Jukuri, open repository of the Natural Resources Institute Finland (Luke) 
All material supplied via Jukuri is protected by copyright and other intellectual property rights. Duplication 
or sale, in electronic or print form, of any part of the repository collections is prohibited. Making electronic 
or print copies of the material is permitted only for your own personal use or for educational purposes.  For 
other purposes, this article may be used in accordance with the publisher’s terms. There may be 
differences between this version and the publisher’s version. You are advised to cite the publisher’s 
version. 
This is an electronic reprint of the original article.  
This reprint may differ from the original in pagination and typographic detail. 
Author(s): Xiao Huang, Hanna Silvennoinen, Bjørn Kløve, Kristiina Regina, Tanka P. Kandel, 
Arndt Piayda, Sandhya Karki, Poul Erik Lærke & Mats Höglind 
Title: Modelling CO2 and CH4 emissions from drained peatlands with grass cultivation by 
the BASGRA-BGC model 
Year: 2021 
Version: Published version 
Copyright:  The Author(s) 2021 
Rights: CC BY-NC-ND 4.0 
Rights url: http://creativecommons.org/licenses/by-nc-nd/4.0/ 
Please cite the original version: 
Huang, X., Silvennoinen, H., Kløve, B., Regina, K., Kandel, T.P., Piayda, A., Karki, S., Lærke, 
P.E., Höglind, M. (2021). Modelling CO2 and CH4 emissions from drained peatlands with grass 
cultivation by the BASGRA-BGC model. Science of The Total Environment 765, 144385. 
https://doi.org/10.1016/j.scitotenv.2020.144385. 
Science of the Total Environment 765 (2021) 144385
Contents lists available at ScienceDirect
Science of the Total Environment
j ourna l homepage: www.e lsev ie r .com/ locate /sc i totenvModelling CO2 and CH4 emissions from drained peatlands with grass
cultivation by the BASGRA-BGC modelXiao Huang a,⁎, Hanna Silvennoinen b, Bjørn Kløve c, Kristiina Regina d, Tanka P. Kandel e, Arndt Piayda f,
Sandhya Karki g, Poul Erik Lærke h, Mats Höglind a
a Norwegian Institute of Bioeconomy Research, Klepp Station, Norway
b Norwegian Institute of Bioeconomy Research, Ås, Norway
c Water, Energy and Environmental Engineering Research Unit, University of Oulu, Oulu, Finland
d Bioeconomy and Environment Unit, Natural Resources Institute Finland, Jokioinen, Finland
e Noble Research Institute, LLC, Ardmore, USA
f Thünen Institute for Climate-Smart Agriculture, Braunschweig, Germany
g Delta Water Management Research Unit, USDA-ARS, Jonesboro, USA
h Department of Agroecology, Aarhus University, Interdisciplinary Centre for Climate Change, Tjele, DenmarkH I G H L I G H T S G R A P H I C A L A B S T R A C T• Dual-porosity framework is used for the
soil properties of drained peatlands.
• Model outputs represent the trade-offs
between CO2 and CH4 under
different WTLs.
• SOM decomposition rate could vary sig-
nificantly under the sameWTLs.
• This model can be used to optimize
water table management for drained
peatlands.⁎ Corresponding author.
E-mail address: xiao.huang@nibio.no (X. Huang).
https://doi.org/10.1016/j.scitotenv.2020.144385
0048-9697/© 2020 The Author(s). Published by Elsevier Ba b s t r a c ta r t i c l e i n f oArticle history:
Received 27 September 2020
Received in revised form 11 November 2020
Accepted 4 December 2020
Available online 25 December 2020
Editor: Jurgen Mahlknecht
Keywords:
Cultivated peatlands
Drainage
WTL
CO2
CH4
BASGRA-BGC modelCultivated peatlands under drainage practices contribute significant carbon losses from agricultural sector
in the Nordic countries. In this research, we developed the BASGRA-BGC model coupled with hydrological,
soil carbon decomposition and methane modules to simulate the dynamic of water table level (WTL), car-
bon dioxide (CO2) and methane (CH4) emissions for cultivated peatlands. The field measurements from
four experimental sites in Finland, Denmark and Norway were used to validate the predictive skills of
this novel model under different WTL management practices, climatic conditions and soil properties. Com-
pared with daily observations, the model performed well in terms of RMSE (Root Mean Square Error;
0.06–0.11 m, 1.22–2.43 gC/m2/day, and 0.002–0.330 kgC/ha/day for WTL, CO2 and CH4, respectively),
NRMSE (Normalized Root Mean Square Error; 10.3–18.3%, 13.0–18.6%, 15.3–21.9%) and Pearson's r (Pearson
correlation coefficient; 0.60–0.91, 0.76–0.88, 0.33–0.80). The daily/seasonal variabilities were therefore
captured and the aggregated results corresponded well with annual estimations. We further provided an
example on the model's potential use in improving the WTL management to mitigate CO2 and CH4 emis-
sions while maintaining grass production. At all study sites, the simulated WTLs and carbon decomposition
rates showed a significant negative correlation. Therefore, controlling WTL could effectively reduce carbon.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385losses. However, given the highly diverse carbon decomposition rates within individual WTLs, adding indi-
cators (e.g. soil moisture and peat quality) would improve our capacity to assess the effectiveness of specific
mitigation practices such as WTL control and rewetting.
© 2020 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).1. Introduction
Northern peatlands sequester carbon dioxide (CO2) effectively (Yu,
2012). With waterlogged and anaerobic conditions, they are also signif-
icant sources of methane (CH4) to the atmosphere (Harriss et al., 1985;
Smith et al., 2004). In the Nordic countries, peat soils are important for
agriculture, accounting for about 2.0% (Nielsen et al., 2013), 7.0%
(Kløve et al., 2010), 8.7% (Berglund and Berglund, 2010) and 10.4%
(Myllys et al., 2012) of the total agriculture area in Denmark, Norway,
Sweden and Finland, respectively. Cultivated peatlands in this region
are extensively drained for crop growth and grasslands of animal hus-
bandry in particular (Kasimir et al., 2018). However, drainage acceler-
ates the soil organic carbon (SOC) decomposition by introducing more
oxygen into the soil, thus shifting the peatlands from C sinks into signif-
icant sources of CO2 (Grønlund et al., 2008). Assessing greenhouse gas
(GHG) emissions from cultivation practices is important in the Nordic
countries as they make efforts to realize their goals for both improved
food security (Forbord and Vik, 2017) and mitigation of GHG emissions
(e.g. EU's 2030 climate & energy framework to cut at least 40% GHG
emissions). In the future, prolonged growing seasons in boreal regions
may provide opportunities for agricultural development (Wiréhn,
2018), while warmer and drier soil environment could further increase
SOC decomposition.With increase temperature and droughts peatlands
can secure grass production, but their carbon stocks will be more vul-
nerable. Due to the high emission rate, mitigating emissions from
drained peatland could be an effective option to tackle climate change
(Leifeld and Menichetti, 2018) compared to the C sequestration strate-
gies for mineral soils as suggested by the 4‰ initiative of the Lima
Paris Action Agenda (Chabbi et al., 2017). Consequently, finding
environment-friendly management practices for drained peatlands is
of great importance for the Nordic agriculture (Kløve et al., 2017).
Estimation of annual GHG emissions using data from field experi-
ments on cultivated peatlands (Berglund and Berglund, 2011; Kløve
et al., 2010; Maljanen et al., 2003; Regina and Alakukku, 2010) that
are usually short in duration and have infrequent sampling, may fail to
capture important feedbacks from management practices. Modelling
methods have been established to simulate the biogeochemical pro-
cesses of cultivated peatlands with higher spatio-temporal resolution
than those available from field experiments. The empirical regression
approaches (e.g. Kandel et al., 2017; Lloyd and Taylor, 1994), which di-
rectly build the relationship between GHG emission rate and certain
measured variables, are popular tools for interpolating and upscaling
the existing measurements (Eickenscheidt et al., 2015; Karki et al.,
2019; Lohila et al., 2003). Such data-driven methods, however, could
be limited in integrating complex environmental factors to predict opti-
mal management schemes.
On the other hand, process-based models describe the interactions
between plants and soil environment in more detail than the empirical
regression approaches. Thesemodels, including biogeochemical models
(Frolking et al., 2001; Kleinen et al., 2012; Mezbahuddin et al., 2016),
global vegetation models (Wania et al., 2009) and land surface models
(Qiu et al., 2018; Qiu et al., 2019; Shi et al., 2015), have been developed
to simulate the C dynamics of Northern peatlands. However, we identi-
fied three key challenges for current process-based models related to
the modelling of biogeochemical processes of the cultivated peatlands
with grass cultivation in the Nordic region: (i) While grass cultivation
accounts for the highest fraction of cultivated peatlands in this region,
most of the current models with specially developed peat soil module
(e.g. Orchidee-Peat, CLIMBER2-LPJ) mainly focus on forest ecosystems.2As the interaction between plants and soil largely controls the GHG
emissions and C balance, a proper module simulating the grass growth
and its winter survival in the Northern environment is needed; (ii) Un-
like pristine peatlands with relatively stable water table level (WTL),
cultivated peatlands exhibit obvious WTL variations due to drainage
and climate. A detailed hydrological module with higher temporal reso-
lution than seasonal or annual step is needed considering the specific
properties of drained peat soils; (iii) Lack of corresponding modelling
for different drainage and irrigation practices restricts model applica-
tions for exploring management effects on the GHG emissions.
To address these challenges, we developed the new model version
BASGRA-BGC (BASic GRAss model – BioGeochemical Cycle). The
model specifically simulates the C balance, including CO2 and CH4 emis-
sions, and biomass productivity, from drained peatlands with grass cul-
tivation. The original BASGRA model is a process-based model for
simulating the daily-step dynamics of leaves, roots, tillers and biomass
(Höglind et al., 2016) with detailed processes for cold hardening and
dehardening. The latestmodel version BASGRA_N also couplesmodules
of N supply from soil and N allocation among plant organs (Höglind
et al., 2020). BASGRA_N and its predecessor has been well validated
for grass growthmodelling in the Nordic region and used to investigate
different schemes to improve grassland management including
ideotype design and optimal fertilization (Hjelkrem et al., 2017;
Korhonen et al., 2018; Van Oijen and Höglind, 2016; Woodward et al.,
2020). We developed the BASGRA-BGC version based on BASGRA_N
by coupling modules from SWAT, Century and DNDC models.
The objectives of this paper are to: (i) describe the new features of
BASGRA-BGC which was developed to simulate WTL，CO2 and CH4
emissions from drained peatlands with grass cultivation; (ii) evaluate
its performance for WTL, CO2 and CH4 prediction against data from
field experiments in Finland, Denmark and Norway; (iii) use the
model to propose improved drainage schemes to mitigate GHG emis-
sions and maintain grass production.
2. Model development
BASGRA_N model, mainly focusing on the simulation of physiologi-
cal processes of the above-ground grass growth, represents soil physical
and biological processes in a rather simplisticway,which limits its capa-
bility to simulate the complex biogeochemical interactions characteriz-
ing cultivated peat soils. For example, in BASGRA_N the vertical soil
column is represented as one single layer, and the current version
does not simulate draining process and CH4 emissions. Therefore, the
new version BASGRA-BGC uses a multi-layer soil structure (user de-
fined) for the vertical soil column with layer-dependent simulation of
hydrological and biogeochemical processes, including CO2 and CH4
emissions. Detailed functions are described in the following Sections
2.1–2.3. The list of variable abbreviations used in this paper is provided
in Table 1.
2.1. Hydrological processes
Compared with mineral soils, peat soils have a higher SOC content
with lower bulk density and larger total porosity. The dominating
macro-scale pores in undecomposed peat can actively transmit water
to infiltration, evapotranspiration and drainage (Rezanezhad et al.,
2016). However, for degraded peat with long drainage history, the
plant debris is broken down into smaller fragments and a high propor-
tion of large pores is therefore turned into small and closed pores
Table 1
Explanation of variable abbreviations.
Name Description Unit Equation
State variable
SWm,i/SWim,i The soil water content of mobile/immobile part in the ith soil layer mm·H2O (1a)/(1b)
θm,i/θim,i The relative soil moisture in mobile/immobile part in the ith soil layer mm/mm (1c)
Tsoilz,j The soil temperature at depth z on the jth day of the year °C (5)
Tair The average annual air temperature °C (5)
Tsurf The surface temperature °C (5)a
CLitk (CPk) Litter carbon pool: k=1, labile pool; k=2, resistant pool g·C/m2 (9)
CSOMk (CPk) Soil organic carbon pool: k=1, very labile pool; k=2, labile pool; k=3, passive pool g·C/m2 (9)
NSH/NRT The nitrogen content in shoot/root g·N/m2 (11a)/(11b)a
CST/CLV/CRT/CSTUB The carbon content in stem/leaf/root/stubble g·C/m2 a
DOC The dissolved organic carbon in the soil layer g·C/m2 (15a)
H2 The hydrogen content in each soil layer g·H/m2 (15b)
anf The anaerobic fraction in each soil layer – (12)
FC The soil water content at field capacity in each soil layer mm·H2O (12)
SAT The saturated soil water content in each soil layer mm·H2O (12)
Non-state variables
Wmm,i/Wmim,i The water melting rate in mobile/immobile part in the ith soil layer mm·H2O/day (1a)/(1b)
Wfm,i/Wfim,i The water freezing rate in mobile/immobile part in the ith soil layer mm·H2O/day (1a)/(1b)
Ii The water infiltration rate in the ith soil layer mm·H2O/day (2)
Es,i The actual soil evaporation rate in the ith soil layer mm·H2O/day (3)
Et,i The actual transpiration rate in the ith soil layer mm·H2O/day (4)
Di The drainage rate in the ith soil layer mm·H2O/day (1a)
Exi The water flux from mobile part to immobile part in the ith soil layer mm·H2O/day (1c)
dzi The depth of the ith soil layer mm (1c)
SWexcess,i The infiltrative volume of water in the ith soil layer mm·H2O/day (S2a)
TT The travel time for infiltration hr (S2b)
Es,lower,i/Es,upper,i The cumulative evaporation rate until the lower/upper boundary of the ith soil layer mm·H2O/day (S3a)
Et,lower,i/Et,upper,i The cumulative transpiration rate until the lower/upper boundary of the ith soil layer mm·H2O/day (S4a)
df The depth factor – (S5a)
Dpot The potential draining rate mm·H2O/day (6)(7)(8)
m The height from water table to draining pipes mm (6)
t The height from water table to soil surface mm (7)
CFk The carbon flux leaving the kth carbon pool g·C /m2/day (9)
CFhr,k The heterotrophic flux from the decomposition of the kth carbon pool g·C/m2/day (10)
MRsh/MRrt The maintenance respiration rate in shoot/root g·C/m2/day (11a)
GRsh/GRrt The growth respiration rate in shoot/root g·C/m2/day a
andeck The anaerobic decomposition rate of the kth SOM carbon pool g·C/m2/day (13)
RTdec The root exudation rate in each soil layer g·C/m2/day (14)
CH4,p1/CH4,p2/CH4,p The CH4 production rate from DOC/H2/total g·C/m2/day (16a)/(16b)/(16c)
CH4,trans The CH4 transport rate g·C/m2/day (18)
θrt The relative soil moisture in the root zone – (19)
θrt,sat The relative saturated soil moisture in the root zone – (19)
Parameter
coefex The coefficient for water flux between mobile and immobile parts day−1 (1c)
coefl The lag coefficient that represent the influence of the previous day's temperature – (5)
Ke The effective lateral hydraulic conductivity mm·H2O/day (S6a)
de The corrected height from draining pipes to soil bottom mm (S6b)
g Dimensionless factor – (S7)
Dm The maximum open ditch drainage rate mm·H2O/day (8)
Ds Scaling coefficient for Dm – (8)
Ws Scaling coefficient for maximum soil moisture – (8)
θs Average soil porosity from water table to soil bottom – (8)
r0,k The maximum decomposition rate coefficient for the kth carbon pool yr−1 (9)
ftotal The combined decomposition scalar considering temperature, soil moisture and depth factors – (S9a)
rfk The respiration fractions of litter and SOM pools – (10)
rm,sh/rm,rt The base maintenance respiration rate in shoot/root g·C/g·N/day (11a)/(11b)
fm,t The temperature scalar for maintenance respiration – (S11)
ran,k The maximum anaerobic decomposition rate coefficient for the kth SOM carbon pool yr −1 (13)
ftotal,an The combined anaerobic decomposition scalar considering temperature and depth factors – (S13b)
rRT The maximum exudation rate coefficient day−1 (14)
CH4,DOC/CH4,H2 The maximum rate of CH4 production from DOC/H2 g·C/m2/day (S16c)
kmDOC/kmH2 The half saturation constant for DOC/H2
g·C/m3
g·H/m3
(16a)/(16b)
CH4,oxid,max The maximum oxidation rate of CH4 in each soil layer g·C/m2/day (S17a)
ftotal,oxid The combined oxidation scalar considering temperature and depth factors – (S17b)
fp/fe/fd The coefficient of plant transport/bubble ebullition/diffusion – (S18)
Kaer The coefficient of deficient aeration conditions – (19)
θair The anaerobiosis point of relative water moisture – (19)
Input
L The distance between draining pipes mm (6)(7)
b The height from soil surface to draining pipes mm (7)
r The radius of draining pipes mm (7)
a Calculated from the original BASGRA_N model.
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385
3
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385(Bragazza et al., 2009). As a result, we used a dual-porosity framework
that includes both “mobile zone”wherewater canmove easily and “im-
mobile zone” where water movement is negligible to simulate the hy-
drological processes in drained peat soils (Binet et al., 2013;
Rezanezhad et al., 2016). As shown in Fig. 1, the immobile zone can
only exchange water and solution with the mobile part. The overall
mass balance equations (one dimension) for soil water modelling are
as follows:
dSWm;i
dt
¼ Ii−1 þWmm;i−Ii−Es;i−Et;i−Di−Wfm;i þ Exi ð1aÞ
dSWim;i
dt
¼Wmim;i−Wf im;i−Exi ð1bÞ
Exi ¼ coef ex∙ θm;i−θim;i

 
∙dzi ð1cÞ
In the BASGRA-BGC model, we used the methods from the SWAT
(Soil &Water Assessment Tool)model (Arnold and Fohrer, 2005) to cal-
culate (i) soil water infiltration (Eq. (2)); (ii) the partition of total soil
evaporation and transpiration in each layer (Eqs. (3) & (4)); (iii) the
soil temperature as well as the thawing/freezing processes (Eq. (5)).
We present the key functions in the main text and refer to the supple-
mentary material for a full description of the detailed processes. Based
on Eq. (5), the soil temperature at the center of each soil layer is com-
puted and then linearly interpolated to get the soil temperature at any
depth within the whole soil column. We assumed that the correspond-
ing soil water in each soil layer is uniformly distributed and that the
water within soil depth with below-zero temperature gets frozen.
Infiltration rate:
Ii ¼ SWexcess;i∙ 1− exp
−24:0
TT
  
ð2Þ
Soil evaporation rate:
Es;i ¼ Es;lower;i−Es;upper;i ð3Þ
Transpiration rate:
Et;i ¼ Et;lower;i−Et;upper;i ð4ÞFig. 1. The dual-porosity framework for hydrological modelling in the BASGRA-BGC
model.
4Soil temperature:
Tsoilz; j ¼ coef l∙Tsoilz; j−1 þ 1:0−coef lð Þ∙ df ∙ Tair−Tsurf

 þ Tsurf
  ð5Þ
Subsurface drainage is a popular practice in the Nordic region to
lower the WTL of peatlands (Kløve et al., 2017). Tile drainage (here re-
fers to subsurface drainage using tile, PVC pipe and other materials) to-
gether with open ditch is a common way to extensively drain the
peatlandwhile open ditch drainage only without pipe is another option
which is less frequently used in this region. In BASGRA-BGC, the
Hooghoudt (1940) steady-state (Eq. (6)) and Kirkham (1957) tile
(Eq. (7)) equations are used to simulate the tile drainage flux with
WTL below and above the soil surface. For open ditch drainage, we
use the conceptual Arno model formulation (Franchini and Pacciani,
1991) to model the baseflow into the nearby ditch (Eq. (8)). After the
potential drainage flux has been calculated, the actual drainage rate Di
in each layer is computed from thewater table surface until the bottom
layer to meet the total drainage potential.
Tile drainage with water table below soil surface:
Dpot ¼ 8Kedemþ 4Kem
2
L2
ð6Þ
Tile drainage with water table above soil surface:
Dpot ¼ 4πKe∙ t þ b−rð ÞgL ð7Þ
Open ditch drainage:
Dpot ¼
DsDm
Wsθs
∙θ; θ≤Wsθs
DsDm
Wsθs
∙θþ Dm∙ 1−DsDmWs
 
∙
θ−Wsθs
θs−Wsθs
 2
; θ > Wsθs
8><
>: ð8Þ
2.2. Soil decomposition and plant respiration
The decomposition of litter material and soil organic matter (SOM) is
modelled using the Century-based cascade method (first-order decay
model) between different carbon pools (Parton, 1996).We define two lit-
ter pools and three SOM carbon pools (see Table 1). The detailed routines
of carbon transition among different soil and plant carbon pools are
shown in Fig. 2. The main functions used in this module are from CLM
(Common Land Model) version 5.0 (Lawrence et al., 2019). The dual-
porosity framework in the multi-layer soil modelling in Section 2.1 is
still applicable for decomposition processes. We do not explicitly label
the soil layer and pore region where the variable belongs to.
In BASGRA-BGC model, decomposition is modelled without consid-
ering nitrogen stress by assuming sufficient fertilizer input could signif-
icantly alleviate nitrogen limitation. Therefore, the carbon fluxes leaving
the upstream pools in each soil layer are calculated as:
CFk ¼
dCPk
dt
¼ CPk∙r0;k∙ f total ð9Þ
The soil respiration, as the CO2 emissions from the soil, aremodelled
as:
CFhr;k ¼ CFk∙rf k ð10Þ
Meanwhile, as the plant maintenance respiration has not been
modelled in the BASGRA_Nmodel,we add the simulation of plantmain-
tenance respiration rate as follows:
MRsh ¼ NSH∙rm;sh∙ f m;t ð11aÞ
MRrt ¼ NRT∙rm;rt ∙ f m;t ð11bÞ
Fig. 2. The pool structure, carbon transition and respiration in the decomposition module of the BASGRA-BGC model (see Table 1 for the explanation of abbreviations).
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 1443852.3. Methane modelling
In BASGRA-BGC, we follow DNDC model's routine for CH4 produc-
tion (Fumoto et al., 2008) and simplify parts of its functions to simulate
the CH4 production, oxidation and transport processes (see Fig. 3).
Hydrogen (H2) and dissolved organic carbon (DOC) released from root
exudation and SOM decomposition are used as electron donors in the
reductive reactions of CH4 production in hydrogenotrophic and
acetoclastic methanogenesis, respectively. Thereafter the produced
CH4 is assumed to be transported and emitted to the atmosphere
through three main pathways: (i) plants' vascular tissues; (ii) ebulli-
tion; (iii) diffusion. Both the atmospheric CH4 and that produced in
deeper soil layers are assumed to be consumed when passing theFig. 3. The schematic description of methane module in the BASGRA
5aerobic zone of soil matrix. The dual-porosity framework in the multi-
layer soil modelling in Section 2.1 is still applicable in this CH4 module.
A simple function based on soil water content is developed to deter-
mine the anaerobic fraction in each soil layer:
anf ¼
0:01; SW < FC
0:01þ 0:99∙ SW−FC
SAT−FC
; SW ≥FC
(
ð12Þ
The SOM anaerobic decomposition (assumed as Eq. (S13)) rate in
each soil layer modelled as:
andeck ¼ CSOMk∙ran;k∙ f total;an∙anf ð13Þ-BGC model (see Table 1 for the explanation of abbreviations).
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385The root exudation process in each soil layer is computed as:
RTdec ¼ CRT ∙rRT ð14Þ
As a result, the DOC and H2 generated in these two processes can be
calculated for soil layers. Meanwhile, DOC can be exchanged between
mobile and immobile pores and the exchange amount is proportional
to the water exchange volume
dDOC
dt
¼ RTdecþ
X3
k¼1
andeck∙ 1:0−rf kð Þð Þ ð15aÞ
dH2
dt
¼
X3
k¼1
andeck∙2:0=72:0ð Þ ð15bÞ
The production of CH4 includes two parts: (i) The reaction of DOC;
(ii) The reaction of H2 and CO2.
CH4;p1 ¼ CH4;DOC ∙ DOCkmDOC þ DOC ð16aÞFig. 4. The locations of four experim
6CH4;p2 ¼ CH4;H2 ∙
H2
kmH2 þ H2
ð16bÞ
CH4;p ¼ CH4;p1 þ CH4;p2 ð16cÞ
The CH4 oxidation is simulated considering the aerobic fraction in
each soil layer:
CH4;oxid ¼ CH4;oxid; max∙ 1:0−anfð Þ∙ f total;oxid ð17Þ
The CH4 transport is modelled using simple linear equation:
CH4;trans ¼ CH4;p∙ f p þ f e þ f d
 
ð18Þ
2.4. Water stress for deficient aeration conditions
Waterlogged conditions in the peat soils could result in the stress of
oxygen deficit for root activities and therefore limit grass growth. As
such stress is not explicitly modelled in BASGRA_N, we used the simple
linear curve as in AquaCropmodel (Raes et al., 2012) to simulate the ef-
fect of deficient aeration on grass transpiration in Eq. (19). Limitedent sites in the Nordic region.
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385transpiration due to oxygen deficit could further lower the photosyn-
thesis rate following the original procedure of the BASGRA_N model.
Kaer ¼
1:0; θrt < θair
θrt;sat−θrt
θrt;sat−θair
; θrt ≥θair
8<
: ð19Þ
3. Materials
3.1. Site description
We used observations from four experimental grassland sites on
peatlands distributed across the Nordic region to parameterize the
model and validate the model performance. The experimental sites
Jokioinen, Rovaniemi, Nørreå and Bodø were located in Southern
and Northern Finland, Denmark and Northern Norway, respectively
(see Fig. 4 for their locations), covering a broad range of climatic con-
ditions spanning from subarctic to temperate and soil properties
with a long history of drainage. The general information is listed in
Table 2. The main grass species are reed canary grass (PhalarisTable 2
The information of experimental sites.
Site Jokioinena,b Rovaniemia,b
Country Finland Finland
Coordinate 60°49′N, 23°30′E 66°35′N, 26°01′E
Annual precipitation (mm) 607 537
Average daily temperature (°C) 4.3 0.0
Period
1999.09.01-
2002.09.30
2000.05.01-
2002.06.30
Drainage tile drainage open ditch draina
Grass species
Phleum pratense and Festuca
pratensis
Phleum pratense a
pratensis
Peat depth (cm) 55 100
Soil bulk density (g/cm3) 0.51 (0–20 cm) 0.29 (0–20 cm)
Organic C (%) 24.0 (0–20 cm) 45 (0–20 cm)
Total N (%) 1.1 (0–20 cm) 2.5 (0–20 cm)
Porosity (%) 71 (0–20 cm) 91 (0–20 cm)
Saturated hydraulic conductivity
(mm/h)
12 (0–20 cm)
12 (0–20 cm)
3 (20–30 cm)
1 (30–40 cm)
a Regina et al., 2004.
b Regina et al., 2007.
c Karki et al., 2019.
d Kløve et al., 2010.
7arundinacea) and poa (Poa spp.) at Nørreå but timothy (Phleum
pratense) at other sites. The experiments at the sites Nørreå (Karki
et al., 2019) and Bodø (Kløve et al., 2010), included multiple plots
with different management treatments. Here we used the measure-
ments from the control treatment (poorly drained by ditch) at Nørreå
and the pipe drained plot at Bodø (other treatments at these two sites
are not for drainage practices) for daily-level model validation.
Meanwhile, we also used the annual estimations of CH4 emissions
from flooded treatment at Nørreå and the natural plot at Bodø. The
water table fluctuation was monitored continuously during the ex-
perimental period by the pressure sensors installed in a perforated
PVC tube (Nørreå) or groundwater wells (Bodø) and then the daily
averages were used in this study. In Finnish experiments, WTLs
were measured periodically in perforated plastic dipwells around
the field plots. At all four sites, ER (ecosystem respiration; CO2) and
CH4 emissions were measured using manual chambers. GPP (gross
primary production) measurements were only available in Nørreå.
These measurements were carried out periodically at intervals that
varied by 7–21 days over the season depending on environmental
conditions and timing of management practices (e.g. fertilization,
harvest).Nørreåc Bodød
Denmark Norway
56°27′N,
9°40′E
67°17′N, 14°28′E
650 1055
7.9 4.3
2015.01.01–2017.03.31
2003.08.09-
2004.11.30
ge
open ditch drainage (control
treatment)
tile drainage (P treatment)
nd Festuca
Festulolium and Tall fescue
Phleum pretense and Elytrigia
repens
83 64
0.33 (0–18 cm);
0.31 (18–45 cm);
0.18 (45–57 cm);
0.15 (57–83 cm);
0.23 (0–24 cm);
0.19 (24–42 cm);
0.15 (42–64 cm)
38.6 (0–18 cm);
39.7 (18–45 cm);
45.0 (45–57 cm);
46.9 (57–83 cm);
42.0 (0–24 cm);
44.7 (24–42 cm);
46.6 (42–64 cm);
3.3 (0–18 cm);
3.3 (18–45 cm);
3.1 (45–57 cm);
2.8 (57–83 cm);
2.4 (0–64 cm)
83 (0–18 cm);
83 (18–45 cm);
87 (45–57 cm);
91 (57–83 cm);
83 (0–24 cm);
86 (24–42 cm);
89 (42–64 cm);
6.4 (0–18 cm);
2.1 (18–45 cm);
1.8 (45–57 cm);
1.4 (57–83 cm);
12 (0–24 cm);
14 (42–64 cm);
Table 3
The parameterization scheme of the BASGRA-BGC model for the four sites.
Parameter Unit Fixed
parameter
(Param1)
Site-specific parameter (Param2)
Jokioinen Rovaniemi Nørreå Bodø
Parameters in the main text
coefex day−1 0.10
coefl – 0.70
Dm mm·H2O/day – – 1.0 0.6 –
Ds – – – 0.38 0.38 –
Ws – – – 0.45 0.45 –
r0,1 (Lit1) yr−1 4.5
r0,2 (Lit2) yr−1 1.4
r0,3 (SOM1) yr−1 1.0 0.6 0.7 1.0 0.7
r0,4 (SOM2) yr−1 0.027 0.03 0.02 0.04 0.02
r0,5 (SOM3) yr−1 0.0004
rfk (k=1,
…,5)
– 0.55, 0.45, 0.26, 0.54, 0.88
rm,sh g·C/g·N/day 0.252 0.25 0.25 0.30 0.25
rm,rt g·C/g·N/day 0.218 0.20 0.20 0.25 0.20
ran,k yr −1 0.15 0.15 0.16 0.30 0.16
rRT day−1 0.001 0.001 0.001 0.002 0.001
kmDOC g·C/m3 61.44
kmH 2 g·H/m
3 0.0266
θair – 0.85 0.75 0.75 0.90 0.75
Parameters in the supplementary materials
βw – 0.5
coefp – 0.7
Q10,dec – 2.0 2.1 2.6 2.1 2.0
zτ m 0.5
θ1 – 0.04
θ2 – 0.40
Q10,m – 2.0 2.2 2.0 1.5 2.0
Q10,P – 3.0
CH4,30 g·C/g·soil/day 8.5E-5
CH4,oxid,0 g·C/m3/day 0.0008
fp,max – 0.8
fe,max – 0.4
fd,max – 0.1
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 1443853.2. Model forcing data
Daily maximum/minimum temperature, global radiation, precipita-
tion, wind speed and relative humidity are required climatic forcing for
model running. We used the in-situ measurements for all these vari-
ables at Jokioinen and Bodø sites during the simulating period. For the
Rovaniemi site, there were no in-situ records for global radiation. In-
stead, we used the corresponding measurements from the nearest sta-
tion (Sodankylä Tähtelä) from the open dataset of the Finnish
Meteorological Institute (Weather Observation). For Nørreå site, all
the climatic forces are from the national weather station located ca.
5 km from the study site. Other site-specific information needed to
drive the model, including peat soil properties (carbon content, poros-
ity, field capacity, wilting point and hydraulic conductivity), manage-
ment records (harvest date, fertilizer amount) and drainage-related
parameters (e.g. the Input in Table 1) were also collected. Soil profile
depth (upper peat layer+ lower soil layer) was set at 1.5m and divided
into over 15 soil layers according to the detailed soil data (see Table S1).
As themeasuredWTL could drop below the peat soil layer, we assumed
the soil texture below the peat soil as loam or sandy soil with low hy-
draulic conductivity as it is the most common condition for cultivated
peatlands.
3.3. Field observations of WTL, CO2 and CH4
Wevalidated the daily-step outputs of the BASGRA-BGCmodel with
field measurements includingWTL, ER and CH4 emissions. The primary
WTL data was used directly without post-processing. At each site, there
were 2–3 replicated plots for the same treatment and 2–3 replicated
samplings of ER and CH4 for each plot and time point. We computed
the average emissions per site, treatment and time point and used
those for comparison with the corresponding simulated values. We
did not interpolate the discontinuous ER and CH4 flux measurements
into daily step to avoid introducing uncertainty with interpolating
methods. Instead, we compared the measured values with model out-
puts on the corresponding measuring day. The ER and CH4 emissions
were usually measured atmid-day for a few hours, andmay thus signif-
icantly differ from daily averages. Therefore, we corrected measured
values of ER to the daily average using themodified van't Hoff equation
(Davidson et al., 2006):
Fave ¼ Fm∙Q10 Tave−Tmð Þ=10:0ð Þ ð20Þ
where Fave and Fm are the emission rates at daily average temperature
Tave andmaximum temperature Tm;Q10 is the scaling parameter.We as-
sumed that the emission measurements were obtained at the daily
maximum temperature and thus used the daily average temperature
to correct it with Q10 = 2.0 (Petersen et al., 2012). We kept the primary
CH4measurements as its emissions are influenced bymore complicated
environmental factors.
3.4. Model setup and parameterization
A spin-up running for drained peatlands could bring significant uncer-
tainty to the soil carbon balance as detailed drainage history is usually un-
available. Therefore, for the four sites in this research, we directly ran the
model during the experimental period (see Table 2, period). The initial C
stocks in CSOM1, CSOM2 and CSOM3 pools accounted for 3, 60 and 37% of
the total organic carbon, respectively (see Table S2). As the four sites had
been drained for decades, the proportions of immobile and mobile pores
in the total porosity were set equally to 50% and 50% (see Table S2). We
used the WTL measurement closest to the first running day and then
set the initial soil water content in the layers below thewater table as sat-
urated and the ones in layers above water table at field capacity. The ini-
tial soil temperature across the soil columnwas set at the air temperature
at the first day. Besides, we used two kinds of parameterization schemes8(see Table 3): (i) Param1: a set of fixed parameters with commonly used
values in other models for all the four sites; and (2) Param2: using site-
specific values for part of parameters to account for the difference in
peat quality that affects the SOM decomposition potential and grass spe-
cies that have different tolerances to oxygen deficit and yield potentials.
We manually adjusted the parameter values to make the daily observa-
tions and simulations better fit. All parameter values and their relevant
sources are presented in Table S2.
3.5. Model evaluation
To evaluate the model, we compared the simulated daily WTL, ER
and CH4 emissions with measurements. We used three indicators
RMSE (RootMean Square Error),NRMSE (Normalized RootMean Square
Error) and Pearson's r (Pearson correlation coefficient) to quantify the
discrepancy and correlation between simulations and observations for
WTL, ER and CH4. In addition, we aggregated the model daily outputs
into annual emission factors and validated the C balance using the esti-
mations from relevant studies for the same sites (labelled in Table 5), in-
cluding annual GPP, NEE (net ecosystem exchange), SR (soil
respiration), ER and grass yield.We also compared the simulated annual
emission factors of CH4 at the four sites with estimations from previous
studies (labelled in Table 6) as an additional quality assessment.
4. Results
4.1. Comparison between two parameterization schemes
The model outputs using two parameterizations schemes were
compared with daily observations and the values of indictors were
Table 4
The evaluation of the daily simulation of the BASGRA-BGC model for the four sites.
Jokioinen Rovaniemi Nørreå Bodø
Param1 Param2 Param1 Param2 Param1 Param2 Param1 Param2
WTL (water table level)
RMSE (m) 0.06 0.06 0.10 0.10 0.105 0.103 0.11 0.11
NRMSE (%) 15.2 15.1 17.8 17.8 18.7 18.3 14.5 14.5
Pearson's r 0.58 0.60 0.91 0.91 0.62 0.63 0.79 0.79
ER (ecosystem respiration)
RMSE (gC/m2/day) 1.50 1.47 1.42 1.22 3.06 2.43 1.44 1.28
NRMSE (%) 13.2 13.0 17.8 15.3 18.3 14.5 21.0 18.6
Pearson's r 0.87 0.88 0.80 0.83 0.83 0.87 0.75 0.76
CH4
RMSE (kgC/ha/day) 0.0022 0.002 0.092 0.085 0.293 0.217 0.36 0.33
NRMSE (%) 19.6 19.0 16.5 15.3 25.1 18.6 24.9 21.9
Pearson's r 0.46 0.47 0.31 0.36 0.25 0.33 0.73 0.80
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385shown in Table 4. The difference between the two approaches for
WTL simulation is negligible as we mainly modified the biological-
related parameters. However, simulations for ER and CH4 were effec-
tively improved by Param2 by using site-specific values to account
for the heterogeneity in peat quality, grass species, as well as the un-
certainty in model structure. We also presented the comparison of
these two parameterization schemes for daily average WTL, SR, PR
(plant respiration), GPP and CH4 emissions in Fig. S1. Due to the
higher accuracy of Param2, we present results from Param2 for fur-
ther analysis in 4.2–4.5.
4.2. Simulation of WTL dynamics
We presented observed daily precipitation, temperature, and WTL
together with simulated WTL at Jokioinen, Rovaniemi, Nørreå and
Bodø in Fig. 5. At Jokioinen (see Fig. 5a), tile drainage was important
in maintaining the simulated and observed WTLs at the level of ~
−0.8mdepth.Heavy rainfall eventsfirstly supplemented thewater def-
icit in the upper soil layer and therefore could not immediately raise
WTL. As the site was drained down to the underlying silty loam layer
with relatively slower hydraulic conductivity than that of the peat
layer, the drainage potential was limited also during normal climatic
conditions. At the temperate Nørreå site (see Fig. 5c) with compara-
tively warm temperature in winter, the simulated and observed WTL
dynamics were thus mainly controlled by the precipitation and evapo-
transpiration patterns. TheWTL approached soil surface due to constant
precipitation input and low drainage rate of the open ditch, although it
dropped occasionally to lower level with high evapotranspiration de-
mand and periodic rain-free conditions. TheWTL observations and sim-
ulations at the two northernmost sites, Rovaniemi and Bodø, had the
largest seasonal variability. At the Rovaniemi site (see Fig. 5b) with an
open ditch, the soil freezing contributed significantly to the decline of
WTL and it was often seen in cold climate. At the Bodø site (see
Fig. 5d), the WTL rose to a high level (~−0.15 m) immediately after
an intensive precipitation in 2003 and dropped close to the depth of
the drainage pipe (~−0.95m) during a severe summer drought in 2004.
4.3. Simulation of CO2 emissions
The simulated ER and SR in daily step are presented in Fig. 6 and
the simulated GPP and NEE dynamics in Fig. 7. We also compared the
simulated and measured GPP at Nørreå in Fig. S2. The simulated ER
series generally captured the temporal pattern and magnitude and
the simulated SR followed the dynamic of daily temperature except
when the WTL was too high (see Fig. 5c, 2016.07–2016.08). The av-
erage simulated SR at northernmost site Rovaniemi (0.9 gC/m2/
day) was lower than at Bodø (1.3 gC/m2/day) and Jokioinen (2.19gC/m2/day). The PR simulations (plant respiration; the difference
between ER and SR) were mainly affected by the grass growth as
PR increased rapidly with grass growth and abruptly dropped after
harvest. In the winter, simulated ER and SR generally equaled as PR
approximated zero due to lack of green biomass. The daily variation
of PR (as well as ER) was greater than that of SR as the radiation force
in the Nordic region has a strong daily-level variation due to frequent
cloudiness, which significantly affects themodelling of photosynthe-
sis and growth respiration.
We aggregated daily outputs into yearly values and compared the C
balance at Jokioinen, Nørreå and Bodø with reported values (see
Table 5). The simulated annual SR was at the rate of 500–700 gC/m2/
yr in these sites (the simulated annual SR at Rovaniemi was 330 gC/
m2/yr, but no estimation was found for this site). However, due to the
differences in the grass species and soil nutrient condition, the simu-
lated annual GPP in Nørreå (reed canary grass with high biomass) was
about 40–100% higher than in Jokioinen and Bodø (timothy with low
biomass). As a result, the grassland in Nørreå was a significant carbon
sink with simulated annual NEE over 500 gC/m2/yr and still remained
carbon neutral with exported biomass C considered. Jokioinen and
Bodø were net carbon sources based on simulated NEE due to the high
decomposition rate.
4.4. Simulation of CH4 emissions
The daily model outputs of CH4 emissions at the four sites demon-
strated that the CH4 simulations correspond well with the measure-
ments, as well as with the WTL dynamics (Fig. 8). However, in
Fig. 8b-d, model outputs failed to capture the emission peaks at
these sites. Comparisons of annual emissions for the ‘flooded’ treat-
ment (WTL constantly at−0.03 m) in Nørreå and the ‘natural’ condi-
tion (tile drainage cancelled) in Bodø are presented in Table 6
alongside with the drainage modelling. This comparison proved
good performance of the BASGRA-BGC model in modelling the an-
nual budget of CH4 dynamics under different managements. At
Jokioinen and Bodø with tile drainage, CH4 emissions were trivial
and could even be a sink for atmospheric CH4. But as the result of
Bodø in ‘natural’ condition shows, the peatlands could still be a sig-
nificant CH4 source once drainage is cancelled and WTL raised. At
Nørreå with higher WTL and reed canary grass (higher biomass) cul-
tivation, both the ‘control’ and ‘flooded’ treatments emitted larger
amounts of CH4 compared to other sites. We attributed the high
emissions at this site (especially under ‘flooded’ treatment) to the
rapid decomposition of grass stubble, to root exudation into the
SOM in comparatively higher temperatures and to the high nutrient
availability. However, the BASGRA-BGC model still underestimated
the annual emissions from the Nørreå site during 2016.03–2017.03.
Fig. 5. The climatic conditions and hydrological simulations in the four sites. [WTL: above the soil surface (positive); below the soil surface (negative)].
X
.H
uang,H
.Silvennoinen,B.K
løve
etal.
Science
ofthe
TotalEnvironm
ent765
(2021)
144385
10
Fig. 6. The simulations of peat decomposition and grass respiration in the four sites. (average and std.: themean and standard deviation ofmultiple replicatedmeasurements. The positive
value of CO2 flux means the CO2 emitted from the soil-plant system into the atmosphere and vice versa.)
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 1443854.5. Model application to predict CO2 and CH4 emissions and yield as
affected by WTL
We used the predictions in Jokioinen during 2000.10–2002.09 to
show the potential of the BASGRA-BGC model to provide guidance in
improving theWTLmanagement. In this case, we assumed that subsur-
face irrigation could be applied to rise the average WTL to different
levels, under which we predicted the corresponding annual grass yield
and CO2 and CH4 emissions (see Fig. 9). We used the 100-yr global
warming potential value 34 (Myhre et al., 2013) to convert CH4 emis-
sions into the CO2 equivalents. In regime I of Fig. 9, grass yield couldFig. 7. The simulation of daily GPP, ER and NEE at the four sites. (The positive value of CO2 flu
11be maintained and ~200 gCO2-C m−2 yr−1 emission reduction could
be achieved with WTL rising from −0.8 to −0.4 m. When the WTL
was between−0.8mand−0.6m, the emission reduction is not obvious
as the water table still remained in the silty loam layer. Emission de-
crease rate accelerated when the WTL was between −0.6 m and
−0.4mandmore of the SOC stockwas under anaerobic condition. In re-
gime II, the emissionswere still reduced by 40 gCO2-Cm−2 yr−1, but the
grass yield decreased simultaneously. The emission reduction rate de-
creased due to the balance between decreased CO2 and increased CH4.
Within this regime, the peatlands were predicted as C sinks with nega-
tive CO2 equivalent. It can be explained by the lower SOC content in thex means the CO2 emitted from the soil-plant system into the atmosphere and vice versa.)
Table 5
The validation of carbon balance (positive value: carbon lost to the environment; negative value: carbon absorbed from the environment).
Jokioinen Nørreå Bodø
Period
2001.10–2002.09 (annual); 2001.10–2002.03 (winter),
2002.04–2002.09 (summer);
2015.03–2016.02 (Year 1);
2016.03–2017.02 (Year 2);
2003.08.20–2004.11.02
Emission factor of ER (ecosystem
resipiration; gC/m2/yr)
Simulated:
1039 (annual)
Literaturec:
1496 ± 22% (Year 1); 1490 ± 14%
(Year 2);
Literaturee:
1185–1236
Simulated:
1514 (Year 1);
1050 (Year 2);
Simulated:
1167
Emission factor of SR (soil respiration;
gC/m2/yr)
Literaturea:
573 ± 245 (annual);
125 ± 71 (winter);
447 ± 145 (summer);
Simulated:
589 (Year 1);
353 (Year 2);
Literaturee:
578–629
Simulated:
659 (annual);
170 (winter);
489 (summer);
Simulated:
609
Annual GPP (gross primary production;
gC/m2/yr)
Simulated:
−1038 (annual);
Literaturec:
−1973 ± 10% (Year 1);
−1862 ± 15% (Year 2);
Literaturee:
−1012
Simulated:
−2218 (Year 1);
−1595 (Year 2)
Simulated:
−1034
Annual NEE
(net ecosystem exchange; gC/m2/yr)
Literatureb:
79 ± 25 (annual)
Simulated:
−704 (Year 1);
−545 (Year 2)
Literaturee:
174–225
Simulated:
36
Simulated:
133
Yield (gC/m2/yr)
Literatureb:
373 (annual)
Literatured:
701 (Year 1);
535 (Year 2);
Literaturee:
405
Simulated:
378 (annual)
Simulated:
696 (Year 1);
532 (Year 2);
Simulated:
408
Annual NEE + Exported Yield (gC/m2/yr)
Simulated:
414 (annual)
Simulated:
−8 (Year 1);
−13 (Year 2);
Literaturee:
579–640
Simulated:
541
a Lohila, 2008.
b Lohila et al., 2004.
c Karki et al., 2019.
d Kandel et al., 2020.
e Kløve et al., 2010.
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385upper peat layer at that site (low SR), and the simulation is for timothy
under deficient aeration (low PR) instead of nature vegetation commu-
nity with high WTL. In regime III with higher WTL, the growth of timo-
thy was greatly limited and the CH4 production accelerated under the
anaerobic condition. The emissions increased unless timothy was re-
placed by other grass species with high anaerobic tolerance. Based on
the model prediction, the current drainage practices with WTL at
−0.8m creates a comfortable zone for grass growth, but had the highest
emissions. As a result, we think that a WTL rise up to (~−0.6 m) would
be feasible for sustained grass production. Farmersmay also take further
actions with the help of the model predictions to meet their specific
targets.
5. Discussion
5.1. Model performance
We used the dual-porosity framework to simulate the soil moisture
under drainage practices for cultivated peatlands. The indicator RMSE
demonstrated the “good” performance of WTL modelling with its
value less than 15 cm at all the four sites (Mohammadighavam and12Kløve, 2016) and it tended to be better than Orchidee-Peat model
(24.40–25.93 cm; Qiu et al., 2018). Meanwhile, compared with the em-
pirical methods to calculate ER, the BASGRA-BGC model (0.76–0.88)
showed similar (0.71–0.90; Kandel et al., 2013) or even better (0.10;
Lohila et al., 2003) skills in terms of r values to capture the temporal dy-
namics of ER. However, unlike these empirical models that generally
target limited variables and need grass-related measurements (e.g.
LAI) as inputs, the BASGRA-BGCmodel could providemore comprehen-
sive outputs including grass growth and soil biogeochemical processes.
Meanwhile, the accuracy of the BASGRA-BGC model for cultivated
peatlands was also comparable with other process-based biogeochemi-
cal models. For example, r values varied between 0.37 and 0.85 and the
averaged errors vary between 0.0% and 26% for CH4 emissions using
DNDC model (Deng et al., 2015). Besides, r value was 0.78 and RMSE
was 0.83 gC/m2/day for ER using Orchidee-Peat model (Qiu et al.,
2018). Compared with ecological models focusing more on soil biologi-
cal processes and hydrological model focusing on soil water dynamics,
the BASGRA-BGC model showed its advantage on integrating different
water management practices and GHG emissions into model simula-
tion. Therefore, it could not onlymodel GHG emissions, but also provide
useful predictions to improve the field management.
Fig. 8. The daily CH4 flux at the four sites. (The positive value of CH4 flux means CH4 emitted from peat soils and vice versa.)
Table 6
The comparison between simulated annual CH4 flux and estimations.
Site Period Simulation
(kgC/ha/yr)
Estimation
(kgC/ha/yr)
Jokioinen
2000.09–2001.08 −0.14 −0.21 ± 0.18a
2001.09–2002.08 −0.04 −0.20 ± 0.15a
Rovaniemi 2000.09–2001.08 3.85 2.73 ± 0.98a
Nørreå
2015.03.05–2016.03.04
Control: 31.8
Flooded: 615
Control: 30 ± 25b
Flooded: 610 ± 170b
2016.03.05–2017.03.04
Control: 68.2
Flooded: 634
Control: 70 ± 30b
Flooded: 870 ± 130b
Bodø 2003.08.20–2004.11.02
Drainage: 2.60
Natural: 43.62
Drainage: 2.06c
Natural: 48.58c
a Regina et al., 2007.
b Kandel et al., 2020.
c Kløve et al., 2010.
Fig. 9. The simulations of average annual CO2 equivalents (2000.10–2002-0
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385
13Moreover, both the site-specific results and across-site comparison
demonstrated negative correlation between SR rates and WTLs and
the positive relationship between CH4 emission rates and WTLs, which
corresponded well with the meta-analysis in previous publications
(Carlson et al., 2015; Moore and Dalva, 1993; Ojanen et al., 2010). The
trade-offs between CO2 and CH4 (Hatala et al., 2012) was illustrated in
Fig. 9 under different WTLs. The temperature also had significant influ-
ence on CO2 and CH4 emissions (Lafleur et al., 2005) as both SR and ER
were positively correlated with daily temperature (see Fig. 6). All these
results guarantee the robustness of our model outputs.
5.2. Uncertainty in observed data
There is inevitable uncertainty in both the measurements to param-
eterize the model and the input data to drive the model. These uncer-
tainty sources include: (i) As peat soils differ a lot in their quality with
respect to the components and drainage history, brief information
about total soil C/N contentsmay not be enough to illustrate the decom-
position potential of certain peat soil. Lack of detailed peat quality9) for grass yield and GHG emission in Jokioinen under different WTLs.
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385measurements makes it hard to systemically explain the model param-
eterization. (ii) The emission measurements with manual chambers
have a relatively low sampling frequency. As explained in Section 3.3,
the measurements had to be adjusted to daily time step to fit the tem-
poral resolution ofmodel outputs. However, the upscalingmethod itself
introduced uncertainty for model validation. (iii) Direct measurements
of SR with similar temporal resolution as ER were not available for any
of the four sites. Althoughwe used estimated annual SR rates from pre-
vious studies, they were derived either from mass balance calculations
or from bare soil measurements, which are both limited in representing
the real field conditions. To fully evaluate themodel with respect to CO2
emissions, and the contribution of above- and below-ground processes
to the total emissions, more detailed measurements for both SR and ER
are needed. The detailed properties in different depths across the soil
column will reduce the predictive uncertainty. We expect the continu-
ous measurements (from automatic chambers and eddy covariance)
will be available for further validation of model simulations. More accu-
rate approaches to estimate the SR rate from cultivated peatland are
needed to determine the loss rate of C stocks.5.3. Uncertainty in model
The site-specific parameterization scheme (Param2) used for model
running was manually determined and therefore subjective and sub-
optimal. Meanwhile, in the BASGRA-BGC model, decomposition and
methane modules focus more on the biological processes of CO2 and
CH4 production, while the physical transport of the gases especially
through the snowpack is not included yet. The simulated ER was sys-
tematically overestimated in winter compared with the measurements
at Jokioinen, Rovaniemi and Bodø, whereas the model's performance
was satisfying at Nørreå with warmer winters (Fig. 6). We believe that
the snow cover has a buffering or storing effect on the short-term CO2
emission pattern but little impact on the annual balance. Besides, most
models, including theBASGRA-BGC, donot describe the anaerobic stress
for crop growth in detail. To our knowledge, the method used in
Section 2.4, as well as other similar methods, are among the few ap-
proaches to model the deficient aeration conditions. However,Fig. 10. The relationship between WTL and
14according to the model simulation in Nørreå, the simulated ER and
GPP in Year 2 (with higher WTL) had a significant discrepancy from
other estimations even though the lower grass yield implies that anaer-
obic stress indeed existed for grass growth (see Table 4). As a result, a
more accurate method for deficient aeration modelling is needed for
the waterlogged environment of peat soils. Besides, Nørreå site under
flooded condition had greater annual CH4 emissions in the second
year of flooding than in the first year (see Table 5). This indicated that
transition period from aerobic and anaerobic conditions should be con-
sidered in modelling of CH4 emissions.5.4. Implications for field management and research
In Section 4.4, we took Jokioinen as an example to illustrate how the
model prediction could be used for decision support to reduce CO2 and
CH4 emissions and maintain yield. In Fig. 10, we showed the simulated
relationship between WTL and SR at the four sites. The corresponding
air temperature was divided by 3 °C intervals to minimize the influence
of temperature on peat decomposition rate. The results clearly demon-
strate that thenegative correlation betweenWTL and SR ratewas signif-
icant at all sites. However, the within-site variabilities of SR rate among
a certain temperature interval reached up to 80%–150% even at the
same WTL. Given the small temperature effect, differences in the soil
moisture above the water table likely accountable for this diversity.
The variation of SR rate was more obvious with deeperWTL as the cor-
responding soil moisture above became more uncertain compared to
shallow WTL. For example, at Jokioinen and Bodø with deep WTL,
both short-term drought and heavy precipitation did not induce imme-
diate changes in WTL, but the peat decomposition rate varied greatly
due to the fluctuation of soil moisture in the upper SOC-rich layers.
This study proved WTL is still the most important indicator for the
balance between climate effects and productivity of cultivated
peatlands, which is very operative for monitoring and field manage-
ment. However, WTL target is most likely insufficient for optimal SOC
management when other environmental factors and specific practices
are not considered. For instance, to rewet the drained peat soils, surface
flooding likely leads to a higher moisture in upper soil layers thanSR in different temperature intervals.
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385subsurface irrigation despite of similar WTLs and therefore the two
methods would have different peat loss performances. As a result,
more indicators and information, including predictions from process-
basedmodel, could be used to support developing effective waterman-
agement practices and decision making. Meanwhile, we suggest that
more environmental factors alongside with WTL should be taken into
consideration in experimental designs to providemore accurate and ob-
jective conclusions about peat carbon balance.
6. Conclusion
The outputs of the BASGRA-BGC model accurately represented the
short-term dynamics of WTL, CO2 and CH4 emissions, as well as the an-
nual factors for GHG and grass yield. Thereafter, we used the BASGRA-
BGC model to predict the effects of WTL control on GHG emissions and
grass yield, which demonstrates the strength of such model to improve
fieldmanagement and to balance the production versus the environmen-
tal effects. Additionally, the simulations in the four sites indicate thatWTL
still appears as the most straightforward indicator to prevent C loss from
peatlands, but given the significant variability of peat decomposition rate
under the sameWTL,more environmental factors and information should
be considered in accordancewith the specific practices. To provide amore
comprehensive and accurate assessment of the cultivated peatlands' dy-
namics, more data with higher temporal resolution across the Nordic re-
gion will be collected, and the biogeochemical processes in the BASGRA-
BGC model will be further improved in the future work.
CRediT authorship contribution statement
Xiao Huang: Writing – original draft, Methodology, Software,
Formal analysis, Visualization. Hanna Silvennoinen: Project adminis-
tration, Funding acquisition, Writing – review & editing. Bjørn Kløve:
Methodology, Resources, Writing – review & editing. Kristiina Regina:
Methodology, Resources, Writing – review & editing. Tanka P. Kandel:
Methodology, Resources, Writing – review & editing. Arndt Piayda:
Methodology, Resources, Writing – review & editing. Sandhya Karki:
Methodology, Resources, Writing – review & editing. Poul Erik
Lærke: Methodology, Resources, Writing – review & editing. Mats
Höglind:Methodology, Resources, Writing – review & editing.
Declaration of competing interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influ-
ence the work reported in this paper.
Acknowledgements
We acknowledge support from the project “Climate smart use of
Norwegian organic soils” (MYR, 2017-2022) funded by the Research
Council of Norway (decision no. 281109). Thanks go to anonymous re-
viewers and editor for their constructive comments on this paper. We
also thank Dr. Marcel Van Oijen for his valuable suggestions on model
development.
Appendix A. Supplementary Information
Supplementary data to this article can be found online at https://doi.
org/10.1016/j.scitotenv.2020.144385.
References
Arnold, J.G., Fohrer, N., 2005. SWAT2000: current capabilities and research opportunities
in applied watershed modelling. Hydrol. Process. 19 (3), 563–572. https://doi.org/
10.1002/hyp.5611.
Berglund, Ö., Berglund, K., 2010. Distribution and cultivation intensity of agricultural
peat and gyttja soils in Sweden and estimation of greenhouse gas emissions from15cultivated peat soils. Geoderma 154 (3), 173–180. https://doi.org/10.1016/j.
geoderma.2008.11.035.
Berglund, Ö., Berglund, K., 2011. Influence of water table level and soil properties on emis-
sions of greenhouse gases from cultivated peat soil. Soil Biol. Biochem. 43 (5),
923–931. https://doi.org/10.1016/j.soilbio.2011.01.002.
Binet, S., Gogo, S., Laggoun-Défarge, F., 2013. A water-table dependent reservoir model to
investigate the effect of drought and vascular plant invasion on peatland hydrology.
J. Hydrol. 499, 132–139. https://doi.org/10.1016/j.jhydrol.2013.06.035.
Bragazza, L., Buttler, A., Siegenthaler, A., Mitchell, E.A., 2009. Plant litter decomposition
and nutrient release in peatlands. Geoph. Monog. 184, 99–110. https://doi.org/
10.1029/2008GM000815.
Carlson, K.M., Goodman, L.K., May-Tobin, C.C., 2015. Modeling relationships between
water table depth and peat soil carbon loss in southeast Asian plantations. Environ.
Res. Lett. 10 (7), 74006. https://doi.org/10.1088/1748-9326/10/7/074006.
Chabbi, A., et al., 2017. Aligning agriculture and climate policy. Nat. Clim. Chang. 7 (5),
307–309. https://doi.org/10.1038/nclimate3286.
Davidson, E.A., Janssens, I.A., Luo, Y., 2006. On the variability of respiration in terrestrial
ecosystems: moving beyond Q10. Glob. Chang. Biol. 12 (2), 154–164. https://doi.
org/10.1111/j.1365-2486.2005.01065.x.
Deng, J., Li, C., Frolking, S., 2015. Modeling impacts of changes in temperature and water
table on C gas fluxes in an Alaskan peatland. Journal of Geophysical Research: Bioge-
osciences 120 (7), 1279–1295. https://doi.org/10.1002/2014JG002880.
Eickenscheidt, T., Heinichen, J., Drösler, M., 2015. The greenhouse gas balance of a drained
fen peatland is mainly controlled by land-use rather than soil organic carbon content.
Biogeosciences 12 (17), 5161–5184. https://doi.org/10.5194/bg-12-5161-2015.
Forbord, M., Vik, J., 2017. Food, farmers, and the future: investigating prospects of in-
creased food production within a national context. Land Use Policy 67, 546–557.
https://doi.org/10.1016/j.landusepol.2017.06.031.
Franchini, M., Pacciani, M., 1991. Comparative analysis of several conceptual rainfall-
runoff models. J. Hydrol. 122 (1), 161–219. https://doi.org/10.1016/0022-1694(91)
90178-K.
Frolking, S., Roulet, N.T., Moore, T.R., Richard, P.J.H., Lavoie, M., Muller, S.D., 2001. Model-
ing northern peatland decomposition and peat accumulation. Ecosystems 4 (5),
479–498. https://doi.org/10.1007/s10021-001-0105-1.
Fumoto, T., Kobayashi, K., Li, C., Yagi, K., Hasegawa, T., 2008. Revising a process-based bio-
geochemistry model (DNDC) to simulate methane emission from rice paddy fields
under various residue management and fertilizer regimes. Glob. Chang. Biol. 14 (2),
382–402. https://doi.org/10.1111/j.1365-2486.2007.01475.x.
Grønlund, A., Hauge, A., Hovde, A., Rasse, D.P., 2008. Carbon loss estimates from cultivated
peat soils in Norway: a comparison of three methods. Nutr. Cycl. Agroecosys. 81 (2),
157–167. https://doi.org/10.1007/s10705-008-9171-5.
Harriss, R.C., Gorham, E., Sebacher, D.I., Bartlett, K.B., Flebbe, P.A., 1985. Methane flux from
northern peatlands. Nature 315 (6021), 652–654. https://doi.org/10.1038/315652a0.
Hatala, J.A., Detto, M., Sonnentag, O., Deverel, S.J., Verfaillie, J., Baldocchi, D.D., 2012.
Greenhouse gas (CO2, CH4, H2O) fluxes from drained and flooded agricultural
peatlands in the Sacramento-San Joaquin Delta. Agric. Ecosyst. Environ. 150, 1–18.
https://doi.org/10.1016/j.agee.2012.01.009.
Hjelkrem, A.R., Höglind, M., van Oijen, M., Schellberg, J., Gaiser, T., Ewert, F., 2017.
Sensitivity analysis and Bayesian calibration for testing robustness of the
BASGRA model in different environments. Ecol. Model. 359, 80–91. https://doi.
org/10.1016/j.ecolmodel.2017.05.015.
Höglind, M., Van Oijen, M., Cameron, D., Persson, T., 2016. Process-based simulation of
growth and overwintering of grassland using the BASGRA model. Ecol. Model. 335,
1–15. https://doi.org/10.1016/j.ecolmodel.2016.04.024.
Höglind, M., Cameron, D., Persson, T., Huang, X., van Oijen, M., 2020. BASGRA_N: a model
for grassland productivity, quality and greenhouse gas balance. Ecol. Model. 417,
108925. https://doi.org/10.1016/j.ecolmodel.2019.108925.
Hooghoudt, S.B., 1940. Bijdragen tot de kenins van enige natuurkundige grootheden van
de ground. No. 7. Versl. Landbouwk. Onderz [contributions to the knowledge of some
physical constants of the soil. No. 7]. Report Agric. Resour. 46, 515–707.
Kandel, T.P., Elsgaard, L., Lærke, P.E., 2013. Measurement and modelling of CO2 flux from
a drained fen peatland cultivated with reed canary grass and spring barley. GCB
Bioenergy 5 (5), 548–561. https://doi.org/10.1111/gcbb.12020.
Kandel, T.P., Elsgaard, L., Lærke, P.E., 2017. Annual balances and extended seasonalmodel-
ling of carbon fluxes from a temperate fen cropped to festulolium and tall fescue
under two-cut and three-cut harvesting regimes. GCB Bioenergy 9 (12), 1690–1706.
https://doi.org/10.1111/gcbb.12424.
Kandel, T.P., Karki, S., Elsgaard, L., Labouriau, R., Lærke, P.E., 2020. Methane fluxes from a
rewetted agricultural fen during two initial years of paludiculture. Sci. Total Environ.
713, 136670. https://doi.org/10.1016/j.scitotenv.2020.136670.
Karki, S., Kandel, T.P., Elsgaard, L., Labouriau, R., Lærke, P.E., 2019. Annual CO2 fluxes from
a cultivated fen with perennial grasses during two initial years of rewetting. Mires &
Peat 25. doi:10.19189/MaP.2017.DW.322.
Kasimir, Å., He, H., Coria, J., Nordén, A., 2018. Land use of drained peatlands: greenhouse
gas fluxes, plant production, and economics. Glob. Chang. Biol. 24 (8), 3302–3316.
https://doi.org/10.1111/gcb.13931.
Kirkham, D., 1957. Theory of Land Drainage. Drainage of agricultural lands. American So-
ciety of Agronomy Madison, Wisconsin, In, pp. 139–181.
Kleinen, T., Brovkin, V., Schuldt, R.J., 2012. A dynamic model of wetland extent and peat
accumulation: results for the Holocene. Biogeosciences 9 (1), 235–248. https://doi.
org/10.5194/bg-9-235-2012.
Kløve, B., Sveistrup, T.E., Hauge, A., 2010. Leaching of nutrients and emission of green-
house gases from peatland cultivation at Bodin, Northern Norway. Geoderma 154
(3), 219–232. https://doi.org/10.1016/j.geoderma.2009.08.022.
Kløve, B., Berglund, K., Berglund, Ö., Weldon, S., Maljanen, M., 2017. Future options
for cultivated Nordic peat soils: can land management and rewetting control
X. Huang, H. Silvennoinen, B. Kløve et al. Science of the Total Environment 765 (2021) 144385greenhouse gas emissions? Environ. Sci. Pol. 69, 85–93. https://doi.org/10.1016/
j.envsci.2016.12.017.
Korhonen, P., et al., 2018. Modelling grass yields in northern climates - a comparison of
three growth models for timothy. Field Crop Res. 224, 37–47. https://doi.org/
10.1016/j.fcr.2018.04.014.
Lafleur, P.M., Moore, T.R., Roulet, N.T., Frolking, S., 2005. Ecosystem respiration in a cool
temperate bog depends on peat temperature but not water table. Ecosystems 8 (6),
619–629. https://doi.org/10.1007/s10021-003-0131-2.
Lawrence, D. et al., 2019. CLM5 documentation. Tech. rep., Boulder, CO: National Center
for Atmospheric Research.
Leifeld, J., Menichetti, L., 2018. The underappreciated potential of peatlands in global cli-
mate change mitigation strategies. Nat. Commun. 9 (1), 1071. https://doi.org/
10.1038/s41467-018-03406-6.
Lloyd, J., Taylor, J.A., 1994. On the temperature dependence of soil respiration. Funct. Ecol.
8 (3), 315–323. https://doi.org/10.2307/2389824.
Lohila, A., 2008. Carbon dioxide exchange on cultivated and afforested boreal peatlands,
Finnish Meteorological Institute Contributions 73, Yliopistopaino, Helsinki. PhD
Thesis.
Lohila, A., Aurela, M., Regina, K., Laurila, T., 2003. Soil and total ecosystem respiration in
agricultural fields: effect of soil and crop type. Plant Soil 251 (2), 303–317. https://
doi.org/10.1023/A:1023004205844.
Lohila, A., Aurela, M., Tuovinen, J., Laurila, T., 2004. Annual CO2 exchange of a peat field
growing spring barley or perennial forage grass. Journal of Geophysical Research: At-
mospheres 109 (D18). https://doi.org/10.1029/2004JD004715.
Maljanen, M., Liikanen, A., Silvola, J., Martikainen, P.J., 2003. Methane fluxes on agricul-
tural and forested boreal organic soils. Soil Use Manag. 19 (1), 73–79. https://doi.
org/10.1111/j.1475-2743.2003.tb00282.x.
Mezbahuddin, M., Grant, R.F., Flanagan, L.B., 2016. Modeling hydrological controls on var-
iations in peatwater content, water table depth, and surface energy exchange of a bo-
real western Canadian fen peatland. Journal of Geophysical Research: Biogeosciences
121 (8), 2216–2242. https://doi.org/10.1002/2016JG003501.
Mohammadighavam, S., Kløve, B., 2016. Evaluation of DRAINMOD 6.1 for hydrological
simulations of peat extraction areas in northern Finland. J. Irrig. Drain. Eng. 142
(11), 4016053. doi:https://doi.org/10.1061/(ASCE)IR.1943-4774.0001086.
Moore, T.R., Dalva, M., 1993. The influence of temperature and water table position on
carbon dioxide and methane emissions from laboratory columns of peatland soils.
J. Soil Sci. 44 (4), 651–664. https://doi.org/10.1111/j.1365-2389.1993.tb02330.x.
Myhre, G., et al., 2013. Anthropogenic and Natural Radiative Forcing. Climate Change
2013: The Physical Science Basis. Contribution ofWorking Group I to the Fifth Assess-
ment Report of the Intergovernmental Panel on Climate Change, 659–740. Cambridge
University Press, Cambridge.
Myllys, M., Lilja, H., Regina, K., 2012. The Area of Cultivated Organic Soils in Finland Ac-
cording to GIS Datasets, Proceedings of the 14th International Peat Congress.
Peatlands in Balance, Stockholm, Sweden.
Nielsen, O., et al., 2013. Denmark’s National Inventory Report 2013: Emission Inventories
1990–2011-Submitted under the United Nations Framework Convention on Climate
Change and the Kyoto Protocol Aarhus Universitet (DCE-Nationalt Center for Miljø og
Energi).
Ojanen, P., Minkkinen, K., Alm, J., Penttilä, T., 2010. Soil–atmosphere CO2, CH4 and N2O
fluxes in boreal forestry-drained peatlands. Forest Ecol. Manag. 260 (3), 411–421.
https://doi.org/10.1016/j.foreco.2010.04.036.16Parton, W.J., 1996. The CENTURY model. In: Powlson, D.S., Smith, P., Smith, J.U. (Eds.),
Springer. Berlin Heidelberg, Berlin, Heidelberg, pp. 283–291 https://doi.org/
10.1007/978-3-642-61094-3_2.
Petersen, S.O., et al., 2012. Annual emissions of CH4 and N2O, and ecosystem respiration,
from eight organic soils inWestern Denmarkmanaged by agriculture. Biogeosciences
9 (1), 403–422. https://doi.org/10.5194/bg-9-403-2012.
Qiu, C., et al., 2018. ORCHIDEE-PEAT (revision 4596), a model for northern peatland CO2,
water, and energy fluxes on daily to annual scales. Geosci. Model Dev. 11 (2),
497–519. https://doi.org/10.5194/gmd-11-497-2018.
Qiu, C., et al., 2019. Modelling northern peatland area and carbon dynamics since the Ho-
locenewith the ORCHIDEE-PEAT land surfacemodel (SVN r5488). Geosci. Model Dev.
12 (7), 2961–2982. https://doi.org/10.5194/gmd-12-2961-2019.
Raes, D., Steduto, P., Hsiao, T.C., Fereres, E., 2012. AquaCrop Version 4.0: Chapter 3 Calcu-
lation Procedures, FAO. Land and Water Division, Rome, Italy.
Regina, K., Alakukku, L., 2010. Greenhouse gas fluxes in varying soils types under conven-
tional and no-tillage practices. Soil Tillage Res. 109 (2), 144–152. https://doi.org/
10.1016/j.still.2010.05.009.
Regina, K., Syväsalo, E., Hannukkala, A., Esala, M., 2004. Fluxes of N2O from farmed peat
soils in Finland. Eur. J. Soil Sci. 55 (3), 591–599. https://doi.org/10.1111/j.1365-
2389.2004.00622.x.
Regina, K., Pihlatie, M., Esala, M., Alakukku, L., 2007. Methane fluxes on boreal arable soils.
Agric. Ecosyst. Environ. 119 (3), 346–352. https://doi.org/10.1016/j.agee.2006.08.002.
Rezanezhad, F., Price, J.S., Quinton, W.L., Lennartz, B., Milojevic, T., Van Cappellen, P., 2016.
Structure of peat soils and implications for water storage, flow and solute transport: a
review update for geochemists. Chem. Geol. 429, 75–84. https://doi.org/10.1016/j.
chemgeo.2016.03.010.
Shi, X., et al., 2015. Representing northern peatland microtopography and hydrology
within the community land model. Biogeosciences 12 (21), 6463–6477. https://doi.
org/10.5194/bg-12-6463-2015.
Smith, L.C., et al., 2004. Siberian peatlands a net carbon sink and global methane source
since the early holocene. Science 303 (5656), 353. https://doi.org/10.1126/
science.1090553.
Van Oijen, M., Höglind, M., 2016. Toward a Bayesian procedure for using process-based
models in plant breeding, with application to ideotype design. Euphytica 207 (3),
627–643. https://doi.org/10.1007/s10681-015-1562-5.
Wania, R., Ross, I., Prentice, I.C., 2009. Integrating peatlands and permafrost into a dy-
namic global vegetation model: 1. Evaluation and sensitivity of physical land surface
processes. Global Biogeochem. Cy. 23 (3). https://doi.org/10.1029/2008GB003412.
Weather observation. Finnish Meteorological Institute. https://en.ilmatieteenlaitos.fi/
download-observations.
Wiréhn, L., 2018. Nordic agriculture under climate change: a systematic review of chal-
lenges, opportunities and adaptation strategies for crop production. Land Use Policy
77, 63–74. https://doi.org/10.1016/j.landusepol.2018.04.059.
Woodward, S.J.R., Van Oijen, M., Griffiths, W.M., Beukes, P.C., Chapman, D.F., 2020. Identi-
fying causes of low persistence of perennial ryegrass (Lolium perenne) dairy pasture
using the Basic Grassland model (BASGRA). Grass Forage Sci. 75 (1), 45–63. https://
doi.org/10.1111/gfs.12464.
Yu, Z.C., 2012. Northern peatland carbon stocks and dynamics: a review. Biogeosciences 9
(10), 4071–4085. https://doi.org/10.5194/bg-9-4071-2012.