. It provides information on the integrated column
precipitation content, cloud liquid water, cloud ice, rain intensity, and rainfall types (e.g.,
stratiform or convective) (Kummerow et al., 1998). TMI measures microwave radiances in
terms of brightness temperature emitted partly at the earth’s surface and partly from
atmospheric constituents. To relate surface rain rate to a linear combination of the
observed brightness temperature, multiple linear regression methods are applied
(Kummerow et al., 1991). However, this linear relationship is limited at higher rainfall
rates. Therefore, Kummerow et al. (2001) proposed a physical technique to account for
non linearity at higher rainfall rates based on work of Grody (1991) and Ferraro and Marks
(1995). Rain rates were observed using a Scattering Index (SI). SI is the difference in 85
GHz brightness temperature estimated from data at the lower frequency (19, 22 GHz)
channels and the observed one. The rainfall rate estimation equation is
Rr = 0.00513 × (SI )
1.9468
4.3
-1
where Rr is the rainfall rate in mm hr , and SI is defined as:
(
)
SI = 451.9 − 0.44Tb19v − 1.775Tb22v + 0.00575Tb222v − Tb85v
4.4
where Tb denotes the brightness temperature at appropriate channels. Tb observed by
satellite can be expressed by the equation suggested by Fujii and Koike (2001):
52
Tb = Tbs exp(− τ c ) exp(− τ r ) + (1 − ω c ){1 − exp(− τ c )}Tc exp(− τ r )
+ ∫ (1 − ω r ( L) )(1 − exp(− τ r ( L) ))Tr ( L)dL
4.5
where Tbs is land surface radiation, τ, T and ω are optical thickness, physical temperature
and single scattering albedo, respectively. c and r denotes vegetation and precipitation
respectively. L is the observational path length from target to a satellite sensor. Equation
4.5 takes care of attenuation from different layers. The surface rainfall thus obtained from
the TMI brightness temperature is provided as product 2A12. The rain profile from the
two sensors (PR and TMI) are combined together to provide best estimate of surface
rainfall rate and vertical structure (TRMM product 2B31).
VIRS is a cross tracking scanning radiometer that senses radiation coming up from the
earth in five spectral regions, ranging from visible to infrared. The radiation intensity thus
estimated can be used to determine the reflectance (visible and near infrared) or brightness
temperature (infrared) of the scene (Kummerow et al., 1998). During clear sky conditions,
the temperature corresponds to that of earth’s surface while in clouds; the temperature will
represent the cloud tops. It is assumed that the colder and bright clouds are associated with
heavier rainfall while warmer and less bright clouds associated with light or no rain. The
VIRS links this cloud top temperature and structure information with microwave
observations (Barnes et al., 2000). It thus provides the cloud context within which the
passive and active microwave observations are made. Data from the VIRS instrument is
available as product 1B01 and is used in rain estimation in combination with the TMI and
PR merged product 2B31.
The LIS and CERES are the additional instruments on board TRMM. The LIS observes
distribution and variability of lightning over the earth. It detect, locate and measure the
intensity of lightning activity by detecting sudden changes in the clouds brightness as they
illuminated by lightning discharges (Christian et al., 1999). Lightning is coupled to storm
convective dynamics and therefore, can be correlated to the global rates, amounts, and
distribution of convective precipitation (Christian et al., 2000).It is used with combination
of PR, TMI and VIRS data to investigate the correlation of the global incidence of
lightning with rainfall and other storm properties (JAXA, 2006). While CERES is a visible
and infrared sensor that measures the total radiant energy balances. It was built to provide
radiative fluxes at top of atmosphere and earth’s surface. In addition, it was providing
cloud properties estimates and their effect on earth’s radiation balance (Wielicki and
Barkstrom, 1997).
Three hourly 25 km × 25 km estimates of rainfall (product 3B42) are then produced using
combined microwave estimates of rain structure with VIRS cloud information. Global
estimates are made by adjusting the GOES Precipitation Index (GPI) to these TRMM
estimates. The information from daily accumulated estimates combined with the monthly
average Special Sensor Microwave/Imager estimates (product 3A46), and the monthly
accumulated Climate Assessment and Monitoring System (CAMS) or GPCC rain gauge
analysis (product 3A45) is used to produce monthly estimates of global rainfall. The final
product is TRMM 3B43(V6) that provides monthly estimates of rainfall at 25 km × 25 km
resolution. More details on the algorithm are available through Huffman et al. (2007).
53
4.2.4
Data availability
The satellite rainfall data was obtained from the TRMM, a joint project between National
Aeronautics and Space Administration (NASA) and the Japanese space agency (JAXA). A
range of orbital and gridded TRMM products are available through NASA’s website
(http://neo.sci.gsfc.nasa.gov/ Search.html? group=39). Monthly rainfall product TRMM
3B43 (V6) for the year 2007 was obtained from NASA’s official distribution site
(Huffman et al., 2007).
The TRMM rainfall estimates contain sampling as well as retrieval or instrumental errors.
The sampling error is caused due to discrete sampling frequency and the sensors’ areal
coverage (Condom et al., 2011). The satellite samples the region only at intermittent time
intervals (92.5 min) that can lead to miss short duration rainfall events. The TRMM 3B42
daily product estimates daily rainfall with sampling frequency of three hours. The PR
revisit frequency is once in every three days for points near the equator (Haddad and Park,
2010). Such a temporal error can range from ± 8 to ±12% per month relative to mean
rainfall (Franchito et al., 2009). The instrumental errors in the rainfall retrieval include
attenuation factor, drop size distribution, density of solid particles etc.
Rain backscatters and backscatters near the surface can interfere resulting in errors in
rainfall estimation by TRMM PR. The emitted radiances from land surfaces can affect the
total radiance from clouds and thereby induce an error in rainfall estimates (Kästner,
2007). Intense rainfall events may cause attenuation of the radar beam by scattering of
radio frequency radiations transmitted and received. The lower rainfall rates are difficult to
detect due to PR detection limitation of 18 dB. Thus the combination of cloud and drizzle
would not be detected. The non uniform rain distribution with the radar resolution cell
may cause error when attenuation is severe (Adeyewa and Nakamura, 2003). The drop
size distribution can cause bias up to 27% under convective rainfall conditions (Sato et al.,
1996).
The TMI uses scattering index based on brightness temperature (Tb) for rainfall estimation.
Tb indicates significant emission signal due to liquid hydrometeors but little or no signal
due to frozen hydrometeors aloft. TMI can miss light and heavy rainfall due to scale and
nature of rainfall (warm rain) (Berg et al., 2006). TMI bias is also dependent on the time of
day (solar hours). Biswas et al.(2010) has observed such biases in TMI retrieval over both
ocean and land.
The major part of the Indus Basin lies within the administrative boundaries of Pakistan
(53% of total). Therefore most of the rainfall station data was obtained from the Pakistan
Meteorological Department (PMD). PMD has established a rain gauge network throughout
the country (see Figure 4-1). This rain gauge network is sparse as compared to the vastness
of the basin (< 4 gauges/10,000 km2). However, this is the only possible source of rainfall
data measured on the ground. The daily rainfall data from 65 stations for the year 2007
was obtained from PMD. The data quality have been checked and verified by PMD. The
consistency between the daily and monthly values has been checked and the extremes
have been validated by comparing the data of one station with the other closer stations.
Although the rain gauges have some errors as discussed in Section 1 yet these are
normally considered correct (Vila et al., 2009). Here it is assumed that the rain gauges are
true representation of the point rainfall in the basin.
54
Weather station data for India came from the National Oceanic and Atmospheric
Administration (NOAA) National Climatic Data Center (NCDC). The NCDC collects in
situ rainfall data from real time reporting stations worldwide in agreement with WMO
regulations (ftp://ftp.ncdc.noaa.gov/pub/data/gsod/). Only six stations have data for the
complete year of study (2007). So for the whole basin, daily rainfall data from 71 stations
were obtained.
A detailed Land Use and Land Cover (LULC) map of the Indus Basin developed by
Cheema and Bastiaanssen (2010) was used to facilitate the interpretation of spatial rainfall
data. To study the topographic effects on rainfall distribution, a 90 m resolution Digital
Elevation Model (DEM) was obtained from the Shuttle Radar Topographic Mission
(SRTM) database (Jarvis et al., 2008).
4.2.5
Methodology
The limitation of having only one rain gauge per TRMM satellite pixel of 25 km x 25 km
creates considerable limitation of this calibration study. However, such uncertainty can be
reduced considerably by averaging rain gauge measurements over time (Harmsen et al.,
2008). Therefore, along with monthly comparisons, the monthly rainfall obtained from
rain gauge stations and TRMM 3B43 (V6) was accumulated into annual, seasonal (based
on cropping patterns) and quarterly (based on weather patterns) periods.
Two growing seasons (rabi and kharif) were selected. Accumulated rainfall over the
months November, December, January, February, March, and April represent the rabi
season. The rainfall accumulated over May, June, July, August, September, and October
represent the kharif season. Four quarters were selected: December, January, and February
(DJF); March, April, and May (MAM); June, July, and August (JJA); September, October,
and November (SON). The quarters represent dry (winter), pre-monsoon (spring), wet
(monsoon) and post-monsoon (autumn), respectively.
In this study, we have adopted two techniques to calibrate TRMM rainfall:
1-
Regression analysis (RA).
2-
Geographical differential analysis (GDA).
For both techniques 85% of the total rain gauge measurements (61 stations) were exposed
to a calibration procedure between satellite and rain gauge measured rainfall. The
remaining 15% (10 stations) were used for validation purpose only.
TRMM rainfall estimates have proven relationship with rain gauge measurements
before(e.g. Din et al., 2008; Omotosho and Oluwafemi, 2009). Such empirical
relationships can be derived from regression equations for annual, seasonal, quarterly and
monthly periods. Regression analysis (technique – 1) were used on data from 61 gauges to
derive equation (4.6) (below) to describe the relationships for different periods:
2
Rcal(1) = a.RTRMM
+ b.RTRMM
4.6
where Rcal(1)is the resultant calibrated TRMM rainfall and RTRMM is the spatial rainfall
obtained from TRMM satellite in LT-1(mm year-1, mm season-1, mm quarter-1 and mm
55
month-1 for annual, seasonal, quarterly and monthly periods, respectively). “a” and “b” are
regression coefficients applicable for the entire Indus Basin. Second order polynomial
equations were applied to get best fit between the TRMM pixel estimates and rain gauge
measurements. These equations provided the required “a” and “b” coefficients for annual,
seasonal, quarterly and monthly periods. Note that the line fits the origin and that a
parabolic shape is assumed. A basic assumption is that TRMM describes the spatial
variations correctly – also in heterogeneous topography and geography, but measures the
quantity of rain with random deviation around an average. Equation (4.6) was used to test
Rcal(1)against rain gauge measurements. New maps of rainfall Rcal(1) were then developed
for different periods.
For technique–2 (GDA method), a map was prepared with differences between TRMM
and gauge values at specific locations. The same data set as with techique–1 was used.
Initially, the difference between the satellite pixel and measurement for each rain gauge
was calculated using equation (4.7). An Inverse Distance Weighted (IDW) interpolation
technique described by Nalder and Wein (1998) was applied for spatial interpolation of
rainfall difference (∆R(x,y)) between points (coordinates (x,y)) for which TRMM and rain
gauge measurements were available. This resulted in a spatially interpolated difference
map (∆R(x,y)ip). The IDW technique was adopted because of its robustness and simplicity
(Brouder et al., 2005; Ahrens, 2006; Babak and Deutsch, 2009). The IDW is the spatially
weighted average of the sample values within the search neighbourhood (Franke, 1982).
The exponent value and the optimal number of neighboring stations used in the
interpolation were selected following the recommendations given by Babak and Deutsch
(2009). Finally, values from the difference maps were subtracted from values from the
satellite rainfall maps to get real estimate of satellite rainfall (see equation 4.9). This
procedure was adopted for all periods and months. Note that geo-statistics are used to
extrapolate the differences in rainfall and not the absolute values. The values for the
differences are small compared to the absolute values and the impact of geo-statistics on
the final calibrated rainfall product is thus limited. The basic assumption in this approach
is that TRMM requires a location specific correction and that the deviations are nonrandom, and relate to a given geo-spatial context.
∆R( x,y ) = RTRMM ( x,y ) − RRGS ( x,y )
4.7
∆R( x,y )ip = ∆R( x,y )
4.8
Rcal(2) = RTRMM − ∆R( x,y )ip
4.9
where ∆R(x,y) is the rainfall difference between satellite and rain gauge data at a given
point, RTRMM(x, y) and RRGS(x, y) is satellite measured rainfall and rain gauge station values at
a specific location (coordinates (x,y)), respectively. ∆R(x,y)ip is the spatially interpolated
difference map. Rcal(2) is the calibrated rainfall after correction and RTRMM is the spatial
rainfall estimate obtained from TRMM satellite, all in dimensions LT-1.
Nash-Sutcliffe Efficiency (NSE) and Standard Error of Estimates (SEE) were used to
indicate performance. The remaining 15% (ten stations) of rain gauge measurements were
used. Nash and Sutcliffe (1970) described a method to determine the relative magnitude of
the noise in estimated data compared to the measured data. NSE indicates how well the
56
plot of observed versus estimated data fits the 1:1 line (Moriasi et al., 2007). The NSE was
computed as:
n
NSE = 1 −
∑ (R
− Rcal(x , y) ) 2
∑ (R
− RRGS(x , y ) ) 2
RGS(x , y)
i =1
n
i =1
RGS(x , y)
4.10
where RRGS( x , y ) is the observed rainfall from rain gauge stations, Rcal( x , y ) is the calibrated
rainfall estimate, RRGS( x , y ) is mean observed rainfall and n is the total number of
observations. NSE ranges between -∞and 1, with NSE = 1 being the optimal value. Values
between 0.0 and 1.0 are generally viewed as acceptable, whereas values <0.0 indicate that
the mean observed value is more accurate than the estimated value, which is unacceptable.
While “0” specifies that the observed mean is as good as the estimated.
SEE is a measure of the deviation of the predicted value from the measured value
(Gravetter and Wallnau, 2006). It gives the deviation of the predicted values from the
regression line. The SEE was computed using equation (4.11).
n
SEE =
∑ (R
i =1
cal ( x , y )
′ ( x,y ) )2
− RRGS
n−2
4.11
′ ( x , y ) is the measured rainfall value
where Rcal( x , y ) is the calibrated rainfall estimate and RRGS
representing the regression line. The best among the two alternative rainfall calibration
methods was decided on basis of these validation tests.
4.3
4.3.1
Results and discussion
Technique -1
Figure 4-2(a – c) show the results of fitting polynomial functions for different periods
using calibration technique–1. Figure 4-2 shows that for all periods a relationship exists
between rain gauge and satellite derived rainfall. The curves show that the satellite is overestimating the total rainfall. This leads to underestimates of rainfall at the higher end of the
rainfall spectrum. Chokngamwong and Chiu (2006) and Islam and Uyeda (2008)
experienced similar trends in Thailand and Bangladesh, respectively. Chokngamwong and
Chiu (2006) found that the satellite overestimated at the low and mid rainfall rate range
(<400 mm month-1), but underestimated at the higher end (>400 mm month-1).
These results were further strengthened by findings of Islam and Uyeda (2008). They
concluded that the TRMM has limitations in accurate rainfall detection at low or high
rainfall rates. This finding pertains to the areas with tropical climate having monsoon
57
rainfall. These findings seem to be in agreement with our results for the Indus Basin. The
shape of the regression line suggests that TRMM overestimates rainfall in the 400 to 1000
mm range, and performs better at higher rainfall amounts.
(a)
(b)
(c)
Figure 4-2 Fitting of rain gauge and TRMM accumulated rainfall of 61 stations for (a)
annual, (b) rabi and (c) kharif season.
The set of regression equation (see equation4.6) was also applied to different rainfall
periods, which resulted in corrected rainfall maps for different periods as shown in Figure
4-4(a)(i) annual and seasonal((ii) rabi and (iii) kharif) periods. This correction is purely
based on RA. The total annual mean rainfall in the basin is 447 mm yr-1(or 519 km3yr-1).
The 5% lowest value is 87 mm yr-1and the 5% highest value is 1514 mm yr-1.
58
4.3.2
Technique -2
Spatial difference structure between satellite and rain gauge datasets was investigated by
preparing the “geographical differential maps” using equation (4.7). A systematic
deviation pattern between rain gauge and satellite data is apparent (see Figure 4-3).
Logarithmic correlation between the elevation (25 km resampled DEM) and annual
rainfall difference (∆R365) is found for the Indus Basin. It indicates that geography has an
impact on rainfall differences. This bias gets larger from lower to higher elevations. The
biases in the satellite estimates range from ±45 mm yr-1 near the coastal areas to ±200 mm
yr-1 in the Himalaya foothills. Similar spatial bias structure in daily rainfall was observed
by Oke et al. (2009) over Australia. They compared 3B42 product with gridded rain gauge
measurements for year 2001-2007. The different Z-Rr relationships and empirical
coefficients explain this result. The variance in TRMM estimates influenced by
topographic variables was also reported by Yin et al.(2008). They estimated biases in
TRMM rainfall over Tibetan Plateau.
Figure 4-3 Relationship between ∆R365 and elevation in the Indus basin. ∆R365 is the
difference between rain gauge measurements and TRMM pixel estimates at annual scale.
The TRMM is known to underestimate in areas with localized, short lived and intense
convection rainfall. Normally, convection rain areas are small (Prabhakara et al., 2002).
Convection storms are more prominent during the monsoon period and visible in the
kharif seasonal period. They mostly occur in the north-west corner of the basin. This
region is at the junction of the Himalayas and Hindu Kush mountain ranges, see also
Barros et al.(2004) and Houze et al. (2007). They studied the orographic effects and
monsoon convections in the Himalayas, respectively. Some coastal areas in the south have
also been identified where satellite underestimates rainfall. A plausible reason is the
59
intermittent measurement system of TRMM. The interval is three hours and rainfall events
can occur in this small time frame that is not detected by the orbiting satellite. The errors
in this region occur during the monsoon period (kharif season). These areas are most prone
to cyclones and intense rainfall during monsoon (Jilani et al., 2008). The larger deviations
in TRMM data thus occur in the areas which have more intense convection and localized
rainfall.
Another possible reason of underestimation could be instrumental errors. An intense
precipitation may cause scattering of radio frequency radiations transmitted and received
by the TRMM PR. The radar beam is attenuated, causing underestimation of precipitation
intensity or even disappearance of the rain cells behind very strong cells (Bringi and
Chandrasekar, 2001). Very low rainfall rates are difficult to measure by PR due to
detection limitation. Rainfall rates lower than 0.7 mm hr-1 cannot be observed due to a
reflectivity detection limit of 18 dB (Anders et al., 2006). So the total annual precipitation
will not account for these events resulting in underestimation.
The TRMM overestimates over mountain ranges in the north and north-eastern parts of the
basin. This effect extended to irrigated plains near foot hills, where it showed maximum
deviation. Rain backscatters and backscatters near the surface can interfere over
mountainous regions (Kozu et al., 2001). The gradient and slopes in such regions can
attenuate radar beam reflectivity (Porcù et al., 2003) thus resulting in erroneous estimates.
Moreover, the rainfall distribution pattern near these mountains is also non uniform (e.g.
Singh et al., 1995; Bhatt et al., 2005; Anders et al., 2006; Bookhagen and Burbank, 2006)
and any comparison with rain gauges is intrinsically complex. The sparse rain gauge
network and high wind conditions cause the rain gauges to underestimate (Franchito et al.,
2009). These systematic errors can cause 30% or more underestimation of rainfall than
actual (WMO, 2006).
There is also an aerosol impact on rainfall estimations. The dust particles due to pollution
in the atmosphere cause accumulation of small droplets. Due to lower temperature, these
droplets freeze, resulting in heavy fog. These droplets do not contribute to rainfall (Khain
et al., 2007). This fog mostly occurs in the middle Indus Basin in the DJF quarter and is
more severe in the months of December and January.
The spatial difference structure obtained from geographical differential maps was
combined with the satellite derived rainfall estimates applying equation(4.9). The resultant
rainfall maps (after applying this correction) are shown in Figure 4-4(b)(i),(ii) and (iii) for
annual, rabi and kharif periods, respectively.
60
(b)(i)
(a)(i)
(a)(ii)
(b)(ii)
(a)(iii)
(b)(iii)
-1
Figure 4-4 Calibrated rainfall (LT ) at (i) annual and seasonal ((ii) rabi and (iii) kharif
maps) scale using (a) technique–1 and (b) technique–2.
The total annual mean rainfall in the basin calculated using technique–2 was 383 mm yr1
(or 445 km3yr-1). The lowest 5% value was 46 mm yr-1and the highest 5% value was 1298
61
mm yr-1. The overall pattern of rainfall observed over the Indus Basin was similar for both
techniques, but the mean difference in real rainfall was 64 mm yr-1.For an area of 1.162
million km2 this adds up to a volumetric difference of 74 km3yr-1, being three times the
outflow of the Indus into the Arabian Sea. It is thus imperative to establish accurate
calibration procedures.
It is evident from the maps of both techniques that most of the rainfall is concentrated
along the arc of the mountain ranges stretching from the north-east to the west of the
basin. Changes in elevation, slope and aspect of these mountain ranges cause unreliable
estimations of spatial rainfall (Buytaert et al., 2006). The foothills and windward sides of
the mountains receive more rainfall due to upward drift of moist air (Anders et al., 2006).
The leeward side receives lower rainfall (<200 mm yr-1) due to rain shadow effects. Due to
less rainfall, “pastures deciduous alpine” and “savanna deciduous” are the dominant land
covers at leeward side. Singh et al. (1995) and Archer and Fowler (2004) observed similar
rainfall patterns in the windward and leeward sides of Himalayas ranges, respectively.
4.3.3
Validation
A comparison of the two techniques was carried out on the rainfall accumulated over
annual, seasonal, quarterly, and monthly periods. The statistics are shown in Table 4.2.
Table 4.2 Nash-Sutcliffe efficiency and standard error of estimate for technique–1 and
technique–2 applied for different periods.
Period
Annual
Seasonal
Quarterly
Monthly
62
Nash-Sutcliffe efficiency
Rabi
Kharif
MAM
JJA
SON
DJF
Jan.
Feb.
Mar.
Apr.
May
Jun.
Jul.
Aug.
Sep.
Oct.
Nov.
Dec.
Technique–1
0.81
0.86
0.81
0.78
0.80
0.63
0.58
0.11
0.71
0.78
0.43
0.61
0.09
0.65
0.68
0.52
-0.95
0.69
0.06
Technique–2
0.86
0.84
0.91
0.94
0.82
0.91
0.25
0.06
0.66
0.85
0.67
0.95
0.32
0.87
0.68
0.84
-0.33
0.76
0.45
Standard error of estimate
(mm)
Technique–1 Technique–2
121.2
79.8
58.7
47.7
81.8
54.7
54.2
37.7
58.9
58.7
15.1
12.6
37.7
44.7
7.0
17.6
26.4
32.4
40.1
31.4
5.7
4.7
12.0
2.5
34.7
40.9
50.5
30.5
28.3
29.3
14.2
12.2
0.9
1.1
1.7
3.9
6.1
7.3
Table 4.2 shows that technique–2 performed better than technique–1. At annual time scale
technique–2 has a higher NSE (0.86) and lower SEE (79.8 mm) as compared to technique–
1. The same trend was observed for seasonal periods. In the rabi season, the NSE of
technique–1 is 0.86, which is slightly higher than technique–2(0.84). However, the SEE
of technique–1 is also higher (58.7 mm) than technique–2(47.7 mm). It is thus important
to use more than one evaluation criterion. A similar trend is also visible for other temporal
periods except for one quarter (DJF) and 2 months (January and February).
TRMM rainfall estimates proofed reliable in some regions and seasons but not in others.
In general, TRMM is reliable in temperate rainfall areas. Rainfall estimates for seasonal
and annual periods are generally better than for shorter periods. The accuracy of estimates
in the kharif season in particular was quite good. Individual quarters showed variable
accuracy. For example, pre-monsoon (MAM) and post-monsoon (SON) quarters with
temperate rainfall showed good agreement. Accurate results were also obtained in the wet
monsoon (JJA) quarter. But the dry winter (DJF) quarter produced less accurate results.
The accuracy of these quarters influenced the accuracy of the seasonal periods. In general,
estimates were reliable where mean rainfall was neither very high nor very low.
Using the results of table 2, it can be concluded that technique–2 performed better. Site
specific calibration considering the geography is thus unavoidable for TRMM data.
Therefore, technique–2 was chosen to analyze the accuracy attainable at different temporal
and spatial scales. It is notable that the maximum efficiency that was attained by
technique–2 is not better than 91%.i.e for the kharif season.
4.3.4
Temporal and spatial deviation analysis
The magnitude of temporal and spatial deviation between the rainfall measurements and
TRMM estimates (both un-calibrated and calibrated) was estimated by calculating average
percentage deviation as suggested by Valderrama and Alvarez (2005). The TRMM
estimates calibrated using GDA technique was used in this analysis. The deviations were
tested for accumulated rain over different periods and different spatial scales. Percentage
deviations were averaged over seasonal, quarterly, and monthly periods. It provided
understanding of how the temporal and spatial scales affect the overall deviations between
the two datasets.
An area of 25 km × 25 km was first considered, i.e. one single TRMM pixel. There were
ten individual TRMM pixels (n=10) with rain gauges not being used for the calibration
process. Each pixel contains one rain gauge. The average percent deviations, before and
after calibration, for different periods are shown in Figure 4-5. The standard deviation
(SD) in the average percent deviation is also provided. The figure shows that the average
deviation is large for monthly time scales even after calibration. The average percentage
deviation for annual and seasonal periods for a 25 km × 25 km area is in the range of
10.9% and 14.3%, respectively. While the quarterly and monthly periods showed
variations up to 24.2% and 34.9%, respectively. For pre-calibrated data, the deviations
were in the range of 48, 51, 79 and 88 % for annual, seasonal, quarterly and monthly
periods, respectively. Figure 4-5 also shows that calibration of TRMM 3B43 product is
necessary, as it has considerably reduced the deviation.
63
Figure 4-5 Average percentage deviation between measured and TRMM rainfall (pre and
post-calibration) for different periods and ten different TRMM pixels. Each point
represents the average of 10 areas encompassing 625 km2. The GDA was applied. The
vertical lines represent standard deviation.
Figure 4-5 shows that uncertainty remains even after calibration of satellite estimates
against rain gauge measurements. This discrepancy could be due to the fact that only one
rain gauge is available in a satellite pixel. This single rain gauge was used to calibrate and
determine the satellite estimated error over a pixel. Capturing spatial variability of rainfall
by a single rain gauge in a 25 km×25 km area is an issue of concern. The existence of both
rain and non-rain areas within the same pixel and random errors in rain gauge
measurements may cause inaccuracy (Gebremichael and Krajewski, 2004). Harmsen, et al.
(2008) concluded that within a satellite pixel rainfall can vary considerably. They found
37% variability in rainfall between two randomly selected rain gauges within a 4 km×4 km
GOES-12 satellite pixel. This implies that the rainfall measured at one station can be 18%
different from the average value of all stations located in the same pixel of 4 km×4 km.
The sub pixel variability can be up to 60% at 2 km×2 km radar pixel (Anagnostou et al.,
1999). Krajewski et al. (2003) found 20% variation in rainfall within a difference of 1 km,
showing the extent of rainfall variability at smaller scale in Guam. Increased number of
validation stations reduces the deviation from the average value (Lebel and Barbe, 1997).
Since the required density of rain gauges is not present in the Indus Basin, it is difficult to
further lower the deviations between large TRMM pixels and single rain gauges. This
discrepancy of scale can be solved if TRMM data is downscaled (Immerzeel et al., 2009),
which requires a different type of research.
To estimate deviation at different spatial scales, the ten validation pixels under
consideration were re-arranged according to their location and elevation. This sorting
helped to make location and elevation specific pixel combinations. Initially, average
percentage deviation was calculated for single (625 km2) pixel. Then two pixels were
combined to get the average deviation for the 1250 km2. Afterwards, combination of three,
64
four and five pixels were made to get 1875 km2, 2500 km2 and 3125 km2 spatial coverage,
respectively. The average deviation in measured and calculated rainfall at different spatial
scales is shown in Figure 4-6.
Figure 4-6 Average percentage deviation in rain gauge measured and calibrated TRMM
rainfall computed at different spatial and temporal scales. The GDA method has been
applied.
Figure 4-6 shows that the average deviation is an inverse function of the spatial area. The
average deviation between single pixels is higher for all time scales (10.9%, 14.3%, 24.2%
and 34.9% for annual, seasonal, quarterly and monthly periods, respectively) than for a
few clumped pixels. The deviations get smaller as the spatial scale is increased by
combining pixels. At 3125 km2 for example the deviations are reduced to 6.1%, 6.1%,
9.1% and 15.4% for annual, seasonal, quarterly and monthly periods, respectively. The
error structure was according to the expectations. Similar error structures was also
reported by Hossain and Huffman (2008) at Oklahoma, US. The information on temporal
and spatial scale TRMM rainfall variability could be vital for the hydrologists, carrying
out water accounting at basin scales especially with complex geographies.
4.3.5
Agricultural landuse – rainfall relationship
The Indus Basin is under water stress in response to meet growing food and fiber
requirements. Therefore, investigating relationship between rainfall and agricultural land
uses is important to improve overall water productivity. Such relationship was explored by
extracting rainfall information from the GDA-calibrated rainfall maps at annual and
seasonal periods for different agricultural land uses (irrigated and rainfed). Figure 4-7
shows the annual and seasonal accumulated rainfall average for each agricultural land use.
The average elevation of these land uses is also shown to depict the variability of rainfall
with changing geography. The contribution of rainfall in rabi and kharif seasons provides
the understanding of inter seasonal variability.
65
The results show that the rainfall has intra-land use and inter-season variability. The
rainfall varies among different land use classes with an increasing trend with elevation
from south to north. The irrigated land uses in the Indus basin lies at elevations ranging
between 100 m to 225 m. Mean annual rainfall in the plain varies from 292 mm at 104 m
(irrigated rice, fodder rotation) to 523 mm at 223 m (irrigated mixed cotton, wheat
rotation/sugarcane). The contribution of kharif season rainfall to annual in irrigated land
uses is 70 – 80 %. Similar percentage contribution (>60%) is observed during kharif
season for all agricultural land uses.
Rainfed croplands are located at an average elevation of 184 m to 841 m. “Rainfed crops
and woods” land use located at average elevation of 841 m receives maximum rainfall of
560 mm yr-1with 60% contribution during kharif. Due to elevation and slope (~ 8°), most
of the rainfall received by this land use is provided as blue water resource in the form of
surface runoff or sub surface drainage.
Figure 4-7 Annual and seasonal rainfall distribution for different agricultural land use
classes in the Indus Basin for the year 2007. Average elevation (m, a.m.s.l) of each land
use is also provided for interpreting influence of geography.
The weighted average rainfall for all agricultural land uses is 426 mm yr-1, while rabi and
kharif seasonal average rainfall is 130 and 293 mm, respectively. The crops with high
water use are grown during the kharif season as compared to rabi that receives meager
rainfall. The crops requiring less amount of water are normally grown during rabi.
66
However, the available amount of water is insufficient to meet the actual crop water
requirement. The average actual evapotranspiration (ET) estimated by Bastiaanssen et al
(2002) for the irrigated land use classes is 352 mm, 603 mm and 957 mm for rabi, kharif
and annual periods, respectively. This study was conducted during 1993-94 using Surface
Energy Balance Algorithm for Land (SEBAL) in the irrigated areas of the Indus Basin.
To supplement this incremental ET, additional amount of water is required to be supplied.
The blue water resource from the rainfed crops and woods land use is thus contributing to
irrigated land uses along with the amount coming from land use classes other than
agricultural.
4.4
Conclusions
In areas with sparse rain gauge station networks, the estimation of rainfall from satellites
contains important information for water management applications. Because of small
runoff coefficients in arid climates, absolute values of basin-wide rainfall need to be
known accurately. The recent floods in Pakistan confirmed that basin scale rainfall is a
vital piece of information. However satellite estimates are biased and need area specific
corrections.
Here we proposed two alternative techniques to calibrate TRMM satellite data using
limited information on rainfall from gauges. The uniqueness of these techniques is that
influence of geo-statistics is reduced and more intensive usage of spatial TRMM data is
used. The first technique is purely based on a regression analysis while the second
technique uses the systematic spatial structure of bias between the satellite and rain gauge
measurements.
Both techniques provided improved estimates of the spatial rainfall distribution. However,
the geographical differential analysis performed better than the regression analysis in the
mountainous Indus Basin. Since the 3B43 (V6) product is a result of three different
spectral sensors, it seems easier to correct the 3B43 product instead of calibrating half
products from individual sensors.
The first research question was addressed by calculating deviations between calibrated
TRMM rainfall and corresponding rain gauge measurements. The average percentage
deviation is found as an inverse function of temporal periods and spatial scales. The precalibrated datasets have more than 50% deviation in measured and estimated rainfall for
the periods under consideration. The calibration of the TRMM rainfall has considerably
reduced the deviation. The annual and seasonal periods showed deviations of 10.9% and
14.3%, respectively for an area of 625 km2. The average deviation percentage increased to
24.2% and 34.9% for quarterly and monthly periods, respectively.
Increased spatial coverage also lowered the deviation. The deviations at five spatial scales
ranging between 625 km2, 1250 km2, 1875 km2, 2500 km2and 3125 km2 were tested. The
deviations decreased from 10.9% to 6.1% for an annual time period. Lower deviations are
thus observed at annual and seasonal time scales and at areas of a few thousand km2 and
larger.
The second research question was addressed by observing influence of geography on
satellite estimates of rainfall. The TRMM deviation seems to increase exponentially with
67
terrain elevation. The TRMM tends to underestimate rainfall in the north western hilly
tracts and southern coastal areas. In the foothill plains and mountainous ranges, the
TRMM overestimated rainfall. The availability of spatially corrected rainfall from TRMM
has the potential to create a daily stream of calibrated rainfall across the Indus Basin,
without becoming reliant on local gauge readings and complex data exchanges and
intermittent data transfers.
68
69
5 Validation of surface soil moisture from AMSR-E using
auxiliary spatial data in the transboundary Indus Basin
Chapter based on: Cheema, M.J.M., Bastiaanssen, W.G.M. and Rutten, M.M., 2011. Validation
of surface soil moisture from AMSR-E using auxiliary spatial data in the transboundary Indus
Basin. Journal of Hydrology 405, 137-149.
5.1
Introduction
Soil moisture is a key parameter that affects hydrological processes at the land surface.
The term soil moisture refers to water present in the pores of the unsaturated soil while
surface soil moisture describes the moisture of the upper few centimeters of soil. Surface
soil moisture determines the partitioning of precipitation into infiltration or surface runoff
(Hellebrand et al., 2009). Soil moisture plays an important role in drought predictions
(Loew et al., 2009), flood forecasting (Parajka et al., 2006) and crop yield predictions (De
Wit and van Diepen 2007).
Soil moisture is spatially and temporally heterogeneous, even in small catchments
(Gomez-Plaza et al., 2000; Dunne et al., 2007). This heterogeneity is due to variability in
soil physical properties (Hawley et al., 1983), geology, land use (Rosnay et al., 2006),
rainfall distribution, irrigation and topography (Western and Blöschl, 1999). The
relationship between these governing hydrological processes and soil moisture is
extremely complex, and it is therefore better to measure it rather than modeling.
Conventional ground based techniques like gravimetric sampling, neutron probes, time
domain reflectometry, capacity probes, etc. are commonly used to measure soil moisture at
a specific depth. These field measurement techniques have some practical limitations,
although very valuable as a direct measurement of soil wetness. For example, the
gravimetric method can produce erroneous estimates due to moisture losses during
sampling and transport to the laboratory for drying. Other direct soil moisture measuring
methods are non-destructive, but need a soil specific calibration against the gravimetric
method. These measurements are generally carried out at small spatial scales. The
measurements represent a relatively small sphere of influences of the sensor. It is laborious
and time consuming to get reasonable large spatial and temporal soil moisture estimates
(Hemakumara et al., 2004a; Van der Kwast, 2009). It can therefore be concluded that in
situ soil moisture measurements are not sufficient to capture consistent spatial and
temporal variability, especially at large scales (Owe et al., 2008). A very comprehensive in
situ observation network is required to have convincingly good spatial estimates (Yoo,
2002). This is technically not feasible in vast areas such as the Indus Basin. Therefore,
alternative solutions need to be developed.
Satellite remote sensing has emerged as an alternative source of providing spatially and
temporally consistent soil moisture datasets. Passive microwave and active radar
microwave technologies can provide measurements of surface soil moisture (e.g.
Schmugge, 1983; Choudhury, 1991; Jackson, 1993; Wagner et al., 1999). Soil moisture
retrieval from active microwave sensors is less reliable due to surface roughness
interferences (Kerr, 2007). Therefore passive microwave technologies are potentially more
70
interesting, assuming that vegetation effects can be removed. There are satellites available
that can estimate soil moisture using passive microwave imagery, including Nimbus
Scanning Multi-channel Microwave Radiometer (SMMR), Special Sensor
Microwave/Imager (SSM/I),Tropical Rainfall Measuring Mission Microwave Imager
(TMI) and Advanced Microwave Scanning Radiometer on the Earth Observing System
(AMSR-E) (Choudhury and Golus, 1988; Ray and Jacobs, 2007). Due to the low natural
emittance capacity in the microwave spectral region, all spaceborne systems have very
large pixel sizes (25 to 50 km).
Remotely sensed soil moisture is often validated by comparing it with in situ
measurements (e.g. Bosch et al., 2004; Jackson et al., 2008; Jackson et al., 2009; Wang et
al., 2009) , with models (e.g. Wagner et al., 2007; Scipal et al., 2008; Draper et al., 2009;
Gruhier et al., 2010), or with related phenomena like precipitation and its known errors
(e.g. McCabe et al., 2005; Crow and Zhan, 2007; Crow et al., 2010). The genuine
validation problem of large size space-borne soil moisture estimates limits its operational
use in water management and hydrological studies. The significant spatial variability in
soil moisture restricts the validation process because sensors cannot be installed wherever
needed (Narayan et al., 2004). Therefore a comprehensive observation network is not
feasible for vast and river basins with complex terrain. The influence of various surface
and meteorological factors makes it cumbersome to compute the average soil moisture
value for a 25 km×25 km area from in situ measurements (Cosh et al., 2004). Without
overcoming these constraints in validation, confidence in the data by the water resources
community will be limited, and it is thus important to develop some alternative validation
procedures. The current paper contributes to that.
This paper explores the validity of operational soil moisture information from AMSR-E
data for the support of water management practices in large scale river basins. Confidence
in NSIDC’s AMSR-E surface soil moisture product will facilitate various applications
such as the calibration/validation of spatially distributed hydrological models that can be
applied on large scale irrigated river basins with scarce data (Droogers et al., 2000; Ines
and Mohanty, 2008).Accurate soil moisture data is also essential for describing droughts
and floods. AMSR-E data for the Indus Basin will be evaluated by comparing it against
auxiliary spatial data. In this study, validation against rainfall, vegetation seasonality
inferred from SPOT-Vegetation NDVI time series and saturated water content (θsat) from
standard soil maps will be conducted.
5.2
Materials and methods
5.2.1
Study area and landuse patterns
The study area is the Indus Basin encompassing four countries (i.e. Pakistan, India, China
and Afghanistan). Total area of the basin is 116.2 million ha (mha) and lies in between 24°
38′ to 37° 03′ N latitudes and 66° 18′ to 82° 28′ E longitudes. The elevation above mean
sea level in the basin ranges between 0 – 8600 m from south to north. The basin exhibits
complex hydrological processes due to variability in rainfall, topography, climatology and
land use. The mean annual rainfall varies from approximately 200 to 1500 mm for year
2007 with a basin average of 383 mm yr-1(Cheema and Bastiaanssen, 2012).
71
The basin hosts the largest contiguous irrigation system in the world. The irrigated
agriculture is concentrated in the middle and lower parts of the basin. Rainfed agriculture
is also practiced in upstream parts of the basin. Other than irrigated and rainfed land uses,
savanna deciduous, pastures deciduous alpine, pastures deciduous lowland, and bare soil
are dominant land use classes (Figure 3-5).
5.2.2
Remote sensing data
Remotely sensed surface soil moisture is retrieved from AMSR-E sensor on board of the
Aqua satellite of the National Aeronautics and Space Administration (NASA). Surface soil
moisture is derived from brightness temperature (TBp) observations using dual polarization
Njoku et al. (2003) retrieval algorithm (see Appendix). The data is freely available through
NASA’s official National Snow and Ice Data Center (NSIDC) website
(http://nsidc.org/data/ae_land3.html)(Njoku, 2008).. The soil moisture (product level 3) is
provided as a re-sampled product at 25 km resolution. Daily soil moisture datasets
(ascending and descending path) covering the Indus Basin were downloaded for the year
2007
A detailed Land Use and Land Cover (LULC) classification map of the Indus Basin
developed by Cheema and Bastiaanssen (2010) was used to explain the surface soil
moisture results. The LULC map was developed using multi-temporal (10-day composite)
Normalized Difference Vegetation Index (NDVI) images. The spatial resolution of the
LULC map is 1 km ×1 km with 77% overall mapping accuracy. This is the first detailed
land use map with 27 classes giving information about different land uses and cropping
patterns in the Indus Basin.
Rainfall data was obtained at a coarser spatial resolution of 25 km using Tropical Rainfall
Measuring Mission (TRMM) processing algorithms described by Huffman et al. (2007). A
range of orbital and gridded TRMM products are available through NASA website
(http://neo.sci.gsfc.nasa.gov/Search.html?group=39). In this study, the global rainfall
algorithm (3B42 V6) was used. 3B42 V6 provides daily accumulated rainfall data. Its
spatial (25 km×25 km) and temporal (daily) resolutions are equivalent to AMSR-E
resolution. The rainfall data has been calibrated against gauge readings.
5.2.3
Methodology
In this study, a new approach was adopted to validate the surface soil moisture data
obtained from AMSR-E. This technique involves a comparison against land use, rainfall,
vegetation seasonality, and saturated moisture content information.
Two hypotheses were tested. (1) The antecedent rainfall is solely responsible for surface
soil moisture variability and (2) the temporal patterns of soil moisture reflect the
phenology of vegetation dynamics expressed by NDVI. The mean, maximum, and
minimum soil moisture for different land uses were mapped at annual and daily scale. The
seasonal soil moisture status was examined for the two growing seasons that normally
prevails in the Indus Basin i.e. rabi (winter) and kharif (summer).
72
Table 5.1 Spearman’s rank correlation coefficient between 8-day mean soil moisture and rainfall in dominant land use class pixels
(column 9). Location and soil types of those pixels are also given.
(1)
Sr
(2)
(3)
(4)
(5)
(6)
(7)
(8)
(9)
(10)
Dominant
Pixel Location
Soil
Annual θmean Annual θmax
θsat
Deg of sat.
rs
p-value
land use class
type*
(cm3cm-3)
(cm3cm-3)
(cm3cm-3)
Se
Lat. N
Long. E
1 Irrigated rice, wheat rotation 32°18΄55˝ 74°40΄38˝
Jc42
0.121
0.180
0.448
0.270
0.49
0.00
2 Irrigated rice, wheat rotation 32°04΄21˝ 74°25΄50˝
Be71
0.115
0.157
0.432
0.266
0.52
0.00
3 Irrigated rice, wheat rotation 31°18΄17˝ 74°40΄23˝
Xk19
0.116
0.179
0.425
0.273
-0.04
0.82
4 Irrigated rice, wheat rotation 30°33΄48˝ 75°25΄56˝
Xk19
0.128
0.243
0.425
0.301
0.13
0.39
5 Irrigated mixed cotton, wheat 32°34΄00˝ 73°55΄55˝
Xh42
0.121
0.177
0.396
0.306
0.42
0.00
rotation/sugarcane
6 Irrigated mixed cotton, wheat 31°49΄16˝ 72°40΄43˝
Yh26
0.107
0.154
0.398
0.269
0.22
0.15
rotation/sugarcane
Yk42
7 Irrigated mixed cotton, wheat 32°18΄55˝ 73°25΄28˝
Jc42
0.124
0.196
0.445
0.279
0.49
0.00
rotation/orchards
8 Irrigated mixed cotton, wheat 30°34΄04˝ 72°25΄38˝
Jc42
0.123
0.221
0.428
0.287
0.05
0.72
rotation/orchards
Yk42
9 Rainfed crops wheat/grams
31°49΄00˝ 71°55΄27˝
Qc47
0.068
0.125
0.448
0.152
0.42
0.03
10 Rainfed crops wheat/grams
32°04΄05˝ 71°55΄27˝
Yh26
0.079
0.139
0.397
0.198
0.14
0.34
11 Rainfed crops general
31°49΄16˝ 71°10΄42˝
Qc47
0.085
0.135
0.451
0.189
0.30
0.04
12 Rainfed crops general
34°19΄12˝ 71°08΄52˝
Qc47
0.086
0.135
0.451
0.189
0.31
0.03
13 Rainfed crops general
25°49΄11˝ 70°40΄47˝
Qc47
0.039
0.072
0.448
0.087
0.52
0.03
14 Rainfed crops general
25°34΄06˝ 67°40΄45˝ I-Rc-Yk
0.047
0.084
0.438
0.107
0.52
0.00
15 Rainfed crops general
28°49΄14˝ 67°40΄45˝
Yh27
0.066
0.257
0.386
0.171
0.51
0.00
16 Bare soil
27°19΄12˝ 69°25΄35˝
Qc47
0.036
0.064
0.448
0.080
0.26
0.08
17 Bare soil
28°03΄57˝ 72°55΄33˝
Qc47
0.049
0.095
0.435
0.113
0.42
0.03
18 Very sparse vegetation
32°04΄21˝ 68°40΄51˝ I-Rc-Yk
0.062
0.105
0.438
0.142
0.55
0.14
19 Very sparse vegetation
31°48΄28˝ 68°25΄45˝ I-Rc-Yk
0.062
0.124
0.438
0.142
0.48
0.00
20 Pastures deciduous
28°04΄13˝ 70°55΄21˝
Qc47
0.044
0.065
0.448
0.098
0.22
0.02
21 Pastures deciduous
29°49΄51˝ 67°40΄45˝ I-Rc-Yk
0.090
0.176
0.438
0.205
0.55
0.00
*Jc: Calcaric Fluvisols
Xh: Haplic Xerosols
Qc: Cambic Arenosols
Be: Eutric Cambisols
Yk: Calcic Yermosols
I: Lithosols
74
Xk: Calcic Xerosols
Yh: Haplic Yermosols
Rc: Calcaric Regosols
The relationship between soil moisture, rainfall and NDVI for specific different land use
classes could be investigated for a few selected AMSR-E pixels with homogeneity in land
use. There was, however, a scale mismatch between LULC (1 km×1 km) and AMSR-E
soil moisture (25 km × 25 km) maps. The AMSR-E map was superimposed on LULC map
and pixels were selected that contain a dominant LULC class. The criterion of more than
80% being one single LULC class within a single AMSR-E pixel was applied. This
satisfies the recommendations posted by Adegoke and Carleton (2002). They suggested
70% of a certain pixel area should have homogeneous LULC to qualify for comparisons
with coarser resolution soil moisture datasets. This was necessary to get pixels with a
single homogenous LULC class in order to have soil moisture information maximally
representing the particular class. It was attempted to get a maximum number of available
pixels covering both irrigated and non-irrigated land use classes. Finally, 21 pixels were
found that met all criteria. These represented eight land uses including irrigated, rainfed,
bare soil, and very sparse vegetation. Table 5.1 defines the location of these pixels. For
soil moisture data, 8-day mean time series were used for the year 2007 (a normal rainfall
year). The daily precipitation data was also accumulated into 8-day totals for convenient
comparison.
To determine concurrence between soil moisture, rainfall and NDVI in the selected pixels,
correlation tests were performed. A spatial correlation between average soil moisture and
accumulated rainfall taking maximum NDVI into consideration was performed for the 21
pixels. The analysis was carried out at annual and seasonal (rabi and kharif) periods to
verify the existence of relationship among these parameters. A temporal correlation
between the soil moisture and rainfall was also performed. For the purpose, Spearman’s
rank correlation coefficient (rs) was calculated using the method described by Press et
al.(1996). It is considered an appropriate non-parametric statistic for data that are not
normally distributed (Cardina and Sparrow, 1996). It expresses percentage association
between two parameters by comparing ranks of the two parameter (Vachaud et al., 1985)
given as:
n
rs = 1 −
6∑ d i
i =1
(
2
)
n n2 −1
5.1
where, “rs”is Spearman’s rank correlation coefficient. “di” is difference between each
rank of corresponding values of soil moisture and rainfall. “n” is number of pairs of
observations. A value of rs lies between −1 and +1. Values closer to 1 show more close
association of the two parameters. A negative sign indicates inverse association.
For testing of the second hypothesis, the trends between soil moisture and NDVI
(describing extent of vegetation in a season) were examined for the 2007 annual cycle.
Time series of 10-day composite NDVI and 10-day mean soil moisture were plotted. This
temporal analysis was performed for each land use as a homogeneous area because
management of land resources differs. The LULC map was re-sampled (with majority
rule) at 25 km. The dependence of the NDVI on soil moisture was checked using
Pearson’s product moment correlation (r) (Eq. 5.2). Pearson’s correlation coefficient was
examined for concurrent and lag time to know whether, and for how long NDVI responds
to antecedent soil moisture.
r=
∑ (x − x )( y − y )
∑ (x − x ) ∑ ( y − y )
2
2
5.2
where, “x” and “y” represents soil moisture and NDVI time series, respectively.
The maximum values captured by AMSR-E were investigated by comparing them to the
saturated soil moisture content (θsat) limit of the soils found in the Indus Basin. The
information on θsat for different soil types in the Indus Basin were inferred from the FAO
digital soil map of the world (FAO, 2008). Pedo-transfer functions were applied to express
soil textural information into θsat (Droogers, 2006). More information on pedo-transfer
functions can be found in Wösten et al. (1995), Batjes (1996) and Nemes et al. (2001).
5.3
Results and discussion
The annual soil moisture values for individual land use classes are shown in Figure 5-1.
These values represent average values for 365 days and for individual land use classes.
The daily maximum soil moisture values, daily minimum, and daily mean soil moisture
values are provided. The maximum and minimum soil moisture values attained for single
days during the year in each class are plotted in addition to show the extreme values in the
AMSR-E dataset. This provides insight in the full range of AMSR-E based soil moisture
values for each land use class. The corresponding total annual rainfall (year 2007) for a
particular land use is added for the sake of extra information. Some concurrence is
observed between the peak soil moisture and rainfall in non irrigated land uses (e.g.
“rainfed crops”, “bare soil”, and “pastures”). For “irrigated land use”, the peak soil
moisture values do not match with rainfall as expected. The peak mean soil moisture for
“irrigated rice-wheat rotation” is higher with lower rainfall due to the extra water supply
from canal water and groundwater. A similar response is observed in “irrigated rice-fodder
rotation” and this agrees with the objective of irrigation systems to enhance soil moisture
values and crop evapotranspiration.
76
Figure 5-1 Relationship between annual rainfall and surface soil moisture for different
land use classes in the Indus Basin. Daily maximum and minimum values are soil moisture
observed once in 365 days.
The mean soil moisture exhibits some concurrence with the accumulated rainfall in LULC
classes. A spatial correlation at annual and seasonal periods between the mean soil
moisture, accumulated rainfall and maximum NDVI for 21 single land use dominant
pixels is given in Table 5.2. At annual time scale, a stronger correlation exists between the
soil moisture and rainfall (rs=0.74) than between rainfall and NDVI (rs=0.70). A stronger
correlation also exists between soil moisture and NDVI (rs=0.85) than between rainfall and
NDVI (rs=0.70). Similar correlation is also found for rabi and kharif seasons.
Table 5.2 The correlation between the AMSR-E mean soil moisture, TRMM accumulated
rainfall and maximum NDVI for the dominant land use pixels.
Period
Parameter 1
Parameter 2
rs
p-value
Annual
Soil moisture
Rainfall
0.74
1.25E-04
Annual
Soil moisture
NDVI
0.85
1.20E-06
Annual
Rainfall
NDVI
0.70
3.97E-04
Rabi
Soil moisture
Rainfall
0.69
4.77E-04
Rabi
Soil moisture
NDVI
0.78
3.14E-05
Rabi
Rainfall
NDVI
0.50
2.23E-02
Kharif
Soil moisture
Rainfall
0.68
7.03E-04
Kharif
Soil moisture
NDVI
0.91
1.51E-08
Kharif
Rainfall
NDVI
0.66
1.26E-03
77
The dynamic range of soil moisture content (0.05 to 0.20 cm3cm-3) is dampened due to
averaging all the soil moisture values encompassed by each LULC class. The maximum
and minimum soil moisture values in response to rainfall during individual days show
large variations. The minimum and maximum soil moisture values in the LULC class
“bare soil” for example are 0.01 and 0.36 cm3cm-3, respectively, while “rainfed crops”
generally show a range varying between 0.01 and 0.38 cm3cm-3. The maximum values are
in accordance with Wang et al. (2009), who observed maximum soil moisture values
ranging from 0.25 to 0.40 cm3cm-3for similar types of land use in North West China. They
used AMSR-E data at 6.9 GHz using annual minimum microwave polarization difference
index. The minimum soil moisture values reveals that AMSR-E (NSIDC) algorithm is
capable of capturing very low soil moisture values on bare soil and rainfed croplands
(~0.01 cm3cm-3).
The Indus Basin lies in arid to semi arid climate zones, soil moisture derived from rainfall
is not sufficient to meet crop water requirements. In dry atmospheric conditions, soil
moisture from rainfall needs to be supplemented by irrigation to grow crops (Ozdogan and
Gutman, 2008). Irrigation in agriculture is expected to be applied at specific time intervals
and according to a pre-defined rotational water supply schedule. Irrigated land exhibits
extremely high soil moisture values (up to 0.37 cm3cm-3) with an average of 0.33 cm3cm-3.
Thus, in a way surface soil moisture information combined with rainfall describes
irrigation practices and can be used to distinguish rainfed from irrigated crops. A scatter
plot describing the spatial relationship between AMSR-E mean soil moisture and annual
accumulated rainfall along with maximum NDVI values for 21 pixels is provided in
Figure 5-2. It is evident from the Figure 5-2 that the irrigated pixels have relatively higher
soil moisture content even with less rainfall as compared to the rainfed pixels.
Figure 5-2 Relationship between TRMM accumulated rainfall for year 2007 and AMSR-E
mean soil moisture (n=21). Size of the dots indicates the maximum NDVI value. Green
filled dots are irrigated.
78
Soil moisture and rainfall time series of year 2007 were considered for testing the first
hypothesis on soil moisture response to rainfall. For this purpose, 21 pixels were
investigated while the temporal variations in two selected pixels are for the sake of
clarification shown in Figure 5-3(a) and Figure 5-3(b).
Irrigated mixed cotton-wheat rotation/sugarcane
0.25
(a)
Rainfed crops general
0.16
0
20
0
20
(b)
0.14
40
0.15
60
80
0.10
100
AMSR-E soil moisture (cm3/cm3)
40
TRMM (3B42) Rainfall (mm)
AMSR-E Soil soisture (cm3/cm3)
0.20
0.12
60
0.10
80
0.08
100
120
0.06
140
0.04
160
0.05
180
rs = 77%
Rainfall
Soil moisture
SM_mov_avg
360
344
328
312
296
280
264
248
232
216
200
184
168
152
136
120
88
72
200
104
0.00
56
360
344
328
312
296
280
264
248
232
216
200
184
168
152
136
88
120
104
72
56
40
8
24
Days of year
140
SM_mov_avg
0.00
40
Soil moisture
25°34΄06˝ N
67°40΄45˝ E
24
Rainfall
0.02
8
31°49΄16˝ N rs = 33%
72°40΄43˝ E
120
Days of year
Figure 5-3 Comparison of 8-day mean AMSR-E soil moisture and accumulated TRMM
rainfall for two AMSR-E pixels (25 km × 25 km) with dominant land use (a: irrigated
cotton-wheat rotations mixed with sugarcane in the location 31°49΄16˝ N and 72°40΄43˝ E
and b: rainfed crops in the location 25°34΄06˝ N and 67°40΄45˝ E). The solid line shows
three periods moving average of soil moisture.
In the irrigated pixel with cotton-wheat rotations mixed with sugarcane (e.g. Figure
5-3(a)), soil moisture time series between day of year (DOY) 120 and 152 decreases due to
higher crop evapotranspiration rate. The soil moisture peaks show some relationship with
rainfall, but the overall relationship is weak. This can be explained by the heterogeneous
character of a mixed pixel in the middle Indus Basin at 31°49΄16˝ N latitude and
72°40΄43˝ E longitude. The peaks in soil moisture are associated with the irrigation season
and time of more intense irrigation caused such peaks. Increase in soil moisture due to
irrigation has also been reported by Singh et al. (2005). They analyzed spatial and
temporal patterns of surface soil moisture from the SSM/I sensor covering eastern parts of
the Indus Basin. Figure 5-3(a) shows that soil moisture increases in June due to irrigation
at the emergence stage of kharif crops. It attains maximum values (0.13 cm3cm-3) during
July and August. Afterwards, decreases to a minimum (0.08 cm3cm-3) at senescence stage.
It should be noted that these magnitudes of soil moisture are systematically lower than the
moisture values of the root zone.
The pixel of the LULC class “rainfed crops” shows a strong response to rainfall (Figure
5-3(b)). Rainfall is the only source of soil moisture. The peaks of soil moisture relate well
to antecedent rainfall in the lower Indus Basin at 25°34΄06˝ N latitude and 67°40΄45˝ E
longitude. The temporal patterns of rainfall and soil moisture for these land uses are
characteristically concurrent. Extremely low soil moisture content (~0.04 cm3cm-3) is
observed during dry spells when top soil is heated up and dries out until the residual soil
moisture content is reached. Soil moisture increases to 0.09 cm3cm-3 after rainfall events.
This range of surface soil moisture is in accordance with the findings of Rao et al. (2006)
and Rao et al. (2008). They found average soil moisture values to be in the range of 0.06
to 0.10 cm3cm-3 (average) for the eastern parts of the Indus Basin with similar land uses.
Rao et al. (2006) studied the soil moisture patterns using AMSR-E (NSIDC) with limited
79
ground truth data for the years 2002-05. Rao et al. (2008) used ALOS PALSAR to
estimate soil moisture for the year 2006-07.
High evaporative demand results in quick decrease of soil moisture. In dry soils and arid
climates, the depletion process is activated soon after a rainfall event (Hu et al., 2008). The
overpass of Aqua satellite occurs at the local solar time of 01:30 and 13:30 for
ascending/descending orbit. This lag in ascending and descending passes can result in less
capturing of quickly decreasing soil moisture.
To further explore association between soil moisture and rainfall a temporal correlation is
given in Table 5.1 (column 9). The Spearman’s rank correlation coefficient (rs) was
calculated for the specific pixels applying Eq.5.1. The Spearman’s rs of soil moisture with
the antecedent rainfall (rainfall occurred in 8 – day period) ranges from -0.04 for irrigated
to 0.55 for non irrigated land uses. High rs is not found for irrigated land uses except
“irrigated rice-wheat rotation” at two locations (32°18΄55˝ N, 74°40΄38˝ E and 32°04΄21˝
N, 74°25΄50˝ E with 0.49 and 0.52, respectively). Irrigated rice-wheat rotation lies outside
the Canal Command Areas (CCA) in a higher rainfall zone thus supplemented with less
irrigation (Cheema and Bastiaanssen, 2012). Therefore, a higher association between
rainfall and soil moisture is observed for these locations.
Soil moisture shows reasonably higher association with antecedent rainfall in the non
irrigated or rainfed land use pixels. The “rainfed crops” and “sparsely vegetated” land use
shows maximum rs values of 0.52 and 0.55, respectively. .In the pixels with dominant
rainfed land uses, higher association is observed between the rainfall events and AMSR-E
captured soil moisture. However, this association is variable at different location for a
similar land use. This spatial variability could be due to the influence of soil physical
properties. Soil properties affect the infiltration and evaporation processes (Canton et al.,
2004). Therefore, one land use class at two different locations with different soil physical
properties may have different correlations.
The mean correlation of the irrigated land cover pixels (0.29) is lower than the mean of the
rainfed land cover (0.36). The significance of this difference in mean correlation was
investigated with a Monte Carlo simulation experiment using 1000 pairs of randomly
generated time series with time length t=46. The probability distribution function was
used to test the significance of the difference between the mean rs values. Based on this
experiment the p-value of the difference is estimated at 0.2. The significance of the
difference is found low. The analysis of longer time series is therefore necessary to proof
the statistical significance of the results.
To investigate the daily variation of soil moisture in response to rainfall occurrences, two
pixels of similar land cover exhibiting relatively higher soil moisture values were selected
(Pixel 1:latitude 28°18΄42˝ N, longitude 67°56΄04˝ E and Pixel 2: latitude 30°33΄39˝ N,
longitude 75°55΄54˝ E). It is found that the retrieved soil moisture does fluctuate according
to the major rainfall events as it can be seen in Figure 5-4(a). The rainfall events with high
intensity and for consecutive days have increased the chances to capture the peak soil
moisture values e.g. during March (0.196 cm3cm-3) and July (0.375 cm3cm-3). The pixel 1
is located in water logged but dry climatic region receiving 204 mm of rainfall during the
whole year. The intense and continuous rainfall events may result in more moisture
80
retention at soil surface due to water logged conditions thus allowing AMSR-E to capture
higher soil moisture values.
Pixel 2 is irrigated pixel with arable soil and in wet region receiving 591mm of annual
rainfall however the soil moisture response to rainfall is not sensitive. The periodic
irrigation and arable conditions has resulted in weaker temporal relationship. The higher
soil moisture values between June and July (0.29 cm3cm-3) correspond to the upraising
period of rice paddies. The other peak during October (0.184 cm3cm-3) is corresponding to
the land preparation for cultivation of wheat crop. However, a peak in month of April is
less understood as the period corresponds to the harvesting of wheat crop. The very low
soil moisture values (~0.013 cm3cm-3) during month of May represent the bare soil
conditions after wheat harvesting. The high temperature increases evaporation thus
reducing the soil moisture.
Figure 5-4 (a) Comparison of AMSR-E daily soil moisture and TRMM rainfall for pixel no
1. The dotted line is three day moving average. (b) Comparison of AMSR-E daily soil
moisture and TRMM rainfall for pixel no 2.The dotted line is three day moving average.
Soil moisture retrieval by AMSR-E was further validated by investigating time series of
soil moisture and growth phenology of each land use. Although this cannot validate the
estimated soil moisture quantitatively, it can at least provide qualitative trend information
using the dynamics of soil moisture. In Figure 5-5, the 10-day mean of soil moisture
values were plotted against 10-day composite SPOT-Vegetation NDVI values (mean of
each land use for each layer).
0.16
0.80
Irrigated mixed cotton,wheat rotation/sugarcane
NDVI
0.15
0.60
0.25
Top soil moisture
Rainfed crops general
Top soil moisture
NDVI
0.70
0.50
0.13
0.50
0.20
AMSR-E soil moisture (cm3/cm3)
0.60
AMSR-E soil moisture (cm3/cm3)
0.14
0.40
0.15
0.40
0.11
0.30
0.10
0.20
0.09
0.10
0.30
0.10
0.20
0.05
Days of year
0.10
0.00
0.00
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
0.00
10
20
30
40
50
60
70
80
90
100
110
120
130
140
150
160
170
180
190
200
210
220
230
240
250
260
270
280
290
300
310
320
330
340
350
360
0.08
NDVI
NDVI
0.12
Days of year
Figure 5-5 Relationship of AMSR-E mean soil moisture and responding NDVI for the year
2007. Soil moisture in 10-days means and NDVI is 10-day composite time series.(a)
Irrigated land use and (b) Rainfed land use.
81
Seasonal trends in Figure 5-5, shows that the land use phenology described by NDVI time
series has a similar pattern as soil moisture. The soil moisture has two peaks, one in rabi
season and the other in kharif season. Corresponding NDVI peaks are also significant for
the two seasons but with a time lag. This trend is observed for all land uses. Farrar et
al.(1994) reported that the NDVI for different soils is controlled by soil moisture of the
concurrent and preceding months. Such a time lagged relationship can also be expected for
surface soil moisture. This time delay varies according to the type of vegetation, i.e. land
uses and also with the growing seasons. In irrigated land uses (e.g. Figure 5-5(a)), soil
moisture values are higher (0.10 cm3cm-3) with lower NDVI (0.25) and suppresses (<
0.085cm3cm-3) as NDVI rises (0.45). The rise in NDVI resulted in complete cover of soil
surface by the crop canopy that can suppress the soil moisture estimates. The resulted soil
moisture suppression in zones with higher NDVI and closed canopies could be due to
inadequately application of vegetation corrections of the emitted microwave radiation,
especially in irrigated areas. A drier top soil can also be the result of drying and active root
water uptake. In rainfed land uses (Figure 5-5(b)), the crop canopies are not completely
closed due to lower NDVI peaks (<0.25). The lag in such land use classes is due to
delayed response of vegetation to soil moisture. The soil moisture peaks correspond to the
wet seasons. One peak in rabi from DOY 30 to 80 receiving rainfall with low evaporation
and secondly, in kharif that receives monsoon rainfall with maxima attained on DOY 160
to 190 in the year 2007.
The NDVI peak occurs zero to 60 days after the soil moisture peak but the lag is different
for both rabi and kharif seasons. Therefore, the analysis was conducted separately on the
two seasons. Pearson’s product moment correlation coefficient was calculated by applying
Eq. 5.2 with different time lags. It quantifies the dependence of NDVI on surface soil
moisture. The computations were carried out with time step variations of 10 days up to 40
days lag in rabi and 60 days for kharif. Beyond these days, Pearson’s correlation
coefficient (r) decreases. The maximum linear dependence of NDVI on soil moisture
observed with time lag is shown in Figure 5-6(a) & (b) for rabi and kharif season,
respectively.
In rabi(Figure 5-6(a)), seven out of 27 land use classes have shown a Pearson’s r being
less than 0.60, testing up to 40 days. The soil moisture signal is related to NDVI in 75% of
the land use classes with 0 – 40 days lag time. It is evident from Figure 5-6(a), that there is
a high linear dependence between NDVI and soil moisture in irrigated land. The maximum
linear association is observed when NDVI is correlated with concurrent or lagged 10 days
soil moisture data, which implies that crop growth is tightly coupled to irrigation supplies
and react after a bit more than a week. The periodic application of water, especially at the
start of the growing season, supplies sufficient moisture to trigger vegetation growth.
Lower temperature and less solar radiation in the dry winter reduce the soil moisture loss
through evaporation. In the LULC class “alpine land” and “snow cover”, vegetation is
dormant during rabi season and soil moisture does not affect NDVI signals.
82
Figure 5-6 (a) Maximum Pearson’s correlation between NDVI and time lagged surface
soil moisture for different land uses during rabi season.(b) Maximum Pearson’s
correlation between NDVI and time lagged surface soil moisture for different land uses
during kharif season.
The results from kharif (Figure 5-6 (b)) indicate a longer time delay for the NDVI to
respond to soil moisture. Twenty two out of 27 land uses show maximum Pearson’s r (>
0.60) with variable time lags (20 – 60 days). This lagged correlation depends upon land
uses and soil properties (Wang et al., 2007). Forests and savanna exhibit the longest time
lag. Irrigated land attains maximum r at 20 – 40 days lag. Irrigated crops in kharif thus
respond slower than in rabi, which is possibly related to the type of crop. Rice is dominant
during kharif and wheat is the major crop during rabi. For different land uses, Philippon et
al. (2005) and Propastin et al. (2006) observed seven to 84 days lag in West Africa and
Middle Asia, respectively. These observations related NDVI response to rainfall.
However, due to close relationship between rainfall and soil moisture, such a time lag may
hold true for NDVI response to surface soil moisture.
One would expect that NDVI responds better to root zone moisture than to surface soil
moisture, as observed by Wang et al. (2007). By absence of root zone moisture values,
NDVI could be compared to soil surface moisture only. NDVI has shown some linear
association with surface moisture (rs=0.85, Table 5.2) but with a time lag; this can be
ascribed to two possible factors: (i) the AMSR-E soil moisture reflects a slightly larger
depth from where some root water uptake occurs or (ii) vegetation parameters (e.g.
canopy, optical depth, and scattering albedo) interfere with the emitted microwave
radiance and by doing so account for some variation in the surface soil moisture values.
Another test was applied on the maximum soil moisture estimated by AMSR-E to check
the expected maximum values are reasonable. The θsat values in the Indus Basin obtained
83
by applying pedo-transfer function on FAO digital soil map of the world appear to be
between 0.35 – 0.55 cm3cm-3 (see Figure 5-7).
Figure 5-7 FAO saturated soil moisture content (θsat ) limits for different soil types. The
pixels are resampled to 25 km resolution.
The range of maximum soil moisture values that a pixel of 25 km × 25 km attained on a
single day during the year 2007 is between 0.08 – 0.38 cm3cm-3 (see Figure 5-8). About
3.9% of all pixels have shown values ≥ 0.30 cm3cm-3 while only 0.2% of all have attained
soil moisture values ≥ 0.37 cm3cm-3.The maximum soil moisture values are in a reasonable
range when compared with the θsat, especially when taking into account that maximum soil
moisture values are undesirable and do not easily occur under actual conditions. The
maximum values (> 0.30 cm3cm-3) are more profound in the irrigated land uses of the
basin during kharif season (not shown). The “irrigated rice-wheat rotation”, “irrigated ricefodder rotation” and “irrigated cotton-wheat rotation/sugarcane” exhibits > 0.30 cm3cm-3
soil moisture values on DOY 168 – 171, 176 – 184 and 170 – 171, respectively. These
DOYs correspond to the rainy season (Monsoon) in the basin. The maximum value (~0.38
cm3cm-3) is observed in “irrigated rice-fodder rotation” land use which comprises of
severely water logged areas in the lower Indus Basin especially in North West, Fuleli and
Begari canal commands (Aslam and Prathapar, 2001). Some higher values are also
observed in “rainfed crop general” and “bare soil” land uses which occurred on DOY 181
– 183. These higher values are normally due to the rainfall occurrence before the satellite
overpass in those areas.
The flood year 2010 has been investigated in addition because large portions of land were
flooded. The maximum values that a pixel attained during this period are given in figure
10. About 6.5% of all pixels have shown values ≥ 0.30 cm3cm-3 while 1.8% of all have
attained soil moisture values ≥ 0.37 cm3cm-3. These values were observed during months
84
of July and August, the duration coinciding with the period of severe flooding in the
region as well as basin irrigation for rice fields. Pixels fulfilling ≥ 0.37 cm3cm-3 reach
close to their physical upper limit and these areas coincide with areas being flooded. These
pixels have reached the θsat values for particular soil type (see Figure 5-7 and Figure 5-9).
This brief analysis confirms that the higher end range of AMSR-E values is appropriate.
Although not systematically investigated, the lower range of soil moisture (see Figure
5-10) reveal values in the range of 0.01 to 0.10 cm3cm-3. For sandy soils and bare land, the
values are often θ <0.03 cm3cm-3, which coincides with residual soil moisture content.
These are plausible ranges for desert surfaces.
The reason of lower maximum soil moisture values compared to θsat was explored by
taking three regions of interest (ROI) marked in Figure 5-8. The ROI 1 and 2 shows
maximum soil moisture values around 0.105 cm3cm-3to 0.145cm3cm-3 in the year 2007 and
0.111.cm3cm-3 to 0.151cm3cm-3 in the year 2010, respectively. The ROI 3 represents the
desert area, with maximum soil moisture values ranging between 0.079 cm3cm-3 to 0.108
cm3cm-3 for years 2007 and 2010. The θsat values for ROI 1, 2 and 3 are 0.438 cm3cm-3,
0.443cm3cm-3 and 0.443cm3cm-3, respectively. The plausible reasons for this variation are
the soil geography, terrain slope and arid climate. The ROI 1 and 2 are located in the areas
receiving lower rainfall and terrain slope ranging between 2 to 15%. The large slopes
restrict the underlying soils to attain the moisture content at saturation. ROI 3 is the desert
area with extremely dry climate. High evaporation and infiltration rates can restrict
underlying soil to attain moisture content equivalent to θsat.
Figure 5-8 Maximum surface soil moisture values attained by the 25 km pixel during the
year 2007 in the Indus Basin. Three regions of interests are also marked on the map.
85
Figure 5-9 Maximum surface soil moisture values attained by the 25 km pixel during the
year 2010 in the Indus Basin.
Figure 5-10 Minimum surface soil moisture values inferred by AMSR-E for the 25 km
pixel during the year 2007 in the Indus Basin.
86
Topsoil volumetric soil moisture content values exceeding 0.30 cm3cm-3are rather rare for
a 25 km pixel and this can only occur if the entire pixel is irrigated. The Indus Basin has
indeed large contiguous areas of irrigated crop land.
5.4
Summary and conclusions
The validation of soil moisture estimated by the AMSR-E satellite and interpreted with the
Njoku method from the National Snow and Ice Data Center was carried out using auxiliary
spatial data sets. The validation against (i) land use, (ii) rainfall from TRMM satellite, (iii)
seasonality of vegetation from SPOT-Vegetation NDVI and (iv) saturated water content
(θsat) from the FAO soil map and pedo-transfer functions was performed. The following
conclusions are drawn:
A strong relationship between TRMM rainfall and AMSR-E surface soil moisture in the
land use classes “rainfed”, “very sparse vegetation”, and “bare soil” exist. Spearman’s
rank correlation coefficient (rs) ranged from 0.30 to 0.55. The rainfall events in these land
uses have higher association with the soil moisture captured by AMSR-E. For irrigated
land, this association was lower (-0.04 to 0.52) due to extra supplies from irrigation – and
thus perturbations of the rainfall – soil moisture relationship. Surface soil moisture also
depends upon soil type, irrigation, and growing season in irrigated land uses. The
significance of differences between mean correlation of irrigated and non irrigated land
uses is low. At annual time scale a stronger spatial correlation exists between the AMSR-E
mean soil moisture and TRMM accumulated rainfall (rs=0.74) than between TRMM
accumulated rainfall and NDVI (rs=0.70). The soil moisture temporal variations are
consistent with the rainfall patterns in rainfed and with irrigation patterns in irrigated land
uses. The spatial correlation between soil moisture, rainfall and NDVI helped to
distinguish between irrigated and non irrigated areas.
Time lagged peaks of soil moisture and vegetation phenology expressed by NDVI time
series was observed. Such lag was expected due to delayed response of vegetation against
moisture in the root zone. The lag time varied between zero to 60 days, and was generally
speaking longer for the wet kharif season. For the dry rabi season, a Pearson’s r> 0.60 was
found for 75% of the cases with zero to 40 days lag. For the wet kharif season, it was
found for 81% cases but with a lag of 20 to 60 days. A strong spatial correlation exists
between AMSR-E mean soil moisture and NDVI (rs=0.85) at annual and seasonal periods.
The maximum values that a pixel could attain in a single day during 2007 ranged from
0.08 to 0.38 cm3cm-3. The maximum values of 0.37 cm3cm-3 and more is similar to the
theoretical expected top layer saturated moisture content and was during 2010 indeed
reported as being flooded. The high end values occur also on irrigated rice fields with
flood irrigation practices. This suggests that absolute values AMSR-E estimates are
plausible for wet land surfaces.
The complete analysis was conducted for one year cycle only, and the behavior of the
relationships between soil moisture and rainfall, NDVI and saturated moisture content is
acceptable for creating trust in the AMSR-E values. It is recommended to expand future
analysis with longer time series as shorter time series may reflect low significance. The
overall conclusion is that the AMSR-E soil moisture product from the public domain
NSIDC data centre is realistic, not only in relative terms, but also in absolute terms. Since
87
the AMSR-E soil moisture product has an acceptable accuracy, it can be used to describe
hydrological conditions and water management practices in large scale river basins.
5.5
Appendix: Soil moisture retrieval algorithm
Soil moisture data was obtained from the AMSR-E sensor due to its frequent revisit time
and the operational access of data. AMSR-E is one of the six sensors onboard Aqua
satellite, which was launched in 2002. Aqua crosses the equator at a local solar time of
01:30 and 13:30 for ascending and descending passes, respectively. This information is
for a surface layer of maximum 5 cm (Ray and Jacobs, 2007). AMSR-E measures
microwave radiations emitted by the earth’s surface expressed in terms of brightness
temperature.
Soil moisture is then retrieved using microwave radiative transfer models. Several
algorithms are available for retrieval (e.g. Njoku et al., 2003; Owe et al., 2008). One of
them has been developed by NASA’s NSIDC following the method outlined in Njoku et
al. (2003). Subsequent refinements were then proposed by Njoku and Chan (2006) to
account for vegetation surface roughness. This method uses 10.7 GHz frequency to
retrieve soil moisture. Alternatively, interpreted AMSR-E data is available from the
radiative transfer models developed by the Free University of Amsterdam in conjunction
with NASA (Owe et al., 2001; de Jeu et al., 2008). They provide soil moisture data
retrieved at lower frequency of 6.9 GHz. After inspection of the measured surface soil
moisture data in the dry areas of the Indus Basin, the data from NSIDC seemed more
plausible and in agreement with expectations for dry soils in the semi-arid and arid
climate of Pakistan and India. The analysis in this paper is therefore based on the NSIDC
interpretation models using 10.7 GHz frequency. The retrievals using this frequency are
less susceptible to RFI problems (Njoku et al., 2003).
Natural surfaces emit radiations in the microwave region, and the total signal is a function
of both land surface and atmospheric attenuation. The surface brightness temperature
observed for these natural surfaces is given by Njoku et al.(2003) as:
[
]
TBp = Tu + exp(− τ a ) {Td rsp exp(− 2τ c )}+ TS {(1 − rsp )exp(− τ c ) + (1 − ω p ){1 − exp(τ c )}{1 + rsp exp(− τ c )}}
5.3
where, “TBp” is the brightness temperature (K), “Tu” and “Td” are the upwelling and down
welling emission, respectively. “TS”is soil surface temperature (K). “τa”and “τc”are
atmospheric and vegetation opacity, respectively., “ωp”is the single scattering albedo that
depends upon vegetation structure and water content. The value of ωp is very small (Njoku
and Entekhabi, 1996) and its effect is minimal (Jackson et al., 1982). The values vary from
0.04 to about 0.13 for different crops (Owe et al., 2008). Since experimental data for this
parameter is limited, it can be considered to be negligible (ωp≈ 0)(Njoku and Chan, 2006).
The parameter “rsp” is the surface reflectivity (subscript “p” denotes vertical (V) or
horizontal (H) polarization).
88
The opacity τa along atmospheric path is dependent on the viewing angle “β”, precipitable
water “qv” and vertical column cloud liquid water path “ql”. It can be expressed as given
by Njoku and Li, (1999):
τ a = (τ o + av qv + al ql ) Cosβ
5.4
where, “τo”is oxygen opacity at nadir, “av” and “al” are the water vapor and cloud liquid
water coefficients, respectively. The vegetation opacity depends upon vegetation water
content and has approximately linear relationship (Bolten et al., 2003).
τ c = b × VWC Cosβ
5.5
-2
where, “VWC” is vegetation water content (kg m ) and “b” is an experimentally derived
vegetation parameter (Jackson and Schmugge, 1991). The values may vary for different
crops, normally taken 0.12 in absence of experimental data. “β” is the incidence angle that
can be affected by topographic features and land cover on the ground (Bartalis et al., 2006;
Friesen, 2008).
The reflectivity from rough soil surface (rsp) can be empirically related to the equivalent
smooth surface (rop) (Wang and Choudhury, 1981), through the expression:
{
= {(1 − Q )r
}
}exp(− h )
rsV = (1 − Q )roV + QroH exp(− h )
rsH
oH
+ QroV
5.6
5.7
where, parameter “h” is related to the surface height standard deviation “σ”, theoretically
given as h = (4πσ λ × Cosβ )2 while “Q” is polarization mixing parameter related to
h. The term in bracket describe mixing of the co and cross polarized scattered radiation.
The “Q” and “h” parameters are determined experimentally and vary spatially. If surface
roughness conditions are not known then “Q” is assigned 0 and “h” is assumed between 0
and 0.3 (Jackson, 1993).
The soil surface reflectivity depend upon the soil dielectric constant, which in turn
depends on soil moisture content (Dobson et al., 1985). The dielectric constant is an
electrical property of matter that measures medium response to an applied electric field (de
Jeu, 2003). Large differences between dielectric constants of dry soil (~3.5) and water
(~80) permits retrieval of soil moisture from microwave emissivity (Schmugge et al.,
1992). The effective dielectric constant is computed by inverting Fresnel equations (Ulaby
et al., 1986). These equations estimate surface reflectivity as a function of the dielectric
constant “ε” of the medium and the incidence angle “β” is based on vertical or horizontal
polarization of the sensor:
89
 ε cos β − ε − sin 2 β
roV = 
 ε cos β + ε − sin 2 β

2

 cos β − ε − sin 2 β


=
r
 o
 cos β + ε − sin 2 β


H




2
5.8
Finally, the dielectric mixing model (Dobson et al., 1985) based on soil texture, soil bulk
density “ρb” and specific density “ρs” gives the volumetric soil moisture “θo” at each pixel
of AMSR-E.
ε α = 1+
(
)
b α
ε s − 1 + θ o℘ε αfw − θ o
s
5.9
where, “θo” is volumetric water content (cm cm ). “εs”is dielectric constant of soil solids
depending upon ρs , “α” is shape factor and “γ” is empirical constant depends on soil
textural composition. “εfw”is dielectric constant of free water calculated by Debye equation
and more description on this mixing model can be found in Dobson et al.(1985).
3
90
-3
6 The surface energy balance and actual evapotranspiration
of the Transboundary Indus Basin estimated from satellite
measurements and the ETLook model
Chapter based on: Bastiaanssen, W.G.M., Cheema, M.J.M., Immerzeel, W.W., Miltenburg, I.
and Pelgrum, H., 2012. The surface energy balance and actual evapotranspiration of the
transboundary Indus Basin estimated from satellite measurements and the ETLook model.
Water Resources Research, (under review).
6.1
Introduction
Planning and monitoring of consumptive water use is necessary for sound management of
scarce water resources. Consumptive use influences social, economic, agricultural, and
environmental development. Water is consumed mainly through evaporation (E) and
transpiration (T) (jointly termed evapotranspiration (ET)) from crops, soil, forests, urban
areas, and natural vegetation, amongst others. If precipitation over a specific land cover
exceeds ET (e.g. forests), such a land cover class is a net producer of water resources.
Non-consumed water from precipitation feeds streams, rivers and aquifers. If, however,
ET exceeds precipitation, such a land cover class will be a net consumer of water
resources. Irrigated lands are a typical example of a net consumer of water. ET information
can be used for irrigation management (Bastiaanssen et al., 1996; Allen et al., 2007),
drought detection (e.g. Calcagno et al., 2007), real water savings (e.g. Seckler, 1996),
water accounting (e.g. Molden and Sakthivadivel, 1999), water productivity (e.g. Zwart et
al., 2010), virtual water trade (e.g. de Fraiture and Wichelns, 2010), model calibration (e.g.
Immerzeel and Droogers, 2008), hydrological model applications (Droogers et al., 2010a)
and groundwater management (Ahmad et al., 2005).
A number of techniques are in use to measure ET, ranging from conventional point
measurements to modeling and spatially distributed remote sensing estimates. At
individual plant and field scales, lysimeters, heat pulse velocity, Bowen ratio,
scintillometry, surface renewal, and eddy correlation are commonly used (e.g. Meijninger
et al., 2002; Nagler et al., 2005). Field scale ET measurements are generally considered
accurate, however the accuracy of these traditional methods are often less than 90%
(Twine et al., 2000; Teixeira and Bastiaanssen, 2011). The equipment cost, extensive
labor, and coverage issues restrict use of these techniques at large scale (Elhaddad and
Garcia, 2008). At the regional scale, earth observations by means of satellite data are
gradually becoming more accepted (e.g. Courault et al., 2005; Anderson et al., 2007; Mu
et al., 2007; Kalma et al., 2008; Guerschman et al., 2009; Wu et al., 2012) although
operational data provision remains rare. This paper aims at contributing to the
development of operational systems that could be applied on a daily time step for areas
with limited ground data. Routine weather data is assumed to be available.
Evapotranspiration computations are often based on surface energy balances (e.g. Price,
1990; Mu et al., 2007; Senay et al., 2007; Tang et al., 2009; Long and Singh, 2012). Many
of these energy balance models require thermal infrared radiation from cloud free images
and atmospheric corrections in order to produce accurate land surface temperature maps
91
(Jia et al., 2009). Cloud free surface temperature images for large areas in basins with
monsoon climates are not straightforward to obtain (e.g. Bastiaanssen and Bandara, 2001).
Thermal infrared radiation is more sensitive to atmospheric water vapor absorption than
visible and near-infrared radiation (Lillesand and Kiefer, 2000), and it is thus more
challenging to acquire land surface temperature maps not being thwarted by clouds. For
instance, the surface temperature product (MOD 11A2) available through Moderate
Resolution Imaging Spectro radiometer (MODIS) is thwarted by cloud cover for the entire
period of monsoon 2007 (June – September). About 50% of the basin area was found
without or with limited surface temperature data from day of year (DOY) 161 to 241. This
illustrates the difficulty in getting continuous information for ET computations in irrigated
areas from thermal infrared data. While it is generally accepted that thermal infrared data
provide reliable results based on sound physics (e.g. Bastiaanssen et al., 2008; Allen et al.,
2010; Allen et al., 2011), the cloud cover is a serious hindrance to routine applications in
various parts of the world.
To circumvent these problems, the current study deployed the ETLook algorithm that was
first introduced by Pelgrum et al. (2010). Soil moisture derived from passive microwave
sensors is the driving force for calculation of the surface energy balance in ETLook.
Surface soil moisture relates typically to a depth of 2 to 3 cm, and the number of surface
soil moisture databases is growing due to an increasing number of operational passive
microwave sensors. This is a good moment to explore and develop ET models that are
based on these data sets. Future soil moisture data layers will be based on active Synthetic
Active Radar (SAR) measurements, once this data become available easily and free of
charge.
Microwave radiometry is less affected by cloud cover (Ulaby et al., 1981; Fily et al., 1995)
and can thus provide continuous surface soil moisture information even in monsoon
periods. Li et al.(2006) have shown the value of using microwave derived near-surface soil
moisture in a two-source energy balance model over an agricultural area in central Iowa
(USA). The ETLook algorithm is a two-source model and surface soil moisture is used for
the computation of E, and a parameterization is introduced to compute sub-soil moisture
content for the determination of T.
Accurate ET information is of paramount importance for the 116.2 million hectares (mha)
Indus Basin, with high elevation water source areas, a distinct monsoon climate with cloud
covered regions, and declining water tables due to over-exploitation. This study was a first
attempt to use microwave technologies to accurately estimate ET over the Indus Basin, and
to detect areas with excessively high ET rates using a spatial resolution of 1 km. Such a
resolution is thought to be good enough for regional scale applications. The main objective
of this study was to demonstrate the validity of a combined optical and microwave based
energy balance model (ETLook) in a vast river basin with large irrigation systems.
Another objective was to use public domain data to estimate ET in the areas where field
data are not available, and to show water managers that spatially discrete ET information
is the basis for describing the major water flow path in ungauged basins.
92
6.2
Study area
The study area is the Indus Basin, which lies between latitude 24°38′ to 37°03′ N and
longitude 66°18′ to 82°28′ E. The total area of the basin is 116.2 mha and encompasses
four countries (Pakistan: 53%, India: 33%, China: 8% and Afghanistan: 6%) (Figure 6-1).
The basin exhibits complex hydrological processes due to variability in topography,
rainfall, and land use. The elevations range from 0–8000 m above mean sea level (a.m.s.l)
and mean annual rainfall varies between approximately 200 to 1500 mm. The basin-wide
average rainfall for 2007 was 383 mm yr-1 (Cheema and Bastiaanssen, 2012). The basin
has two distinct agricultural seasons, being the wet kharif monsoon season (May to
October) and the dry rabi season (November to April). Wheat is the major rabi crop while
rice and cotton are major kharif crops. The basin provides food for more than half a billion
inhabitants. To meet the water demand for such food production, and because rainfall is
inadequate for meeting the full crop water requirements, the world’s largest contiguous
irrigation system was built in the Indus Basin. The irrigation system supplies surface water
to the middle and lower parts of the Indus. Irrigated agriculture is practiced in 26.02 mha
(22.6%) area of the basin (see Figure 6-1). The era of tube well installations with
subsidized rates and direct access to water has motivated farmers to augment shortages in
surface water with groundwater resources (Shah et al., 2000). Currently 40–50 % of
agricultural water needs are met through groundwater used in conjunction with surface
water (Sarwar and Eggers, 2006). Groundwater quality is decreasing and phreatic surfaces
diminishing across the Indus Basin.
Figure 6-1 Location of the Indus Basin and provinces of different countries in the basin.
PK stands for Pakistan and IN for India. The irrigated areas in the basin are also shown.
93
6.3
6.3.1
Material and methods
Satellite data and pre-processing
Key input data for ETLook are: surface soil moisture, spectral vegetation index, surface
albedo, atmospheric optical depth, land use and land cover (LULC), soil physical
properties, and routine weather data. Surface soil moisture was obtained from the
Advanced Microwave Scanning Radiometer (AMSR-E) on the Aqua satellite. Daily soil
moisture datasets with 25 km foot print (ascending and descending path) covering the
Indus Basin were downloaded from the National Snow and Ice Data Center (NSIDC)
website (http://nsidc.org/data/ae_land3.html) for the complete year of 2007 (Njoku, 2008).
The year 2007 was selected because it was the last year prior to the start of this study, and
all required auxiliary data were available. The actual spatial resolution of C-band AMSRE soil moisture is large (approximately 70 km×40 km). AMSR-E collects 60 km resolution
C-band brightness temperature with a sampling interval of 10 km, which allows AMSR-E
C-band data to be gridded at 25 km resolution. The operational character of surface soil
moisture in NSIDC contributes to the construction of a routine provision of spatial ET data
bases. A comprehensive soil moisture data validation study in the Indus Basin was
performed by Cheema et al., (2011). It was found that both the behavior as well as the
absolute values are realistic and provide sufficient information on the spatial and temporal
changes of topsoil moisture in the Indus Basin. The daily layers were in the current study
averaged to obtain 8-day soil moisture layers to be compatible with the MODIS optical
satellite data.
This Indus Basin ETLook study required topsoil moisture at 1 km scale. Various
sophisticated methods are documented in the literature to downscale the available coarse
resolution soil moisture data to 1 km pixels (Hemakumara et al., 2004b; Merlin et al.,
2006; Friesen et al., 2008; Merlin et al., 2008; Gharari et al., 2011). All these downscaling
methods require a number of parameters and have an empirical character related to the
physiographical setting of a specific area. More research studies are required to find more
generic solutions to this problem. Due to the absence of detailed soil moisture data in the
Indus Basin, a simple method of downscaling based on effective saturation has been
adopted in this study. Each AMSR-E pixel was downscaled to 1 km using a bilinear resampling technique first. This is simplistic, but is necessity to remove step changes in the
data layers due to the texture of the large scale AMSR-E pixels. The information on
saturated and residual moisture content (θsat and θres, respectively) for each soil type was
used to calculate effective saturation (Setop xy) at 1 km grid using the definition proposed by
van Genuchten (1980) as:
S etop
, xy =
θ AMSRE − θ res , xy
θ sat , xy − θ res , xy
6.1
where Se ,xy , θAMSRE , θsat,xy and θres,xy represent the effective saturation, AMSRE soil
moisture, saturated and residual moisture content at 1 km pixel (x,y), respectively. The
values for θsat,xy and θres,xy were inferred from the Food and Agriculture Organization
(FAO) soil map (FAO, 1995) using pedo-transfer functions (Droogers, 2006). Soils with a
top
94
large pore volume (θsat,) contain more air and have a lower degree of saturation. Their drier
conditions reduce soil evaporation because soil moisture is retained stronger to the soil
matrix, and the volume with water filled pores that are needed to transport water will be
lower under dry conditions. In addition, the saturation of the subsoil (Sesub) is required for
the computation of root water uptake and subsequent crop transpiration. The preprocessing of saturation of the subsoil was done by using an empirical relationship
between Setop, the vegetation photosynthetically activity (that reflect soil water availability)
and Sesub. The following relationship is imbedded in ETLook 1.0 and was applied in the
current study:
[
]
S esub = 0.1LAI + (1 − 0.1LAI ) 1 − exp{S etop (− 0.5LAI − 1)}
6.2
where LAI is the Leaf Area Index. The basic assumption is that the degree of saturation of
the subsoil is exceeds the saturation of the topsoil when vegetation is photosynthetically
active, and that Setop affects the level of Sesub under all conditions. The green LAI reflects
the access of vegetation to soil water. In absence of green plants, moisture in the sub-soil
holds a direct analytical relationship with the moisture in the top-soil (e.g. Hillel, 1998).
Hence, passive microwave data in combination with LAI describes the daily variation of
root zone soil water content.
The Normalized Difference Vegetation Index (NDVI) is an undisputed indicator of active
vegetation and was used to compute LAI as explained further down. It has been
demonstrated by, for instance Nagler et al. (2005) and Burke et al.(2001), that NDVI is an
indicator of soil wetness and ET fluxes, which is in line with Eq. (6.2). NDVI data are
distributed by the Land Processes Distributed Active Archive Center (LP DAAC), located
at the U.S. Geological Survey (USGS) Earth Resources Observation and Science (EROS)
Center (lpdaac.usgs.gov). Two 16-day NDVI datasets (MOD13A2 and MYD13A2
(collection 5) starting from day 1 and day 9 respectively) at 1 km were used to create 8day NDVI layers. The vegetation cover (VC) was derived from NDVI following Jiang et
al., (2006) as:
 NDVI fv − NDVI
VC = 1 − 
 NDVI − NDVI
fv
bs




0.7
6.3
Threshold values of NDVI = 0.8 and 0.125 were used as boundary condition for full
vegetation cover (NDVIfv) and bare soil (NDVIbs), respectively. The LAI was computed
from NDVI values using standard asymptotic relationships between LAI and VC (e.g.
Curran and Steven, 1983; Carlson and Ripley, 1997):
 1 − VC 
LAI = − ln 

 kl 
6.4
where kl is the light extinction coefficient with a value range of 0.40 to 0.65. An average
value of 0.5 was taken for all representative vegetation types (e.g. Kale et al., 2005). The
LAI (VC) relationship was similar for all land use classes because ki values for all classes
95
were not available and we assumed that ki differences between classes were small enough
to justify the use of a few selective ki values, for all classes.
Surface albedo was also derived from standard MODIS products. The 8-day albedo data
product MCD43B3 (collection 5) at 1 km resolution was downloaded from
(https://wist.echo.nasa.gov/~wist/api/imswelcome/) server provided by LP DAAC.
Solar radiation is classically computed from the extra-terrestrial radiation in association
with an atmospheric transmissivity in the solar spectrum. The atmospheric transmissivity
of shortwave radiation can be inferred from optical depth information provided by the
MODIS cloud product (King et al., 1997). One km resolution MYD06_L2 values of the
optical
depth
product
were
downloaded
from
https://wist.echo.nasa.gov/~wist/api/imswelcome/ to estimate atmospheric transmissivity
for the Indus Basin. The cloud optical depth from MODIS products was used to infer
atmospheric transmissivity of shortwave radiation τMODIS (Barnard and Long, 2004).
A detailed LULC map of the Indus Basin developed by Cheema and Bastiaanssen (2010)
was used to infer information on different LULC classes in the basin. Twenty-seven
LULC classes were identified. This LULC classification was used to create look-up tables
for the definition of certain bio-physical parameters required for ET computations, such as
minimum stomatal resistance, moisture sensitivity and maximum obstacle height.
Rainfall (R) data are used to determine interception evaporation. Interception (I) is
computed on a daily scale with the classical von Hoyningen model following von
Hoyningen (1983) and Braden (1985).
 
 
1
I = 0.2 LAI 1 − 
(VC ) R
  1 +
0.2 LAI
 





6.5
which assumes that maximum a water film of 0.2 mm is stored per unit LAI. This
coefficient can be modified. ET cannot exceed R without being augmented by additional
water resources. Rainfall is therefore a good measure to validate against. By absence of
sufficient rain gauges, rainfall was obtained at spatial resolution of 25 km using Tropical
Rainfall Measuring Mission (TRMM) processing algorithms described by Huffman et al.
(2007). The global rainfall algorithm (3B43 V6) available through NASA website
(http://neo.sci.gsfc.nasa.gov/Search.html?group=39) was used. It provides monthly
accumulated rainfall data, which has been calibrated and validated according to the
Geographical Differential Analysis (GDA) as outlined in Cheema and Bastiaanssen
(2012).
6.3.2
Meteorological data
The major portion of the Indus Basin (53%) lies within the administrative boundaries of
Pakistan. Most of the meteorological data (e.g. air temperature, relative humidity and wind
speed) were therefore obtained from 65 meteorological stations under the aegis of the
Pakistan Meteorological Department (PMD). Weather station data for India, China and
96
Afghanistan were extracted from the National Oceanic and Atmospheric Administration
(NOAA) National Climatic Data Center (NCDC). The NCDC collects meteorological data
from real time reporting stations worldwide in agreement with World Meteorological
Organization regulations (ftp://ftp.ncdc.noaa.gov/pub/data/gsod/). Data from 16 stations
with complete datasets were downloaded. Hence, air temperature, relative humidity and
wind speed data from 81 stations collected at standard height of 2 m were obtained.
ETLook requires gridded meteorological data for air temperature (Tair), relative humidity
(RH) and wind speed (U2) at 1 km resolution. Topography, land use, sun angle, and
distance from water bodies directly affects the spatial variability of near surface
meteorological parameters (Brutsaert, 1982; Schulze et al., 1993). Ordinary geo-spatial
interpolation techniques do not take these variables into account. The meteorological
distribution model (Daymet) described by Thornton et al. (1997) was therefore used to
convert point data to spatial meteorological data. Daymet uses a truncated Gaussian
weighting filter for regional distribution of climatic variables in relation with topography.
A 1 km Digital Elevation Model (DEM) obtained from GTOPO30 database
(http://eros.usgs.gov/#/ Find_ Data/ Products_ and_ Data_ Available/gtopo30_info) was
used to establish relationships between the climatic variables and topography.
The weather grids for 2007 were independently validated against values from the
International Water Management Institute (IWMI) world water and climate atlas that is
based on long term field measurements and specific spatial interpolation procedures
(http://www.iwmi.cgiar.org/ WAtlas/Default.aspx). The atlas provides monthly summaries
of rainfall, temperature, humidity, wind speed, and sun shine hours at 18 km grid averaged
over the period 1961-1990, as produced by the University of East Anglia (New et al.,
1999). The 8-day Daymet estimates were aggregated to monthly values in order to make
them comparable with the IWMI atlas. A high coefficient of determination (R2 > 0.85) was
obtained for air temperature estimates. However, for relative humidity and wind speed,
moderate coefficients of determination (R2 =0.70–0.80 and 0.60–0.70 respectively) were
achieved. These correlations are considered reasonable because IWMI atlas values are
monthly averages for 1961-1990, which may be different for different years. The IWMI
atlas values are also interpolated and extrapolated, and associated with a certain
uncertainty. Nevertheless, the impression is that air humidity and wind speed values
display more uncertainty than air temperature.
6.3.3
Theoretical background of ETLook
The surface energy balance can be written as:
Rn − G = λ E + H
(Wm-2)
6.6
where Rn is net radiation, G is soil heat flux, λE is latent heat flux and H is the sensible
heat flux. λE is associated with ET. The ETLook algorithm uses a two layer approach to
solve the Penman – Monteith equation (Pelgrum et al., 2010). The Penman – Monteith
equation for E and T can be written as:
97
 ∆
∆ ( Rn , soil − G ) + ρ c p  e
 ra , soil
E=

r 
∆ + γ 1 + soil 
 ra , soil 



6.7
 ∆e
∆ ( Rn ,canopy ) + ρ c p 
r
 a ,canopy
T=


r
∆ + γ 1 + canopy 
 r

a , canopy 




6.8
where E and T are evaporation and transpiration respectively in Wm-2; Δ (mbar K-1) is the
slope of the saturation vapor pressure curve, which is a function of air temperature (Tair,
°C) and saturation vapor pressure (es, mbar); Δe (mbar) is vapor pressure deficit, which is
the difference between the saturation vapor content and the actual vapor content; ρ (kg m3
) is the air density, and cp is specific heat of dry air =1004 J kg-1 K-1; γ (mbar K-1) is the
psychometric constant; Rn,soil and Rn,canopy are the net radiations at soil and canopy
respectively; rsoil and rcanopy are resistances of soil and canopy, while ra,soil and ra,canopy are
aerodynamic resistances for soil and canopy respectively. All resistances are in s m-1. The
E and T fluxes (W m-2) are converted to rates (mm d-1) using a temperature dependent
function of the latent heat of vaporization.
The LAI can be used to partition the net radiation into net radiation of the soil (Rn,soil) and
the canopy (Rn,canopy) (Shuttleworth and Wallace, 1985). The increase in LAI results in an
exponential decrease in the fraction of radiation available for the soil, and vice versa for
the canopy. The energy dissipation due to interception losses is subtracted from the total
net radiation. This energy is computed from the actual interception evaporation rates and
the latent heat of vaporization being associated with that. The net radiation at the soil and
canopy can be calculated using Beer’s law as follows:
Rn , soil =
Rn ,canopy
{(1 − α ) R − L − I } exp ( −aLAI )
= {(1 − α ) R − L − I }{1 − exp ( −aLAI )}
↓
o
n
↓
o
n
6.9
6.10
where αo is surface albedo (–); R↓ (Wm-2) is the incoming shortwave radiation; Ln (Wm-2)
is the net longwave radiation; I is the interception of water by leaves expressed in Wm-2;
and a is the light extinction coefficient for net radiation. The incoming shortwave radiation
can be calculated using daily measurements of shortwave transmissivity (τsw) and the
theoretical extraterrestrial radiation (Rtoa). The parameterization for R↓ and Ln is taken from
the FAO Irrigation and Drainage Paper 56 (Allen et al., 1998). The sum of Rn,soil and
Rn,canopy constitute total net radiation Rn , after being corrected for interception losses.
98
The surface resistances in equations 6.7 and 6.8 describe the influence of the soil on
evaporation or canopy transpiration. The soil resistance (rsoil) is a function of the topsoil
effective saturation (Setop), estimated using equation 6.1. A power function defines this
relationship (e.g. Clapp and Hornberger, 1978; Camillo and Gurney, 1986; Wallace et al.,
1986; Dolman, 1993):
rsoil = b(S etop )
c
6.11
where b and c are soil resistance parameters, which can vary with soil type and are taken
here as 30 and –3, respectively. Equation (6.11) describes the transfer of water in the
liquid phase through the topsoil. Such transfer processes can also be computed with an
unsaturated Darcian flow type of equation, but the accuracy is not necessarily better, due
to high and uncertain gradients of moisture and suction in the upper 5 cm of the
unsaturated soil. The soil hydraulic conductivity of a porous soil being exposed to
anthropogenic influences under dry conditions is rather uncertain and reliance on detail
soil physical data would jeopardize the potential of ETLook. The format of Eq. (6.11) is
attractive for calibration purposes.
Canopy resistance describes the resistance of vapor flow through the transpiring
vegetation and is a function of the minimum stomatal resistance rs,min (s m-1), in association
with a number of reduction factors and the leaf area that integrates the vaporization
process from leaf to canopy scale. The value rs,min represents the resistance to transpiration
from canopy under ideal conditions (no moisture stress, enough sunshine etc.). The
resistance rs,min can have different values for the different land use classes. The rs,min is
defined for a single layer of leaves, therefore effective leaf area index LAIeff, which
describes the actual transpiring leaf mass, was used for integration from leaf to canopy.
The following equation, as described by Mehrez et al. (1992) and Allen et al. (2006), was
used to infer LAIeff:
LAI eff =
LAI
0.3LAI + 1.2
6.12
The canopy resistance under actual growing conditions can be computed using the
common Jarvis-Stewart parameterization (Jarvis, 1976; Stewart, 1988). The Jarvis-Stewart
parameterization describes the response of stomata to environmental factors in the form of
minimal resistance multiplied by the product of interacting stresses on plants, and is
computed as follows:
 r
rcanopy =  s ,min
 LAI
eff



1
 

  St S v S r S m 
6.13
where St is temperature stress, and a function of minimum, maximum and optimum
temperatures, as defined by Jarvis (1976); Sv is vapor pressure stress induced due to
persistent vapor pressure deficit; Sr is radiation stress induced by the lack of incoming
shortwave radiation; and Sm is soil moisture stress originating from the root zone. The
Jarvis-Stewart parameterization is common in many soil-vegetation-transfer models and
has not been disputed in this work. It describes the joint response of soil moisture and LAI
99
on transpiration fluxes in a bio-physically justified manner. Sm is defined using a sinusoidal
relationship with sub soil effective saturation (Sesub) and tenacity factor (Ksf) defined in
ASCE (1996) as:
S m = K sf S
sub
e
sin (2πS esub )
−
2π
6.14
where Ksf describes the ability of plants to extract soil moisture under different moisture
conditions. It ranges from 1 for sensitive plants to 1.5 for moderately sensitive plants to 3
for insensitive (tenacious) plants.
The aerodynamic resistance for soil (ra,soil) and canopy (ra,canopy) can be computed
(Holtslag, 1984; Choudhury et al., 1986; Allen et al., 1998) as:
ra ,soil
 z
ln obs
z0,soil
= 
ra ,canopy
  zobs
 ln
  0.1z
0 , soil
 
2
k uobs




 z − d   zobs − d 

 ln
ln obs
z 0,m   0.1z 0,m 

=
k 2 uobs
6.15
6.16
where k is von Karman constant = 0.41[-], uobs is the wind speed at observation height [ms1
], d is displacement height [m], z0,soil is the soil surface roughness = 0.001 m. z0,m is the
surface roughness. The land use map is used to prescribe values for z0,m. Research is in
progress to derive surface roughness from radar imagery.
The soil heat flux (G) for land surface is calculated using a sine function as described by
Allen et al.(1998). The maximum value for G is recorded in May for northern latitudes,
which coincides with a phase of–π/4. For southern latitudes the phase is –π/4 +π.
G
 2π J π 
− 
2 At , year k sin 
4
 P
exp ( − aLAI )
zd
6.17
where At,year is the yearly amplitude for air temperature; J is the Julian day measured in
seconds; k is the soil thermal conductivity (W m-1 K-1), which has a linear relationship
with top soil moisture; a is the same light extinction coefficient as used in Beers law, see
Eqs. 6.9 and 6.10; zd.(m) is the damping depth that is calculated as:
100
zd =
2kP
2πρ c p
6.18
where P is the period in seconds; and ρcp is the volumetric heat capacity (a function of the
porosity and Setop). Eq. (6.17) includes light interception effects on soil heat flux.
6.3.4
Calibration and validation approaches
The cloud optical depth measures the attenuation of solar radiation passing through the
atmosphere due to scattering and absorption by cloud droplets. The cloud optical depth can
be defined as the negative algorithm of the fraction of the incoming radiation that is not
scattered or absorbed in the atmosphere (Kitchin, 1987). Maximum and minimum
threshold atmospheric transmissivity values were taken into consideration to account for
latitude, zenith angle and diffuse radiation. The resulting atmospheric transmissivity
(τMODIS ) was checked and calibrated using the simplified - but doable - field methods
suggested by Angstrom (1924) and Hargreaves and Samani (1985). Records of sunshine
hours were used for the Angstrom equation. Sunshine records were available from 24
stations in the study area. The same 24 stations were used to get diurnal air temperature
differences for the Hargreaves equation. Results from the Angstrom and Hargreaves
methods were used to determine a linear fit through the origin for each time interval of 8
days, to obtain calibrated short wave transmissivity (τsw):
τ sw = e.τ MODIS
6.19
where e is the regression coefficient.
Minimum stomatal resistance values rs,min for each LULC were used to fine-tune ETLook.
The rs,min values for agricultural classes were accepted to be between 40 to 140 s m-1
(ASCE, 1996; Radersma and de Ridder, 1996; Bastiaanssen and Bandara, 2001). Various
researchers (e.g. Monteith, 1981; Sharma, 1985; Vanderkimpen, 1991; Allen et al., 1998)
suggested a value of rs,min = 100 s m-1 for various agricultural crops like wheat, rice, beans,
etc. Firstly, this default rs,min value was used for all agricultural classes (n=11). The
following values were assigned to the remaining classes: pastures 125 s m-1, savannas 150
s m-1, forests 150 and 300 s m-1 (for broad leaf and needle leaf forests, respectively), sparse
vegetation 200 s m-1, and urban and industrial settlements 60 s m-1. Water bodies were
assigned 0 s m-1 because water vapor molecules can be transported into the atmosphere
without physical barriers. During the second run of ETLook, all LULC classes with
irrigated crops were assigned 80 s m-1 and the rainfed crops were assigned a value of 150 s
m-1. During the third run, adjustments were made to the urban and industrial settlements
land use, and a rs,min value of 500 s m-1 was assigned. During the third run, the soil
resistance functions were also adjusted to improve congruity of the seasonal and annual
total rainfall for terrain with sparse vegetation and desert surfaces. The annual bare soil
evaporation must be lower than rainfall. After three runs, the ETLook results provided
acceptable agreement between rainfall and ET. As expected, areas with ET<R
corresponded to LULC classes “forests”, “rainfed crops”, “mountains”, and “deserts”.
Areas with ET>R appeared to be irrigated areas and water bodies. Hence, calibrated
101
rainfall patterns were used to tune the minimum stomatal resistances and the coefficients
in the formulation of the bare soil resistance.
The ET output data cannot be used for water resources management without testing its
accuracy. Results from previous studies based on soil moisture and lysimeter experiments
were used for validation. Pakistan Agricultural Research Council (PARC) measured actual
ET at Peshawar, Bhalwal, Faisalabad, Bhakkar, Mian Channu, and Tandojam representing
upper, middle and lower parts of the basin (PARC, 1982). The ET results of PARC are for
the years 1975-80 following an internationally funded study. Data collection discontinued
when the project ended, yet it seems to be one of the most basic databases in Pakistan.
More recent field measurement study was conducted by Ahmad (2002) at the Soil Salinity
Research Institute, Pindi Bhattian (31°52′34.2˝N, 73°20′50.2˝ E) and Ayub Agricultural
Research Institute, Faisalabad (31°23′26.2˝N, 73°02′49.8˝E). As part of a field
investigation program during 2000 and 2001, he measured actual ET in rice/wheat and
cotton/wheat systems by a temporarily installed Bowen ratio energy balance system.
ETLook estimates were also checked against previously conducted remote sensing and
modeling studies. The ET estimates provided by Bastiaanssen et al. (1999) for the Sirsa
irrigation circle in India were checked. Other studies e.g. Shakoor et al. (2006), Sarwar
and Bill (2007), Ahmad et al. (2009) and Shakir et al. (2010) determined ET in selected
areas within the basin for different years. Previous studies were synthesized and used to
compare with ETLook estimates. The coefficient of determination (R2), Root Mean Square
Error (RMSE) and Relative Error (RE) were calculated to estimate the difference of the
ETLook estimates with the previous studies.
6.3.5
Sensitivity and uncertainty analysis
A sensitivity analysis was performed to check the contribution of selected main input
parameters to the output results. The sensitivity of ET was tested for a number of input
parameters, i.e. θAMSRE, NDVI, rs,min, rsoil and θsat. Annual mean climatic conditions were
assumed for the analysis. One factor at a time methodology was adopted to check the
variance in the outputs due to input variability (e.g. Pitman, 1994). The analysis was
conducted on two representative land uses i.e. “bare soil” and “irrigated rice - wheat
rotation” at locations 71°22'54.123"E, 28°38'50.042"N and 75°23'53.59"E,
30°40'37.719"N, respectively. Randomly generated uniform distribution of AMSR-E
based soil moisture values (n = 100) were used while keeping other parameters constant to
check the variations in E, T and ET. The analysis was performed using representative
NDVI values of 0.05 for bare soil and 0.67 for irrigated land use. For the parameters for
which only a range was known, the defined parameter change was used to estimate the
sensitivity between input parameters and output E, T and ET.
A complete sensitivity analysis representing the change in the response variable caused by
a unit change of an explanatory variable, while holding the rest of parameters constant,
was performed. A sensitivity coefficient (SC = ∆out/∆in) was then calculated for each input
parameter as described by Gu and Li. (2002). The sensitivity coefficient was normalized
by the mean values representing the range of each pair of output and input variable. This
normalized sensitivity coefficient is called sensitivity index (SI) and can be positive or
negative. SI makes it feasible to compare the results of different input parameters. A
102
higher absolute value indicates higher sensitivity. A negative SI indicates an inverse
relationship between input parameter and response variable. SI can be represented as:
SI = 

M in
 ∆ out

∆ in 
M out 
6.20
where, Min and Mout are the mean values of the input and output range, respectively.
In addition, a stochastic uncertainty analysis was performed. A Monte Carlo simulation
experiment using 1000 pairs of randomly generated input parameters was performed to
investigate the model uncertainty. The values of the sensitive parameters were varied,
while other climatic variables were kept constant.
6.4
6.4.1
Results and discussion
Surface energy balance
The temporal variation of each component of the surface energy balance of the Indus
Basin for the hydrological year 2007 is presented in Figure 6-2. The values represent the
spatial averages for the whole Indus Basin. The average values attained by the surface
energy fluxes with their standard deviations (SD) are provided in Table 6.1. A high
variability from the mean is observed for the year, especially for net shortwave radiation
(R↓), net radiation (Rn) and sensible heat flux (H). The large variation in climate during
summer and winter is the probable cause of the high SD.
Table 6.1 The minimum, maximum and average values of surface energy fluxes in the
Indus Basin attained during the year 2007. The entire basin is covered and the values
represent average flux densities for periods of eight days including daytime and nighttime.
Fluxes
Minimum
Maximum
Mean
SD
Net shortwave radiation(Wm-2)
95.70
237.50
170.10
45.10
Net longwave radiation(Wm-2)
–75.60
–36.90
–57.80
10.80
-2
Net radiation(Wm )
46.20
177.40
112.30
46.70
Soil heat flux(Wm-2)
–7.10
8.10
0.34
5.20
-2
Sensible heat flux(Wm )
37.20
131.60
79.50
29.80
Latent heat flux(Wm-2)
10.90
57.00
32.40
14.30
Evapotranspiration(mm d-1)
0.39
2.10
1.20
0.50
Evaporative fraction (–)
0.19
0.36
0.28
0.05
Rn is the dominant source of energy for land surface processes. The annual average value
for Rn was 112.3 Wm-2 with a standard deviation of 46.7 Wm-2. The lower Rn values (<80
Wm-2) prevailed during the winter season (DOY 305–361 and 1–65). This low Rn is
probably due to the lower net shortwave radiation. After DOY 66, the Rn continuously
increased to a maximum of 177.4 Wm-2 for an 8 day period (DOY 161). In the summer
season the maximum Rn (>150 Wm-2) was observed during DOY 113–225 while the
average was 163.5 Wm-2. For the same period, Rn values fluctuated considerably with
sudden depressions during DOY 145–225, corresponding to the monsoon season with
103
clouds. Afterwards, the net radiation decreased gradually and reached its minimum values
again in winter.
Figure 6-2 Temporal variation of components of the surface energy equation during 2007
in the entire Indus Basin (116.2 mha). The dashed lines represent 24 days moving average
values.
The dry arid environment of the Indus Basin (annual rainfall is 383 mm) causes the net
radiation to dissipate mainly into sensible heat flux (H). H followed the same temporal
pattern as that of the net radiation and the daily mean value varied between a minimum of
37.2 and maximum of 131.6 Wm-2. The average annual value for a 24-hour period of H
was 79.5 Wm-2. When the soil is moist, a significant part of the energy is dissipated into
evaporation. λE showed two peaks during its annual cycle (Figure 6-2). The seasonal
peaks for the entire Indus Basin correspond to the two agricultural seasons, once in rabi
and once in kharif. The λE varied in the range from 10.9 (during winter with more cloud
covers and lower temperatures) to 57 Wm-2 (during the periods of more canopy cover and
higher temperature), with an annual average of 32.4 Wm-2. The average λE for the entire
Indus Basin coincided with an ET of 1.2 mm d-1, but large variability among LULC
classes occurred. The basin-wide evaporative fraction (Λ) is calculated as 0.28, equivalent
to a Bowen ratio of 2.5. Hence, the amount of heat released into the atmosphere is 2.5
times more than for water vapor, if both are expressed in energy terms.
Soil heat flux (G) is normally ignored when seasonal averages are considered because of
its small scale. However, G can account for a significant portion (3 – 5 %) of the total
104
energy during summer (DOY 113 – 171); indicating that G is transferred from shallow to
deep soil while for the rest of the year, the reverse process occurs.
6.4.2
Actual evapotranspiration estimates
The total transpiration and evaporation in the basin was estimated at 233 km3 yr-1 and 263
km3 yr-1, respectively. The major portion of water was consumed as non-beneficial
evaporation (E), mainly from water-logged soils, dry soils and open water bodies. High
annual ET values occurred on the alluvial plains as depicted in Figure 6-3. Irrigated
agriculture is the major land use class (22.6%) in the basin and is a major consumer of
water. It accounts for the annual ET rates of between 700 – 1200 mm and represents the
middle part of the frequency distribution in Figure 6-4. The highest values (1200 – 1550
mm yr-1) were found in the tail end of the basin: in particular in the right bank of the Indus
River, and southern parts towards the Indian Ocean, in the Sindh Province of Pakistan.
Water-logged soils, rice paddies with shallow phreatic surfaces, and flooded areas
normally occur in these parts of the basin, especially during kharif. Besides higher soil
water content, factors such as higher solar radiation, higher air temperatures, more rainfall,
and cultivation of higher consumptive use crops are the reasons for the higher ET.
Figure 6-3 ETLook estimated cumulative actual evapotranspiration for the hydrological
year 2007 (January to December). The canal command areas for irrigated cropland are
superimposed on the ET map.
105
Figure 6-4 provides the frequency distribution of annual ET. The average ET for all land
use classes was 426 mm yr-1 during 2007. The 2% lowest value was 60 mm yr-1 and the
2% highest value was 1550 mm yr-1.
Figure 6-4 Frequency distribution of the ETLook estimated annual ET in the Indus Basin
at spatial resolution of 1 km×1km for 2007.
A sensitivity analysis was performed to understand the role of topsoil moisture data in the
ET estimation procedure. The results are provided in Figure 6-5(a) and 6-5(b). Two land
use classes and the average climatological condition of the year were used.
Figure 6-5 The response of evaporation, transpiration and evapotranspiration rates to
surface soil moisture (n=100) for two representative land uses (a) Bare soil
(71°22'54.123"E, 28°38'50.042"N) and (b) Irrigated rice-wheat rotation (75°23'53.59"E,
30°40'37.719"N).NDVI values of 0.05 and 0.67 were selected for bare soil and full grown
irrigated rice – wheat land use, respectively.
106
The curves in the Figure 6-5(a) and Figure 6-5(b) display the response of model outputs (E
and T) to variation in surface soil moisture. It is evident from the figure that the ET
responds to surface soil moisture variability. Bare soil shows a fast response in E to
surface soil moisture, while T remains negligibly small by absence of leaves
(NDVI=0.05). E is the dominant flux in the overall ET process of bare soil. The response
is curvilinear with the highest sensitivity occurring between 0.05 and 0.25 cm3 cm-3. The
effects are lower when θ>0.25 cm3 cm-3 prevails. The parameter rsoil and the non-linearity
of Equation (6.11) is one reason for this result. Another explanation is the non-linear
relationship between the resistance and the latent heat fluxes that generally exists (not
shown in this paper). The combined effect yields the S-type curve that is portrayed in
Figure 6-5(a), and to a lesser extent in Figure 6-5(b).
Figure 6-5(b) reveals that, in closed canopies, T dominates E. The net radiation is absorbed
partially by the canopy, and the bare soil surface receives less energy for evaporation. At
an NDVI of 0.67, E increases with increasing topsoil moisture, up to 0.18 cm3 cm-3.
Apparently there is always soil evaporation in rice-wheat rotation systems, which is
confirmed by many other agro-hydrological studies (e.g. Sarwar and Bastiaanssen, 2001;
Ahmad et al., 2002). Canopy transpiration depends entirely upon root zone soil moisture
rather than on the surface soil moisture. Therefore, T shows less sensitivity to surface soil
moisture. The same can be concluded on the ET response to surface soil moisture changes.
The effect of other input parameters on ET is summarized in Table 6.2. The lower and
higher ranges of model input parameters are given, together with ET estimates for the
average climate in the Indus Basin. The values of the input parameters were changed with
specific increments. The sensitivity index (SI) was determined and the parameters were
ranked based on the absolute values. The surface soil moisture appears to be the most
important parameter for describing ET variability, with ET values ranging from 2.3 to 6.3
mm d-1, followed by the coefficient c in rsoil with a range of 2.5 to 6.2 mm d-1. The
measurements of AMSR-E are thus essential for achieving proper ET modeling results,
and form the key input parameter as was suggested.
Table 6.2 Sensitivity of estimates of ET to model parameter values for irrigated rice-wheat
land use. The last column depicts the sensitivity in terms of slope. ∆ is change, and M is
mean.
Parameter
θAMSRE
(cm3cm-3)
rsoil, c
(s m-1)
rs,min
(s m-1)
NDVI
(-)
rsoil, b
(s m-1)
Min
0.05
Input value
Base Max
0.15
0.35
∆out
Mout
SC
SI
0.2
Resulted ET (mm)
Min
Base Max
2.3
5.1
6.3
4.0
4.3
13.3
0.62
∆in
Min
0.30
−10.0
−3.0
5.0
15
7.5
2.5
5.1
6.2
3.7
4.3
0.23
0.40
40.0
80.0
500
460
270
5.6
5.1
3.4
−2.2
4.5
−0.005
−0.3
0.05
0.45
0.67
0.62
0.36
3.9
5.1
5.9
2.0
4.9
3.2
0.24
10.0
30.0
70.0
60
40
5.8
5.1
4.3
−1.5
5.1
−0.025
−0.2
Base=Definition of fixed reference values during sensitivity test.
107
Model parameter sensitivity was investigated using a Monte Carlo simulation experiment
with 1000 pairs of randomly generated input parameters. Based on this experiment the
mean ET for “irrigated rice-wheat rotation” was 3.2 mm d-1 with an SD of 1.7 mm. The
standard error for this distribution was 0.05 mm. A 95% confidence interval was used to
determine the 2.5th and 97.5th percentiles, which ranged between 3.1 and 3.3 mm d-1. This
level of uncertainty reflects that the model generates results with a potential error of 3.4%.
6.4.3
Validation
6.4.3.1 Field measurements
Several field methods to measure ET fluxes can be used to validate the results. AsiaFlux
has erected flux towers in China and India, but not in Pakistan (Mizoguchi et al., 2009).
Therefore, to evaluate performance, ET estimates by ETLook were compared with the
measured values given by PARC (1982) and Ahmad (2002) for 1975-80 and 2000-01
respectively (onwards referred to as “measured” values). Figure 6-6 shows the results of
irrigated crops.
Figure 6-6 A comparison of evapotranspiration in rice, wheat rotation measured by
previous studies, and those estimated by ETLook for 2007 in the Indus Basin.
The correlations were good with an R2 of 0.70, and an RMSE of 163 mm (0.45 mm d-1).
The RE between ETLook and measured ET values ranged from –1.9% to –28% with an
average of –11.5%. The negative RE means ET figures from ETLook were lower than the
field measurements. The regression line fitted through the origin has a slope of 0.89. This
implies that ETLook estimates for 2007 were 11% lower than ET from previous studies.
This difference of 11% is acceptable, considering the climatic differences between the
years, the scale difference between in-situ measurements, and the 1 km remote sensing
pixel size, as well as the uncertainty embedded in field measurements.
Figure 6-7 shows the comparison of annual and seasonal ET from ETLook, from previous
remote sensing and modeling studies (year 1995-96, 2001-02), and from other models
108
(year 2000, 1999-2006) (onwards referred to as “modeled” values). There is a reasonably
good agreement at annual scale, with an R2 of 0.76 and an RMSE of 108 mm yr-1 (or 0.29
mm d-1). The values for the rabi season are reasonable (R2 of 0.60 and RMSE of 47.9
mm). However, the kharif season shows a relatively low R2 (0.54) and a high RMSE of
70.7 mm (or 0.39 mm d-1). The RE ranges between –13% and 18%, with an average of
6.5%. The RE for rabi ranges from –13% to 32%, with an average of 8%, while for kharif,
the range is between –20% and 13%, with an average of –3%. This confirms the difficulty
of modeling ET under cloudy conditions, and supports the inclusion of microwave data in
computational processes, such as in ETLook. Note that there is no bias towards the lower
or higher end of the ET data, and that the average slope is 1.05. Since Figure 6-6 suggests
an underestimation of ET, and Figure 6-7 an overestimation, we believe that the ETLook
performance is satisfactory for this data scarce basin.
Figure 6-7 Comparison of evapotranspiration modeled/estimated by previous studies
conducted during the years 1995-96 (Bastiaanssen et al., 1999), 1995-2000 (Sarwar and
Bill, 2007), 2001-02 (Shakoor et al., 2006; Ahmad et al., 2009) and 1999-2006 (Shakir et
al., 2010) and ET estimated by ETLook for the year 2007 in the Indus Basin.
ETLook has also been validated in regions other than the Indus, e.g. Australia and China.
Some of these unpublished results are presented as a demonstration of the model
performance under different climates and landscapes of ETLook. The National Water
Commission of the Australian Government has provided Australian Water Resources
(AWR) data for the year 2005. The water use data of eight states and 23 jurisdictional
areas are publically available through http://www.water.gov.au/. The ET is computed as
the difference between rainfall and runoff; storage changes and ground water are not
considered. ET values from the water balance were compared against ETLook (Figure
6-8). Considering that the annual values were averaged over a large area, correlation was
reasonable with an R2 of 0.70 and an RMSE of 112 mm (0.31 mm d-1). RE between the
ETLook and AWR ET values ranged from –40% to 36% with an average of – 2.8%.
109
Figure 6-8 Comparison of evapotranspiration estimated by ETLook and estimates
provided by Australian water commission for the year 2005.
In China, ETLook estimated latent heat flux in the year 2009 was compared with flux
tower measurements obtained from the eddy covariance flux measurement station at
Heibei, Qinghai, China (37°36′ N, 101°20′ E) (Figure 6-9). Annual values correlated well
with an R2 of 0.92 and an RMSE of 11mm (0.04 mm d-1). The RE of 9.5% between the
two datasets is satisfactory since there is always a mismatch of scales between 1 km
ETLook pixel estimates and ground measurements (flux tower).
Figure 6-9 Comparison of latent heat flux estimated by ETLook and measured by flux
tower at Heibei, Qinghai, China (37°36′ N, 101°20′ E), for the year 2009. Each point
represents 8-day average value.
ETLook compared with other results and relevant statistics are summarized in the Table
6.3. The mean R2 of comparisons was 0.77 and the RMSE 0.28 mm d-1, while the absolute
110
mean of RE was 6.3%. This level of uncertainty needs to be considered in the presentation
of ET mapping results using ETLook.
Table 6.3 Validation statistics for ETLook results compared with previous studies’ at
various spatial scales.
Annual scale
Study Areas
Spatial Scale
2
R
RMSE (mmd-1)
RE (%)
*
Field
0.70
0.45
–11.50
Indus Basin
Regional†
0.76
0.29
6.50
Basin
–
–
1.00
Field
–
–
–
Australia
Regional
–
–
–
Basin¶
0.70
0.31
–2.80
Field‡
0.92
0.04
9.50
China
Regional
–
–
–
Basin
–
–
–
Absolute Mean
0.77
0.28
6.3
Sources: * PARC (1982); Ahmad (2002). † Bastiaanssen et al., 1999; Shakoor et al. (2006);
Sarwar and Bill (2007); Ahmad et al. (2009); Shakir et al. (2010). ¶AWR, ‡ Flux tower
Heibei, China.
6.4.3.2 Water balance
A map depicting differences in rainfall and evapotranspiration (R–ET) was prepared using
TRMM rainfall data, calibrated by Cheema and Bastiaanssen (2012), and the ET results
from this study (Figure 6-10). It shows areas with net water production (R>ET) and areas
with net water consumption (ET>R). This indicates the value of spatial data to describe
hydrological processes and withdrawals. The pixels that produce water (R>ET) are
discharge areas responsible for streamflow and ground water recharge. These areas are in
the upstream parts of the basin, and are the source of the rivers Indus, Jhelum and Sutlej
that feed the large reservoirs Tarbela, Chashma, Mangla and Bhakra, respectively. Areas
with sparse vegetation and low ET also have higher rainfall than ET and are water
producing areas. Large parts of the Tibetian Plateau comprise such areas. The Rajasthan
Desert between India and Pakistan also exhibits positive values of R–ET, which suggests
groundwater recharge.
Net water consumption areas are generally the irrigated areas, lakes and reservoirs.
Irrigation increases crop ET far beyond the level of rainfed crops. In the Indus Basin, 30.3
% of the total land area is composed of net consumer areas, and 22.6% (26.02 mha) is
irrigated land. The mountain valleys are net water consumers; the valleys receive both
seepage water through the groundwater system and surface water from the higher elevated
mountains, which generally results in shallow water table areas in the vicinity of streams.
111
Figure 6-10 Rainfall-Evapotranspiration (R – ET) difference map of the Indus Basin for
the hydrological year 2007.
The water balance of the irrigated areas covering 26.02 mha was computed to validate ET
results on a large scale (Table 6.4). Total annual groundwater abstraction in Pakistan’s part
of the Indus Basin is given by Qureshi et al. as 51 km3. Chadha (2008) estimated that for
the Indian part of the Indus Basin 18.5 km3 is being abstracted from the groundwater
system. This totals to 69.5 km3 yr-1. The surface water releases into the main canals add up
to 122 and 36 km3 yr-1 in Pakistan and India, respectively. These data on releases from
Tarbela, Mangla, Chashma, Thein, Pong and Bhakra reservoirs, as well as flows into the
main irrigation canals were obtained from Punjab Irrigation Department and Indus Water
Commission, Pakistan. If we assume a conveyance efficiency of 80% that is locally
checked and verified (Habib, 2004; Jeevandas et al., 2008), then 126.4 km3yr-1 will arrive
at the farm gate through the network of canals. Adding the 69.5 km3 of groundwater from
locally operating tube wells, the total amount of water used is about 196 km3. If we take an
on-farm irrigation efficiency of 80% to describe losses of water that is not properly stored
in the root zone, the total ET from irrigation will be 156.8 km3 yr-1. Note that a regional
scale on-farm efficiency for the total irrigation system includes recycling of non-consumed
irrigation water (Perry, 2007). A total irrigation efficiency of 64% (0.8×0.8) for one
contiguously irrigated alluvial plain is realistic. It can however also be 60% or 70%. The
rainfall over the irrigated area is 117 km3 yr-1. The net rainfall infiltrated into the soil –
112
after runoff and percolation losses - and available for uptake by roots is 94 km3 yr-1
(assuming 80% efficiency). The total ET for the irrigated land on the basis of water
balance is 94 + 156.8 = 250.8 km3 yr-1, or 964 mm yr-1. ETLook results provided an
estimate on the basis of the energy balance as being 254 km3 yr-1, or 974 mm yr-1. While
this is a difference of 1 % only, we conclude that the results are congruent and within
acceptable ranges that are usually related to water balances.
Table 6.4 Water balance for the irrigated areas in the Indus Basin during the hydrological
year 2007.
Annual
rainfall
R
km3
mm
6.5
117
451
From
surface
water
158
607
Irrigation
(IRR)
At
From
farm ground
gate
water
126.4
69.5
486
267
Total
(column
4+5)
196
753
Evapotranspiration
(ET)
IRR
R
Total
156.8
603
94
361
250.8
964
ET
ETLook
254
974
Summary and conclusions
The first requirement for an operational ET monitoring system is that the satellite data
must be available at all times. Microwave satellite data are operationally provided – even
under all weather conditions – and their growing number of standard databases form an
attractive source for developing ET models. ETLook can assess the spatial and temporal
(daily, 8-day, or monthly) patterns of the surface energy balance and actual
evapotranspiration. Computing E and T separately, on the basis of the energy balance, has
the advantage that complex transient moisture flow computations in the unsaturated
topsoil can be circumvented. The novelty of this paper is a doable computational method
for non-beneficial E and beneficial T that can be applied under conditions of persistent
overcast skies, and in data scarce environments. The sensitivity analysis revealed that the
surface soil moisture is the most important parameter for describing ET variability.
Variability of surface soil moisture revealed that the ET values for rice-wheat rotation
system on an average day ranged between 2.3 to 6.3 mm d-1, followed by the coefficient c
of soil resistance, with a range of 2.5 to 6.2 mm d-1.
Good agreement was attained between ETLook and previously conducted field
measurements and remote sensing studies. R2 varied between 0.70 and 0.76 at annual time
scale (RMSE: 0.45 and 0.29 mm d-1 respectively). Tests in Australia and China provided
similar agreements based on watershed measurements. The water balance of 26.02 mha of
irrigated land is congruent and matches generic data on surface water and groundwater
supply. There are discrepancies in timescales shorter than a year. However, no bias was
evident towards the lower or higher end of the ET values. The observed errors could be
due to the meteorological differences between the years of study. The determination of
wind speed and air humidity needs more attention in future studies. Better quality soil
maps will also improve the quality of the ET results.
The average value for latent heat flux in the Indus Basin is 32 Wm-2, which corresponds
with an ET of 1.2 mm d-1 (426 ± 14.5 mm yr-1). The average value for rainfall is 383 mm
113
yr-1. Over-exploitation and negative storage changes of water occur at the basin scale (ET
>R). The negative change in storage can be ascribed to reduced volumes of water stored in
reservoirs and aquifers. Retirement of glaciers also contribute to water storage changes.
The power of having access to daily soil moisture data from passive microwave
measurements onboard satellites is at the same time limited by the low resolution of
AMSR-E surface soil moisture pixels (25 km). Several methods exist to deal with the
downscaling of soil moisture, but the best method that is doable under a wide range of
conditions still needs to be found. More sophisticated solutions on downscaling can be
gleaned from topographic information (e.g. height above drain, distance to drain,
accumulated upstream drainage area) and soil properties (infiltration capacity, water
holding capacity, drainage capacity). It is expected that satellites with synthetic aperture
radar will provide high-resolution soil moisture values in an operational context in the near
future, in addition to the thermal data that are already used for routine mapping of soil
moisture under clear sky conditions. The analytical relationships between topsoil and subsoil moisture need improvement and more testing.
This analysis was conducted for a one year cycle only, to raise confidence in using the first
version of ETLook algorithm (ETlook 1.0). Future analysis with longer time series is
recommended, since shorter time series may be of low significance. Despite the limitations
mentioned, the current paper has demonstrated that the ET results show potential for
determining water depletion in ungauged basins.
114
7 Spatial quantification of groundwater abstraction for
irrigation in the Indus Basin using pixel information, GIS
and the SWAT model
Chapter based on: Cheema, M.J.M., Immerzeel, W.W. and Bastiaanssen, W.G.M. 2012. Spatial
quantification of groundwater abstraction for irrigation in the Indus Basin using pixel
information, GIS and the SWAT model. Groundwater, (under review).
7.1
Introduction
Quantification of groundwater abstraction, especially in arid regions where recharge is
genuinely small, is of prime importance for sustainable basin scale water resources. Rapid
population growth and increased irrigation development for food security has resulted in
exhaustive groundwater abstractions in many alluvial plains (e.g. Foster and Chilton,
2003; Shah et al., 2007). Wada et al. (2010) created a global map of groundwater
abstractions in 2010, which indicates several areas with abstractions exceeding 100 mm yr1
. Siebert et al. (2010) developed a global inventory on groundwater which estimates that
43% of the total consumptive irrigation water use is met through groundwater.
Groundwater abstractions are temporally episodic and spatially variable and depend upon
the crop irrigation needs, surface water availability and water quality. The spatial
variability in groundwater availability and water requirement by crops complicate the
quantification of abstractions. The Indus Basin is a typical example showing high
variability in land use, climate, canal water availability, soil types and irrigation practices
without any regulation in place to measure the groundwater abstraction.
Irrigated agriculture is common land use in the trans-boundary Indus Basin covering 23%
of the total area. Groundwater is utilized solely or in conjunction with surface water to
augment the insufficient and unreliable surface water supplies. Different studies (e.g. Scott
and Shah, 2004; Sarwar and Eggers, 2006) have estimated that about 40 to 50% of
irrigation needs of the basin are met through groundwater abstraction. Arshad et al. (2008)
has estimated that the groundwater abstraction is up to 60% in the areas having rice –
wheat rotation land use. The continuous abstractions, in high quantities, can adversely
affect the overall water balance when the average value consistently exceeds the recharge
over a long period. Therefore, accurate information on spatial groundwater abstraction and
depletion is urgently required to support development of management plans. The use of
tabular values per district or province is no longer sufficient for achieving progress in
controlling over-exploitation of the aquifers. Policy makers can take more effective actions
if groundwater activities are expressed by means of pixels. The benefits of pixel-based
information include the availability of information in terms of geographical coordinates,
coverage of a discrete land area, quantified abstraction rate, and the identification of land
owners. This paper employs new methods to obtain groundwater information from pixels,
using satellite measurements, GIS systems and hydrological models.
The groundwater use estimation is normally carried out using the tubewell utilization
factor technique or water table fluctuation methods (Maupin, 1999; Healy and Cook, 2002;
Qureshi et al., 2003). These methods become less suitable when applied at basin scale due
115
to the poor spatial density of the point measurements. Alternatively, abstraction data on
groundwater use can be derived from hydrological models. The success of these models
depends primarily on availability of comprehensive input data and how well the models
are calibrated (Zhang et al., 2008). Long-term time series data sets with high spatial detail
are difficult to obtain in spatially heterogeneous basins with a limited gauging network
(Sivapalan et al., 2003). Extreme spatio-temporal variability in precipitation,
evapotranspiration in combination with low density surface and groundwater point
measurements makes models prone to errors. Measurements at few stations and their
spatial extrapolation for the entire basin may yield unreliable estimates of water use. The
uncertainties associated with the measured input data may also lead to biases in the model
estimations (Srinivasan et al., 2010).
The use of remote sensing techniques in combination with spatially distributed
hydrological models has shown great potential to overcome these difficulties. Many
researchers, including Houser et al. (1998), Boegh et al.(2004), Bastiaanssen et al. (2007),
and Immerzeel et al. (2008a) have successfully used remote sensing to parameterize
hydrological models. Remotely sensed evapotranspiration has also been used successfully
in calibrating hydrological models (e.g. Droogers and Bastiaanssen, 2002; Immerzeel and
Droogers, 2008; Jhorar et al., 2011).
In the present study we develop, for the first time, a detailed Soil Water Assessment Tool
(SWAT) model application that encompasses the entire transboundary Indus Basin. The
SWAT model is forced and parameterized, using remote sensing derived datasets of
elevation, land use, temperature, and precipitation. We then calibrate the SWAT, at the
highest possible spatial detail, using estimates of actual evapotranspiration (ET) based on
the ETLook algorithm as described in Cheema et al. (2012). We then use the calibrated
model to spatially estimate the total irrigation water supply at the farm gate of irrigated
areas following the principles similar to Droogers et al. (2010b). The surface water
diverted at the head of the canal command areas has been integrated with this dataset to
isolate the spatial distribution of irrigation by 1 km pixels of gross groundwater
abstractions. The main objective of this paper is to explain, demonstrate and validate the
methodology to determine groundwater abstractions using 1 km pixels. It identifies the
hotspot areas with a discretization of 100 ha and consequently groundwater-pumping
activities are no longer a hidden piece of information.
7.2
7.2.1
Material and methods
Study area
The Indus Basin lies between latitude 24°38′ to 37°03′ N and longitude 66°18′ to 82°28′ E
located in four countries (Figure 7-1). The lifeline of the Indus Basin is the Indus River
that traverses China, Afghanistan, India and Pakistan, moving from upstream to the
downstream end of the basin. The total size of the basin is 116.2 million ha (mha). In total
53% of the area of the basin is located in Pakistan. The area in India is 33% followed by
China and Afghanistan with 8% and 6%, respectively. The elevations range from 0–8600
m above mean sea level (a.m.s.l). The basin exhibits complex hydrological processes due
to variability in topography, rainfall, land use, and water use. The average annual rainfall
116
varies from less than 200 mm in the desert area to more than 1500 mm in the north and
north-east of the basin. The thirty year (1961 – 90) average reference crop
evapotranspiration (ETo) varies between 650 mm in the northern parts and 2000 mm in the
southern desert areas of the basin.
Water is diverted from the Indus River and its major tributaries (Jhelum, Chenab, Ravi,
Beas and Sutlej) through a network of canals to irrigate the agricultural lands. The main
reason for this diversion is that the rainfall is inadequate to fulfill crop water requirements.
However, the availability of canal water is unreliable and that has motivated farmers to
augment shortages in surface water by groundwater resources (Shah et al., 2000).
Two agricultural seasons kharif (May to October) and rabi (November to April) are in
practice. The main crops cultivated are wheat, cotton, rice, fodder, sugar cane and fruit
orchards. Vegetable are also raised in some areas. Perennial sugarcane and seasonal fodder
is also grown in tracts. Wheat is the major crop grown in rabi. Rice and cotton are the
major crops of kharif season.
Figure 7-1 Location of the Indus Basin and land use scheme used in the SWAT model. The
codes are explained in Table 7.1.
7.2.2
Soil and Water Assessment Tool
The Soil and Water Assessment Tool (SWAT) is a process based distributed hydrological
model which provides spatial coverage of the integral hydrological cycle including
atmosphere, plants, unsaturated zone, surface water, and groundwater. A comprehensive
description of the model can be found in literature (e.g. Arnold et al., 1998; Srinivasan et
al., 1998; Neitsch et al., 2005), however, for the convenience of our reader, we summarize
the SWAT model in the following paragraphs.
117
SWAT provides continuous simulation of evapotranspiration, percolation, return flow,
storage change, surface runoff, channel routing, transmission losses, crop growth and
sediment transport (Kannan et al., 2011). We selected SWAT because it represents a
simple groundwater reservoir that acts as an interface between soil moisture in the
unsaturated zone, groundwater storage in the saturated zone and surface water systems.
The latter is essential for water exchanges and for understanding recycling mechanisms of
non-consumed water. The spatial water balance of the unsaturated zone reads as:
7.1
where ΔSus is the change in storage of the unsaturated zone (mm), RSWAT is the amount of
precipitation (mm), IRRSWAT is the amount of total irrigation applied (mm), Qsurf is the
amount of surface runoff (mm), ETSWAT is the actual evapotranspiration (mm), Qlat is the
amount of lateral flow through the unsaturated zone (mm), Qperc is the amount of
percolation (mm) and Cr is the capillary rise (mm). Net Groundwater Use (NGU) is
introduced by Ahmad et al. (2005) as the difference between vertical exchanges
(abstraction, recharge, capillary rise), and NGU can be computed from this data.
SWAT computes daily ETo and potential plant transpiration (Tp) according to
meteorological input data and crop coefficients based on Penman-Monteith method
(Monteith, 1965). Daily crop height and leaf area index (LAI) are controlling aerodynamic
and canopy resistances and are used in calculating Tp. Potential soil evaporation is an
exponential function of ETo and the soil cover, which is reduced during periods with high
plant water use. Actual soil evaporation is limited by the soil water content (θ) and is
reduced exponentially when θ drops below field capacity.
The potential plant water uptake can be defined as follows for calculating the actual plant
transpiration:
7.2
where wup,z (mm) is the potential plant water uptake from the soil surface to a specified
depth from the soil surface on a particular day, Tp(mm) is the maximum plant transpiration
on a given day, βw (-) is the water use distribution parameter, z is the depth from the soil
surface (mm), and zroot is the depth of root development in the soil (mm). Actual plant
water uptake equals actual plant transpiration and shows exponential reduction as θ drops
below field capacity. Hence, soil moisture regulates actual transpiration fluxes, and
deviations of soil moisture will be propagated into deviation in actual transpiration. Actual
evapotranspiration modeled by SWAT (ETSWAT) is the sum of interception, actual soil
evaporation, and actual plant transpiration.
Soil moisture in the root zone is a function of irrigation water supply at the farm gate.
Irrigation has been modified consecutively until ETSWAT on the water balance of Eq (7.1)
was matching ET from ETLook. Some major principles of ETLook will be explained in
the latter section. This technical approach provides a vehicle to infer the actual irrigation
applications in a realistic way, without the need to construct large databases on irrigation
water distribution. Santos et al. (2010) used a similar approach for irrigation systems in
Spain.
118
The water balance of the saturated zone, i.e. -shallow aquifer- can also be computed using
SWAT. The net groundwater depletion (DEPgw: amount of water leaving the shallow
aquifer) can be estimated for the irrigated areas using information of canal water losses
(LOSScw) as:
7.3
where Qgw is the return flow from the shallow aquifer towards the river (mm). Note that
the capillary rise is considered as a component of the irrigation water supply. IRRgw is the
gross groundwater that is abstracted from the shallow aquifer.
The SWAT model subdivides the Indus Basin into sub-basins, which are further divided
conceptually into hydrological response units (HRU). HRUs are based on unique and
homogenous combination of land use and soil type. We delineated 132 sub-basins with
average area of 8000 km2 and 2459 HRUs in the Indus Basin. We identified 489 HRU as
irrigated where conjunctive use is practiced. We used the average data on radiation, wind
speed, relative humidity, air temperature, and rainfall for each sub-basin. We fed the above
mentioned data into SWAT for the computation of Tp and wup,z. We used a model spin up
period of two years to initialize the model, in particular soil moisture.
7.2.3
Data
We obtained the rainfall data in SWAT as 25 km grids from the Tropical Rainfall
Measurement Mission (TRMM) as described by Huffman et al. (2007). The products 3B42
(daily) and 3B43 (monthly) were collected for the year 2007. Daily 3B42 data were
aggregated per month and a correction factor for each month was established using
calibrated 3B43 monthly data (Cheema and Bastiaanssen, 2012). Using these correction
factors, daily grids of rainfall (RTRMM) were generated from January 1, 2007 to December
31, 2007. The rainfall that was deposited below a threshold temperature was classified as
snow.
Solar radiation has been computed from the extra-terrestrial radiation in association with
an atmospheric transmissivity. The atmospheric transmissivity was inferred from optical
depth information obtained at 1 km pixel resolution Moderate Resolution Imaging Spectro
radiometer
(MODIS)
cloud
product
(MYD06_L2)
downloaded
from
https://wist.echo.nasa.gov/~wist/ api/imswelcome/.
The spatially distributed meteorological data for maximum (Tmax) and minimum (Tmin) air
temperature, relative humidity (RH) and wind speed (U2) was prepared using
meteorological distribution model (Daymet) described by Thornton et al.(1997). The
meteorological station data was obtained from 65 meteorological stations under the aegis
of the Pakistan Meteorological Department (PMD). Weather station data for India, China
and Afghanistan were obtained from the National Oceanic and Atmospheric
Administration (NOAA) National Climatic Data Center (NCDC). The NCDC collects
meteorological data from real time reporting stations worldwide in agreement with World
Meteorological Organization (WMO) regulations (ftp://ftp.ncdc.noaa.gov/pub/data/gsod/).
Data from 16 stations with complete datasets were downloaded. Air temperature, relative
humidity and wind speed data from 81 stations, collected at standard height of 2 m, were
119
obtained. The gridded daily datasets were aggregated per sub-basin, which provided 132
hypothetical meteorological stations with uniform distribution.
A detailed land use and land cover (LULC) map developed by Cheema and Bastiaanssen
(2010) was used to infer information on different LULC classes in the Indus basin. Twenty
seven LULC classes were classified. Those classes were clustered into 21 land use classes
based on SWAT land use library. The details of seven irrigated land use classes including
growing season and irrigation depths are provided in the Table 7.1. The actual dates vary
spatially and temporally. For example, wheat crop is sown between 1–30 November and
irrigation depths may vary from 45 mm to 105 mm, while number of irrigations may vary
from 3 to 5. These irrigation practices, provided by Pakistan Agricultural Research
Council, PARC (1982) and Ahmad (2009), were adopted for initiation of the SWAT
model.
The FAO digital soil map of the world (FAO, 1995) was used to derive soil properties
with the aid of pedo-transfer functions (Droogers, 2006). Forty eight different soil units
were found in the basin and the alluvial plains are predominantly characterized by
vertisols and the steeper slopes by fluvisols. A GTOPO30 – DEM was used in watershed
delineation and defining streams. A stream network developed by International Water
Management Institute (IWMI) was used in stream delineation process where topography
was relatively flat. The surface water supplies at canal heads for various CCA were
obtained from Punjab Irrigation department (PID), Water and Power Development
Authority (WAPDA) and Indus Water Commission (IWC), Lahore, Pakistan.
Table 7.1 Land use and land cover classes in the Indus Basin. The cropping period and
number of irrigation with depths are also provided for irrigated land use classes.
SWAT
Land use class
Area Growing
Cropping
Irrigation No
Code
(Depth,mm)
(%)
season
period
15 May – 15 Nov 5 (120)
Irrigated mixed
AGI1
3.6 kharif
15 Nov – 30 Apr
cotton,wheat
rabi
4 (75)
rotation/orchards
15 May – 15 Nov 5 (120)
Irrigated mixed
AGI2
4.3 kharif
15 Nov – 30 Apr
cotton,wheat
rabi
4 (75)
rotation/sugarcane
15 Jun – 30 Oct
Irrigated rice,wheat
AGI3
8.3 kharif
15 (100)
15 Nov – 30 Apr
rotation
rabi
4 (75)
15 Jun – 30 Oct
Irrigated mixed rice,wheat
AGI4
2.2 kharif
15 (100)
15 Nov – 30 Apr
rotation/cotton
rabi
4 (75)
01 Aug – 30 Oct
Irrigated fodder,wheat
AGI5
2.3 kharif
6 (75)
15 Nov – 30 Apr
rotation
rabi
4 (75)
15 Jun – 30 Oct
Irrigated rice,fodder
AGI6
1.7 kharif
15 (100)
15 Jan – 30 Apr
rotation
rabi
6 (75)
15 Jun – 30 Oct
Irrigated mixed rice,wheat
AGI7
0.2 kharif
15 (100)
15 Nov – 30 Apr
rotation/sugarcane
rabi
4 (75)
Forests/cropland alpine
AGR1
2.5
Rainfed crops wheat/grams AGR2
1.2 rabi
Rainfed crops mixed
AGR3
1.5 kharif
cotton,wheat
rabi
120
rotation/fodder
Rainfed crops general
Rainfed crops and woods
Bare soil
Forests deciduous alpine
Forests evergreen broadleaf
Forests evergreen
needleleaf
Pastures deciduous alpine
Pastures deciduous lowland
Pastures evergreen
Savanna deciduous
Savanna evergreen closed
Savanna evergreen open
Very sparse vegetation
Snow and ice permanent
Snow and ice temporary
Urban and industrial
settlements
Water bodies
7.2.4
AGR4
AGR5
BARE
FRSD
FRSD
FRSE
10.1
3.3
6.3
3.2
0.5
3.8
PAST
PAST
PAST
RNGB
RNGB
RNGB
RNGE
SAIP
SAIT
URBN
6.7
6.5
3.0
11.1
1.9
2.9
1.9
3.6
4.8
1.6
WATR
1.0
ETLook
The ETLook algorithm uses a two layer Penman–Monteith equation (Monteith, 1965) by
dividing each pixel of the image into bare soil and canopy to infer evaporation (E) and
transpiration (T), respectively(Pelgrum et al., 2010) . The Penman–Monteith equation for
E and T can be written as:
7.4
7.5
where E represents evaporation and T represents transpiration in Wm-2. Δ (mbar K-1) is the
slope of the saturation vapor pressure curve which is a function of air temperature (Tair,
°C) and saturation vapor pressure (es, mbar); Δe(mbar) is vapor pressure deficit, which is
the difference between the saturation vapor content and the actual vapor content; ρ (kg m3
) is the air density and cp is specific heat of dry air =1004 J kg-1 K-1; γ (mbar K-1) is the
psychometric constant; Rn,soil and Rn,canopy are the net radiations at soil and canopy
respectively; rsoil and rcanopy are resistances of soil and canopy, while ra,soil and ra,canopy are
aerodynamic resistances for soil and canopy. All resistances are in s m-1. The resistance for
E is based on top soil moisture measurements from satellite data. The resistance to T is
based on root zone soil moisture that is estimated from topsoil moisture and the presence
of photosynthetically active vegetation. Total ET is the sum of T and E and the units can
121
be converted to mm day-1. Details of the ETLook algorithm can be found in Pelgrum et al.
and Cheema et al. (2012). In the current study, the ET data is available with an eight day
interval for the period of one calendar year from January 1st to December 31st, 2007.
7.2.5
Model calibration procedure
The calibration of the SWAT model was performed by comparing SWAT modeled ET
(ETSWAT) with spatially observed ET (ETETLook) for all HRUs. As in complex distributed
hydrological model with numerous parameters with a high spatial and temporal
heterogeneity, conventional stream flow calibration has a large risk of equifinality
problems (Beven, 2006). Moreover, it becomes ineffective in basins, such as the Indus,
where stream flow is under human control (Immerzeel and Droogers, 2008).
A stepwise heuristic iterative approach was therefore adopted to perform calibration by
adjusting the key soil and groundwater parameters. A number of important model
parameters, which have a large influence on ET, were used in the model calibration. The
sensitive parameters that have control on E and T fluxes were identified from the previous
literature (e.g. Immerzeel and Droogers, 2008; Immerzeel et al., 2008a; Githui et al., 2011;
Kannan et al., 2011). The soil water holding capacity (Ф), capillary rise (Cr), depth of the
evaporation front (Ψ) and the relative water uptake by plant roots as a function of soil
moisture (wup,z) have been calibrated. Their allowable ranges were bound to 0.06 to 0.60
mm/mm for Ф, 0.02 to 0.9(Pelgrum et al., 2010) for Cr, 0.01 to 1.0 for Ψ and 0.01 to 1.0
for wup,z. Forty eight soil types with two layers resulted in 96 different parameters for Ф to
be optimized. Ψ, Crand wup,z coefficients were optimized for each HRU where irrigation
was absent. Most parameters were optimized per HRU of each land use class to capture
the spatial heterogeneity because land use information is available at relatively detailed
level compared with the soil type information. Default values of these parameters were
adopted for base run and the implemented adjustments were constrained by the ranges of
parameters suggested by Neitsch et al.(2005). Three common statistical indicators, as
described by Hoffmann et al.(2004), were used to quantify the achieved level of
calibration and to evaluate the SWAT model’s overall performance .The Nash-Sutcliffe
model efficiency (NSE) (Nash and Sutcliffe, 1970), Pearson’s correlation coefficient (r)
and percent bias (Pbias) between modeled and observed ET were determined, which are
given as:
7.6
7.7
7.8
where, O represents the observed (ETETLook) and M represents the modeled (ETSWAT) ET. Ō
is mean observed and n is the total number of observations. NSE ranges between −∞ and
1, with NSE = 1 being the optimal value. The Pbias reveals to which degree the modeled
122
value is smaller or larger than the observed values given in percentage. Small values of
Pbias are preferred.
7.2.6
Pixel based groundwater abstraction data
The pixel information on actual evapotranspiration using ETLook (ETETLook) and rainfall
(RTRMM) from TRMM is valuable additional spatial information. This information can be
used to infer the total irrigation water supply at the farm gate for each pixel (IRRRS), when
being integrated with HRU fluxes obtained from SWAT calculations of Eq. (7.3).
Assuming that capillary rise and storage changes to be part of applied water, the analytical
expression becomes:
7.9
Irrigation water, diverted to main canals irrigating a specific canal command area (CCA),
was aggregated to monthly and annual irrigation volumes. The resulting vector maps of
canal water supplies for each CCA were prepared. The supplies were then converted into
depths by dividing over the area of each CCA. The map of irrigated land use was overlaid
with the map of CCA to identify the irrigation command areas matching with each HRU.
The result is a canal irrigation vector map (IRRcw) that has a spatial refinement as
compared to the HRU. The overlay helped to partition the total irrigation water supply
(IRRRS) into canal water use (IRRcw) and finally gross groundwater abstraction from
shallow aquifer (IRRgw) (Figure 7-2):
7.10
The annual depths of canal water vary from 200 mm to 1700 mm per diversion.
Conveyance efficiency of 70% was considered for canals in Pakistani part of the Indus
Basin (Habib, 2004; Arshad et al., 2005; Kreutzmann, 2011) and 80% for canals in Indian
part of the Indus Basin (Bastiaanssen et al., 1999; Kroes et al., 2003; Jeevandas et al.,
2008).
Figure 7-2 Schematic diagram showing data sources to infer groundwater abstraction
information at 1 km pixel.
123
7.3
7.3.1
Results and discussion
Model calibration
The model performance was evaluated using three statistical indicators namely NSE,
Pearson’s correlation r and Pbias. The performance was assessed at sub-basin and HRU
levels. At sub-basin level, NSE of the calibrated model was 0.93, while under uncalibrated conditions the NSE was 0.52. Improvement in Pearson’s correlation was
observed from 0.78 to 0.97, suggesting a strongly improved correlation between ETSWAT
and ETETLook when model parameters where adjusted according to the existing ET layers
from the energy balance. The Pbias resulted in -0.4, which is very low as compared to 17.3 (base run) indicating no systematic under or over prediction of ET is observed at subbasin level. Figure 7-3 shows the correspondence between the modeled and observed ET at
sub-basin scale.
Figure 7-3 Comparison between modeled (ETSWAT) and observed (ETETLook) actual
evapotranspiration for 132 sub-basins in the Indus Basin.
Figure 7-4 shows the results for 489 HRUs that contain irrigated land only. The NSE,
Pbias and Pearson’s r were 0.93, -2.3 and 0.97. This level of agreement shows that SWAT
can produce ET values from the soil water balance that is very similar to the ET of
irrigated crops as interpreted from satellite images. Figure 7-5 shows the spatial patterns of
the annual sum of ET. The spatial pattern of ET modeled by SWAT was in good
agreement with ETETLook. However, some local differences were observed and the reason is
that ETSWAT results show a larger within land use variation.
124
Figure 7-4 Comparison between modeled (ETSWAT) and observed (ETETLook) actual
evapotranspiration for 489 irrigated HRUs.
Figure 7-5 Cumulative ET from 1st January 2007 to 31st December 2007 derived with
ETLook (left) and modeled with SWAT (right). Both figures display aggregated data per
HRU for the sake of compatibility.
Figure 7-6 provides more insight in the temporal ET patterns. The monthly ET shows good
agreement between the modeled monthly ETSWAT and ETETLook with a correlation coefficient
125
(R2) of 0.87. November and December show a low ET rate due to lower solar altitudes and
low ambient temperature. The warm atmosphere and large rainfall amounts due to
monsoon system are the reason for peak ET rates during July. The strong reduction in ET
during the month of May –when land is prepared for the summer crop– is picked up well
by SWAT.
Figure 7-6 Comparison of monthly ETETLook against ETSWAT for all 489 irrigated HRU’s
during 2007.
It is notable that in the months of September and October ETETLook is higher than the
modeled values while the reverse is observed for the months of June and July. One of the
main reasons is that most of the fields became fallow due to the harvest of kharif crops.
Normally harvesting starts at different dates at different locations depending upon the crop
maturity. However, in SWAT parameterization each land use was assigned with a single
date of harvesting. The moisture retained especially in paddy fields, contributed to
evaporation thus causing higher ETETLook than the ETSWAT. In the contrary, during the
months of June and July, ETSWAT has shown higher values. These two months corresponds
to the monsoon months with higher rainfall and considerable irrigation is supplied to the
crops especially to rice and cotton. It suggests that ETSWAT over estimates ET during these
months.
Moreover, during model simulations, the specified irrigation dates for a particular crop
were the same for the HRUs representing that crop. It was assumed that the total area
under that particular crop is irrigated on the specified date and specified depths. In reality,
the irrigation of a particular crop is completed over a period of time depending on the
farmer’s rotation for canal supplies. Another source of temporal discrepancy could be that
during the simulation periods canal water supply to a certain CCA is taken as constant for
126
the entire command area that in reality varied based on the distance from the canal head.
All these factors can be the cause of the deviations.
Overall it is concluded that the calibration on actual ET is highly satisfactory given the
high correlation, NSE and low biases. The Indus SWAT model is calibrated at the highest
possible spatial detail (HRU level) and the temporal ET patterns are also simulated with
reasonable accuracy. The adopted calibration strategy is effective and outperforms earlier
work in this field (Immerzeel and Droogers, 2008).
7.3.2
Spatial patterns of water supply and consumption
The spatial distribution of total irrigation estimated by applying pixel information (IRRRS),
surface water supplied at farm gate (IRRcw), percolation to aquifer (Qperc), gross irrigation
from groundwater (IRRgw) and related groundwater depletion (DEPgw) are presented in
Figure 7-7 (a – e). The total canal water available at the farm gate for each canal command
is estimated at 113 km3 (or 434 mm) (Figure 7-7(a)). This amount is computed from the
reservoir releases and reported conveyance losses. Canal water available at farm gates
varies from 200 to 900 mm yr-1. This spatial variability in canal supplies is due to the nonperennial system and variability in water released from the reservoirs. The highest rate of
IRRcw is observed in lower Indus especially in Sindh province. These areas are suffering
from salinity and a deteriorated groundwater quality (Qureshi et al., 2010b) owing to the
high intensity of rice cultivation on this low laying river plain areas.
The total irrigation estimated by pixel information (IRRRS) using Equation 7.9 is 181 km3
(696 mm). Total applied water varies between 200 to 1400 mm yr-1 in the irrigated areas
across the basin (Figure 7-7(b)). The irrigation application is found to be largest (700 –
1400 mm yr-1) in Jacobabad, Shikarpur and Larkana districts of northern Sindh. Higher
values are also found in Noshehroferoz and Nawabshah districts in southern Sindh.
Narowal and Gujranwala districts in northern Punjab, Jhang, Toba Tek Singh and
Pakpattan districts in central Punjab and Khanewal, Multan, Lodhran and Vehari districts
of southern Punjab have received higher total irrigation. Districts of Amritsar, Ludhiana,
Jalandhar, and Ferozpur in Indian Punjab and Patiala district of Haryana also have
received higher irrigation rate. The reason of this higher irrigation is the large-scale
cultivation of high water consumptive crops like rice, sugarcane, cotton etc.
Aquifer recharge (Qperc) ranges between 10 – 600 mm during the year as depicted in
Figure 7-7(c). A percolation of 71 mm from irrigated fields (especially in irrigated rice –
wheat land use) and high rainfall during monsoon are the sources of this recharge. The
losses from canals (LOSScw = 144 mm) also contributes to the aquifer. This LOSScw does
not indicate that water is lost permanently from the system but it represents non-consumed
water that can be potentially recycled.
127
(a)
(b)
(d)
128
(c)
(e)
Figure 7-7 Spatial maps of (a) Canal water supplies (b) Irrigation estimated using remote
sensing products (c) Percolation (d) Gross groundwater abstraction and (e) Net
groundwater depletion
The canal water supplies are not sufficient to meet the crop water requirements and hence
do not match with IRRRS. The deficit is met through groundwater irrigation and Figure
7-7(d) shows gross groundwater abstraction rates (IRRgw) for each pixel estimated by using
Eq 7.10. The data shows that, on annual basis, an amount of 300 to 900 mm is abstracted
from the aquifers for irrigating crops. The highest values for IRRgw are observed in middle
and northeastern parts of the basin. These areas contain relatively good quality
groundwater resources (Arshad et al., 2007) and are located in the Punjab province of
Pakistan and the Indian state of Haryana.
129
The largest groundwater abstractions in Pakistan occur in the province of Punjab. The
IRRgw in the districts of Multan, Khanewal, Pakpattan, Vehari and Lodhran covering lower
Bari doab (area between Ravi and Sutlej) ranges between 300 – 700 mm. The average
abstraction in lower Bari doab is 400 mm. The middle and lower Rechna doab - area
between Chenab and Ravi rivers - also shows similar trends and IRRgw ranges between 200
– 600 mm. In the upper Rechna doab, fragmented pockets of high IRRgw (500 -700 mm)
are observed especially in the districts of Narowal, Gujranwala and Sialkot. The reason of
this high groundwater abstraction is that the rice is extensively cultivated but canal water
do not suffice. IRRgw is also observed in pockets of the Sargodha district in Chaj doab
(area between Jhelum and Chenab rivers). Groundwater is mostly saline here but
fragmented lenses of fresh water are available. Conjunctive use of groundwater with
surface water is a normal practice in these areas. In Sindh province, groundwater
abstraction is fragmented with significant groundwater abstractions occurring in district of
Larkana, Jacobabad and Shikarpur. These areas have higher annual ET because of rice
cultivation or high cropping intensities. Groundwater recharge by percolation from fields
and canals can be recycled which results in conjunctive use of groundwater and surface
water in these northern parts of Sindh province (Siebert et al., 2010).
The districts of Jalandhar, Phagwara, Ludhiana and Bathinda of Indian Punjab are
vulnerable to extensive groundwater pumping. The value ranges between 400 to 900 mm.
Irrigated rice,wheat rotation is the dominant land use that requires extensive irrigation to
meet crop water demand. The surface water supplies are not sufficient to meet the needs
therefore the deficit is met through groundwater. Large number of small capacity
tubewells is installed to pump groundwater. According to Shankar et al.(2011), tubewells
density in the Punjab and Haryana state is 27 and 14.1 tubewells per km2 in 2001.and the
number is increasing.
The information on groundwater abstraction in different LULC is important to identify the
opportunities for saving water or relocation of water. The analysis for irrigated LULC is
provided in the Table 7.2 which shows that the maximum groundwater was abstracted
(64% of total groundwater abstraction) in the “irrigated rice,wheat rotation” land use. It is
53% of the total irrigation water supplied to this land use. The groundwater supply in
“irrigated mixed cotton,wheat rotation/orchards” is 43% of the total irrigation water
supply. Groundwater contribution to the total irrigation supplies in the “irrigated mixed
cotton,wheat rotation/sugarcane” and “irrigated rice,fodder rotation” land uses is 23%. The
results show that the total irrigation supplies in the irrigated areas of the Indus Basin is 181
km3, and this extra water should originate from irrigation practices. An amount of 68 km3
originates from groundwater, while the surface water contribution is 113 km3. This
diagnosis suggests that groundwater supplies 68/181 or 38% of the total water applied at
the farm gate. The result is in agreement with the 40 to 50% groundwater contribution
reported by Sarwar and Eggers (2006).
130
Table 7.2 Amount of water supplied through various sources in irrigated land uses of the
Indus Basin during 2007.
Irrigated land uses
Area
ET
R
Total
Surface
Ground
Irrigation
water
water
supply
supply
supply
mm
mm
mm
mm
mm
(mha) (km3)
(km3)
(km3)
(km3)
(km3)
Irrigated mixed
3.99
1130
474
783
448
334
cotton, wheat
(45.1)
(18.9)
(31.3)
(17.9)
(13.4)
rotation/orchards
Irrigated mixed
5.78
805
523
489
376
111
cotton, wheat
(46.5)
(30.2)
(28.2)
(21.7)
(6.4)
rotation/sugarcane
Irrigated rice,
9.97
1102
462
830
396
441
wheat rotation
(110)
(46.1)
(82.8)
(39.5)
(43.9)
Irrigated mixed
2.77
785
316
601
538
64
rice, wheat
(21.7)
(8.7)
(16.6)
(14.9)
(1.8)
rotation/cotton
Irrigated wheat,
2.27
748
493
409
427
-18
fodder rotation
(16.9)
(11.2)
(9.3)
(9.7)
(-0.4)
Irrigated rice,
fodder rotation
1.25
1073
(13.4)
178
(2.2)
963
(12.1)
745
(9.3)
218
(2.7)
Total
26.02
974
(254)
451
(117.4)
696
(181)
434
(113)
262
(68)
The gross groundwater abstraction can be explored further to quantify the aquifer
depletion (Figure 7-7(e)). The total depletion of 31 km3 (121 mm yr-1) in the aquifer has
been computed from IRRgw and the return flow Qgw. The return flow, e.g. base flow from
the groundwater to the surface water system, of non-consumed water that is fed back into
the river network of 20 km3 yr-1 is included in the analysis. The net groundwater abstracted
(gross abstraction 68 km3 yr-1 minus recharge from leaking fields and canals 57 km3 yr-1)
became 11 km3 or 42 mmyr-1. The average net groundwater use (NGU) in Rechna doab is
101 mm which is within 20% of the NGU estimated by Ahmad et al (2005). They
estimated NGU of 82 mm in the Rechna doab using geo information techniques for the
year 1993-94.
The largest net groundwater depletion (DEPgw) occurs in Punjab province of India (200 to
800 mm yr-1). Jeevandas et al. (2008) has estimated a net deficit of 260 mm between crop
consumptive use and surface supplies in Indian Punjab. The Haryana state of India is also
vulnerable to serious groundwater depletion developments (400 to 600 mm yr-1). A recent
assessment of groundwater abstractions by NASA showed that the three states of India
(i.e. Punjab, Haryana and Rajasthan) lost about 109 km3 of water during 2002 to 2008
leading to decline in water table of 330 mm per year (Rodell et al., 2009). The
groundwater overdraft at this alarming rate could potentially change the transboundary
groundwater flow between India and Pakistan as also documented in IUCN (2010).
131
Another independent validation of the present results was performed against the GRACE
terrestrial water storage change estimates for the year 2007. The GRACE provides change
in terrestrial water storage observed at monthly time scale. Figure 7-8 shows the change
for the whole year. The change in storage in the irrigated areas of the basin is from 0 to
1000 mm yr-1(Figure 7-8). The spatial patterns of high groundwater abstraction coincide
with a broad region of intensive groundwater extraction and water table decline estimated
by GRACE. The differences between our estimates are, in combination with the errors
associated with GRACE groundwater change estimates (Schrama et al., 2007; Huang et
al., 2008; Duan et al., 2009), acceptable.
Figure 7-8 Terrestrial water storage changes observed from GRACE satellite for the
irrigated areas in the Indus Basin during the year 2007.
7.3.3
Accuracy assessment
The usability of the IRRgw information for carrying out water management plans depends
on the accuracy of the estimates. The influence of uncertainty in evapotranspiration,
rainfall, SWAT outputs and canal supplies on IRRgw computations based on Eq (7.9) and
(7.10) were tested. Seven pixels representing “irrigated cotton,wheat rotation/sugarcane”,
“irrigated cotton,wheat rotation/orchards”, “irrigated rice, wheat rotation” and “irrigated
rice,fodder rotation” were randomly selected. The locations of these land uses are provided
in the Table 7.3. Cheema et al. (2012) estimated ±4% uncertainty in ETETLook estimates
while RTRMM have deviation of ±6% at annual scale (Cheema and Bastiaanssen, 2012). The
errors in IRRcw are taken as ±15% (Habib, 2004; Ahmad et al., 2005). The error in SWAT
model output parameters (e.g. Qsurf, Qperc, Qlat ) are taken as ±15% (Harmel et al., 2006).
132
One thousand pairs of data series were randomly generated to estimate IRRgw using the
uncertainty range of ±4, ±6, ±15 and ±15% for ETETLook, RTRMM, IRRcw and SWAT model
outputs, respectively. For all seven locations the absolute deviation was plotted against its
probability of occurrence. The maximum absolute error (100% probability), ranges
between 122 to 213 mm with an average of 151 mm (Table 7.3). There is 70% probability
that the absolute error in IRRgw will be within 62 mm yr-1. The mean error at 50%
probability of exceedance is 41 mm yr-1. In the areas with high groundwater abstraction
rate (> 400 mm yr-1), this error can be considered within acceptable range (Ahmad et al.,
2005).
Table 7.3 Annual water balance components and computed gross groundwater abstraction
at seven selected locations. The absolute deviation of IRRgw at 100%, 70% and 50%
exceedance probability is also provided.
Pixel
Location
32°25'
48.12"N
73°24'
39.23"E
31°17'
34.54"N
73°04'
37.52"E
31°07'
05.83"N
75°42'
18.35"E
30°08'
49.02"N
75°28'
58.89"E
32°06'
42.44"N
73°49'
32.15"E
28°07'
17.29"N
67°59'
16.69"E
28°27'
13.85"N
68°28'
35.80"E
Average
Land
use
ET
R
ETLook
TRMM
Qsurf
Qperc
Qlat
IRRcw
IRRgw
(mm)
AGI1
Absolute deviation
of IRRgw(mm) at
probability of
exceedance
100
70
50
(%)
(%)
(%)
(mm)
(mm)
(mm)
(mm)
(mm)
(mm)
1311
767
131
70
0
722
23
213
84
54
AGI2
1174
409
50
48
0
365
498
122
47
30
AGI3
1202
580
237
169
0
230
798
142
51
33
AGI3
1265
429
59
53
3
359
592
124
50
33
AGI3
1098
579
163
105
0
348
439
148
52
35
AGI6
1316
148
24
37
0
534
695
131
62
42
AGI6
1361
183
16
23
0
800
417
175
86
60
1246
442
97
72
0.4
480
494
151
62
41
133
7.3.4
Water balance
The combination of pixel information on ET and rainfall, GIS information on canal water
supplies and losses and SWAT model outputs make it practically possible to analyze the
entire water balance of unsaturated and saturated zones in the irrigated areas of the Indus
Basin separately. The annual water balance components in unsaturated and saturated zones
are summarized in Figure 7-9 (a) and (b), respectively.
Figure 7-9 Annual water balance in (a) unsaturated and (b) saturated zones over the
entire irrigated areas of the Indus Basin for the year 2007. The total area is 26.02 mha
It is evident from the Figure 7-9(a), that the total inflows to the unsaturated zone is
estimated at 1147 mm. This inflow includes rainfall (451 mm), irrigation from surface
(434 mm) and groundwater (262 mm). The major outflow component from the unsaturated
zone is evapotranspiration (974 mm). A considerable amount (97 mm) leaves the system
as surface runoff and 75 mm percolates down to the shallow aquifer. The water balance is
closed in case of unsaturated zone (zero storage change).
Figure 7-9(b) shows the inflow and outflow components of the saturated zone. The inflows
to saturated zone include seepage and leakage from fields and canals that are equivalent to
219mm. An amount of 262mm gross amount of groundwater is abstracted from the
aquifer, while 79mm leaves the system as return flow. This outflow causes a depletion of
121mm in the saturated zone. For an area of 26.02mha, this is a net depletion of 31km3yr-1.
The water balance in the irrigated areas of the Indus Basin is further explored by
separating the basin into Pakistan and Indian parts. The detailed water balance components
134
are provided in Table 7.4. The part of Indus in Pakistan and India are written as Indus
Basin Pakistan (IB-PK) and Indus Basin India (IB-IN), respectively.
Table 7.4 Water balance components for the irrigated areas in Pakistani and Indian parts
of the Indus Basin for year 2007. The values in bracket represent water volume in km3.
Region
IB-PK
IB-IN
Indus
Basin
ET
R
ETLook
TRMM
(mm)
(km3)
978
(154)
963
(99)
974
(254)
(mm)
(km3)
381
(60)
558
(57)
451
(117)
Qsurf
Qperc
IRRRS
IRRcw
IRRgw
LOSScw
Qgw
DEPgw
(mm)
(km3)
85
(13)
117
(12)
97
(25)
(mm)
(km3)
56
(8)
103
(11)
75
(19)
(mm)
(km3)
740
(117)
629
(64)
696
(181)
(mm)
(km3)
485
(77)
357
(36)
434
(113)
(mm)
(km3)
255
(40)
271
(28)
262
(68)
(mm)
(km3)
195
(31)
67
(7)
144
(38)
(mm)
(km3)
60
(9)
107
(11)
79
(20)
(mm)
(km3)
64
(10)
208
(21)
121
(31)
Table 7.4 shows that intense groundwater irrigation is carried out in the Indian part of the
Indus Basin to meet consumptive use of crops especially paddy. The groundwater
proportion to the total irrigation is 44% in IB-IN while in IB-PK; it is 34% that is in
agreement with Foster and Chilton (2003). Foster and Chilton (2003) estimated
groundwater proportion of 34% in Pakistan based on information extracted from FAO
Aquastat database system. The total consumptive use of water in IB-PK is 117km3,
77km3is supplied through surface water and remaining 40 km3 is from groundwater. These
values are in close agreement with the estimates of Siebert et al. (2010). They made a
global inventory on groundwater use for irrigation and computed surface and groundwater
contribution of 78 km3 and 39 km3, respectively in Pakistan. Annual gross groundwater
abstraction in IB-PK is given by Qureshi et al. (2010a) as 51 km3. Groundwater
abstraction in 2007 estimated by Government of Pakistan (GOP) for entire Pakistan is 60
km3(GOP, 2010). All these estimates are based on the estimated tubewell density and their
approximated working hours. Taking into account the fact that the tubewell withdrawal
estimates have higher deviations as reported by Ahmad et al, (2005), the groundwater
abstraction estimated at 1 km pixel resolution in this study can be considered reasonable.
The groundwater depletion in IB-IN is estimated at 21 km3 as shown in Table 7.4. These
results are also consistent with the findings of Rodell et al. (2009). They estimated average
groundwater depletion of 17.7±4.5 km3 yr-1 over the Indian states of Punjab, Haryana and
Rajasthan. They used terrestrial water storage change observations from GRACE satellite
combined with the hydrological models for the years 2002 to 2008.
7.4
Conclusions
The Indus Basin is facing water shortage in time and space domains due to intensive
development in irrigation, domestic and industrial water needs. Irrigation is the largest
consumer that is using both surface (113 km3 or 434 mm) and groundwater (68 km3or 262
mm) to meet the crop water requirements. Uncontrolled groundwater abstraction,
consistently exceeding recharge, is threatening the groundwater reserves in the basin.
Groundwater is the hydrological component with the largest uncertainty. By using remote
sensing information in combination with GIS data on canal flows and SWAT model
135
outputs, a spatial estimate of the groundwater abstractions and depletions over the entire
irrigated area of the Indus Basin is obtained at a resolution of 1 km.
Certain areas in Pakistani and Indian provinces of Punjab and Haryana have experienced
extensive groundwater abstractions to augment surface supplies at unsustainable levels.
The gross groundwater abstractions within the irrigated areas accumulate to 68 km3 (262
mm) in the year 2007 and the corresponding groundwater depletion equaled 31 km3 (121
mm) due to the recharge of leaking canals and irrigation fields also considering return flow
to rivers. Sustainable groundwater management is under threat as groundwater abstraction
exceeds recharge consistently in these areas. The spatial maps of groundwater abstraction
and depletion identify the hot spots that need special attention of water management
experts. The districts of Jalandhar, Phagwara, Ludhiana, Bathinda in Indian part of the
basin and Narowal, Sialkot, Khanewal, Jhang and Larkana in Pakistani part of the basin
are most vulnerable.
The only solution to safeguard access to water for food and environment is to reduce
groundwater abstractions. Net depletion should be virtually neutral averaged over longer
time period. This can be achieved by negotiating groundwater abstractions using the maps
provided in Figure 7-7 (d) and (e). Monitoring of groundwater abstraction can be
implemented using the same methodology and procedures as outlined in this paper. The
technological procedures are outlined and validated. Recharge by constructing wells or
delay action dams should be facilitated. The role of trans-boundary aquifers should be
given equal importance as the attention that goes to surface water exchanges between
administrative boundaries.
The analysis is based on only one year needs more years to consider. There is also a
potential problem of limited validation as no information on spatial groundwater
abstraction is available. The use of single conveyance efficiency is cautious. Conveyance
efficiency based on per unit length of the canal should be tested in the future studies.
136
8 Summary and conclusions
8.1
Rationale
This PhD research investigates the use of multi-sensor satellite information to map
complex hydrological processes and water management practices in data scarce river
basins. The internationally shared Indus Basin is taken as an example. The international
character of the basin has made it difficult to study water flows and resource management
using traditional point based data sets. The Indus Water Commission has virtually no
access to any data, and consequently little power to monitor the developments and changes
of the water system. The limited data sharing and lack of trust between the riparian states
with political conflicts make it even worse. A first pre-requisite for regional scale cooperation on water issues and comparison between riparian states is a standardized
description of water flows, not only in streams, but also in aquifers and withdrawals to
irrigation systems. It is recognized that reliable data on water resources conditions is
insufficient in the Indus Basin. The lack of a dense network of hydro-meteorological
observatories is the plausible cause. Pixel based information obtained from satellites can
therefore be an attractive alternative solution that is worth investigating, especially
because the entire basin can be encompassed and measured in the same standard manner.
Transparent procedures to convert raw satellite data into water management information
need to be developed. This thesis provides the scientific grounds for developing
transparent data collection procedures based on earth observations. A knowledge base
using intelligent pixels is developed and validated.
The major goal of the research was to develop scientific methodologies to efficiently
utilize satellite measurements for quantifying conjunctive water use in data scarce river
basins. The first step to achieve this prime objective was to establish a reliable water
balance that can be used for further hydrological analysis. The water balance structure
includes an interactive link with water use, which makes it possible for users to appraise
the overall water management situation. If this is feasible for the Indus Basin, then it can
be used also for other basins in the world.
Four knowledge gaps were identified which impede a successful basin scale hydrological
analysis: (i) lack of fundamental data (e.g. hydrological, water management,
meteorological),(ii) lack of knowledge on land use, (iii) missing information on
groundwater resources for the entire basin, and (iv) lack of analytical tools to study
alternative solutions to combat over-exploitation of water resources and become more
climate resilient.
The specific contribution of the study and innovative aspects are highlighted in each of the
chapters which include, but are not limited to the development of methodologies to
determine the spatial variability of land use (chapter 3), rainfall (chapter 4), soil moisture
(chapter 5), evapotranspiration (chapter 6), and groundwater abstraction (chapter 7).
Groundwater abstraction is one of the biggest problems in sustainable water management,
but information on abstractions cannot be obtained from thousands of individual
piezometer readings. It is clear that there are major water balance constituents of which
137
temporal patterns can never be accurately simulated by traditional point based hydrometeorological monitoring programs.
Spatial data can be used for preparing water accounts, prohibiting new upstream water
resources developments, groundwater restoration plans, developing fair irrigation
management practices (adequate, reliable, uniform), providing access to water during
droughts, estimating impact of retiring glaciers, reducing non-beneficial water use,
enhancing recycling of drainage water, introducing green water credits in upstream
catchments, etc. Some tangible recommendations are provided at the end of this chapter.
This study can be considered as a first in its kind on the water balance of the entire Indus
Basin including Pakistan, India, China and Afghanistan. Making available reliable
information on the data scarce and politically divided Indus Basin is fundamental for
reviewing and implementing international agreements. The study was carried out using
freely available public domain satellite data from the World Wide Web. The original
source of satellite data is freely accessible and transparency is guaranteed. Such
information enables water managers and decision makers of developing countries to
efficiently manage water resources in an era of population growth, diminishing per capita
water resources, and threats of climate change.
This thesis describes the research results of preparing digital data layers on land use,
rainfall, soil moisture, evapotranspiration and groundwater abstractions, without
undertaking sophisticated field measurement campaigns. The only field data used
originates from weather stations and measurements of water released from large reservoirs
into the main irrigation canals.
8.2
Pixel land use
A land use database is essential to provide information on the type of water consumers and
the returns in terms of food production, wood production, hydropower, environmental
services, economic benefits, etc. For judicious allocation of water, the crops grown in the
area have to be identified. In chapter 3, an innovative way of discerning land uses is
presented. Spatially consistent 1 km × 1 km pixel information of various land uses was
generated using NDVI data freely available from vegetation sensors on board the SPOT
satellite for the year 2007.The temporal profiles of NDVI were used to describe the
phenological cycle of all agro-ecosystems. There were considerable differences in NDVI
patterns between the land uses. It is the first land use map with 27 classes that covers the
entire Pakistan. The irrigated land uses showed two peaks in a year while evergreen forests
have consistently higher NDVI throughout the year. The rainfed land uses have peaks in
phase with rainfall and soil moisture conditions. The phenological cycle thus obtained,
supported by expert knowledge and ground information, was used to identify specific
crops and crop rotations. Overall accuracy of 77% (Kappa = 0.73) was attained, which is
deemed sufficient for supporting water management analyses of large heterogeneous
landscapes.
138
8.3
Pixel rainfall
Any water resources analysis starts with quantification of rainfall. Aerial rainfall for
catchments and basins is normally interpolated from rain gauge networks. A density of
less than four gauges per 10,000 km2, is however insufficient to capture the spatial
heterogeneity of rainfall processes. Any small error of a large water balance component
such as rainfall will produce significant errors in the smaller components such as runoff.
The spatial interpolation of point measurements in heterogeneous landscapes and
mountains result in erroneous estimates. Dense networks are needed that are difficult to
establish and maintain in developing countries. Rainfall estimation by means of satellites
is therefore essential.
Satellite based sensors provide an integrated measurement of rainfall in time and space.
They are an excellent alternative for rainfall measurement. Sensors on board the TRMM
satellite provide three-hourly global rainfall estimates that can be freely downloaded. In
chapter 4, details of these sensors are provided, as well as results from accuracy tests of
the TRMM in the Indus Basin. It was concluded that these satellite derived rainfall
products still show relatively large uncertainties and are difficult to validate. The influence
of geography on satellite estimates was apparent with an exponential increase in deviations
between TRMM and rain gauge measurements with increasing elevations. It demonstrates
that standard statistical relationships with topography cannot be applied.
Two methodologies were therefore used to calibrate the monthly estimates of the TRMM
rainfall (product 3B43) with low density rain gauge measurements. One method was
regression analysis and the other geographical differential analysis. Both techniques
improved the results with Nash Sutcliffe efficiency by 81% and 86%, respectively.
However, the standard error of estimate of the geographical differential analysis was 41
mm lower than the regression analysis.
The advantage of the geographical differential analysis is that the differences in rainfall
are used for spatial interpolation and not the absolute values of rainfall. The values for the
differences reflect certain underlying terrain features, and the corrections at specific
locations were thus bigger than in other locations. The deviations in measured and
estimated rainfall can be reduced considerably to 6% This is equivalent to an amount of
water of 26 km3 yr-1, which is 17 % of the withdrawals to the irrigation sector, being
approximately 150 km3yr-1.Any small error in rainfall measurement thus affects the
calculation of water availability for diversions.
8.4
Pixel surface soil moisture
Soil moisture is considered as another state variable that is informative of the water
balance. Under arid conditions soil moisture patterns are a direct indication of the presence
of water. Wet top soils reflect irrigation water distribution or shallow water table areas. It
is practically impossible to attain soil moisture information at large spatial scales with in
situ sensors, and for this reason it is not common to include soil moisture in management
decision making. Passive microwave sensors on board satellites such as AMSR-E, SMOS
and Feng Yung provide global scale estimates of daily surface soil moisture for free.
These sensors provide continuous soil moisture estimates without being affected by
139
weather conditions. That is also the main reason for applying a new evapotranspiration
(ET) algorithm based on soil moisture measurements.
Surface soil moisture is estimated by applying inversion techniques to the brightness
temperature measured by satellites. The technique is not error free; therefore it was
necessary to validate the satellite soil moisture prior to its use in estimating other
hydrological processes. Due to the non-availability of in situ soil moisture measurements
in vast river basins, classical validation techniques are not technically feasible. Therefore,
alternative validation approaches were needed to build confidence in using satellite soil
moisture products. The response of vegetation to soil moisture and soil moisture to rainfall
was studied to explain the soil moisture behavior (chapter 6).
Strong relationships between TRMM rainfall and AMSR-E surface soil moisture in the
land use classes “rainfed”, “very sparse vegetation”, and “bare soil” were observed. The
rainfall events in these land uses have a high association with the soil moisture measured
by AMSR-E. For irrigated land, this association was lower due to extra supplies from
irrigation – and thus perturbations of the rainfall – soil moisture relationship. At annual
time scale a stronger correlation exists between the AMSR-E mean soil moisture and
TRMM accumulated rainfall (Spearman’s rank correlation coefficient rs=0.74) than
between TRMM accumulated rainfall and NDVI (rs=0.70), which is explicable on the
basis of soil physical processes. Mean soil moisture and NDVI have stronger correlation
(rs=0.85) compared to TRMM rainfall and NDVI (rs=0.70) which is also according to
expectations.
A time lag between soil moisture and NDVI time series was observed. Such a lag was
expected due to delayed response of vegetation against moisture in the root zone. The lag
time varied between zero to 60 days, and was generally longer for the wet kharif season.
For the dry rabi season, a Pearson’s r> 0.60 was found for 75% of the cases with zero to
40 days lag. For the wet kharif season, it was found for 81% cases but with a lag of 20 to
60 days. The maximum surface soil moisture value of 0.35 to 0.45 cm3 cm-3for a pixel was
similar to the top layer saturated moisture content expected on the basis of soil texture
maps and pedo-transfer functions. Higher values occurred in flooded lands and paddy
fields. This suggests that absolute AMSR-E values properly describe soil moisture under
wet land surface conditions.
8.5
Pixel evapotranspiration
Evapotranspiration (ET) accounts for the dominant part of the outgoing fluxes of vegetated
land in semi-arid climates. Because ET can be managed partially, it provides a vehicle to
control the available water resources for agriculture, forests, swamps, wetlands, and other
types of land cover. The actual rate of evapotranspiration cannot be inferred from routine
weather data. Rainfall can be measured easily with cheap gauges, but not ET from land
surfaces (except at sites equipped for scientific energy balance equipment). Even though
water institutions and international river basin commissions have the duty to manage the
water resources of river basins, they do not use basin scale evapotranspiration information
at all. The financial constraints in the developing countries restrict installation of
permanent flux observation towers. There is no single flux tower present in the Indus
140
Basin, while one seventh of the world population lives here and a volume of 496 km3of
water evaporate every year.
Therefore, a new methodology (ETLook) has been tested that provides spatial estimates of
evapotranspiration from satellite measurements and a surface energy balance. There are
several ET algorithms available in the literature, but the advantage of this particular
algorithm is that it provides continuous estimates of ET throughout the year without being
affected by weather. ETLook is a two-source model that can infer information on nonbeneficial evaporation and beneficial transpiration separately using microwave derived
surface soil moisture. Microwave radiometry is least affected by cloud cover and can thus
provide continuous surface soil moisture information even in monsoon periods and for
high altitude regions with persistent clouds. This study was a first attempt to use
microwave technologies to accurately estimate evapotranspiration over vast areas of the
Indus Basin, using public domain datasets. Due to technical problems, the data from
AMSR-E is not available since October 4, 2011. However, passive microwave radiometers
like SMOS (European Space Agency) of Feng Yung (Chinese Space Agency) also provide
surface soil moisture estimates.
The estimated evapotranspiration at 1 km pixel resolution correlated well with some
historic lysimeter measurements, Bowen ratio measurements, and remote sensing studies
(R2of 0.70 to 0.76 at annual time scale; RMSE of 0.29 and 0.45 mm d-1).It was
demonstrated that the pixel scale ET fluxes at daily, 8-day, or monthly time scales can be
estimated. The total basin wide ET in 2007 was 496 km3yr-1 while rainfall was 443 km3yr1
. This revealed that more water is evaporated from the land surface than what was
received through rainfall. Multiple line agencies from the region and the gravity mission
from NASA also suggested a net storage change related to net groundwater withdrawals,
declining groundwater tables, snowmelt, and retiring glaciers. The only solution to make
the environment of the Indus Basin more sustainable is to reduce ET, and in particular the
non-beneficial ET.
ET information can greatly assist water managers and policy makers with water
allocations. It is also important to gain knowledge of net water producing (R>ET) and
water consuming areas (ET>R). The stream flow resulting from net water producing areas
can be managed by land use change and adjusted cultivation practices. The concept of
green water credits is based on certain agreed upstream ecosystem services to generate
sufficient stream flow. Net consuming areas–such as irrigation systems - can be controlled
by regulating the water diversions and withdrawals. The irrigation sector is often criticized
for inefficient use of water resources. By comparing water diversions to consumptive use,
estimates of non-consumed water flows can be made. Irrigation water supply in Pakistan is
based on equal access to water for all farmers. Without doubt, such aspiration is difficult to
achieve anywhere in the world and especially in areas where water resources are limited.
Farmer communities often claim that they are not receiving the volume of water they are
entitled to. Pixel values of ET will provide a new source to verify these conditions; not
only to certify that they receive water, but at the same time also to verify that they are not
using water non-beneficially and non-productively.
141
8.6
Pixel groundwater abstraction
Unmetered and large groundwater withdrawals are two of the biggest water resources
problems in the Indus Basin. An adequate assessment of the groundwater withdrawals can
be made from the spatial tools presented in this thesis. With spatial estimation of rainfall
and ET, it is possible to determine the amount of irrigation water supplies that match ET
rates. A hydrological model was used for this purpose. Pixel information on topography,
land use, soils, rainfall, and evapotranspiration was used to provide input data into the
SWAT hydrological model. SWAT is a well-known and accepted distributed hydrological
model that describes many relevant hydrological, soil physical and bio-physical processes.
It was used because of its free availability, global adoptability, and the ability to compute
atmospheric, land surface, soil, groundwater, and stream flow processes.
The SWAT model parameterization and calibration was supported by 8-day ET layers
from ETLook. A calibration procedure with variable irrigation water supply and soil
physical properties was carried out for each hydrological response unit. The calibrated
model was then applied to generate output maps of total irrigation water supply at the farm
gate, surface runoff, and combined drainage and percolation, which provide fundamental
insights in the breakdown of irrigation water flows into consumed and non-consumed
outflows of all irrigated fields present in the Indus basin in an uniform manner.
The release of surface water from the major reservoirs to the canal command areas was
151 km3yr-1. By integrating irrigation from canal water that arrives at the farm gate (113
km3yr-1) with the total irrigation water supply estimated from the SWAT model (181
km3yr-1); it was possible to isolate the total groundwater abstractions for irrigation (68
km3yr-1). Pakistan supplies 40 km3and India 28 km3 every year. Table 8.1 provides an
overview of the bulk water balance of the irrigated areas of the Indus Basin.
Table 8.1The water balance of all irrigated areas combined and considered as being one
single“ big field” of 26.02 million ha
Inflow
km3 yr-1 Outflow
km3yr-1
Rainfall
117
Surface runoff
25
Canal water supply India
36
Interception evaporation 5
Canal water supply Pakistan
77
ET from rainfall
99
Groundwater supply India
28
ET from irrigation
154
Groundwater supply Pakistan 40
Drainage and percolation 19
from fields
Total
298
298
The vector maps of canal water supplies for every canal command area were
superimposed on the vector maps of surface runoff and drainage / percolation and the
raster maps of rainfall and ET. This yielded a map with 1 km pixels of gross groundwater
abstractions for the entire Indus Basin (Figure 8-1). This is the first map with this level of
detail. It shows the hotspot areas at a 100 ha resolution. Information on groundwater
pumping activities can thus be made public knowledge.
142
Figure 8-1 Schematic diagram showing data sources to infer groundwater abstraction
information at 1 km pixel.
The groundwater abstraction can be explored further to quantify the groundwater
depletion. The groundwater depletion (gross abstraction 68km3yr-1 minus recharge from
leaking fields, amounting to19km3 yr-1,and canals amounting to 38km3 yr-1) was 31km3 yr1
. The groundwater return flow of 20km3 yr-1 in the irrigated areas of the Indus Basin was
included in the analysis. The actual groundwater depletion is larger due to outflow of
groundwater into the river system. The largest groundwater depletion occurs in the Punjab
province of India (350 – 800 mm yr-1). The Haryana state of India is also threatened by
serious groundwater depletion. The only solution to safeguard access to water for food and
environment is to reduce net groundwater usage to virtually nothing over a longer time
period. Figure 8-2 provides information on various water balance components for irrigated
areas of Indus Basin lying in Pakistan and India separately.
Figure 8-2 Water balance components in the irrigated areas of Pakistani and Indian part
of the Indus Basin
143
8.7
New data sources
Classical data (e.g. data from governmental organizations) are generally point
measurements and its accessibility is generally poor. Remotely sensed information is an
effective alternative source. It has a public domain status, and everybody can have access
to raw satellite information. Data from an international fleet of sensors can be found in the
Earth Observing System Data and Information System (EOSDIS). Eight Data Active
Archive Centers (DAAC’s), representing a wide range of earth science disciplines, are
operational under NASA to process, archive, and distribute EOSDIS data. The Earth
Resources Observation System (EROS) Data Center of the USGS provides, in addition,
access to land processes data from both satellite and aircraft platforms.
Most of the satellite data at relative coarse resolution, which serves the purpose at basin
scale, is freely available. The fine resolution data is also available at a fee. NASA and
ESA have new policies to keep prices of images low, so that satellite information becomes
everybody’s business. Some examples of internationally opened satellite databases can be
found at http:// www. daac. gsfc. nasa. gov,https:// www. wist. echo. nasa. Gov /~wist
/api/imswelcome/orhttp://www.nsidc.org/data/ae_land3.html.
The classical sources of data and associated problems with their alternative solutions are
summarized in Table 8.2. The use of satellite data as an alternative for getting firsthand
knowledge on hydrology, agriculture, environment and geography, was explored in this
thesis. The climatic data available through meteorological departments was measured
using routine weather stations. Space borne measurements made it possible to get these
datasets at higher temporal and spatial scales.
Table 8.2 Applicability of pixel information for database generation in vast and data
scarce international Indus Basin
Database
Classical data
Associated problems
Alternative Solution and
acquisition
applicability
sources
Land use
- Global
- Generalised land
NDVI time series at 1
databases*
cover classes
km pixels with intervals
- International
- No information on
of 8 to 10 days
organizations§
specific crop
- Governmental
rotations
organizations
- Outdated
Cropped
- Governmental - No real time
Land use map
area
organizations
information
- International
- Late dissemination
organizations
- Tabular data
Biomass
- Global net
- No data
Advanced algorithms
production
primary
turning raw data into
production
quantified information
maps
Crop yield
- Governmental - Late dissemination
The pixel information on
organizations’ - Administrative unit
land use and biomass
statistics
wise information
production can provide
144
Rainfall
Snow cover
Soil
moisture
- Meteorological
department
- World
meteorological
organization
- Global
databases‡
- Global datasets
- Specific
Projects
- Field
experiments
Evapo - Field
transpiration
experiments
Solar
- Meteorological
radiation
department
- Absence of spatial
data
- Point measurements
- Sparse raingauge
networks (<4
gauges/ 10,000km2)
- Not proper
representation of
spatial hydrology
- Not available
- Not available
crop yield at pixel scale.
3-hourly, daily and
monthly global estimates
of rainfall by TRMM
satellite.
Optical and radar
satellites
Passive microwave and
radar satellites
- No flux tower in the Advanced algorithms
Indus Basin
such as ETLook
- Point data available Geostationary and polar
through few stations orbiting satellites
- Not enough for any
measuring cloud cover
basin scale study
Ground
- Groundwater
- Point data available Pixel information on
water
monitoring
through sparse
related hydrological
organizations
piezometer network variables can provide
- Only two
spatial estimates of
measurements in a
groundwater abstraction.
year
- Data not easily
accessible
Reservoir
- Single gauge
- Single gauge is
Radar and lidar
levels
insufficient
altimeters
- Data not accessible
*
GLCC-IGBP, Glob Cover-ESA, GLC2000, § IWMI, ¶ GMAO-MERRA, ‡ GPCC, CMAP,
ERA
The most outstanding conclusion therefore is that satellite data archives should be
explored more often to study the water resources conditions in basins with water and data
scarcity. Data on land use, rainfall, soil moisture, and evapotranspiration can be derived.
Integration with other data sources such as GIS (e.g. canal water flow) and hydrological
models (e.g. irrigation supply, surface runoff, drainage / percolation) is preferred. Once
properly computed, the spatial dataset has the potential to drastically improve the
knowledge base on the hydrology of data scarce trans-boundary basins with high climate
variability.
Institutes, organizations, or consultants could process the raw satellite data into water
resources information systems across basins with national and international rivers and
aquifers. The emerging trends in space technology, geographical information systems, and
145
their applications, coupled with developments in numerical hydrological modeling, should
be oriented towards maximizing benefits of all stakeholders. This thesis is an “Indus Basin
Data Generator” and an example for other internationally shared river basins. The same
procedure could be applied elsewhere because it has been demonstrated that the only data
source from the national line agencies was the release of water flows from reservoirs. All
other data was drawn from generally accessible data bases.
8.8
Development of applications
The world agricultural production is vulnerable to climate change. Global warming in the
past few decades has influenced a number of components of the hydrological cycle and
hydrological systems. For example, intensity and extremes of rainfall; widespread glacier
melts; increased non beneficial evaporation; and changes in soil moisture and runoff.
Therefore, the need to manage international waters in an integrated manner is amplified.
The management of international water has implications at the local, regional, and global
levels, and needs a standard knowledge framework that describes the resource base and the
management conditions. The introduction of water accounting procedures by the UN and
IWMI is a good direction to achieve a common understanding of water resources. A
stronger institutional and legal framework is needed to improve management of
international river basins. This is more likely to happen if the supporting data sources are
available and reliable.
This thesis propagates the notion that international basin initiatives and commissions
should rely on transparent and unbiased satellite measurements for their water resources
information. The availability of spatially distributed and temporally consistent
information on various hydrological parameters opens complete new opportunities to
study the hydrological process, water resources depletion, food security, and environmental
development in international river basins. It opens a new protocol where central
governmental bodies and internationally controlling agencies get uniform information.
There are a number of hydrological challenges that need to be given attention while
carrying out future basin scale integrated water resources management.
One of the biggest challenges is the increased water demand due to population growth and
growing need for food security, which increase pressure on aquifers. Groundwater
withdrawals have become a major problem, exacerbated by climate change. It is
anticipated that the groundwater and related environmental issues will worsen in near
future. An integrated approach of conjunctive use and management of surface and
groundwater is needed. The groundwater management should be carried out beyond the
political boundaries of the countries sharing trans-boundary aquifers. The unchecked and
extensive groundwater abstractions in one part of the basin can adversely affect the flow
paths and thus availability in other parts. The mandate of river commissions needs to be
revised and UN proposed “Law of Transboundary Aquifers” should be approved by the
member states and should also be included in the Indus Water Treaty. The draft law has
clear instructions on management, restoration, and data exchange between the riparian
countries of internationally shared aquifers.
Global warming will cause glacier melts resulting in massive river flows during the first
half of the century, followed by 30 to 40% decrease in rivers flows at the end of this
146
century. Therefore, to combat the impending calamity of floods and droughts it is
important to construct new large capacity reservoirs under the mandate of the Indus Water
Treaty. These reservoirs will not only serve the purpose of water storage and temporal
regulation, but also provide a source of power generation.
Besides construction of large reservoirs, small dams should also be considered especially
in the areas where abstraction exceeds recharge. Such interventions will help to replenish
depleting aquifer and will ensure sustainability of the resource for future generations.
An increase in water productivity (crop yield per unit of water consumed) is deemed
necessary to feed growing population. This could be achieved by reducing non-beneficial
evaporation (263 km3yr-1) from fallow lands and minimizing non-beneficial transpiration
from weeds and grasses grown within the crops.
A flexible approach to irrigation water supplies is needed based on crop water needs
instead of a fixed supply system currently in practice in the Indus Basin. This can be
achieved by continuous monitoring of ET over the irrigated areas of the basin. It is
unfortunate that there is not a single permanent flux tower in the entire basin available to
measure the actual evapotranspiration, which is essential for efficient irrigation. An effort
should be made to install a few flux towers in the basin for ET monitoring along with the
setup of real time ET monitoring through satellites.
There is a need to monitor land use changes continuously, as any change in land use will
affect downstream water flows. The anthropogenic land use changes upstream can result
in land degradation that severely affects the downstream riparians. For example, upstream
deforestation can result in changing rainfall patterns. Soil erosion from upstream
watersheds is a major cause of sedimentation in the large reservoirs. Here a relatively new
concept of “green water credits” can play a role in incentives for upstream farmers to stop
deforestation and carryout soil and water conservation practices to benefit downstream
water users.
The limitation of this investigation is the use of only a single year of data. The reason was
to explore the academic possibilities and limitations of using pixel information in
hydrological studies and to build confidence in the use of satellite products. It was not
intended to solve the water resources management problems in the river Indus. For sound
decision making, it is recommended to expand the future analysis with longer time series.
8.9
Conclusions
The aim of this study was to develop methodologies to efficiently utilize satellite
information that leads to a better understanding of the hydrological processes and water
management practices in data scarce river basins. At the same time it proposes
development of tools that can support decision makers and water managers in testing
alternative solutions and combating over exploitation and poor management of water
resources.
A description of the water flow path of the Indus Basin can be based on factual
information by using satellite information. The national scale components of the Indus
Basin can be computed including the cross boundary flows, also for un-gauged basins. The
147
riparian states’ fear of cheating can be diminished. They can utilize this objective
information on water flows to have unbiased databases.
The main conclusions are summarized as:
148

Data scarcity issues can be solved successfully by using satellite measurements as
proxies. They are always an indirect measurement though.

Satellite data requires intelligent pixel interpretations and basin-wide
hydrological modeling to generate quantified spatial information on, for example,
land use, rainfall, soil moisture, actual evapotranspiration, and groundwater
abstractions.

Only limited support is needed from ground information observatories to interpret
satellite data.

Satellite measurements of temporal vegetation phenological cycles were applied
to create a comprehensive land use map of the entire Indus Basin with 27 land
use classes. Such a map was not available earlier. The knowledge of dominant
crop rotation schemes can play an essential role in the planning of food security
and rural development.

The integration of land and water use offers a robust data set to categorize
beneficial and non-beneficial water use, and hence make water allocation
processes more efficient.

In areas with sparse rain gauge networks, the estimation of rainfall from satellites
adds important information for water management applications. Mountains and
forests do not have the required density of gauges. Because of the small runoff
proportion in arid climates, absolute values of basin-wide rainfall need to be
known accurately. The recent floods in Pakistan confirmed that basin scale
rainfall information is vital.

Accurate information on E and T at a daily and weekly basis is a basic
requirement for efficient irrigation and can be obtained at 1 km pixel using
passive microwave soil moisture without being affected by climatic conditions.

Combining ET and rainfall maps provides valuable insights in net water
withdrawals.

The availability of groundwater information is one of the biggest water resources
management problems. A map with 1 km annual abstraction rates has been
prepared, which is the first of its kind, not only for the Indus Basin, but likely
worldwide. The largest groundwater abstractions (64% of total) occur in irrigated
rice, wheat rotation land use followed by irrigated cotton, wheat rotation mixed
with orchards with 20% of total abstractions. Hot spots need special attention to
restore the underlying aquifers.

The abstractions in irrigated areas are more than recharge thus resulting in 30km3
of net groundwater usage that needs to be reduced to maintain aquifer
sustainability.

The trans-boundary basin water resources must be managed in an integrated
manner considering both surface and groundwater. ET management should be
introduced with the aim to minimize ET in certain areas to ensure a feasible crop
production, while other areas need to be given a quota that they are expected not
to exceed. The irrigation in every sub-basin and local administrative boundary
should be based on an ET quota. This system is applied on the North China plain
and is known as “ET controlled groundwater management”.

The continuous stress on trans-boundary aquifer demands emphasizes the
importance of the Indus Water Treaty to account for trans-boundary groundwater
along with laws on sharing surface water.
The looming water crisis in international river basins can be partially alleviated
by creating trust and faith through the use of objective data relevant to water
resources management. International databases such as the one of which the
applicability is shown in this thesis can contribute to that. Such information is
first of its kind in the Indus Basin and can be of great use for water resources
managers in basin level water management and protection.
149
150
9 Samenvatting
9.1
Motivatie
Dit proefschrift beschrijft de resultaten van een onderzoek naar het gebruik van multisensor satellietinformatie voor het beschrijven van complexe hydrologische processen en
de waterhuishouding in data arme stroomgebieden. Het internationaal gedeelde Indus
stroomgebied is als voorbeeld genomen. De internationale aard van het stroomgebied
maakt het moeilijk om de waterstromen en de waterhuishouding via traditionele punt
gebaseerde datasets te beschrijven. De ‘Indus Water Commission’ heeft vrijwel geen
toegang tot gegevens en heeft daardoor weinig mogelijkheden om de ontwikkelingen en
veranderingen van het watersysteem te monitoren. Het beperkt delen van gegevens en het
gebrek aan vertrouwen tussen de aangrenzende landen met politieke conflicten versterkt
dit. Een eerste vereiste voor regionale samenwerking op het gebied van water en voor het
maken van een vergelijking tussen de aangrenzende staten is een gestandaardiseerde
beschrijving van de waterstromen. Niet enkel de waterstromen in de rivier, maar ook de
aquifers en de onttrekkingen uit het irrigatiesysteem moeten worden beschreven. Het is
algemeen erkend dat betrouwbare gegevens over de waterhuishouding van de Indus Basin
niet toereikend zijn. Een intensief hydro-meteorologisch meetnetwerk ontbreekt. Pixel
gebaseerde satellietinformatie kan daarom een aantrekkelijke alternatieve oplossing
bieden, en het is waard om dit te onderzoeken, zeker omdat het gehele stroomgebied
meegenomen kan worden en op eenzelfde standaard manier kan worden gemeten. Heldere
procedures voor het converteren van de ruwe satelliet data naar waterbeheersinformatie
moet worden ontwikkeld. Dit proefschrift levert de wetenschappelijke onderbouwing voor
de ontwikkeling van transparante procedures voor het verzamelen van data uit
aardobservaties. Een datasystem dat gebruik maakt van intelligente pixels is ontwikkeld en
gevalideerd.
Het hoofdonderzoeksdoel is de ontwikkeling van wetenschappelijke methodes om
efficiënt gebruik te maken van satellietmetingen voor het kwantificeren van
gemeenschappelijk watergebruik in stroomgebieden met schaarse grondgegevens. De
eerste stap om dit doel te bereiken was het opzetten van een betrouwbare waterbalans die
gebruikt kan worden voor verdere hydrologische analyses. De waterbalans bevat een
interactieve link met het watergebruik, waardoor het voor de gebruikers mogelijk is om het
gevoerde beheer te beoordelen. Als dit mogelijk is voor de Indus Basin, dan zal het tevens
mogelijk zijn voor andere stroomgebieden in de wereld.
Een succesvolle hydrologische analyse op stroomgebiedsschaal wordt door vier
kennisgebreken belemmerd: (i) gebrek aan fundamentele data (bijv. hydrologisch,
waterhuishoudkundig, meteorologisch), (ii) gebrek aan landgebruik kennis, (iii) missende
informatie over de grondwaterbronnen voor het gehele stroomgebied, en (iv) gebrek aan
analytisch modellen voor het bestuderen van alternatieve oplossingen om overexploitatie
van waterbronnen te bestrijden en om meer klimaatbestending te worden.
De specifieke bijdrage van het onderzoek en innovatieve aspecten komen in elk hoofdstuk
aan bod. De ontwikkeling van methodes voor het bepalen van de ruimtelijke spreiding van
landgebruik (hoofdstuk 3), neerslag (hoofdstuk 4), bodemvocht (hoofdstuk 5), verdamping
151
(hoofdstuk 6) en grondwateronttrekking (hoofdstuk 7) wordt beschreven.
Grondwateronttrekking is een van de grootste problemen voor een duurzaam waterbeheer.
Echter informatie over de onttrekkingen kan niet worden verkregen uit de duizenden
individuele peilbuisobservaties omdat ze niet regelmatig worden uitgelezen. Het is
duidelijk dat de temporele patronen van de waterbalans nooit met hoge nauwkeurigheid
gesimuleerd kunnen worden op basis van traditionele punt gebaseerde hydrometeorologische meetprogramma’s.
Ruimtelijke gegevens over het waterbeheer kunnen worden gebruikt voor het voorbereiden
van de wateraccounts, het verbieden van nieuwe bovenstroomse waterhuishoudkundige
ontwikkelingen, grondwaterherstelplannen, ontwikkeling van een eerlijk irrigatie
management (adequaat, betrouwbaar, uniform), leveren van toegang tot water tijdens
droogte, inschatten van de gevolgen van terugtrekkende gletsjers, reduceren van
waterverspilling, bevorderen van hergebruik van drainage water, introduceren van groen
water kredieten in bovenstroomse stroomgebieden, etc. Enkele tastbare aanbevelingen
worden aan het einde van dit hoofdstuk gegeven.
Dit onderzoek is – in zover bekend - de eerste keer dat de waterbalans van het gehele
stroomgebied van de Indus Basin, inclusief Pakistan, India, China en Afghanistan, wordt
meegenomen. Het beschikbaar stellen van betrouwbare informatie over het politiek
verdeelde Indus Basin met schaarste gegevens is essentieel voor het herzien en
implementeren van internationale overeenkomsten. De studie is uitgevoerd met vrij
beschikbare openbare satelliet data van het Word Wide Web. De originele databronnen
zijn vrij beschikbaar en transparantie is gewaarborgd. Dit soort informatie stelt
watermanagers en besluitvormers in ontwikkelingslanden in een tijd van bevolkingsgroei,
vermindering van waterbeschikbaarheid per capita en bedreigingen van
klimaatsverandering in staat om efficiënt waterbeheer uit te voeren
Dit proefschrift beschrijft de onderzoeksresultaten van de toepassing van digitale kaarten
van landgebruik, regenval, bodemvocht, verdamping en grondwateronttrekkingen, zonder
dat uitgebreide veldmeetcampagnes zijn ondernomen. De enige veldgegevens die zijn
gebruikt komen van weerstations en metingen van watertoelevering vanuit de grote
reservoirs naar de hoofdirrigatiekanalen.
9.2
Pixel landgebruik
Een bestand van het landgebruik is essentieel voor het leveren van informatie over het type
watergebruiker en geeft informatie over de voedselproductie, houtproductie, waterkracht,
milieuwaarden, economische voordelen, etc. Om de allocatie van water te kunnen
beoordelen, moet bekend zijn welke gewassen er groeien in het gebied. In hoofdstuk 3
wordt een innovatieve manier om landgebruik te onderscheiden gepresenteerd.
Ruimtelijke consistente pixel informatie in een grid van 1 km x 1 km van verschillende
vormen van landgebruik is gegenereerd uit NDVI gegevens. Deze gegevens zijn vrij
verkrijgbaar van sensoren aan boord van de SPOT satelliet voor het jaar 2007. De
temporele profielen van NDVI zijn gebruikt om de fenologische cycli van alle landbouwen ecosystemen te beschrijven. Er waren grote verschillen in de NDVI patronen tussen de
verschillende typen landgebruik. Het is de eerste landgebruikskaart in Pakistan met 27
klassen. Het geïrrigeerde land laat twee fenologische periodes per jaar zien, terwijl
152
permanent groene bossen een periode met een hogere NDVI heeft. Land met
regenafhankelijke vegetatie vertoont NDVI pieken in fase met de neerslag en
bodemvochtcondities. De fenologische cyclus, samen met veldinformatie is gebruikt om
specifieke gewassen en gewasrotaties te identificeren. De algehele nauwkeurigheid van
77% (Kappa=0.73) werd bereikt, welke voor het ondersteunen van watermanagement
analyses over grote heterogene landschappen voldoende wordt geacht.
9.3
Pixel neerslag
Elke analyse van het waterbeheer begint met het kwantificeren van de neerslag. De
ruimtelijke neerslag voor stroomgebieden wordt normaal gesproken verkregen door het
interpoleren van regenmeters in een netwerk. Een dichtheid van minder dan vier
regenmeters per 10.000 km², is echter onvoldoende om de ruimtelijke heterogeniteit van
neerslagprocessen weer te geven. Een kleine fout in een grote waterbalanscomponent,
zoals neerslag, zal significante fouten in de kleinere waterbalanscomponenten, zoals
afvoer, tot gevolg hebben. De ruimtelijke interpolatie van puntmetingen in heterogene
landschappen en gebergten leidt tot foutieve schattingen. Intensieve meetnetwerken zijn
nodig, die moeilijk in ontwikkelingslanden op te zetten en te onderhouden zijn.
Neerslagschattingen van satellieten zijn daarom essentieel.
Sensoren op satellieten geven een integrale meting van neerslag in tijd en ruimte en zijn
een perfect alternatief voor neerslagmetingen op de grond. Sensoren aan boord van de
TRMM satelliet geven drie-uurlijkse mondiale neerslagschattingen welke vrij downloadbaar zijn. In hoofdstuk 4 worden de verdere details over deze sensoren gegeven, net als de
testresultaten over de nauwkeurigheid van TRMM in de Indus Basin. We concluderen dat
deze door satelliet verkregen neerslagproducten echter nog steeds relatief grote
onzekerheden hebben en moeilijk te valideren zijn. De invloed van geografie op de
satellietschattingen is zichtbaar met een exponentieel toenemend verschil tussen de
TRMM en regenmetingen met toenemende hoogte. We laten zien dat standaard statistische
relaties met topografie niet kunnen worden toegepast.
Twee methodes zijn uitgeprobeerd om de maandelijkse schattingen van TRMM (product
3B43) met een geringe dichtheid van regenmeters te kalibreren. Eén methode is een
regressie analyse en de andere een geografische differentieel analyse. Beide technieken
verbeteren de resultaten met een Nash Sutcliff efficiëntie van respectievelijk 81% en 86%.
De standaard fout bij de geografische differentieel analyse was 41 mm lager dan bij de
regressie analyse.
Het voordeel van de geografische differentieel analyse is dat de verschillen in neerslag
gebruikt worden voor de ruimtelijke interpolatie en niet de absolute waarden van de
neerslag. De verschillen geven bepaalde onderliggende terreineigenschappen weer, en de
correcties op specifieke plaatsen zijn dus groter dan op andere plaatsen. De afwijkingen in
gemeten en geschatte neerslag kan verkleind worden tot 6%. Dit is gelijk aan een
hoeveelheid water van 26 km3 jr-1, wat 17% van de onttrekkingen in de irrigatiesector is
(ca 150 km3 yr-1). Een kleine fout in de neerslagmetingen heeft dus veel effect op de
berekening van de waterbeschikbaarheid.
153
9.4
Pixel oppervlak bodemvocht
Bodemvocht is een belangrijke variabele voor de waterbalans. Onder aride condities zijn
bodemvochtpatronen een directe aanwijzing voor de aanwezigheid van water. Natte
bodems zijn een indicatie voor irrigatiewater of ondiepe grondwaterstanden. Het is bijna
onmogelijk om bodemvochtinformatie op grote schaal te verkrijgen met in situ sensoren,
en daarom is het niet gebruikelijk om bodemvocht mee te nemen in de besluitvorming van
het waterbeheer. Passieve microgolf sensoren aan boord van bijvoorbeeld de AMSR-E,
SMOS en Feng Yung geven gratis mondiale schattingen voor het dagelijkse bodemvocht.
Deze sensoren leveren continue bodemvochtschattingen zonder beïnvloeding van de
weercondities. Dit is dan ook de hoofdreden om een een nieuw type verdampingsalgoritme
(ET) te gaan inzetten (zie volgende sectie).
Oppervlak bodemvocht is geschat door middel van inversietechnieken op de ‘brightness’
temperatuur gemeten met de satelliet. De techniek is niet foutloos, daarom was het nodig
om het satellietbodemvocht te valideren voordat het gebruikt kan worden om andere
hydrologische processen te schatten. Omdat er geen in situ bodemvochtmetingen aanwezig
zijn in het stroomgebied, was een klassieke validatietechniek niet mogelijk. Alternatieve
validatietechnieken zijn nodig om vertrouwen te winnen in de satelliet
bodemvochtproducten. De reactie van vegetatie op bodemvocht alsmede bodemvocht op
neerslag is bestudeerd om een realistisch gedrag van het bodemvocht te verklaren
(hoofdstuk 6).
Er zijn goede relaties tussen TRMM neerslag en AMSR-E top bodemvocht in de
landgebruik klasses ‘geïrrigeerd’, ‘schaarse vegetatie’, en ‘kale grond’ gevonden. De
neerslaggebeurtenissen in deze landgebruik klassen hebben een groot verband met het
bodemvocht gemeten met AMSR-E. Voor geïrrigeerd land was dit verband lager, vanwege
de extra levering van irrigatiewater – en dus verstoringen in de relatie met neerslag. Op
jaarbasis is er een betere relatie tussen AMSR-E gemiddeld bodemvocht en TRMM
geaccumuleerde neerslag (Spearman rank correlatiecoëfficiënt rs=0.74) dan tussen TRMM
geaccumuleerde neerslag en NDVI (rs=0.70), wat verklaarbaar is door bodem fysische
processen. Gemiddeld bodemvocht en NDVI hebben een grotere correlatie (rs=0.85)
vergeleken met neerslag van TRMM en NDVI (rs=0.70), wat ook te verwachten was.
Tussen de tijdseries van bodemvocht en NDVI werd een tijdsverschil geobserveerd. Dit
verschil was te verwachten vanwege een vertraagde reactie van de vegetatie op het vocht
in de wortelzone. Het tijdsverschil varieerde tussen nul en 60 dagen en was over het
algemeen langer tijdens het natte kharif seizoen. Tijdens het droge rabi seizoen werd een
Pearson r > 0.60 gevonden voor 75% van de gevallen met een tijdsverschil van nul tot 40
dagen. Tijdens het natte kharif seizoen was dit 81% voor de gevallen met een verschil van
20 tot 60 dagen. De maximale waarde voor het bovenste bodemvocht van 0.35 tot 0.45
cm3 cm-3 voor een pixel was gelijk aan de verwachte toplaag
verzadigingsbodemvochtwaarde op basis van bodemtextuurkaarten en pedo-transfer
functies. Hogere waarden kwamen in de overstroomde gebieden en in de rijstvelden voor.
Dit suggereert dat absolute AMSR-E waarden bodemvocht juist beschrijven tijdens natte
bodemcondities.
154
9.5
Pixel verdamping
Verdamping is de meest dominante uitgaande term van de waterbalans van begroeid land
in semi-aride klimaten. Omdat verdamping gedeeltelijk bestuurd kan worden, heeft het de
mogelijkheid om de hoeveelheid beschikbaar water voor landbouw, bossen, moerassen, en
andere landtypes te controleren. De werkelijke verdamping kan niet worden bepaald uit
standaard meteorologische reeksen. Neerslag kan direct worden gemeten met goedkope
regenmeters, maar verdamping niet (behalve op specifieke locaties die zijn uitgerust met
geavanceerde instrumenten). Ondanks dat waterinstellingen en internationale
stroomgebiedscommissies de verplichting hebben om de waterhuishouding te regelen,
gebruiken zij geen verdampingsinformatie op stroomgebiedsschaal. De financiële
beperkingen in ontwikkelingslanden staan de installatie van permanente verdampingsflux
observatietorens in de weg. In de Indus Basin is geen enkele fluxtoren aanwezig, terwijl
een zevende van de wereldbevolking hier leeft en een volume van 496 km3 water elk jaar
verdampt.
Daarom is een nieuwe methode (ETLook) getest, die ruimtelijke schattingen van
verdamping geeft op basis van satellietmetingen en de energiebalans van het
landoppervlak. In de literatuur zijn er verschillende ET-algoritmes beschikbaar, maar het
specifieke voordeel van dit algoritme is dat het continue schattingen van ET gedurende het
gehele jaar geeft, zonder dat het wordt beïnvloed door het weer. ETLook maakt gebruik
van twee bronnen, waaruit informatie over onnuttige evaporatie (E) en nuttige transpiratie
(T) kan worden verkregen door gebruik te maken van bodemvocht op basis van
microgolfmetingen. Microgolf radiometers wordt het minst beïnvloed door bewolking en
kan daarom dus continue bodemvocht informatie geven, zelfs tijdens de moesonperiode en
voor hooggelegen gebieden met aanhoudend bewolking. Dit onderzoek was een eerste
poging om microgolftechnieken te gebruiken om de verdamping nauwkeurig te schatten
over grote gebieden van de Indus Basin door gebruik te maken van openbare datasets.
Wegens technische problemen, is de AMSR-E data niet beschikbaar sinds 4 oktober 2011.
Echter passieve microgolf radiometers, zoals SMOS (Europese ruimtevaart organisatie) of
Feng Yung (Chinese ruimtevaart organisateie) leveren ook vergelijkbare
bodemvochtschattingen.
De geschatte verdamping op 1 km pixel resolutie correleerde goed met enkele historische
lysimeter metingen, Bowen ratio metingen en eerder uitgevoerd remote sensing
onderzoeken (R2 van 0.70 tot 0.76 op jaarschaal; RMSE van 0.29 en 0.45 mm d-1). De
vergelijking laat zien dat de verdampingflux op pixelschaal per dag, 8-daags, of
maandelijkse tijdsschaal geschat kan worden. De totale stroomgebiedsverdamping in 2007
was 496 km3jr-1 terwijl de regen 443 km3jr-1 was. Dit laat zien dat de er meer water
verdampt dan dat er ontvangen wordt door regenval. Meerdere regionale
overheidsinstanties en de zwaartekrachtmetingen van de Grace satelliet suggereren ook
een netto bergingsverschil dat gerelateerd is aan netto grondwateronttrekkingen, dalende
grondwaterniveaus, sneeuwsmelt en terugtrekkende gletsjers. De enige oplossing om het
Indus stroomgebied duurzamer te maken is het verminderen van verdamping en
voornamelijk de onnuttige verdamping (E).
Vlakdekkende informatie over verdamping kunnen watermanagers en besluitvormers
enorm helpen bij de allocatie van het water. Het is ook belangrijk om kennis over de netto
155
waterproductie (R>ET) en waterconsumptie (ET>R) gebieden te vergaren. De afvoer van
de netto water producerende gebieden kan worden beheerd door veranderingen in het
landgebruik en aangepaste landbouwmethodes in de hand te houden. Het concept van
‘green water credits’ is gebaseerd op bepaalde overeengekomen bovenstroomse
ecosysteem diensten te hanteren waar er voldoende rivierafvoer wordt gegenereerd. Netto
water consumptiegebieden, zoals irrigatiesystemen, kunnen worden gecontroleerd door het
regelen van de waterverdeling en onttrekkingen. De irrigatiesector wordt vaak bekritiseerd
over inefficiënt watergebruik. Door het vergelijken van de wateronttrekking met de
waterconsumptie, kunnen schattingen van de niet-geconsumeerde waterstromen gemaakt
worden. De toevoer van irrigatiewater in Pakistan is gefundeerd op gelijkwaardige toegang
tot water voor alle boeren. Zonder twijfel is zo’n aspiratie overal in de wereld moeilijk te
bereiken en met name in gebieden waar water beperkt is. Boerengemeenschappen beweren
vaak dat zij niet het volume water krijgen waar zij recht op hebben. Verdampingswaardes
op pixelniveau is een nieuwe informatiebron, die gebruikt kan worden om dit te
controleren. Niet alleen om te bevestigen dat zij water ontvangen, maar tegelijkertijd ook
om te verifiëren dat zij water niet-nuttig en niet-productief gebruiken.
9.6
Pixel grondwateronttrekking
Ongecontroleerde en forse grondwateronttrekkingen is een groot waterproblemen in de
Indus Basin. Een passende schatting van de grondwateronttrekkingen kan worden gemaakt
met de ruimtelijke tools die in dit proefschrift worden gepresenteerd. Met ruimtelijk
schattingen van neerslag en verdamping, blijkt het mogelijk te zijn om de hoeveelheid
irrigatiewater te bepalen dat nodig is om bepaalde verdampingswaarden te bereiken. Een
hydrologisch model is gebruikt om dit doel te bereiken. Pixel informatie over topografie,
landgebruik, bodems, neerslag en verdamping is gebruikt als input data voor het
hydrologische SWAT model. SWAT is een bekend en algemeen aanvaard gedistribueerd
hydrologisch model dat vele relevante hydrologische, bodemfysische en bio-fysische
processen beschrijft. Dit model is gekozen omdat het vrij beschikbaar is, mondiaal
toepasbaar is en het de mogelijkheid heeft om atmosferische, landoppervlakte, bodem,
grondwater en stromingsprocessen te berekenen.
De parametrisatie en kalibratie van het SWAT model werd ondersteund door kaartlagen
van 8-daagse verdamping verkregen uit het ETLook model. Voor elke hydrologische
response eenheid werd een kalibratie procedure uitgevoerd met variabele irrigatiewater
toevoer en variabele bodemfysische eigenschappen. Het gekalibreerde model werd
vervolgens toegepast voor het genereren van kaarten van de irrigatiewatergift per perceel,
de afvoer en de gecombineerde drainage en percolatie. Dit verschaft essentiële inzichten
in de verdeling van het irrigatiewater naar geconsumeerd en niet-geconsumeerde
watergebruik van alle geïrrigeerde velden in de Indus Basin op een uniforme manier.
De uitstroming vanuit de belangrijkste reservoirs naar de hoofdirrigatiekanalen was 151
km3jr-1. Door het koppelen van de hoeveelheid kanaalwater dat aankomt bij een
irrigatieveld (113 km3 jr-1) met de totale irrigatiegift geschat op basis van het SWAT model
(181 km3 jr-1), is het mogelijk om de totale grondwateronttrekking te kwantificeren (68
km3 jr-1). De verdeling over de landen is dat Pakistan elk jaar 40 km3 levert en India 28
km3. Tabel 9.1 geeft een overzicht van de bulk waterbalans van de geïrrigeerde gebieden
van de Indus Basin.
156
Tabel 9.1: De waterbalans van alle geïrrigeerde gebieden gecombineerd en beschouwd
als een enkel ‘groot veld’ met een omvang van 26.02 miljoen/ ha
Instroming
km3 jr-1
Uitstroming
km3jr-1
Neerslag
117
Afvoer
25
Watertoevoer kanalen India 36
Interceptie verdamping
5
Watertoevoer kanalen
77
Verdamping uit neerslag 99
Pakistan
Grondwaterlevering India
28
Verdamping uit irrigatie
154
Grondwaterlevering
40
Drainage en percolatie
19
Pakistan
van velden
Totaal
298
298
De vectorkaarten van het kanaalwater voor elk irrigatieservicegebied zijn gesupponeerd op
de vectorkaarten van de afvoer en drainage/percolatie en de rasterkaarten van neerslag en
verdamping. Dit levert een kaart met 1 km pixels op van grondwateronttrekkingen voor de
gehele Indus Basin (Figuur 9-1). Dit is de eerste kaart met zo’n gedetailleerd niveau. Het
laat de hotspot gebieden zien met een resolutie van 100 ha. Informatie over
grondwateronttrekkingen kan dus nu openbare kennis worden.
Figuur 9-1 Schematisch diagram van de databronnen gebruikt voor het verkrijgen van
pixelinformatie over grondwateronttrekkingen met een resolutie van 1km.
De grondwateronttrekkingen kunnen nog verder worden onderzocht voor het bepalen van
de grondwateruitputting. Grondwateruitputting (bruto onttrekking 68 km3 jr-1 minus
aanvoer vanuit percolerence velden 19 km3 jr-1,en kanalen 38 km3 jr-1) was 31 km3 jr-1. De
terugstroming van grondwater van 20 km3 jr-1 in de geïrrigeerde gebieden van de Indus
Basin is in de analyse meegenomen. De werkelijke grondwateruitputting is groter door de
uitstroming van grondwater naar het riviersysteem. De grootste grondwateruitputting vindt
plaats in de Punjab provincie van India (350-800 mm jr-1). De staat Haryana in India wordt
ook bedreigt door grondwateruitputting. De enige oplossing om toegang tot water voor
voedsel en de omgeving te kunnen waarborgen is door het reduceren van het netto
grondwatergebruik tot vrijwel niets over een langere periode. Figuur 9-2 geeft een beeld
van de verschillende waterbalans componenten apart voor de geïrrigeerde gebieden van de
Indus Basin gelegen in Pakistan en India.
157
Figuur 9-2 Componenten van de waterbalans van de Pakistaanse en Indiaase geïrrigeerde
delen van de Indus Basin.
9.7
Nieuwe gegevensbronnen
Klassieke gegevens (bijvoorbeeld data van overheidsinstanties) zijn over het algemeen
puntmetingen en zijn vaak moeilijk toegangbaar. Remote sensing informatie is een goed
alternatief. Het is openbaar en iedereen heeft toegang tot de ruwe satelliet informatie.
Gegevens van een internationale vloot van sensoren kan gevonden worden in de Earth
Observing System Data and Information System (EOSDIS). Acht Data Active Archive
Centers (DAAC’s), die een ruime verscheidenheid aan aardwetenschappelijke disciplines
vertegenwoordigen, zijn operationeel onder de NASA om te verwerken, archiveren en
distribueren van EOSDIS gegevens. De Earth Resources Observation System (EROS)
Data Center van de USGS biedt tevens toegang tot gegevens over land fysische processen
verkregen vanuit zowel satellieten als vliegtuigen.
De meeste satellietgegevens met een relatief grove resolutie, wat voldoende is voor de
stroomgebiedsschaal, is vrij beschikbaar. Gegevens met een fijnere resolutie is
beschikbaar tegen betaling. De NASA en ESA voeren het beleid om de kosten voor
beelden laag te houden, zodat iedereen gebruik kan maken van deze gegevens. Een paar
voorbeelden van internationaal open databases kan gevonden worden ot http:// www. daac.
gsfc. nasa. gov,https:// www. wist. echo. nasa. Gov /~wist /api/imswelcome/ of
http://www.nsidc.org/data/ae_land3.html.
In Tabel 9.2 zijn de klassieke bronnen met bijbehorende problemen en alternatieve
oplossing van gegevens samengevat. Het gebruik van satellietgegevens als een alternatief
om uit de eerste hand kennis over hydrologie, landbouw, milieu en geografie te verkrijgen
is onderzocht in dit proefschrift. Klimaatgegevens worden gemeten met standaard
158
meteorologische stations. Vlakdekkende metingen maken het mogelijk deze datasets op
een hogere tijd en ruimteschaal te verkrijgen.
Tabel 9.2 Toepasbaarheid van pixel informatie voor de generatie van geografische
bestanden in het grote en data schaarse stroomgebied van de Indus.
Onderwerp
Landgebruik
Akkerbouw
gebieden
Biomassa
productie
Gewas
opbrengst
Regenval
Sneeuwbedek
king
Bodemvocht
Verdamping
Zonnestraling
Klassieke bronnen
voor de gegevens
- Globale
databases*
- Internationale
organisaties§
- Overheidsinstelli
ngen
- Overheids
instellingen
- Internationale
organisaties
Gerelateerde problemen
- Zeer algemen klasses
van landgebruik
- Geen informative over
specifieke
gewasrotaties
- Verouderd
- Geen real time
informatie
- Late verspreiding van
gegevens
- Data in tabelvorm
- Geen directe metingen
voorhanden
Alternatieve oplossing en
toepasbaarheid
NDVI tijdseries op 1 km
pixels met intervallen van 8
tot 10 dagen
Landgebruikskaart
- Globale netto
primaire
productie kaarten
- Statistieken van
- Late verspreiding van
overheidsinstellin
gegevens
- Informatie alleen per
gen
administratieve
eenheid
- Afwezigheid van
vlakdekkende
gegevens
- Meteorologische - Punt metingen
afdeling
- Schaarse regenmeters
- World
netwerken (<4 meters/
meteorological
10,000km2)
- Geen juiste weergave
organization
- Mondiale
van ruimtelijke
gegevensbestand
hydrologie
en‡
- Mondiale
- N.v.t.
gegevens
Geavanceerde algoritmes
om ruwe data om te zetten
in kwalitatieve informatie
Pixel informatie van
landgebruik en biomassa
productie met
gewasopbrengst op
pixelschaal
- Specifice
projecten
- Veld
experimenten
- Veld
experimenten
- Geen fluxtorens
in de Indus Basin
- Meteorologische
afdeling
- N.v.t.
Passieve microwave en
radar satellieten
- N.v.t.
Geavanceerde algoritmes,
zoals ETLook
- Puntmetingen
beschikbaar via enkele
stations
Geostationaire en polar
orbiting satellieten die de
bewolking meten
3-uurlijkse, dagelijkse en
maandelijkse mondiale
schattingen van regen van
de TRMM satelliet.
Optische en radar
satellieten
159
Grondwater
- Grondwater
monitoring
organisaties
Reservoir
niveaus
- Enkele meting
- Niet genoeg voor
enige
stroomgebiedsstudie
- Punt gegevens
beschikbaar via
schaarse
peilbuisnetwerken.
- Enkel twee metingen
per jaar
- Data niet ontsloten
- Enkele meting is
onvoldoende
- Data niet beschikbaar
Pixel informatie over
gerelateerde hydrologische
variabele kunnen
ruimtelijke schattingen van
grondwateronttrekkingen
geven
Radar en lidar hoogtmeters
*
GLCC-IGBP, Glob Cover-ESA, GLC2000, § IWMI, ¶ GMAO-MERRA, ‡ GPCC, CMAP,
ERA
De voornaamste conclusie is daarom dat archieven met satellietmetingen vaker zouden
moeten worden geraadpleegd om de waterhuishouding van gebieden met schaarse
gegevens te onderzoeken. Gegevens over landgebruik, neerslag, bodemvocht en
verdamping kunnen via indirecte afleidingen worden verkregen. Integratie met andere data
bronnen zoals GIS (bijvoorbeeld kanaalstromingen) en hydrologische modellen
(bijvoorbeeld irrigatietoevoer, afvoer, drainage/percolatie) heeft de voorkeur. Bij
zorgvulding uitgevoerde berekeningen heeft de ruimtelijke dataset de potentie om de
kennis over de hydrologie van grensoverschrijdende stroomgebieden met hoge klimaat
variabiliteit en schaarse gegevens, drastisch te verbeteren.
Instituten, organisaties of consultants zouden de ruwe satelliet gegevens kunnen
verwerken tot waterbeheersinformatiesystemen voor stroomgebieden met nationale en
internationale
rivieren
en
aquifers.
Opkomende
ontwikkelingen
in
de
ruimtevaarttechnologie, geografische informatie systemen en hun toepassing, samen met
de ontwikkelingen in numeriek hydrologisch modelleren, zou moeten leiden tot het
maximaliseren van de voordelen voor alle belanghebbenden. In dit proefschrift is een
‘Indus Basin Data Generator’ als voorbeeld gebruikt voor andere internationaal gedeelde
stroomgebieden. Eenzelfde procedure zou elders ook toegepast kunnen worden, omdat de
enige databron van de verantwoordelijke overheid de wateruitstroming van de grote
reservoirs was. Alle andere data was ontleent van algemeen toegankelijke gegevens
bestanden.
9.8
Toepassingsontwikkeling
De mondiale landbouwkundige productie is kwetsbaar voor klimaatsverandering. De
opwarming van de aarde heeft in de afgelopen decennia een aantal componenten van de
hydrologische kringloop en systemen beïnvloed. Bijvoorbeeld de intensiteit en extremen
van neerslag; wijdverbreid smelten van gletsjers; toename van niet-nuttige verdamping; en
veranderingen in bodemvocht en afvoer. Daarom is het nog belangrijker om de
internationale wateren op een integrale wijze te beheren. Het waterbeheer van
internationale wateren heeft invloed op het lokale, regionale en mondiale niveau en heeft
een standaard kennisraamwerk nodig dat de waterhuishouding en de condities van het
waterbeheer beschrijft. De introductie van water accounting methodes door de Verenigde
160
Naties en IWMI is een stap in de goede richting om wederzijds begrip te bevorderen over
de gevoerde waterhuishouding. Een sterker institutioneel en juridisch raamwerk is
noodzakelijk om het beheer van internationale stroomgebieden te verbeteren. Het is
waarschijnlijker dat dit gebeurd als de ondersteunende gegevens beschikbaar en
betrouwbaar zijn.
Dit proefschrift onderschrijft de notie dat internationale stroomgebiedsinitiatieven en
commissies zouden moeten kunnen vertrouwen op transparante en onbevooroordeelde
satellietmetingen en de waterhuishoudkundige informatie die daarvan is afgeleid. De
beschikbaarheid van ruimtelijk gedistribueerde en temporele consistente informatie over
verschillende hydrologische parameters opent compleet nieuwe mogelijkheden voor het
bestuderen van hydrologische processen, wateruitputting, voedselzekerheid en
omgevingsontwikkeling in internationale stroomgebieden. Het opent een nieuw protocol
waar centrale overheden en internationale agentschappen uniforme informatie krijgen. Er
zijn een aantal hydrologische uitdagingen die moeten worden meegenomen in toekomstig
integraal waterbeheer.
Een van de grootste uitdagingen is de toenemende watervraag door bevolkingsgroei en de
groeiende behoefte naar voedselzekerheid, die de druk op aquifers doet toenemen.
Grondwateronttrekking is een groot probleem geworden, en is verergerd door de
klimaatverandering. Het is de verwachting dat grondwater en gerelateerde milieu kwesties
erger zullen worden in de nabije toekomst. Een integrale aanpak van gecombineerd
gebruik en beheer van oppervlakte en grondwater is noodzakelijk. Grondwaterbeheer zou
buiten de politieke grenzen van de aangrenzende landen, die een aquifer delen, moeten
plaatsvinden. Ongecontroleerde en intensieve grondwateronttrekkingen in een gedeelte
van het stroomgebied beïnvloed de waterbeweging en dus de beschikbaarheid van water in
andere delen op een negatieve wijze. Het mandaat van de riviercommissies moet worden
herzien en de UN voorgestelde "Law of transboundary aquifers" dient te worden
goedgekeurd door de lidstaten en moeten ook worden opgenomen in de ‘Indus Water
Treaty’.Het wetsontwerp heeft duidelijke instructies over beheer, restauratie en
gegevensuitwisseling tussen de aangrenzende landen van internationaal gedeelde aquifers.
De opwarming van de aarde kan smeltende gletsjers tot gevolg hebben, wat resulteert in
enorme rivierstromen gedurende de eerste helft van de volgende eeuw, gevolgd door een
afname van 30% tot 40% in de rivierafvoeren tegen het einde van deze eeuw. Daarom, om
de dreigende calamiteiten van overstromingen en droogte te bestrijden, is het belangrijk
om nieuwe grote reservoirs te bouwen onder het mandaat van de ‘Indus Water Treaty’.
Deze reservoirs zullen niet enkel dienen als wateropslag en temporele regulatie, maar
dienen ook als bron voor waterkracht.
Naast de bouw van grote reservoirs, zullen ook kleine dammen moeten worden beschouwd
in die gebieden waar de onttrekking de aanvulling overschrijd. Zulke ingrepen zullen
helpen om de uitgeputte aquifers weer aan te vullen en verzekerd een duurzaam beheer
voor toekomstige generaties.
Een toename in de waterproductiviteit (gewasopbrengst per eenheid van geconsumeerd
water) wordt nodig geacht om de groeiende bevolking te voeden. Dit zou bereikt kunnen
worden door het reduceren van de niet-nuttige verdamping (263 km3 jr-1) van braakliggend
161
land en het minimaliseren van niet-nuttige transpiratie van onkruid en gras dat tussen de
gewassen groeit.
Een flexibele aanpak in de irrigatiewatertoevoer op basis van de waterbehoefte van
gewassen is noodzakelijk in plaats van een gefixeerd waterverdelingssysteem wat
momenteel gebruikelijk is in de Indus Basin. Dit kan worden bereikt door het continue
monitoren van verdamping over de geïrrigeerde gebieden. Het is spijtig, dat er geen enkele
permanente fluxtoren in het stroomgebied aanwezig is om de werkelijke verdamping te
meten, wat essentieel is voor efficiënte irrigatie. Enkele fluxtorens zouden in het
stroomgebied geïnstalleerd moeten worden om verdamping te monitoren samen met het
monitoren van verdamping met behulp van satellieten.
Er is behoefte om het landgebruik continue te monitoren, omdat elke wijziging in
landgebruik de benedenstroomse waterstroom beïnvloed. Antropogene wijzigingen in
landgebruik kunnen resulteren in degradatie van land dat ernstige gevolgen heeft voor
benedenstroomse gebieden. Bijvoorbeeld, bovenstroomse ontbossing kan resulteren in
veranderende regenpatronen. Bodemerosie van bovenstroomse stroomgebieden is een
belangrijke reden voor sedimentatie in grote reservoirs. Hiervoor kan het relatief nieuwe
concept van ‘green water credits ’ een rol spelen door bovenstroomse boeren te prikkelen
om de ontbossing te stoppen en bodem en waterconservatietechnieken gebruiken uit te
voeren die nut hebben voor benedenstroomse watergebruikers.
De beperking van dit onderzoek is het gebruik van slechts een enkel jaar aan
satellietgegevens. De reden was om de academische mogelijkheden en de beperkingen in
het gebruik van pixel informatie in hydrologische studies te verkennnen. Het was niet de
intentie om de waterhuishoudingproblemen in de Indus Basins op te lossen. Voor goede
besluitvorming wordt het aangeraden om de toekomstige analyse uit te breiden met
langere tijdseries.
162
10 References
Adegoke, J.O., Carleton, A.M., 2002. Relations between soil moisture and satellite
vegetation indices in the U.S. corn belt. Journal of Hydrometeorology 3, 395-405.
Adeyewa, Z.D., Nakamura, K., 2003. Validation of TRMM radar rainfall data over major
climatic regions in Africa. Journal of Applied Meteorology and Climatology 42, 331347.
AghaKouchak, A., Nasrollahi, N., Habib, E., 2009. Accounting for uncertainties of the
TRMM satellite estimates. Remote Sensing 1, 606 - 619, doi:610.3390/rs1030606.
Agrawal, S., Joshi, P.K., Shukla, Y., Roy, P.S., 2003. SPOT VEGETATION multi
temporal data for classifying vegetation in South Central Asia. Current Science 84,
1440-1448.
Ahmad, M.-u.-D., Bastiaanssen, W.G.M., Feddes, R.A., 2002. Sustainable use of
groundwater for irrigation: a numerical analysis of the subsoil water fluxes. Irrigation
and Drainage 51, 227-241.
Ahmad, M.D., 2002. Estimation of net groundwater use in irrigated river basins using geoinformation techniques: A case study in Rechna Doab, Pakistan. PhD dissertation,
Wageningen University, Wageningen, the Netherlands, p. 143.
Ahmad, M.D., Bastiaanssen, W., Feddes, R., 2005. A new technique to estimate net
groundwater use across large irrigated areas by combining remote sensing and water
balance approaches, Rechna Doab, Pakistan. Hydrogeology Journal 13, 653-664.
Ahmad, M.D., Turral, H., Nazeer, A., 2009. Diagnosing irrigation performance and water
productivity through satellite remote sensing and secondary data in a large irrigation
system of Pakistan. Agricultural Water Management 96, 551-564.
Ahmad, S., 2009. Water availability in Pakistan: Paper presented by Dr. Shahid Ahmad,
Member PARC. In National Seminar on "Water Conservation, Present Situation and
Future Strategy". Project Management & Policy Implementation Unit (PMPIU) of the
Ministry of Water & Power, Islamabad, Pakistan, p. 114.
Ahrens, B., 2006. Distance in spatial interpolation of daily rain gauge data. Hydrology and
Earth System Sciences 10, 197-208.
Alam, M., Bhutta, M.N., 1996. Availability of water in Pakistan during 21st century.
Proceedings of the International Conference on Evapotranspiration and Irrigation
Scheduling. November 3-6, 1996, San Antonio,Texas,USA.
Allen, R., Irmak, A., Trezza, R., Hendrickx, J.M.H., Bastiaanssen, W.G.M., Kjaersgaard,
J., 2011. Satellite-based ET estimation in agriculture using SEBAL and METRIC.
Hydrological Processes 25, 4011–4027.
Allen, R.G., Hendrickx, J., Bastiaanssen, W.G.M., Kjaersgaard, J., Irmak, A., Huntington,
J., 2010. Status and continuing challenges in operational remote sensing of ET. ASABE
5th National Decennial Irrigation Conference Proceedings, Phoenix Convention Center,
Phoenix, Arizona USA IRR10-9971.
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop evapotranspiration-guidelines
for computing crop water requirements. FAO Irrigation and Drainage Paper, No. 56.
FAO Irrigation and Drainage Paper, No. 56, FAO,Rome, Italy.
Allen, R.G., Pruitt, W.O., Wright, J.L., Howell, T.A., Ventura, F., Snyder, R., Itenfisu, D.,
Steduto, P., Berengena, J., Yrisarry, J.B., Smith, M., Pereira, L.S., Raes, D., Perrier, A.,
Alves, I., Walter, I., Elliott, R., 2006. A recommendation on standardized surface
163
resistance for hourly calculation of reference ETo by the FAO56 Penman-Monteith
method. Agricultural Water Management 81, 1-22.
Allen, R.G., Tasumi, M., Trezza, R., 2007. Satellite-Based Energy Balance for Mapping
Evapotranspiration with Internalized Calibration (METRIC)---Model. Journal of
Irrigation and Drainage Engineering 133, 380-394.
Anagnostou, E.N., Krajewski, W.F., Smith, J., 1999. Uncertianty quantification of meanareal radar-rainfall estimates. Journal of Atmospheric and Oceanic Technology 16, 206215.
Anders, A.M., Gerard, H.R., Bernard, H., David, R.M., Noah, J.F., Jaakko, P., 2006.
Spatial patterns of precipitation and topography in the Himalaya. In: Willett, S.D.,
Hovius, N., Brandon, M.T., Fisher, D. (Eds.), Tectonics, Climate, and Landscape
Evolution. Geological Society of America, Special Paper 398, pp. 39-53.
Anderson, M.C., Norman, J.M., Mecikalski, J.R., Otkin, J.A., Kustas, W.P., 2007. A
climatological study of evapotranspiration and moisture stress across the continental
United States based on thermal remote sensing: 2. Surface moisture climatology. Journal
of Geophysical Research 112, D11112.
Andersson, L., Wilk, J., Todd, M.C., Hughes, D.A., Earle, A., Kniveton, D., Layberry, R.,
Savenije, H.H.G., 2006. Impact of climate change and development scenarios on flow
patterns in the Okavango River. Journal of Hydrology 331, 43-57.
Andréassian, V., Perrin, C., Michel, C., Usart-Sanchez, I., Lavabre, J., 2001. Impact of
imperfect rainfall knowledge on the efficiency and the parameters of watershed models.
Journal of Hydrology 250, 206-223.
Angstrom, A., 1924. Solar and terrestrial radiation. Quarterly Journal of the Royal
Meteorological Society 50, 121-126.
Archer, D.R., Fowler, H.J., 2004. Spatial and temporal variations in precipitation in the
Upper Indus Basin, global teleconnections and hydrological implications. Hydrology
and Earth System Sciences 8, 47-61.
Arino, O., Bicheron, P., Achard, F., Latham, J., Witt, R., Weber, J.-L., 2008.
GLOBCOVER: The most detailed portrait of Earth. European Space Agency,Bulletin
136,November 2008.
Arnold, J.G., Srinivasan, R., Muttiah, R.S., Williams, J.R., 1998. Large-area hydrologic
modeling and assessment: Part I. Model development. Journal of the American Water
Resources Association 34, 73−89.
Arshad, M., 2004. Contribution of irrigation conveyance system components to the
recharge potential in Rechna Doab under lined and unlined options. PhD dissertation,
Department of Irrigation and Drainage. University of Agriculture, Faisalabad, Pakistan,
p. 229.
Arshad, M., Ahmed, N., Cheema, J.M., 2008. Modeling approach for the assessment of
recharge contribution to groundwater from surface irrigation conveyance system.
Irrigation and Drainage Systems 22, 67-77.
Arshad, M., Cheema, M.J.M., Ahmed, S., 2007. Determination of lithology and
groundwater quality using electrical resistivity survey. International Journal of
Agriculture and Biology 9, 143-146.
Arshad, M., Choudhry, M.R., Ahmed, R.N., 2005. Groundwater recharge contribution
from various components of irrigation water conveyance system of Rechna Doab of
Punjab-Pakistan. Pakistan Journal of Water Resources 9, 17-24.
164
ASCE, 1996. Hydrology handbook. American Society of Civil Engineers Task Committee
on Hydrology Handbook, ASCE Publications No 28. pp 784.
Babak, O., Deutsch, C.V., 2009. Statistical approach to inverse distance interpolation.
Stochastic Environmental Research and Risk Assessment 23, 543-553.
Barnard, J.C., Long, C.N., 2004. A simple empirical equation to calculate cloud optical
thickness using shortwave broadband measurements. Journal of Applied Meteorology
43, 1057-1066.
Barnes, R.A., Barnes, W.L., Lyu, C.-H., Gales, J.M., 2000. An overview of the visible and
infrared scanner radiometric calibration algorithm. Journal of Atmospheric and Oceanic
Technology 17, 395-405.
Barros, A.P., Kim, G., Williams, E., Nesbitt, S.W., 2004. Probing orographic controls in
the Himalayas during the monsoon using satellite imagery. Natural Hazards and Earth
System Sciences 4, 29-51.
Bartalis, Z., Scipal, K., Wagner, W., 2006. Azimuthal anisotropy of scatterometer
measurements over land. IEEE Transactions on Geoscience and Remote Sensing 44,
2083-2092.
Bastiaanssen, Bandara, 2001. Evaporative depletion assessments for irrigated watersheds
in Sri Lanka. Irrigation Science 21, 1-15.
Bastiaanssen, W.G.M., 1998. Remote sensing in water resources management:The state of
the art. International Water Management Institute,Colombo, Sri Lanka.
Bastiaanssen, W.G.M., 2000. Shared water resources information from space: new
management opportunities or unwanted interference? , Inaugural address, International
Institute for Aerospace Survey and Earth Sciences, Enschede, The Netherlands, p. 22.
Bastiaanssen, W.G.M., Allen, R.G., Droogers, P., D'Urso, G., Steduto, P., 2007. Twentyfive years modeling irrigated and drained soils: State of the art. Agricultural Water
Management 92, 111-125.
Bastiaanssen, W.G.M., Cheema, M.J.M., Immerzeel, W.W., Miltenburg, I., Pelgrum, H.,
2012. The surface energy balance and actual evapotranspiration of the transboundary
Indus Basin estimated from satellite measurements and the ETLook model. Water
Resources Research, (Under review).
Bastiaanssen, W.G.M., Molden, D.J., Thiruvengadachari, S., Smit, A.A.M.F.R.,
Mutuwatte, L., Jayasinghe, G., 1999. Remote sensing and hydrologic models for
performance assessment in Sirsa irrigation circle, India. Research Report 27.
International Water Management Institute, Colombo, Sri Lanka.
Bastiaanssen, W.G.M., Pelgrum, H., Soppe, R.W.O., Allen, R.G., Thoreson, B.P.,
Teixeira, A.H., 2008. Thermal-infrared technology for local and regional scale irrigation
analyses in horticultural systems. Acta Horticulture (ISHS) 792, 33-46.
Bastiaanssen, W.G.M., Prathapar, S.A., 2000. Satellite observations of international river
basins for all. In: Florinsky, I. (Ed.), International Archives of Photogrammetry and
Remote Sensing, ISPRO Congress, Amsterdam, the Netherlands, XXXIII Part B7, pp.
439-451.
Bastiaanssen, W.G.M., Van der Wal, T., Visser, T.N.M., 1996. Diagnosis of regional
evaporation by remote sensing to support irrigation performance assessment. Irrigation
and Drainage Systems 10, 1-23.
Bastiaanssen, W.G.M., Zwart, S.J., Pelgrum, H., 2003. Remote sensing analysis. In: Van
Dam, J.C., Malik, R.S. (Eds.), Water productivity of irrigated crops in Sirsa District,
165
India. Integration of remote sensing, crop and soil models and geographical information
systems. WATPRO final report, pp. 85-100.
Batjes, N.H., 1996. Development of a world data set of soil water retention properties
using pedotransfer rules. Geoderma 71, 31-52.
Berg, W., L'Ecuyer, T., Kummerow, C., 2006. Rainfall climate regimes: The relationship
of regional TRMM rainfall biases to the environment. Journal of Applied Meteorology
and Climatology 45, 434-454.
Beven, K., 2006. A manifesto for the equifinality thesis. Journal of Hydrology 320, 18-36.
Beven, K.J., Fischer, J., 1996. Remote sensing and scaling in hydrology. In: Stewart, J.B.,
Engman, T., Feddes, R.A., Kerr, Y. (Eds.), Scaling up in hydrology using remote
sensing. John Wiley, New York.
Bhatt, B.C., Higuchi, A., Nakamura, K., 2005. TRMM observations of the precipitation
around the Himalayan region. "Maximization of the use of satellite data for
understanding the earth environment". The 11th CEReS International Symposium on
Remote Sensing, University Convention Hall, Chiba University, Chiba, Japan.
Biswas, S.K., Gopalan, K., Jones, W.L., Bilanow, S., 2010. Correction of time-varying
radiometric errors in TRMM microwave imager calibrated brightness temperature
products. IEEE Geoscience and Remote Sensing Letters 7, 851 - 855.
Boegh, E., Thorsen, M., Butts, M.B., Hansen, S., Christiansen, J.S., Abrahamsen, P.,
Hasager, C.B., Jensen, N.O., van der Keur, P., Refsgaard, J.C., Schelde, K., Soegaard,
H., Thomsen, A., 2004. Incorporating remote sensing data in physically based
distributed agro-hydrological modelling. Journal of Hydrology 287, 279-299.
Bolten, J.D., Lakshmi, V., Njoku, E.G., 2003. Soil moisture retrieval using the
passive/active L- and S-band radar/radiometer. IEEE Transactions on Geoscience and
Remote Sensing 41, 2792-2801.
Bookhagen, B., Burbank, D.W., 2006. Topography, relief, and TRMM-derived rainfall
variations along the Himalaya. Geophysical Research Letters 33, L08405, 08401-08405.
Bosch, D., Lakshmi, V., Jackson, T., Jacobs, J., Moran, S., 2004. In situ soil moisture
network for validation of remotely sensed data. Geoscience and Remote Sensing
Symposium, IGARSS '04 Proceedings, IEEE International pp. 3188-3190
Braden, H., 1985. Ein energiehaushalts und verdunstungsmodell fur Wasser und
Stoffhaushaltsuntersuchungen landwirtschaftlich genutzer einzugsgebiete. Mittelungen
Deutsche Bodenkundliche Geselschaft 42, 294-299.
Bringi, V.N., Chandrasekar, V., 2001. Polarimetric Doppler weather radar: principles and
applications. Cambridge University Press, Cambridge, U.K.
Brouder, S.M., Hofmann, B.S., Morris, D.K., 2005. Mapping soil pH: accuracy of
common soil sampling strategies and estimation techniques. Soil Science Society of
America Journal 69, 427-442.
Brutsaert, W.H., 1982. Evaporation into the atmosphere: Theory, history, and applications.
Reidel Publishing Co, Dordrecht, the Netherlands.
Burke, E.J., Shuttleworth, W.J., French, A.N., 2001. Using vegetation indices for soilmoisture retrievals from passive microwave radiometry. Hydrology and Earth System
Sciences 5, 671-678.
Buytaert, W., Celleri, R., Willems, P., Bievre, B.D., Wyseure, G., 2006. Spatial and
temporal rainfall variability in mountainous areas: A case study from the south
Ecuadorian Andes. Journal of Hydrology 329, 413-421.
166
Calcagno, G., Mendicino, G., Monacelli, G., Senatore, A., Versace, P., 2007. Distributed
estimation of actual evapotranspiration through remote sensing techniques. Methods and
Tools for Drought Analysis and Management. Springer Netherlands, pp. 125-147.
Camillo, P.J., Gurney, R.J., 1986. A resistance parameter for bare-soil evaporation models.
Soil Science 141, 95-105.
Campbell, J.B., 2002. Introduction to remote sensing. The Guilford Press, New York.
Canton, Y., Sole-Benet, A., Domingo, F., 2004. Temporal and spatial patterns of soil
moisture in semiarid badlands of SE Spain. Journal of Hydrology 285, 199-214.
Cardina, J., Sparrow, D.H., 1996. A comparison of methods to predict weed seedling
populations from the soil seedbank. Weed Science 44, 46-51.
Carlson, T.N., Ripley, D.A., 1997. On the relation between NDVI, fractional vegetation
cover, and leaf area index. Remote Sensing of Environment 62, 241-252.
Chadha, D.K., 2008. Development, management and impact of climate change on
transboundary aquifers of Indus Basin. 4th International Symposium on Transboundary
Water Management, Thessaloniki, Greece, p. 4.
Cheema, M.J.M., Bastiaanssen, W.G.M., 2010. Land use and land cover classification in
the irrigated Indus Basin using growth phenology information from satellite data to
support water management analysis. Agricultural Water Management 97, 1541-1552.
Cheema, M.J.M., Bastiaanssen, W.G.M., 2012. Local calibration of remotely sensed
rainfall from the TRMM satellite for different periods and spatial scales in the Indus
Basin. International Journal of Remote Sensing 33, 2603-2627.
Cheema, M.J.M., Bastiaanssen, W.G.M., Rutten, M.M., 2011. Validation of surface soil
moisture from AMSR-E using auxiliary spatial data in the transboundary Indus Basin.
Journal of Hydrology 405, 137-149.
Chokngamwong, R., Chiu, L., 2006. TRMM and Thailand daily gauge rainfall
comparison. . The 86th AMS Annual Meeting, Atlanta, GA, p. P1.2.
Chokngamwong, R., Chiu, L., Vongsaard, J., 2005. Comparison of TRMM rainfall and
daily gauge data in Thailand. American Geophysical Union H23A-10.
Choudhury, B.J., 1991. Passive microwave remote sensing contribution to hydrological
variables. Surveys in Geophysics 12, 63-84.
Choudhury, B.J., Golus, R.E., 1988. Estimating soil wetness using satellite data.
International Journal of Remote Sensing 9, 1251-1257.
Choudhury, B.J., Reginato, R.J., Idso, S.B., 1986. An analysis of infrared temperature
observations over wheat and calculation of latent heat flux. Agricultural and Forest
Meteorology 37, 75-88.
Christian, H.J., Blakeslee, R.J., Goodman, S.J., Mach, D.A., Stewart, M.F., Buechler,
D.E., Koshak, W.J., Hall, J.M., Boeck, W.L., Driscoll, K.T., Boccippio, D.J., 1999. The
lightning imaging sensor. ICAE 99 - International Conference on Atmospheric
Electricity, Guntersville, United States pp. 746-749.
Christian, H.J., Blakeslee, R.J., Goodman, S.J., Mach, D.M., 2000. Algorithm theoretical
basis document for the lightning imaging sensor. Earth Science Department, National
Aeronautics and Space Science Administration, USA.
Ciach, G., Morrissey, M., Krajewski, W.F., 2000. Conditional bias in radar rainfall
estimation. Journal of Applied Meteorology 39, 1941-1946.
167
Cihlar, J., Latifovic, R., Beaubien, J., Guindon, B., Palmer, M., 2003. Thematic mapper
(TM) based accuracy assessment of a land cover product for Canada derived from SPOT
VEGETATION (VGT) data. Canadian Journal of Remote Sensing 29, 154-170.
Clapp, R.B., Hornberger, G.M., 1978. Empirical equations for some soil hydraulic
properties. Water Resources Research 14, 601–604.
Cohen, J., 1960. A coefficient of agreement for nominal scales. Educational and
Psychological Measurement 20, 37-46.
Condom, T., Rau, P., Espinoza, J.C., 2011. Correction of TRMM 3B43 monthly
precipitation data over the mountainous areas of Peru during the period 1998–2007.
Hydrological Processes, n/a-n/a.
Congalton, R.G., 1991. A review of assessing accuracy of classifications of remotely
sensed data. Remote sensing of Environment 37, 35-46.
Congalton, R.G., Green, K., 1999. Assessing the accuracy of remotely sensed data:
Principles and practices. Boca Raton: Lewis Publications, London.
Cosgrove, W.J., Rijsberman, F.R., 2000. World water vision, making water everybody’s
business. World Water Council, Earthscan Publications Ltd, London, UK.
Cosh, M.H., Jackson, T.J., Bindlish, R., Prueger, J.H., 2004. Watershed scale temporal and
spatial stability of soil moisture and its role in validating satellite estimates. Remote
Sensing of Environment 92, 427-435.
Courault, D., Seguin, B., Olioso, A., 2005. Review on estimation of evapotranspiration
from remote sensing data: From empirical to numerical modeling approaches. Irrigation
and Drainage Systems 19, 223-249.
Crow, W.T., Miralles, D.G., Cosh, M.H., 2010. A quasi-global evaluation system for
satellite-based surface soil moisture retrievals. IEEE Transactions on Geoscience and
Remote Sensing 48, 2516-2527.
Crow, W.T., Zhan, X., 2007. Continental-scale evaluation of remotely sensed soil
moisture products. IEEE Geoscience and Remote Sensing Letters 4, 451-455.
Curran, P.J., Steven, M.D., 1983. Multispectral remote sensing for the estimation of green
leaf area index [and discussion]. Philosophical Transactions of the Royal Society of
London. Series A, Mathematical and Physical Sciences 309, 257-270.
de Fraiture, C., Wichelns, D., 2010. Satisfying future water demands for agriculture.
Agricultural Water Management 97, 502-511.
de Jeu, R.A.M., 2003. Retrieval of land surface parameters using passive microwave
remote sensing. PhD dissertation,Vrije Universiteit, Amsterdam, the Netherlands, p.
122.
de Jeu, R.A.M., Wagner, W., Holmes, T.R.H., Dolman, A.J., van de Giesen, N.C., Friesen,
J.C., 2008. Global soil moisture patterns observed by space borne microwave
radiometers and scatterometers. Surveys in Geophysics 29, 399-420.
De Wit, A.J.W., van Diepen , C.A., 2007. Crop model data assimilation with the Ensemble
Kalman filter for improving regional crop yield forecasts. Agricultural and Forest
Meteorology 146, 38-56.
Dimri, A.P., 2006. Surface and upper air fields during extreme winter precipitation over
the Western Himalayas. Pure and Applied Geophysics 163, 1679-1698.
Din, S.U., Dousari, A.A., Ramdan, A., Ghadban, A.A., 2008. Site-specific precipitation
estimate from TRMM data using bilinear weighted interpolation technique:An example
from Kuwait. Journal of Arid Environments 72, 1320-1328.
168
Dinku, T., Ceccato, P., Grover-Kopec, E., Lemma, M., Connor, S.J., Ropelewski, C.F.,
2007. Validation of satellite rainfall products over East Africa's complex topography.
International Journal of Remote Sensing 28, 1503-1526.
Dobson, M.C., Ulaby, F.T., Hallikainen, M.T., El-Rayes, M.A., 1985. Microwave
dielectric behavior of wet soil—Part II: Dielectric mixing models. IEEE Transactions on
Geoscience and Remote Sensing 23, 35-46.
Dolman, A.J., 1993. A multiple-source land surface energy balance model for use in
general circulation models. Agricultural and Forest Meteorology 65, 21-45.
Draper , C.S., Walker , J.P., Steinle, P.J., de Jeu, R.A.M., Holmes, T.R.H., 2009. An
evaluation of AMSR–E derived soil moisture over Australia. Remote Sensing of
Environment 113, 703-710.
Draper, C.S., Walker, J.P., Steinle, P.J., deJeu, R.A.M., Holmes, T.R.H., 2009. An
evaluation of AMSR–E derived soil moisture over Australia. Remote Sensing of
Environment 113, 703-710.
Droogers, P., 2006. Unpublished data on pedo-transfer functions. Future Water,
Wageningen, the Netherlands.
Droogers, P., Bastiaanssen, W., 2002. Irrigation performance using hydrological and
remote sensing modeling. Journal of Irrigation and Drainage Engineering 128, 11-18.
Droogers, P., Bastiaanssen, W.G.M., Beyazgül, M., Kayam, Y., Kite, G.W., Murray-Rust,
H., 2000. Distributed agro-hydrological modeling of an irrigation system in western
Turkey. Agricultural Water Management 43, 183-202.
Droogers, P., Immerzeel, W.W., Lorite, I.J., 2010a. Estimating actual irrigation application
by remotely sensed evapotranspiration observations. Agricultural Water Management
97, 1351 - 1359.
Droogers, P., Immerzeel, W.W., Lorite, I.J., 2010b. Estimating actual irrigation
application by remotely sensed evapotranspiration observations. Agricultural Water
Management 97, 1351-1359.
Duan, X., Guo, J., Shum, C., van der Wal, W., 2009. On the postprocessing removal of
correlated errors in GRACE temporal gravity field solutions. Journal of Geodesy 83,
1095-1106.
Dunne, S.C., Entekhabi, D., Njoku, E.G., 2007. Impact of multiresolution active and
passive microwave measurements on soil moisture estimation using the Ensemble
Kalman Smoother. IEEE Transactions on Geoscience and Remote Sensing 45, 10161028.
Elhaddad, A., Garcia, L.A., 2008. Surface energy balance-based model for estimating
evapotranspiration taking into account spatial variability in weather. Journal of
Irrigation and Drainage Engineering ASCE 134, 681-689.
Fahlbusch, H., Schultz, B., Thatte, C.D., 2004. The Indus Basin history of irrigation,
drainage and flood management. ICID, New Delhi, India.
Falkenmark, M., Rockström, J., 2006. The new blue and green water paradigm: Breaking
new ground for water resources planning and management. Journal of Water Resources
Planning and Management 132, 129-132.
Fang, W., Chen, J., Shi, P., 2005. Variability of the phenological stages of winter wheat in
the North China Plain with NOAA/AVHRR NDVI data (1982-2000). IEEE 7803-90504, 3124-3127.
169
FAO, 1995. FAO-UNESCO digital soil map of the world and derived soil properties.
UNESCO, Paris.
FAO, 2008. Harmonized world soil database (version 1.0). FAO/IIASA/ISRIC/ISSCAS/JRC, Rome, Italy.
Farrar, T.J., Nicholson, S.E., Lare, A.R., 1994. The influence of soil type on the
relationships between NDVI, rainfall, and soil moisture in semi-arid Botswana. II.
NDVI response to soil moisture. Remote Sensing of Environment 50, 121-133.
Ferraro, R.R., Marks, G.F., 1995. The development of SSM/I rain rate retrieval algorithms
using ground-based radar measurements. Journal of Atmospheric and Oceanic
Technology 12, 755-770.
FFC, 2009. Annual flood report 2008. Federal Flood Commission Pakistan, Islamabad,
Pakistan, p. 57.
Fily, M., Dedieu, J.-P., Surdyk, S., 1995. A SAR image study of a snow-covered area in
the French Alps. Remote Sensing of Environment 51, 253-262.
Foody, G.M., 2002. Status of land cover classification accuracy assessment. Remote
Sensing of Environment 80, 185-201.
Foster, S.S.D., Chilton, P.J., 2003. Groundwater:The processes and global significance of
aquifer degradation. Philosophical Transactions of the Royal Society of London. Series
B, Biological Sciences 358, 1957-1972
Franchito, S.H., Rao, V.B., Vasques, A.C., Santo, C.M.E., Conforte, J.C., 2009. Validation
of TRMM precipitation radar monthly rainfall estimates over Brazil. Journal of
Geophysical Research 114, D02105.
Franke, R., 1982. Scattered data interpolation: tests of some method. Mathematics of
Computation 38, 181-200.
Friedl, M.A., Woodcock, C., Gopal, S., Muchoney, D., Strahler, A.H., Schaaf, C.B., 2000.
A note on procedures used for accuracy assessment in land cover maps derived from
AVHRR data. International Journal of Remote Sensing 21, 1073-1077.
Friesen, J., Rodgers, C., Oguntunde, P.G., Hendrickx, J.M.H., van de Giesen, N., 2008.
Hydrotope-based protocol to determine average soil moisture over large areas for
satellite calibration and validation with results from an observation campaign in the
Volta Basin, West Africa. Geoscience and Remote Sensing, IEEE Transactions on 46,
1995-2004.
Friesen, J.C., 2008. Regional vegetation water effects on satellite soil moisture estimations
for West Africa. PhD Dissertation No 63, Center for Development Research. University
of Bonn, Bonn, Germany, p. 121.
Fujii, H., Koike, T., 2001. Development of a TRMM/TMI algorithm for precipitation in
the Tibetan Plateau by considering effects of land surface emissivity. Journal of the
Meteorological Society of Japan 79, 475-483.
Fuyane, B., Madai, F., 2001. The Hungary-Slovakia Danube River dispute: implications
for sustainable development and equitable utilization of natural resources in
international law. International Journal of Global Environmental Issues 1, 329-344.
Gastélum, J., Valdés, J., Stewart, S., 2010. A System dynamics model to evaluate
temporary water transfers in the Mexican Conchos Basin. Water Resources
Management 24, 1285-1311.
170
Gebremichael, M., Anagnostou, E.N., Bitew, M.M., 2010. Critical steps for continuing
advancement of satellite rainfall applications for surface hydrology in the Nile river
basin. Journal of the American Water Resources Association 46, 361-366.
Gebremichael, M., Krajewski, W.F., 2004. Assessment of the statistical characterization of
small-scale rainfall variability from radar: Analysis of TRMM ground validation
datasets. Journal of Applied Meteorology 43, 1180-1199.
Gharari, S., Hrachowitz, M., Fenicia, F., Savenije, H.H.G., 2011. Hydrological landscape
classification: investigating the performance of HAND based landscape classifications
in a central European meso-scale catchment. Hydrology and Earth System Sciences 15,
3275-3291.
Giri, C., Jenkins, C., 2005. Land cover mapping of greater mesoamerica using MODIS
data. Canadian Journal of Remote Sensing 31, 274-282.
Githui, F., Selle, B., Thayalakumaran, T., 2011. Recharge estimation using remotely
sensed evapotranspiration in an irrigated catchment in southeast Australia. Hydrological
Processes DOI:10.1002/hyp.8274, n/a-n/a.
Gleick, P., 2008. Water conflict chronology. Pacific Institute, Oakland, p. Available online
at: www.worldwater.org/conflictIntro.htm.
Gomez-Plaza, A., Alvarez-Rogel, J., Albaladejo, J., Castillo, V.M., 2000. Spatial patterns
and temporal stability of soil moisture across a range of scales in a semi-arid
environment. Hydrological Processes 14, 1261-1277.
GOP, 2010. Pakistan statistical year book 2010. Federal Bureau of Statistics, Statistics
Division, Islamabad, p. 68.
GOP, 2011. Pakistan statistical year book 2011. Agriculture statistics. Federal Bureau of
Statistics, Statistics Division, Islamabad.
Gravetter, F.J., Wallnau, L.B., 2006. Statistics for the behavioral sciences. Thomson
Wadsworth Publishers, USA.
Grimes, D.I.F., Pardo-Iguzquiza, E., Bonifacio, R., 1999. Optimal areal rainfall estimation
using raingauges and satellite data. Journal of Hydrology 222, 93-108.
Grody, N.C., 1991. Classification of snow cover and precipitation using the Special Sensor
Microwave Imager. Journal of Geophysical Research 96, 7423-7435.
Gruhier, C., de Rosnay, P., Hasenauer, S., Holmes, T., de Jeu, R., Kerr, Y., Mougin, E.,
Njoku, E., Timouk, F., Wagner, W., Zribi, M., 2010. Soil moisture active and passive
microwave products: intercomparison and evaluation over a Sahelian site. Hydrology
and Earth System Sciences 14, 141-156.
Gu, R.R., Li, Y., 2002. River temperature sensitivity to hydraulic and meteorological
parameters. Journal of Environmental Management 66, 43-56.
Guerschman, J.P., Van Dijk, A.I.J.M., Mattersdorf, G., Beringer, J., Hutley, L.B., Leuning,
R., Pipunic, R.C., Sherman, B.S., 2009. Scaling of potential evapotranspiration with
MODIS data reproduces flux observations and catchment water balance observations
across Australia. Journal of Hydrology 369, 107–119.
Habib, Z., 2004. Scope for reallocation of river waters for agriculture in the Indus Basin.
PhD Dissertation. ENGREF, Montpellier, France.
Haddad, Z.S., Park, K.W., 2010. Vertical profiling of tropical precipitation using passive
microwave observations and its implications regarding the crash of Air France 447.
Journal of Geophysical Research 115, D12129.
171
Hansen, M.C., Defries, R.S., Townshend, J.R.G., Sohlberg, R., 2000. Global land cover
classification at 1 Km spatial resolution using a classification tree approach.
International Journal of Remote Sensing 21, 1331-1364.
Hargreaves, G.H., Samani, Z.A., 1985. Reference crop evapotranspiration from
temperature. Applied Engineering in Agriculture 1, 96-99.
Harmel, R.D., Cooper, R.J., Slade, R.M., Haney, R.L., Arnold, J.G., 2006. Cumulative
uncertainty in measured streamflow and water quality data for small watersheds.
Transactions of the ASABE 49, 689-701.
Harmsen, E.W., Mesa, S.E.G., Cabassa, E., Ramírez-Beltran, N.D., Pol, S.C., Kuligowski,
R.J., Vasquez, R., 2008. Satellite sub-pixel rainfall variability. International Journal of
Systems Applications, Engineering & Development 2, 91-100.
Hawley, M.E., Jackson, T.J., McCuen, R.H., 1983. Surface soil moisture variation on
small agricultural watersheds. Journal of Hydrology 62, 179-200.
Healy, R., Cook, P., 2002. Using groundwater levels to estimate recharge. Hydrogeology
Journal 10, 91-109.
Hellebrand, H., van den Bos, R., Hoffmann, L., Juilleret, J., Krein, A., Pfister, L., 2009.
Spatio-temporal variability of behavioral patterns in hydrology in meso-scale basins of
the Rhineland Palatinate (1972–2002). Climate Change 93, 223–242.
Hemakumara, M., Kalma, J., Walker, J., Willgoose, G., 2004a. Downscaling of low
resolution passive microwave soil moisture observations. In: Teuling, A., Leijnse, H.,
Troch, P., Sheffield, J., Wood, E. (Eds.), 2nd international CAHMDA workshop on: The
Terrestrial Water Cycle: Modelling and Data Assimilation Across Catchment Scales,
Princeton, NJ, pp. 67-71.
Hemakumara, M., Kalma, J.D., Walker, J.P., Willgoose, G., 2004b. Downscaling of low
resolution passive microwave soil moisture observations. In: Teuling, A.J., Leijnse, H.,
Troch, P.A., Sheffield, J., Wood, E.F. (Eds.), 2nd International CAHMDA workshop on:
The Terrestrial Water Cycle: Modelling and Data Assimilation Across Catchment
Scales, Princeton, NJ, pp. 67–71.
Hillel, D., 1998. Environmental soil physics. Academic Press, London, UK.
Hoffmann, L., El Idrissi, A., Pfister, L., Hingray, B., Guex, F., Musy, A., Humbert, J.,
Drogue, G., Leviandier, T., 2004. Development of regionalized hydrological models in
an area with short hydrological observation series. River Research and Applications 20,
243-254.
Holtslag, A.A.M., 1984. Estimates of diabatic wind speed profiles from near-surface
weather observations. Boundary Layer Meteorology 29, 225-250.
Hong, Y., Hsu, K.l., Moradkhani, H., Sorooshian, S., 2006. Uncertainty quantification of
satellite precipitation estimation and Monte Carlo assessment of the error propagation
into
hydrologic
response.
Water
Resources
Research
42,
W08421,doi:08410.01029/02005WR004398.
Hossain, F., Anagnostoub, E.N., Bagtzoglou, A.C., 2006. On Latin hypercube sampling
for efficient uncertainty estimation of satellite rainfall observations in flood prediction.
Computers & Geosciences 32, 776–792.
Hossain, F., Huffman, G.J., 2008. Investigating error metrics for satellite rainfall data at
hydrologically relevant scales. Journal of Hydrometeorology 9, 563-575.
172
Houser, P.R., Shuttleworth, W.J., Famiglietti, J.S., Gupta, H.V., Syed, K.H., Goodrich,
D.C., 1998. Integration of soil moisture remote sensing and hydrologic modeling using
data assimilation. Water Resour. Res. 34, 3405-3420.
Houze, R.A., Wilton, D.C., Smull, B.F., 2007. Monsoon convection in the Himalayan
region as seen by the TRMM Precipitation Radar. Quarterly Journal of the Royal
Meteorological Society 133, 1389-1411.
Hu, W., Shao, M.A., Wang, Q.J., Reichardt, K., 2008. Soil water content temporal-spatial
variability of the surface layer of a loess plateau hillside in China. Scientia Agricola 65,
277-289.
Huang, J., Véronneau, M., Mainville, A., 2008. Assessment of systematic errors in the
surface gravity anomalies over North America using the GRACE gravity model.
Geophysical Journal International 175, 46-54.
Huffman, G.J., Adler, R.F., Bolvin, D.T., Gu, G., Nelkin, E.J., Bowman, K.P., Hong, Y.,
Stocker, E.F., Wolff, D.B., 2007. The TRMM Multisatellite Precipitation Analysis
(TMPA): quasi-global, multiyear, combined-sensor precipitation estimates at fine scales.
Journal of Hydrometeorology 8, 38-55.
Huffman, G.J., Adler, R.F., Morrissey, M.M., Bolvin, D.T., Curtis, S., Joyce, R.,
McGavock, B., Susskind, J., 2001. Global precipitation at one-degree daily resolution
from multisatellite observations. Journal of Hydrometeorology 2, 36-50.
Hussain, I., 2011. Water demand management and economic value of water in the Indus
Basin. PhD dissertation, Department of Economics. University of Sargodha, Sargodha,
Pakistan, p. 238.
Hussain, M.Z., 1998. NOAA-measurements for identifying agro-ecological zones in the
province of Sindh, Pakistan. International Institute for Aerospace and Earth Sciences,
Enschede, the Netherlands.
ICID, 2009. Irrigation and drainage in the world – A global review (Pakistan Chapter).
International
Commission
on
Irrigation
and
Drainage,
http://www.icid.org/i_d_pakistan.pdf, p. 16.
Iguchi, T., Meneghini, R., Awaka, J., Kozu, T., Okamoto, K., 2000. Rain profiling
algorithm for TRMM Precipitation Radar data. Advances in Space Research 25, 973976.
Immerzeel, W.W., Droogers, P., 2008. Calibration of a distributed hydrological model
based on satellite evapotranspiration. Journal of Hydrology 349, 411-424.
Immerzeel, W.W., Gaur, A., Zwart, S.J., 2008a. Integrating remote sensing and a processbased hydrological model to evaluate water use and productivity in a south Indian
catchment. Agricultural Water Management 95, 11-24.
Immerzeel, W.W., Gaur, A., Zwart, S.J., 2008b. Integrating remote sensing and a processbased hydrological model to evaluate water use and productivity in a south Indian
catchment. Agricultural Water Management 95, 11-24.
Immerzeel, W.W., Rutten, M.M., Droogers, P., 2009. Spatial downscaling of TRMM
precipitation using vegetative response on the Iberian Peninsula. Remote Sensing of
Environment 113, 362-370.
Ines, A.V.M., Mohanty, B.P., 2008. Near-surface soil moisture assimilation for
quantifying effective soil hydraulic properties under different hydroclimatic conditions.
Vadose Zone Journal 7, 39-52.
173
Islam, N.M., Uyeda, H., 2008. Vertical variations of rain intensity in different rainy
periods in and around Bangladesh derived from TRMM observations. International
Journal of Climatology 28, 273-279.
IUCN, 2010. Beyond Indus Water Treaty: Ground water and environmental management
– Policy issues and options. IUCN, Karachi,Pakistan, p. 10.
IUCN, 2011. Water resources of Pakistan: The Goverment's main objectives. Pakistan
Water Gateway http://waterinfo.net.pk/cms/?q=node/19.
Jackson, T.J., 1993. Measuring surface soil moisture using passive microwave remote
sensing. Hydrological Processes 7, 139-152.
Jackson, T.J., Bindlish, R., Cosh, M., 2009. Validation of AMSR-E soil moisture products
using in situ observations. Journal of The Remote Sensing Society of Japan 29, 263-270.
Jackson, T.J., Moran, M.S., O'Neill, P.E., 2008. Introduction to Soil Moisture Experiments
2004 (SMEX04) special issue. Remote Sensing of Environment 112, 301-303.
Jackson, T.J., Schmugge, T.J., 1991. Vegetation effects on the microwave emission of
soils. Remote Sensing of Environment 36, 203-212.
Jackson, T.J., Schmugge, T.J., Wang, J.R., 1982. Passive microwave sensing of soil
moisture under vegetation canopies. Water Resources Research 18, 1137-1142.
Jain, S.K., Agarwal, P.K., Singh, V.P., 2007. Indus Basin. Hydrology and Water
Resources of India. Springer Netherlands pp. 473-511.
Jarvis, A., Reuter, H.I., Nelson, A., Guevara, E., 2008. Hole-filled SRTM for the globe
version 4, available from the CGIAR-CSI SRTM 90m database. http://srtm.csi.cgiar.org.
Jarvis, P.G., 1976. The Interpretation of the Variations in Leaf Water Potential and
Stomatal Conductance Found in Canopies in the Field. Philosophical Transactions of the
Royal Society of London. B, Biological Sciences 273, 593-610.
JAXA, 2006. TRMM data users hand book. Japan Aerospace Exploration Agency, Earth
Observation Centre, Japan.
Jeevandas, A., Singh, R.P., Kumar, R., 2008. Concerns of groundwater depletion and
irrigation efficiency in Punjab agriculture: A micro-level study. Agricultural Economics
Research Review 21, 191-199.
Jhorar, R.K., Smit, A., Bastiaanssen, W.G.M., Roest, C., 2011. Calibration of a distributed
irrigation water management model using remotely sensed evapotranspiration rates and
groundwater heads. Irrigation and Drainage 60, 57-69.
Ji, Y., Stocker, E., 2003. Ground validation of TRMM and AMSU microwave
precipitation estimates. Geoscience and Remote Sensing Symposium, 2003. IGARSS
'03. Proceedings. 2003 IEEE International pp. 3157-3159.
Jia, L., Xi, G., Liu, S., Huang, C., Yan, Y., Liu, G., 2009. Regional estimation of daily to
annual regional evapotranspiration with MODIS data in the Yellow River Delta wetland.
Hydrology and Earth System Sciences 13, 1775-1787.
Jianchu, X., Shrestha, A., Vaidya, R., Eriksson, M., Hewitt, K., 2007. The melting
Himalayas regional challenges and local impacts of climate change on mountain
ecosystems and livelihoods. ICIMOD Technical Paper :The Melting Himalayas,
Kathmandu, Nepal, p. pp15.
Jiang, Z., Huete, A.R., Chen, J., Chen, Y., Li, J., Yan, G., Zhang, X., 2006. Analysis of
NDVI and scaled difference vegetation index retrievals of vegetation fraction. Remote
Sensing of Environment 101, 366-378.
174
Jilani, R., Sidiqui, P., Munir, S., Haq, M., 2008. Revalidation of TRMM precipitation data
with ground based measurements for selected cities of Pakistan. Asian Conference on
Remote Sensing (ACRS), Colombo, Sri Lanka, p. 1.20.
Jonathan, M., Meirelles, M.S.P., Berroir, J.P., Herlin, I., 2006. Regional scale land
use/land cover classification using temporal series of MODIS data. ISPRS Commission
VII Mid-term Symposium "Remote Sensing: From Pixels to Processes", Enschede, the
Netherlands, pp. 522-527.
Kale, M., Singh, S., Roy, P.S., 2005. Estimation of leaf area index in dry deciduous forests
from IRS-WiFS in central India. International Journal of Remote Sensing 26, 48554867.
Kalma, J.D., McVicar, T.R., McCabe, M.F., 2008. Estimating land surface evaporation: A
review of methods using remotely sensed surface temperature data. Surveys in
Geophysics 29, 421-469.
Kannan, N., Jeong, J., Srinivasan, R., 2011. Hydrologic modeling of a canal irrigated
agricultural watershed with irrigation best management practices:A case study. Journal
of Hydrologic Engineering 16, 746-758.
Kästner, M., 2007. EURAINSAT algorithm validation and intercomparison exercise. In:
Levizzani, V., Bauer, P., Turk, F.J. (Eds.), Measuring Precipitation From Space:
EURAINSAT and the Future Springer Netherlands, pp. 369-380.
Kawanishi, T., Kuroiwa, H., Kojima, M., Oikawa, K., Kozu, T., Kumagai, H., Okamoto,
K.i., Okumura, M., Nakatsuka, H., Nishikawa, K., 2000. TRMM Precipitation Radar.
Advances in Space Research 25, 969-972.
Kerr, Y.H., 2007. Soil moisture from space: Where are we? Hydrogeology Journal 15,
117-120.
Khain, A., Rosenfeld, D., Pokrovsky, A., 2007. Aerosol impact on precipitation from
convective clouds. In: Levizzani, V., Bauer, P., Turk, F.J. (Eds.), Measuring
Precipitation From Space: EURAINSAT and the Future Springer Netherlands, pp. 421434.
Khan, A.F., 2009. India building small dams on Indus. DAWN, Karachi.
Khan, A.R., 1999. An analysis of the surface water resources and water delivery systems
in the Indus Basin. Report R-93. International Water Management Institute (IWMI),
Lahore, Pakistan, p. 66.
Kijne, J.W., 1999. Improving the productivity of Pakistan’s irrigation: the importance of
management choices. International Water Management Institute, Colombo, Sri Lanka.
King, M.D., Tsay, S.C., Platnick, S.E., Wang, M., Liou, K.N., 1997. Cloud retrieval
algorithms for MODIS: Optical thickness, effective particle radius, and thermodynamic
phase. MODIS algorithm theoretical basis document no. ATBD-MOD-05 MOD06 –
Cloud product.
Kitchin, C.R., 1987. Stars, nebulae and the interstellar medium: Observational physics and
astrophysics. Adam Hilger, Bristol, England.
Knight, J.F., Lunetta, R.L., Ediriwickrema, J., Khorram, S., 2006. Regional scale landcover characterization using MODIS-NDVI 250 m multi-temporal imagery: A
phenology based approach. GIScience and Remote Sensing 43, 1-23.
Kozu, T., Kawanishi, T., Kuroiwa, H., Kojima, M., Oikawa, K., Kumagai, H., Okamoto,
K.i., Okumura, M., Nakatsuka, H., Nishikawa, K., 2001. Development of precipitation
175
radar onboard the Tropical Rainfall Measuring Mission (TRMM) satellite. IEEE
Transactions on Geoscience and Remote Sensing 39, 102-116.
Krajewski, W.F., Ciach, G.J., Habib, E., 2003. An analysis of small-scale rainfall
variability in different climatic regimes. Hydrological Sciences 48, 151-162.
Kreutzmann, H., 2011. Scarcity within opulence: Water management in the Karakoram
Mountains revisited. Journal of Mountain Science 8, 525-534.
Kroes, J.G., Droogers, P., Kumar, R., Immerzeel, W.W., Khatri, R.S., Roelevink, A., ter
Maat, H.W., Dabas, D.S., 2003. A regional approach to model water productivity. In:
Van Dam, J.C., Malik, R.S. (Eds.), Water productivity of irrigated crops in Sirsa
District, India. Integration of remote sensing, crop and soil models and geographical
information systems. WATPRO final report, pp. 101-119.
Kummerow, C., Barnes, W., Kozu, T., Shiue, J., Simpson, J., 1998. The tropical rainfall
measuring mission (TRMM) sensor package. Journal of Atmospheric and Oceanic
Technology 15, 809-817.
Kummerow, C., Hakkarinen, I.M., Pierce, H.F., Weinman, J.A., 1991. Determination of
precipitation profiles from airborne passive microwave radiometric measurements.
Journal of Atmospheric and Oceanic Technology 8, 148-158.
Kummerow, C., Hong, Y., Olson, W.S., Yang, S., Adler, R.F., McCollum, J., Ferraro, R.,
Petty, G., Shin, D.B., Wilheit, T.T., 2001. The evolution of the goddard profiling
algorithm (GPROF) for rainfall estimation from passive microwave sensors. Journal of
Applied Meteorology 40, 1801-1820.
Kummerow, C., Simpson, J., Thiele, O., Barnes, W., Chang, A.T.C., Stocker, E., Adler,
R.F., Hou, A., Kakar, R., Wentz, F., Ashcroft, P., Kozu, T., Hong, Y., Okamoto, K.,
Iguchi, T., Kuroiwa, H., Im, E., Haddad, Z., Huffman, G., Ferrier, B., Olson, W.S.,
Zipser, E., Smith, E.A., Wilheit, T.T., North, G., Krishnamurti, T.N., Nakamura, K.,
2000. The status of the tropical rainfall measuring mission (TRMM) after two years in
orbit. Journal of Applied Meteorology 39, 1965-1982.
Kureshy, K.U., 1977. A geography of Pakistan, 4th edition. Oxford University Press,
Karachi, Pakistan.
Lang, T.J., Barros, A.P., 2004. Winter storms in the central Himalayas. Journal of the
Japanese Meteorological Society 82, 829-844.
Lashkaripou, G.R., Hussaini, S.A., 2007. Water Resource Management in Kabul River
Basin, Eastern Afghanistan. The Environmentalist 10.1007/s10669-007-9136-2.
Latifovic, R., Olthof, I., 2004. Accuracy assessment using sub-pixel fractional error
matrics of global land cover products derived from satellite data. Remote sensing of
Environment 90, 153-165.
Lebel, T., Barbe, L.L., 1997. Rainfall monitoring during HAPEX-Sahel: Point and areal
estimation at the event and seasonal scale. Journal of Hydrology 188-189, 97-122.
Li, F., Kustas, W.P., Anderson, M.C., Jackson, T.J., Bindlish, R., Prueger, J.H., 2006.
Comparing the utility of microwave and thermal remote-sensing constraints in twosource energy balance modeling over an agricultural landscape. Remote Sensing of
Environment 101, 315-328.
Lillesand, T., Kiefer, R.W., 2000. Remote sensing and image interpretation. John Wiley
and Sons, New York.
176
Loew, A., Holmes, T., de Jeu, R., 2009. The European heat wave 2003: Early indicators
from multisensoral microwave remote sensing? Journal of Geophysical Research 114,
D05103.
Long, D., Singh, V.P., 2012. A modified surface energy balance algorithm for land (MSEBAL) based on a trapezoidal framework. Water Resources Research 48,
W02528,doi:02510.01029/02011WR010607.
Loveland, T.R., Reed, B.C., Brown, J.F., Ohlen, D.O., Zhu, Z., Yang, L., Merchant, J.W.,
2000. Development of a global land cover characteristics database and IGBP DISCover
from 1km AVHRR data. International Journal of Remote Sensing 21, 1303-1330.
Mahajan, S., Panwar, P., Kaundal, D., 2001. GIS application to determine the effect of
topography on landuse in Ashwani Khad watershed. Journal of the Indian Society of
Remote Sensing 29, 243-248.
Mandel, R., 1992. Sources of international river basin disputes. Conflict Quarterly 12, 2556.
Marshall, J.S., Palmer, W.M., 1948. The distribution of raindrops with size. Journal of the
Atmospheric Sciences 5, 165-166.
Maupin, M.A., 1999. Methods to determine pumped irrigation-water withdrawals from the
Snake River between upper Salmon Falls and Swan Falls dams, Idaho, using electrical
power data,1990–95. U.S. Geological Survey Water-Resources Investigation Report 99–
4175. USGS, Littleton, Colorado, p. 20.
McCabe, M.F., Wood, E.F., Gao, H., 2005. Initial soil moisture retrievals from AMSR-E:
Multiscale comparison using in situ data and rainfall patterns over Iowa. Geophysical
Research Letters 32, L06403.
Mehrez, M.B., Taconet, O., Vidal-Madjar, D., Valencogne, C., 1992. Estimation of
stomatal resistance and canopy evaporation during the HAPEX-MOBILHY experiment.
Agricultural and Forest Meteorology 58, 285-313.
Meijninger, W.M.L., Hartogensis, O.K., Kohsiek, W., Hoedjes, J.C.B., Zuurbier, R.M., De
Bruin, H.A.R., 2002. Determination of area-averaged sensible heat fluxes with a large
aperture scintillometer over a heterogeneous surface-Flevoland field experiment.
Boundary-Layer Meteorology 105, 37-62.
Melesse, A.M., Jordan, J.D., 2003. Spatially distributed watershed mapping and modeling:
thermal Maps and vegetation indices to enhance land cover and surface microclimate
mapping: part 1. Journal of Spatial Hydrology 3, 1-29.
Merlin, O., Chehbouni, A., Kerr, Y.H., Goodrich, D.C., 2006. A downscaling method for
distributing surface soil moisture within a microwave pixel: Application to the Monsoon
'90 data. Remote Sensing of Environment 101, 379-389.
Merlin, O., Chehbouni, A., Walker, J.P., Panciera, R., Kerr, Y.H., 2008. A simple method
to disaggregate passive microwave-based soil moisture. IEEE Transactions on
Geoscience and Remote Sensing 46, 786 - 796.
Meyer, W.B., Turner, B.L., 1994. Changes in land use and land cover: a global
perspective. . Cambridge University Press, Cambridge, U.K.
MINFAL, 2007. Agricultural Statistics of Pakistan. 2006-07. Ministry of Food,
Agriculture and Livestock, Islamabad, Pakistan, pp. 139-140,178-179.
MINFAL, 2008. Crop area and production (By Districts) 2006-2007. Minstry of Food,
Agriculture and Livestock, Islamabad, Pakistan.
177
Mizoguchi, Y., Miyata, A., Ohtani, Y., Hirata, R., Yuta, S., 2009. A review of tower flux
observation sites in Asia. Journal of Forest Research 14, 1-9.
MODIS, 2004. Moderate Resolution Imaging Spectroradiometer (MODIS) Land Cover
(MOD12Q1) Product. Boston University
Molden, D., 1997. Accounting for water use and productivity., SWIM Paper 1., Colombo,
Sri Lanka.
Molden, D., Sakthivadivel, R., Keller, J., 2001. Hydronomic zones for developing basin
water conservation strategies. Research Rep. No. 56. International Water Management
Institute, Colombo, Sri Lanka.
Molden, D.J., Sakthivadivel, R., 1999. Water accounting to assess use and productivity of
water. International Journal of Water Resources Development 15, 55-71.
Monteith, J.L., 1965. Evaporation and environment. Symposia of the Society for
Experimental Biology 19, 205-234.
Monteith, J.L., 1981. Evaporation and surface temperature. Quarterly Journal of the Royal
Meteorological Society 107, 1-27.
Moriasi, D.N., Arnold, J.G., Van Liew, M.W., Bingner, R.L., Harmel, R.D., Veith, T.L.,
2007. Model evaluation guidelines for systematic quantification of accuracy in
watershed simulations. Transactions of the ASABE 50, 885-900.
Mu, Q., Heinsch, F.A., Zhao, M., Running, S.W., 2007. Development of a global
evapotranspiration algorithm based on MODIS and global meteorology data. Remote
Sensing of Environment 11, 519-536.
Mucher, C.A., de Badts, E.P.J., 2002. Global land cover 2000: Evaluation of the SPOT
VEGTATION sensor for land use mapping. Alterra, Green World Research.
Wageningen, the Netherlands., Wageningen.
Muslehuddin, M., Mir, H., Faisal, N., 2005. Sindh summer (June-September) monsoon
rainfall prediction. Pakistan Journal of Meteorology 2, 95-112.
Nagler, P.L., Scott, R.L., Westenburg, C., Cleverly, J.R., Glenn, E.P., Huete, A.R., 2005.
Evapotranspiration on western U.S. rivers estimated using the Enhanced Vegetation
Index from MODIS and data from eddy covariance and Bowen ratio flux towers.
Remote Sensing of Environment 97, 337-351.
Nalder, I.A., Wein, R.W., 1998. Spatial interpolation of climatic normals: test of a new
method in the canadian boreal forest. Agricultural and Forest Meteorology 92, 211-225.
Narayan, U., Lakshmi, E.V., Njoku, E.G., 2004. Retrieval of soil moisture from passive
and active L/S band sensor (PALS) observations during the soil moisture experiment in
2002 (SMEX02). Remote Sensing of Environment 92, 483-496.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models part 1A discussion of principles. Journal of Hydrology 10, 282-290.
Neitsch, S.L., Arnold, J.G., Kiniry, J.R., Williams, J.R., 2005. Soil and Water Assessment
Tool - Theoretical documentation. Version 2005. Blackland Research & Extension
Center, Texas, USA.
Nemes, A., Wösten, H., Lilly, A., 2001. Development of soil hydraulic pedotransfer
functions on a European scale: Their usefulness in the assesment of soil quality. In: D.E.
Stott, Mohtar, R.H., Steinhardt, G.C. (Eds.), Sustaining the Global Farm.10th
International Soil Conservation Organization Meeting, Purdue University and
USDAARS National Soil Erosion Research Laboratory, pp. 541-549.
178
Nespor, V., Sevruk, B., 1999. Estimation of wind-induced error of rainfall gauge
measurements using a numerical simulation. Journal of Atmospheric and Oceanic
Technology 16, 450-464.
New, M., Hulme, M., Jones, P., 1999. Representing twentieth-century space–time climate
variability. Part 1: Development of a 1961–90 mean monthly terrestrial climatology.
Journal of Climate 12, 829-856.
New, M., Lister, D., hulme, M., Makin, I., 2002. A high-resolution data set of surface
climate over global land areas. Climate Research 21, 1-25.
Niblack, M., Sanchez, C., 2008. Implementation of a water accounting system to assist
growers in managing current and future water restrictions. Annual meeting of the Soil
and Water Conservation Society, Tucson, Arizona, USA.
Njoku, E.G., 2008. AMSR-E/Aqua Daily L3 Surface Soil Moisture, Interpretive
Parameters, & QC EASE-Grids V002, January to December 2007. National Snow and
Ice Data Center, Boulder, Colorado USA.
Njoku, E.G., Chan, S.K., 2006. Vegetation and surface roughness effects on AMSR-E land
observations. Remote Sensing of Environment 100, 190-199.
Njoku, E.G., Entekhabi, D., 1996. Passive microwave remote sensing of soil moisture.
Journal of Hydrology 184, 101-129.
Njoku, E.G., Jackson, T.J., Lakshmi, V., Chan, T.K., Nghiem, S.V., 2003. Soil moisture
retrieval from AMSR-E. IEEE Transactions on Geoscience and Remote Sensing 41,
215-229.
Njoku, E.G., Li, L., 1999. Retrieval of land surface parameters using passive microwave
measurements at 6–18 GHz. IEEE Transactions on Geoscience and Remote Sensing 37,
79-93.
Oke, A.M.C., Frost, A.J., Beesley, C.A., 2009. The use of TRMM satellite data as a
predictor in the spatial interpolation of daily precipitation over Australia. 18th World
IMACS / MODSIM Congress,Cairns, Australia, pp. 3726-3732.
Omotosho, T.V., Oluwafemi, C.O., 2009. One-minute rain rate distribution in Nigeria
derived from TRMM satellite data. Journal of Atmospheric and Solar-Terrestrial
Physics 71, 625-633.
Owe, M., de Jeu, R., Holmes, T., 2008. Multi-sensor historical climatology of satellitederived global land surface moisture. Journal of Geophysical Research 113, F01002,
doi:01010.01029/02007JF000769.
Owe, M., de Jeu, R., Walker, J., 2001. A methodology for surface soil moisture and
vegetation optical depth retrieval using the microwave polarization difference index.
IEEE Transactions on Geoscience and Remote Sensing 39, 1643-1654.
Ozdogan, M., Gutman, G., 2008. A new methodology to map irrigated areas using multitemporal MODIS and ancillary data: An application example in the continental US.
Remote Sensing of Environment 112, 3520-3537.
Parajka, J., Naeimi, V., Blöschl, G., Wagner, W., Merz, R., Scipal, K., 2006. Assimilating
scatterometer soil moisture data into conceptual hydrologic models at the regional scale.
Hydrology and Earth System Sciences 10, 353-368.
PARC, 1982. Consumptive use of water for crops in Pakistan. Final Technical Report.
Pakistan Agricultural Research Council, Islamabad, Pakistan, pp. 20-30.
179
Pelgrum, H., Miltenburg, I.J., Cheema, M.J.M., Klaasse, A.K., Bastiaanssen, W.G.M.,
2010. ETLook: a novel continental evapotranspiration algorithm. Remote Sensing and
Hydrology Symposium, Jackson Hole, Wyoming USA, pp. IAHS Publ. 3XX, 2011.
Perry, C., 2007. Efficient irrigation;inefficient communication;flawed recommendations.
Irrigation and Drainage 56, 367–378.
Philippon, N., Mougin, E., Jarlan, L., 2005. Analysis of the linkages between rainfall and
land surface conditions in the West African monsoon through CMAP, ERS-WSC, and
NOAA-AVHRR data. Journal of Geophysical Research 110, D24115.
PILDAT, 2003. Issues of water resources in Pakistan. Briefing paper for Pakistani
parliamentarians. Pakistan Institute of Legislative Development and Transparency
(PILDAT), Islamabad. Briefing paper 7.
PILDAT, 2010. Pakistan-India relations: Implementation of Indus Water Treaty. A
pakistani Narrative. Pakistan Institute of Legislative Development and Transparency
(PILDAT), Islamabad. Background paper 6-020, p. 14.
Pitman, A.J., 1994. Assessing the sensitivity of a land-surface scheme to the parameter
values using a single column model. Journal of Climate 7, 1856-1869.
Porcù, F., Prodi, F., Pinori, S., Dietrich, S., Panegrossi, G., Tripoli, G., 2003. On the
capabilities of VIS/IR satellite data to resolve orographic precipitation. Proceedings of
the 5th EGS Plinius Conference, Ajaccio, Corsica, France.
Portmann, F.T., Siebert, S., Döll, P., 2010. MIRCA2000-Global monthly irrigated and
rainfed crop areas around the year 2000: A new high-resolution data set for agricultural
and hydrological modeling. Global Biogeochemical Cycles 24, GB1011.
Prabhakara, C., Iacovazzi, J.R., Yoo, J.M., 2002. TRMM Precipitation Radar and
Microwave Imager Observations of convective and stratiform rain over land and their
theoretical implications. Journal of the Meteorological Society of Japan 80, 1183-1197.
Press, W.H., Teukolsky, S.A., Vetterling, W.T., Flannery, B.P., 1996. Numerical recipes
in C. Cambridge University Press.
Price, J.C., 1990. Using spatial context in satellite data to infer regional scale
evapotranspiration. IEEE Transactions on Geoscience and Remote Sensing 28, 940-948.
Probert-Jones, J.R., 1962. The radar equation in meteorology. Quarterly Journal of the
Royal Meteorological Society 88, 486-495.
Propastin, P.P., Kappas, M., Muratova, N.R., 2006. Temporal responses of vegetation to
climatic factors in Kazakhstan and Middle Asia. Shaping the Change, XXIII FIG
Congress, Munich, Germany, p. 16.
Qureshi, A.S., Gill, M.A., Sarwar, A., 2010a. Sustainable groundwater management in
Pakistan:Challenges and Opportunities. Irrigation and Drainage 59, 107-116.
Qureshi, A.S., Hussain, A., Makin, I., 2002. Integrated database development for river
basin management: An example from Rechna Doab. Working Paper 53. International
Water Management Institute.(IWMI), Lahore, Pakistan.
Qureshi, A.S., McCornick , P.G., Sarwar, A., Sharma, B.R., 2010b. Challenges and
prospects of sustainable groundwater management in the Indus Basin, Pakistan. Water
Resources Management 24, 1551-1569.
Qureshi, A.S., Shah, T., Akhtar, M., 2003. The groundwater economy of Pakistan.
Working Paper 64. International Water Management Institute (IWMI), Lahore, Pakistan.
180
Radersma, S., de Ridder, N., 1996. Computed evapotranspiration of annual and perennial
crops at different temporal and spatial scales using published parameter values.
Agricultural Water Management 31, 17-34.
Rao, Y.S., Sharma, S., Garg, V., Venkataraman, G., 2006. Soil moisture mapping over
India using Aqua AMSR-E derived soil moisture product. IEEE International
Conference on Geoscience and Remote Sensing Denver, USA, pp. 2999-3002.
Rao, Y.S., Singh, G., Venkataraman, G., 2008. Soil moisture mapping using ALOS
PALSAR Quad-pol data. International Conference on Microwave 08, Jaipur, pp. 214216.
Rashid, M.R., 2007. Monitoring habitat change and its relation to sand lizard population
dynamics with multi temporal remote sensing: A case study of Terschelling and
Vlieland, the Netherlands. ICT, Enschede, the Netherlands. ITC, Enschede, the
Netherlands.
Ray, R.L., Jacobs, J.M., 2007. Relationships among remotely sensed soil moisture,
precipitation and landslide events. Natural Hazards 43, 211-222.
Reed, B.C., Brown, J.F., VanderZee, D., Loveland, T.R., Merchant, J.W., Ohlen, D.O.,
1994. Measuring phenological variability from satellite imagery. Journal of Vegetation
Science 5, 703-714.
Rodell, M., Velicogna, I., Famiglietti, J.S., 2009. Satellite-based estimates of groundwater
depletion in India. Nature 460, 999-1002.
Rosnay, P.d., Calvet, J.-C., Kerr, Y., Wigneron, J.-P., Lemaître, F., Escorihuela, M.J.,
Sabater, J.M., Saleh, K., Barrié, J., Bouhours, G., Coret, L., Cherel, G., Dedieu, G.,
Durbe, R., Fritz, N.E.D., Froissard, F., Hoedjes, J., Kruszewski, A., Lavenu, F., Suquia,
D., Waldteufel, P., 2006. SMOSREX: A long term field campaign experiment for soil
moisture and land surface processes remote sensing. Remote Sensing of Environment
102, 377-389.
Rubel, F., Hantel, M., 1999. Correction of daily rain gauge measurements in the Baltic Sea
drainage basin. Nordic Hydrology 30, 191-208.
Santos, C., Lorite, I.J., Tasumi, M., Allen, R.G., Fereres, E., 2010. Performance
assessment of an irrigation scheme using indicators determined with remote sensing
techniques. Irrigation Science 28, 461-477.
Sarwar, A., 2000. A transient model approach to improve on-farm irrigation and drainage
in semi-arid zones. PhD dissertation. Wageningen University, Wageningen, the
Netherlands, p. 147.
Sarwar, A., Bastiaanssen, W.G.M., 2001. Long-term effects of irrigation water
conservation on crop production and environment in semiarid areas. Journal of Irrigation
and Drainage Engineering 127, 331-338.
Sarwar, A., Bill, R., 2007. Mapping evapotranspiration in the Indus Basin using ASTER
data. International Journal of Remote Sensing 28, 5037-5046.
Sarwar, A., Eggers, H., 2006. Development of a conjunctive use model to evaluate
alternative management options for surface and groundwater resources. Hydrogeology
Journal 14, 1676–1687.
Sato, T., Teraoka, T., Kimura, I., 1996. Validation and ground truth for TRMM
precipitation radar using the. Geoscience and Remote Sensing Symposium, 1996.
IGARSS '96. 'Remote Sensing for a Sustainable Future.', International, pp. 1361-1363
vol.1362.
181
Sawunyama, T., Hughes, D.A., 2008. Application of satellite-derived rainfall estimates to
extend water resource simulation modelling in South Africa. Water SA 34, 1-9.
Schmugge, T.J., 1983. Remote sensing of soil moisture: Recent advances. IEEE
Transactions on Geoscience and Remote Sensing GE-21, 336 - 344
Schmugge, T.J., Jackson, T.J., Kustas, W.P., Wang, J.R., 1992. Passive microwave remote
sensing of soil moisture: results from HAPEX, FIFE and MONSOON 90. ISPRS
Journal of Photogrammetry and Remote Sensing 47, 127-143.
Schrama, E.J.O., Wouters, B., Lavallée, D.A., 2007. Signal and noise in Gravity Recovery
and Climate Experiment (GRACE) observed surface mass variations. J. Geophys. Res.
112, B08407.
Schulze, R.E., Kiker, G.A., Kunz, R.P., 1993. Global climate change and agricultural
productivity in southern Africa. Global Environmental Change 3, 330-349.
Schumacher, C., Houze, R.A., 2003. Stratiform rain in the tropics as seen by the TRMM
Precipitation Radar. Journal of Climate 16, 1739-1756.
Schwarz, M., Zimmermann, N.E., 2005. A New GLM-based method for mapping tree
cover continuous fields using regional MODIS reflectance data. Remote sensing of
Environment 95, 428-443.
Scipal, K., Holmes, T., De Jeu, R., Naeimi, V., Wagner, W., 2008. A possible solution for
the problem of estimating the error structure of global soil moisture data sets.
Geophysical Research Letters 35, L24403.
Scott, C.A., Shah, T., 2004. Groundwater overdraft reduction through agricultural energy
policy: insights from India and Mexico. International Journal of Water Resources
Development 20, 149-164.
Seckler, D., 1996. The new era of water resources management:From "dry" to "wet" water
savings. Research Report 1. International Water Management Institute, Colombo, Sri
Lanka.
Senay, G.B., Budde, M., Verdin, J.P., Melesse, A.M., 2007. A coupled remote sensing and
simplified surface energy balance approach to estimate actual evapotranspiration from
irrigated fields. Sensors 7, 979-1000.
Shah, T., Bruke, J., Vullholth, K., Angelica, M., Custodio, E., Daibes, F., Hoogesteger
Van Dijk, J.D., Giordano, M., Girman, J., Kendy, E., Kijne, J., Llamas, R.,
Masiyandama, M., Margat, J., Marin, L., Peck, J., Rozelle, S., Sharma, B., Vincent,
L.F., Wang, J., 2007. Groundwater: A global assessment of scale and significance. In:
Molden, D. (Ed.), Water for food: water for life. A comprehensive assessment of water
management in agriculture. Earthscan, London, U.K, Colombo, Sri Lanka, IWMI, pp.
395-423.
Shah, T., Molden, D.J., Sakthivadivel, R., Seckler, D., 2000. The global groundwater
situation: Overview of opportunities and challenges. International Water Management
Institute, Colombo, Sri Lanka.
Shakir, A.S., Khan, N.M., Qureshi, M.M., 2010. Canal water management: Case study of
upper Chenab Canal in Pakistan. Irrigation and Drainage 59, 76-91.
Shakoor, A., Shehzad, A., Asghar, M.N., 2006. Application of remote sensing techniques
for water resources planning and management. International Conference on Advances in
Space Technologies, 2006, Islamabad, Pakistan, pp. 142-146.
Shankar, P.S.V., Kulkarni, H., Krishnan, S., 2011. India’s groundwater challenge and the
way forward. Economic and Political Weekly XLVI, 37-45.
182
Sharma, M.L., 1985. Estimating evapotranspiration. Advances in Irrigation, Volume 3,
Academic Press Orlando, Florida.
Shuttleworth, W.J., Wallace, J.S., 1985. Evaporation from sparse crops-an energy
combination theory. Quarterly Journal of the Royal Meteorological Society 111, 839855.
Siebert, S., Burke, J., Faures, J.M., Frenken, K., Hoogeveen, J., Doll, P., Portmann, F.T.,
2010. Groundwater use for irrigation – a global inventory. Hydrology and Earth System
Sciences 14, 1863-1880.
Singh, P., Kumar, N., 1997. Effect of orography on precipitation in the western Himalayan
region. Journal of Hydrology 199, 183-206.
Singh, P., Ramasastri, K.S., Kumar, N., 1995. Topographical influence on precipitation
distribution in different ranges of Western Himalayas. Nordic Hydrology 26, 259-284.
Singh, R., Kroes, J.G., van Dam, J.C., Feddes, R.A., 2006. Distributed ecohydrological
modelling to evaluate the performance of irrigation system in Sirsa district, India: I.
Current water management and its productivity. Journal of Hydrology 329, 692-713.
Singh, R.P., Oza, S.R., Chaudhari, K.N., Dadhwal, V.K., 2005. Spatial and temporal
patterns of surface soil moisture over India estimated using surface wetness index from
SSM/I microwave radiometer. International Journal of Remote Sensing 26, 1269 - 1276.
Sivapalan, M., Takeuchi, K., Franks, S.W., Gupta, V.K., Karambiri, H., Lakshmi, V.,
Liang, X., McDonnell, J.J., Mendiondo, E.M., O'Connell, P.E., Oki, T., Pomeroy, J.W.,
Schertzer, D., Uhlenbrook, S., Zehe, E., 2003. IAHS Decade on Predictions in
Ungauged Basins (PUB), 2003-2012: Shaping an exciting future for the hydrological
sciences. Hydrological Sciences Journal 48, 857-880.
Srinivasan, R., Ramanarayanan, T.S., Arnold, J.G., Bednarz, S.T., 1998. Large area
hydrologic modeling and assessment: part II. Model application. Journal of the
American Water Resources Association 34, 91-101.
Srinivasan, R., Zhang, X., Arnold, J., 2010. SWAT ungauged: Hydrological budget and
crop yield predictions in the Upper Mississippi River Basin. Transactions of the ASABE
53, 1533-1546
Stewart, J.B., 1988. Modelling surface conductance of pine forest. Agricultural and Forest
Meteorology 43, 19-35.
Tang, Q., Peterson, S., Cuenca, R.H., Hagimoto, Y., Lettenmaier, D.P., 2009. Satellitebased near-real-time estimation of irrigated crop water consumption. Journal of
Geophysical Research 114, D05114.
Teixeira, d.C.A., Bastiaanssen, W.G.M., 2011. Five methods to interpret field
measurements of energy fluxes over a micro-sprinkler-irrigated mango orchard.
Irrigation Science, 1-16.
Thapa, R.B., Murayama, Y., 2009. Urban mapping, accuracy, & image classification: A
comparison of multiple approaches in Tsukuba City, Japan. Applied Geography 29, 135144.
Thatte, C.D., 2008. Indus waters and the 1960 treaty between India and Pakistan. In:
Varis, O., Biswas, A.K., Tortajada, C. (Eds.), Management of Transboundary Rivers and
Lakes. Springer Berlin Heidelberg, pp. 165-206.
Thayyen, R.J., Gergan, J.T., 2009. Role of glaciers in watershed hydrology: “Himalayan
catchment” perspective. The Cryosphere Discussions 3, 443-476.
183
Thenkabail, P.S., Schull, M., Turral, H., 2005. Ganges and Indus river basin Land
Use/Land Cover (LULC) and irrigated area mapping using continuous streams of
MODIS data. Remote sensing of Environment 95, 317-341.
Thornton, P.E., Running, S.W., White, M.A., 1997. Generating surfaces of daily
meteorological variables over large regions of complex terrain. Journal of Hydrology
190, 214-251.
Thunnissen, H.A.M., Noordman, E., 1997. National land cover database of the
Netherlands: Classification methodology and operational implementation. Netherlands
Remote Sensing Board, Delft, the Netherlands, p. 95.
Tobin, K.J., Bennett, M., E., 2010. Adjusting satellite precipitation data to facilitate
hydrologic modeling. Journal of Hydrometeorology 11, 966-978.
Tucker, C.J., 1979. Red and photographic infrared linear combinations for monitoring
vegetation. Remote sensing of Environment 8, 127-150.
Turner, G.M., Baynes, T.M., McInnis, B.C., 2008. A water accounting system for strategic
water management. Socio-Economics and the Environment in Discussion (SEED).
CSIRO Working Paper 2008-14, p. 59.
Twine, T.E., Kustas, W.P., Norman, J.M., Cook, D.R., Houser, P.R., Meyers, T.P.,
Prueger, J.H., Starks, P.J., Wesely, M.L., 2000. Correcting eddy-covariance flux
underestimates over a grass land. Agricultural and Forest Meteorology 103, 279–300.
Ulaby, F.T., Moore, R.K., Fung, A.K., 1981. Microwave remote sensing: active and
passive. Artech House, Dedham, MA.
Ulaby, F.T., Moore, R.K., Fung, A.K., 1986. Microwave remote sensing: active and
passive Vol. 3, from theory to application. Artech House, Dedham, MA.
Ullah, M.K., Habib, Z., Muhammad, S., 2001. Spatial distribution of reference and
potential evapotranspiration across the Indus Basin Irrigation Systems. Working paper
24. International Water Management Institute (IWMI), Lahore, Pakistan.
Vachaud, G., Passerat De Silans, A., Balabanis, P., Vauclin, M., 1985. Temporal stability
of spatially measured soil water probability density function. Soil Science Society of
America Journal 49, 822-828.
Valderrama, J.O., Alvarez, V.H., 2005. Correct way of reporting results when modelling
supercritical phase equilibria using equations of state. The Canadian Journal of
Chemical Engineering 83, 578-581.
Van der Kwast, J., 2009. Quantification of top soil moisture patterns; Evaluation of field
methods, processbased modelling, remote sensing and an integrated approach.
Knag/Faculteit Geowetenschappen. Universiteit Utrecht, Utrecht, p. 313.
van Genuchten, M.T., 1980. A closed-form equation for predicting the hydraulic
conductivity of unsaturated soils. Soil Science Society of America Journal 44, 892-898.
Vanderkimpen, P.J., 1991. Estimation of crop evapotranspiration by means of the PenmanMonteith equation. Dept. of Biology and Irrigation Engineering. PhD dissertation, Dept.
of Biology and Irrigation Engineering, Utah State Univ., Logan, Utah.
Verdin, J., Klaver, R., 2002. Grid-cell-based crop water accounting for the famine early
warning system. Hydrological Processes 16, 1617-1630.
Vila, D.A., de Goncalves, l.G.G., Toll, D.L., Rozante, J.R., 2009. Statistical evaluation of
combined daily gauge observations and rainfall satellite estimates over Continental
South America. Journal of Hydrometeorology 10, 533-543.
184
Villarini, G., Krajewski, W.F., 2007. Evaluation of the research version TMPA threehourly 0.25°×0.25° rainfall estimates over Oklahoma. Geophysical Research Letters 34,
L05402.
Vliet, J.v., 2009. Assessing the accuracy of changes in spatial explicit land use change
models. 12th AGILE International Conference on Geographic Information Science
2009, Leibniz Universität Hannover, Germany.
von Hoyningen-Hune, J., 1983. Die interception des niederschlags in landwirtschaftlichen
bestanden. Schriftenreihe des DVWK 57, 1-53.
Wada, Y., van Beek, L.P.H., van Kempen, C.M., Reckman, J.W.T.M., Vasak, S.,
Bierkens, M.F.P., 2010. Global depletion of groundwater resources. Geophysical
Research Letters 37, L20402, doi:10.1029/2010GL044571.
Wagner, W., Lemoine, G., Rott, H., 1999. A method for estimating soil moisture from
ERS scatterometer and soil data. Remote Sensing of Environment 70, 191-207.
Wagner, W., Naeimi, V., Scipal, K., de Jeu, R., Martínez-Fernández, J., 2007. Soil
moisture from operational meteorological satellites. Hydrogeology Journal 15, 121-131.
Wallace, J.S., Gash, J.H.C., McNeil, D.D., Sivakumar, M.V.K., 1986. Measurement and
prediction of actual evaporation from sparse dryland crops-scientific report on Phase II
of ODA Project 149. ODA Rep. OD 149/3.
Wang, J., Rich, P.M., Price, K.P., 2003. Temporal responses of NDVI to precipitation and
temperature in the central Great Plains, USA. International Journal of Remote Sensing
24, 2345-2364.
Wang, J.R., Choudhury, B.J., 1981. Remote sensing of soil moisture content over bare
field at 1.4 GHz frequency. Journal of Geophysical Research 86, 5277-5282.
Wang, L., Wen, J., Zhang, T., Zhao, Y., Tian, H., Shi, X., Wang, X., Liu, R., Zhang, J.,
Lu, S., 2009. Surface soil moisture estimates from AMSR-E observations over an arid
area, Northwest China. Hydrology and Earth System Sciences Discussions 6, 10551087.
Wang, Q., Tenhunen, J.D., 2004. Vegetation mapping with multitemporal NDVI in North
Eastern China Transect (NECT). International Journal of Applied Earth Observation and
Geoinformation 6, 17-31.
Wang , X., Xie, H., Guan, H., Zhou, X., 2007. Different responses of MODIS-derived
NDVI to root-zone soil moisture in semi-arid and humid regions. Journal of Hydrology
340, 12-24.
WAPDA, 1965. Lower Indus report, physical resources-groundwater. Volume 6,
Supplement 6.1.3, 4 & 5, West Pakistan Water and Power Development Authority.
Wardlow, B.D., Egbert, S.L., 2008. Large area crop mapping using time-series MODIS
250 m NDVI data: An assessment for the U.S. Central Great Plains. Remote sensing of
Environment 112, 1096-1116.
Western, A.W., Blöschl, G., 1999. On the spatial scaling of soil moisture. Journal of
Hydrology 217, 203-224.
Wielicki, B.A., Barkstrom, B.R., 1997. Clouds and the earth’s radiant energy system
algorithm theoretical basis document. Atmospheric Sciences Division, NASA Langley
Research Center,Hampton, Virginia,USA.
Wilson, S., 2011. Preparation of sub regional plan for Haryana sub-region of NCR-2021.
Interim Report -II.
185
Winsemius, H.C., Savenije, H.H.G., Bastiaanssen, W.G.M., 2008. Constraining model
parameters on remotely sensed evaporation: justification for distribution in ungauged
basins? Hydrology and Earth System Sciences 12, 1403-1413.
Wipfler, E.L., Metselaar, K., van Dam, J.C., Feddes, R.A., van Meijgaard, E., van Ulft,
L.H., van den Hurk, B., Zwart, S.J., Bastiaanssen, W.G.M., 2011. Seasonal evaluation of
the land surface scheme HTESSEL against remote sensing derived energy fluxes of the
Transdanubian region in Hungary. Hydrology and Earth System Sciences 15, 12571271.
WMO, 2006. World meteorological organization guide to meteorological instruments and
methods of observation. WMO-No.8. Secretariat of the World Meteorological
Organization, Geneva, Switzerland.
Wolf, A.T., 1998. Conflict and cooperation along international waterways. Water Policy 1,
251-265.
Wolf, A.T., Natharius, J.A., Danielson, J.J., Ward, B.S., Pender, J.K., 1999. International
river basins of the world. International Journal of Water Resources Development 15,
387-427.
Wösten, J.H.M., Finke, P.A., Jansen , M.J.W., 1995. Comparison of class and continuous
pedotransfer functions to generate soil hydraulic characteristics. Geoderma 66, 227-237.
Wu, B., Yan, N., Xiong, J., Bastiaanssen, W.G.M., Zhu, W., Stein, A., 2012. Validation of
ETWatch using field measurements at diverse landscapes: A case study in Hai Basin of
China. Journal of Hydrology, doi: 10.1016/j.jhydrol.2012.1002.1043.
Xie, P., Arkin, P.A., 1996. Analysis of global monthly precipitation using gauge
observations, satellite estimates, and numerical model predictions. Journal of Climate 9,
840-858.
Yin, Z.Y., Zhang, X., Liu, X., Colella, M., Chen, X., 2008. An assessment of the biases of
satellite rainfall estimates over the Tibetan Plateau and correction methods based on
topographic analysis. Journal of Hydrometeorology 9, 301-326.
Yoo, C., 2002. A ground validation problem of remotely sensed soil moisture data.
Stochastic Environmental Research and Risk Assessment 16, 175-187.
Zawahri, N.A., 2006. Stabilizing Iraq’s water supply: What the Euphrates and Tigris
Rivers can learn from the Indus. Third World Quarterly 27, 1041-1058.
Zawahri, N.A., 2008. International rivers and national security:The Euphrates, Ganges–
Brahmaputra, Indus, Tigris, and Yarmouk rivers. Natural Resources Forum 32, 280-289.
Zawahri, N.A., 2009. India, Pakistan and cooperation along the Indus River system. Water
Policy 11, 1-20.
Zhang, X., Srinivasan, R., Liew, M.V., 2008. Multi-site calibration of the SWAT model
for hydrologic modeling Transactions of the ASABE 51, 2039-2049
Zwart, S.J., Bastiaanssen, W.G.M., de Fraiture, C., Molden, D.J., 2010. WATPRO: A
remote sensing based model for mapping water productivity of wheat. Agricultural
Water Management 97, 1628-1636.
186
Curriculum Vitae
Muhammad Jehanzeb Masud (Cheema) was born in Sargodha, Pakistan, on 31st July 1976.
In 2000, he completed B.Sc (Agricultural Engineering) with distinction from Faculty of
Agricultural Engineering and Technology, University of Agriculture, Faisalabad Pakistan. He received his M.Sc (Hons) degree in Agricultural Engineering with major in
Irrigation and Drainage in the year 2006. He has worked as an Asst. Agricultural Engineer
in Water Resources Research Institute-PARC and afterwards joined Department of
Irrigation and Drainage, University of Agriculture, Faisalabad as a lecturer in 2005.
In December 2007, Cheema joined TUDelft as PhD researcher, with the focus on spatial
modeling of water resources in the Indus Basin (the largest contiguous basin) which is
under threat of accelerated water scarcity but have limited data. He investigated the use of
multi-sensor satellite information to map complex hydrological processes and water
management practices in data scarce river basins. He carried out six months extensive
fieldwork in Pakistan to crosscheck the results of satellite data. Cheema presented his
work on several international conferences and seminars. The findings of his PhD research
resulted in publications in peer-reviewed journals.
187
Publications
Cheema, M.J.M., Immerzeel, W.W. , Bastiaanssen, W.G.M., 2012. Spatial quantification
of groundwater abstraction for irrigation in the Indus Basin using pixel information, GIS
and the SWAT model. Groundwater, (under review).
Bastiaanssen, W.G.M., Cheema, M.J.M., Immerzeel, W.W., Miltenburg, I., Pelgrum, H.,
2012. The surface energy balance and actual evapotranspiration of the transboundary
Indus Basin estimated from satellite measurements and the ETLook model. Water
Resources Research, (under review).
Cheema, M.J.M., Bastiaanssen, W.G.M., 2012. Local calibration of remotely sensed
rainfall from the TRMM satellite for different periods and spatial scales in the Indus
Basin. International Journal of Remote Sensing 33, 2603-2627.
Cheema, M.J.M., Bastiaanssen, W.G.M., Rutten, M.M., 2011. Validation of surface soil
moisture from AMSR-E using auxiliary spatial data in the transboundary Indus Basin.
Journal of Hydrology 405(1/2): 137-149.
Pelgrum, H., Miltenburg, I.J., Cheema, M.J.M., Klaasse, A., Bastiaanssen, W.G.M., 2011.
ETLook: A novel continental evapotranspiration algorithm. Remote Sensing and
Hydrology 2010, Proceedings of a Symposium held at Jackson Hole, Wyoming, USA,
September 2010, IAHS Publ. 3XX (in press) (2011).
Cheema, M.J.M., Bastiaanssen, W.G.M., 2010. Land use and land cover classification in
the irrigated Indus Basin using growth phenology information from satellite data to
support water management analysis. Agricultural Water Management 97, 1541-1552.
Kiptala, J., Mohamed, Y., Cheema, M.J.M., Mul, M., Van der Zaag, P., 2012. Land use
and land cover classification using Phenological variability from MODIS vegetation in the
upper-Pangani river basin, Tanzania. Agricultural Water Management (under review).
188
189
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Download PDF

advertisement