Local- and landscape-scale impacts of clear-cuts organic carbon in boreal forests

Local- and landscape-scale impacts of clear-cuts organic carbon in boreal forests
PUBLICATIONS
Journal of Geophysical Research: Biogeosciences
RESEARCH ARTICLE
10.1002/2015JG003190
Key Points:
• Climate change and forestry will have
a limited impact on boreal [DOC]
• Climate and forestry effects are not
additive at a local scale
• Climate and forestry effects are
additive at a landscape scale
Correspondence to:
M. N. Futter,
[email protected]
Citation:
Oni, S. K., T. Tiwari, J. L. J. Ledesma,
A. M. Ågren, C. Teutschbein, J. Schelker,
H. Laudon, and M. N. Futter (2015),
Local- and landscape-scale impacts of
clear-cuts and climate change on surface
water dissolved organic carbon in boreal
forests, J. Geophys. Res. Biogeosci., 120,
2402–2426, doi:10.1002/2015JG003190.
Received 24 AUG 2015
Accepted 23 OCT 2015
Accepted article online 28 OCT 2015
Published online 25 NOV 2015
Local- and landscape-scale impacts of clear-cuts
and climate change on surface water dissolved
organic carbon in boreal forests
Stephen K. Oni1, Tejshree Tiwari1, José L. J. Ledesma2, Anneli M. Ågren1, Claudia Teutschbein3,
Jakob Schelker4, Hjalmar Laudon1, and Martyn N. Futter2
1
Department of Forest Ecology and Management, Swedish University of Agricultural Sciences, Umeå, Sweden,
Department of Aquatic Science and Assessment, Swedish University of Agricultural Sciences, Uppsala, Sweden,
3
Department of Earth Sciences, Uppsala University, Uppsala, Sweden, 4Department of Limnology and Bio-Oceanography,
University of Vienna, Vienna, Austria
2
Abstract Forest harvesting and climate change may significantly increase concentrations and fluxes of
dissolved organic carbon (DOC) in boreal surface waters. However, the likely magnitude of any effect will
vary depending on the landscape-element type and spatial scale. We used a chain of hydrological, empirical,
and process-based biogeochemical models coupled to an ensemble of downscaled Regional Climate Model
experiments to develop scenario storylines for local- and landscape-scale effects of forest harvesting and
climate change on surface water DOC concentrations and fluxes. Local-scale runoff, soil temperature, and DOC
dynamics were simulated for a range of forest and wetland landscape-element types and at the larger
landscape scale. The results indicated that climate change will likely lead to greater winter flows and earlier,
smaller spring peaks. Both forest harvesting and climate change scenarios resulted in large increases in summer
and autumn runoff and higher DOC fluxes. Forest harvesting effects were clearly apparent at local scales. While
at the landscape scale, approximately 1 mg L1 (or 10%) of the DOC in surface waters can be attributed to
clear-cuts, both climate change and intensified forestry can each increase DOC concentrations by another
1 mg L1 in the future, which is less than that seen in many waterbodies recovering from acidification. These
effects of forestry and climate change on surface water DOC concentrations are additive at a landscape scale
but not at the local scale, where a range of landscape-element specific responses were observed.
1. Introduction
The boreal forest, which covers much of the northern hemisphere landmass, plays a pivotal role in the global
carbon cycle [Apps et al., 1993; Magnani et al., 2007]. Not only are the wood and other biomass products from
boreal forests economically important but they are widely used for climate change mitigation through fossil
fuel substitution. Boreal soils store up to 500 Gt of organic carbon (C) [Pan et al., 2011], large amounts of which
are exported as dissolved organic carbon (DOC) to surface waters [Battin et al., 2009; Tranvik et al., 2009],
forming an important component of the global carbon budget [Aufdenkampe et al., 2011; Cole et al., 2007;
Öquist et al., 2014]. DOC is a substrate for microbial respiration and contributes to CO2 evasion to the atmosphere [Koehler et al., 2014; Wallin et al., 2013] or can be a pathway for C accumulation in lake or river sediments [Cole et al., 2007]. However, with the changing climate and increasing pressure of forestry, there are
additional stresses on boreal stream DOC, which may have severe consequences.
©2015. The Authors.
This is an open access article under the
terms of the Creative Commons
Attribution-NonCommercial-NoDerivs
License, which permits use and distribution in any medium, provided the
original work is properly cited, the use is
non-commercial and no modifications
or adaptations are made.
ONI ET AL.
Much of the boreal forest is in a delicate balance between acting as a C source or sink [Laudon et al., 2013b;
Tetzlaff et al., 2013]. Climate change will likely affect net primary production and increase litter production by
altering the length of the growing season and climatic conditions [Jansson et al., 2008; Kleja et al., 2008]. It also
affects soil organic matter mineralization rates as soils become warmer and wetter [Davidson and Janssens,
2006]. Modeling studies suggest both large increases [Larsen et al., 2011] and small changes [Holmberg
et al., 2014] in boreal lake DOC concentrations ([DOC]) under a changing climate. However, it has also been
shown that the magnitude of simulated change in [DOC] under a changing climate depends on the modeling
approach [Oni et al., 2014].
Similarly, forest harvesting can have significant impacts on water quality [Kreutzweiser et al., 2008; Laudon
et al., 2011b]. The mechanisms causing increases in [DOC] are related primarily to reductions in interception
FUTURE BOREAL DOC
2402
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 1. Conceptual diagram showing connections in the chain of models used in assessing present and future local- and
landscape-scale climate and forestry effects on stream DOC concentration and fluxes. Measured climate data and downscaled outputs from 15 Regional Climate Model (RCM) experiments feed into the hydrological model (PERSiST). Outputs
from the PERSiST model are used as hydrological forcings to the local-scale biogeochemical models (INCA-C and RIM) for
the different landscape end-members (FT = Forest on till, FS = Forest on sediment, W = Wetland, DS = downstream). The
output of the local-scale biogeochemical models feed into the large-scale landscape mixing model (LMM), where they are
combined with a DOC signal from deep groundwater flow paths (as a function of discharge).
and evapotranspiration. Forest harvesting can lead to increased winter snow accumulation and spring melts
[Schelker et al., 2013b] and higher runoff compared to intact forests [Futter et al., 2010]. Warmer soils may also
be observed following clear-cuts as the shading effect of tree cover is absent [Schelker et al., 2013a]. There are
well-documented studies showing increases in DOC concentrations and fluxes following clear-cutting in the
boreal forest [Palviainen et al., 2014; Schelker et al., 2012]. Studies of forestry impacts on surface water DOC are
typically based on paired catchments [Palviainen et al., 2014; Schelker et al., 2012] or long-term monitoring
[Lepistö et al., 2014]. There have been some attempts to upscale local results to a regional level. For example,
Schelker et al. [2014] showed that when forest harvesting affects more than a critical threshold of catchment
area, detectable increases in downstream [DOC] will occur. However, to date, there have been no studies on
the potential combined effects of future forest management and climate change on surface water [DOC] and
its implications for the overall C balance in boreal forests.
Currently, less than 1% of productive Swedish forests are harvested annually [Christiansen, 2014], but this may
increase in the future, both due to greater demand and faster forest growth. Industrialized forestry is
expected to grow in many countries due to increasing demands for forest products for bioenergy to meet
green energy targets including the EU Renewable Energy Directive [Egnell et al., 2011]. In recent years,
Sweden has pursued more intensive forest harvesting as a way of fulfilling the requirements of the EU renewable energy directive [Egnell et al., 2011]. A major concern is how intensified forest activities will impact water
quality and particularly DOC.
The boreal forest can be conceptualized as a mosaic of landscape-element types defined by vegetation cover,
drainage, and Quaternary geology. These landscape element types display markedly different biogeochemical and hydrological signatures. In the boreal regions of Sweden, landscape-element types representing
forest on till, forest on sediment, and wetlands define the mosaic. Forests on till are typically found above
the highest coastline, while forests on lacustrine or marine sediments are found below the highest coastline.
ONI ET AL.
FUTURE BOREAL DOC
2403
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Wetlands, usually mires, are found in areas of impeded drainage. Each of these landscape-element types has
a distinct biogeochemical signature [Grabs et al., 2012; Ledesma et al., 2013]. At a larger landscape scale, much
of the unique signature can be lost [Ågren et al., 2014] and local-scale effects of forest harvesting on water
quality diluted [Schelker et al., 2014]. While climate change and forestry will affect all landscape element
types, much remains unknown about the individual local-scale responses, or their combined effect at a
downstream landscape scale.
Spatial scale is extremely important when assessing the impacts of anthropogenic activity on water quality.
Forestry impacts on water quality are most apparent at small spatial scales typical of individual stands or headwaters [Cui et al., 2012; Futter et al., 2010; Palviainen et al., 2014; Sanders and McBroom, 2013] and can be masked
or impossible to detect at the river mouths of large catchments [Lepistö et al., 2014]. In what follows, we refer to
two spatial scales: the local and the landscape. The former has a spatial extent of approximately 10 ha and
includes individual forest stands and headwater catchments. The latter has a spatial extent of tens of square
kilometers and larger and is representative of medium size rivers and mesoscale catchments. Landscape-scale
water quality is not simply the sum of local-scale inputs but can also include the effects of in-stream processing
or inputs from deep groundwaters [Hagedorn et al., 2000; Tiwari et al., 2014].
Both empirical [Hejzlar et al., 2003; Larsen et al., 2011] and process-based [Futter et al., 2009; Holmberg et al., 2014]
models have been used to project future [DOC]. There are advantages and disadvantages to each approach;
empirical models generally produce better goodness of fit statistics [Futter and de Wit, 2008] but process-based
modes are considered to be more appropriate for simulating conditions outside the range of current observations [Leavesley, 1994; Adams et al., 2012; Crossman et al., 2014]. Models can be used to simulate hydrological
and biogeochemical processes at different scales, ranging from hillslopes to landscapes (Figure 1) [Futter
et al., 2007; Seibert et al., 2009; Winterdahl et al., 2011]. By using an ensemble of biogeochemical models and
climate scenarios, it is possible to ensure that the range of outcomes reflects different spatial scales as well as
the different biogeochemical processes operating within the catchment. This range of possible outcomes contributes to the development of storylines about possible futures. While storylines can be based on quantitative
outputs, they are more useful as internally consistent plausible, qualitative descriptions of what may happen, not
necessarily what will happen [Börjeson et al., 2006; Milestad et al., 2014; Rounsevell and Metzger, 2010].
Here we present simulations based on an ensemble of hydro-biogeochemical models and downscaled climate models outputs (Figure 1) to develop storylines about possible forest harvesting and climate change
effects on local- and landscape-scale seasonal and interannual patterns in DOC concentrations and fluxes
in boreal streams. We used the steady state empirical Riparian Flow-Concentration Integration Model (RIM)
[Seibert et al., 2009] and the dynamic, process-based Integrated Catchment Model for Carbon (INCA-C)
[Futter et al., 2007] to reflect different scales of processes occurring in the boreal landscape (Figure 1).
These two modeling approaches have been used previously to simulate DOC dynamics in headwater boreal
catchments representative of the landscape element types simulated here [Futter et al., 2009, 2011; Futter and
de Wit, 2008; Oni et al., 2014; Winterdahl et al., 2011, 2014]. However, other models, such as those developed
by Naden et al. [2010] or Temnerud et al. [2014], could also have been used. A landscape mixing model is used
to integrate signals from headwater to downstream. Local-scale inputs do not necessarily completely define
landscape-scale water quality signatures mainly because, as transit times increase, the influence of groundwater discharge becomes more prominent [Ågren et al., 2014] and there is more opportunity for aquatic
processing of carbon [Müller et al., 2013]. Thus, the effect of groundwater at the landscape scale was also
included (Figure 1).
We addressed the following questions related to the development of plausible, internally consistent storylines about future water quality in the boreal forest: (i) Do empirical and process-based models of surface
water DOC give similar projections of future surface water quality under climate change and more intensive
forest harvesting? (ii) Are the projected effects of clear-cutting on surface water quality the same for different
landscape-element types? (iii) What are the interactions between climate change and forestry effects on
surface water DOC concentrations and fluxes at local and landscape-scales?
2. Materials and Methods
The climate and forestry scenarios presented here are summarized in Table 1. All results were based on measurements made at the Krycklan catchment and Balsjö experimental clear-cut (Figure 2). All present-day
ONI ET AL.
FUTURE BOREAL DOC
2404
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Table 1. Summary of the Scenarios, Their Data Sources, and Properties
Period
Climate Forcing
Abbreviation
Name
Scale
Brief Description of Key Features
Present
Observed (Krycklan)
meteorology.
KC
Krycklan calibrated
Local, Landscape
PDF
Present-Day Forestry
Local
BAU
Business as Usual
Landscape
IFH
Intensive forest harvest
Landscape
CC
Climate Change
Local, Landscape
FF
Future Forestry
Local
CCF
Climate Change and “Business
as Usual” Forest Harvest
Landscape
CCI
Climate change and intensive
forest harvesting
Landscape
Calibrated to streamflow (C7) and [DOC] (C2, C4, and C20)
from Krycklan study catchments
KC parameterization adjusted to reproduce runoff and soil
temperature effects of clear-cutting observed at Balsjö
KC and PDF were provided as inputs to the Landscape
Mixing Model (LMM) using current climate with an
assumption of 0.86% of the area felled annually
KC and PDF were provided as inputs to the LMM using
current climate and an assumption of 1.72% of
the area felled annually
Based on KC hydrological and biogeochemical model
parameterization with future climate from RCMs
Based on PDF hydrological and biogeochemical model
parameterization with future climate from RCMs
CC and FF were provided as inputs to the LMM using future
climate with an assumption of 0.86% of the area
felled annually
CC and FF were provided as inputs to the LMM using
future climate with an assumption of 1.72% of the
area felled annually
Future
Ensemble of 15
downscaled RCM
experiments
simulations were forced by observed meteorological data from the Krycklan catchment study. Future projections were based on model parameter sets obtained during present-day simulations and forced by outputs
from an ensemble of downscaled outputs from 15 RCM experiments.
2.1. Krycklan Catchment Study
The modeling and analyses presented here were performed using data from the Krycklan catchment (68 km2)
(Figure 2), a well-instrumented and monitored site representative of the boreal landscape [Laudon et al.,
2013a]. The catchment is located in northern Sweden (64°16′N, 19°46′E) and has been set aside for
process-based, experimental, and hypothesis-driven research needed to improve scientific knowledge,
policy, and management of boreal ecosystems. Across the whole Krycklan catchment, Quaternary till deposits
with low permeability cover the upper part (51%) of the catchment while thick freely draining marine-derived
glacio-fluvial sediments cover the lower parts (49%) with deep characteristic incised channeled streams
[Ågren et al., 2007]. Forest soils are mostly well-developed iron podzols with organic carbon-rich riparian
zones. Catchment elevation ranges from 126 to 372 m above sea level, and the highest postglacial coastline
crosses the catchment at around 255–260 m [Ågren et al., 2007]. In areas with low slopes, water tends to
accumulate to form minerogenic mires, characterized by acidophilic Sphagnum species. Forests (87%),
wetlands (peaty mires; 9%), arable land (3%), and lakes (1%) cover the catchment, and low-intensity forestry
is conducted throughout. The dominant tree species in the catchment are Norway spruce (Picea abies, 27%)
and Scot pine (Pinus sylvestris, 63%). The remaining 10% are deciduous trees, mainly birches (Betula spp.) that
are found on and around riparian zones.
The catchment has a long-term (1981–2012) mean annual temperature of 1.8°C (peaking in July), mean
annual precipitation of 631 mm yr1 (±102 mm yr1), and mean annual runoff of 324 mm yr1 (± 96 mm yr1).
Up to 50% of annual precipitation falls as snow in the winter, and the snowpack can last up to 170 days
(November–May) [Laudon et al., 2013a; Oni et al., 2013]. Subsurface pathways dominate the delivery of runoff
to streams and rivers. This influences the overall hydrology of Krycklan between limited groundwater influences at the headwaters to a dominating groundwater contribution at the outlet of the 68 km2 catchment.
Long-term hydrologic simulations were based on daily precipitation and air temperature monitored at a
weather station at the center of the Krycklan catchment and specific discharge data from site C7 (Figure 2)
[Laudon et al., 2013a]. Site C7 has been used because it has the longest consistent record of streamflow of
any of the catchments in Krycklan.
At the Krycklan catchment outlet, water chemistry can be conceptualized as a time-dependent mixture of
local-scale upstream surface waters and landscape-scale groundwater contributions, in which the dominant
ONI ET AL.
FUTURE BOREAL DOC
2405
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 2. Map showing the location of the study catchments within (a) Sweden, (b) the Krycklan catchment, and (c) the Balsjö clear-cut experiment. The Krycklan
map (Figure 2b) shows the location of the forest on till (C2), wetland (C4), and forest on sediment (C20) headwater catchments used in the local-scale simulations
presented here. The landscape-scale simulations were made for the catchment outlet (C16). Balsjö data used here were from the clear-cut (CC4, BF5) and reference
(NR-7) subcatchments.
landscape types (forest on till, forest on sediment, and wetlands) exhibit unique biogeochemical signatures.
While forests on till and wetlands are DOC rich, forests on sediments are characterized by lower [DOC]
[Ågren et al., 2014]. Three well-monitored headwater catchments representative of forests on till (C2),
wetlands (C4), and forests on sediments (C20) were used to characterize local-scale DOC dynamics (Figure 2)
[Laudon et al., 2011a, 2013a; Oni et al., 2013].
2.2. Balsjö Paired Catchment Clear-Cut Experiment
Simulations of forest harvesting effects on hydrology and soil conditions influencing DOC production and
export were based on observations made between 2005 and 2011 for hydrology and 2009 for soil temperature from the Balsjö paired catchment study (64°2′N, 18°56′E; 22.9 km2). Balsjö is close to Krycklan, about
70 km west of the Baltic coast in northern Sweden (Figure 2). Studies at Balsjö have documented the effects
of clear-cutting on soil temperature [Schelker et al., 2013a], hydrologic flow paths [Schelker et al., 2013b;
Sørensen et al., 2009b], and stream chemistry [Laudon et al., 2009; Löfgren et al., 2009; Schelker et al., 2012;
Schelker et al., 2014; Sørensen et al., 2009a]. The study includes an unharvested control catchment and two
catchments with differing degrees of clear-cutting. The simulations used here were based on observations
made at the unharvested NR7 reference catchment and the clear-cut catchments (Figure 2).
ONI ET AL.
FUTURE BOREAL DOC
2406
Journal of Geophysical Research: Biogeosciences
Future projections
Present day
140
a
120
100
Observed
PERSiST
80
60
40
Runoff (mm)
Present day forestry
Ensemble median
40
0
140
Balsjö
120
Reference
100
Ensemble max
60
20
b
Present day
Ensemble min
80
0
120
c
100
20
140
Runoff (mm)
140
Krycklan
80
60
40
Runoff (mm)
Runoff (mm)
120
10.1002/2015JG003190
100
d
Present day forestry
Ensemble min
Ensemble median
Ensemble max
80
60
40
20
20
0
0
Figure 3. Monthly runoff summaries including present-day (a and b) and future (c and d) values. (a) Monthly average
observed runoff, PERSiST model calibration for the KC scenario and the PDF scenario, which simulates clear-cut effects
at Krycklan site C7 between 2005 and 2012. (b) Monthly average observed runoff at Balsjö clear-cut and control sites.
The difference in runoff between control and clear-cut sites was used to identify parameter values for the PDF scenario.
(c) A summary of results from the CC scenario with possible climate effects on runoff including the monthly minimum,
maximum, and overall average runoff obtained from the ensemble of 15 RCMs. Average monthly present-day runoff from
the KC scenario is included to facilitate comparison. (d) Results from the FF scenario which simulates possible local-scale
combined effects of future climate change and forest harvesting on runoff dynamics. Average values from the PDF scenario
are presented to facilitate comparison. Values in Figure 3d were obtained using the 15 RCM realizations and the PDF
parameter set used for simulating present-day forestry.
Balsjö shares some similar attributes to Krycklan. Soils in Balsjö are dominated by well-drained till in recharge
areas and mire-dominated discharge areas cover 15% of the catchment[Schelker et al., 2013b]. Tree cover is
also similar. The Balsjö catchment has annual temperature and precipitation of 0.6°C and 613 mm yr1,
respectively, and a stable snow cover between November and May. Spring melt runoff occurred over an
extended time period (April–May) at the Balsjö reference site (no clear-cut) but peaked later in May in
Krycklan (Figure 3). The spring peak at Balsjö was higher than that in Krycklan (Figure 3), and both summer
and autumn runoff were higher at the Balsjö reference site compared to Krycklan. No difference in spring
melt runoff between the nonharvested reference site and the clear-cut sites in Balsjö (Figure 3) was observed.
Forest clear-cuts at Balsjö have intensified the rate of snow accumulation and melt [Schelker et al., 2013b],
increased summer runoff [Sørensen et al., 2009a], and increased summer soil temperatures [Schelker
et al., 2013a].
2.3. Climate Projections
Current (1981–2010) and future (2061–2090) simulations of precipitation and temperature were obtained
from the ENSEMBLES project data archive [Van der Linden and Mitchell, 2009]. An ensemble of 15 regional
climate models (RCMs) driven by different global climate models (GCMs) were used in the simulations
presented here. The ensemble is further described in Oni et al. [2014]. Hereafter, individual members of
the ensemble are referred to as RCM experiments. Each RCM experiment had a resolution of 25 km × 25 km
and was run under the Special Report on Emissions Scenarios A1B. The A1B scenario is a medium-high
emission-scenario based on a future assumption of strong economic growth and balance use of all sources
of energy with an associated increase in the rate of greenhouse gas emissions [Intergovernmental Panel on
Climate Change, 2007]. It is broadly comparable to Representative Concentration Pathway (RCP) 6.0/Shared
Socio-Economic Pathway 2 (SSP2) in the new Representative Concentration Pathways (RCP) and Shared
Socio-Economic Pathways (SSP) scenarios [van Vuuren and Carter, 2014]. Precipitation and temperature
values were obtained by averaging the values of the RCM grid cell with center coordinates closest to the
ONI ET AL.
FUTURE BOREAL DOC
2407
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
center of the catchment and of its eight neighboring grid cells. The spatial mismatch between RCM grid cell
size (~625 km2) and the size of the relatively small study catchments can, in combination with the often
debated issue of systematic errors (so-called biases) in RCMs, heavily influence subsequent hydrological modeling results [Ehret et al., 2012; Muerth et al., 2013; Teutschbein and Seibert, 2012]. Postprocessing of the RCM
data (i.e., bias correction) is therefore required. To bias-correct the RCM-simulated precipitation and temperature series, a distribution mapping procedure was used [Déqué et al., 2007; Ines and Hansen, 2006], which has
been used previously for climate change impact studies in Sweden [Teutschbein and Seibert, 2012, 2013]. The
distribution mapping adjusts a theoretical cumulative distribution function (CDF) of RCM-simulated climate
values so that it matches the observed CDF of observed temperature or precipitation (measured at a weather
station within the Krycklan catchment). The Gamma distribution was used to simulate the CDF of precipitation events [Piani et al., 2010], and a Gaussian distribution was used to bias correct temperature time series
[Teutschbein and Seibert, 2012]. A precipitation threshold was introduced for each RCM so as to avoid distorting the precipitation distribution during dry conditions. This was needed as RCM outputs often include large
numbers of days with light rains instead of dry conditions. Biases are assumed to be stationary, and the same
correction algorithm (i.e., CDF parameters) was applied to both current and future climate conditions.
2.4. Forest Harvest Scenarios
Two forest harvest scenarios were used here: business as usual (BAU) representing current conditions and
intensive forest harvesting (IFH) representing a plausible increase in harvest rate. According to the Swedish
Statistical Yearbook of Forestry [Christiansen, 2014] 0.86% of productive forest land is clear-cut annually.
Thus, the landscape-scale BAU clear-cut rate was set to 0.86% yr1. It should be noted that this value is
slightly higher than the actual clear-cut rate at Krycklan, where approximately 25% of the catchment is not
harvested. The IFH scenario assumed that forest harvesting will intensify due to shorter rotation periods
and increasing demand for forest products for bioenergy. A doubling of clear-cut rates to 1.72% yr1 was
assumed to represent the maximum feasible increase in harvest intensity rates for the IFH scenario.
2.5. Biogeochemical Modeling
The biogeochemical modeling was performed as follows. Present-day (2005–2012) conditions were simulated
in the following four steps: (i) A rainfall-runoff model: the Precipitation Evapotranspiration and Runoff Simulator
for Solute Transport (PERSiST) [Futter et al., 2014], was calibrated to long-term present-day (1981–2012) conditions. (ii) Outputs from the rainfall-runoff model (including soil moisture status) were then used as inputs to a
series of local-scale landscape-element specific simulations (2005–2012) using INCA-C, a process-based model
of DOC production and transport [Futter et al., 2007, 2009]. (iii) Simulated flows from (i) and soil temperatures
from (ii) were used as inputs to local-scale landscape-element specific simulations (2005–2012) using RIM, an
empirical model for riparian and stream DOC dynamics [Seibert et al., 2009]. Finally, (iv) landscape-scale DOC
fluxes were estimated using local-scale, landscape-element-specific results from both models (ii) and (iii) as
inputs to a landscape mixing model [Cooper et al., 2000; Tiwari et al., 2014] (Figure 1).
2.5.1. Hydrologic Modeling
The Precipitation, Evapotranspiration and Runoff Simulator for Solute Transport (PERSiST) is a catchment-scale,
daily time scale rainfall-runoff model requiring inputs of precipitation and air temperature [Futter et al., 2014].
Precipitation is partitioned into rain or snow depending on whether air temperature is above or below a
user-specified threshold value. Canopy interception for snowfall and rainfall is simulated, and the model uses
simple degree-day evapotranspiration and temperature-index snowmelt routines [Futter et al., 2014]. PERSiST
simulates the hydrologically effective rainfall (HER) and soil moisture deficit (SMD) time series needed for
INCA-C (Figure 1). HER represents the fraction of precipitation that contributes to the runoff generation. SMD
is the difference between maximum soil water holding capacity and an average soil water depth.
PERSiST was set up using four vertically stacked buckets through which precipitation was routed. The uppermost bucket was used to manage canopy interception as well as rain and snowmelt inputs, while the lower
three buckets contributed to streamflow. These three buckets were representative of direct runoff having
limited contact with the soil, soilwater, and shallow groundwater, respectively. All evapotranspiration was
simulated in the soilwater bucket. The catchment was simulated as a single reach with one land cover type.
The model was calibrated using long-term (1981–2012) observed streamflow data from site C7 and meteorological data from the Krycklan weather station to simulate present-day (KC scenario) runoff conditions.
ONI ET AL.
FUTURE BOREAL DOC
2408
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Manual calibration was first performed by fine tuning the parameters to minimize difference between the
simulated and observed runoff. Model performance was assessed using Nash and Sutcliffe efficiencies
(NSE) [Nash and Sutcliffe, 1970] from untransformed and log-transformed series. A Monte Carlo analysis
was performed, in which parameter values from the manual calibration were allowed to vary by ± 20%,
and the best performing parameter set was retained for scenario projections, in which the ensemble of
bias corrected future climate series obtained from the 15 RCM experiments was used to project future
runoff conditions.
2.5.2. Process-Based Integrated Catchment Model for Carbon (INCA-C)
INCA-C is a process-based model that simulates both terrestrial and aquatic carbon production and transport.
The model requires daily time series of air temperature, precipitation, SMD, and HER (Figure 1). Terrestrial process rates are dependent on soil temperature and moisture. Soil temperature is simulated independently of
DOC using routines developed for boreal conditions [Rankinen et al., 2004]. Nonmonotonic relationships
between soil moisture and carbon processing can be simulated, which allow maximum rates at intermediate
soil wetness and slower processing when soils are excessively wet or dry.
In a nutshell, change in terrestrial DOC is simulated in INCA-C as the net rate of sorption to and desorption
from soil organic carbon (SOC) minus the mass of DOC running off to surface waters. Both sorption and desorption are dependent on soil temperature and moisture (f(TSoil, M)) , and kS and kD represent rates of sorption and desorption, respectively. The change in mass of SOC includes additional terms for mineralization (kM)
and time-varying production from litter inputs and breakdown of plant material (p). Equations (1) and (2)
present simplified representations of the changes in mass of SOC and DOC. Full descriptions of process
equations are described in earlier literature [Futter et al., 2007, 2009].
dDOC
¼ f ðT Soil ; MÞðk D SOC k S DOCÞ Q½DOC
dt
(1)
dSOC
¼ f ðT Soil ; MÞðk s DOC k D SOC k ´M SOCÞ þ p
dt
(2)
INCA-C was manually calibrated against 2005–2012 stream water [DOC] in the C2, C4, and C20 headwater
catchments. The manual calibration was refined using the method of Futter et al. [2014] to generate a
present-day INCA-C Krycklan Calibrated (KC) scenario. Model performance was assessed using NSE statistics
for flow and [DOC]. For the Climate Change (CC scenarios), local-scale impacts of future climate on [DOC]
were simulated by running the models with bias corrected temperature and precipitation time series
(Figure 4).
2.5.3. Empirical Riparian Flow-Concentration Integration Model (RIM)
The RIM (equation (3)) assumes that stream [DOC] can be simulated as a nonlinear temperature-dependent
function of runoff (equation (3)).
½DOC ¼ ðC base þ k 1 Tsoil Þ XðQ Qo Þðf base þ k2 T soil Þ
(3)
where Cbase (mg L1) is baseline stream [DOC], k1 (mg L1°C1) is the stream water concentration
temperature multiplier, and Tsoil (°C) is the soil temperature. Q is the measured runoff, Qo (mm d1) is a
calibrated runoff offset. Paramater fbase is the baseline flow: DOC exponent and k2 (°C1) is an empirical
temperature modifier for the flow: concentration relationship. RIM was calibrated against 2005–2012
observed stream [DOC] in the C2, C4, and C20 catchments and driven by PERSiST-simulated flow and
INCA-C simulated soil temperature using the solver routine in MS-Excel to identify parameter values which
minimized the sum of squared differences between simulated and observed stream [DOC] to generate a
present-day Krycklan Calibrated (KC) scenario from RIM. The Climate Change (CC) scenario, in which
impacts of future climate on [DOC] were evaluated, was performed by running the model with the above
parameters and forcings from the 15 downscaled RCM experiments.
2.5.4. Simulation of Forestry Effects
The “Present Day Forestry” (PDF) (Table 1) results were generated in the following manner. Evidence from the
Balsjö clear-cutting experiment showed increases in runoff and warmer soil temperatures following final
harvesting [Schelker et al., 2012]. The effects of forest harvest on runoff were simulated by adjusting parameters
from the PERSiST Krycklan calibration related to snowfall, snowmelt temperature, canopy interception, and
evapotranspiration so as to match the differences in runoff between the Balsjö reference and clear-cut sites
ONI ET AL.
FUTURE BOREAL DOC
2409
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
prorated to 100% harvested. While
PERSiST generates daily estimates of
runoff, adjustments to simulate final
harvesting effects were made on a
monthly basis so as to preserve the
seasonal pattern in changes in runoff
resulting from final harvesting. The
increase in local snowfall associated
with clear-felling was simulated by
increasing the snow multiplier in
PERSiST. This parameter scales measured precipitation to match the
amount of snow accumulating in the
area of interest. As a consequence of
reducing canopy interception and
decreasing evapotranspiration, wetter
soils were simulated.
In the simulations presented here, soil
temperature is simulated as a function of air temperature [Rankinen
et al., 2004; Futter et al., 2007]. So as
to reproduce the increase in soil
temperature associated with forest
harvesting, soil thermal conductivity
parameters in INCA-C were adjusted
so that simulated soil temperatures
matched the observed soil temperaFigure 4. Summary plots of monthly precipitation (a) and average tempera- ture increase at the Balsjö clear-cut.
Future local-scale forestry scenarios
ture (b) during the observation period (KC scenario; red lines; 1981–2012)
and for the climate scenario (CC; green lines; 2061–2090). Red lines show
(FF) were simulated using downscaled
present conditions. Thin red lines show the minimum and maximum monthly climate from the 15 RCM experiprecipitation (Figure 4a) and the minimum and maximum average monthly
ments to force present-day forestry
temperature (Figure 4b). Thick red lines show average monthly precipitation
and temperature. Green lines show possible future conditions based on the parameterizations for INCA-C, RIM,
and PERSiST.
15 climate scenarios. Thin green lines represent minimum and maximum
monthly precipitation (Figure 4a) and the minimum and maximum average
2.5.5. Combining Climate Change
monthly temperature (Figure 4b) from the 15 scenarios. Dashed green lines and Effects of Forestry at the
represent scenario averages values. The thick dashed line represents the
Local Scale
wettest or warmest average value from the 15 scenarios while the thin
Climate change (CC scenario) and fordashed line represents the driest or coldest average value.
estry (PDF) effects on [DOC] (Figure 5)
and fluxes (Figure 6) were first modeled separately. The Future Forestry (FF scenario) was run to assess whether or not the combined impacts
of expected climate change and forest harvesting increase or reduce stream water [DOC]. By superimposing
the temperature and hydrological signal of climate change and forestry effects and using these as input variables in the biogeochemical models (RIM and INCA-C), it was possible to examine the combined effects of
climate change and forestry on [DOC] (Figure 7) and fluxes (Figure 8) in the future for the three landscapeelement types. At the local scale, these are referred to as the “FF scenario” (Table 1).
2.5.6. Landscape-Scale Mixing Model
As noted earlier, landscape scale water quality is not merely the sum of local scale inputs. A number of
scenarios were evaluated to explore the differences between local and landscape scale patterns (Table 1).
Local scale outputs from the KC and PDF scenarios were used as inputs representing intact and clear-cut
forests. Three scenarios were evaluated using present-day climate and differing degrees of clear-cut intensity.
In the KC scenario, 0% clear-cutting was assumed while the BAU and IFH scenarios assumed 0.86% and 1.72%
respectively of the landscape was clear-cut annually.
ONI ET AL.
FUTURE BOREAL DOC
2410
Journal of Geophysical Research: Biogeosciences
45
Present day
35
Present day forestry
30
25
35
Forest on till
40
10.1002/2015JG003190
Forest on till
30
Ensemble median
25
Ensemble min
20
Ensemble max
20
15
15
10
10
5
5
0
0
50
45
40
35
30
25
20
15
10
5
0
30
70
Wetland
Wetland
60
50
40
30
20
10
0
20
18
16
14
12
10
8
6
4
2
Forest on sediment
25
20
15
10
5
Forest on sediment
Present day
Present day forestry
Ensemble median
Ensemble min
Ensemble max
0
Figure 5. Local-scale simulated present-day (KC) and projected future [DOC] from (left column) INCA-C and (right column)
RIM for representative boreal landscape element types showing the effects of forestry (PDF) and the possible effects of
climate change (CC). Each figure shows average monthly [DOC] from the KC scenario (filled blue shapes), the PDF scenario
(red line), and a summary of the CC scenario results. The black dashed line shows the lowest average [DOC] from the
ensemble of 15 RCM experiments used to generate CC results, while the blue line shows the maximum of the scenario
averages. The green line shows the overall scenario median value. Note the differences in scale between INCA-C and RIM
simulations and between landscape element types.
All landscape scale simulations were based on Tiwari et al. [2014] who showed that a landscape-scale
mixing model (LMM) using headwater catchments dominated by till forest, sediment forest, and wetlands landscape-element types as headwater end members (C2, C20, and C4, respectively) can be used
to predict downstream concentrations by mixing local DOC signals in proportion to their areal coverage
(equation (4)).
½DOCs ¼
X 3
x
mx ½DOCx
(4)
where [DOC]s is the landscape-scale DOC concentration in downstream surface waters, mixed from local
headwater concentration signals from x identified landscape-element types (x; 1 = forest on till,
2 = wetland, 3 = forest on sediment) and m is the proportion of the areal coverage of respective landscapeelement. Tiwari et al. [2014] also showed that by incorporating the deep groundwater flow paths into the
model the RMSE improved from 5.2 mg L1 to 2.2 mg L1. The magnitude of the base flow component
varied as a function of discharge, f(Q), from 75% groundwater during base flow to 0% during high flow
(equation (6)). In short, concentration of DOC at the outlet is a mixture of the DOC concentration signal
from surface (local) and groundwater (landscape) inputs (equation (5))
½DOCoutlet ¼ pg ½DOCg þ 1-pg ½DOCs
(5)
where pg is the percent of runoff from groundwater and [DOC]g and [DOC]s are the concentrations of DOC in
groundwater and surface waters, respectively. We have been using a value of 1.25 mg L1 for [DOC]g based
ONI ET AL.
FUTURE BOREAL DOC
2411
Journal of Geophysical Research: Biogeosciences
Forest on till
Forest on till
2
2
1
1
0
0
5
5
Wetland
4
4
3
3
2
2
1
1
0
0
2
10.1002/2015JG003190
2
Forest on sediment
Wetland
Forest on sediment
Present day
Present day forestry
Ensemble min
Ensemble median
Ensemble max
1
1
0
2
1
Figure 6. Local-scale simulated DOC fluxes in g DOC m month from (left column) INCA-C and (right column) RIM for
representative boreal landscape element types showing present-day conditions (KC scenario), the effects of present-day
forestry (PDF) and the range of possible future climate change effects from the CC scenarios. Each figure shows average
monthly DOC fluxes from the KC scenario (filled blue shapes), the PDF scenario (red line), and a summary of the CC scenario
results. The black dashed line shows the lowest average DOC flux from the ensemble of 15 RCM scenarios used to generate
CC results, while the blue line shows the maximum of the scenario averages. The green line shows the overall scenario
median flux. Note the differences in scale between landscape element types.
on measurements from deep groundwater in a nearby well [Tiwari et al., 2014]. The flow-dependent
groundwater fraction in runoff was simulated using a relationship from Tiwari et al. [2014] (equation (6))
pg ¼ 0:139 lnðr outlet Þ þ 0:3362
(6)
1
where routlet is runoff at the outlet, in mm d , and runoff is measured at C7. In the forestry scenarios
(see below), runoff was adjusted to account for forestry impacts.
Landscape-scale effects of forest harvesting on DOC concentrations and fluxes (BAU and IFH scenarios; Table 1)
were modeled in the following manner. The percent of the landscape affected by forestry (f) was determined
based on the fraction of the landscape experiencing forestry impacts. This, in turn, was determined by the
fraction harvested every year (h, % yr1), the duration of the effect (d, years), and the fraction of the local
landscape-element subject to clear felling effects (g = 1 for forest on till and forest on sediment, 0.67 for
wetlands (67% of the mires in the catchment are covered by forests)), and the fractional coverage of the
different local-scale landscape-elements across the catchment (m, %)
X3
f ¼ hd x ðmx gx Þ
(7)
The value of h was set to 0.86% yr1 in the BAU scenario and 1.72% yr1 in the IFH scenario
The effects of clear-cutting on local hydrology, soil temperature, and C biogeochemistry decline as forest
cover re-establishes [Löfgren et al., 2009; Schelker et al., 2012]. It is assumed that the full clear-cut effect on
stream water biogeochemistry is seen the first 10 years after felling and that canopy closure is reached after
30 years. The clear-cut signal is assumed to show a linear decline from total effects 10 years after felling to no
ONI ET AL.
FUTURE BOREAL DOC
2412
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 7. Local-scale simulated present-day and projected future [DOC] from (left column) INCA-C and (right column) RIM
for representative boreal landscape element types showing the effects of present-day forestry (PDF) and future forestry
(FF). Each figure shows average monthly [DOC] from the KC scenario (filled blue shapes), the PDF scenario (red line), and a
summary of the FF scenario results where the parameterization for present-day forestry was forced with downscaled RCM
outputs. The black dashed line shows the lowest average [DOC] from the ensemble of 15 RCM experiments used to generate
FF results, while the blue line shows the maximum of the scenario averages. The green line shows the overall scenario median
value. Note the differences in scale between INCA-C and RIM simulations and between landscape element types.
effect after 30 years. At a landscape scale, d is based on the time of 100% effect and 0.5 times the number of
years for the effect to show a linear decline to zero. Thus, d = 10 + 0.5 × 20 = 20 years.
Thus, routlet can be calculated as runoff from forestry affected areas plus runoff from intact forests
r outlet ¼ f r forestry þ ð1 f Þr intact
(8)
where rforestry and rintact are the simulated runoff values, in mm d1, for clear-felled and intact forests,
respectively. Runoff values are obtained from PERSiST simulations.
The surface water [DOC] can be derived from the forestry and intact forest scenario [DOC] for each landscapeelement. Concentrations must be prorated by runoff. Thus,
½DOCs ¼ r intact ½DOCintact þ r forestry ½DOCforestry = r intact þ r forestry
(9)
Daily [DOC]intact and [DOC]forestry values for each landscape-element type are available from the INCA-C and RIM
simulations. The surface water DOC signal at the outlet includes DOC from unharvested areas (equation (10)),
plus DOC the fraction of harvested landscape-elements for which forestry effects do not apply (equation (11))
plus the DOC signal from clear-cut areas (equation (12)).
X 3
ð1 – f Þ r intact x mx ½DOCintact;x
(10)
X3
mx ½DOCintact;x ð1 gx Þ
(11)
f r intact x
X 3
f r forestry x mx ½DOCforestry;x gx
(12)
ONI ET AL.
FUTURE BOREAL DOC
2413
Journal of Geophysical Research: Biogeosciences
2
10.1002/2015JG003190
1
Figure 8. Local-scale simulated DOC fluxes in g DOC m month from (left column) INCA-C and (right column) RIM for
representative boreal landscape element types showing the effects of present-day forestry (PDF) and future forestry (FF).
Each figure shows average monthly DOC fluxes from the KC scenario (filled blue shapes), the PDF scenario (red line), and a
summary of the FF scenario results where the parameterization for present-day forestry was forced with downscaled RCM
outputs. The black dashed line shows the lowest average DOC flux from the ensemble of 15 RCM scenarios used to generate
FF results, while the blue line shows the maximum of the scenario averages. The green line shows the overall scenario median
flux. Note the differences in scale between landscape element types.
Equations (10)–(12) must be prorated by runoff adjusted for landscape-scale forestry impacts (equation (13))
ð1 – f Þ r intact þ f r intact X
x
3
X3
ðmx ð1-gx ÞÞ þ f r forestry x ðmx gx Þ
(13)
Combining equations (10)-(13), it is possible to estimate landscape-scale DOC concentrations at the catchment
outlet [DOC]s:
½DOCs ¼ ðð10Þ þ ð11Þ þ ð12ÞÞ =ð13Þ
(14)
As a reference point to the scenarios we also give the present-day observed DOC. Measurements of [DOC]
made at the catchment outlet (Figure 2, Site 16) were interpolated in time (2005-2012) to obtain daily values.
The frequency of the measurements varies from monthly during low winter flow to 2 days-weekly during
spring flood. Daily interpolated values were used to calculate a mean observed concentration and flux representative of present-day conditions.
Future landscape-scale scenarios representing climate change with no forestry (CC), “Business as Usual” forestry under a future climate (CCF) and more intensive forestry in a future climate (CCI; Table 1) were simulated
in the following manner. Outputs from both RIM and INCA-C, for the three headwater landscape-elements
estimated under the CC and FF scenarios, were used as inputs in the LMM to model landscape-scale DOC concentrations and fluxes at the outlet during the years 2061–2090 using the above mixing model (equation (4)).
This was done so as to integrate the local-scale climate change DOC signals into possible future overall
landscape-scale DOC signals.
ONI ET AL.
FUTURE BOREAL DOC
2414
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
3. Results
3.1. Model Calibrations to
Present-Day Conditions
Figure 9. Plots of (a) observed (blue dots) and modeled (grey line) average
daily runoff from site C7 for 2005–2012 and (b) differences between the forestry runoff scenario based on Balsjö flows (PDF; Table 1) and present-day
simulations at Krycklan (KC scenario). Note that the runoff simulations were
performed for 1981–2012. Only 2005–2012 results are shown for comparison
with DOC results and to better display seasonal patterns. The negative spikes
in Figure 9b are the result of lower snowmelt threshold temperatures in
the Balsjö clear-cut scenario.
The “Krycklan Calibrated” KC scenario
(Table 1) includes all model calibrations to present conditions observed
at Krycklan. The PERSiST model parameterization used here simulated
present-day C7 streamflow with
NSE values of 0.66 for both logtransformed and untransformed flows
between 1981 and 2012 (Figure 9).
While the total volume of runoff simulated matched observations, there are
some discrepancies in summer 2011
and 2012 where precipitation measurements did not match streamflow.
Between 2005 and 2012, runoff simulations captured the monthly patterns
(Figure 9). However, the spring peak
was, in general, slightly underestimated; summers were slightly too
wet and late autumn simulations
slightly too dry.
Both INCA-C and RIM were able
to reproduce the short-term [DOC]
dynamics in the three landscapeelement types (Figure 10) with NS
values of 0.30 (forest on till), 0.39 (mire), and 0.47 (forest on sediment) for INCA-C and NS values of 0.62
(forest on till), 0.57 (mire), and 0.27 (forest on sediment) for RIM. It is noteworthy that while both models
had similar performance as evaluated by NS statistics, the overall patterns simulated were different, both
in terms of simulated average concentrations and fluxes and temporal dynamics. This effect is most
pronounced for the wetland and forest on sediment simulations. For wetlands, the INCA-C simulation
did a better job of capturing the snowmelt associated DOC dilution than RIM. The INCA-C simulation for
forest on sediment displays considerably more short-term variability than the RIM simulations for that
landscape-element type.
3.2. Downscaled Climate Projections
Outputs from the 15 downscaled RCM experiments suggested warmer (5.5°C versus 1.8°C) and wetter
(749 mm precipitation annually versus 631 mm) conditions in 2061–2090 as compared to the present day
(Figure 4). Downscaled average monthly temperatures from the 15 RCM experiments were similar to the
highest average monthly temperatures observed during the present day (1981–2012). Maximum monthly
average temperatures are projected to be significantly warmer in all months while minimum monthly
average temperatures from the 15 RCM experiments are relatively similar to the coldest monthly average
temperatures under present-day conditions. The projected average monthly temperatures for 2061–2090
are similar to the maximum monthly average present-day temperatures (Figure 4).
The RCM outputs suggested that there is likely to be more precipitation in the future (Figure 4). However, the
range of average monthly precipitation values projected by 15 RCM experiments was similar to average
monthly present-day precipitation. There was little change in minimum monthly precipitation, but it is
noteworthy that maximum future precipitation was much higher than today. There was greater variation
in projected future precipitation than observed today (not shown). Projected future temperatures were less
variable than under present conditions.
ONI ET AL.
FUTURE BOREAL DOC
2415
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 10. Results of local-scale model calibration (KC) to present-day [DOC] using (a–c) INCA-C and (d–f) RIM for landscape element types characteristic of the boreal ecozone. Solid lines represent model simulations while yellow dots are observed [DOC].
3.3. Local-Scale Present-Day Forestry Effects
The “Present-Day Forestry” PDF scenario was based on adjusting hydrological and biogeochemical model
parameters from the KC scenario so as to simulate clear-cutting of the representative landscape-element
types at Krycklan based on observed changes at the Balsjö clear-cut experiment. The increase in runoff
observed at Balsjö following forest harvest was successfully reproduced at Krycklan in the PDF scenario
(Figures 3 and 9). Winter flows were unchanged, but there was a larger spring peak associated with greater
snow accumulation. Higher summer and autumn runoff were simulated as rainfall contributed to runoff
instead of recharging soils, which had dried out due to evapotranspiration.
Soil thermal conductivity parameters in INCA-C were adjusted so as to match the monthly average differences in soil temperature between clear-cut and intact forest reported by Schelker et al. [2012] (Figure 11).
INCA-C and RIM simulations were then run with the adjusted soil temperatures and hydrology from the
PERSiST forestry scenario simulations. Results differed between landscape-element types and between RIM
and INCA-C (Figures 5 and 6). In all cases except INCA-C wetland simulations, models simulated higher summer [DOC] (Figure 5). Winter [DOC] was similar between KC and PDF simulations in all cases, due to the lack of
significant change in soil temperatures under the snowpack (Figure 11) or runoff (Figures 3 and 9). Higher
DOC fluxes were simulated by both models for all landscape-element types (Figure 6). INCA-C and RIM simulated similar seasonal DOC flux patterns in both of the forest landscape-element types, with the majority of
the increase occurring in summer and autumn. The largest difference in model simulations was obtained for
wetlands, where INCA-C projected small flux increases associated with forest harvesting, while RIM simulated
extremely high summer fluxes (Figure 6).
3.4. Local-Scale Climate Change Simulations
Local-Scale Climate Change (CC) scenario results showed that possible future climate would lead to shifts in
both the timing and amount of runoff and DOC. Runoff projections under a future (2061–2090) climate
(Figure 3) showed a clear shift in timing and suggested a possibility of both increases and decreases in total
amount. Because of the warmer temperatures (Figure 11), higher winter flows were simulated and the spring
peak was relatively smaller due to less snow accumulation and more of the total precipitation falling as rain.
The smallest changes in flow between present day (KC) and under a future climate (CC) were projected for
ONI ET AL.
FUTURE BOREAL DOC
2416
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 11. Plots showing the simulated average monthly deviations in soil temperature between and PDF and KC scenarios
used to drive INCA-C and RIM simulations. (a–c) Based on observations of soil temperature changes after clear-cut at (d) Balsjö
as reported by Schelker et al. [2012]. Positive values mean that the soil was warmer after clear felling while negative values
represent cooler soils.
June. The minimum runoff projected from the CC scenario runs was considerably lower (192 mm yr1) than
average present-day conditions (308 mm yr1), primarily due to greatly reduced spring flows. The average
runoff from the 15 CC projections did not differ substantially from present-day (KC) conditions in terms of
absolute amount (308 versus 314 mm yr1). The maximum runoff projected under the CC scenario
(513 mm yr1) was much higher than present day with high winter flows and a pronounced snowmelt peak
in April.
The simulated changes in runoff are driven by projected changes in both temperature and precipitation
(Figure 4). In PERSiST, warmer temperatures result in a shift in the balance between snowfall and rain as well
as higher rates of evapotranspiration (and potentially lower runoff). Higher precipitation results in increased
runoff and wetter soils if it exceeds the increased evapotranspiration associated with warmer temperatures.
Both INCA-C and RIM simulated similar climate change effects on [DOC] (Figure 5). Slight increases in concentration were simulated in all cases except for INCA-C wetland simulations, where the range of projected
future concentrations matched present-day conditions. Seasonal patterns of projected fluxes differed
markedly between models (Figure 6). For all landscape-element types, INCA-C projected a large increase in
spring DOC flux, while RIM simulations suggested that spring flux would be reduced and late autumn and
winter fluxes would increase. There were only slight increases in projected summer fluxes (Figure 6).
3.5. Combined Local-Scale Effects of Climate Change and Forestry
A similar pattern to that observed for CC versus KC scenario simulations was seen when comparing the future
climate change + forestry scenario (FF; Table 1) to the PDF scenario at Krycklan (Figure 3). The minimum average runoff from the 15 FF simulations (466 mm yr1) was similar to the PDF results (547 mm yr1) in all seasons
except spring. The FF ensemble median runoff (645 mm yr1) was slightly higher than that for the PDF scenario, while the maximum average runoff simulated by one of the 15 FF runs was much wetter (863 mm yr1)
than present-day\ observed (KC) conditions (308 mm yr1). To put these results into context, the August
ensemble maximum runoff under the FF scenario is similar to the present-day spring runoff in May.
A summary of FF scenario results showing the combined possible effect of climate change and clear-cutting
on stream DOC concentrations and fluxes is presented in Figures 7 and 8. In general, INCA-C simulations
suggest small decreases in concentration between future and present-day clear-cuts, while RIM projections
suggest increased future [DOC] as compared to present-day clear-cut impacts (Figure 7). It was only during
summer months on forest on sediment land cover type that INCA-C simulations projected higher future
ONI ET AL.
FUTURE BOREAL DOC
2417
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
[DOC] than under the PDF scenario (Figure 7). The two biogeochemical models diverge in their projections of
DOC dynamics in wetlands as INCA-C projected decreasing concentrations in the future while RIM projected
increases. Both models projected increased DOC export from till and sediment forests, while INCA-C projected fluxes from wetlands were similar to present conditions (Figure 8).
3.6. Summary of Local Effects and Projected Landscape-Scale DOC Response
Local-scale average DOC concentrations and fluxes simulated by INCA-C and RIM under present and future
conditions are presented in Figure 12, while landscape-scale results are summarized in Figure 13. When
summarized at an annual scale, there were notable differences in both concentration and flux both between
landscape-element types and between models (Figure 12). In general, there was good correspondence
between present-day simulations and observed data for concentrations and fluxes in all landscapeelement types.
For forests on till, INCA-C simulations suggested the most extreme increase in [DOC] would occur due to
present-day forestry effects (PDF scenario; Figure 12a). Future [DOC] simulated under the CC and FF simulations were intermediate to present-day calibration (KC) and present-day forestry (PDF) scenario concentrations. RIM simulated concentrations for present-day forestry (PDF scenario) were within the range of
climate (CC) scenario concentrations (Figure 12a). Unlike INCA-C, RIM projected the highest [DOC] for the
combined climate + forestry effects (FF) in forests on till (Figure 12a).
In wetlands, INCA-C and RIM projected different directions of change in [DOC] (Figure 12b). INCA-C projected
declines in [DOC] under all scenarios. The largest declines were in the climate change + forestry effects (FF),
followed by present-day forestry (PDF) and climate change (CC) scenarios (Figure 12b). By contrast, RIM projected similar or increasing [DOC] under all scenarios. RIM-projected CC scenario [DOC] was not appreciably
different from the present-day simulated (KC) or observed, and future forestry (FF) effects [DOC] was slightly
higher than simulated for present-day forestry (PDF). RIM simulations of [DOC] under the FF were higher than
any present-day observed or simulated values (Figure 12b).
INCA-C simulations of [DOC] for forests on sediment were similar to those for forests on till. The range of
values projected under the climate change (CC) and climate change + forestry effects (FF) was similar and
higher than present-day modeled (KC) and observed [DOC] (Figure 12c). Forestry effect simulations displayed
high interannual variability in [DOC] and spanned a similar concentration range to that of the climate scenarios. The RIM simulations for sediment forests presented a [DOC] pattern more similar to RIM simulated [DOC]
from wetlands. Projected [DOC] under the climate scenarios (CC) was similar to present-day concentrations,
and forestry effects [DOC] was intermediate to the range of values simulated in the CC and FF scenarios
(Figure 12c). An important finding was that on the local scale (Figure 12) the combined effects of climate
change and forestry effects on [DOC] were not additive.
INCA-C simulated increased DOC fluxes from both till and sediment forests (Figures 12d and 12f) under both
climate change (CC) and future forestry (FF) scenarios. For both forest types, INCA-C projected slight flux
increases under a changing climate (CC) and a large increase under the climate change + forestry effects scenarios (FF). In both cases, projected DOC fluxes for forestry effects were similar to values simulated for climate
change + forestry effects (Figures 12d and 12f). Very little change in flux from wetlands was simulated by
INCA-C for any of the climate scenarios (Figure 12e).
RIM simulated similar fluxes for till forests and wetlands (Figures 12d and 12e). Present-day (KC) fluxes were
similar to those under the CC scenario. The largest flux and most variable flux increases were projected for FF
scenario. Simulated present-day forestry (PDF scenario) fluxes were intermediate to CC and FF fluxes. For
forests on sediment, RIM projected a slightly smaller increase in flux for the PDF scenario than for either
CC or FF scenarios.
When the individual landscape-element type responses were integrated across a larger catchment using the
LMM a different pattern emerged (Figure 13). Observed average [DOC] at the Krycklan outlet was 9.9 mg L1,
which was in between the KC scenarios which assumed no clear-cuts (8.9 and 9.0 mg L1 for INCA-C and RIM)
and the present-day BAU scenarios (10.0 and 10.4 mg L1 for INCA-C and RIM, respectively). This is to be
expected since approximately 25% of the Krycklan catchment has been protected from harvesting since
1922 [Laudon et al., 2013a] and the catchment as a whole therefore should have a smaller DOC signal from
ONI ET AL.
FUTURE BOREAL DOC
2418
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 12. Summary of present-day (KC and PDF) and projected future (CC and FF) local-scale INCA-C (horizontal axis) and
RIM (vertical axis) simulations showing annual average (left column) DOC concentrations and (right column) fluxes for
forest on (a and d) till, (b and e) wetland, and (c and f) forest on sediment. Lengths of the bars on the crosses show standard
errors of the annual values. Grey dots represent average observed concentrations and fluxes between 2005 and 2012. Red
dots show present-day calibrated values from the KC scenario. The blue cross represents values from the PDF scenario.
Yellow dots represent the average values from each of the 15 RCM forcings in the CC scenario. White diamonds show the
results of individual RCM forcings in the future forestry (FF) scenario. Note that the effects of climate change and forestry
are not additive. If effects were additive, projections would move upwards along the diagonal instead of displaying the
trajectories presented here.
forestry than the average forest landscape in Sweden that was considered in the BAU scenarios. A range of
average concentrations between 9.3 and 11.3 mg L1 was projected for the CC scenario, in which there were
no clear-cuts simulated. The average INCA-C projected concentration (10.6 mg L1) was slightly higher than
that projected by RIM (10.2 mg L1). For the future “Business As Usual” (CCF) scenario, a range of annual average [DOC] between 10.5 and 13.3 mg L1 was projected. The average INCA-C projected concentration
ONI ET AL.
FUTURE BOREAL DOC
2419
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Figure 13. Present day (KC, BAU, and IFH scenarios) and potential future (CC, CCF, and CCI scenarios) landscape-scale
annual average (a) DOC concentrations and (b) fluxes from INCA-C (horizontal axis) and RIM (vertical axis) as simulated the
Landscape Mixing Model at the Krycklan catchment outlet. The bars on the crosses show standard errors of the mean values.
Observed values are represented using a grey circle. The two landscape scale, present-day forestry scenarios are shown
as light blue circles (BAU) and bright yellow circles (IFH). The points labeled “No Forestry” represent the upscaled KC scenario
assuming no clear-cutting in the catchment. White diamonds represent individual RCM forcings under the CC scenario which
assumes no clear-cutting and a changing climate. Dark blue dots represent individual RCM forcings under the CCF climate
change and business as usual forestry scenario (0.86 % of the catchment harvested annually), while dark yellow dots
represent individual RCM forcings under the CCI climate change and intensive (2X business as usual) forest harvest scenario.
At the landscape scale, effects of forestry and are additive as the scenario realizations move approximately along the diagonal
as the intensity of forest harvest increases.
(11.3 mg L1) was lower than that projected by RIM (12.0 mg L1). Higher average concentrations were projected under the more intensive forest harvesting (CCI scenario; 10.8 to 14.6 mg L1), the average INCA-C projected concentration was substantially lower than the RIM estimate (11.6 versus 13.2 mg L1).
The pattern of modeled and observed fluxes (Figure 13b) was harder to interpret. The observed flux
(4.2 g m2 yr1) was more similar to the KC scenario, which assumed no clear-cuts (3.8 and 4.3 g m2 yr1
for INCA-C and RIM, respectively) than it was to the present-day BAU (4.7 and 5.2 g m2 yr1 for INCA-C
and RIM, respectively) or IFH scenarios (5.6 and 6.1 g m2 yr1 for INCA-C and RIM). Projected future fluxes
showed a similar pattern to concentrations, increasing forest harvest intensity led to higher fluxes. Under
the CC scenario, which assumed 0% clear-cut, a range of average fluxes between 4.1 and 6.4 g m2 yr1
was projected. The range of RIM projected values (2.3 g m2 yr1) was higher than the range of INCA-C projections (0.8 g m2 yr1), and the average RIM projected value was higher than that obtained using INCA-C
(5.2 versus 4.6 g m2 yr1). Similar behaviors were seen for the two climate change + forestry scenarios.
Under the CCF scenario, projected fluxes ranged between 5.0 and 7.8 g m2 yr1. The range of RIM projections (2.7 g m2 yr1) was higher than those for INCA-C (0.8 g m2 yr1), and the average RIM projected value
was higher than that from INCA-C simulations (6.5 versus 5.3 g m2 yr1). Under the more intensive forestry
CCI scenario, projected fluxes ranged between 5.7 and 9.4 g m2 yr1. The range of RIM projections
(3.1 g m2 yr1) was higher than those for INCA-C (1.4 g m2 yr1) and the average RIM projected value
was higher than that from INCA-C simulations (7.9 versus 6.1 g m2 yr1).
An interesting result was that when the individual landscape-element type responses (Figure 12) were integrated across a larger catchment to the landscape scale a different pattern emerged where the combined
signal of climate change + forestry effects went from not being additive to mostly being additive (Figure 13).
4. Discussion
Many studies have shown the vulnerability of boreal forest aquatic ecosystems to climate change impacts
[Bonan, 2008; Davidson and Janssens, 2006; Laudon et al., 2012, 2013b; Nabuurs et al., 2013; Tetzlaff et al.,
2013] and the impacts of forest harvesting on surface water quality [Kreutzweiser et al., 2008; Laudon et al.,
2011b; Palviainen et al., 2014; Schelker et al., 2012, 2013a, 2014]. To date, there has been limited evaluation
ONI ET AL.
FUTURE BOREAL DOC
2420
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
of the combined impact of these two stressors on water quality in the boreal forest. Uncertainties associated with the possible effects of multiple stressors on boreal water resources can be evaluated by the
use of ensembles of climate scenarios and multiple models of biogeochemical processes. However, the
huge range of sources of uncertainty including (but not limited to) emissions scenario, GCM representation of climate processes, downscaling technique, data availability, model lack of fit, and model structural
and parameter uncertainty makes it impossible to perform a full uncertainty analysis. In light of this limitation, this study presents simulations of possible future water quality at local and landscape scales
under a range of climate and forest harvest scenarios for use in storylines, or internally consistent
descriptions of plausible futures.
4.1. Hydrological Responses
Adequate hydrological simulations are needed for catchment biogeochemical modeling. The simulations presented here were able to capture the pattern of hydrological response at the C7 gauging site downstream of
the forest on till (C2) and wetland (C4) headwater catchments. It should be noted that relying on one parameter
set realization from a single rainfall-runoff model introduces uncertainties in the projection of future conditions.
Specifically, climate change effects on evapotranspiration are still inadequately understood [Watts et al., 2015].
Using multiple model runs with a range of values for parameters related to evapotranspiration would result in
an improved representation of uncertainties, but would require significant computer resources.
Parameter values related to effects of clear-cuts were adjusted so as to reproduce the forest harvesting effect
observed at the nearby Balsjö catchment. There has been a large number of paired catchment studies in
Sweden [Futter et al., 2010] and elsewhere [Brown et al., 2005] documenting the effects of forest harvesting
on runoff. While the general trend toward increasing summer runoff is apparent in almost all studies, the
exact magnitude of effects varies. Balsjö is the closest operational study for which data are available during
the calibration period used here. Some uncertainty has been introduced into the results presented here as it
was assumed that Krycklan C7 runoff is representative of boreal Swedish headwaters and that hydrological
response at Balsjö would have been observed at other clear-cuts. Unfortunately, the necessary data are
not currently available to conclusively test these hypotheses.
At a landscape scale, simulations suggested a slight increase in runoff associated with forest harvesting,
mostly during the growing season. This is consistent with another study from north eastern Ontario where
moderate increases in low flows in medium to large rivers with different degrees of forest harvest intensity
were reported [Buttle and Metcalfe, 2000].
When using meteorological data from the RCM experiments to drive the rainfall-runoff model, a shift in
spring melt regimes due to shortened and warmer winters was simulated (Figures 3 and 4). This is consistent
with other climate change impact studies in the boreal forest [Holmberg et al., 2014; Oni et al., 2014; Thorne,
2011; Woo et al., 2008]. The general trend toward slight increases in future precipitation is also consistent with
other studies [Chen et al., 2015].
4.2. Calibrating Local-Scale Biogeochemical Models
There are a number of approaches to modeling [DOC] including neural network [Aitkenhead-Peterson et al.,
2005; Aitkenhead et al., 2007], process-based [Futter et al., 2007; Naden et al., 2010; Wu et al., 2014], and empirical [Hejzlar et al., 2003; Herrmann et al., 2015; Temnerud et al., 2014] approaches. The use of two models in the
simulations presented here should not be construed as spanning the range of modeling possibilities, but is
more akin to “getting a second opinion” from the medical profession.
Both RIM and INCA-C were able to simulate seasonal and long term patterns in present-day [DOC] at each of the
headwater sites (Figure 10). RIM simulations had better goodness of fit statistics than were obtained produced
using INCA-C. This is not surprising as empirical regression models almost always produce simulations with
smaller residuals than can be obtained using process-based models [Futter and de Wit, 2008]. The differences
between INCA–C and RIM scenario projections are related to model structure. RIM is a nonlinear regression
model which simulates monotonic responses to soil temperature and runoff. Thus, it can give more extreme
projections than obtained by INCA-C, which simulates both production and consumption of DOC, and is therefore able to simulate both positive and negative responses to the same environmental forcing. This means that
the INCA-C model structure is consistent with Laudon et al. [2012], who suggest that [DOC] displays a modal
response to temperature with concentrations increasing up to a point and then declining with further warming.
ONI ET AL.
FUTURE BOREAL DOC
2421
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
While empirical models such as RIM are, in general, easier to apply and produce better goodness of fit statistics than process-based models such as INCA-C, their use in scenarios is problematic. Empirical models
are developed using correlative patterns, and thus reflect current relationships between dependent and
independent variables [Crossman et al., 2014]. If these relationships do not hold in the future, empirical
models will give spurious results [Leavesley, 1994]. On the other hand, process-based models such as
INCA-C have a greater focus on representing the underlying biogeochemical and physical processes that
describe system behavior [Adams et al., 2012; Crossman et al., 2014]. As model parameters have a specific
physical meaning, which can be defined for future altered conditions, process-based models such as INCAC can be applied with more confidence outside the range of data under which they were developed, and
hence have a greater ability to deliver credible projections under possible future conditions [Leavesley,
1994; Crossman et al., 2014].
4.3. Climate Change and Forestry Impacts on DOC in the Local Scale
Boreal forest landscapes are made up of distinct landscape elements with unique biogeochemical signatures.
This heterogeneity results in complex biogeochemical responses across the spatial gradient from local to
landscape scale in large catchments [Löfgren et al., 2009; Schelker et al., 2014; Sørensen et al., 2009a].
Whereas warmer and wetter soils associated with a changing climate and forest harvesting often enhance
DOC production and transport [Hongve et al., 2004; Öquist et al., 2014; Tipping et al., 1999], it is unclear
whether the combined impacts of climate change and forest harvesting will increase or reduce stream
[DOC] due to uncertainties associated with landscape heterogeneity, hydrologic connectivity, incomplete
process understanding, and other local or regional influences [Clark et al., 2010]. In this study, we modeled
the effects of climate change only and forestry effects only (Figures 5 and 6), but also the combined effect
of climate change and forestry (Figures 7, 8, and 12). Combined climate and forestry effects were assessed
by incorporating the hydrological and temperature change signals associated with clear-cutting into the
parameterizations used to drive the hydrological and biogeochemical models, thereby allowing them to
interact and give a combined DOC signal. Both climate change and forestry lead to changes in timing of flow
and amount of runoff. On their own, climate change leads to earlier, smaller spring peaks and clear-cuts promote sustained summer base flow. The combined effects of climate change and forestry will be unprecedented in terms of their wetness. This will almost certainly impact soil processes and hence [DOC]. At a
local scale the combined signals from climate change and forestry was not additive, suggesting that there
are complex, nonlinear interactions between them that affect the stream biogeochemistry of headwaters
draining different landscape elements. However, when aggregated to the landscape scale this pattern
disappeared and the diversified signals from the landscape-elements combined into a general signal from
the forest landscape that was mostly additive (Figure 13).
The assumed duration of harvest effects adds additional uncertainty to the results presented here. The
long-term modeled impact of forestry on stream DOC is highly dependent on the duration of increased
[DOC] following a harvest; it is also important to support long-term monitoring studies, which evaluate the
mechanisms and duration of increased mobilization of DOC after clear-cutting. Recent studies suggest that
fluvial DOC responses will result from forestry-induced hydrologic changes/flow paths as a primary driver,
but that the higher soil temperature, which control biogeochemical process and transpiration rates, may also
contribute [Palviainen et al., 2014; Schelker et al., 2013b].
4.4. Fluvial Carbon Responses at the Landscape Scale and Environmental Implications for the Future
This study is a modeling exercise, in which regional climate change models were linked to hydrological and
biogeochemical models so as to simulate local water chemistry, which is then mixed in a landscape mixing
model to provide overall landscape DOC concentration and export scenarios for the whole boreal forest landscape. Environmental modeling always encompasses uncertainties; one way of addressing this is to use an
ensemble modeling approach. Using an ensemble of multiple models makes explicit the existence of uncertainty, but it does not set quantitative bounds on the full range of future possibilities. Instead, the model outputs presented here should be treated as multiple plausible outcomes or scenario storylines [Börjeson et al.,
2006; Milestad et al., 2014; Rounsevell and Metzger, 2010], which can assist decision making. This is analogous
to how Intergovernmental Panel on Climate Change uses scenario storylines to present possible effects of
climate change.
ONI ET AL.
FUTURE BOREAL DOC
2422
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Previous research works on critical harvesting thresholds suggest that at least 11% of a catchment needs to
be clear-cut in order for a significant (p < 0.05) increase in DOC concentrations can be detected in downstream larger stream reaches [Schelker et al., 2014]. In the BAU scenario 0.86% was clear-cut annually and
the effects were assumed to last 20 years, which means that on average 17% of the catchment could be considered to be affected by forest harvesting at all times. Thus, it is not surprising that a clear signal of forestry
effects on [DOC] was simulated.
There are a number of possible explanations for the additivity of climate and forestry effects at the landscape
scale. Wetlands cover a relatively small amount (9%) of the catchment; thus, their influence will be relatively
minor at the landscape scale and their local-scale nonadditivity will be hard to detect. The local-scale forest
on till and forest on sediment [DOC] projections are similar and closer to additive. Because they are the dominant land cover types in the catchment, they will drive landscape-scale concentration inputs. Runoff is additive across scenarios and this will tend to make fluxes additive. Furthermore, any differences in land-cover
specific concentrations are masked by the homogeneous groundwater inputs assumed for all scenarios.
Although the responses to the future scenarios differed between different landscape units, the overall signal
from the whole forest landscape is probably the most important result as this is what affects the downstream
lakes and rivers. If the modeling results presented here hold true, then climate change can be expected to
increase DOC concentrations in stream water draining the forest landscape in the future in the order of
1 mg L1 (Figure 13). Add to this 1 mg L1, (approximately 10%) driven by effects of forestry and if forestry
will intensify in the future an additional 1 mg L1 (Figure 13). The results from the storyline scenarios in this
study also showed that DOC fluxes from the forest landscape could increase in the future, from somewhere
around a 50% increase to in the worst-case scenarios doubling, in the upcoming 75 years. This might affect
the carbon balance of the forest landscape as much of the carbon that enters into a stream network is quickly
released back into the atmosphere. This vertical CO2 flux can make up 50% of the entire stream flux [Wallin
et al., 2013]. Between 30 and 80% of the carbon entering into Scandinavian freshwater system has been
shown to be lost in lakes through sedimentation and mineralization before reaching the Baltic Sea
[Algesten et al., 2004]. This still means that 20–70% of the increased loadings are likely to reach the sea.
The brownification of surface waters and coastal waters suggested in this study will most likely affect the
nutrient status and light regime and shift food webs toward net heterotrophy [Andersson et al., 2013;
Wikner and Andersson, 2012]. While eutrophication is one of the major environmental concerns for the
Baltic Sea [Karlson et al., 2002; Saikku and Asmala, 2010], increasing fluxes of DOC which acts as a transport
vector for nutrients [Stepanauskas et al., 2002], metals, and persistent organic pollutants [Bergknut et al., 2010]
may also harm Baltic ecosystem health.
Acknowledgments
This project was funded by two larger
projects ForWater and Future Forest,
studying the effect of climate and forest
managements on boreal water
resources. More information about the
data used in this study is available in
Laudon et al. [2013a] or can be requested
at http://www.slu.se/en/departments/
forest-ecology-management/research/
krycklan-catchment-study-new/
dataservice/. Funding for KCS came from
Swedish Science Council (SITES),
Formas (ForWater), SKB, MISTRA
(Future Forest), and Kempe Foundation.
The ENSEMBLES data used in this work
(http://ensembles-eu.metoffice.com/
docs/Ensembles_final_report_Nov09.pdf)
were funded by the EU FP6 Integrated
Project ENSEMBLES (Contract number
505539), whose support is gratefully
acknowledged. M.N.F. acknowledges the
support of the NordForsk DomQua
project. The comments of two
anonymous reviewers greatly improved
the quality of the final manuscript.
ONI ET AL.
5. Conclusion
No one can predict exactly what will happen in the future, but, by using a set of plausible and internally consistent scenarios and different models, we can constrain uncertainties about the future. So while all scenarios
give different results and while the results of this study cannot be verified until 2090, these modeling exercises can be used to explore plausible possible futures and to guide decision-making in Sweden and elsewhere. This is the first study in which the combined impact of climate change and forestry effects on
boreal stream biogeochemistry has been assessed. The combined impacts of expected climate change and
forest harvesting on DOC concentrations and fluxes indicated complex interactive processes on the local
scale as the results were not additive. However, when aggregating to the landscape scale the results were
found to be mostly additive and both forestry and climate change will increase DOC concentrations and
fluxes in large-scale boreal surface waters. While the average projected magnitude of any increase in
[DOC] is less than that already observed in many boreal waterbodies recovering from acidification, the
plausible increases in DOC flux may have negative impacts on the Baltic Sea.
References
Adams, H. D., A. P. Williams, C. Xu, S. A. Rauscher, X. Jiang, and N. G. McDowell (2012), Empirical and process-based approaches to climateinduced forest mortality models, Front. Plant Sci., 4, 438–438.
Ågren, A., I. Buffam, M. Jansson, and H. Laudon (2007), Importance of seasonality and small streams for the landscape regulation of dissolved
organic carbon export, J. Geophys. Res., 112, G03003, doi:10.1029/2006JG000381.
FUTURE BOREAL DOC
2423
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Ågren, A., I. Buffam, D. Cooper, T. Tiwari, C. Evans, and H. Laudon (2014), Can the heterogeneity in stream dissolved organic carbon be
explained by the contributing landscape elements?, Biogeosciences, 11(4), 1199–1213.
Aitkenhead, M., J. Aitkenhead-Peterson, W. McDowell, R. Smart, and M. Cresser (2007), Modelling DOC export from watersheds in Scotland
using neural networks, C. R. Geosci., 33(3), 423–436.
Aitkenhead-Peterson, J., J. Alexander, and T. Clair (2005), Dissolved organic carbon and dissolved organic nitrogen export from forested
watersheds in Nova Scotia: Identifying controlling factors, Global Biogeochem. Cycles, 19, GB4016, doi:10.1029/2004GB002438.
Algesten, G., S. Sobek, A. K. Bergström, A. Ågren, L. J. Tranvik, and M. Jansson (2004), Role of lakes for organic carbon cycling in the boreal
zone, Global Change Biol., 10(1), 141–147.
Andersson, A., I. Jurgensone, O. F. Rowe, P. Simonelli, A. Bignert, E. Lundberg, and J. Karlsson (2013), Can humic water discharge counteract
eutrophication in coastal waters?, PLoS One, 8(4e61293), doi:10.1371/journal.pone.0061293.
Apps, M., W. Kurz, R. Luxmoore, L. Nilsson, R. Sedjo, R. Schmidt, L. Simpson, and T. Vinson (1993), Boreal forests and tundra, Water Air Soil
Pollut., 70(1-4), 39–53.
Aufdenkampe, A. K., E. Mayorga, P. A. Raymond, J. M. Melack, S. C. Doney, S. R. Alin, R. E. Aalto, and K. Yoo (2011), Riverine coupling of
biogeochemical cycles between land, oceans, and atmosphere, Front. Ecol. Environ., 9(1), 53–60.
Battin, T. J., S. Luyssaert, L. A. Kaplan, A. K. Aufdenkampe, A. Richter, and L. J. Tranvik (2009), The boundless carbon cycle, Nat. Geosci., 2(9),
598–600.
Bergknut, M., S. Meijer, C. Halsall, A. Ågren, H. Laudon, S. Köhler, K. C. Jones, M. Tysklind, and K. Wiberg (2010), Modelling the fate of
hydrophobic organic contaminants in a boreal forest catchment: A cross disciplinary approach to assessing diffuse pollution to surface
waters, Environ. Pollut., 158(9), 2964–2969.
Bonan, G. B. (2008), Forests and climate change: Forcings, feedbacks, and the climate benefits of forests, Science, 320(5882), 1444–1449.
Börjeson, L., M. Höjer, K.-H. Dreborg, T. Ekvall, and G. Finnveden (2006), Scenario types and techniques: Towards a user’s guide, Futures, 38(7),
723–739.
Brown, A. E., L. Zhang, T. A. McMahon, A. W. Western, and R. A. Vertessy (2005), A review of paired catchment studies for determining changes
in water yield resulting from alterations in vegetation, J. Hydrol., 310(1), 28–61.
Buttle, J., and R. Metcalfe (2000), Boreal forest disturbance and streamflow response, northeastern Ontario, Can. J. Fish. Aquat. Sci., 57(S2),
5–18.
Chen, D., C. Achberger, T. Ou, U. Postgård, A. Walther, and Y. Liao (2015), Projecting future local precipitation and its extremes for Sweden,
Geogr. Ann. Ser. A Phys. Geogr., 97(1), 25–39.
Christiansen, L. (2014), Swedish Statistical Yearbook of Forestry, 363 pp., Swedish For. Agency, Jönköping, Sweden.
Clark, J., S. Bottrell, C. Evans, D. Monteith, R. Bartlett, R. Rose, R. Newton, and P. Chapman (2010), The importance of the relationship between
scale and process in understanding long-term DOC dynamics, Sci. Total Environ., 408(13), 2768–2775.
Cole, J. J., Y. T. Prairie, N. F. Caraco, W. H. McDowell, L. J. Tranvik, R. G. Striegl, C. M. Duarte, P. Kortelainen, J. A. Downing, and J. J. Middelburg
(2007), Plumbing the global carbon cycle: Integrating inland waters into the terrestrial carbon budget, Ecosystems, 10(1), 172–185.
Cooper, D., A. Jenkins, R. Skeffington, and B. Gannon (2000), Catchment-scale simulation of stream water chemistry by spatial mixing: Theory
and application, J. Hydrol., 233(1), 121–137.
Crossman, J., M. N. Futter, P. G. Whitehead, E. Stainsby, H. M. Baulch, L. Jin, S. K. Oni, R. L. Wilby, and P. J. Dillon (2014), Flow pathways and
nutrient transport mechanisms drive hydrochemical sensitivity to climate change across catchments with different geology and
topography, Hydrol. Earth Syst. Sci., 18(12), 5125–5148.
Cui, X., S. Liu, and X. Wei (2012), Impacts of forest changes on hydrology: A case study of large watersheds in the upper reaches of Minjiang
River watershed in China, Hydrol. Earth Syst. Sci., 16(11), 4279–4290.
Davidson, E. A., and I. A. Janssens (2006), Temperature sensitivity of soil carbon decomposition and feedbacks to climate change, Nature,
440(7081), 165–173.
Déqué, M., D. Rowell, D. Lüthi, F. Giorgi, J. Christensen, B. Rockel, D. Jacob, E. Kjellström, M. De Castro, and B. van den Hurk (2007), An
intercomparison of regional climate simulations for Europe: Assessing uncertainties in model projections, Clim. Change, 81(1), 53–70.
Egnell, G., H. Laudon, and O. Rosvall (2011), Perspectives on the potential contribution of Swedish forests to renewable energy targets in
Europe, Forests, 2(2), 578–589.
Ehret, U., E. Zehe, V. Wulfmeyer, K. Warrach-Sagi, and J. Liebert (2012), HESS Opinions “Should we apply bias correction to global and regional
climate model data?”, Hydrol. Earth Syst. Sci., 16(9), 3391–3404.
Futter, M., D. Butterfield, B. Cosby, P. Dillon, A. Wade, and P. Whitehead (2007), Modeling the mechanisms that control in-stream dissolved
organic carbon dynamics in upland and forested catchments, Water Resour. Res., 43, W02424, doi:10.1029/2006WR004960.
Futter, M., M. Forsius, M. Holmberg, and M. Starr (2009), A long-term simulation of the effects of acidic deposition and climate change on
surface water dissolved organic carbon concentrations in a boreal catchment, Hydrol. Res., 40, 291–305.
Futter, M., E. Ring, L. Högbom, S. Entenmann, and K. Bishop (2010), Consequences of nitrate leaching following stem-only harvesting of
Swedish forests are dependent on spatial scale, Environ. Pollut., 158(12), 3552–3559.
Futter, M., S. Löfgren, S. Köhler, L. Lundin, F. Moldan, and L. Bringmark (2011), Simulating dissolved organic carbon dynamics at the Swedish
integrated monitoring sites with the integrated catchments model for carbon, INCA-C, Ambio, 40(8), 906–919.
Futter, M., M. Erlandsson, D. Butterfield, P. Whitehead, S. Oni, and A. Wade (2014), PERSiST: A flexible rainfall-runoff modelling toolkit for use
with the INCA family of models, Hydrol. Earth Syst. Sci., 10, 8635–8681.
Futter, M. N., and H. A. de Wit (2008), Testing seasonal and long-term controls of streamwater DOC using empirical and process-based
models, Sci. Total Environ., 407(1), 698–707.
Grabs, T., K. Bishop, H. Laudon, S. W. Lyon, and J. Seibert (2012), Riparian zone hydrology and soil water total organic carbon (TOC):
Implications for spatial variability and upscaling of lateral riparian TOC exports, Biogeosciences, 9(10), 3901–3916.
Hagedorn, F., S. Patrick, W. Peter, and F. Hannes (2000), Export of dissolved organic carbon and nitrogen from Gleysol dominated catchments—
The significance of water flow paths, Biogeochemistry, 50(2), 137–161.
Hejzlar, J., M. Dubrovský, J. Buchtele, and M. Růžička (2003), The apparent and potential effects of climate change on the inferred
concentration of dissolved organic matter in a temperate stream (the Malše River, South Bohemia), Sci. Total Environ., 310(1), 143–152.
Herrmann, M., R. G. Najjar, W. M. Kemp, R. B. Alexander, E. W. Boyer, W. J. Cai, P. C. Griffith, K. D. Kroeger, S. L. McCallister, and R. A. Smith
(2015), Net ecosystem production and organic carbon balance of US East Coast estuaries: A synthesis approach, Global Biogeochem.
Cycles, 29, 96–111, doi:10.1002/2013GB004736.
Holmberg, M., M. N. Futter, N. Kotamäki, S. Fronzek, M. Forsius, P. Kiuru, N. Pirttioja, K. Rasmus, M. Starr, and J. Vuorenmaa (2014), Effects of
changing climate on the hydrology of a boreal catchment and lake DOC--Probabilistic assessment of a dynamic model chain, Boreal
Environ. Res., 19, 66.
ONI ET AL.
FUTURE BOREAL DOC
2424
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Hongve, D., G. Riise, and J. F. Kristiansen (2004), Increased colour and organic acid concentrations in Norwegian forest lakes and drinking
water–A result of increased precipitation?, Aquat. Sci., 66(2), 231–238.
Ines, A. V., and J. W. Hansen (2006), Bias correction of daily GCM rainfall for crop simulation studies, Agric. For. Meteorol., 138(1), 44–53.
Intergovernmental Panel on Climate Change (2007), The physical science basis. Contribution of working group I to the fourth assessment
report of the intergovernmental panel on climate change, in Climate Change 2007: The Physical Science Basis, edited by S. Solomon et al.,
996 pp., Cambridge Univ. Press, Cambridge, U. K., and New York.
Jansson, P. E., M. Svensson, D. B. Kleja, and D. Gustafsson (2008), Simulated climate change impacts on fluxes of carbon in Norway spruce
ecosystems along a climatic transect in Sweden, Biogeochemistry, 89(1), 81–94, doi:10.1007/s10533-007-9147-6.
Karlson, K., R. Rosenberg, and E. Bonsdorff (2002), Temporal and spatial large-scale effects of eutrophication and oxygen deficiency on
benthic fauna in Scandinavian and Baltic waters: A review, Oceanogr. Mar. Biol., 40, 427–489.
Kleja, D. B., et al. (2008), Pools and fluxes of carbon in three Norway spruce ecosystems along a climatic gradient in Sweden, Biogeochemistry,
89(1), 7–25, doi:10.1007/s10533-007-9136-9.
Koehler, B., T. Landelius, G. A. Weyhenmeyer, N. Machida, and L. J. Tranvik (2014), Sunlight-induced carbon dioxide emissions from inland
waters, Global Biogeochem. Cycles, 28, 696–711, doi:10.1002/2014GB004850.
Kreutzweiser, D. P., P. W. Hazlett, and J. M. Gunn (2008), Logging impacts on the biogeochemistry of boreal forest soils and nutrient export to
aquatic systems: A review, Environ. Rev., 16, 157–179.
Larsen, S., T. Andersen, and D. O. Hessen (2011), Climate change predicted to cause severe increase of organic carbon in lakes, Global Change
Biol., 17(2), 1186–1192.
Laudon, H., J. Hedtjärn, J. Schelker, K. Bishop, R. Sørensen, and A. Ågren (2009), Response of dissolved organic carbon following forest
harvesting in a boreal forest, AMBIO: A J. Human Environ., 38(7), 381–386.
Laudon, H., M. Berggren, A. Ågren, I. Buffam, K. Bishop, T. Grabs, M. Jansson, and S. Köhler (2011a), Patterns and dynamics of dissolved
organic carbon (DOC) in boreal streams: The role of processes, connectivity, and scaling, Ecosystems, 14(6), 880–893.
Laudon, H., R. A. Sponseller, R. W. Lucas, M. N. Futter, G. Egnell, K. Bishop, A. Ågren, E. Ring, and P. Högberg (2011b), Consequences of more
intensive forestry for the sustainable management of forest soils and waters, Forests, 2(1), 243–260.
Laudon, H., J. Buttle, S. K. Carey, J. McDonnell, K. McGuire, J. Seibert, J. Shanley, C. Soulsby, and D. Tetzlaff (2012), Cross-regional prediction of
long-term trajectory of stream water DOC response to climate change, Geophys. Res. Lett., 39, L18404, doi:10.1029/2012GL053033.
Laudon, H., I. Taberman, A. Ågren, M. Futter, M. Ottosson-Löfvenius, and K. Bishop (2013a), The Krycklan Catchment Study—a flagship
infrastructure for hydrology, biogeochemistry, and climate research in the boreal landscape, Water Resour. Res., 49, 7154–7158,
doi:10.1002/wrcr.20520.
Laudon, H., D. Tetzlaff, C. Soulsby, S. Carey, J. Seibert, J. Buttle, J. Shanley, J. J. McDonnell, and K. McGuire (2013b), Change in winter climate
will affect dissolved organic carbon and water fluxes in mid-to-high latitude catchments, Hydrol. Processes, 27(5), 700–709, doi:10.1002/
hyp.9686.
Leavesley, G. H. (1994), Modeling the effects of climate change on water resources-a review, Clim. Change, 28(1–2), 159–177.
Ledesma, J., T. Grabs, M. Futter, K. H. Bishop, H. Laudon, and S. Köhler (2013), Riparian zone control on base cation concentration in boreal
streams, Biogeosciences, 10(6), 3849–3868.
Lepistö, A., M. N. Futter, and P. Kortelainen (2014), Almost 50 years of monitoring shows that climate, not forestry, controls long-term organic
carbon fluxes in a large boreal watershed, Global Change Biol., 20(4), 1225–1237, doi:10.1111/gcb.12491.
Löfgren, S., E. Ring, C. von Brömssen, R. Sørensen, and L. Högbom (2009), Short-term effects of clear-cutting on the water chemistry of two
boreal streams in northern Sweden: A paired catchment study, AMBIO: A J. Human Environ., 38(7), 347–356.
Magnani, F., M. Mencuccini, M. Borghetti, P. Berbigier, F. Berninger, S. Delzon, A. Grelle, P. Hari, P. G. Jarvis, and P. Kolari (2007), The human
footprint in the carbon cycle of temperate and boreal forests, Nature, 447(7146), 849–851.
Milestad, R., Å. Svenfelt, and K. H. Dreborg (2014), Developing integrated explorative and normative scenarios: The case of future land use in
a climate-neutral Sweden, Futures, 60, 59–71.
Muerth, M., B. Gauvin St-Denis, S. Ricard, J. Velázquez, J. Schmid, M. Minville, D. Caya, D. Chaumont, R. Ludwig, and R. Turcotte (2013), On the
need for bias correction in regional climate scenarios to assess climate change impacts on river runoff, Hydrol. Earth Syst. Sci., 17(3),
1189–1204.
Müller, R. A., M. N. Futter, S. Sobek, J. Nisell, K. Bishop, and G. A. Weyhenmeyer (2013), Water renewal along the aquatic continuum offsets
cumulative retention by lakes: Implications for the character of organic carbon in boreal lakes, Aquat. Sci., 75(4), 535–545.
Nabuurs, G.-J., M. Lindner, P. J. Verkerk, K. Gunia, P. Deda, R. Michalak, and G. Grassi (2013), First signs of carbon sink saturation in European
forest biomass, Nat. Clim. Change, 3, 792–796, doi:10.1038/NCLIMATE1853.
Naden, P. S., N. Allott, L. Arvola, M. Järvinen, E. Jennings, K. Moore, C. N. Aonghusa, D. Pierson, and E. Schneiderman (2010), Modelling the
impacts of climate change on dissolved organic carbon, in The Impact of Climate Change on European Lakes, pp. 221–252, Springer,
Netherlands.
Nash, J. E., and J. Sutcliffe (1970), River flow forecasting through conceptual models part I—A discussion of principles, J. Hydrol., 10(3),
282–290.
Oni, S., M. Futter, K. Bishop, S. Köhler, M. Ottosson-Löfvenius, and H. Laudon (2013), Long-term patterns in dissolved organic carbon, major
elements and trace metals in boreal headwater catchments: Trends, mechanisms and heterogeneity, Biogeosciences, 10(4), 2315–2330.
Oni, S., M. Futter, C. Teutschbein, and H. Laudon (2014), Cross-scale ensemble projections of dissolved organic carbon dynamics in boreal
forest streams, Clim. Dyn., 42(9–10), 2305–2321, doi:10.1007/s00382-014-2124-6.
Öquist, M., K. Bishop, A. Grelle, L. Klemedtsson, S. Köhler, H. Laudon, A. Lindroth, M. Ottosson Löfvenius, M. B. Wallin, and M. B. Nilsson (2014),
The full annual carbon balance of boreal forests is highly sensitive to precipitation, Environ. Sci. Technol. Lett., 1(7), 315–319.
Palviainen, M., L. Finér, A. Laurén, S. Launiainen, S. Piirainen, T. Mattsson, and M. Starr (2014), Nitrogen, phosphorus, carbon, and suspended
solids loads from forest clear-cutting and site preparation: Long-term paired catchment studies from eastern Finland, Ambio, 43(2),
218–233.
Pan, Y., R. A. Birdsey, J. Fang, R. Houghton, P. E. Kauppi, W. A. Kurz, O. L. Phillips, A. Shvidenko, S. L. Lewis, and J. G. Canadell (2011), A large and
persistent carbon sink in the world’s forests, Science, 333(6045), 988–993.
Piani, C., J. Haerter, and E. Coppola (2010), Statistical bias correction for daily precipitation in regional climate models over Europe, Theor.
Appl. Climatol., 99(1–2), 187–192.
Rankinen, K., T. Karvonen, and D. Butterfield (2004), A simple model for predicting soil temperature in snow-covered and seasonally frozen
soil: model description and testing, Hydrol. Earth Syst. Sci., 8(4), 706–716.
Rounsevell, M. D., and M. J. Metzger (2010), Developing qualitative scenario storylines for environmental change assessment, Wiley
Interdiscip. Rev. Clim. Change, 1(4), 606–619.
ONI ET AL.
FUTURE BOREAL DOC
2425
Journal of Geophysical Research: Biogeosciences
10.1002/2015JG003190
Saikku, L., and E. Asmala (2010), Eutrophication in the Baltic Sea, J. Ind. Ecol., 14(3), 482–495.
Sanders, L., and M. W. McBroom (2013), Stream water quality and quantity effects from select timber harvesting of a streamside
management zone, South. J. Appl. For., 37(1), 45–52.
Schelker, J., K. Eklöf, K. Bishop, and H. Laudon (2012), Effects of forestry operations on dissolved organic carbon concentrations and export in
boreal first-order streams, J. Geophys. Res., 117, G01011, doi:10.1029/2011JG001827.
Schelker, J., T. Grabs, K. Bishop, and H. Laudon (2013a), Drivers of increased organic carbon concentrations in stream water following forest
disturbance: Separating effects of changes in flow pathways and soil warming, J. Geophys. Res. Biogeosci., 118, 1814–1827, doi:10.1002/
2013JG002309.
Schelker, J., L. Kuglerova, K. Eklöf, K. Bishop, and H. Laudon (2013b), Hydrological effects of clear-cutting in a boreal forest–Snowpack
dynamics, snowmelt and streamflow responses, J. Hydrol., 484, 105–114.
Schelker, J., K. Öhman, S. Löfgren, and H. Laudon (2014), Scaling of increased dissolved organic carbon inputs by forest clear-cutting–What
arrives downstream?, J. Hydrol., 508, 299–306.
Seibert, J., T. Grabs, S. Köhler, H. Laudon, M. Winterdahl, and K. Bishop (2009), Linking soil-and stream-water chemistry based on a Riparian
Flow-Concentration Integration Model, Hydrol. Earth Syst. Sci., 13(12), 2287–2297.
Sørensen, R., M. Meili, L. Lambertsson, C. von Brömssen, and K. Bishop (2009a), The effects of forest harvest operations on mercury and
methylmercury in two boreal streams: Relatively small changes in the first two years prior to site preparation, AMBIO: A J. Human Environ.,
38(7), 364–372.
Sørensen, R., E. Ring, M. Meili, L. Högbom, J. Seibert, T. Grabs, H. Laudon, and K. Bishop (2009b), Forest harvest increases runoff most during
low flows in two boreal streams, AMBIO: A J. Human Environ., 38(7), 357–363.
Stepanauskas, R. n., N. O. JØrgensen, O. R. Eigaard, A. Žvikas, L. J. Tranvik, and L. Leonardson (2002), Summer inputs of riverine nutrients to
the Baltic Sea: bioavailability and eutrophication relevance, Ecol. Monogr., 72(4), 579–597.
Temnerud, J., J. K. Hytteborn, M. N. Futter, and S. J. Köhler (2014), Evaluating common drivers for color, iron and organic carbon in Swedish
watercourses, Ambio, 43(1), 30–44.
Tetzlaff, D., C. Soulsby, J. Buttle, R. Capell, S. Carey, H. Laudon, J. McDonnell, K. McGuire, S. Seibert, and J. Shanley (2013), Catchments on the
cusp? Structural and functional change in northern ecohydrology, Hydrol. Processes, 27(5), 766–774, doi:10.1002/hyp.9700.
Teutschbein, C., and J. Seibert (2012), Bias correction of regional climate model simulations for hydrological climate-change impact studies:
Review and evaluation of different methods, J. Hydrol., 456-457, 12–29.
Teutschbein, C., and J. Seibert (2013), Is bias correction of regional climate model (RCM) simulations possible for non-stationary conditions?,
Hydrol. Earth Syst. Sci., 17(12), 5061–5077.
Thorne, R. (2011), Uncertainty in the impacts of projected climate change on the hydrology of a subarctic environment: Liard River Basin,
Hydrol. Earth Syst. Sci., 15(5), 1483–1492.
Tipping, E., C. Woof, E. Rigg, A. Harrison, P. Ineson, K. Taylor, D. Benham, J. Poskitt, A. Rowland, and R. Bol (1999), Climatic influences on the
leaching of dissolved organic matter from upland UK moorland soils, investigated by a field manipulation experiment, Environ. Int., 25(1),
83–95.
Tiwari, T., H. Laudon, K. Beven, and A. M. Ågren (2014), Downstream changes in DOC: Inferring contributions in the face of model
uncertainties, Water Resour. Res., 50, 514–525, doi:10.1002/2013WR014275.
Tranvik, L. J., J. A. Downing, J. B. Cotner, S. A. Loiselle, R. G. Striegl, T. J. Ballatore, P. Dillon, K. Finlay, K. Fortino, and L. B. Knoll (2009), Lakes and
reservoirs as regulators of carbon cycling and climate, Limnol. Oceanogr., 54(6), 2298–2314.
Van der Linden, P., and J. F. B. Mitchell (2009), ENSEMBLE: Climate Change and its Impacts: Summary of Research and Results From the
ENSEMBLES Project, Met Office Hadley Centre, FitzRoy Road, Exeter EX1 3PB, U. K.
van Vuuren, D. P., and T. R. Carter (2014), Climate and socio-economic scenarios for climate change research and assessment: Reconciling the
new with the old, Clim. Change, 122(3), 415–429.
Wallin, M. B., T. Grabs, I. Buffam, H. Laudon, A. Ågren, M. G. Öquist, and K. Bishop (2013), Evasion of CO2 from streams–The dominant
component of the carbon export through the aquatic conduit in a boreal landscape, Global Change Biol., 19(3), 785–797.
Watts, G., R. W. Battarbee, J. P. Bloomfield, J. Crossman, A. Daccache, I. Durance, J. A. Elliott, G. Garner, J. Hannaford, and D. M. Hannah (2015),
Climate change and water in the UK–past changes and future prospects, Prog. Phys. Geogr., 39(1), 6–28.
Wikner, J., and A. Andersson (2012), Increased freshwater discharge shifts the trophic balance in the coastal zone of the northern Baltic Sea,
Global Change Biol., 18(8), 2509–2519.
Winterdahl, M., M. Futter, S. Köhler, H. Laudon, J. Seibert, and K. Bishop (2011), Riparian soil temperature modification of the relationship
between flow and dissolved organic carbon concentration in a boreal stream, Water Resour. Res., 47, W08532, doi:10.1029/2010WR010235.
Winterdahl, M., M. Erlandsson, M. N. Futter, G. A. Weyhenmeyer, and K. Bishop (2014), Intra-annual variability of organic carbon concentrations in running waters: Drivers along a climatic gradient, Global Biogeochem. Cycles, 28, 451–464, doi:10.1002/2013GB004770.
Woo, M.-K., R. Thorne, K. Szeto, and D. Yang (2008), Streamflow hydrology in the boreal region under the influences of climate and human
interference, Philos. Trans. R. Soc. B Biol. Sci., 363(1501), 2249–2258.
Wu, H., C. Peng, T. Moore, D. Hua, C. Li, Q. Zhu, M. Peichl, M. Arain, and Z. Guo (2014), Modeling dissolved organic carbon in temperate forest
soils: TRIPLEX-DOC model development and validation, Geosci. Model Dev., 7(3), 867–881.
ONI ET AL.
FUTURE BOREAL DOC
2426
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

Related manuals

Download PDF

advertisement