J. Meteor. Res.  2018, Vol. 32 Issue (6): 862-880   PDF    
http://dx.doi.org/10.1007/s13351-018-8055-0
The Chinese Meteorological Society
0

Article Information

ZHANG, Guo, Jianduo LI, Xinyao RONG, et al., 2018.
An Assessment of CAMS-CSM in Simulating Land–Atmosphere Heat and Water Exchanges. 2018.
J. Meteor. Res., 32(6): 862-880
http://dx.doi.org/10.1007/s13351-018-8055-0

Article History

Received April 24, 2018
in final form July 26, 2018
An Assessment of CAMS-CSM in Simulating Land–Atmosphere Heat and Water Exchanges
Guo ZHANG, Jianduo LI, Xinyao RONG, Yufei XIN, Jian LI, Haoming CHEN, Jingzhi SU, Lijuan HUA     
State Key Laboratory of Severe Weather, Chinese Academy of Meteorological Sciences, Beijing 100081
ABSTRACT: The Chinese Academy of Meteorological Sciences (CAMS) has been devoted to developing a climate system mo-del (CSM) to meet demand for climate simulation and prediction for the East Asian region. In this study, we evaluated the performance of CAMS-CSM in regard to sensible heat flux (H), latent heat flux (LE), surface temperature, soil moisture, and snow depth, focusing on the Atmospheric Model Intercomparison Project experiment, with the aim of participating in the Coupled Model Intercomparison Project phase 6. We systematically assessed the simulation results achieved by CAMS-CSM for these variables against various reference products and ground observations, including the FLUXNET model tree ensembles H and LE data, Climate Prediction Center soil moisture data, snow depth climatology data, and Chinese ground observations of snow depth and winter surface temperature. We compared these results with data from the ECMWF Interim reanalysis (ERA-Interim) and Global Land Data Assimilation System (GLDAS). Our results indicated that CAMS-CSM simulations were better than or comparable to ERA-Interim reanalysis for snow depth and winter surface temperature at regional scales, but slightly worse when simulating total column soil moisture. The root-mean-square differences of H in CAMS-CSM were all greater than those from the ERA-Interim reanalysis, but less than or comparable to those from GLDAS. The spatial correlations for H in CAMS-CSM were the lowest in nearly all regions, except for North America. CAMS-CSM LE produced the lowest bias in Siberia, North America, and South America, but with the lowest spatial correlation coefficients. Therefore, there are still scopes for improving H and LE simulations in CAMS-CSM, particularly for LE.
Key words: Climate System Model of the Chinese Academy of Meteorological Sciences     Atmospheric Model Intercomparison Project     sensible heat flux     latent heat flux    
1 Introduction

The land surface is an important component of the earth’s system and acts as a transitional layer between the boundary layer and the subsurface soil column. Exchanges of heat, water, and carbon between the land and the atmosphere are significantly important in hydrological and biological processes (Li et al., 2017; Valayamkunnath et al., 2018). Accurate prediction of the turbulent transfer of heat and water from the land surface has been identified as being very important in numerical weather prediction and climate modeling studies, because the land surface and atmospheric processes were strongly connected (Betts et al., 1997; Pielke et al., 2002).

Sensible heat flux (H) is a major heat source component, and its effect has recently been extensively investigated, particularly with regard to rapid climate change in different regions (Yanai and Wu, 2006; Yang et al., 2011). Latent heat flux (LE), including bare soil evaporation, canopy-intercepted water evaporation, and transpiration, has been directly linked to the amount of water evaporated from the land surface into the atmosphere (Valayamkunnath et al., 2018).

Soil moisture is a crucial variable in hydrological processes, controlling the partitioning of energy in H and LE at the land–atmosphere interface, and affecting surface albedo and ground thermal heat capacity (Dorigo et al., 2015). Thus, it is important to understand the spatial distribution and temporal changes of soil moisture at the regional scale (Qin et al., 2017).

Accurate estimation of snowpack states is crucial for improving initial conditions for numerical weather forecasts, climate prediction model simulation, water resources management, etc. Because snow has special properties, such as high reflectivity and low thermal conductivity (Essery, 1997), it can influence the exchange of energy and water between the atmosphere and land, and can eventually lead to changes in atmospheric circulation and climate (Xia et al., 2014).

There have been many investigations on the impacts of these land surface model-related variables, on the global or regional scale, in the Earth System Models (ESMs) in the previous phases of the Coupled Model Intercomparison Project (CMIP). For instance, Mueller and Seneviratne (2014) analyzed global evapotranspiration (ET) from the historical CMIP5 and twentieth century CMIP3 simulations. They indicated that the identified biases in ET could explain systematic biases in temperature noted for many of the considered regions. Li et al. (2018) evaluated global land surface models (LSMs), from 12 ESMs in CMIP5, regarding their simulations of gross primary production (GPP) and ET, and reported that the majority of models overestimated GPP and ET. Li et al. (2014) analyzed the simulated responses of land surface processes on the Tibetan Plateau to global warming, from the BCC_CSM1.1 simulations of CMIP5.

In this regard, evaluations of the heat and water transfer simulations from offline land surface models have been conducted as well (Niu and Yang, 2006; Cai et al., 2014; Li et al., 2017), in order to provide a basis for improving LSMs.

In our study, we evaluated the model performance of the Climate System Model of the Chinese Academy of Meteorological Sciences (CAMS-CSM) on surface heat flux (H and LE), surface temperature, total column soil moisture, and snow depth, focusing on the Atmospheric Model Intercomparison Project (AMIP) experiment.

The paper has been organized as follows: models and the observational (reanalysis) data used are described in Section 2, the results are analyzed in Section 3, and conclusions are presented in Section 4.

2 Experiments and data 2.1 Model description

The coupling framework of CAMS-CSM is based on a modified version of the atmospheric general circulation model ECHAM5 (v5.4) developed by the Max Planck Institute for Meteorology coupled with the Modular Ocean Model version 4 (Roeckner et al., 2003; Griffies, 2010). The resolution of the current version is T106L31, which corresponds to a horizontal resolution of ~1°, with 31 vertical layers extending from the surface to 10 hPa. Detailed descriptions of the ECHAM5 model and CAMS-CSM can be found in Roeckner et al. (2003) and Rong et al. (2018), respectively.

In the current version of CAMS-CSM, a two-step Shape Preserving Advection Scheme (TSPAS), for the passive tracer transport (Yu, 1994), and a correlated k-distribution scheme, developed by Zhang et al. (2006a, b) for shortwave and longwave radiation transfer parameterization, have been adopted in ECHAM5. The Common Land Model (CoLM; Dai et al., 2003) has been employed as the land component of CAMS-CSM.

The initial CoLM was a combination of the best features of the biosphere–atmosphere transfer scheme (BATS; Dickinson et al., 1993), the Chinese Academy of Sciences Institute of Atmospheric Physics LSM (IAP94; Dai and Zeng, 1997), and the NCAR LSM (Bonan, 1996). Compared to the initial version, the current CoLM was developed with an improved two-stream approximation model of radiation transfer in the canopy, a photosynthesis–stomatal conductance model (Ji andDai, 2010), an unfrozen water process (Niu and Yang, 2006), a quasi-Newton–Raphson method for sunlit and shaded leaves, etc. (Ji et al., 2014).

2.2 Numerical experiments

The focus of this study was the mean monthly outputs from the AMIP simulations of CAMS-CSM. The forcing used for the AMIP simulation was from the CMIP6 AMIP input dataset (Eyring et al., 2016) (https://esgf-node.llnl.gov/projects/input4mips), including sea surface temperature, ozone, solar forcing, and greenhouse gases, as well as anthropogenic and stratospheric aerosols. CAMS-CSM was started with a resting ocean with a climatological temperature and salinity, the initial atmospheric conditions of 1 January 1989, and the land states from a separate 20-yr pre-spinup run with a land model coupled to a standalone atmospheric model. In our discussion of the results, we focused mainly on comparing CAMS-CSM and ERA-Interim, based on their simulations for a 30-yr period (1982–2011), during which more observational data were available. ForH and LE, the results from the GLDAS dataset, in which H and LE were from offline land scheme simulations, were presented as a reference as well.

2.3 Validation data

The datasets used for the evaluation of CAMS-CSM LSM-related simulations have been listed in Table 1. The variables were evaluated with available observations that we acquired.

Table 1 Evaluated variables and validation and reference datasets
Variable Validation dataset Reference dataset
Soil moisture (mm) CPC ERA-Interim
Snow depth (m) ETAC, CMA ERA-Interim
Surface temperature (°C) CMA ERA-Interim
H, LE (W m–2) FLUXNET MTE ERA-Interim, GLDAS

The spatial distribution of soil moisture can be determined from many in situ observations, but it is challenging to scale these up to a regional scale, since the heterogeneity of factors such as soil texture and topography affects soil moisture variability (Su et al., 2016; Wang et al., 2016). Many studies have attempted to use LSMs (Robock et al., 2000; Dirmeyer et al., 2004) and a land data assimilation system (Dorigo et al., 2012; Chen et al., 2013) to obtain high-resolution products in place of in-situ soil moisture observations. However, as mentioned in previous studies (Koster et al., 2009; Zhang et al., 2014), different LSMs may have different soil moisture dynamic regimes—because of the various methods used to deal with evaporation and runoff, i.e., the simulated soil moisture state is a model-specific quantity. Generally speaking, soil moisture dynamics in LSMs are tuned more towards capturing observed surface heat fluxes, and consequently LSMs do not necessarily exactly reproduce observed soil moisture.

In this regard, monthly gridded soil moisture data from 1979 to 2016 with a horizontal resolution of 0.5° × 0.5° were acquired from the NOAA Climate Prediction Center (CPC) [available online https://www.esrl.noaa.gov/psd/; Fan and van den Dool (2004)]. This soil moisture data were model-calculated and not directly measured either, however, the difference is that this model is a one-layer ‘‘bucket’’ water balance model (Huang et al., 1996), while the driving input fields were CPC monthly global precipitation over land, which used over 17,000 gauges worldwide, and monthly global air temperature from global reanalysis. The model needed only air temperature and precipitation as inputs. Thus, it reduced the model uncertainties from other forcing data in LSMs.

Long-term monthly mean, 0.5° × 0.5° gridded, snow depth data were acquired from the United States Air Force (USAF) / Environmental Technical Applications Center (ETAC), whose global snow depth climatology has been considered as one of the most credible sources of snow depth data. It was obtained based on a set of station measurements of snow depth (Foster and Davy, 1988), and has been used in several studies to validate snow depth simulations (Roesch and Roeckner, 2006; Niu and Yang, 2007; Xia et al., 2014; Li et al., 2017). In addition, Chinese ground observations of daily snow depth and surface temperature from ~2400 stations, collected by the National Meteorological Information Center of China Meteorological Administration (CMA), were used to further validate CAMS-CSM simulations.

Monthly, 0.5° × 0.5° gridded, surface H and LE (Jung et al., 2011), based on the model tree ensemble (MTE) upscaling of FLUXNET eddy covariance measurements, from the Max Planck Institute for biogeochemistry (available online https://www.bgc-jena.mpg.de/geodb/projects/Home.php), were used. This dataset has been widely employed to evaluate model simulations (Ji et al., 2014; Xia et al., 2016; Ma et al., 2017), and satellite-derived water and energy exchanges (Frankenberg et al., 2011).

There are a number of products of land surface heat flux and water budget components provided by numerical model simulations, such as the Global Land Data Assimilation System (GLDAS; Rodell et al., 2004), NCEP (Kalnay et al., 1996), ECMWF Interim reanalysis (ERA-Interim; Dee et al., 2011), Japanese 55-yr Reanalysis (JRA55; Ebita et al., 2011), etc. In this study, the model performance of CAMS-CSM was mainly compared to those of the ERA-Interim reanalysis and GLDAS dataset, because they have been widely used in validation studies (Ji et al., 2014; Ma et al., 2017).

ERA-Interim is the latest global atmospheric reanalysis produced by the ECMWF (Dee et al., 2011). The ERA-Interim reanalysis was produced with the ECMWF fixed Integrated Forecasting System (IFS), which incorporated a forecast model with three fully coupled components for the atmosphere, land surface, and ocean waves. It has the advantage of being consistent over the whole period from 1979 onwards, and by design, reanalysis products have been more suitable than their operational counterparts for use in climatic studies (Albergel et al., 2012).

The land scheme used in ERA-Interim was the tiled ECMWF scheme for surface exchange over land (TESSEL; Van den Hurk et al., 2000). Snow depth, snow water equivalent, and snow density estimates generated by the forecast model were updated, based on a Cressman analysis of station observations of snow depth and (when available) snow cover data, from satellites. A detailed description of the ERA-Interim product archive has been provided by Berrisford et al. (2009).

GLDAS drives multiple, offline (not coupled to the atmosphere) land schemes, integrates a huge quantity of observation based data, and executes globally at high resolutions (2.5° to 1 km), enabled by the Land Information System (LIS; Kumar et al., 2006). In this study, 1.0° × 1.0°, monthly H and LE data generated by GLDAS with the Noah land scheme were used.

2.4 Evaluation methods

All simulations and reanalysis were mapped onto a 0.5° × 0.5° spatial grid, using bilinear interpolation to match the grid of validation data. Evaluations of the simulated seasonal cycles were completed for six selected regions (as shown in Fig. 1), including the following: (1) the Tibetan Plateau (25°–40°N, 80°–100°E), (2) Siberia (50°–65°N, 60°–130°E), (3) North America (13°–72°N, 168°–56°W), (4) South America (53°S–12.5°N, 81°–35°W), (5) the Indo-China Peninsula (10°–20°N, 92°–110°E), and (6) the Indian subcontinent (8°–30°N, 65°–90°E).

Figure 1 Locations of the six focused regions. A: Siberia, B: Tibetan Plateau, C: Indian subcontinent, D: Indo-China Peninsula, E: North America, and F: South America.

Spatial correlations (COR), the mean bias (BIAS, model simulations minus validation data), and root mean square difference (RMSD) were used for quantitative evaluation.

3 Results and discussion 3.1 Soil moisture

CAMS-CSM coupled with CoLM defines 10 soil layers: 0–1.8, 1.8–4.5, 4.5–9.1, 9.1–16.6, 16.6–28.9, 28.9–49.3, 49.3–82.9, 82.9–138.3, 138.3–229.6, and 229.6–343.3 cm. Four soil layers are used in the land scheme of ERA-Interim reanalysis: 0–7, 7–28, 28–100, and 100–289 cm.

Because the soil layers between the two models have been defined differently, it is difficult to compare soil moisture in any one layer. Therefore, we compared the total soil moisture in CAMS-CSM and ERA-Interim to the CPC monthly gridded soil moisture in a single 160-cm column. We assumed in CAMS-CSM and ERA-Interim reanalysis that, in this case, the soil moisture in one layer was evenly distributed. Therefore, added soil moisture from the eight CAMS-CSM layers (138.3 cm in total) and at a depth of 21.7 cm (i.e., 160 minus 138.3 cm) from the 9th layer, and added soil moisture from the three ERA-Interim layers (100 cm in total) and at a depth of 60 cm from the 4th layer, were compared with CPC soil moisture.

The spatial distributions of CPC soil moisture, and the BIAS in the CAMS-CSM simulation and ERA-Interim reanalysis, are shown in Fig. 2. Overall, there was more soil moisture in middle and low latitude regions, except in areas of the Saharan Desert and Asian plateau (Fig. 2a). The pattern of soil moisture distribution in CAMS-CSM was similar to that for ERA-Interim reanalysis (Figs. 2b, c). Both of them produced more soil moisture in most regions, except in southern North America, most areas of South America, and the maritime continent. The area of soil moisture underestimation in CAMS-CSM in North America was larger than that in ERA-Interim reanalysis. In addition, CAMS-CSM underestimated soil moisture in northwestern Europe and central Africa.

Figure 2 Annual mean spatial distributions for total soil moisture (mm) in a 160-cm column, using (a) a dataset from the CPC, and its bias (mm) in (b) CAMS-CSM and (c) ERA-Interim reanalysis.

Overall, compared to ERA-Interim reanalysis, CAMS-CSM produced more soil moisture in the high latitudes, but less soil moisture in other regions. This was consistent with previous experimental findings (Mahfouf and Noilhan, 1991; Balsamo et al., 2011; Albergel et al., 2012), which found that ERA-Interim tended to overestimate soil moisture, particularly for arid areas, although it worked well in capturing surface soil moisture variability.

The seasonal variations of latitudinal mean soil moisture in the CPC dataset and BIAS in CAMS-CSM and ERA-Interim reanalysis are shown in Fig. 3. Overall, the CPC total soil moisture in the Southern Hemisphere was greater than that in the Northern Hemisphere (Fig. 3a). Only in a belt between approximately 20°S and 10°N was there a clear seasonal variation, with higher soil moisture during the first half of the year (Fig. 3a). Both CAMS-CSM and ERA-Interim reanalysis overestimated the soil moisture in most areas, particularly ERA-Interim reanalysis (Figs. 3b, c). Only in regions around the equator and 45°S was the bias smaller for both. The BIAS in the Southern Hemisphere was smaller than for the Northern Hemisphere. Compared to ERA-Interim reanalysis, CAMS-CSM provided better soil moisture estimates around 30°S during the first half of the year and around midlatitudes in the Northern Hemisphere (Figs. 3b, c). In other places, CAMS-CSM was comparable to or slightly worse than ERA-Interim reanalysis.

Figure 3 The seasonal evolution of latitudinal mean soil moisture (mm) in a 160-cm column in (a) a dataset from CPC and its bias (mm) with (b) CAMS-CSM and (c) ERA-Interim reanalysis.

Seasonal variations of total soil moisture in the six focused regions are shown in Fig. 4. There were seasonal variations of CPC soil moisture, with higher values during the summer on the Tibetan Plateau, in South America, on the Indo-China Peninsula, and on the Indian subcontinent. No obvious seasonal change was observed in Siberia or North America.

Figure 4 Climatological annual cycles of the simulated and validated total soil moisture (mm) in a 160-cm colomn for (a) the Tibetan Plateau, (b) Siberia, (c) North America, (d) South America, (e) the Indo-China Peninsula, and (f) the Indian subcontinent.

Both CAMS-CSM and ERA-Interim reanalysis depicted the seasonal pattern well, although both of them overestimated the total soil moisture in a 160-cm column for all six regions—probably because both schemes in CAMS-CSM and ERA-Interim reanalysis assumed the soil column bottomed on to bedrock (Viterbo and Beljaars, 1995; Dai et al., 2003). CAMS-CSM and ERA-Interim reanalysis produced the best simulation of soil water in South America (Fig. 4d), with the least bias and RMSDs (Table 2). Compared to ERA-Interim reanalysis, CAMS-CSM produced slightly more soil moisture for nearly all the regions, except Siberia (Fig. 4), with larger bias and RMSDs, and comparable COR (Table 2). Li et al. (2017) also found that offline CoLM underestimated soil moisture in Siberia, relative to ERA-Interim reanalysis.

Table 2 Statistical summary of the total soil moisture in a 160-cm colomn for different regions
Tibetan Plateau Siberia North America
BIAS (mm) RMSD (mm) COR BIAS (mm) RMSD (mm) COR BIAS (mm) RMSD (mm) COR
CAMS-CSM 251.43 288.10 0.40 202.85 243.00 0.34 155.37 252.08 0.07
ERA-Interim 239.02 257.07 0.80 255.37 261.59 0.48 154.96 227.58 0.05
South America Indo-China Peninsula Indian subcontinent
BIAS (mm) RMSD (mm) COR BIAS (mm) RMSD (mm) COR BIAS (mm) RMSD (mm) COR
CAMS-CSM 70.01 141.41 0.64 108.55 162.64 0.10 167.66 210.01 0.44
ERA-Interim 58.12 139.97 0.59 91.05 172.75 0.06 144.32 184.35 0.44

Vertical water flow within soil in the land schemes used in CAMS-CSM and ERA-Interim was similar. Darcy’s law covered both of them, and the soil properties were calculated by using the parametric relations of Clapp and Hornberger (1978). The main difference was that ERA-Interim scheme specified only one soil type everywhere to represent the average response of medium textured soils and it is quite possible that the properties of the soil type in Siberia may be significantly different from the average soil properties of medium textured soils. Thus, the soil moisture in CAMS-CSM was systematically less than that in ERA-Interim reanalysis.

3.2 Snow depth

Figure 5 shows the spatial distribution of snow depth from the ETAC dataset, and its BIAS in CAMS-CSM and ERA-Interim reanalysis. There was greater snow depth from the ETAC climatology at high latitudes (Fig. 5a). CAMS-CSM performed better and slightly overestimated snow depth overall, except that it indicated much more snow on the northwestern Tibetan Plateau compared with ERA-Interim reanalysis (Fig. 5b). ERA-Interim reanalysis produced much more snow in Antarctica and Greenland, but apparently less in other areas (Fig. 5c). The snow scheme used in ERA-Interim was simple, with an explicit snow layer similar to the scheme described in Douville et al. (1995). This scheme tended to underestimate snow depth in both forest and open sites, due to overestimation of snow density (Dutra et al., 2010).

Figure 5 Annual mean spatial distribution of snow depth (m) in (a) ETAC dataset and its bias (m) with (b) CAMS-CSM and (c) ERA-Interim reanalysis.

From the annual cycles of snow depth in different regions (Fig. 6), both CAMS-CSM and ERA-Interim reanalysis depicted the seasonal pattern well on the Tibetan Plateau, in Siberia, and in North America (Figs. 6ac). The timing of snow accumulation was depicted well in CAMS-CSM, but the thaw started one month earlier than in the ETAC dataset, in Siberia and North America (Figs. 6b, c). Swenson and Lawrence (2012) also demonstrated that the parameterization used to determine snow-cover-fraction-based snow depth, in CLM4, exhibited an early thaw.

When the Weather Research and Forecasting /Noah model was applied for multiyear snow simulations, with winter precipitation, it verified well against Natural Resources Conservation Service Snowpack Telemetry (SNOTEL) observations; the snowpack characteristics modeled by the Noah LSM depicted an early snow period end, and an early seasonal maximum snow water equivalent (Barlage et al., 2010; Ikeda et al., 2010).

CAMS-CSM provided better snow depth estimates for Siberia and North America, with fewer errors and a higher spatial correlation, compared with ERA-Interim reanalysis (Figs. 6b, c, and Table 3). It overestimated the snow depth on the Tibetan Plateau, with a higher bias and RMSD.

Table 3 Statistical summary of snow depth for different regions
Tibetan Plateau Siberia
BIAS (m) RMSD (m) COR BIAS (m) RMSD (m) COR
CAMS-CSM 0.0305 0.1100 0.41 –0.0028 0.0923 0.57
ERA-Interim –0.0053 0.0196 0.65 –0.1013 0.1666 0.35
North America South America
BIAS (m) RMSD (m) COR BIAS (m) RMSD (m) COR
CAMS-CSM 0.0083 0.1477 0.55 –0.0102 0.0638 0.33
ERA-Interim –0.0803 0.2039 0.35 –0.0104 0.0609 0.53

The start time of snow accumulating and melting in ERA-Interim reanalysis agreed well with ETAC in Siberia and North America, but the amount of snow was significantly underestimated. ERA-Interim reanalysis did not depict the seasonal change in snow depth on the Tibetan Plateau well, and produced almost no snow during the entire year (Fig. 6a). The current ERA-Interim snow-depth analysis relied on real-time observations of snow depth, the short-range forecast, and snow-depth climatic data. This analysis did not make use of first guess, model-generated values.

Figure 6 Annual cycles of simulated and validated snow depths (m) for (a) theTibetan Plateau, (b) Siberia, (c) North America, and (d) South America.

It was difficult to evaluate the snow scheme through its performance, since its initial state may have already been largely in error, especially in regions of poor data coverage or regions, which did not report snow or precipitation. In Armstrong and Brodzik (2001), the Tibetan Plateau was characterized by intermittent and patchy snow cover. The absence of reliable real-time observations, and the existence of questionable climatic data, made this area extremely difficult to analyze with the snow scheme in ERA-Interim. The National Oceanic and Atmospheric Administration/National Environmental Satellite, Data, and Information Service (NOAA/NESDIS) snow cover product has been used to constrain the ERA-Interim analysis since July 2003, but it made no difference to the average pattern for the period 1982–2011.

In addition, Chinese ground observations of snow depth from the website (http://data.cma.cn/) are shown in Fig. 7. The maximal observed average snow depth over ~2400 stations in China was 0.1047 m, with an average value of 0.0037 m. The maximal simulated average snow depths from CAMS-CSM and ERA-Interim reanalysis at observation locations were 0.0904 and 0.0490 m, with averaged values of 0.0048 and 0.0012 m, respectively. The mean biases of snow depth from CAMS-CSM and ERA-Interim were 0.0011 and –0.0024 m, respectively, but their RMSDs were all 0.0082 m.

Overall, CAMS-CSM overestimated snow depth, while ERA-Interim reanalysis significantly underestimated it. CAMS-CSM captured the heavy snow events of northwestern China and northeastern China well, but slightly overestimated the amount of snow in northeastern China, while underestimating that for northwestern China (Fig. 7b). CAMS-CSM significantly overestimated snow depth at the southern edges of the Tibetan Plateau.

Figure 7 Comparison of snow depth (m) from (a) China ground observations, (b) CAMS-CSM simulations, and (c) ERA-Interim reanalysis at ~2400 observation locations.

The annual mean spatial correlation coefficients for snow depth, for CAMS-CSM and ERA-Interim reanalysis, were 0.73 and 0.80, respectively. CAMS-CSM had a lower spatial correlation coefficient than ERA-Interim reanalysis. The reason is probably that CAMS-CSM produced more snow events at the southern edge and downstream region of the Tibetan Plateau.

As indicated in previous studies (Roesch and Roeckner, 2006; Xia et al., 2014), the correctness of snow thickness simulations was closely connected to temperature and precipitation. Therefore, the Chinese ground observation data of surface temperature in winter, from the website, have also been shown in Fig. 8.

Figure 8 Comparison of winter surface temperature (°C) from (a) Chinese ground observations, and simulations at observation locations in China from (b) CAMS-CSM and (c) ERA-Interim reanalysis.

The observed average minimum surface temperature was –25.49°C among ~2400 stations in China, with an average of 1.97°C. The simulated mean minimum surface temperatures in winter, from CAMS-CSM and ERA-Interim reanalysis, at observation locations, were –23.37 and –22.81°C, with average values of 1.33 and 1.51°C, respectively. The mean surface temperature biases in winter, from CAMS-CSM and ERA-Interim reanalysis, were –0.61 and –0.47°C, and with RMSD of 2.68 and 3.02°C, respectively. Both CAMS-CSM and ERA-Interim produced warmer minimum surface temperatures in winter, but colder maximum surface temperatures.

Both CAMS-CSM and ERA-Interim reanalysis captured well the spatial pattern of winter surface temperature, with spatial correlation coefficients of 0.96 and 0.94, respectively. CAMS-CSM performed slightly better, with less RMSD and a higher spatial correlation coefficient. However, both CAMS-CSM and ERA-Interim reanalysis underestimated the surface temperature along the southern and southeastern edges of the Tibetan Plateau, and north of Guangdong and Guizhou provinces in China. This significant cold bias over steep topography probably resulted from the overestimated precipitation in these areas (omitted). This was consistent with Ji et al. (2014), who found that the Beijing Normal University Earth System Model (BNU-ESM), coupled with CoLM, exhibited a cold bias in annual mean surface-air temperature over eight selected regions including China.

3.3 Surface heat fluxes

The annual mean spatial distributions of H in CAMS-CSM, ERA-Interim reanalysis and GLDAS were compared to H observations using the Model Tree Ensemble (MTE) algorithm, as shown in Fig. 9. The annual mean MTE H for most areas was at most 40 W m–2 in the mid-to-high latitudes, and ~80 W m–2on the west coast of America, northern and southern Africa, and central Australia (Fig. 9a). These patterns were depicted well in GLDAS, CAMS-CSM, and ERA-Interim reanalysis, although they all underestimated H north of 50° N (Figs. 9bd), indicating that all the LSMs used in GLDAS, CAMS-CSM, and ERA-Interim reanalysis needed further improvement in Alpine regions. GLDAS overestimated H in the relatively high value regions of MTE product (Fig. 9b).

The pattern of CMAS-CSM was similar to that for ERA-Interim reanalysis. Both underestimated H in central Africa (Figs. 9c, d). CAMS-CSM overestimated H in eastern and south western Africa, and Australia (Fig.9c), while ERA-Interim underestimated H on northeastern Indian subcontinent, and on southwestern Tibetan Plateau (Fig. 9d).

Figure 9 Annual mean spatial distributions of H (W m–2) in (a) MTE product, (b) GLDAS, (c) CAMS-CSM, and (d) ERA-Interim.

The seasonal variation of latitudinal mean H and bias are shown in Fig. 10. Note that the label bar in Fig. 10a is different from those in Figs. 10bd. The seasonal variations of H are observed for most areas (higher during the summer and lower during the winter), but with a slight variation of average values of approximately 30–40 W m–2 near the equator (Fig. 10a). GLDAS, CAMS-CSM, and ERA-Interim all underestimated H in the mid-to-high latitudes of the Northern Hemisphere, particularly during the summer (Figs. 10bd). GLDAS overestimated H in most low-to-mid latitude regions, particularly in the area between 0° and 20°N, during the first half of the year (Fig. 10b). CAMS-CSM performed better than ERA-Interim reanalysis in the Northern Hemisphere, but worse in the Southern Hemisphere, and overestimated H in most regions of the Southern Hemisphere (Figs. 10c, d).

Figure 10 Annual variation of monthly zonal mean H (W m–2) in (a) MTE, and its bias in (b) GLDAS, (c) CAMS-CSM, and (d) ERA-Interim.

The annual mean spatial distributions of LE simulated by the different models were compared to LE in the MTE product, as shown in Fig. 11. The annual mean MTE LE north of 30° N was approximately 20–40 W m–2, and approximately 60–100 W m–2 in most low-latitude regions (Fig. 11a). These patterns were well represented in GLDAS, CAMS-CSM, and ERA-Interim reanalysis (Figs. 11bd), but it appeared that GLDAS and ERA-Interim reanalysis overestimated LE in South America and central Africa.

Figure 11 As in Fig. 9, but for latent heat flux (LE; W m–2).

Except for the high LE value belt between 20°S and 5°N, the seasonal variation in LE was clear, and peaked during the summer (Fig. 12a). GLDAS produced a bias of –10 to 10 W m–2 in most regions (Fig. 12b). CAMS-CSM underestimated LE from June to October between 0° and 15°S but overestimated it between 0° and 15°N (Fig. 12c). ERA-Interim overestimated LE in most areas (Fig. 12d).

Figure 12 As in Fig. 10, but for latent heat flux (LE; W m–2).

The annual cycles of H and LE in CAMS-CSM, GLDAS, and ERA-Interim reanalysis for the six regions are shown in Figs. 13 and 14. All of them captured the seasonal patterns well. CAMS-CSM underestimated H on the Tibetan Plateau, in Siberia and North America, and on the Indian subcontinent, and overestimated H in South America and on the Indo-China Peninsula. Particularly during the spring on the Indo-China Peninsula, CAMS-CSM produced more H up to 60 W m–2 compared to the observations (Fig. 13e). ERA-Interim reanalysis underestimated H in all regions, producing less H compared to CAMS-CSM, except on the Tibetan Plateau. The reason was probably that the surface temperature in ERA-Interim reanalysis was slightly lower throughout the year than CAMS-CSM (figure omitted).

GLDAS overestimated H on the Tibetan Plateau, Indo-China Peninsula, and Indian subcontinent. Its patterns on the Indo-China Peninsula and Indian subcontinent were similar to those of CAMS-CSM. The parameter H increased sharply during the spring, and this increase continued until the end of the dry season, before dropping quickly after the onset of the Asian monsoon season (Figs. 13e, f). This discrepancy in H could be partly attributed to large uncertainties in the surface temperature in CAMS-CSM, ERA-Interim reanalysis, and GLDAS. The surface temperatures in CAMS-CSM and GLDAS were much higher than that in ERA-Interim reanalysis during the spring (omitted). In the land scheme used in ERA-Interim, leaf area index (LAI) was set as a constant value (LAI = 4) (Viterbo and Beljaars, 1995), and Balsamo et al. (2011) found that this scheme in ERA-Interim produced large near-surface temperature errors in the midlatitude areas and monsoon regions, which were particularly evident in spring and summer. The higher LAI during spring enhanced evaporation, but reduced H, and indicated cooling.

Figure 13 Annual cycles of simulated and observed H (W m–2) for (a) the Tibetan Plateau, (b) Siberia, (c) North America, (d) South America, (e) Indo-China Peninsula, and (f) Indian subcontinent.

The bias of H, simulated by using CAMS-CSM, was lowest in Siberia and North America, and on the Indian subcontinent (Table 4).The RMSDs of H in CAMS-CSM were all greater than those for ERA-Interim reanalysis, but less than or comparable to those from GLDAS. The spatial correlations of H in CAMS-CSM were the lowest in nearly all regions, except for North America.

Table 4 Statistical summary of H for different regions (units for BIAS and RMSD: W m–2)
Tibetan Plateau Siberia North America
BIAS RMSD COR BIAS RMSD COR BIAS RMSD COR
CAMS-CSM –10.31 20.40 0.46 –8.35 17.87 0.48 –4.74 22.46 0.77
ERA-Interim –7.83 13.61 0.71 –14.16 17.02 0.58 –11.22 21.68 0.72
GLDAS 2.27 30.97 0.56 –19.40 27.52 0.60 –11.63 25.30 0.83
South America Indo-China Peninsula Indian subcontinent
BIAS RMSD COR BIAS RMSD COR BIAS RMSD COR
CAMS-CSM 8.18 28.99 0.62 16.74 30.03 0.22 –9.19 24.49 0.54
ERA-Interim –6.04 16.95 0.76 –10.84 15.90 0.35 –13.26 19.14 0.74
GLDAS –7.32 27.33 0.68 9.87 25.63 0.39 7.85 27.57 0.61

The LE peak tended to lag the H peak by a few months in MTE product (Fig. 14)—probably due to the start time of summer rains. For instance, the onset time of the monsoon season was May and June, on the Indo-China Peninsula and Indian subcontinent, respectively. The corresponding jumps for LE from April and May were quite significant.

CAMS-CSM, GLDAS, and ERA-Interim reanalysis all depicted this pattern well (Fig. 14), with LE in CAMS-CSM the most responsive to the start of the rains. This can be partly attributed to the sensitivity of surface soil moisture to rains (not shown), due to the fact that the CAMS-CSM land scheme had the thinnest first soil layer of the three land schemes (the effective soil depth of first layer in Noah land scheme used in GLDAS was 5 cm). In contrast to H, the LE produced by CAMS-CSM was less than that in ERA-Interim reanalysis, particularly during spring, on the Indo-China Peninsula (Fig. 14).

ERA-Interim reanalysis overestimated LE in all six regions and this was probably one reason why the total soil moisture in ERA-Interim reanalysis was less than that in CAMS-CSM in all selected regions, except in Siberia.

GLDAS simulated the best LE with the lowest bias and RMSDs, and highest spatial correlation coefficients on the Tibetan Plateau, Indo-China Peninsula, Indian subcontinent, and in North America (Table 5). However, the GLDAS LE on the Indo-China Peninsula was similar to that for CAMS-CSM, apparently underestimating LE during the spring (Fig. 14e). CAMS-CSM LE produced the lowest bias in Siberia, North America, and South America, but with the lowest spatial correlation coefficients (Table 5).

Table 5 Statistical summary of LE for different regions (units for BIAS and RMSD: W m–2)
Tibetan Plateau Siberia North America
BIAS RMSD COR BIAS RMSD COR BIAS RMSD COR
CAMS-CSM 11.96 24.77 0.57 0.54 13.99 0.42 1.76 17.04 0.82
ERA-Interim 8.87 18.17 0.87 3.05 7.07 0.72 8.09 18.03 0.89
GLDAS –1.72 16.10 0.92 4.61 12.33 0.51 2.34 14.62 0.92
South America Indo-China Peninsula Indian subcontinent
BIAS RMSD COR BIAS RMSD COR BIAS RMSD COR
CAMS-CSM –5.37 26.52 0.69 –16.37 27.72 0.38 –0.98 27.01 0.64
ERA-Interim 16.34 23.11 0.88 19.03 27.27 0.26 17.55 27.40 0.82
GLDAS 6.94 22.74 0.89 –5.78 25.01 0.45 –1.33 22.02 0.86
Figure 14 As in Fig. 13, but for latent heat flux (LE; W m–2).
4 Conclusions

In this study, the modelling performance of the CAMS-CSM AMIP experiment for H, LE, surface temperature, total column soil moisture, and snow depth was evaluated. These results were compared to the ERA-Interim reanalysis dataset for the period 1982–2011, against global diagnostic data and ground observations. For H and LE, offline land scheme simulations in GLDAS were also evaluated, as a reference.

Both CAMS-CSM and ERA-Interim reanalysis depicted the patterns of soil moisture well, although both of them overestimated the total soil moisture in a 160-cm column, in all the six regions. For soil moisture, among the six selected regions, they produced the best simulation in South America.

CAMS-CSM outperformed ERA-Interim reanalysis for snow depth, but slightly overestimated it overall. Both CAMS-CSM and ERA-Interim reanalysis depicted the seasonal pattern well on the Tibetan Plateau, and in Siberia and North America. The timing of snow accumulation was depicted well by CAMS-CSM, but the start of the thaw was one month earlier than that of the ETAC dataset for Siberia and North America. CAMS-CSM provided better snow depth in Siberia and North America, with lower errors and higher spatial correlation compared to that of ERA-Interim reanalysis. Overall, CAMS-CSM overestimated the snow depth, but ERA-Interim reanalysis significantly underestimated it, when averaged at all observation locations in China. CAMS-CSM had a lower spatial correlation coefficient than that of ERA-Interim reanalysis, but with a lower bias and comparable RMSD. The winter surface temperature of CAMS-CSM, averaged over all observation locations in China, exhibited a lower RMSD and a higher spatial correlation coefficient.

CAMS-CSM was better at simulating H than ERA-Interim reanalysis in the Northern Hemisphere, but worse in the Southern Hemisphere, overestimating H in most regions there. GLDAS, CAMS-CSM, and ERA-Interim reanalysis all produced lower spatial correlation coefficients for H compared to those for LE. The RMSDs of H in CAMS-CSM were all greater than those for ERA-Interim reanalysis, but were less than or comparable to those from GLDAS. The spatial correlations of H in CAMS-CSM were the lowest in nearly all regions, except for North America. GLDAS simulated the best LE, with the lowest bias and RMSDs, and highest spatial correlation coefficients, on the Tibetan Plateau, the Indo-China Peninsula, and the Indian subcontinent, and in North America. However, the GLDAS LE on the Indo- China Peninsula behaved similarly to CAMS-CSM, apparently underestimating LE during the spring. The CAMS-CSM LE produced the lowest bias in Siberia, North America, and South America, but with the lowest spatial correlation coefficients.

References
Albergel, C., P. De Rosnay, G. Balsamo, et al., 2012: Soil moisture analyses at ECMWF: Evaluation using global ground-based in situ observations. J. Hydrometeorol., 13, 1442–1460. DOI:10.1175/JHM-D-11-0107.1
Armstrong, R. L., and M. J. Brodzik, 2001: Recent Northern Hemisphere snow extent: A comparison of data derived from visible and microwave satellite sensors. Geophys. Res. Lett., 28, 3673–3676. DOI:10.1029/2000GL012556
Balsamo, G., S. Boussetta, E. Dutra, et al., 2011: Evolution of land-surface processes in the IFS. ECMWF Newsletter, No. 127, 17–22. doi:10.21957/x1j3i7bz.
Barlage, M., F. Chen, M. Tewari, et al., 2010: Noah land surface model modifications to improve snowpack prediction in the Colorado Rocky Mountains. J. Geophys. Res. Atmos., 115, D22101. DOI:10.1029/2009JD013470
Berrisford, P., D. Dee, K. Fielding, et al., 2009: The ERA-Interim Archive. ECMWF, Reading, UK, 16 pp.
Betts, A. K., F. Chen, K. E. Mitchell, et al., 1997: Assessment of the land surface and boundary layer models in two operational versions of the NCEP Eta model using FIFE data. Mon. Wea. Rev., 125, 2896–2916. DOI:10.1175/1520-0493(1997)125<2896:AOTLSA>2.0.CO;2
Bonan, G. B., 1996: A Land Surface model (LSM version 1.0) for Ecological, Hydrological, and Atmospheric Studies: Technical Description and User’s Guide. NCAR Technical Note: NCAR/TN-417+STR, National Center for Atmospheric Research, Boulder, CO, doi:10.5065/D6DF6P5X.
Cai, X. T., Z. L. Yang, Y. L. Xia, et al., 2014: Assessment of simulated water balance from Noah, Noah-MP, CLM, and VIC over CONUS using the NLDAS test bed. J. Geophys. Res. Atmos., 119, 13751–13770. DOI:10.1002/2014JD022113
Chen, Y. Y., K. Yang, J. Qin, et al., 2013: Evaluation of AMSR-E retrievals and GLDAS simulations against observations of a soil moisture network on the central Tibetan Plateau. J. Geophys. Res. Atmos., 118, 4466–4475. DOI:10.1002/jgrd.50301
Clapp, R. B., and G. M. Hornberger, 1978: Empirical equations for some soil hydraulic properties. Water Resour. Res., 14, 601–604. DOI:10.1029/WR014i004p00601
Dai, Y. J., and Q. C. Zeng, 1997: A land surface model (IAP94) for climate studies Part I: Formulation and validation in off-line experiments. Adv. Atmos. Sci., 14, 433–460. DOI:10.1007/s00376-997-0063-4
Dai, Y. J., X. B. Zeng, R. E. Dickinson, et al., 2003: The common land model. Bull. Amer. Meteor. Soc., 84, 1013–1024. DOI:10.1175/BAMS-84-8-1013
Dee, D. P., S. M. Uppala, A. J. Simmons, et al., 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597. DOI:10.1002/qj.828
Dickinson, R. E., A. Henderson-Sellers, and P. J. Kennedy, 1993: Biosphere-Atmosphere Transfer Scheme (BATS) Version 1E as Coupled to the NCAR Community Climate Model. NCAR Technical Note NCAR/TN-387+STR, National Center for Atmospheric Research, Boulder, CO, doi: 10.5065/D67W6959.
Dirmeyer, P. A., Z. C. Guo, and X. Gao, 2004: Comparison, validation, and transferability of eight multiyear global soil wetness products. J. Hydrometeorol., 5, 1011–1033. DOI:10.1175/JHM-388.1
Dorigo, W., R. de Jeu, D. Chung, et al., 2012: Evaluating global trends (1988–2010) in harmonized multi-satellite surface soil moisture. Geophys. Res. Lett., 39, L18405. DOI:10.1029/2012GL052988
Dorigo, W. A., A. Gruber, R. A. M. de Jeu, et al., 2015: Evaluation of the ESA CCI soil moisture product using ground-based observations. Remote Sens. Environ., 162, 380–395. DOI:10.1016/j.rse.2014.07.023
Douville, H., J. F. Royer, and J. F. Mahfouf, 1995: A new snow parameterization for the Météo-France climate model. Part I: Validation in stand-alone experiments. Climate Dyn., 12, 21–35. DOI:10.1007/BF00208760
Dutra, E., G. Balsamo, P. Viterbo, et al., 2010: An improved snow scheme for the ECMWF land surface model: Description and offline validation. J. Hydrometeorol., 11, 899–916. DOI:10.1175/2010JHM1249.1
Ebita, A., S. Kobayashi, Y. Ota, et al., 2011: The Japanese 55-year reanalysis " JRA-55”: An interim report. SOLA, 7, 149–152. DOI:10.2151/sola.2011-038
Essery, R., 1997: Seasonal snow cover and climate change in the Hadley Centre GCM. Ann. Glaciol., 25, 362–366. DOI:10.3189/S0260305500014282
Eyring, V., S. Bony, G. A. Meehl, et al., 2016: Overview of the Coupled Model Intercomparison Project phase 6 (CMIP6) experimental design and organization. Geosci. Model Dev., 9, 1937–1958. DOI:10.5194/gmd-9-1937-2016
Fan, Y., and H. van den Dool, 2004: Climate Prediction Center global monthly soil moisture data set at 0.5° resolution for 1948 to present. J. Geophys. Res. Atmos., 109, D10102. DOI:10.1029/2003JD004345
Foster, D. F. Jr., and R. D. Davy, 1988: Global Snow Depth Climatology. USAF Environmental Technical Application Center, Scott Air Force Base, Illinois, USA, 48 pp.
Frankenberg, C., J. B. Fisher, J. Worden, et al., 2011: New global observations of the terrestrial carbon cycle from GOSAT: Patterns of plant fluorescence with gross primary productivity. Geophys. Res. Lett., 38, L17706. DOI:10.1029/2011GL048738
Griffies, S. M., 2010: Elements of MOM4p1. GFDL Ocean Group Technical Report No. 6, NOAA/Geophysical Fluid Dynamics Laboratory, 444 pp.
Huang, J., H. M. van den Dool, and K. P. Georgarakos, 1996: Analysis of model-calculated soil moisture over the United States (1931–1993) and applications to long-range temperature forecasts. J. Climate, 9, 1350–1362. DOI:10.1175/1520-0442(1996)009<1350:AOMCSM>2.0.CO;2
Ikeda, K., R. Rasmussen, C. H. Liu, et al., 2010: Simulation of seasonal snowfall over Colorado. Atmos. Res., 97, 462–477. DOI:10.1016/j.atmosres.2010.04.010
Ji, D., L. Wang, J. Feng, et al., 2014: Description and basic evaluation of Beijing Normal University Earth System Model (BNU-ESM) version 1. Geosci. Model Dev., 7, 2039–2064. DOI:10.5194/gmd-7-2039-2014
Ji, D. and Y. Dai, 2010: The Common land model (CoLM) technical guide. [available online at http://globalchange.bnu.edu.cn/download/doc/CoLM/CoLM_Technical_Guide.pdf].
Jung, M., M. Reichstein, H. A. Margolis, et al., 2011: Global patterns of land–atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res. Biogeosci., 116, G00J07. DOI:10.1029/2010JG001566
Kalnay, E., M. Kanamitsu, R. Kistler, et al., 1996: The NCEP/NCAR 40-year reanalysis project. Bull. Amer. Meteor. Soc., 77, 437–472. DOI:10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2
Koster, R. D., Z. C. Guo, R. Q. Yang, et al., 2009: On the nature of soil moisture in land surface models. J. Climate, 22, 4322–4335. DOI:10.1175/2009JCLI2832.1
Kumar, S. V., C. D. Peters-Lidard, Y. Tian, et al., 2006: Land Information System: An interoperable framework for high resolution land surface modeling. Environ. Modell. Softw., 21, 1402–1415. DOI:10.1016/j.envsoft.2005.07.004
Li, C. W., H. Lu, K. Yang, et al., 2017: Evaluation of the common land model (CoLM) from the perspective of water and energy budget simulation: Towards inclusion in CMIP6. Atmosphere, 8, 141. DOI:10.3390/atmos8080141
Li, L. H., Y. P. Wang, V. K. Arora, et al., 2018: Evaluating global land surface models in CMIP5: Analysis of ecosystem water- and light-use efficiencies and rainfall partitioning. J. Climate, 31, 2995–3008. DOI:10.1175/JCLI-D-16-0177.1
Li, Z. C., Z. G. Wei, S. H. Lv, et al., 2014: Effect of land surface processes on the Tibetan Plateau’s past and its predicted response to global warming: An analytical investigation based on simulation results from the CMIP5 model. Environ. Earth Sci., 72, 1155–1166. DOI:10.1007/s12665-013-3034-3
Ma, N., G. Y. Niu, Y. L. Xia, et al., 2017: A systematic evaluation of Noah-MP in simulating land–atmosphere energy, water, and carbon exchanges over the continental United States. J. Geophys. Res. Atmos., 122, 12245–12268. DOI:10.1002/2017JD027597
Mahfouf, J. F., and J. Noilhan, 1991: Comparative study of various formulations of evaporations from bare soil using in situ data. J. Appl. Meteor., 30, 1351–1362. DOI:10.1175/1520-0450(1991)030<1354:CSOVFO>2.0.CO;2
Mueller, B., and S. I. Seneviratne, 2014: Systematic land climate and evapotranspiration biases in CMIP5 simulations. Geophys. Res. Lett., 41, 128–134. DOI:10.1002/2013GL058055
Niu, G. Y., and Z. L. Yang, 2006: Effects of frozen soil on snowmelt runoff and soil water storage at a continental scale. J. Hydrometeorol., 7, 937–952. DOI:10.1175/JHM538.1
Niu, G.-Y., and Z.-L. Yang, 2007: An observation-based formulation of snow cover fraction and its evaluation over large North American river basins. J. Geophys. Res. Atmos., 112, D21101. DOI:10.1029/2007JD008674
Pielke, R. A., G. Marland, R. A. Betts, et al., 2002: The influence of land-use change and landscape dynamics on the climate system: Relevance to climate-change policy beyond the radiative effect of greenhouse gases. Philosophical Transactions of The Royal Society A: Mathematical, Physical and Engineering Sciences, 360, 1705–1719. DOI:10.1098/rsta.2002.1027
Qin, Y. H., T. H. Wu, X. D. Wu, et al., 2017: Assessment of reanalysis soil moisture products in the permafrost regions of the central of the Qinghai–Tibet Plateau. Hydrol. Processes, 31, 4647–4659. DOI:10.1002/hyp.11383
Robock, A., K. Y. Vinnikov, G. Srinivasan, et al., 2000: The global soil moisture data bank. Bull. Amer. Meteor. Soc., 81, 1281–1300. DOI:10.1175/1520-0477(2000)081<1281:TGSMDB>2.3.CO;2
Rodell, M., P. R. Houser, U. Jambor, et al., 2004: The global land data assimilation system. Bull. Amer. Meteor. Soc., 85, 381–394. DOI:10.1175/BAMS-85-3-381
Roeckner, E., G. Bäuml, L. Bonaventura, et al., 2003: The Atmospheric General Circulation Model ECHAM 5. PART I: Model Description. MPI-Report No. 349, Max Planck Institute for Meteorology, Hamburg, 140 pp.
Roesch, A., and E. Roeckner, 2006: Assessment of snow cover and surface albedo in the ECHAM5 general circulation model. J. Climate, 19, 3828–3843. DOI:10.1175/JCLI3825.1
Rong, X. Y., J. Li, H. M. Chen, et al., 2018: The CAMS climate system model and a basic evaluation of its climate state and variability simulation. J. Meteor. Res., 32, 839–861. DOI:10.1007/s13351-018-8058-x
Su, B. D., A. Q. Wang, G. J. Wang, et al., 2016: Spatiotemporal variations of soil moisture in the Tarim River basin, China. International Journal of Applied Earth Observation and Geoinformation, 48, 122–130. DOI:10.1016/j.jag.2015.06.012
Swenson, S. C., and D. M. Lawrence, 2012: A new fractional snow-covered area parameterization for the Community Land Model and its effect on the surface energy balance. J. Geophys. Res. Atmos., 117, D21107. DOI:10.1029/2012JD018178
Valayamkunnath, P., V. Sridhar, W. G. Zhao, et al., 2018: Intercomparison of surface energy fluxes, soil moisture, and evapotranspiration from eddy covariance, large-aperture scintillometer, and modeling across three ecosystems in a semiarid climate. Agric. Forest Meteor., 248, 22–47. DOI:10.1016/j.agrformet.2017.08.025
van den Hurk, B. J. J. M., P. Viterbo, A. C. M. Beljaars, et al., 2000: Offline Validation of the ERA40 Surface Scheme. Technical Memorandum No. 295, ECMWF, Reading, UK, 43 pp.
Viterbo, P., and A. C. M. Beljaars, 1995: An improved land surface parameterization scheme in the ECMWF model and its validation. J. Climate, 8, 2716–2748. DOI:10.1175/1520-0442(1995)008<2716:AILSPS>2.0.CO;2
Wang, A. H., X. B. Zeng, and D. L. Guo, 2016: Estimates of global surface hydrology and heat fluxes from the community land model (CLM4.5) with four atmospheric forcing datasets. J. Hydrometeorol., 17, 2493–2510. DOI:10.1175/JHM-D-16-0041.1
Xia, K., B. Wang, L. J. Li, et al., 2014: Evaluation of snow depth and snow cover fraction simulated by two versions of the flexible global ocean–atmosphere–land system model. Adv. Atmos. Sci., 31, 407–420. DOI:10.1007/s00376-013-3026-y
Xia, Y. L., B. A. Cosgrove, K. E. Mitchell, et al., 2016: Basin-scale assessment of the land surface water budget in the National Centers for Environmental Prediction operational and research NLDAS-2 systems. J. Geophys. Res. Atmos., 121, 2750–2779. DOI:10.1002/2015JD023733
Yanai, M., and G. X. Wu, 2006: Effects of the Tibetan Plateau. The Asian Monsoon, B. Wang, Ed., Springer, Berlin, Heidelberg, Germany, 513–549.
Yang, K., X. F. Guo, and B. Y. Wu, 2011: Recent trends in surface sensible heat flux on the Tibetan Plateau. Sci. China Earth Sci., 54, 19–28. DOI:10.1007/s11430-010-4036-6
Yu, R. C., 1994: A two—step shape—preserving advection scheme. Adv. Atmos. Sci., 11, 479–490. DOI:10.1007/BF02658169
Zhang, G., G. S. Zhou, F. Chen, et al., 2014: A trial to improve surface heat exchange simulation through sensitivity experiments over a desert steppe site. J. Hydrometeorol., 15, 664–684. DOI:10.1175/JHM-D-13-0113.1
Zhang, H., G. Y. Shi, T. Nakajima, et al., 2006a: The effects of the choice of the k-interval number on radiative calculations . Journal of Quantitative Spectroscopy and Radiative Transfer, 98, 31–43. DOI:10.1016/j.jqsrt.2005.05.090
Zhang, H., T. Suzuki, T. Nakajima, et al., 2006b: Effects of band division on radiative calculations. Optical Engineering, 45, 016002. DOI:10.1117/1.2160521