Boreal Eurasia is a region where the interaction between droughts and the carbon cycle may have significant impacts on the global carbon cycle. Yet the region is extremely data sparse with respect to meteorology, soil moisture, and carbon fluxes as compared to e.g. Europe. To better constrain our vegetation model SiBCASA, we increase data usage by assimilating two streams of satellite-derived soil moisture. We study whether the assimilation improved SiBCASA's soil moisture and its effect on the simulated carbon fluxes. By comparing to unique in situ soil moisture observations, we show that the passive microwave soil moisture product did not improve the soil moisture simulated by SiBCASA, but the active data seem promising in some aspects. The match between SiBCASA and ASCAT soil moisture is best in the summer months over low vegetation. Nevertheless, ASCAT failed to detect the major droughts occurring between 2007 and 2013. The performance of ASCAT soil moisture seems to be particularly sensitive to ponding, rather than to biomass. The effect on the simulated carbon fluxes is large, 5–10 % on annual GPP and TER, tens of percent on local NEE, and 2 % on area-integrated NEE, which is the same order of magnitude as the inter-annual variations. Consequently, this study shows that assimilation of satellite-derived soil moisture has potentially large impacts, while at the same time further research is needed to understand under which conditions the satellite-derived soil moisture improves the simulated soil moisture.
The interest of this publication is to explore the potential of assimilating satellite-derived soil moisture in the vegetation model SiBCASA over Boreal Eurasia with particular focus on the impact on simulated carbon fluxes. In remote regions such as Boreal Eurasia meteorological driver data for vegetation models (temperature, precipitation, etc.) are poorly constrained by surface observations, leaving room for improvement in the soil moisture content simulated in vegetation models. Boreal Eurasia is also a region with large carbon stocks in biomass and vegetation (Schepaschenko et al., 2013; McGuire et al., 2009; Tarnocai et al., 2009), which are subject to the fastest climatic change rates on Earth (Goetz et al., 2007), making it a relevant region in the context of ecosystem carbon sequestration. Furthermore, large parts of the region are located in continuous and discontinuous permafrost soils. In the spring, meltwater from the accumulated winter precipitation cannot percolate into the still-frozen soil – it runs off in hilly terrain, and forms ponds on the soil in flat terrain. This causes a bi-modal spatial distribution in soil moisture, with dry hills and wet plains. This process is probably hard to catch by land surface models and satellite observations alike. Satellite-derived soil moisture is observed to have large variability in the northern regions, both within and between different approaches (Mladenova et al., 2014). The derivation of satellite soil moisture is difficult in snow-, ice-, and surface-water-rich areas (Högström et al., 2014; Naeimi et al., 2012a). These aspects make a working soil moisture data assimilation system very relevant for vegetation modelling in the region, and challenging. The few tower sites running in the region now have fairly long measurement records, so that they can be used for validation of inter-annual variation (e.g. the 2010 drought). An effort specifically targeted at Boreal Eurasia and carbon fluxes has not been tried before.
Earlier efforts to assimilate satellite-derived soil moisture in vegetation
models were often focused on the improvement of soil moisture itself and/or
validated with in situ observations over short vegetation in temperate and
Mediterranean climate zones (Reichle and Koster, 2005; Reichle et al.,
2007; Draper et al., 2009a, b; Miralles et al., 2011b).
Other studies focus on the effect on crop yield and carbon fluxes in Europe
(Verstraeten et al., 2010; de Wit and van Diepen, 2007; Han et al., 2014).
The Global Land Data Assimilation Systems (GLDAS,
Soil moisture affects vegetation carbon fluxes through photosynthesis and
respiration rates. Photosynthesis rates depend on the stomatal conductance,
which the plants regulate according to the water potential in the leaf and
the atmospheric vapour pressure deficit. The water potential in the leaf is
a function of water supply by the roots and the water use by transpiration
(Katul et al., 2010). Heterotrophic respiration depends on the soil
moisture content, which is the substrate in which microbes and bacteria
consume organic matter and release CO
Based on simulations with global climate models, it is expected that global warming is associated with more extreme precipitation regimes, resulting in more frequent and more intense flooding and drought events (Dai, 2011). It is probable that this trend is already becoming visible in the large number of recent droughts, e.g. in Amazonia (2005 and 2010), the US (2008–2012, 2014), European Russia and Siberia (2010, 2013), West Europe (2003), East Africa (2011–2012), Australia (2003–2007), China (2010–2011), as well as floodings in Austria and Germany (2013), Southwest China (2013), India and Pakistan (2014), and the UK (2014). These extremes may have a large effect on the carbon balance of the affected regions, sometimes undoing 10 years of “normal” carbon uptake by natural vegetation and causing crop yield reduction or crop failure (van der Molen et al., 2011; Peters et al., 2010; Reichstein et al., 2007; Ciais et al., 2005).
Soil moisture in land surface models is the balance of precipitation and evaporation, transpiration, runoff, and lateral outflow. Land surface models are often primarily calibrated to simulate water, heat, and carbon exchange with the atmosphere correctly (Williams et al., 2009; Morales et al., 2005), and the water stress function is often one of the functions that is used for calibration. As a result, the absolute value of soil moisture in land surface models, and its variation, has often been subsidiary to simulating exchange fluxes correctly. In the perspective of the expected increasing occurrence of extremes in soil moisture, it is questionable whether this procedure results in satisfactory representation of droughts in land surface models, and their effects on the carbon balances and the disturbance of pools.
Two independent databases of remotely sensed soil moisture have been published recently: the passive microwave soil moisture data set, based on the land parameter retrieval model (LPRM) (Owe et al., 2008), and the active microwave data set, METOP ASCAT 25 (Wagner et al., 1999; Naeimi et al., 2009, 2012b). See Sect. 2.2 for details.
These data sets provide globally consistent, satellite-observed soil moisture
data. As such, they provide ideal soil moisture information to analyse the
development of droughts on a regional scale. In this paper, we
describe the implementation of a data assimilation system into the land
surface model SiBCASA (Schaefer et al., 2008). This
scheme adjusts the simulated soil moisture towards the satellite-observed
soil moisture, accounting for the errors in observation and model. We
subsequently evaluate the performance of this soil moisture assimilation
system by comparing the simulated soil moisture, and the resulting change in
carbon fluxes against observations collected at four sites across Boreal
Eurasia (defined here as 3
The objective of this paper is therefore to evaluate the use of satellite observed soil moisture for data assimilation in vegetation models in a data-poor region like Boreal Eurasia and to evaluate the impact on the simulated carbon fluxes.
The increasing availability of remote sensing products, e.g. the Soil Moisture and Ocean Salinity mission (SMOS) (Kerr et al., 2001), the Global Change Observation Mission – Water (GCOM-W) (Imaoka et al., 2010; Oki et al., 2010), the Global Land Evaporation: the Amsterdam Methodology (GLEAM) (Miralles, 2011; Miralles et al., 2011b), and the Soil Moisture Active Passive mission (SMAP) (Entekhabi et al., 2010), make this evaluation timely and relevant. If the assimilation of soil moisture proves worthwhile, it may also be applied in other land surface models, e.g. in mesoscale atmosphere models or in agro-meteorological models.
SiBCASA is an interactive vegetation–atmosphere model simulating how the growth of vegetation depends on the exchange of water, energy, and carbon between vegetation, atmosphere, and soil (Schaefer et al., 2008). SiBCASA consists of two coupled components, the SiB component simulates the exchange of heat and water and the uptake of carbon dioxide as a function of temperature, radiation density, humidity, and wind speed, as well as root-zone soil moisture conditions with several time steps per hour (Sellers et al., 1996; Jafarov and Schaefer, 2015). It uses the Farquhar photosynthesis parametrization (Farquhar et al., 1980) in combination with a Ball–Berry type stomatal conductance formulation (Collatz et al., 1991).
The CASA component simulates how the carbon taken up by photosynthesis is allocated to different parts of the vegetation, with specified residence times (Potter et al., 1993). The seasonal development of leaf area index is a function of the amount of carbon allocated to leaves, but the amount of absorbed photosynthetically active radiation (fPAR) is constrained by remote sensing. A more detailed summary of the general features of SiBCASA can be found in van der Velde et al. (2014).
SiBCASA is intended for use as a lower boundary condition for large-scale atmospheric transport models, such as general circulation models and data assimilation systems such as CarbonTracker (Peters et al., 2007). Therefore, it is specifically required to correctly simulate the regional effects of climate variations. In this paper we focus on the role of soil moisture in SiBCASA, with the aim to better describe the effects of climate extremes on the terrestrial carbon dioxide balance. Therefore, we will briefly describe SiBCASA's method of simulating soil water uptake and the relation with stomatal conductance.
SiBCASA is configured with 25 soil layers with a depth up to 15 m and
thicknesses ranging from 2 cm near the surface to 3 m at depth. Roots
extract water from the part of the soil they penetrate. The plant-available
water fraction (
Soil moisture also affects the turnover times of organic matter in the soil. The turnover times are shortest at an optimal soil moisture saturation fraction (which varies around 60 % of the pore space) and from there increases towards drier and wetter soils (Fig. 1) (Raich et al., 1991). The respiration rates are a function of the carbon pools and turnover times, which are temperature and soil moisture dependent.
The net effect of soil moisture on Net Ecosystem Exchange (NEE) is the sum of the effect on photosynthesis (GPP, Gross Primary Production) and on respiration rates (TER, Total Ecosystem Respiration).
The satellite-observed soil moisture data used in this study come from two independent sources (Liu et al., 2011, 2012): the first is from passive microwave observations, the second from active radar observations. Passive microwave radiation sensors have been on board of various satellite platforms, e.g. Scanning Multichannel Microwave Radiometer (SMMR, 1978–1987), Special Sensor Microwave Imager (SSM/I, 1987–present), Tropical Rainfall Measuring Mission Microwave Imager (TMI, 1997–2015) and the Advanced Microwave Scanning Radiometer – Earth Observing System (AMSR-E, 2002–2011). The retrieval algorithm, the Land Parameter Retrieval Method (LPRM) (De Jeu et al., 2009; De Jeu and Owe, 2003; Owe et al., 2008), is based on a simple radiative transfer equation and used dual-polarized brightness temperature observations in an optimization routine to solve for soil moisture. The retrieved soil moisture is representative for the top few centimetres of the soil. LPRM accounts for the vegetation opacity in the microwave domain. Dense canopies attenuate the microwave signal from the underlying soil surface which results in a lower soil moisture retrieval accuracy. Therefore, the soil moisture retrieval is most reliable for bare and sparsely vegetated areas (de Jeu et al., 2008). As this study is focused on the period 2007–2013, we use only the AMSR-E data with version v05 of the LPRM with a 50 km spatial resolution, and a 2–3 day revisit time (Owe et al., 2008).
Complementarily, active radar soil moisture retrievals from the Advanced
Scatterometer (ASCAT, 2006–present) in combination with the change detection
algorithm is representative for the soil moisture in the top few
centimetres (Bartalis et al., 2007; Wagner et al., 1999; Naeimi et al.,
2009, 2012b). ASCAT soil moisture retrievals are reliable for
sparse and moderately vegetated areas, and less for bare soils
(Liu et al., 2012). We use METOP ASCAT25,
version WARP5.5, release 2.1, with a 0.25
The objective of this paper is to attempt an improvement of soil moisture
dynamics in SiBCASA by assimilating the passive microwave and/or ASCAT
satellite-derived data described above. We use the same assimilation method
as used in GLEAM (Miralles et al., 2011a; Miralles, 2011):
Characteristics of the flux tower sites used in this study.
Soil moisture anomalies in August 2009 with respect to the average
soil moisture in the months August 2007–2011.
Since the satellite-observed soil moisture data essentially carry
information only about the temporal variations in soil moisture, and not
about the absolute value of mean and the amplitude of the variations, we
normalize the satellite data (see below) before assimilating the satellite
data in SiBCASA. The entire assimilation procedure consists of the following steps:
Step 1: run SiBCASA without data assimilation to equilibrium in 2007 and
then run until 2013, storing daily model results. Step 2: take the spatial average of satellite-derived soil moisture within
the 1 Step 3: run SiBCASA from equilibrium in 2007 until 2013 with assimilation of
the satellite-derived soil moisture. Step 4: evaluate the simulated soil moisture, water, and carbon fluxes with
in situ flux tower data as described in Sect. 2.4.
For evaluation of the model results, particularly the carbon fluxes, we use 25 site-years of data from four flux tower sites in Boreal Eurasia available in the period 2007–2013 (Table 1). The sites vary in vegetation type, continentality, and permafrost. Flux data were taken in Siberia at more locations, although predominantly in the period 1997–2005, when the ASCAT satellite was not yet launched (Dolman et al., 2012). The sites are in Hyytiälä, Finland (Ilvesniemi et al., 2010; Kolari et al., 2009; Mammarella et al., 2009; Rannik et al., 2004), Tver, European Russia (Kurbatova et al., 2008; Milyukova et al., 2002), Yakutsk, East Siberia (Dolman et al., 2004; Ohta et al., 2008), and Elgeeii, East Siberia (Kotani et al., 2014). The eddy covariance data have been processed according to the harmonized LaThuille FLUXNET database (Baldocchi et al., 2001; Reichstein et al., 2005; Papale et al., 2006; Moffat et al., 2007). All sites are covered with needle leaf forests, evergreen in the western sites, and deciduous in the eastern sites. The stations Yakutsk and Elgeeii are on permafrost and have maximum snow depths of the order of 40 and 50 cm, respectively.
We will first evaluate the spatial coherence between the simulated and satellite-observed soil moisture, then the temporal coherence. All satellite and in situ data are CDF-matched to the SiBCASA soil moisture. Next, we will compare model and satellite soil moisture data with in situ observations. Finally, we will evaluate the impact of satellite soil moisture assimilation on the simulated carbon fluxes.
The spatial coherence between SiBCASA and satellite-observed soil moisture
is studied by comparing maps of monthly soil moisture anomalies and by
quantifying the spatial correlation between the anomalies. We compute the
anomalies for each month with respect to the average soil moisture in that
month in the years 2007 to 2011. We use the reference period until 2011 (and
not 2013), because the AMSR-E satellite became dysfunctional in November 2011,
and we have no passive microwave soil moisture after that date. Figure 2
shows an example of the spatial coherence in August 2009. This is the month
with the largest spatial correlation between SiBCASA and ASCAT soil moisture
in the period that AMSR-E data are available (
These findings are also quite typical for August months in other years. Only
in August 2013 was the spatial correlation coefficient between SiBCASA and ASCAT
soil moisture slightly larger,
Considering the seasonal evolution of the correlation coefficients between SiBCASA and passive microwave soil moisture, there is no coherence between the two, except perhaps for the steppe regions, for which the correlation coefficients reach to 0.50 in Septembers. However, the slope of the regression curve is only about 1 : 3 (third panel of Fig. 2), whereas it is near 1 : 1 for ASCAT soil moisture (second panel of Fig. 2). Because the prior agreement between SiBCASA and passive microwave soil moisture is too low for Boreal Eurasia, we do not consider it meaningful to proceed with assimilation in SiBCASA. We will focus on the ASCAT soil moisture alone in the remainder of this paper. This decision will be further addressed in the discussion section.
The spatial correlation is a measure of how well satellite and SiBCASA agree
on the location of drought and wet
The seasonal variation of the spatial correlation coefficient of SiBCASA and ASCAT soil moisture for all grid cells (black x), tundra cells (red *), forest (green triangles), and steppe (brown diamonds). The errorbars indicate the variation between the years 2007–2013.
The temporal correlation coefficient of SiBCASA and ASCAT soil
moisture for all August months in the period 2007–2013 (7 years
The temporal correlations shown in Fig. 4 were computed with soil moisture on a daily basis. When computed on a monthly average basis, the correlation coefficients generally improve considerably, but the variability in soil moisture is smaller accordingly. This shows that day-to-day noise, particularly in the satellite soil moisture data, is responsible for the low correlation coefficients, and that the remaining correlation is dominated by inter-annual variations. This suggests that it may be worthwhile to investigate assimilation of low-pass filtered satellite soil moisture instead of instantaneous measurements.
To better understand what these large-scale spatial and temporal correlation coefficients imply for the use of satellite soil moisture for assimilation in SiBCASA, time series of soil moisture were compared for four stations across Eurasia. In these time series we show the original SiBCASA soil moisture, the ASCAT soil moisture (CDF-matched), and in situ soil moisture. In addition, we show the SiBCASA soil moisture after assimilation of ASCAT soil moisture according to Sect. 2.3. The first station is Hyytiälä in Finland. The time series are shown in Fig. 5. Simulated and in situ observed soil moisture generally change slowly in time, because of the soil moisture retention in the soil. Since satellite observations lack this memory effect, the satellite observations are noisier. Over the years ASCAT and in situ soil moisture are larger than SiBCASA soil moisture in the spring period (May to early June). This causes the assimilation procedure to increase soil moisture in SiBCASA (red line is higher than the blue line). This increase of soil moisture in the spring period improves the match with in situ observed soil moisture to the degree specified by the uncertainties (Eq. 2).
Time series of soil moisture in SiBCASA original (without assimilation) (blue), and with assimilation (red), in situ soil moisture (orange), and ASCAT soil moisture (marine) in Hyytiälä, Finland. Each panel shows 1 year of soil moisture. Grey shades indicate periods when the top soil is frozen. The three asterisks indicate the date when the top soil is last frozen in the spring, 46 days after that, and the date when the top soil is first frozen again in the fall. Error bars in the top panel indicate the uncertainty in ASCAT soil moisture, which is for clarity only shown for 1 year. The bottom panel shows the average seasonal cycle of the each soil moisture type. In situ and ASCAT soil moisture are CDF-matched to SiBCASA soil moisture, which explains why they have the same mean and standard deviation.
As Fig. 5, but for the Tver wet forest site.
As Fig. 5, but for the Yakutsk Larix site.
As Fig. 5, but for the Elgeeii site.
We loosely define a drought as a period when in situ observed soil moisture is more below the average soil moisture in that period (see bottom panels of Figs. 5–8) than the day-to-day variation during three or more weeks. Subsequently, two distinct drought periods in the time series occurred in July/August 2010 and in July/August 2013. In 2010 the original SiBCASA simulation also “saw” the drought, but ASCAT did not. The assimilation therefore decreased the match with in situ soil moisture. In 2013 the in situ observed drought in Hyytiälä was picked up by neither SiBCASA nor ASCAT.
The second site we analyse is “Tver wet forest”, for which the time series are shown in Fig. 6. In Tver, a similar springtime behaviour emerges. ASCAT is generally larger than SiBCASA soil moisture in April to early May, and the assimilation improves the match with in situ observed soil moisture. Field workers confirm that the water table is generally high or even above the soil surface after snowmelt (April) and decreases quickly in May. In the summers, soil moisture is generally quite constant, except during the 2010 drought, which caused extensive fires in European Russia. The drought was picked up by in situ observations and SiBCASA, but not by ASCAT. As a consequence, the assimilation decreased the match of SiBCASA with in situ soil moisture.
The third site is Yakutsk Larix. At this site, ASCAT soil moisture is noisier than for the other sites (Fig. 7). There is a tendency that in situ soil moisture is high in spring, due to melted snow, and decreasing during the season. This trend is reproduced by SiBCASA in 2007 and 2008 and perhaps in 2009, but not in 2011 to 2013. The ASCAT signal tends to be smaller than average in the spring, and increases somewhat in the summer. There are four intense “droughts” in the in situ time series – in 2008, 2011, 2012, and 2013. Ohta et al. (2014) show that drought conditions at the site occurred between 2001 and 2004, and that the site was actually water logged from 2005 to 2009, returning to normal conditions afterwards. The water logging had severe impact on the ecosystem, with reduced photosynthesis rates and tree browning and mortality. This water logging may be a larger-scale process in eastern Siberia (Muskett and Romanovsky, 2009; Vey et al., 2013). Therefore the term “drought” is relative to the studied period. SiBCASA sees the in situ observed droughts in 2008 and 2012 to some extent, but not the 2011 and 2013 ones. This inter-annual variation in soil moisture data is reflected in the minimum summertime in situ observed soil moisture in Fig. 7, which is much lower in 2011–2013 than in 2007–2008. ASCAT does not detect this inter-annual variation. ASCAT also does not observe the droughts, whereas the ASCAT soil moisture is generally even larger than in SiBCASA, causing the assimilation to decrease the match with in situ soil moisture.
The fourth site is Elgeeii, for which the time series are shown in Fig. 8. SiBCASA soil moisture is again low in the springtime (May), and ASCAT soil moisture is larger. The assimilation increases the soil moisture in SiBCASA, and this seems to improve the match with in situ observed soil moisture, although the early springtime in situ observations were unreliable. In 2012 a drought occurred in July and August. SiBCASA sees the drought too, although with a much earlier development. ASCAT does not see the drought, and as a consequence, the assimilation moves the soil moisture in SiBCASA away from the in situ observations. It is interesting to see that on 5 August 2012 the soil moisture in SiBCASA increases due to a precipitation event, but this is not seen in the in situ observations. However, an increase is seen at that time in the in situ observations in Yakutsk, some 340 km to the northeast, possibly suggesting a displacement of the precipitation event in the SiBCASA driver data from ECMWF ERA-interim.
Of the 11 droughts observed in the in situ time series at the four sites, SiBCASA reproduces four, and ASCAT none (Table 2). The poor skills of ASCAT to reproduce local drought conditions are an apparent contradiction given the good skills in positioning the major drought regions shown in Figs. 2 and 3. Particularly at the locations of our in situ observation sites ASCAT does not perform well in reproducing the temporal variability (see Fig. 4), which is confirmed at the site level in Figs. 5–8. To explain this better, we look at monthly mean soil moisture maps for the five drought occurrences which SiBCASA observes, but ASCAT does not.
Monthly mean soil moisture in
Detection of in situ observed drought occurrences by SiBCASA and ASCAT.
In August 2010 there was a large drought and heat wave in European Russia
and western Siberia, which was caused by a strong blocking situation. The
drought was accompanied with many wildfires (Miralles et al., 2014; Krol
et al., 2013). The drought was also apparent in the in situ measurements
performed in Hyytiälä and Tver. Figure 9 shows that SiBCASA simulates
a drought extending from Scandinavia to Novosibirsk (55
In July 2012 SiBCASA simulates an intense drought that was located around the city Yakutsk, extending eastward to the region between the Lena and Aldan rivers (Fig. S3). The Elgeeii site was located just on the eastern border of the simulated drought region. ASCAT does not observe this drought region, not in July, nor in earlier or later months. Where Elgeeii was on the perimeter of the 2012 drought region, as were Hyytiälä and Tver in 2010, Yakutsk was in the centre of the drought region, which ASCAT does not observe at all. Therefore a site's location on the rim of a drought does not explain why ASCAT does not observe the drought. Rather it appears that ASCAT has limited capability to observe droughts in the forested zone where the in situ observations were made.
GPP (green) and TER (red) simulated with SiBCASA without (solid lines) and with assimilation (dashed lines) of ASCAT soil moisture for Yakutsk Larix.
The changes in soil moisture in springtime and during drought, induced by the assimilation of ASCAT observed soil moisture in SiBCASA (Sect. 3.1, Figs. 5–8), may have substantial effects on the representation of the carbon fluxes, which we will look at next. We will show separately how GPP, TER, and NEE depend on the change in soil moisture and the season.
An interesting case is presented at the Yakutsk Larix site (Fig. 10). At this location, the change in soil moisture due to assimilation of ASCAT soil moisture was large relative to the other sites (Fig. 7). However, although the absolute value of the change in soil moisture was more or less constant throughout the years, the change in GPP shows a distinct seasonal cycle, with large changes in spring, small changes in summer, and hardly any change in fall. This is because of two reasons: (i) the sensitivity of GPP to soil moisture is simulated as a function of the plant-available water fraction (Sect. 2.1, Fig. 1a). In Yakutsk in spring, the permafrost soil has only thawed for a couple of centimetres, resulting in a very small plant available water fraction and very strong soil moisture sensitivity (Fig. 1a: the soil moisture sensitivity curve is steepest on the low plant-available water fraction side). This results in a strong GPP effect of assimilating ASCAT soil moisture in SiBCASA. Note that soil thawing does not automatically mean that more soil moisture becomes available for root uptake. The soils in Yakutsk often freeze after a relatively dry summer, so that the frozen soil may be quite dry. In the spring, the snow meltwater cannot penetrate the soil, which is still frozen, and may run off; (ii) the Yakutian spring is almost simultaneous with the solar maximum on 21 June, so that the potential GPP is large. In the course of the growing season, the permafrost active layer thaws deeper, resulting in a larger plant-available water fraction, reducing the drought sensitivity. This explains the smaller change in GPP in the summer. In the fall, GPP is limited more by the lack of available sunlight than by water stress, explaining the absence of change in GPP with assimilation of satellite soil moisture data.
In a similar way, the change in TER (Fig. 10) does not only depend on the
change in soil moisture with satellite soil moisture assimilation, but also
in the absolute value of soil moisture (Fig. 1b) and temperature limitation
on TER. In June, when the soil is still cold, the changes in TER are small.
In July and August the changes in TER are larger than in GPP, because the
soil is warm and TER is a function of absolute soil moisture change. In this
example, the changes in GPP and TER have the same direction. Figure 1
shows that this is always the case when the soil moisture saturation
fraction is below its minimum value of
Accumulated over a year (Table 3), the changes in GPP, TER, and NEE are of
the order of tens of gC m
The effects of ASCAT soil moisture assimilation in SiBCASA are also
significant when integrated over the entire study domain (27.8
GPP, TER, and NEE in SiBCASA without ASCAT soil moisture assimilation, and the changes dGPP, dTER, dNEE with assimilation for the site Yakutsk Larix.
Top panel: cumulative NEE in Boreal Eurasia for the years 2007 to 2013
according to SiBCASA without assimilation of satellite-observed soil
moisture. The text describes the 7-year mean NEE and the inter-annual
variation (as standard deviation). Bottom panel: cumulative NEE anomaly relative
to the 7-year mean. Solid lines represent the SiBCASA NEE anomaly (
The second largest effect of soil moisture assimilation occurred in 2012,
with an extra anomaly of
The spatial and temporal correlation coefficients between SiBCASA and satellite-observed soil moisture shown in Sect. 3 suggest that ASCAT and passive microwave satellite signals have a certain skill in observing land surface soil moisture. The absence of perfect correlations implies that assimilating the satellite-observed soil moisture in SiBCASA will have an effect. The question is whether that effect is an improvement.
The performance of passive microwave data was low over the entire study
region and in all months (Fig. 2, Sect. 3.1). Only in steppe regions were the
temporal correlations large (
The spatial and temporal correlation coefficients vary with the months and with land cover. The spatial correlation between SiBCASA and ASCAT soil moisture is largest in August and quickly decreases towards the spring and fall. What processes may cause this? Ecologically there are large differences between the seasons in Siberia. Large parts of Siberia are snow covered and particularly the region north of Mongolia and east of the Yenissei River is subject to continuous permafrost. This hampers a correct retrieval of soil moisture from satellite-observed signals (Naeimi et al., 2012a; Högström et al., 2014), while correctly simulating soil moisture under snow conditions is also difficult in vegetation models. However, even in the northern tundra regions most snow and ice have disappeared by June. Considering that the grid cells with frozen top soil in SiBCASA and snow/ice detection in ASCAT have been excluded from the statistical analysis, the lower correlations in June, July, and September (Fig. 3) are probably not only caused by the presence of snow and ice on the land surface.
Other important changes from May to July are the expansion of leafs, the drying out of the topsoil after snowmelt on frozen ground, and the deeper thawing of the permafrost active layer. The increase of the leaf area index (LAI) does not seem beneficial for better satellite soil moisture retrievals, as is also suggest by the smaller correlation coefficients for forests than for steppe zone (Fig. 3). The decrease in the ponded area fraction after snowmelt on frozen ground is a potential explanation for the improving correlation coefficients (Högström et al., 2014), since they occur particularly in the forest and the tundra zones, which contain the wettest parts of the region, and not for the steppe zone, which is drier and outside the permafrost zone.
With the same arguments the increasing depth of the permafrost ice front may also be a potential explanation of the improving spatial correlation coefficients towards August. Indeed, ice and frozen soil at some depth may disturb the satellite signal (Way et al., 1997; Wegmüller, 1990). Maximum active layer thicknesses of a mere 10–20 cm are not uncommon in the northern tundra, although the penetration depth of microwave radiation in the soil is of the order of one to a few centimetres.
It is interesting that the spatial correlation coefficients for steppe zones are larger and for tundra zones smaller than average. Both steppe and tundra vegetation are characterized by short vegetation, but tundra regions are generally much wetter than steppe regions and with continuous permafrost. This implies that the presence of short vegetation alone is not the only prerequisite to obtain a good match between SiBCASA and ASCAT soil moisture.
On the site level, Figs. 5–8 show that ASCAT soil moisture has much more
day-to-day variability than SiBCASA soil moisture. While SiBCASA soil
moisture has a significant, physically meaningful auto-correlation with lag
times up to 10–17 days (
Variables possibly explaining the temporal correlation
coefficient of SiBCASA and ASCAT soil moisture:
On the positive side, in Figs. 5–8 we found a quite consistent pattern in springtime, which indicates that ASCAT soil moisture was larger than SiBCASA soil moisture, and assimilation seemed to improve the match with in situ observed soil moisture. Is this a realistic pattern? Experimentalists confirm that ponding after snowmelt occurs on the sites. However, it is known that ASCAT soil moisture is unreliable when the footprint of the observation is (partially) covered with snow, ice, or surface water, which is likely to happen in springtime. At the same time, SiBCASA soil moisture in spring depends on the amount of snow accumulated in the winter, the time of snowmelt, the fate of the meltwater on frozen ground (runoff or ponding). Since it is hard to simulate these processes correctly, also considering the coarse resolution of SiBCASA relative to dependency of these processes on topography, springtime soil moisture in SiBCASA may also be questioned. Nevertheless, this springtime underestimation pattern is also observed at other steppe and forest grid cells where the temporal correlations are large. Thus there are indications that the spring wetting with assimilation of ASCAT data in SiBCASA improves the soil moisture. Field workers (see author contributions) confirm the springtime water logging and ponding at the four sites. High water tables during springtime are followed by drying out of the soil, depending on the weather conditions. The low soil moisture in SiBCASA could be caused by overestimation of the evapo-transpiration rates in the spring.
In an attempt to explain the variation in temporal correlation coefficients
over the region, Fig. 12 shows the temporal correlation coefficient of
SiBCASA and ASCAT soil moisture in August 2013 as a function of several
variables. Each dot in the figures represents a grid point. With increasing
LAI the correlation coefficient
In conclusion, (partial) ponding of the soil appears to be a good potential explanation of why the poor performance of ASCAT soil moisture improves into the summer months in Boreal Eurasia. The presence of dense leafs rather than aboveground biomass disturbs the satellite signal.
It has been shown in Sect. 3.2 that assimilation of ASCAT soil moisture in SiBCASA has an effect of 5 to 10 % on GPP and TER, and of a few tens of percent on NEE, at the site of Yakutsk, over the entire year. This represents the higher end of the range, since the effect of assimilation on soil moisture and carbon fluxes was relatively large in Yakutsk. The reason why Yakutsk is so sensitive is because the plant-available water fraction is small there, so that the drought sensitivity is large (Fig. 1a). Integrated over the entire region, assimilation causes changes in the order of half the inter-annual variability, or 2 % of the mean annual NEE. We consider this quite large, given the fact that we only applied the assimilation to the top soil moisture. However, the temporal correlation coefficients were quite low in large parts of the region (Fig. 4). This implies that simulated and observed soil moisture are quite different. Assimilation will thus have a large effect when the observational errors are small. A comparison between observed and simulated NEE is made and discussed in the Supplement (Fig. S4).
The effect of changing soil moisture on GPP is largest in SiBCASA when the plant-available water fraction is smaller than 0.3. The area where this occurs is confined to the steppe zone in Southwest Siberia in South European Russia, where it is dry and in the northeast Siberian forest zone, where water availability is limited by permafrost. If the drought stress function in Fig. 1a were to be defined more linearly, the effects of soil moisture would be spread more evenly over the study domain. Note that it may not be realistic to prescribe identical water stress formulations for all biome types, as SiBCASA does. Furthermore, Fig. 10 shows that the drought sensitivity in Fig. 1 only represents the potential drought sensitivity. The actual sensitivity of GPP to change in soil moisture also depends on the temperature, radiation, and vapour pressure deficit (Fig. 10). This applies to TER in a similar way too. As a result, changes in NEE are not linearly dependent on the change in soil moisture due to assimilation of satellite observed soil moisture. Consequently, local effects may be much larger than 2 % of the mean annual NEE. Furthermore, Ohta et al. (2014) show that, in reality, water logging at high plant-available water fractions may also reduce photosynthesis rates and affect the water use efficiency.
The spatial and temporal correlation between SiBCASA and ASCAT soil moisture are considerable in the summer period and the steppe zone. However, ASCAT-derived soil moisture fails to detect the eight major droughts observed in situ at four sites during 7 years, while SiBCASA reproduces half of those droughts. At site level, temporal correlations between SiBCASA and in situ observed soil moisture are larger than between SiBCASA and ASCAT soil moisture. These facts suggest that SiBCASA soil moisture is more reliable than ASCAT soil moisture at those four locations and that assimilation of ASCAT soil moisture does not improve SiBCASA soil moisture.
The temporal correlation between SiBCASA and ASCAT soil moisture is best in the steppe zone, and in a selection of forest locations where LAI is low, soil temperature is high, and soil moisture is low (Fig. 12). Unfortunately, we do not have ground observations to prove whether assimilation in such conditions would lead to improved soil moisture in SiBCASA.
There is evidence that assimilation of ASCAT soil moisture improves the match of SiBCASA soil moisture with in situ observations in springtime (Figs. 5–8). However, these results should be taken carefully, because ice and ponding occur often in the spring. Irrespective of the question of whether assimilation improves soil moisture in SiBCASA, assimilation of ASCAT soil moisture causes considerable changes in GPP, TER, and NEE. At individual locations these changes may reach up to 5 to 10 % of annual GPP and TER, and tens of percent of annual NEE, and integrated over the entire region, the changes cause changes of the order of half the inter-annual variability in NEE or 2 % of annual NEE.
Ultimately, this study shows that assimilation of satellite-observed soil moisture in vegetation models potentially has large impacts on the simulated carbon fluxes, but that further research is needed to clarify when, where, and in which conditions assimilation leads to more reliable soil moisture simulations. In the near future, important improvements in the quality and spatial resolution of soil moisture are expected to be realized with the SMAP L-band instrument and Sentinel-1. Additionally, the benefit of more advanced assimilation techniques, e.g. by assimilating low-pass filtered satellite signals, may be investigated.
M. K. van der Molen implemented the assimilation scheme and performed the model runs and analysis together with I. R. van der Velde. R. A. M. de Jeu and W. Wagner contributed by discussing the reliability of the satellite-observed soil moisture data. W. Peters and M. C. Krol contributed by discussing the research plan and results. The in situ soil moisture observations were collected and provided by P. Kolari (Hyytiälä), J. Kurbatova, A. Varlagin (Tver), T. C. Maximov, A. V. Kononov, T. Ohta, A. Kotani (Yakutsk and Elgeeii), who also contributed by discussing the reliability of the in situ observations.
The research on which this publication is based has been partly funded by the Dutch Organization for Scientific Research (NWO) under grant number 864.08.012 (VIDI WP “A multiple constraint data assimilation system for the carbon cycle”). J. Kurbatova and A. Varlagin, and the in situ observations in Tver were funded by the Russian Science Foundation, Grant 14-27-00065. Edited by: P. Saco