Widely present in boreal regions, peatlands contain large carbon stocks
because of their hydrologic properties and high water content, which makes
primary productivity exceed decomposition rates. We have enhanced the global
land surface model ORCHIDEE by introducing a hydrological representation of
northern peatlands. These peatlands are represented as a new plant functional
type (PFT) in the model, with specific hydrological properties for peat soil.
In this paper, we focus on the representation of the hydrology of northern
peatlands and on the evaluation of the hydrological impact of this
implementation. A prescribed map based on the inventory of
Peatlands are widely present in northern latitudes and in permafrost
regions. They contain large carbon stocks that are estimated between 473 and
621 Gt C in boreal regions
Given the importance of peatlands in the carbon and hydrological cycle,
studies have attempted to include their representation in global models.
Peatlands have specific properties concerning vegetation, hydrology and carbon.
As ecosystems, peatlands are very different than other land areas,
because the vegetation can survive in a permanently inundated area
The representation of peatlands in a land surface model therefore requires
specific parameterizations of biological and hydrological processes. In order
to improve the
Some studies have used the flooded area to model potential peat areas and
then estimate their amount of methane emissions
In this study, we focus on a better representation of the hydrological
processes occurring in peatlands located in northern latitudes. These
developments were carried out in the high-latitude version of the model
ORCHIDEE (ORCHIDEE-MICT;
In the present study, we introduce the northern peatland (version
ORC-HL-PEAT, rev. 3058) in a specific high-latitude version
(ORCHIDEE-HL/ORCHIDEE-MICT v1, rev. 1255;
The model is driven by a prescribed atmospheric forcing, rather than coupled with the atmospheric model, in order to facilitate the assessment of the newly introduced processes, which will be described in the following sections. In addition to the meteorological driving data, the model requires distributed parameters such as vegetation distribution, soil texture, topography and watershed location to represent land surface properties.
Including bare soil, there are 13 different PFTs in
the model. In this study, dynamic vegetation is not activated and the
fraction of each PFT is prescribed. We use the
Some hydrological variables, such as transpiration and interception, vary as
a function of the vegetation. The transport of water in the soil is described
by the 11-layer scheme of
The runoff and drainage transport to the river and oceans are accounted for
by a routing scheme which is separated into three reservoirs with different
velocities and residence times
This section describes the developments and methods used to incorporate
peatlands into the model. The inclusion of peatland processes requires
locating them. Here, we choose to use a fixed map of peatlands rather than a
model that describes only inundated areas, as this already has been studied
in ORCHIDEE
The structure of the hydrological scheme in ORCHIDEE implies that the peatlands have to be considered as linked to a new PFT with adapted biological parameters; only this allows for a separated calculation of the water balance.
Given the difficulty of locating large-scale peatlands, many methods exist to
define these surfaces globally, but all of them have some biases. The
TOPMODEL approach, which determines areas where the soil is temporarily
saturated, does not allow for restoring the hydrology of peatlands; these are
then described as wetlands
To represent the evolution of the hydrology of peatlands over time, we chose
to locate peatlands without integrating other wetlands or other soils with
high carbon content. We therefore use the peatland map obtained by
We constructed a new land cover map including peatlands as a combination of
the original map of 13 PFTs and the global peatland map from
This method leads to a reduction of 19 % of grassland area north of
45
Map of fraction of peatlands as defined in the vegetation map of
0.5
Mosses, sphagnum and grassland mainly compose the vegetation in peatlands.
These vegetation types can be grouped as flood-tolerant C
Peatland vegetation can survive in saturated areas. The representation of
inundation stress is taken into account in some models such as LPJ-WHy
On an energetic level, peatland primary production is lower than for grasses.
Peatland vegetation has a nutrient deficiency caused in part by high water
content, resulting in slower plant growth
Because the ORCHIDEE high-latitude version (ORC-HL) model does not include
the nitrogen cycle, the lower net primary productivity
(NPP) observed in peatlands due to the nitrogen
limitation has to be taken into account by strengthening RuBisCo limitation
on carboxylation, which allows for the fixation of carbon dioxide
(
The lower productivity in peatlands is represented empirically by reducing
The reduction of the maximum rate of carboxylation leads to a lower leaf area index (LAI), which also affects the water balance.
In the model, the roots are represented by a root vertical profile
In summary, the new peatland PFT corresponds to a flood-tolerant C
Peatlands have to be represented with a specific hydrological scheme. High
water content in peatlands is maintained by poor drainage. In our model, the
water balance is described for groups of PFTs. In the original version of
ORCHIDEE, three different water columns are defined. These columns pertain to
bare soil, trees and grasses. To represent the hydrological
conditions of peatlands, we have added a new water column corresponding to
the peatland PFT. This leads to a separation of the water balance of peatland
soils that is crucial to represent the water content of these soils. The
calculation of evaporation is therefore separated for these soils, where an
adjustment can subsequently counterbalance the non-representation of the
mosses often present in these environments, which have a significant capacity
to retain water. We aimed at improving the representation of the hydraulic
properties of relevant large-scale peat soils by using appropriate Van
Genuchten parameters of organic peat soils as described in
Table
Van Genuchten parameters applied to organic peat soils for the column of peat soil in this study and their corresponding references.
Peatlands are characterized by a higher average soil moisture content. Peatland water inflow comes from precipitation, surface runoff and from nearby soils. To represent these processes, we choose to infiltrate the soil column of peatlands with the entire runoff generated in the non-peatland soil columns of the same grid box at the same time step.
This water supply is gradually infiltrated into the deeper soil layers.
Infiltration depends on the hydraulic conductivity of each layer as described
by
In the peatland scheme, the amount of standing water above the soil surface
is taken into account as an additional reservoir. In this study, we use the
standard 11-layer scheme of ORCHIDEE, as described by
Water supply to peatlands is produced by the total runoff from other soils in
addition to precipitation. The total mass of water in the grid cell must be
conserved. This leads to a dependence of the water supply to peatlands on the
peatland fraction in a grid cell: if the peatland fraction is small,
peatlands receive a larger amount of water per square metre of peatland than when
the peatland fraction is large, all other things being equal. The water
supply
We studied the dependence of water routing on the peatland fraction for the
Degero site
Amount of total humidity contained in the column of peat soil of 2 m depth at the Degero site in 2001 as a function of the percentage of peatland imposed in a grid cell.
The hydrological variations in peatlands are assessed using the WTD. Perched water tables within peatlands are unusual
The amount of water evaporated depends on the water availability of the soil.
The evaporation
Peatland soils are flooded for part of the year, which leads to a vegetation
saturated with water during this time. In that case, the relative humidity of
the air at the surface of the soil is close to saturation. The calculation of
evaporation is defined with the air humidity at 2 m above the surface
per grid cell, taking into account the water balance of peatland as well as
other PFTs. This unique grid-cell average air humidity is typically lower
than the air humidity at the surface of flooded areas. To address this
problem, we add a resistance to evaporation
We choose a resistance
Observed (black) and modelled (green) turbulent latent heat flux
(LE) in W m
In this study, we made two peatland simulations: the first one, referred to as PEAT-LOWET in the following, includes the resistance to evaporation. In the second one, referred to as PEAT, no such resistance is applied.
The flow of transpiration by the vegetation of peatlands is also reduced by
reducing the GPP with the parameter
We evaluate the modelled processes at different spatial scales in order to evaluate the hydrological behaviour on peatland sites, and also to evaluate the impact of the inclusion of peatlands on large-scale hydrology, which must be carried out for global climate models.
Firstly, we evaluate the modelled peatland processes with site measurements
from the FLUXNET meteorological database
The site of Fajemyr is located in southern Sweden (56
The site of Siikaneva is located in Ruovesi in the south of Finland
(61
The site of Degero is located in the county of Västerbotten in northern
Sweden in the middle boreal zone (64
The data obtained for these three sites are based on available FLUXNET data from
2004 to 2005 for Siikaneva, from 2001 and 2005 for the Degero site, and from
2005 to 2006 for the Fajemyr site
Descriptions of peatlands sites used for site evaluation. The “Years” column corresponds to the available years of FLUXNET meteorological data.
The site evaluation was performed using WFDEI meteorological forcing at the
0.5
The years for which the WTD measurements coincide with the available years of FLUXNET meteorological forcing at the same site are 2002, 2005 and 2006 for the Degero, Siikaneva and Fajemyr sites, respectively. For the site evaluation, the simulated site is assumed to be covered by 100 % of the PFT of peatlands.
The impact of the inclusion of peatland in the model has been studied at
large spatial scales, considering all northern peatlands above
45
We have studied the impact of peatlands on the terrestrial water storage
(TWS) variations north of 45
The impact of peatlands on river discharge has been evaluated for
different catchments. Mean annual cycles of monthly mean discharge are
computed with the CRUNCEP and WFDEI meteorological forcing at 0.5
The simulations were performed with a spin-up of several hundred years to ensure that hydrological processes have reached equilibrium.
The mean diurnal cycle of NEE of the modelled
peatland PFTs is compared to observations with a 10-day running-mean
smoothing that eliminates day-to-day variations. This allows for evaluating the
simulated evolution of the diurnal cycle of NEE on seasonal timescales
(Fig.
The simulated amplitude of the maximum net uptake of
Diurnal cycle of NEE smoothed with a 10-day running-mean filter of
modelled PFT peatland (green), modelled PFT grass (blue) and observation
(black) from the peatland sites of Degero
Monthly mean temperature (black line) and monthly precipitation
(blue bars) at the top and seasonal cycle of modelled (green) and observed
(black) water table depth (WTD) at the bottom for the peatland sites of
Degero
The hydrology of peatlands is evaluated by comparing the modelled and
observed WTD as shown on Fig.
In the setup of site simulations, the amount of water in the modelled peat soil is filled only with precipitation. However, in reality, the water supply of the minerotrophic sites such as Degero and Siikaneva also comes from lateral input. This phenomenon cannot be represented in the model since the amount of water coming from sub-surface runoff and drainage remains unknown. Therefore, the modelled WTD rather directly follows the quantity of precipitation. Model results for the minerotrophic sites (Degero and Siikaneva) show that the water supply from precipitation only is almost enough to reproduce the observed water table position. For the three sites of this study, the modelled WTD is in agreement with the observations during the summer, when the soil is no longer frozen. Results from the Degero fens site slightly underestimate the WTD during the summer. This small bias can be explained by the amount of water from groundwater, which is not represented in the model. The opposite is observed at the Siikaneva site, where the WTD is overestimated during the summer, which could come from an outgoing flow such as a low drainage rate. In winter, the modelled WTD is underestimated when the soil is frozen. The single-layer snow scheme used in this study does not represent percolation of water. When the soil is frozen, the infiltration of water is blocked, which leads to underestimating the water content in the soil leading to an underestimation of the WTD.
The modelled WTD at the Fajemyr site reached 64 cm during March 2006. This
value can be explained by low rainfall in the previous year, where the
annual precipitation is under 75 % of the amount of the year 2006 with a
monthly value less than 50 mm month
The meteorological conditions of the minerotrophic peatlands allows for a better representation of the hydrology of peatlands than the ombrotrophic bogs such as Fajemyr, where during February and March the simulated frozen soil prevents infiltration and thus maintains undersaturated soil conditions.
In an experiment where we add water content to force saturated soil conditions below 30 cm depth, the model simulates a WTD for the Fajemyr site that matches the observed water table even in winter.
The misfit between modelled WTD and site measurements could be caused in part by the unknown lateral water inputs, which are not represented in site simulations where peatlands represent 100 % of the grid cell. Since we cannot separate bogs and fens at the spatial scales relevant here, we consider that all peatlands are fed by runoff in the model. Due to the lack of large-scale information on the distinction between peatland types, we have chosen to create this map in order to discern the hydrological behaviour of the different types of peatlands. Here, bogs are distinguished from fens from the WTD when peatlands are fed only by precipitation. A precipitation sensitivity study of the different types of peatlands is carried out by modifying the precipitation according to different multiplicative factors.
We modelled ombrotrophic bogs, i.e. peatland fed only by rainfall that do not
receive input from other soil columns, in two steps. First, we made a
simulation switching off runoff transfer from other PFTs and defined the
peatland fraction in a grid cell as ombrotrophic bogs if the water table in
this grid cell was not deeper than 30 cm (in accordance with observations;
The ombrotrophic bogs are diagnosed as localized in areas where peatlands are
flooded during the summer. These bogs are located in north-eastern Canada, on
the west coast of Canada, central Russia, United Kingdom, Norway and
north-west Russia near the White Sea (represented in light blue in
Fig.
Map of northern peatlands separating the type of bogs (light blue) from fens (dark blue) modelled peatlands based on conditions of the water table depth (WTD) fed by the precipitation only. The region of western Siberia (zoomed area) has been used for a sensitivity study of peatland to precipitation.
On a regional scale, we carried out a sensitivity study of the simulated
peatland hydrology to precipitation. The selected study area concerns western
Siberia between 55–75
Snowfall in these Siberian areas leads to persistent snow cover from December
to April, while the averaged simulated WTD in summer reaches
more than 1.5 m depth for ombrotrophic bogs soils that are not supplied with
water by runoff (as shown in Fig.
Seasonal cycle of WTD of minerotrophic (FENS), ombrotrophic (BOGS) and minerotrophic in regions localized as ombrotrophic (FENS in BOGS areas) with the standard precipitation (STD, full line), 50 % of the precipitation (50 % PRECIP, dashed line) and 150 % of the precipitation (150 % PRECIP, dotted line).
The mean WTD of soil of minerotrophic fens reaches 14 cm in summer. The
water supply from the runoff allows the fen soils to have a WTD closer to the
surface than bog soils. When we reduce the precipitation by 50 %, the
WTD of fen soils increases by 1 m, which corresponds to a
relative increase of 7.19 (Fig.
Relative change of water table position of peatlands in August as a function of precipitation multiplication factor ranging from 0.5 to 1.5 depending on the type of peatlands: minerotrophic (FENS), ombrotrophic (BOGS) and minerotrophic in regions localized as ombrotrophic (FENS in BOGS areas).
However, minerotrophic peatlands in this study are located south of
64
To summarize, our simulations show that minerotrophic peatlands are more sensitive to precipitation than ombrotrophic peatlands. The reduction of the precipitation by a factor of 2 leads to a rise of the WTD by up to 8 times deeper for minerotrophic soils while this change does not exceed 2 times deeper for the ombrotrophic soils. This sensitivity is also seen when the runoff process is applied to areas defined as ombrotrophic (case “FENS in BOG areas”).
The inclusion of peatlands in the ORCHIDEE land surface model had to be assessed on a larger scale in order to determine the influence that peatlands have on large-scale hydrology in northern latitudes. After evaluating the processes of simulated peatlands on measurement sites, this study evaluates the impact of peatland implementation on the simulated river flow in boreal watersheds, and on the water mass changes of northern latitudes studied at different timescales.
Implementing peatlands leads to the redirection of runoff from the other soil
columns to the peat soils. Here, we evaluate the impact of these changes on
the simulated river discharge. We compare the modelled river discharge of the
Ob basin in three different simulations. The standard simulation (STD)
corresponds to the version of ORCHIDEE-HL which includes soil freezing
The modelled river discharge with the original ORCHIDEE-HL version
underestimated the river flow of watersheds located in boreal regions. This
underestimation is known with the version of ORCHIDEE-HL that does not
include the snow scheme by
The modelled mean seasonal cycle of the river discharge for the boreal Ob
watershed is shown in Fig.
The simulated river discharge PEAT-LOWET, which includes the reduction of
evapotranspiration as a function of the water supply (translated by the
fraction of peatland in a grid cell), slightly reduces the underestimation of
the river flow of the boreal basins. The average evapotranspiration of Ob
basin peat soils reaches an average of 2.4 mm day
The reduction of evaporation leads to an increase in surface runoff by
577 mm year
These results are not very sensitive to the meteorological forcing used. The seasonal peak of runoff from peat soils occurs one month earlier with the WFDEI than with the CRUNCEP forcing. This results in a large river flow that occurs earlier in the season with the WFDEI forcing. The behaviour of the modelled river discharge of the Ob basin is similar for both meteorological forcings after the month of June.
In all cases, the introduction of the peatland scheme does not alleviate the
underestimate of the spring peak and summer discharge. This tends to confirm
that this underestimate, at least the underestimate of the springtime maximum
linked to snowmelt, is due to the known overestimate of snow evaporation
mentioned before. The new multi-layer snow scheme, not included in the model
version used here, better represents snow depth and snow water equivalent, which were previously
both underestimated in ORCHIDEE. This corrects the underestimation of
snowmelt runoff, and consequently improves the modelled river discharge in
northern high latitudes
We now compare the simulated total terrestrial water storage (TWS) variations
north of 45
The observed TWSs are compared with the simulated total water storage calculated from the water reservoirs represented in ORCHIDEE: surface runoff (FAST), deep drainage (SLOW), lateral flux (STREAM), floodplains (FLOOD), snow mass (SNOW) and humidity of the soil (SOIL).
Figure
At a seasonal scale, the negative contribution of the modelled variation in
TWS occurs too early compared to the observations (figure not shown). This
trend is due to the same shift of the contribution of the modelled TWS,
linked to the snowmelt that occurs about one month too early
The seasonal and interannual variation in TWS in boreal regions is mainly
affected by the changes in snow mass and changes in water contained in soils.
Changes in snow mass contribute alone to more than 5 cm of amplitude, which
represents more than 74 % of the TWS variations north of 40
Modelled TWS in Fig.
In our model, peatlands store a fraction of runoff water that is not
transported to the ocean. To evaluate the impact of peatlands on the
variation of TWS, we selected only the grid-cells containing some peat
(non-zero peat fraction) and performed two simulations. We evaluated the
contribution of the SOIL reservoir both for all soil columns
(Fig.
In summary, the inclusion of the peatland hydrology leads to weaker water loss, which reduces the annual variation in the TWS of these soils. Since peatlands represent a small proportion of soils in northern latitudes, these changes do not significantly affect the large-scale average TWS in northern latitudes.
The hydrology of peatlands in northern latitudes is difficult to assess on a
large scale since the measurements are scarce. One option is to compare
flooded peatlands with satellite observations of flooded areas by
Figure
The flooded surfaces observed by satellite are present from April to August for Siberia and until September for Canada. Concerning modelled flooded peatlands, the seasonality is lower, especially in the northeastern region of Canada where peatlands are flooded throughout the year. Conversely, the extent of the modelled flooded peatlands is in sharp decline during the summer in western Siberia (not shown).
Extent of flooded areas (in km
We compared the seasonality of flooded peatlands with satellite observation
by considering only flooded peatlands when the soil is not frozen and in the
absence of snow. Figure
According to observations, the extent of flooded areas increases from April
following the melting of snow. The increase in the modelled extent of flooded
peatlands is less pronounced and occurs one month later. The simulated total
extent of flooded peatland reaches 0.55
Mean seasonal cycle of flooded land areas of 0.5
The observed area of flooded peatlands is maximum between May and August. Precipitation amounts are also significant in summer in boreal flooded areas. As a result, the extent of flooded peatlands is well represented in July when summer precipitation occurs.
Underestimation of flooded peatlands in spring occurs in Canadian and Siberian
regions with high peatland fractions. In Siberia, the observed flooded areas
are more concentrated in the centre of the region, while the simulated
flooded peatlands are more uniformly distributed over the entire region. In
western Siberia, the model underestimates the flooded peatlands by
0.04
This study has focused on northern peatlands since the majority is located at these latitudes where the increase in temperature is the most important. These peatlands are also often located in permafrost regions that are particularly sensitive to climate warming. We aimed at a better representation of hydrology of peatlands to account for their high carbon content in the soil and to include them in global climate models.
The developments related to the vegetation present in peatlands, however, remain limited. In this study, we suppose a vegetation of peatlands close to the PFT grass already included in the model, with a lower root depth and a lower productivity. The thermal insulation of mosses which is particularly present in peatland is not taken into account.
The hydrological processes of the peatlands were mainly based on the
significant water supply of peatlands from runoff with negligible drainage
and stagnant water on the soil surface. The hydraulic properties of peat
differ greatly from those of mineral soils. Peat soils have a high soil
porosity with an important hydraulic conductivity near the surface, which
decreases much faster than mineral soils
Although these parameters have been measured for peat and parameterized in
the model from the studies of
The water supply of peatlands that comes from the surface runoff of the other soils depends on their fraction within the grid cell. As a result, all other things being equal, the water supply increases as the fraction of peatland within a grid cell decreases. In an extreme case, in which peatlands would be spatially concentrated in a small area within a larger region (because of topographic constraints, for example), this phenomenon would make the peatland hydrology resolution-dependent; however, the large-scale hydrology would certainly also be resolution-dependant in such a case. On the scales of interest here, peatlands mostly represent a small fraction of the grid, and moderate resolution increases will not necessarily lead to very strong variations of the peatland fractions at the grid scale.
The site evaluation showed that the NEE is in agreement with observations and
the internal hydrology of the model, with a negligible deep drainage and
stagnant water above the surface, allows us to represent the water table
profiles of peatland sites when the local weather data are
known. The differences in the water table
profile between observations and simulations can come from the water supply
of the runoff, which is not known and not represented in the simulations on
site. The model underestimates the WTD in winter for the Siikaneva and
Fajemyr sites. This can be explained by the overestimate of snow sublimation
and the underestimate of the snowmelt runoff known in this version of the
model
The development of peatlands in the model does not distinguish the different types of peatlands, since this information is not known on a large scale. In this peatland scheme, we have chosen to feed the peat soils by surface runoff in addition to precipitation in order to obtain shallow WTD, which corresponds to the type of minerotrophic peatlands. This limitation enhances an overestimation of the WTD of ombrotrophic peatlands, which are not represented in this study. However, we located areas where, according to ORCHIDEE, peatlands could subsist as ombrotrophic bogs, in order to evaluate the sensitivity of these types of peatlands to precipitation. The results showed that the WTD of minerotrophic peatlands are more sensitive to precipitation than ombrotrophic peatlands. Areas where ombrotrophic bogs may exist on large scales (that is, under the applied large-scale meteorological forcing) are also less sensitive to precipitation than other regions because the weather conditions are sufficient to supply wetlands with water. These results were obtained with runs at the typical resolution of a climate model. At much smaller scales, variability of topography, meteorological forcing, soil parameters, etc., will of course enable ombrotrophic peatlands to subsist in areas where the large-scale conditions appear unfavourable.
The evaluation of the inclusion of peatlands in the model was carried out at
different spatial scales. We studied the impact of this implementation, which
influences the routing of surface runoff, on the river discharge of the
largest boreal watershed located in Siberia. We have found that hydrological
processes of peatlands including the re-infiltration of surface runoff from
other soils into peat soil induced a reduction of 20 % of the river
discharge of the Ob basin. However, the reduction of the evaporation from
these soils counterbalances this process. The modelled river flow remains
underestimated compared to the GRDC observations
The hydrological impact of this inclusion has also been studied on a large
scale and at different timescales. We have shown that this inclusion has a
negligible impact on terrestrial water mass variations, since the fraction of
peatlands remains sufficiently small compared to the area of high latitudes.
The variations in water mass linked to soil moisture of all soils contribute
to 34 % of the annual mean TWS. This is insufficient to influence the
seasonal and interannual variations in the TWS in northern latitudes.
However, the ORCHIDEE model in the standard version (STD) underestimates the
interannual variability in TWS compared to the observations. In boreal
regions, the interannual variability in TWS is best explained by the change
of precipitation or discharge
The seasonal and interannual variations in water mass from the humidity of
peat soils are reduced by 65 % when the hydrological processes of
peatlands are activated. Changes in snow mass contribute to an average of
74 % of the total variation in TWS. These simulated variations are
underestimated compared to the observations of the GRACE satellite
At the northern latitude scale, the modelled hydrology of peatlands has been
evaluated by comparing the area of modelled flooded peatland with the
satellites observations from
We further showed that the peatland scheme is consistent with observations
both on site
These developments can be used to estimate current and future associated
methane emissions and to understand the sensitivity of methane emissions to
hydrological variations from these northern peatlands. For this, an
adaptation of the
This approach possibly may be used for tropical peatlands but its limitation is that runoff is only received from PFT in the same grid and this would need to be evaluated for tropical conditions. If a tropical peat system is connected to a large-scale hydrological network with water routing connecting grid-cells, then our approach cannot be used.
We have implemented peatlands in the high-latitude version of the land
surface model ORCHIDEE, in order to take into account their important role in
the carbon and hydrological cycle. We have represented peatlands as a new PFT
based on a global inventory peatland map from
The hydrology of peatlands follows the internal hydrology of the model with a redistribution of the surface runoff from other soils, that is redirected into peat soils with a negligible deep drainage and a possible accumulation of water above the surface of these soils. These modifications are evaluated on site measurements that have shown the ability to represent the hydrological profile of peatlands.
We have shown that the reintroduction of the surface runoff of peatlands make these more vulnerable to changes in precipitation than peatlands fed only by precipitation. In this context, we considered that minerotrophic peatlands are more sensitive to precipitation than ombrotrophic peatlands. The location of ombrotrophic peatlands has a lower sensitivity to precipitation than minerotrophic peatland areas.
The impact study of this implementation on the river flow and on the variation of terrestrial water storage showed that the incorporation of peatlands does not significantly affect the continental hydrology of the northern latitudes.
This study showed that the location and the seasonality of flooded peatlands are well represented, despite the low extent in early spring. The runoff of soils re-infiltrated into peat soils has resulted in a reduction of river flow of 20 % continuously for the growing season for the case of the Ob basin. This reduction becomes negligible when the corrected evaporation flux is activated.
At the interannual scale, the variations in the modelled terrestrial water storage of northern latitudes is in accordance with the GRACE satellite observations. At these latitudes, the variations in mass of water from snow and soil moisture are the largest contributors, representing, on average from 2002 to 2014, 74 % and 36 % of the total variations in terrestrial water, respectively. The incorporation of peatlands induces a reduction of the variation in soil moisture of peat soils, both seasonally and interannually. This reduction enhances a total reduction of 6 % of the total variations in terrestrial water, which can be neglected at this large scale.
The new scheme represents peatland hydrology relatively well on a large scale, without disrupting the large-scale hydrology of the surface model. The peatland hydrology has only small effects on the simulation of northern river discharge. This implementation will be further used to estimate the future evolution of the hydrology of peatlands and their associated methane emissions at the end of the century.
The documentation and the code of the trunk version of
ORCHIDEE are open source and can be found here:
External data set from GRACE land are available at
Model development, validation and evaluation was principally carried out by CL. This study has been supervised by the co-authors.
The authors declare that they have no conflict of interest.
This study has been supported by the PAGE21 project, funded by the European Commission FP7-ENV-2011 (grant agreement no. 282700) and the European Research Council Synergy grant ERC-2013-SyG-610028 IMBALANCE-P. Simulations with ORCHIDEE were performed using computational facilities of the TGCC-Curie (CEA, France). Edited by: Gerd A. Folberth Reviewed by: two anonymous referees