Yangtze Delta floods and droughts of the last millennium: Abrupt

Yangtze Delta floods and droughts of the last millennium: Abrupt
Theor. Appl. Climatol. 82, 131–141 (2005)
DOI 10.1007/s00704-005-0125-4
1
2
Nanjing Institute of Geography and Limnology, Chinese Academy of Sciences (CAS), Nanjing, China
Meteorologisches Institut, Universit€at Hamburg, Hamburg, Germany
Yangtze Delta floods and droughts of the last millennium:
Abrupt changes and long term memory
T. Jiang1 , Q. Zhang1 , R. Blender2, and K. Fraedrich2
With 8 Figures
Received March 2, 2004; revised December 6, 2004; accepted January 1, 2005
Published online April 26, 2005 # Springer-Verlag 2005
Summary
Climate variability and flood events in the Yangtze Delta,
which is a low-lying terrain prone to flood hazards, storm
tides and typhoons, are studied in terms of a trend and
detrended fluctuation analysis of historical records. The data
used in this paper were extracted from historical records
such as local annuals and chronologies from 1000–1950 and
supplemented by instrumental observations since 1950. The
historical data includes frequencies of floods, droughts and
maritime events on a decadal basis. Flood magnitudes
increase during the transition from the medieval warm
interval into the early Little Ice Age. Fluctuating climate
changes of the Little Ice Age, which are characterised by
arid climate events, are followed by wet and cold climate
conditions with frequent flood hazards. For trend analysis,
the Mann-Kendall test is applied to determine the changing
trends of flood and drought frequency. Flood frequency
during 1000–1950 shows a negative trend before 1600 A.D.
and a positive trend thereafter; drought frequency increases
after 1300. The detrended fluctuation analysis of the flood
and drought frequencies reveals power law scaling up to
centuries; this is related to long-term memory and is similar
to the river Nile floods.
1. Introduction
Climate warming, flood hazards and their impacts on human society have received increasing attention in China during recent years,
particularly as agriculture, industry and even the
development of national economies suffer losses
from flood and water logging hazards (IPCC
2001). The Yangtze Delta is an extremely densely populated and economically developed area
with Mega-cities such as Shanghai, Nanjing, and
Hangzhou. Shi (2003) and Zhang et al. (2001)
show that changes of flood occurrences may
reflect climate transitions. Therefore, understanding the principles governing flood occurrence is
theoretically and practically meaningful for local
mitigation and the reduction of flood induced
disasters. A first step towards an understanding
of the underlying processes is the analysis of
floods and climate change in the past. Research
on past climatic fluctuations in China (An, 1986;
Wang et al., 1991; Zhang et al., 1995; Gong and
Harneed, 2000; James et al., 2000; Zhang et al.,
2001) has identified abundant historical records
and exceptional information, which is a valuable
historical heritage to the study of climate variability. Documentation of Yangtze Delta floods is
based on relatively complete and long historical
records of climatic events and human activities.
Historical records (Zhang et al., 2002) indicate
that the Yangtze Delta is typical for the lower
Yangtze River in both its geomorphologic characteristics and climatic change patterns. In this
study, the historical records are extended by
132
T. Jiang et al.
instrumental observations of precipitation, temperature and discharge available from 1950 to
2002.
The analysis presented here is an extension of
ongoing research on China’s water cycle and
climate. Commencing with an in-depth description of the quantification of historical climate
data and its transfer into quantitative time series
(Wang et al., 1991), the necessary next step
was to demonstrate historical climate variability.
Fluctuations of flood and drought indicators,
which characterise the eastern parts of China,
were analysed (by wavelet methods) to identify
water cycle related regimes over the past millennium (Jiang et al., 1997). Based on instrumental
precipitation records, a novel monitoring algorithm, the standardised precipitation index (SPI),
was introduced (Bordi et al., 2004). Thus, a
unique set of analyses has been accomplished,
ranging from historical data extraction and climate variability analysis to monitoring and practical prediction of wetness=drought regimes.
These analyses will be extended with the application of a novel nonlinear systems analysis technique used to detect long term memory as an
underlying mechanism for water cycle variability
in eastern China, in particular, in the Yangtze
Delta area.
First steps towards analysing regional climate
variability are commonly associated with climate
change, which is described and quantified in
terms of internal fluctuations manifested in
power spectra or similar methods. An example of
these methods is the detrended fluctuation analysis (DFA), which has been developed recently
(Peng et al., 1994) to extract the scaling properties of the fluctuations and thereby to characterise the complex dynamics underlying climate
variability. This analysis technique has been applied systematically to temperature time series of
last century observations and of coupled global
ocean-atmosphere circulation models (Fraedrich
and Blender, 2003, loc. cit.). The time scales
resolved by this method are, due to sampling,
limited by about a tenth of the length of the time
series. Beyond that, trend analysis is used to
identify an abrupt climate change in the time
series. The millennium time series of Yangtze
Delta floods provides an ideal testbed for the
application of both the established and novel
methods to gain insight into the underlying
dynamics and to provide appropriate measures
for comparison with climate models simulating
both control (trend-free) and close to reality
(trend) situations.
We describe the climatological embedding and
the data sets to be analysed (Section 2) followed
by the methods applied (Section 3). The results
for the historical climate, trends and long term
memory are presented in Section 4 and summarised in Section 5.
2. Climatological setting and data
The Yangtze Delta (30 N to 33 N, 119 E to
122 E) is located in Eastern China and characterised by the subtropical monsoon climate
(Fig. 1). Mixed deciduous and evergreen forest
determines the natural vegetation of the region.
The mean annual temperature is 15.5 C. In summer, the area is dominated by the Subtropical
High with maximum temperatures reaching
28.9 C; in winter, the region is influenced by
the Mongolian High with minimum temperatures
around 3.3 C. The mean annual precipitation is
1235 mm with summer rainfall (June–August) accounting for 40% of the total and only 11% occur
during winter months (December–February).
Climatologically, the region is sensitive because
it lies along the demarcation between subtropical
and temperate climates characterised by significantly disparate air masses (Daniel et al., 1996).
Thus, it is affected by maximum floods that
Fig. 1. The Yangtze Delta with the key pilot study region
(shaded)
Yangtze Delta floods and droughts of the last millennium
mostly result from excess rainfall during summer, especially June and July (the so-called Plum
Rainy Season) when slow-drifting cold fronts
encounter moist air masses of subtropical origin
and also in August which is characterised by
typhoon-induced excessive precipitation. A nearly level plain with an elevation of 2 to 7 m above
sea level covers 75% of the region. This lowlying terrain makes the region vulnerable to
floods, maritime tidal and other natural hazards
(Daniel et al., 1996). In addition, low-lying terrain causes retarded discharge of ground surface
water and therefore, excessive precipitation
usually results in floods. The Yangtze Delta is a
133
well-developed area in industry and agriculture
and densely populated with sufficient historical
and instrumental climatological data available.
The information used for our analysis are
annual time series characterising flood and
drought occurrence of the middle Yangtze and
the Yangtze Delta for the period 1000 to 1950.
The data base consists of historical documents
(1000–1950, Hu and Luo, 1988 and MICMB,
1981) from which quantitative information can
be extracted. A grading, GðtÞ, is used for each
year t during the last millennium to establish a
time series, which characterises individual flood
(G ¼ 2 and 1 for heavy flood and flood),
Table 1. Frequency of the floods, drought events, maritime tidal events (Number of natural hazards per ten years) extracted
from historical documents
Year
Flood
Drought
Maritime
events
Year
Flood
Drought
Maritime
events
Year
Flood
Drought
Maritime
events
1005
1015
1025
1035
1045
1055
1065
1075
1085
1095
1105
1115
1125
1135
1145
1155
1165
1175
1185
1195
1205
1215
1225
1235
1245
1255
1265
1275
1285
1295
1305
1315
1325
1335
1345
1355
1
0
4
0
0
0
1
0
2
1
1
3
0
1
0
2
0
0
0
2
5
1
1
0
0
0
0
0
2
2
1
0
1
0
0
0
2
3
0
3
2
0
0
2
1
0
3
2
1
5
2
1
1
0
4
3
4
2
0
0
2
0
0
0
0
3
1
0
2
2
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
1
0
0
0
1
0
0
0
0
0
0
0
0
0
1
0
0
0
1
2
1
1
0
1365
1375
1385
1395
1405
1415
1425
1435
1445
1455
1465
1475
1485
1495
1505
1515
1525
1535
1545
1555
1565
1575
1585
1595
1605
1615
1625
1635
1645
1655
1665
1675
1685
1695
1705
0
0
1
0
1
1
0
1
4
4
2
1
0
1
0
2
3
2
2
4
7
9
4
6
3
4
3
4
1
2
3
7
1
4
3
0
0
0
1
0
0
1
3
1
1
1
3
5
0
5
3
4
3
2
4
1
0
2
0
1
4
2
6
3
5
1
2
1
0
1
0
1
1
1
0
0
1
0
0
0
1
0
0
0
0
2
1
1
0
0
0
0
2
0
0
0
0
0
1
1
3
0
0
0
0
1715
1725
1735
1745
1755
1765
1775
1785
1795
1805
1815
1825
1835
1845
1855
1865
1875
1885
1895
1905
1915
1925
1935
1945
1950
1955
1960
1965
1970
1975
1980
1985
1990
1995
2000
3
0
4
4
3
1
1
1
0
1
2
3
5
2
1
1
1
2
0
4
3
1
1
1
2
1
1
2
2
2
2
2
0
2
4
3
1
1
1
2
1
3
5
1
7
3
1
2
1
3
2
5
1
3
0
2
4
3
0
0
1
1
2
3
2
2
3
2
3
2
0
2
3
1
2
0
1
0
3
3
0
0
0
1
2
1
0
1
0
0
0
0
0
0
3
1
0
0
0
0
1
1
1
1
2
134
T. Jiang et al.
normal (G ¼ 0), and drought (drought and heavy
drought, G ¼ 1 and 2) events, hence GðtÞ ¼ 2,
1, 0, 1, 2. For example, ‘‘1335, tidal inundation
occurred accompanying floods, many people
died’’ (2-grade flood), ‘‘1512, July, it is windy,
tidal inundation occurred. Houses and people
were damaged and in inundation in thousands’’
(1-grade flood), ‘‘1539, tide level rose rapidly,
the depth of the ground surface water is tens of
meters, thousands of people died’’ (1-grade
flood). Some historical records indicate that
floods are partly a direct result of tidal events.
Here we use frequency statistics derived from
the grade time series GðtÞ, to make it comparable
with other millennium data and define decadal
flood and drought events F and D (Table 1).
Droughts are defined by dðtÞ ¼ 1 for G>0, and
floods, f ðtÞ ¼ 1 for G<0. These yield the decadal sums Dðtc Þ ¼ t dðtÞ and Fðtc Þ ¼ t f ðtÞ for
the decadal times tc during the last millennium,
tc ¼ 1005, 1015, 1025, . . . , and the years involved
in the sums are t ¼ tc þ ð5; 4; . . . ; þ4Þ. In
addition, the number M of marine tidal events
per decade is extracted from historical documents in a similar manner (Table 1). Finally, a
humidity parameter I is introduced (Zheng,
1997), which depends on the decadal frequencies
of floods and droughts, F and D (published in
MICMB, 1981, see Table 1)
FD
ð1Þ
I¼2
FþD
If the occurrences of flood and drought events are
equal, I ¼ 0 (also if no flood and drought events
are measured); if the climate is wet or dry, I >1
or I <1. The decadal time series of floods,
Table 2. The flood=drought grade criteria for instrumental
precipitation data (1950–2002) in the Yangtze Delta. Ri is
the precipitation at each station in the Yangtze Delta from
the multi-year average
May to September since 1950, R
precipitation and is the standard deviation (MICMB, 1981)
Index
þ 1:17Þ
Ri > ðR
þ 0:33Þ < Ri ðR
þ 1:17Þ
ðR
0:33Þ < Ri ðR
þ 0:33Þ
ðR
1:77Þ < Ri ðR
ðR 0:33Þ
1:77Þ
Ri ðR
Grade
Type
2
1
Heavy flood
Flood
0
1
Normal (harvest year
or missing records)
Drought
2
Heavy drought
drought and maritime events, D, F, M, and the
humidity index, I, are subjected to trend and
detrended fluctuation analysis.
The historical data are complemented by instrumental observations after 1950 based on the
NCEP reanalysis data set (Kalnay et al., 1996).
The grades GðtÞ during the instrumental period
are determined by the precipitation using the criteria in Table 2. An SST (sea surface temperature) index is used which characterises El Nino=
Southern Oscillation events (COAPS, 2004). This
index is defined as the 5-month running mean of
the monthly SST anomalies averaged over the
areas 4 N to 4 S and 150 W to 90 W.
3. Analysis methods
Two methods are applied to determine climate
change and low frequency variability in the historical and instrumental data: (1) The detection
of trends and trend changes is performed by the
Mann-Kendall-test, a rank-test that determines
a single trend change in a time series. (2) The
low frequency variability is determined by the
detrended fluctuation analysis, an extension of
the fluctuation analysis with the particular capability to investigate long time memory in time
series. These methods of analysis are described
below.
3.1 Abrupt changes: Mann-Kendall test
Abrupt changes and trends of floods and maximum summer temperatures are diagnosed by the
Mann-Kendall rank test (MK, as suggested by
Sneyers, 1975; Goossens and Berger, 1987). This
technique has been applied to dryness (in China,
Jiang and Qian, 1997) and to flood levels (Nile
river, Fraedrich et al., 1997, see also Jiang et al.,
2002).
Estimating abrupt changes using the MannKendall test (MK) is based on a progressive analysis to test the beginning of a single trend
(within a given sample time series) utilising
related progressive and retrogressive rank series.
MK is suitable to detect one abrupt change in a
sample time series of anomalies that are independent and Gaussian distributed. First, assume
that the time series is in progressive sequence,
x1 ; x2 ; x3 . . . xN for a given sample size N. The
accumulated total of values xj smaller than xi
Yangtze Delta floods and droughts of the last millennium
for j<i up to i is denoted as mi . The sum of the
first of these terms is the rank statistic dk
k
X
dk ¼
mi ; for k ¼ 2; . . . ; N
ð2Þ
i
This sum is standardised to a normal distribution
by
dk E½dk ð3Þ
Z1 ðkÞ ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi
var½dk with the mean E½dk ¼ kðk 1Þ=4 and the variance var½dk ¼ kðk 1Þð2k þ 5Þ=72.
A trend of the time series is significant at
the threshold Ua if Z1 ðkÞ>Ua . For the standard
significance level a ¼ 0:05 the threshold is
U0:05 ¼ 1:96. To identify an abrupt climate
change, this analysis is performed with the same
time series but with the retrograde sequence, i.e.
xN ; xN1 ; . . . ; x1 , and yields Z1 ðkÞ ¼ Z2 ðkÞ for
k ¼ N; N 1; . . . ; 1. An abrupt change is
initiated at the intersection Z1 ðkÞ and Z2 ðkÞ.
3.2 Long term memory: Detrended
fluctuation analysis (DFA)
Long term memory is a particular case of enhanced low frequency variability when the time
integral of the auto-correlation function diverges.
Due to this, the correlation time is infinitely
large. In practice, due to the finite range of any
observed or modelled time series, long term
memory is considered to exist if there is no exponential decay up to the maximum time scale.
The detrended fluctuation analysis (DFA) has
been developed to determine the variability of
time series on different time scales (Peng et al.,
1994) and is becoming an ideal tool to derive
long term memory (e.g. Fraedrich and Blender,
2003). The DFA determines the fluctuation function FðtÞ, however, and this is the difference to
the standard fluctuation analysis, for the integrated time series, the so-called profile. The effect of
the integration is reversed after the calculation.
For this profile, linear fits are determined for all
segments of different widths t and the fluctuation
function FðtÞ is calculated as the standard deviation between the profile and the time series. The
main advantage of the DFA is that long time
correlations become visible by the integration.
For scaling behaviour, which shows up as a power
law in FðtÞ, there are direct relationships to the
135
correlation function and the power spectrum of
the time series: if FðtÞ t the power spectrum
follows Sð f Þ f with ¼ 2 1, and the correlation function is CðtÞ t with ¼ 1 .
The exponent equals the Hurst exponent
(1951) obtained by the rescaled range analysis.
Stationary long term memory is present within
0< <1 (1=2<<1). The two limiting cases
are uncorrelated time series (white noise) with
¼ 0 ( ¼ 1=2) and the 1=f-spectrum (flicker
noise) with ¼ 1 ( ¼ 1). For >1 (>1) the
time series is non-stationary and conventional
statistical analysis cannot be applied. In the case
of dominant oscillations the fluctuation function
FðtÞ shows step-like transitions at the oscillation
periods, and hence, the result may be difficult to
interpret for a multitude of oscillations. Therefore, the DFA should be combined with a power
spectrum analysis.
The DFA analysis, which does not – in contrast to its name – eliminate trends in the data,
can be extended to higher order versions DFA-N,
which fit and subtract higher order polynomials
of degree N to the profile (DFA-1 represents the
standard DFA). In order to avoid any confusion
with the trend analysis above, it is necessary to
differentiate the time scales involved. For DFA-N
the time scales of interest are well below the total
available duration of the time series since the
same conditions for significance as for a power
spectrum apply. Therefore, the time scale of the
polynomials subtracted in DFA-N is below about
100 years. On the other hand, the trend analysis,
which is another main topic in the present study,
is applied to climate changes on the centennial
time scale. The essential result of the DFA is the
decadal variability and thus complements the
Mann-Kendall analysis.
4. Results: Climate history, trends
and long term memory
The historical data are decadal flood and drought
frequencies during the last millennium (from
these a humidity parameter is derived) and the instrumental data precipitation, temperature and the
SST index characterising the El Nino=Southern
Oscillation (COAPS, 2004). They are analysed in
three steps: First the historical and instrumental
data time series are presented with an overview
of wet and dry periods. Then, a Mann-Kendall
136
T. Jiang et al.
trend analysis is applied to the historical and
instrumental data, and finally, the long term
memory is determined in the historical data.
4.1 The historic and instrumental data
Historical data: Figure 2 demonstrates the connections between flood events, climatic changes
and maritime tidal events. Eight frequentlyoccurring flooding periods are marked as the
shaded zones a–h. A time interval during 1500
and 1800, which is characterised by a high frequency of floods, is the Little Ice Age (see, for
example, Bradley et al., 1992; Huang, 2000;
Yang et al., 1995). In our key region the Little
Ice Age is characterised by high wetness, I >1,
with alternating wet and dry climate episodes.
Larger floods are noted from 1500 to 1700,
which characterises the transition from the medieval warm interval to the cooler Little Ice Age
(temperatures taken from Wang et al., 2002). The
same transition occurred in the upper Mississippi
River (James, 1993). It can also be seen from
Fig. 2 that the occurrences of maritime tidal and
flood events are correlated. As these eight shaded
zones show, periods with tidal events and floods
occurred almost simultaneously (except zone a
and zone b), the maritime tidal events had a more
significant impact on the occurrence of the floods
in the Yangtze Delta than other more direct
effects. The low-lying terrain makes the study
region prone to impacts of floods and extreme
tides. Figure 2 also demonstrates that drought
and flood events usually occurred concurrently.
Fig. 2. Flood events, tidal events
and temperature changes evolving
in the Yangtze Delta. The dashed
lines in panel IV and V denote the
50-year moving average, in panel I
they denote zero degrees. The shaded
zones mark periods of higherfrequency floods (the data are listed
as the appendix) coinciding with
climate changes of China (Wang
et al., 2002)
Yangtze Delta floods and droughts of the last millennium
137
Fig. 3. Mean sea surface temperature
(SST) index, flood discharge, and annual
precipitation in the lower reaches of the
Yangtze River from instrumental records
(1950 to 2002); discharge with 4-year
moving average (dashed) of rivers in the
Yangtze Delta
We can see that these eight flooding periods are
also the drought periods (flooding events and
drought events occurred at the same year, e.g.
the flood events occurred in the spring and the
drought event occurred in the autumn). It should
be mentioned here that the connections between
flood events and tidal inundation are not significant, because tidal inundation events are not the
decisive cause of flood occurrence.
Instrumental data: Close connections between
annual precipitation, flood discharge and SST
index, demonstrated in Fig. 3, occur in certain
periods marked by shading. This may be the reason why correlations between SST, annual precipitation and discharge are not significant. These
three shaded zones (Fig. 3) show three periods,
which are characterised by greater flood discharge (see 4-year running mean). Figure 3 also
indicates that these three periods correspond to
higher mean SSTs. These three periods (shaded
zones) also match the frequently-occurring
periods of floods in the lower Yangtze River.
Therefore, we can tentatively draw the conclusion that higher SSTs appear to enhance the
hydrological cycle linking ocean and continent,
with excessive precipitation and the increasing
possibility of larger floods. Figure 4 shows correlations between the annual precipitation and
flood discharge of the Yangtze Delta, with linear
connections between flood-season discharge
(averages from June to September) and annual
precipitation. The regression fit is high ðR2 ¼
0:87Þ and demonstrates significant impacts of
precipitation on floods in the study region.
Flood-season discharge, annual precipitation
and SST index: The Pearson correlations between
SST index and annual precipitation and discharge
is weakly positive. A highly significant correlation of 0.935 exists between annual precipitation
and discharge (two-tailed significance at 99%).
Note that a general characteristic of the monsoon
climate is the strong correlation between precipitation and floods in the rainy summer season
(June to August).
4.2 Trend analysis of historical
and instrumental data
Fig. 4. Correlation between annual precipitation and the
flood discharge during 1950–2002 at a tributary river basin
of the Yangtze Delta
The Mann-Kendall outcome is presented in Fig. 5.
The historical sample (1000 to 1950) shows a
negative trend in flood frequency before 1600
(Z1 curve, at 95% significance) and a weak positive trend thereafter, from 1600 to 1950. The discontinuity of the trend in flood frequency occurs
around 1550 (95% significance). For drought
frequency a positive trend is found from 1400
138
T. Jiang et al.
Fig. 5. Mann-Kendall analysis results of the flood frequency (a) and drought frequency (b) during 1000–1950
in the Yangtze Delta. The two dashed lines denote the 95%
confidence level
Fig. 6. Mann-Kendall analysis result of the annual precipitation (a) and that of the extreme high summer temperature (b) during 1950–2002 in the Yangtze Delta. The two
dashed lines denote the 95% confidence level
(Z1 curve) until present-day, however we do not
detect a discontinuity in the time series.
In the instrumental sample (Fig. 6) an abrupt
change in the annual precipitation time series
occurs in the mid 1960s (passing the significance
level of 95%). The Z1 -curve demonstrates a
negative tendency in the summer precipitation
(also passing the 95% level). After the 1970s,
precipitation changes are negligible.
However, looking at the fit line with negative
slope, the general trend of annual precipitation
is negative. The original annual precipitation
curves (Fig. 6a) demonstrate this negative trend
after 1990. Figure 6b indicates that the discontinuity of the trend of the maximum summer temperature is in the mid 1990s, around 1993. The
Z1 -curve demonstrates that the maximum summer temperature changes follow a negative trend
before 1990s and positive trend after the 1990s
(satisfying a 95% significance). This changing
point can be seen in the 4-year moving average
curves (Fig. 6b). While the flood frequency
shows a negative trend before the 1990s, a positive tendency is found after the 1990s, which
matches the maximum summer temperatures of
the study region. This suggests an increasing
probability of maximum summer temperature
(under climatic warming scenarios, James, 2000;
Mirza, 2002; Deng et al., 2000), which may lead
to an increasing occurrence of flood events.
4.3 Long term memory of historical data
The frequencies of floods, droughts and the
deduced humidity parameter Eq. (1) are subjected to DFA to determine the functional relationship of the long term memory. For the
duration of 1000 years, results in the time range
up to 100 years can be reliably derived. Figure 7
shows the fluctuation functions in a log–log-plot
where power laws follow straight lines. Floods
and droughts reveal power law exponents above 1=2 which represent long term memory
up to different time scales: 200 years for floods
and without restriction for droughts. The exponents are ¼ 0.69 ( ¼ 0.38) for droughts and
¼ 0.76 ( ¼ 0.52) for the frequency of floods.
These results have been compared with the
Yangtze Delta floods and droughts of the last millennium
Fig. 7. Fluctuation functions FðtÞ for the decadal time
scale t for floods (), droughts (), and the humidity parameter (þ) obtained by DFA. The lines describe power laws
with exponents as indicated
higher order DFA-2 and DFA-3 to elucidate the
role of superimposed linear and quadratic trends
on time scales below 100 years: both methods
(not shown) reveal the analogous power laws
and the exponents agree with DFA-1 findings.
Note that the trends investigated in the preceding
section are determined above this time scale. A
possible origin for the long term memory of the
hydrological cycle in this region is the sea surface temperature of the East China Sea. The
humidity parameter shows a white spectrum
( ¼ 0.5, ¼ 0) and no long term memory. The
most probable reason is that, by its definition
Eq. (2), the long term memory effects of floods
and droughts may be obscured.
Fig. 8. Power spectrum Sðf Þ for spectral frequency f of the
maritime events with the maximum intensity at 70 years.
Sðf Þ is averaged in three non-overlapping segments of the
total time series
139
The frequency of maritime events shows a different outcome for the long term variability. Maritime events are rarely found before 1300 A.D.
which reduces the total variability and the significance of the results. According to a power spectrum analysis (Fig. 8, determined as an average
of three disjoint segments) an oscillation with a
period around 70 years may exist, however, this
remains speculative in the available time range.
The overall behaviour of the power spectrum
hints to an absence of long term correlations.
Due to the presence of the oscillation the DFA
fluctuation function (not shown) cannot reveal a
power law behaviour and, from this point of view,
the question of long term variability in the frequency of maritime events remains unsettled.
5. Summary and conclusions
Yangtze Delta flood data covering up to a 1000
years are analysed in terms of trend changes and
low frequency variability in the range of decades
to centuries. Rapid and strong fluctuations of the
climate (frequent changes of wet and arid conditions) occur during the Little Ice Age (1500 to
1850, see Bradley, 1992) with frequent flood
events in the Yangtze Delta. Larger floods occurred from about 1500 to 1700 and characterise
the transition from the medieval warm episode to
the cooler Little Ice Age (similar to the upper
Mississippi River (James, 1993)). The low-lying
terrain of the study region means that flood tides
are an important contributor to floods events.
Indeed, according to historical records, some
floods are the direct result of flood tidal events.
This result appears to be beneficial when considering future mitigation and flood hazard reduction for the Yangtze Delta.
Analyses of abrupt changes and trends in the
historical records indicate that, during 1000 to
1950, the flood frequency reveals a negative
trend before 1600 A.D. and a positive trend
thereafter (from 1600 to 1950, but not at 95%
significance); from 1700 to 1800 this trend is
positive with 95% significance. Instrumental
records of annual precipitation indicate an abrupt
change occurred in the 1990s. Prior to this, annual precipitation shows a negative trend, whilst
post-1990s the discharge trend reverses to become positive. Similarly, the maximum monthly
summer temperatures show an abrupt change in
140
T. Jiang et al.
the mid 1990s. The Z1 ðkÞ curves of annual precipitation and maximum temperature series
show a negative trend before 1990s and positive
trend after 1990. Deng et al. (2000) suggest that
climatic warming may lead to a high occurrence
probability of maximum summer temperatures.
Based on the close relationship between changes
in annual precipitation and the maximum summer temperature, leads us to surmise that maximum summer temperatures may be greater in
the future under a warming scenario, due to
the anticipated increases in serious summer rainfall. Still, flood control remains an arduous political and engineering problem which requires
serious consideration if future climatic warming
is expected for the Yangtze area. Pearson correlation analysis indicates that close connections
occur between annual precipitation, flood discharge and mean SST (99% level). Therefore,
external forcing, that is changing SSTs, may
be helpful in estimating future flood hazard
probabilities.
The detrended fluctuation analyses of the flood
and drought fluctuations show long term memory
with a power law scaling. The estimated power
law exponents for the millennium data set are
distinctly different from white noise and yield
the related spectra Sð f Þ f 0:5 for the flood frequency and Sð f Þ f 0:4 for the drought frequency. The long term memory extends at least
up to 100 years and appears not to be restricted
by the total available millennium time range for
the flood frequency. The frequencies of the maritime events however, indicate a period of 70
years and long term correlation cannot be proven. The most likely reason for the observed long
term memory is the sea surface temperature of
the adjacent East China Sea which delivers warm
and moist air during the monsoon phases.
Acknowledgements
This paper is financially supported by the key project of
Chinese Academy of Sciences (KZCX3-SW-331), Preparation grant of the Nanjing Institute of Geography and Limnology, CAS (No. SS220007), National Natural Science
Foundation of China (Grant No. 40271112), National Postdoctoral Research Programme of China and the K. C. Wong
Education Foundation, Hong Kong. We also would like to
thank the National Climate Centre of China Meteorological
Agency to provide us the instrumental data. We should
extend our thanks to Prof. Chen Jiaqi in the Nanjing Institute
of Geography and Limnology, CAS, for providing part
of valuable historical data of climatic changes of the
study region. The support by the Deutsche Forschungsgemeinschaft (FR450) is acknowledged.
References
An ZS (1986) Environmental variation of past 20000 years
of China. In: Liu TS, An ZS (eds) Loess Global changes.
Beijing: Science Press, pp 1–26 (In Chinese)
Bordi I, Fraedrich K, Jiang J, Sutera A (2004) Fluctuations
and patterns of dry and wet periods: Eastern China. Theor
Appl Climatol 79: 81–91
Bradley RS, Jones PP (eds) (1992) Climate since A.D. 1500.
London: Routledge, pp 1–31
COAPS (2004) ENSO Index According to JMA SSTA,
Center for Ocean-Atmospheric Prediction Studies, ftp:==
www.coaps.fsu.edu=pub=JMA_SST_Index
Daniel JS, Chen ZY (1996) Neolithic settlement distribution
as a function of sea level-controlled topography in the
Yangtze delta, China. Geology 12: 1083–1086
Deng ZW, Ding YG, Chen YG (2000) Influences of global
warming on the occurrence probability of the maximum
summer temperature in the Yangtze Delta. Journal
of Nanjing Institute of Meteorology 23(1): 42–47
(In Chinese)
Fraedrich K, Jiang J, Gerstengarbe FW, Werner PC (1997)
Multiscale detection of abrupt climate changes: Application to Nile river flood levels. Int J Climatol 17: 1301–1315
Fraedrich K, Blender R (2003) Scaling of atmosphere and
ocean temperature correlations in observations and climate models. Phys Rev Lett 90: 108501
Goossens C, Berger A (1987) How to recognize an abrupt
climate change? In: Berger WH, Labeyrie LD (eds)
Abrupt climate change, evidence and implications.
NATO ASI series C: Mathematical and Physical Sciences,
Vol. 216. Dortrecht: D. Reidel, pp 31–46
Hu MS, Luo CZ (1988) Chinese historical great flooding
events. Beijing: Book Store Press
Huang CC (2000) Climatic changes. Beijing: Science Press,
pp 140–156 (In Chinese)
Hurst HE (1951) Long-term storage capacity of reservoirs.
Trans Am Soc Civ Eng 116: 770–808
IPCC (Intergovernmental Panel on Climate Change) (2001)
Climate Change 2001: The Scientific Basis. Cambridge:
Cambridge University Press, 785 pp
James CK (1993) Large increase in flood magnitude in
response to modest changes in climate. Nature 361:
430–432
James CK (2000) Sensitivity of modern and Holocene floods
toclimate change.Quaternary ScienceReviews19: 439–457
Jiang J, Zhang D, Fraedrich K (1997) Historic climate
variability of wetness in East China (960–1992): A wavelet analysis. Int J Climatol 17: 969–981
Jiang J, Mendelssohn R, Schwing F, Fraedrich K (2002)
Coherency detection of multiscale abrupt changes in historic Nile flood levels. Geophys Res Lett 29(8): 112-(1-4)
Kalnay E et al (1996) The NMC=NCAR reanalysis project.
Bull Amer Meteor Soc 77: 437–471
Yangtze Delta floods and droughts of the last millennium
Meteorological Institute of the Chinese Meteorological
Bureau (Chief editor) (MICMB) (1981) Atlas of the
Flood=Dryness in China for the Last 500-Year Period.
Beijing, China: Cartological Press
Mirza MMQ (2002) Global warming and changes in the
probability of occurrence of floods in Bangladesh
and implications. Global Environmental Changes 12:
127–138
Peng CK, Buldyrev SV, Havlin S, Simons M, Stanley HE,
Goldberger AL (1994) On the mosaic organization of
DNA sequences. Phys Rev E 49: 1685–1689
Shi YF (2003) Estimation of climatic transition from warm
dry to warm moist in the north-western China. Beijing:
Meteorological Press, pp 1–120 (In Chinese)
Sneyers R (1975) Sur l’analyse statistique des series d’observation. WMO TN 143, World Meteorological Organisation, Geneva
Wang RS, Wang R, Fraedrich K (1991) An approach
to reconstruction of temperature on seasonal basis
using historical documents from China. J Climatol 11:
381–392
Wang SW, Xie ZH, Cai JN, Zhu JH, Gong DY (2002) Global
mean temperature during past 1000 years. Progress in
Natural Sciences 12: 1145–1149
Yang HR, Xu X, Yang DY (1995) Environment changes and
Geo-ecosystem of the lower Yangtze River. Nanjing:
Press of Hohai University, pp 20–50 (In Chinese)
141
Zhang D, Liu C, Gu X (1996) A research of reconstruction of 6
regional dry=wetness grades series in the east China for the
last 1000 years. In: Proceedings on the National Research
Project (85–913) (eds) Collection of research papers on
climate change and its numerical simulation (vol. 1).
Beijing: Meteorological Press, pp 69–74 (in Chinese)
Zhang Q, Zhu C, Chen Ji (2002) Preliminary study on the
flooding and drought calamity during past 1500 years in
the Haian region. Chinese Geographical Science 12(2):
146–152
Zhang Q, Zhu C, Jiang FQ (2001) Environmental archaeological research of the Zhangjiangwan site of past 2000
years, Wushan, Chongqing, Acta Geographica Sinica 3:
353–362 (In Chinese)
Zheng SZ (1977) Changes of climatic humidity during past
2000 years in the South-eastern China. In: China Meteorological Agency (eds) Nanjing Institute of Geography and
Limnology Collection of climatic changes and super-long
prediction. Beijing: Science Press, pp 29–32 (In Chinese)
Authors’ addresses: Tong Jiang, Qiang Zhang, Nanjing
Institute of Geography and Limnology, Chinese Academy
of Sciences (CAS), Nanjing 21008, China; Richard Blender,
Klaus Fraedrich (e-mail: [email protected]), Meteorologisches Institut, Universit€at Hamburg, 20146 Hamburg,
Germany.
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