Introduction
The high latitudes are experiencing greater temperature increases than the
global average (AMAP, 2012; IPCC, 2013). Low decomposition
rates due to the cold environment have led to an accumulation of large
pools of carbon (C) in litter, soils and peatlands, much of which is currently
held in permafrost (Tarnocai et al., 2009; Koven et al., 2011; Hartley et
al., 2012). However, these C stores may be mineralized rapidly to the
atmosphere due to the warming effects on soil microbial activity and thereby
increase atmospheric concentrations of both carbon dioxide (CO2) and
methane (CH4). Meanwhile, temperature-induced vegetation changes may
mitigate those effects by photosynthetic enhancement, which is, however,
greatly influenced by disturbances such as plant–herbivore interactions as
well as soil water and nutrient contents (Jonasson and Michelsen, 1996;
Van Bogaert et al., 2009; Keuper et al., 2012). It is becoming crucial to
understand those aspects of vulnerable high-latitude ecosystems and their
responses to climate warming (Callaghan et al., 2013).
Ecosystems fix atmospheric C through photosynthesis and return this C back
through diverse paths and in different forms. In recent years, many field
measurements have been carried out in subarctic and arctic environments to
quantify C exchanges between the atmosphere and the biosphere. Those
measurements enable us to better understand possible feedbacks from
terrestrial biota and responses to the changing climate (Bäckstrand
et al., 2010; Christensen et al., 2012). However, some concerns about field
measurements in the subarctic/arctic environment have been raised and the
following research needs have been identified:
Complete year-round observations are generally missing, and many studies focus only on growing season measurements
(Grogan and Jonasson, 2006; Roulet et al., 2007; Christensen et al., 2012; McGuire et al., 2012). Year-around observations
are needed because there is clear evidence that C fluxes in the cold seasons are very important (Larsen et al., 2007b; Mastepanov et al., 2008).
Observations of interactions between terrestrial and aquatic systems are lacking (Lundin et al., 2013; Olefeldt et al., 2013).
Quantifications of terrestrial lateral loss of C are needed not only because they represent a significant fraction of net ecosystem
exchange (NEE) but also because they are intrinsically linked to downstream aquatic C cycling (Lundin et al., 2013).
Integrating terrestrial and aquatic C cycling is of high importance for our understanding of the C balance at the catchment
scale, particularly at high latitudes. Northern peatlands are large sources of dissolved organic carbon (DOC), while receiving
lakes are generally net sources of both CO2 and CH4 (Tranvik et al., 2009).
Many environmental characteristics of the Stordalen catchment, located in
the subarctic discontinuous permafrost region of northern Sweden, have been
measured since the 1950s (Bäckstrand, 2008) and many studies
covering a variety of disciplines are still ongoing (Callaghan et al.,
2010, 2013). Observations related to the dynamics of
almost all the relevant C components have been made in different land-cover
types (Christensen et al., 2012; Callaghan et al., 2013). However, the
various observations over different land-cover types have not yet been
integrated into a comprehensive, year-round catchment-level C budget. A
significant aspect of this area is that it contains permafrost that is
rapidly thawing (Åkerman and Johansson, 2008). This makes
more C hydrologically available (Olefeldt and Roulet, 2012), and the
large stock of C in these tundra soils becomes available to microbial
processes (Sjögersten and Wookey, 2002; Fox et al., 2008; Hartley et
al., 2012). A rapidly changing environment together with comprehensive
observations has established the unique importance of this area as a model
system for furthering our process-based understanding of the role of climate
changes in northern regions. Furthermore, this understanding, gained in an
accessible and highly instrumented area, can be applied to vast areas where
large C stocks exist but long-term measurements are lacking.
The Stordalen catchment contains several distinct land-cover types, including
tundra heath, birch forest with heath understory, peatlands and
lakes/rivers. Hereafter, the peatland is divided into three groups named
“palsa”, “Sphagnum site” and “Eriophorum site”, based on surface hydrology, permafrost
condition and characteristic plant communities. An earlier
compilation of the C balance of the larger Torneträsk catchment, which
encompasses the Stordalen catchment, indicated that there was a significant
sink capacity in the birch forest as well as across the peatland
(Christensen et al., 2007). This assessment, however,
lacked year-round measurement of CO2 and CH4 emissions and did
not include direct measurements of aquatic C fluxes
(Christensen et al., 2007). Subsequent observations in
the Stordalen catchment have focused on filling the missing components.
Consequently, recently updated year-round CO2 and CH4
measurements in the peatland identified the wetter non-permafrost
Eriophorum sites to be strong C sinks (-46 g C m-2 yr-1) with high CH4 emissions
(18–22 g C m-2 yr-1; Jackowicz-Korczyński et al., 2010; Christensen
et al., 2012), while measurements conducted on the drier palsa (where
permafrost is present) showed a relatively weaker uptake (-39.44 g C m-2 yr-1; Olefeldt et al., 2012). Total waterborne C exports
(DOC plus particulate organic C (POC) and dissolved inorganic C (DIC)) from
the terrestrial ecosystems (both peatland and forest) were also monitored (Olefeldt
et al., 2013) and found to represent a significant component of the net
ecosystem C balance, ranging from 2.77 to 7.31 g C m-2 yr-1. In contrast,
4 years of continuous eddy covariance (EC)-tower-based CO2
measurements in the birch forest revealed very variable C sink/source
functionality, which in 2 out of 4 years has been found to be a C
source to the atmosphere (Heliasz, 2012). Tundra heath, another important
land-cover type in this region, has lower C uptake (Christensen et al.,
2007; Fox et al., 2008) and both birch forest and tundra heath were found to
have high spatial heterogeneity (Fox et al., 2008; Heliasz, 2012).
Altogether, different land-cover types show diverse contributions to this
subarctic ecosystem's C balance. With pronounced future warming expected in
this region, the structure and function of the different vegetation types
are expected to vary dramatically as has been observed during warming in the
recent past (Callaghan et al., 2013). In addition, changes to
soil conditions due to warming and permafrost thaw will likely stimulate C
fluxes to the atmosphere and affect the long-term accumulated C (Wolf et
al., 2008a; McGuire et al., 2012), but this likely C release needs to be
weighed against the possibility of increased uptake by increased primary
productivity resulting from longer growing seasons and/or potential CO2
fertilization.
The field measurements described above provide an insight into the ongoing
processes and current ecosystem status, but until now, no modeling exercises
have been implemented in this region in combination with the comprehensive
measured data. Moreover, high-spatial-resolution predictions of future
potential dynamics of both vegetation and soil processes and their responses
to the projected climate are lacking in this region. In this study,
therefore, we aim to assess the Stordalen catchment C budget in a
retrospective as well as in a prognostic way by implementing a process-based
dynamic ecosystem model (Smith et al., 2001; Miller and Smith, 2012)
integrated with a distributed hydrology model (Tang et al., 2014a, b) at high spatial resolution (50 by 50 m; Yang et
al., 2012). We quantify the overall C budget of the study catchment by
synthesizing diverse C fluxes and specifically address the following
questions: (1) will this subarctic catchment become a C source or a larger C
sink in the near future? (2) How differently will the catchment's vegetation
microtypes respond to the climate drivers? (3) What are the major
limitations in the model's prognostic ability?
To answer these questions, we implemented an Arctic-enabled version of the
dynamic ecosystem model, LPJ-GUESS WHyMe (Wania et al., 2009a, b,
2010). This model has been widely and successfully implemented for
estimating and predicting ecosystem function in high-latitude regions
(McGuire et al., 2012; Miller and Smith, 2012; Zhang et al., 2013).
LPJ-GUESS WHyMe includes comprehensive process descriptions to capture the
interactions between atmosphere–vegetation–soil domains and it explicitly
describes permafrost and peatland processes, which are important components
of our study catchment. Importantly, previous studies (Tang et al.,
2014a, b) dedicated to integrating a distributed hydrology
scheme to LPJ-GUESS WHyMe have demonstrated the necessity of considering
lateral water movements to accurately capture water and C cycling in this
region. The model with distributed hydrology is called LPJG-WHyMe-TFM, where
TFM stands for “triangular form-based multiple-flow algorithm”
(Pilesjö and Hasan, 2014). The study presented here is the
first modeling exercise to combine all the available year-round measured
data in the Stordalen catchment and at high spatial resolution.
Model descriptions
The process-based dynamic ecosystem model LPJG-WHyMe-TFM was chosen as the
platform for studying the subarctic catchment C balance. The processes in
the model include vegetation growth, establishment and mortality,
disturbance, competition between plant individuals for light and soil water
(Smith et al., 2001, 2014), and soil biogeochemical
processes (Sitch et al., 2003). These
processes are operated in a number of independent and replicate patches.
Vegetation in the model is defined and grouped by plant function types
(PFTs), which are based on plant phenological and physiognomical features
combined with bioclimatic limits (Hickler et al., 2004; Wramneby et al.,
2008). Bioclimatic conditions determine which PFTs can potentially grow in
study regions, and vertical stand structure together with soil water
availability further influence PFT establishment based on PFTs' shade and
drought tolerance characteristics. A list of the simulated PFTs in this
catchment can be found in Table S1 in the Supplement. The parameterization and
choice of non-peatland PFTs in this study cover all the main vegetation
types found in the study region (boreal forest, shrubs, open-ground grasses) and are based on previous
studies for the arctic region using LPJ-GUESS (Wolf et al., 2008a;
Hickler et al., 2012; Miller and Smith, 2012; Zhang et al., 2013). For
peatland grid cells, two new peatland PFTs, flood-tolerant graminoids and
Sphagnum moss, are introduced by Wania et al. (2009b).
The model is driven by monthly/annual climate data and includes both
non-peatland and peatland hydrological processes (Fig. 1). The vertical
water movement between atmosphere, vegetation and soil is based on Gerten et al. (2004), while the lateral water movement between grid cells was
implemented by Tang et al. (2014b) based on topographical
variations. More recently, an advanced multiple-flow algorithm, TFM (Pilesjö and Hasan, 2014), was chosen to distribute water among
grid cells, due to its better treatment of flow continuity and flow
estimation over flat surfaces (Tang et al., 2014a). Within the
catchment boundary, surface and subsurface runoff can move from one upslope
cell to multiple downslope cells, which greatly improves hydrological flux
estimations (Tang et al., 2014a) and results in a better
estimate of C fluxes in peatland region. The soil temperature estimation is
driven by surface air temperature, and the Crank–Nicolson heat diffusion
algorithm (Crank and Nicolson, 1996) is implemented to calculate
the soil temperature profile on a daily time step (Wania et al.,
2009a). The C cycling descriptions in LPJG-WHyMe-TFM for peatland cells are
based on Wania's developments (Wania et al., 2009a, b, 2010), while
the non-peatland grid cell C cycling is kept the same as in LPJ-GUESS
(Smith et al., 2001; Sitch et al., 2003). The full hydrological processes
and peatland C cycling descriptions in LPJG-WHyMe-TFM can be found in Sect. S1 in the
Supplement. A brief summary of the most relevant C cycling processes in the
model will be presented below.
Schematic of carbon components and cycling in LPJG-WHyMe-TFM. The
solid-line box includes carbon cycling for non-peatland cell, whereas the
items inside the dashed-line box represent processes particular to the
peatland cells. Only DOC&DIC in red text is not explicitly presented in
the current model. T: air temperature; P: precipitation; R: radiation;
CO2: atmospheric CO2 concentration; Decom_CCH4: decomposed materials allocated to CCH4; Decom:
decomposition.
A modified Farquhar photosynthesis scheme (Haxeltine and Prentice, 1996;
Haxeltine et al., 1996; Sitch et al., 2003) is used to estimate gross
primary production (GPP), which is related to air temperature (T),
atmospheric CO2 concentration, absorbed photosynthetically active
radiation (PAR) and stomatal conductance. Part of the GPP is respired to the
atmosphere by maintenance and growth respiration (Ra), and the
remaining part is net primary production (NPP) for each PFT. The
reproduction costs are subtracted from NPP, and thereafter the remaining NPP
is allocated to different living tissues in accordance with a set of
PFT-specific allometric relationships (Smith et al.,
2001). Leaves, fine root biomass and root exudates are transferred to the
litter pool with a given turnover rate, and above-ground plant materials can
also provide inputs to the litter pool due to stochastic natural disturbance
events and mortality (Smith et al., 2001; Thonicke et al., 2001). The
majority (70 %) of the litter is respired as CO2 directly to
the atmosphere, with a fixed fraction entering into fast- and slow-turnover soil
organic pools (fSOM and sSOM; Sitch et al.,
2003). The overall decomposition rate in the model is strongly influenced by
soil temperature and moisture (Sitch et al.,
2003). Additionally, the model also estimates emissions of biogenic volatile
organic compounds (BVOC) per PFT (Arneth et al., 2007; Schurgers et al.,
2009). However, the modeled BVOC values are not compared in the current
study, because they only represent a very small fraction of the modeled NEE
and, additionally, there are insufficient growing season measurement data in
the study domain to evaluate model performance. The major C cycling pathways
can be found in the solid-line box area of Fig. 1. For peatland cells, an
extra potential C pool for methanogens (CCH4) has been added (see
dashed-line box area in Fig. 1) and mainly includes root exudates and
easily decomposed materials (Wania et al., 2010). The majority of CCH4
is located in the acrotelm layer (Sect. S1) and the oxidation and
production of CH4 together determine the net emission of CH4. In
the model, the oxidation of CH4 through methanotrophic bacteria is
turned into CO2, whereas the unoxidized CH4 can be released to
the atmosphere by plant transport, diffusion and ebullition (see the solid-line arrows in Fig. 1). The oxidation level is mainly determined by
the location of water table position (WTP) in the model (Wania et al.,
2010).
The biomass production in the current version of LPJ-GUESS WHyMe has no
representation of nitrogen (N) limitation and neither N fluxes nor C–N
interactions are included (Sitch et al., 2007). The latest
version of LPJ-GUESS does include N cycling and N limitation on plant
production (Smith et al., 2014), but this capability has
not yet been integrated with the customized arctic version of the model
adopted in the present paper. Moreover, processes determining concentrations
of DOC and DIC in soil water have not yet been explicitly described in the
model (Fig. 1). To cover the majority of dissolved C losses and gains in our
assessment of the catchment C budget, DOC is estimated by combining modeled
runoff with observed DOC concentrations. Measured DIC export data are used
directly, based on observations by Olefeldt et al. (2013).
Object-based vegetation classification map of the Stordalen
catchment (a) and schematic of land-cover types and features of the
approximate transect A–B in (b).
Results
The modeled C fluxes for birch forest, peatland and tundra heath are first
compared with the measured data for the period of observation. The
seasonality and magnitudes of the fluxes are evaluated and then used to
estimate the catchment-level C budget. The modeled long-term C dynamics
during the period 1913–2080 are then presented for different land-cover
types, and the catchment-level C budget considering all the available C components
is also assessed.
Evaluation of the carbon balance in the historical period
Peatland
Both the plant communities and hydrological conditions in the peatland
differ among the palsa, Sphagnum and Eriophorum sites. Notably, the measured NEE (see grey
bars in Fig. 3a, b) covers both dry hummock (palsa) and semi-wet (Sphagnum site)
vegetation (Olefeldt et al., 2012), whereas our model cannot
represent the dry conditions of the palsa. Therefore, the observed fluxes
over both palsa and Sphagnum sites were compared with the modeled fluxes at the
Sphagnum site. The C fluxes magnitudes (including both wintertime emission and
summertime uptake) are larger at the Eriophorum site when compared with the
palsa/Sphagnum site for both measured and modeled data. The modeled NEE at both
sites generally captures the seasonality and magnitude of measured NEE, from
being a strong sink (negative NEE) of CO2 during the summer (mainly
June–August) to being a wintertime CO2 source. However, the model is
unable to fully capture the C source/sink functionality in September at both
sites. Furthermore, the modeled winter respiration at the Eriophorum site is very
close to the observations, though the model overestimates the wintertime
emissions at the palsa/Sphagnum site. The mean annual cumulative NEE reveals that
the model estimations of C fluxes for both parts of the peatland are within
the observed ranges, though with around 20 % of underestimation in the
3-year-averaged annual uptake (-39.76 and -50.18 g C m-2 yr-1 for the modeled and observed fluxes at the palsa/Sphagnum site; -71.54
and -90.34 g C m-2 yr-1 for the modeled and observed fluxes
at the Eriophorum site). For the Eriophorum site, the 3-year mean growing season uptake of C
is underestimated by the model (Fig. 3d), which indicates that the modeled
photosynthetic rates may be too low, that summer respiration rates may be too
high, or both.
Monthly (left) and mean annual cumulative (right) CO2 NEE and
CH4 fluxes for the peatland. Positive values indicate ecosystem release
to the atmosphere and negative values indicate ecosystem uptake. Msd stands
for measured.
Two full years (2006 and 2007) of EC-tower-measured CH4 emissions were
used to evaluate modeled estimations, and three different pathways
of modeled CH4 are also presented (Fig. 3e, f). The seasonal
variability is well described by the model and the modeled cumulative
CH4 over the year shows accurate representation of CH4 emissions
when compared with the observations (20.23 and 22.57 g C m-2 yr-1 for the
measured and modeled, respectively). Specifically, the wintertime emissions
are slightly underestimated, but this underestimation is compensated for by the
overestimated summer emissions. Plant-mediated transport of CH4
dominates during the growing season, while the ebullition and diffusion
transport reach a maximum in August. Additionally, the plant-mediated
CH4 emission is the main pathway active during the late spring and
early autumn.
Birch forest
The modeled average leaf area index for the birch forest and understory vegetation is
around 1.4 and 0.3, respectively, which are consistent with observations
made in this area (Heliasz, 2012). Modeled and measured monthly and
cumulative NEE are compared for the years 2007–2010 in Fig. 4. From the
monthly NEE comparisons (Fig. 4a), we see that the model underestimates
ecosystem respired C both before and after the growing season. The maximum
photosynthesis-fixed C in July is lower than that measured for the year
2007–2009, but not for the year 2010. The comparisons in Fig. 4b clearly
show the cumulative discrepancies between the modeled NEE and the
observations. The years 2009 and 2010 become C sources over the year (with
the average annual NEE value of 26.77 g C m-2 yr-1), even though the
measured air temperature for the winter and spring is relatively lower than
for the years 2007 and 2008 (mean air temperature for all months apart from
June–September is -3.11, -3.56, -4.26 and -5.75 ∘C
for the years 2007, 2008, 2009 and 2010, respectively). The abrupt emergence
of strong respiration in 2009 and 2010 is not captured by the model.
Furthermore, a comparison of the measured and modeled cumulative NEE
reveals that the main discrepancy occurs for the winter fluxes. As seen in
Fig. 4b, the observed birch forest cumulative CO2 emissions from January to
May reaches 50 g C m-2, a value which exceeds the range of the model's
predictions. Furthermore, the observed CO2 fluxes in September indicate
a C source but the modeled fluxes are close to zero.
Monthly (a) and cumulative (b) NEE between the years 2007 and 2010
for the birch forest. Modeled NEE with spatial variability (red shadow) for
each year is shown.
The red-shaded areas in Fig. 4a reveal high spatial variations during the
growing season and the accumulated variations over the year (the average
monthly standard deviations of the 4 years; see Fig. 4b) demonstrate a
remarkable spatial variability of the model-estimated annual NEE.
Interestingly, the observed mean annual NEE in both 2007–2008 and 2009–2010
fall within the wide spatial variations indicated by the model.
Tundra heath
Around 29.8 % of the catchment with alpine terrain was covered by heaths
and dwarf shrubs during the reference period 2001–2012. The model predicts
that low-growing evergreen shrubs (PFT: LSE, e.g., Vaccinium vitis-idaea) currently dominate in
this area, with tall summer-green shrub (PFT: HSS, e.g., Salix spp.) dominant in
the future predictions (2051–2080; Fig. 5). Since there are no available
year-round observations of the C balance in this part of the Stordalen
catchment, a synthesis of published data from similar environments is used
to evaluate the model estimations (Table 1).
Tundra heath summer and whole year NEE comparisons between the
modeled and published data.
Tundra type
Arctic heath
Arctic heath
Arctic heath
Subarctic heath
LSE & HSS
LSE & HSS
LSE & HSS
LSE & HSS
Period
1997, 2000–2005,
2007, summer
2007, summer
2004, 13
1961–1990,
1991–2000,
2001–2012,
2051–2080,
summer
Jul–21 Aug, 40 days
summerc/whole year
summerc/whole year
summerc/whole year
summerc/whole year
Location
NE Greenland
NE Greenland
NE Greenland
Northern Sweden
Northern Sweden
Northern Sweden
Northern Sweden
Northern Sweden
Methodsa
EC
CH
CH
EC & CH
LPJG
LPJG
LPJG
LPJG
Cumulated NEE
-1.4 ∼ -23.3d
-22.5e
-18e
-38.2 ∼ -68.7f,b
-26.89/-4.31g
-41.83/-12.82g
-58.23/-31.38g
-67.93/-26.88g
a EC stands for eddy covariance method, CH stands for chamber method and
LPJG stands for model estimations with LPJG-WHyMe-TFM. b Values vary from
sparsely vegetated areas (less than 10 % cover) to fully covered
low-growing Empetrum areas. c The modeled values in summer
include data for June, July, August and September. LSE: low-growing
(< 50 cm) evergreen shrubs. HSS: tall (< 2 m) deciduous
shrubs. d Groendahl et al. (2007).
e Tagesson et al. (2010). f
Fox et al. (2008). g This study.
Four periods with averaged C uptake values from our model are presented (see
the last four columns in Table 1) and a clear increase in summer uptake can
be found with increased temperature and CO2 concentration. The
modeled, whole-year uptake also follows the increasing trend, except for
the period of 2051–2080. The model-estimated summer C uptake is much
stronger than the observations made at the high-Arctic heath of NE
Greenland, which is reasonable when considering the longer growing season in
our catchment. The study conducted in an alpine area in Abisko by Fox et al. (2008) shows a wide range of estimated NEE over 40 days during the
summertime of 2004. The wide range of NEE values are because three levels
of vegetation coverage were studied and the lowest uptake is from sparse
vegetation dominated by bare rock and gravels, which is similar to the
situation in our tundra heath sites (Fig. 2a). The modeled summer NEE
between the years of 2001 and 2012 integrates the whole summer season (longer
than 40 days), and falls within the observed ranges presented by Fox et al. (2008), with values slightly higher than the lowest observed
values.
Aquatic systems
Lateral waterborne C fluxes
The modeled DOC exports based on runoff estimations are compared with the
measured DOC fluxes. The model generally underestimated annual runoff (the
measured and modeled mean annual runoff for six points are 279.65 and
207.75 mm, respectively), but the modeled accuracy varies from point to
point as well as for different years (Tang et al., 2014a). The
average DOC export of the birch forest over 3 years across sampling
sites from 2007 to 2009 are 3.65, 2.77 and 2.33 g C m-2 yr-1 from the
observations and 3.09, 2.03 and 1.69 g C m-2 yr-1 from the model
estimations, respectively. The downward trend over the 3 years is
captured and is related to lower precipitation during the latter years. The
observed DOC export rates at the palsa/Sphagnum and Eriophorum sites in 2008 are directly used
to represent different types of peatland export level and the values are
3.35 and 7.55 g C m-2 yr-1 for the palsa/Sphagnum and Eriophorum sites, respectively
(Olefeldt and Roulet, 2012). DIC export is currently beyond the
scope of our model but nonetheless contributes to the whole C budget. We
used an averaged DIC export value of 1.22 g C m-2 yr-1 based on the
published data in Olefeldt et al. (2013) and DIC export is
included in the estimation of the catchment C budget below (see Table 2).
Summary of catchment C budget for different time periods. Both mean
and annual variations (1-standard-deviation value in the parentheses) of
each period are presented. Negative mean values indicate ecosystem carbon uptake,
while positive values indicate that mean ecosystem carbon is lost through
respiration or CH4 emission. The mean temperature (T, ∘C) of each period is
listed.
Periods
Birch
Eriop.
Palsa/Sphag.
Peatland
Tundra
Streams
Lakes
Streams
Lakes
Birch
Peatland
Birch
Eriop.
Palsa/Sphag.
C budget
CO2a
CO2a
CO2a
CH4
heath
CO2c
CO2c
CH4c
CH4c
DIC
DIC
DOC
DOC
DOC
1961–1990
-17.52
-36.83
-18.04
14.08
-4.31
2.52
-1.47
(T = -1.26)
(22.46)
(30.80)
(31.39)
(1.57)
(15.81)
(0.73)
(18.25)
2000–2005
-3.48
-32.85
-10.49
17.65
-1.63
3.25
7.96
(T = 0.13)
(34.10)
(28.91)
(27.36)
(2.21)
(21.77)
(0.56)
(27.00)
2006–2011
-56.68
-60.10
-37.26
18.60
-50.27
2.31
-38.71
(T = -0.19)
(44.26)
(25.98)
(24.58)
(1.27)
(27.28)
(0.67)
(34.55)
2051–2080
-24.56
-28.84
-8.43
22.94
-26.88
2.86
-11.23
(T = 0.90)
(39.05)
(32.25)
(27.67)
(3.71)
(32.22)
(0.40)
(33.11)
Measured
0.88e
-90.34f
-50.18g
20.23f,b
-3d,h
10.1 ± 4.4i
0.5 ± 0.2i
0.06i
0.1 ± 0.1i
1.22j
1.22j
3.17j
7.55k
3.35k
(years)
(2007–2010)
(2006–2008)
(2008–2009)
(2006–2007)
(2009–2011)
(2008–2011)
(2009–2011)
(2010)
(2007–2009)
(2007–2009)
(2008–2009)
(2008)
(2008)
a Fluxes minus DOC and DIC export.
b Observed CH4
fluxes for the Eriophorum site.
c Variables are
normalized to the catchment area. d Reference-estimated
value.e Heliasz (2012). f Jackowicz-Korczyński et al. (2010).
g Olefeldt et al. (2012). h Christensen et al. (2007).
i Lundin et al. (2015).
j Olefeldt et al. (2013). Eriop.: Eriophorum; k Olefeldt and Roulet (2012) Sphag.: Sphagnum.
Dominant PFTs (a, b) and their biomass (c, d) in the
study catchment, for the periods 1961–1990 (a, c) and 2051–2080
(b, d).
Carbon fluxes from lakes and streams
Investigations of aquatic system C emission in the Stordalen catchment were
conducted by Lundin et al. (2013) during the years 2008–2011.
Around 5 % of the total catchment area (0.75 km2) is classified as
aquatic systems, with lakes accounting for 96 % of the aquatic area. Both
lakes and streams contribute to the emissions of CH4 and CO2, but
the streams dominate CO2 emission while the lakes dominate CH4
emissions. Averaged across the catchment area, the measured annual CO2
and CH4 emissions from streams are 10.1 and 0.06 g C m-2 yr-1,
respectively, while the lake emitted 0.5 g C m-2 yr-1 as CO2 and 0.1 g C m-2 yr-1 as CH4 (Lundin et al., 2015). Since river
and lake processes are at present beyond the scope of the model, the
contribution of aquatic systems to the catchment C fluxes is purely based on
the observed data (see Table 2).
Modeled whole-catchment carbon balance
One of the benefits of using a dynamic vegetation model is that it allows us
to investigate the vegetation and C-budget responses to climate change. In
this section, annual variations of both climate variables (temperature and
precipitation) and different ecosystem C fluxes are presented for the whole
simulated period. In addition, the normalized catchment C budget is also
estimated.
Our simulation results suggest that the temperature increase (around
2 ∘C; see Fig. 6a–c) together with a CO2 increase to 639 ppm by
2080 could greatly increase the productivity of the birch forest (Figs. 6,
2a–c and 5) as well as peatland CH4 emissions (see Fig. 6a–c). The DOC export from birch sites also increases slightly over the
study period (Fig. 6, 3a–c). Besides leading to an increase in ecosystem
biomass in grid cells currently occupied by birch forest (Fig. 5c, d),
warming will also result in the current tundra heath close to the birch
treeline being replaced by an upward expansion of the birch forest (Fig. 5a, b). This is indicated by the changes to birch treeline's uppermost
elevation in Fig. 6, 8a–c. The dramatic increase in C uptake in the tundra
heath region (Fig. 6, 5a–c) is largely a result of this vegetation
succession. Meanwhile, the successive degradation of permafrost and slightly
higher annual precipitation may result in more anaerobic conditions for the
modeled peatland. Combined with warmer soil conditions, these result in
both higher decomposition rates of soil organic matter and greatly increased
CH4 production. Moreover, a stronger response of respiration than NPP
to the temperature increase reduces the net C uptake in the peatland (see
Fig. 6, 4a–c). The catchment, as a whole, shows an increased C uptake (see
the trend line in Fig. 6, 7a) over the 1913–2080 period. The averaged C
budgets for the two selected periods (1961–1990 and 2051–2080) are -1.47 and
-11.23 g C m-2 yr-1, respectively, with the increase being dominated by the
large increase seen in the uptake rates at the birch forest and tundra heath
sites.
Time series (a) of annual mean temperature, precipitation
sum, NEE (g C m2 yr-1), averaged DOC export (g C m2 yr-1) from the birch
forest, birch treeline elevation (m) and catchment C budget (g C m2 yr-1)
values between 1913 and 2080. The NEE data for the birch forest and the
peatland have had the corresponding DOC and DIC values subtracted. Here the
averaged C fluxes from the Stordalen peatland (the northeast side of the
catchment) are used to represent for the averaged C fluxes for the whole-catchment peatlands since only the Stordalen peatland DOC and DIC
observations are currently available. The aerial photo-based classification
of the Eriophorum and Palsa/Sphagnum peatland fractions within the Stordalen peatland is used
to scale up the C fluxes. The trend of each data set is shown with a red
dashed line. The second and third columns (b, c) of the figure focus on
the periods 1961–1990 and 2051–2080 (these two periods are also indicated in
the first column with shaded area). The numbers in bold in columns (b) and (c) show the annual average of each quantity for the respective time
period. The fractions of peatland, birch forest, tundra and lakes/rivers are
5.7, 57.69, 29.76 and 5 %, respectively. Approximately
1.79 % of the catchment area is estimated as being dominated by C3G and
HSE, which are not included in the above classification. The last row shows
the birch treeline elevation changes over the period.
During the 2051–2080 period, both boreal needleleaf shade-tolerant spruce
(PFT: BNE, e.g., Picea abies) and boreal needleleaf (but less shade-tolerant) pine
(PFT: BINE, e.g., Pinus sylvestris) start to appear in the birch-dominated regions (Fig. 5).
In the current tundra regions, the coverage of HSS greatly increases at the
expense of LSE. For the northern parts of the catchment, birch forest
densification is observed with future warming, while the greatest relative
changes in biomass occur near the treeline. The increased temperature
together with increased CO2 concentration by 2080 are very likely to
increase CO2 uptake in both tundra heath and forested areas, though the
nutrient limitations are not included in this version of the model (but see
Smith et al., 2014). A summary of the modeled catchment
C components and C fluxes from different sources is given in Table 2.
Additionally, the modeled estimations during the warm period of 2000–2005
are also presented in Table 2, and the positive mean value of catchment
C budget is seen to be an exception with reference to the main reference
period. Furthermore, the annual variations of the modeled C fluxes in
different periods are also presented in Table 2. We find that different
land-cover types could shift rapidly from being a C sink to a C source in the
future (see the mean plus 1 SD value).
To illustrate the impact of CO2 increases alone on the modeled C
budget, the differences between simulations (ΔC fluxes) with and
without a CO2 increase since 1960 are shown in Fig. 7. The original
outputs of those two simulations are plotted separately in
Fig. S1 in the Supplement, together with the statistical significance values (p).
Interestingly, the simulation with constant CO2 forcing after 1960
significantly reduces the birch and tundra uptake (positive values in Fig. 7), whereas the peatland NEE and CH4 are not strongly influenced by the
CO2 increase. The catchment C-budget dynamics are consistent with the
changes seen in the birch and tundra heath regions. Furthermore, the
magnitudes of ΔC fluxes for the birch and tundra NEE show an
increasing trend after 1971, which is also seen in the ΔC fluxes for
the annual GPP and respiration (see birch forest site in Fig. 7a). Since
tundra heath shows a similar trend to birch forest, it is not presented in
the figure. However, the relative differences of ΔC fluxes for the
annual GPP and ecosystem respiration widen over time, which indicates a
stronger response of GPP to the increased CO2 concentration than
ecosystem respiration.
Carbon flux differences (ΔC fluxes) for different
land-cover types with and without a CO2 increase since 1960. Positive
values of ΔC NEE imply a higher uptake or a lower emission of
CH4 in the simulations with a CO2 increase compared to the
simulation without a CO2 increase. For the birch forest land-cover type,
the differences in gross primary production (GPP) and ecosystem respiration
are shown in the panel (a), where the positive values indicate a higher
photosynthesis rate and a higher respiration rate in the simulations with a
CO2 increase compared to the simulations without a CO2 increase.
To account for the fact that CH4 is a more potent greenhouse gas than
CO2, an estimation of the global warming potential (GWP) of the
two simulations can be made. Assuming the relative climate impact of
CH4 is 28 times greater than CO2 over a 100-year period
(IPCC, 2013), the calculated GWP for the simulation with
atmospheric CO2 increase is 3 times larger in the period 1961–1990
(27.3 g CO2-eq) than the period 2051–2080 (8.8 g CO2-eq). However,
in the simulation without CO2 increase, the GWP in 1961–1990 (40.7 g CO2-eq) is approximately half of the GWP value for the 2051–2080 period
(93.3 g CO2-eq). This shows that the change in global warming potential
found in the model simulations is strongly influenced by the CO2
concentration used to force the model, as the CO2 trajectory can alter
the balance between the GWP changes resulting from C uptake in the birch
forest and tundra and the peatland CH4 emissions.
Discussion
To our knowledge, our model simulation is the first attempt to create a C
budget of a subarctic catchment based on a dynamic global vegetation model
applied at the local scale and to predict its evolution in response to a
changing climate. The C-budget estimations in this paper include major flux
components (CO2, CH4 and hydrological C fluxes) based on a
process-based approach. The magnitude and seasonality of modeled C fluxes
compare well with the measurements (Figs. 3, 4 and Tables 1 and 2), which
gives us confidence in the ability of our model to represent the main
processes influencing the C balance in this region. Our hope is that, by
using a process-based modeling approach at high spatial resolution, our
methodology will be more robust in estimating the C budget in a changing
future climate than other budget estimation methods (Christensen et al.,
2007; Worrall et al., 2007). In response to a climate warming scenario, our
model shows a general increase in the C sink in both birch forest and tundra
heath ecosystems, along with greater CH4 emissions in the catchment
(Fig. 6). Integrated over the catchment, our modeled C budget indicates
that the region will be a greater sink of C by 2080, though these estimates
are sensitive to the atmospheric CO2 concentration. Nevertheless, the
current model setup and simulations still contain limitations in both the
historical estimations and predictions. Below we discuss the model's current
performance as well as a few existing limitations, and further propose some
potential model developments. Most of the studies referred to below have
been conducted specifically in the area near the Stordalen catchment or in
other regions with similar environments.
Modeling carbon fluxes for the historical period
Peatland CO2 and CH4 emission
The model estimations in the peatland demonstrated skill in representing the
relative differences of CO2 fluxes between the palsa/Sphagnum site and
Eriophorum site. The current model version cannot represent the very dry conditions of
the palsa due to the acrotelm–catotelm soil structure (Wania et
al., 2009a); therefore, the overestimated emissions rates of CO2 (Fig. 3a, b) are expected when using the modeled fluxes from the Sphagnum site to compare
with the measured data covering both the palsa and Sphagnum sites (Larsen et al.,
2007a; Bäckstrand et al., 2010). The cold-season respiration at the
Sphagnum site can account for at least 22 % of the annual CO2 emissions
(Larsen et al., 2007a), which is a further reason for the
model overestimations. Moreover, at the Eriophorum site, the model slightly
underestimated summertime uptake, which was found to be related to modeled
summer respiration being higher than dark-chamber-measured respiration,
though with high spatial variations (Tang et al., 2014a).
The accurate representations of annual CH4 emissions at the
Eriophorum sites reflect the model's improved estimations of both hydrological
conditions and dynamic C inputs to the CCH4 pools (Fig. 1). With the
inclusion of the distributed hydrology at 50 m resolution (Tang et al.,
2014a, b), the lower-lying peatland can receive surface
runoff from the upslope regions, and the fact that water can rise to a depth
of 10 cm above the surface both creates anaerobic conditions in the model
and favors the establishment of flood-tolerant graminoid PFTs such as
Carex spp., which can transport CH4 to the atmosphere. Under such anaerobic
conditions, decomposition rates are restricted and part of the decomposed C
becomes CH4 (Wania et al., 2010). However, the overestimation of
CH4 emissions during summertime is at least partly likely to be linked
to the complexity of modeling the ebullition process (Wania et al.,
2010). Based on sensitivity testing shown in Wania et al. (2010), the
parameter that controls the CH4 / CO2 production ratio under anaerobic
conditions can strongly impact the ebullition process, so it is important to
determine this parameter more accurately in future studies. Nevertheless, it
remains the case that excluding the palsa type could result in a general
overestimation of CH4 emission over the whole peatland.
Wintertime carbon fluxes, thickness of organic layer and disturbance
in the birch forest
The high wintertime CO2 fluxes to the atmosphere observed in the birch
forest and the underestimations by our model (Fig. 4) highlight the
importance of the representation of winter season C fluxes for the birch
forest, particularly as winter temperatures are increasing more than summer
temperatures (Callaghan et al., 2010). In the model, the
respiration rate is linked to soil temperatures and moisture and the
underestimated respiration could be attributed to the lower temperature
estimations in the wintertime (covering all months except June–September) when
compared with the observed soil temperature at 5 and 10 cm depth in 2009 and
2010 (the modeled wintertime soil temperature is around 1.9 ∘C
lower than that observed at both depths). Also, the investigations of snow
depth insulation influence on soil temperatures and respiratory activity by
Grogan and Jonasson (2003) have shown that soil temperature
contributes to the greatest variations in respiratory activity at our birch
sites. The birch forest is located on the relatively lower parts of the
catchment (Fig. 2) and traps the wind-shifted lighter snow from the upland
tundra heath, creating a much thicker snow pack and therefore significantly
increasing the soil temperatures (Groendahl et al., 2007; Larsen et al.,
2007b; Luus et al., 2013). Snow depth measurements in the Abisko birch and
tundra heath sites from 24 March to 7 April 2009 (http://www.nabohome.org/cgi-bin/explore.pl?seq=131) revealed that the snowpack in the birch forest was 26.62 cm deeper on average
than the snowpack in the tundra heath. The snow density for the birch forest
was artificially decreased to 100 kg m-3 in the model in order to
increase the snow depth in the absence of a wind-redistribution mechanism in
our model, but it is still hard to capture the high soil temperatures as
well as high emission rates in winter.
The thickness of organic soil layer is another crucial component controlling
birch site respiration in our catchment (Sjögersten and Wookey,
2002; Heliasz, 2012). The thickness of organic soil is most likely connected
to the past transformation of the sites from heath to forest and is expected
to decrease if birch biomass continues to increase (Hartley et al., 2012, 2013). Furthermore, Hartley et al. (2012)
showed evidence of the decomposition of older soil organic matter during the
middle of the growing season and concluded that, with more productive forests
in the future, the soil stocks of C will become more labile. From a model
perspective, faster turnover rates of soil C pools (Sitch
et al., 2007) could be used in climate warming model experiments to reflect
the accelerated C mobilization in the soil. This could, to some degree,
offset the stronger C uptake in the birch forest site in the simulation
results shown here. In addition, the current parameterization of the organic
fraction in the birch forest may not fully represent the organic horizons in
real-world soil.
The birch forest is cyclically influenced by the outbreak of moths (E. autumnata), and an
outbreak in 2004 greatly affected the birch forest, resulting in a much
lower C uptake than average (Heliasz et al., 2011). Even
though it is assumed that the fluxes during 2007–2010 had returned to the
pre-defoliation levels (M. Heliasz, personal communication, 2013), it is still difficult to
completely exclude the influence of insect disturbance in this forest. For
the current model simulation, stochastic mortality and patch-destroying
disturbance events have been included to account for the impacts of these
random processes on ecosystem C cyclings. However, to give a more accurate
and reliable representation of the C fluxes in the birch forest, an explicit
representation of insect impacts, outbreaks and their periodicity should be
included (Wolf et al., 2008b), as well as other disturbances
(Callaghan et al., 2013; Bjerke et al., 2014). For example, a warm event
in winter 2007 caused a 26 % reduction in biomass over an area of over
1424 km2 in summer 2008 (Bokhorst et al., 2009).
Benefits of high spatial resolution and limitations from monthly
temporal resolution
Subarctic ecosystems are characterized by small-scale variations in
vegetation composition, hydrological conditions, nutrient characteristics
and C fluxes (Lukeno and Billings, 1985; McGuire et al., 2002; Callaghan
et al., 2013). The spatial resolution of 50 m in this model application
allowed us to capture the diverse vegetation microtypes and their
altitudinal gradient in the catchment, as well as their differential
responses to climate changes (Figs. 5 and 6). This is unlikely to be well
represented by a simple averaging approach across the landscape, e.g.,
regional or global climate models. Furthermore, it is worth noting that the C cycling in the
peatland, especially CH4 fluxes, is sensitive to the WTP estimations
(Tang et al., 2014a). Without spatially distinguished climate
and topographical data, it becomes impossible to implement our distributed
hydrological scheme and thereby capture the peatland WTP dynamics. To our
knowledge, the use of 50 m climate data as forcing of LPJG-WHyMe-TFM
(Yang et al., 2011) is among the most detailed and
comprehensive modeling exercises related to C cycling. Although this study
focuses on C cycling, the innovative projects' changes in vegetation
microtypes at high spatial resolution are relevant to local stakeholders
such as conservation managers and reindeer herders.
Nevertheless, the high-spatial-resolution data were produced with
coarser monthly temporal resolution, which could restrict the model's
ability to accurately estimate C fluxes at the start and end of the growing
season (Figs. 3 and 4). Due to the dramatic variations of day length at high
latitudes, a few days of misrepresenting the starting date of the growing
season could significantly alter the estimated plant C uptake (Heliasz,
2012). The daily variations are difficult to capture from the interpolated
quasi-daily values used in the model. Indeed, such variations strongly
highlight the need for climate forcing at a higher temporal resolution for
this region, due to the long daylight duration in summer and short growing
seasons.
Projection uncertainties
A temperature increase of 2 ∘C (Fig. 6) and elevated CO2
concentrations could greatly increase vegetation growth and thus the C sink
of the whole catchment. However, the densification and expansion of birch
forest as well as the increased presence of boreal spruce and pine PFTs in
our projected period (Fig. 5) could be strongly influenced by reindeer
grazing and herbivore outbreaks (Hedenås et al., 2011; Callaghan et
al., 2013), even though those detected changes are consistent with other
models simulations (Wolf et al., 2008a; Miller and Smith, 2012) and
general historical trends (Barnekow, 1999). Furthermore, climate
warming may favor the spread of insect herbivores, so an assessment of
ecosystem responses to future climate change cannot ignore these
disturbances (Wolf et al., 2008b) or other factors such as
winter warming events (Bjerke et al., 2014).
Temperature increase results in a larger extent of permafrost degradation in
the future. Meanwhile, the increased amount of available water from
precipitation and lateral inflow may increase the degree of anoxia and
further favors the flood-tolerant WetGRS PFT growth as well as CH4
production. However, the exact extent of wetting or drying of peatland in
the future is still highly uncertain, and the model prediction depends
strongly on the climate scenario, permafrost thawing, and the resulting
balance between increased water availability and increased
evapotranspiration. If the peatland drying is large enough, the reduced
degree of anoxia could reduce CH4 emissions in the future (Wrona et
al., 2006; Riley et al., 2011). In our case, the peatland, located at the
lower part of the catchment, receives water from the southern mountain
region, and is more likely to become wetter in the near future in response
to the increases in water availability (Wrona et al.,
2006). Additionally, the determination of grid cell peat fractions in our
simulation is dependent on the current (historical) soil map; therefore, the
projections of peatland expansion to non-peatland cells cannot be reflected
in the current model predictions. This could bring some additional
uncertainties into the catchment C-budget estimations (Malmer et al.,
2005; Marushchak et al., 2013).
To cover all the major C components in this catchment, the current
estimations of the C budget used available, observed aquatic emissions and
DOC concentration and DIC export values, all of which were assumed constant
for the whole simulation period, 1913–2080. However, the direction and
magnitude of changes to aquatic C fluxes are hard to quantify without
modeling additional processes in these systems. Previous studies have found
that the substantial thawing of permafrost as well as increased
precipitation in recent decades has significantly increased total organic
carbon (TOC) concentrations in lakes (Kokfelt et al., 2009; Karlsson et
al., 2010). The increased loadings of nutrients and sediments in lakes are
very likely to increase productivity of aquatic vegetation, but the effects
may be offset by the increased inputs from terrestrial organic C and
increases in respiration (Wrona et al., 2006; Karlsson et al., 2010).
Emissions of CH4 from aquatic systems are more likely to increase due
to the longer ice-free season (Callaghan et al., 2010), increased
methanogenesis in sediments and also increased CH4 transport by
vascular plants (Wrona et al., 2006). Furthermore, a
recent study by Wik et al. (2014) found that CH4 ebullition
from lakes is strongly related to heat fluxes into the lakes. Therefore,
future changes to energy fluxes together with lateral transports of
dissolved C from terrestrial ecosystems to the aquatic ecosystems are
especially important for predicting C emissions from aquatic systems.
As we have discussed, the dynamics of birch forest, and to a lesser extent
tundra heath C assimilation, largely determine the catchment's C budget
(Figs. 5 and 6), whereas the dramatic increase in CH4 can slightly
offset the net climate impact of the projected C uptake. Furthermore, both
modeled C budget and GWP values are very sensitive to the atmospheric
CO2 levels. However, to date, there is no clear evidence showing
significant long-term CO2 fertilization effects in the arctic region (Oechel et al., 1994; Gwynn-Jones et al., 1997; Olsrud et al., 2010). Two
possible reasons for this lack of CO2 fertilization response might be
that CO2 levels close to the moss surface in birch forest can reach as
high as 1140 ppm and are normally within the range of 400–450 ppm
(Sonesson et al., 1992), while tall vegetation such as shrubs and
birch trees as well as peatland species have not been manipulated by
CO2 fertilization. Over the long term, vegetation growth is likely a
result of complex interactions between nutrient supplies (McNown
and Sullivan, 2013), UV-B exposure (Schipperges and Gehrke, 1996; Johnson
et al., 2002), temperature and growing season length
(Heath et al., 2005) and forest longevity
(Bugmann and Bigler, 2011). Therefore, more field experiments are
urgently needed in order to quantify and understand the CO2
fertilization effects on the various vegetation microtypes of the subarctic
environment, and particularly tall vegetation types.
Overall, the current model application has been valuable in pointing to
these gaps in process understanding and meanwhile shows the importance of
including vegetation dynamics in studies of C balance. Furthermore, a
current inability to include the potential impacts of peatland expansion,
potential increases of emissions from aquatic systems and the
potential nutrient limitations on plants (but see Smith et al., 2014) and disturbances (Bjerke et al., 2014) makes it likely
that our projections of the catchment C budget and CO2 GWP will vary
from those that may be observed in the future. However, our high-spatial-resolution, process-based modeling in the subarctic catchment provides an
insight into the complexity of responses to climate change of a subarctic
ecosystem while simultaneously revealing some key uncertainties that ought
to be dealt with in future model development. These developments would be
aided by certain new observations and environmental manipulations,
particularly of CO2 with FACE experiments of shrubs and trees, in order
to improve our understanding and quantification of complex subarctic
processes.