Evaluation of a Conjunctive Surface–Subsurface Process Model (CSSP) over the Contiguous United States at Regional–Local Scales

Xing Yuan Division of Illinois State Water Survey, Institute of Natural Resource Sustainability, University of Illinois at Urbana–Champaign, Urbana, Illinois

Search for other papers by Xing Yuan in
Current site
Google Scholar
PubMed
Close
and
Xin-Zhong Liang Department of Atmospheric Sciences, and Division of Illinois State Water Survey, Institute of Natural Resource Sustainability, University of Illinois at Urbana–Champaign, Urbana, Illinois

Search for other papers by Xin-Zhong Liang in
Current site
Google Scholar
PubMed
Close
Full access

Abstract

This study presents a comprehensive evaluation on a Conjunctive Surface–Subsurface Process Model (CSSP) in predicting soil temperature–moisture distributions, terrestrial hydrology variations, and land–atmosphere exchanges against various in situ measurements and synthetic observations at regional–local scales over the contiguous United States. The CSSP, rooted in the Common Land Model (CoLM) with a few updates from the Community Land Model version 3.5 (CLM3.5), incorporates significant advances in representing hydrology processes with realistic surface (soil and vegetation) characteristics. These include dynamic surface albedo based on satellite retrievals, subgrid soil moisture variability of topographic controls, surface–subsurface flow interactions, and bedrock constraint on water table depths. As compared with the AmeriFlux tower measurements, the CSSP and CLM3.5 reduce surface sensible and latent heat flux errors from CoLM by 10 W m−2 on average, and have much higher correlations with observations for daily latent heat variations. The CSSP outperforms the CLM3.5 over the crop, grass, and shrub sites in depicting the latent heat annual cycles. While retaining the improvement for soil moisture in deep layers, the CSSP shows further advantage over the CLM3.5 in representing seasonal and interannual variations in root zones. The CSSP reduces soil temperature errors from the CLM3.5 (CoLM) by 0.2 (0.7) K at 0.1 m and 0.3 (0.6) K at 1 m; more realistically captures seasonal–interannual extreme runoff and streamflow over most regions and snow depth anomalies in high latitude (45°–52°N); and alleviates climatological water table depth systematic bias (absolute error) by about 1.2 (0.4) m. Clearly, the CSSP performance is overall superior to both the CoLM and CLM3.5. The remaining CSSP deficiencies and future refinements are also discussed.

Current affiliation: Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey.

Current affiliation: Department of Atmospheric and Oceanic Science, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland.

Corresponding author address: Dr. Xin-Zhong Liang, Department of Atmospheric and Oceanic Science, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, 2419 Computer and Space Science, College Park, MD 20742-2452. E-mail: xliang@umd.edu

Abstract

This study presents a comprehensive evaluation on a Conjunctive Surface–Subsurface Process Model (CSSP) in predicting soil temperature–moisture distributions, terrestrial hydrology variations, and land–atmosphere exchanges against various in situ measurements and synthetic observations at regional–local scales over the contiguous United States. The CSSP, rooted in the Common Land Model (CoLM) with a few updates from the Community Land Model version 3.5 (CLM3.5), incorporates significant advances in representing hydrology processes with realistic surface (soil and vegetation) characteristics. These include dynamic surface albedo based on satellite retrievals, subgrid soil moisture variability of topographic controls, surface–subsurface flow interactions, and bedrock constraint on water table depths. As compared with the AmeriFlux tower measurements, the CSSP and CLM3.5 reduce surface sensible and latent heat flux errors from CoLM by 10 W m−2 on average, and have much higher correlations with observations for daily latent heat variations. The CSSP outperforms the CLM3.5 over the crop, grass, and shrub sites in depicting the latent heat annual cycles. While retaining the improvement for soil moisture in deep layers, the CSSP shows further advantage over the CLM3.5 in representing seasonal and interannual variations in root zones. The CSSP reduces soil temperature errors from the CLM3.5 (CoLM) by 0.2 (0.7) K at 0.1 m and 0.3 (0.6) K at 1 m; more realistically captures seasonal–interannual extreme runoff and streamflow over most regions and snow depth anomalies in high latitude (45°–52°N); and alleviates climatological water table depth systematic bias (absolute error) by about 1.2 (0.4) m. Clearly, the CSSP performance is overall superior to both the CoLM and CLM3.5. The remaining CSSP deficiencies and future refinements are also discussed.

Current affiliation: Department of Civil and Environmental Engineering, Princeton University, Princeton, New Jersey.

Current affiliation: Department of Atmospheric and Oceanic Science, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, College Park, Maryland.

Corresponding author address: Dr. Xin-Zhong Liang, Department of Atmospheric and Oceanic Science, and Earth System Science Interdisciplinary Center, University of Maryland, College Park, 2419 Computer and Space Science, College Park, MD 20742-2452. E-mail: xliang@umd.edu

1. Introduction

Land surface and subsurface processes interact with the atmospheric circulation through exchanges of energy, water, and carbon. They define lower boundary conditions with varying degrees of memory that are essential to enhancing weather and climate predictability at time scales from weeks to seasons (Xue et al. 1996; Pielke et al. 1999; Schlosser and Milly 2002; Koster et al. 2004; Dirmeyer 2005; Chen et al. 2010). Thus, land surface models (LSMs) have been continuously developed and greatly improved during the past four decades (Manabe 1969; Dickinson et al. 1986; Sellers et al. 1986, 1996; Xue et al. 1991; Pollard and Thompson 1995; Bonan 1996; Foley et al. 1996; Dai et al. 2003; Oleson et al. 2008). It has been widely accepted that LSMs play an important role, not only in global general circulation models (GCMs; Bonan 1998; Zeng et al. 2002; Dickinson et al. 2006), but also in mesoscale regional climate models (RCMs; Avissar and Pielke 1989; Xue et al. 2001; Chen and Dudhia 2001; Liang et al. 2005a).

After decades of development, there still exist large uncertainties in representing land water and heat storages and related fluxes (Dirmeyer et al. 2006), especially at regional–local scales. These may partially result from inaccurate specification of land surface boundary conditions that define the characteristics and properties of soil and vegetation (Liang et al. 2005a,b; Lawrence and Chase 2007). More importantly, numerical representations of various processes are incomplete. For example, surface albedo greatly influences the surface energy budget and partitioning, but its parameterizations in the state-of-the-art LSMs produce substantial errors from satellite retrievals (Liang et al. 2005c). Even with a perfect bulk albedo, accurate partitioning of the total surface absorbed radiation among soil and canopy components is problematic (Dickinson et al. 2008), especially for the two-big-leaf model (Wang and Leuning 1998; Dai et al. 2004). Recent measurements from flux towers may help to narrow down some model deficiencies (Friend et al. 2007; Stöckli et al. 2008).

Intensive development during the last decade has focused on improving the representation of surface and subsurface hydrological processes. Most current LSMs can simulate water exchanges between aquifers and unsaturated zones (Liang et al. 2003; Yeh and Eltahir 2005; Maxwell and Miller 2005; Niu et al. 2007; Yuan et al. 2008b; Zeng and Decker 2009) and groundwater lateral transports affecting soil moisture redistribution at finer scales (Fan et al. 2007; Maxwell and Kollet 2008). These models may, however, require additional constraints such as bedrock distributions (Choi and Liang 2010) or baseflow estimates (Lo et al. 2008) to better predict water table depth and aquifer flow variations. Some LSMs have incorporated subgrid topographic effects on runoff generation (Famiglietti and Wood 1994; Stieglitz et al. 1997; Chen and Kumar 2001; Niu et al. 2005), but have not accounted for their direct impacts on soil water movement. Another key deficiency is the general lack of explicit treatment for interactions between overland and subsurface flows, which is essential to the hydrological cycle (Govindaraju and Kavvas 1991; LaBolle et al. 2003; Kollet and Maxwell 2006). Surface and subsurface flows are usually calculated independently, and the resulting runoffs are simply removed from the local grid column without recycling. This may cause model errors in infiltration, streamflow, and soil moisture, especially at fine resolutions (Wallach et al. 1997; Choi 2006; Miguez-Macho et al. 2007; Maxwell and Kollet 2008). As a result, leading LSMs still contain large biases. For example, the latest publicly available Community Land Model version 3.5 (CLM3.5) developed at the National Center for Atmospheric Research (NCAR) substantially underestimates water table depth and soil moisture variability in the root zone (Oleson et al. 2008; see also section 6).

Along these lines, an advanced Conjunctive Surface–Subsurface Process Model (CSSP) has been developed as the core LSM of the mesoscale Climate–Weather Research and Forecasting Model (CWRF; Liang et al. 2005a,b,c,d; X.-Z. Liang et al. 2010, unpublished manuscript). The CSSP is rooted in the Common Land Model (CoLM; Dai et al. 2003, 2004) with a few updates from CLM3.5 (Oleson et al. 2008). The most prominent advances of the CSSP include an improved land surface albedo parameterization (Liang et al. 2005c), a scalable representation of subgrid topographic control on soil moisture (Choi et al. 2007), and an explicit treatment of surface–subsurface flow interaction (Choi 2006; Choi and Liang 2010), all of which are built upon realistic distributions of surface (soil and vegetation) characteristics (Liang et al. 2005a,b). The full CCSP coupling with the CWRF and the resulting climate effect will be documented in an upcoming paper (X.-Z. Liang et al. 2010, unpublished manuscript). The present study provides a comprehensive evaluation of the CSSP at regional–local scales over the contiguous United States during 1979–2008 in a standalone mode as driven by atmospheric conditions from an observational reanalysis.

Section 2 gives a brief description of the LSMs used in this study, including the major features of the CoLM and CLM3.5 that are adopted and the most recent updates to the CSSP. Sections 3 and 4 present respectively the model simulations and observational data. Section 5 compares the CSSP simulations against in situ measurements at flux tower sites of distinct surface characteristics. Section 6 elaborates the CSSP performance using a 30-year continuous simulation driven by the observational reanalysis of meteorological conditions, focusing on soil moisture–temperature, runoff–streamflow, and water table depth at regional scales over the contiguous United States. For both evaluations in sections 5 and 6, the CSSP results are compared with those of the CoLM and CLM3.5 to depict the improvements. This is followed by the conclusions and discussions in section 7.

2. Brief model description

a. CoLM

The CoLM was developed by a large scientific community (Dai et al. 2003), and updated by Dai et al. (2004) to incorporate a two-big-leaf (sunlit and shaded leaves) canopy model of Wang and Leuning (1998). The canopy model assumes that the sunlit leaf receives both diffuse and direct radiation while the shaded leaf gets only diffuse light (Spitters et al. 1986). The total solar radiation is thus partitioned among the two big leaves and ground, and their respective energy budget equations are nonlinearly coupled and so jointly solved for corresponding temperatures by an iterative quasi-Newton–Raphson method.

The original CoLM utilizes the two-leaf model during the daytime and the one-leaf model at night. There exist, however, nonconvergent solutions for leaf temperatures at dawn, especially on grid cells with small leaf area index (LAI), causing model abortion. To avoid the numerical instability, we introduce two modifications. First, the one-leaf model is used if local LAI is smaller than both 1 and the stem area index (SAI). Second, at dawn, the absorbed solar radiation is repartitioned between the sunlit and shaded leaves linearly by their cover fractions. In addition, an aquifer model (Choi and Liang 2010) is implemented into the CoLM to facilitate comparison of water table depth.

b. CLM3.5

The CLM3.5 was documented in detail by Oleson et al. (2004, 2008). Its earlier version shares with the CoLM its representation of major physics processes. Here listed are the new hydrological parameterizations that have been incorporated into the CSSP (X.-Z. Liang et al. 2010, unpublished manuscript): 1) introducing the supercooled soil water and fractional impermeable area to determine the freezing process and soil hydraulic properties (Niu and Yang 2006), corrected for unphysical negative soil moisture solution (Choi and Liang 2010); 2) specifying soil water potential at stomata fully open or closed by plant functional types (Thornton and Zimmermann 2007), made consistent with the root distributions (Schenk and Jackson 2002); 3) adding a soil resistance term to limit the excessive evaporation (Sellers et al. 1992; Oleson et al. 2008); 4) adopting the topography controlled fractional saturation area index to represent runoff generation from infiltration, saturation excess, and water table depth changes (Niu et al. 2005); and 5) coupling the unconfined aquifer model (Niu et al. 2007), constrained above by local bedrock.

c. CSSP

The CSSP, as coupled with the CWRF, incorporates the comprehensive land surface boundary conditions based on the best available observational data (Liang et al. 2005a,b). These include surface topographic attributes (mean elevation, slope, curvature, and their standard deviations), surface and base flow directions, bedrock depth, soil sand and clay fraction vertical profiles, land cover category, fractional vegetation cover, and LAI and SAI. It is also built with an advanced dynamic-statistical parameterization of snow-free land surface albedo that significantly reduces biases from MODIS satellite retrievals as compared with the CoLM (Liang et al. 2005c). The dynamic component represents the predictable albedo dependences on solar zenith angle, surface soil moisture, fractional vegetation cover, LAI and SAI, and vegetation type and greenness, while the statistical part depicts the correction for static effects that are specific to local surface characteristics.

As a unique but critical improvement to the CoLM and CLM3.5, the CSSP incorporates a scalable representation of subgrid topographic control on soil moisture (Choi et al. 2007) and an explicit treatment of surface–subsurface flow interaction (Choi 2006; Choi and Liang 2010). The soil moisture is solved by 3D-averaged, localized Richards equation as decomposed into the mean and fluctuation components using the small perturbation approach (Montaldo and Albertson 2003; Kumar 2004). Integration over the model grid box results in a volume-averaged formulation that includes both the grid-resolved mean and subgrid spatial variability for both vertical and lateral subsurface moisture fluxes. The flux contribution from the subgrid variability, generally ignored in most LSMs, was found comparable to that of the mean flux, particularly under drier moisture conditions. Coupled with this 3D subsurface model is a 1D surface routing model that predicts surface water depth in the presence of overland and channel flow by the Saint Venant equations, neglecting local and convective inertia terms (Morita and Yen 2002). The predicted surface water depth affects surface infiltration and thus runoff generation and soil water movement in the subsurface model, which in turn determines surface runoff as the boundary constraint to the surface routing model solution. In this way, the surface and subsurface flows are fully interactive, driving the runoff to reenter the hydrologic cycling process. As discussed later, the surface flow is coupled with groundwater through the variation of soil water.

Another important CSSP feature is its integration of a realistic geographic distribution of bedrock depth (Liang et al. 2005a,b) to improve estimation of the actual soil water capacity. The bedrock depth is defined as the depth of the boundary between soil and unconsolidated material that lies between the land surface and the geologic substratum. Local bedrock topography can be significant for runoff generation (Freer et al. 2002), hillslope–riparian linkage (Katsuyama et al. 2005), and river discharge (Choi and Liang 2010). Most LSMs have, however, generally neglected the bedrock or fixed it at the bottom of the lowest soil layer, with or without adding an unconfined aquifer below. In the present study, soil moisture is predicted only in the layers above the bedrock, while the bedrock is treated as an unconfined aquifer (Niu et al. 2007) with specific yield Sy assumed as 0.2. In Niu et al. (2007), the unconfined aquifer is defined as the part below the model soil column (3.43 m), while in the present study, it is defined as the part within the bedrock, with the upper boundary specified by bedrock depth. The tendency of the total soil water below the water table Ws (mm) is calculated by
e1
where Qr is the recharge rate (mm s−1) representing the exchanges between soil water and groundwater, and Rsb is the base flow or groundwater discharge (mm s−1), which is computed according to Eq. (26) in Choi and Liang (2010). For the case when the water table is above the bedrock, the Qr and the water table depth z (m) are estimated similarly as Choi and Liang (2010). When the water table is within the bedrock layer, the Ws becomes negative, and the recharge rate (now at the soil–bedrock interface) is given as
e2
where ka (mm s−1) is the hydraulic conductivity of the unconfined aquifer, ψjwt (mm) and zjwt (m) are the matric potential and node depth of the soil layer directly above the bedrock, respectively; hence the water table depth is updated as
e3
where dbed (m) is the bedrock depth. The saturated hydraulic conductivity Ks (mm s−1) at depth z (m) is parameterized as
e4
where K0 = 7.0556 × 10−3.884+1.53sand (mm s−1), z* = min(2, dbed), and sand is the soil sand fraction. Note that the Ks is also applied in the bedrock, and the effects of bedrock on the soil water are as follows: 1) if the water table is within bedrock, groundwater can only interact with soil water through the recharge rate at the soil–bedrock interface, so the bedrock is the lower boundary of the soil water; and 2) the bedrock is shallower than 2 m in many regions based on the data used in this study, where it influences the hydraulic conductivity directly. The e-folding length f is defined as (Fan et al. 2007)
e5
where β is the terrain slope; a and b are adjustable parameters specified as respectively 20 and 3000 based on sensitivity experiments in 30-km resolution. Note that our values for parameters a and b are not the same as Fan et al. (2007) because of the different grid resolution and the bedrock treatment. However, the parameters’ dependence on grid resolution needs further investigation as terrain slope varies largely with the calculation scale. Given the above parameterizations, the hydraulic conductivity of the unconfined aquifer is calculated as
e6

3. Model simulations

Two sets of simulations are conducted in this study to evaluate the CSSP against the best available observations at local and regional scales as compared with the CoLM and CLM3.5, respectively. The results will be examined in sections 4 and 5.

The first set is a multiple-year integration driven by the Ameriflux Level 2 meteorological data at tower monitoring sites (http://public.ornl.gov/ameriflux/available.shtml). All models are run in a single-column mode, where the lateral transport processes of surface and subsurface water in the CSSP are turned off. Following Stöckli et al. (2008), the integration at each site is spun up by repeating five cycles for the entire duration of available data (Table 1) to obtain the equilibrium of surface fluxes (Yang et al. 1995). Daily model outputs from the last cycle are used for comparison with concurrent observations.

Table 1.

Information of flux tower sites used in this study. Biome types are grassland (GRA), crop (CRO), closed shrub (CSH), mixed forest (MF), evergreen needleleaf forest (ENF), and deciduous broadleaf forest (DBF).

Table 1.

The second set is a 30-yr integration over the contiguous United States domain driven by the North American Regional Reanalysis (NARR) 3-hourly data during 1979–2008 (Mesinger et al. 2006). The integration is conducted for the CLM3.5, CoLM, and CSSP at the horizontal resolution of respectively 0.25° (~28 km), 30 km, and 30 km. The NARR meteorological conditions—including temperature, wind, humidity, pressure and height at the lowest atmospheric model layer, convective and resolved precipitation (separating to rainfall and snowfall by 2-m temperature), downwelling shortwave and longwave radiation, planetary boundary layer height, and surface pressure—are interpolated from the 32-km data grid cells onto the 0.25° or 30-km LSMs’ grid cells by an inverse quadratic distance weighting method. However, we found that the NARR contains an abrupt discontinuity over southeastern Canada near the U.S. border in December, January, February, and March mean precipitation distributions during 1979–96. This discontinuity may be due to the absence of dense rain gauge data over the region before 1997. As a result, the LSMs driven by the original NARR precipitation also simulate an abrupt discontinuity in snow cover distributions that were not consistent with observations. We therefore replace these problematic NARR precipitation data with the values interpolated from the adjacent grid cells.

Note that the CoLM and CSSP use the same surface boundary conditions as the CWRF (Liang et al. 2005a,b), while the CLM3.5 adopts its own built-in data (Lawrence and Chase 2007). We believe that maintaining the overall consistency between the physics representations and boundary conditions is essential to an objective skill assessment. Thus, both the boundary conditions and computational grid cells are kept identical with those originally developed for each respective LSM. These differences are expected to have a minor effect on the result comparisons and major conclusions in this study. Note also that the first 5-yr (1979–83) outputs from all LSMs are excluded in the subsequent analyses to reduce the effect of uncertainties in model initialization.

4. Observational data

The observational data used for model evaluation in this study are derived from direct measurements or objective analyses of key variables, including surface sensible and latent heat fluxes, soil moisture and temperature, streamflow and runoff, snow depth and water equivalent, and water table depth. Their data sources and processing procedures are briefly described below.

a. Surface sensible and latent heat fluxes

Hourly sensible and latent heat fluxes are collected from measurements at the towers of the Ameriflux network (http://public.ornl.gov/ameriflux/available.shtml). The network provides continuous observations of ecosystem-level exchanges of carbon, water, energy, and momentum spanning diurnal, synoptic, seasonal, and interannual scales, and currently includes sites in the North, Central, and South America (Baldocchi et al. 2001). The flux data were found valuable for LSMs’ development and evaluation before their general application in GCMs or RCMs (Friend et al. 2007; Stöckli et al. 2008). The present study selects 15 towers that have over four years of continuous forcing and flux data from the Ameriflux Level 2 database (Table 1). They represent the typical vegetation types within the study domain, including grassland, cropland, shrubland, mixed forest, evergreen needleleaf forest, and deciduous broadleaf forest. Gaps in the flux data are neither filled nor used in analysis.

b. Soil moisture

Soil moisture includes two data sources. One is from the Illinois network as measured by the neutron probe technique and calibrated by gravimetric observations (Hollinger and Isard 1994). The data contain 11 soil layers above 2 m, with the top and bottom layers 10 cm and others 20 cm in depth. There are 19 sites before 2004, but only 8 remain afterward. In general, these sites are evenly distributed over Illinois. We aggregate the data into three layer—top 0.1, 1, and 2 m—as averages over the entire state of Illinois at a monthly interval using all available records during 1984–2007.

Other data are from the Soil Climate Analysis Network (SCAN), established in 1999 by the U.S. Department of Agriculture (USDA) and focusing on agricultural areas (Schaefer et al. 2007). There are currently more than 150 stations in 39 states across the nation (http://www.wcc.nrcs.usda.gov/scan) collecting hourly soil moisture data by a dielectric constant measuring device at 5, 10, 20, 50, and 100 cm below the surface (Seyfried et al. 2005). The present study selects four regions with relatively dense SCAN stations, each containing at least 3-year records during 2003–08. These include Alabama–Tennessee (AL–TN: 34.3°–35.3°N, 87.2°–85.9°W) with 11 sites, Arkansas–Mississippi (AR–MS: 32.6°–35.4°N, 92.9°–89.7°W) with 11 sites, Iowa–Missouri–Nebraska–Kansas (IA–MO–NE–KS: 38.7°–42.6°N, 96.8°–93.6°W) with 7 sites, and Maryland–North Carolina–Pennsylvania–Virginia (MD–NC–PA–VA: 35.7°–40.9°N, 80.3°–76.5°W) with 8 sites. The annual precipitation over the above four regions are about 1415, 1348, 883, and 1115 mm, respectively. Therefore, only the IA–MO–NE–KS is a semihumid region, and all others are humid regions. The hourly data at each measuring depth are averaged at a monthly interval and over the four regions using all available records.

c. Soil temperature

Monthly mean soil temperatures are available over the United States from the National Climatic Data Center (NCDC) at depths of 0.1 and 1 m, with a much smaller number of stations in the latter (Hu and Feng 2003; Zhu and Liang 2005). They are interpolated onto the respective LSMs’ grid cells by the Cressman analysis method, assuming an effective radius of 1° for each station. The analyzed data for soil temperature at 0.1 m cover the period of 1984–2008 over almost the entire United States except for the Rocky Mountains and northeastern United States (1560 grid points), while the data at 1 m cover only the period of 1984–2001, mostly over Indiana, Iowa, South Dakota, and Minnesota (299 grid points).

d. Runoff and streamflow

Runoff is from the climatological composite analysis combining observed river discharge with an offline water balance model (Fekete et al. 2000, 2002). The data are made available in monthly means on grid cells of 0.5° × 0.5° by the Global Runoff Data Center (GRDC) at the University of New Hampshire. They are mapped onto the respective LSMs’ grid cells by bilinear interpolation. In addition, monthly discharges are acquired from 18 U.S. Geological Survey (USGS) streamflow gauges (http://waterdata.usgs.gov/nwis/sw) in different hydrologic units over the continental United States (http://water.usgs.gov/GIS/huc.html). Table 2 lists the key information of these gauges. Each gauge contains at least 12 years of continuous data records during 1984–2008.

Table 2.

Selected USGS streamflow gauges from different hydrologic units.

Table 2.

e. Snow depth and water equivalent

Daily snow depth and water equivalent data are available during 1979–96 at a grid spacing of 0.25° over North America from the Canadian Meteorology Center (CMC). They are produced by a simple snow accumulation, aging, and melt model that assimilates in situ daily observations from about 8000 U.S. cooperative stations and Canadian climate stations while driven by a 6-hourly air temperature and precipitation reanalysis. The gridded snow depth and estimated snow water equivalent agree well with available independent in situ and satellite data over midlatitude regions (Brown et al. 2003). They are mapped onto the respective LSMs’ grid cells using bilinear interpolation.

f. Water table depth

Water table depth data are taken from the USGS measurements as compiled by Fan et al. (2007). There were in total 549 616 observations during 1927–2005, while 81% of the sites had only one record. Some sites were affected by pumping nearby and exhibited long-term declining trends of water level. Nonetheless, the data provide a reasonable constraint and have been used for model evaluation on climatological water table depths in recent studies (Fan et al. 2007; Miguez-Macho et al. 2008). The present study concerns most shallow water table depths (less than 10 m) as they are highly correlated with anomalies of precipitation surplus or infiltration (Yuan et al. 2008a, 2009). Thus, all observations having a shallow water table (about 318 367 sites in total) within a respective LSM’s grid cell are averaged to define the climatological mean water table depth as a reference for model evaluation. It is assumed here that the lack of lateral saturated flow in the current LSMs has a minor effect on comparison in the 0.25° or 30-km resolution.

5. Evaluation against flux tower measurements

Table 3 illustrates correlation coefficients (CC) and root-mean-square errors (RMSE) between modeled and observed daily sensible heat (SH) and latent heat (LH) fluxes using all concurrent available data. All CC values are statistically significant at the 99% confidence level. For both measures, the performances of the CLM3.5 and CCSP are very close to each other while overall better than the CoLM’s. Since SH is mainly determined by net shortwave radiation and skin temperature and is less dependent on subsurface hydrologic conditions (Betts 2004), the CoLM performs similarly to the CLM3.5 and CSSP at a few sites (Los, NR1, PFa, and SO2). However, the CLM3.5 and CSSP improve notably over the CoLM in simulating SH variations at 9 of the 15 sites, increasing CC from 0.31–0.9 to 0.48–0.94 while decreasing RMSE from 44–80 to 29–60 W m−2. The improvement for LH is more substantial because of its stronger dependence on the advanced hydrological parameterizations in the CLM3.5 and CSSP. In particular, the CC score increases from the CoLM by about 0.3 for some forest sites (Me5, MMS, MOz, UMB, and WCr). On average, the CLM3.5 and CSSP decrease the CoLM surface flux errors by 10 W m−2.

Table 3.

The performances of CoLM, CLM3.5, and CSSP in simulating SH and LH fluxes. CC and RMSE (W m−2, in parenthesis) are based on daily means as compared with concurrent observations.

Table 3.

Figure 1 compares the mean annual cycles of SH and LH fluxes at selected sites including all vegetation types listed in Table 1. For croplands (Figs. 1a,b), the CoLM has large SH biases, especially before growing season, while the CLM3.5 and CSSP perform well, including spring and winter; only the CSSP successfully captures both the spring and summer LH peaks at sites ARM and Bo1 (Figs. 1f,g). On the other hand, the CLM3.5 (CSSP) produces the best annual cycle of SH (LH) over grasslands (Figs. 1c,h), whereas the CoLM (CSSP) generates the best result for SH (LH) over shrublands (Figs. 1d,i). For evergreen needleleaf forests, the CSSP reduces error from the CLM3.5 by 6–8 W m−2 for SH (Figs. 1k,l), but conversely by 4–8 W m−2 for LH (Figs. 1p,q). The results of the CSSP and CLM3.5 are comparable over deciduous broadleaf and mixed forests (Figs. 1e,j,m–o,r–t), where the CSSP (CLM3.5) is slightly better for SH (LH). Note that, in this single-column simulation, neglecting lateral transport processes of surface and subsurface water may degrade the CSSP skill to some extent. Even with this limit, the performance of the CSSP is comparable to the CLM3.5, and both exhibit obvious advances over the CoLM.

Fig. 1.
Fig. 1.

Mean annual cycle of sensible heat (SH) and latent heat (LH) fluxes from CoLM, CLM3.5, and CSSP as compared to Ameriflux observations.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

6. Evaluation of geographic distributions

This section focuses on evaluation of geographic distributions of the key surface and soil variables that critically determine the energy and water cycles over land. It compares the CSSP, CoLM, and CLM3.5 simulations during 1979–2008 (section 3) against concurrent observations (section 4). Recall that the initial 5 yr are considered as a spin up while the simulations during 1984–2008 are actually used in the comparison.

a. Soil moisture

Soil moisture is difficult to evaluate mainly because of its sparse observations and high heterogeneities. Model ability is generally examined at specific regions where sufficient data are available (Entin et al. 1999; Zhu and Liang 2005; Guo and Dirmeyer 2006; Qian et al. 2006; Oleson et al. 2008). This study selects Illinois and other four regions in the southern, central, and northeastern United States that have the best in situ measurements.

Figure 2 compares soil moisture simulations with observations averaged over Illinois in the top 0.1, 1, and 2 m. For interannual variability (Figs. 2a–c), the CoLM roughly simulates the major wet and dry soil conditions that occurred during 1984–2007. The CLM3.5 with new hydrologic parameterizations (Oleson et al. 2008) makes a significant improvement, having higher CC and lower mean absolute error (MAE) than the CoLM. The CSSP presents a further improvement, with large MAE reductions (in mm) from the CLM3.5 (6.0, 21.6, and 26.1) to 4.0, 13.3, and 20.7 for the top three soil layers. The improvement is especially pronounced near the surface, where the CLM3.5 substantially underestimates the observed interannual variability.

Fig. 2.
Fig. 2.

Simulated soil moisture (mm) averaged over Illinois by CoLM, CLM3.5, and CSSP in comparison with observations for top 0.1, top 1, and top 2 m soil. CC is the correlation coefficient, and MAE is the mean absolute error.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

Similar improvements are also reflected in the Illinois soil moisture annual cycle (Figs. 2d–f), where the CSSP best reproduces observations while the CoLM remains the poorest performer for all top (0.1, 1, and 2 m) layers. This is particularly obvious in MAE (mm), for which the CSSP yields (2.8, 7.2, and 11.5) are much smaller than even the CLM3.5 (5.0, 17.2, and 19.1). The large MAE values are identified with low rooting zone soil moisture variability, which was acknowledged by Oleson et al. (2008) as one of the major deficiencies remaining in the CLM3.5. The model skill in depicting the annual cycle amplitude can be measured by the ratio of standard deviation simulated over observed. For the rooting zone layers, the CSSP produces the highest ratios (0.58 and 0.83) as compared with the CoLM (0.44 and 0.79) and CLM3.5 (0.29 and 0.48). Thus, the CSSP generates not only the most realistic phase (highest CC) but also the best amplitude of the soil moisture annual cycle systematically throughout the root zone. Interestingly, the CLM3.5 has a better phase but weaker amplitude than the CoLM.

Figure 3 illustrates the Taylor (2001) plots comparing monthly soil wetness (moisture relative to saturation) anomalies (relative to its mean) simulated with the SCAN observations at depths of 5, 10, 20, 50, and 100 cm as averaged over the humid (AL–TN, AR–MS, and MD–NC–PA–VA) and semihumid (IA–MO–NE–KS) areas. Correlations are all statistically significant at the 95% confidence level, except for two depths in IA–MO–NE–KS by the CoLM. The models perform better over the humid areas than the semihumid region. Similar to the result in Illinois, the CSSP has a CC score comparable to the CLM3.5, while both are generally better than the CoLM, especially over the semihumid region. For the variability, the CSSP makes an obvious improvement over the semihumid region, where the ratio of standard deviation modeled over observed is 0.90–1.12, as compared with large underestimates by the CoLM (0.47–0.79) and CLM3.5 (0.57–0.78). Among all regions and all levels, the ratios fall within the interval (1 ± 0.25) for 70% of cases by the CSSP, 60% by the CoLM, and 40% by the CLM3.5. Thus, the CSSP is overall more realistic in both phase and magnitude.

Fig. 3.
Fig. 3.

Statistics of the monthly soil wetness anomalies (mm mm−1) from CoLM, CLM3.5, and CSSP compared to SCAN observations at the depths of 5, 10, 20, 50, and 100 cm for 2003–08 averaged over four regions. The radial distance from the origin to the letters is the ratio of standard deviations simulated over observed; the azimuthal position of the letters is the linear correlation, and the distance from “REF” is the normalized root-mean-square error based on law of cosines (Taylor 2001).

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

Figure 4 compares annual means and interannual standard deviations of monthly soil moisture in the top 0.1 and 1 m simulated by the three LSMs during 1984–2008. The CSSP produces a more detailed fine structure over the Rocky and Appalachian mountain regions than the other two. The CoLM generates the wettest soil over northeastern and central United States, while CLM3.5 calculates the wettest soil over the western United States. On average over the entire domain, the top 1 m soil in the CSSP is drier than the CoLM and CLM3.5 by respectively 12.2% and 13.4%. The CoLM gives the largest deviation for the top 1 m soil moisture over the eastern United States, which may likely be an overestimation of variability as suggested by the AL–TN analysis above. The underlying mechanism for this CoLM overestimation, though not yet clear, can be helpful in remedying the CLM3.5 deficiency of weak variability over the humid areas, as both are single-column models. As compared with the CLM3.5, the CSSP enhances soil moisture variability in the top 0.1 and 1 m by 52% and 20%, respectively. As such, the CSSP substantially alleviates the variability underestimations in mid- to high latitudes by the CLM3.5 and in the western United States by the CoLM. One possible explanation for the variability enhancement is that the CSSP provides more reasonable water table depth (see section 6e), which helps to constrain the soil moisture variations.

Fig. 4.
Fig. 4.

Model simulated mean values (mm) and standard deviations (mm) of monthly top 0.1 and top 1 m soil moisture (SM) from 1984 to 2008.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

b. Soil temperature

Soil temperature, while containing important land memory (Beltrami 2002; Hu and Feng 2003), has rarely been subject to model evaluation (Zhu and Liang 2005), perhaps partly because of the scarcity of observations. This study compares its simulation by the three LSMs.

Figure 5 compares the frequency distributions of four statistics based on monthly soil temperature at the depths of 0.1 and 1 m. The mean error (ME) gives the systematic bias; RMSE and MAE depict the overall accuracy, with the former more sensitive to large departures; and standard deviation of error (SDE) measures the closeness of the model to observed temporal fluctuations (Yuan et al. 2008a). The percentage of large systematic biases is smallest by the CSSP at 0.1 m and the CLM3.5 at 1 m, respectively. The ME frequency is more evenly distributed at 0.1 than 1 m, the latter being skewed toward systematic warm biases peaking around 1 K. Correspondingly, the RMSE and MAE frequency distribution peaks occur at larger errors for 1 than 0.1 m. Overall, the CSSP is the best, reducing average errors from the CLM3.5 (CoLM) by 0.2 (0.7) K at 0.1 m and 0.3 (0.6) K at 1 m. The SDE distribution shows a similar skill at 0.1 m by the CSSP and CLM3.5 (both better than CoLM) but a consistent higher performance at 1 m by the CSSP than both CLM3.5 and CoLM. On average, the observed variance at 1 m is accounted for 94% by the CSSP and 89% by the CLM3.5. Thus, the CSSP enhances interannual variability both in soil moisture and temperature, which is closer to observations than the CLM3.5 or CoLM.

Fig. 5.
Fig. 5.

Frequency distributions of ME, RMSE, MAE, and SDE for the monthly soil temperature (K) at the depth of 0.1 m (1 m) over the grid cells with observations during 1984–2008 (1984–2001). The average values of error for CoLM, CLM3.5, and CSSP are included in the brackets from left to right.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

c. Runoff and streamflow

The GRDC data have been used to evaluate the climatological mean and annual cycle of runoff generated by global LSMs (Niu et al. 2005, 2007; Niu and Yang 2006; Oleson et al. 2008). They are applied here to examine the model performance in simulating runoff seasonal variations and geographic distributions over the contiguous United States. Figure 6 compares the 1984–2008 mean total runoff for winter [December–February (DJF)], spring [March–May (MAM)], summer [June–August (JJA)], and fall [September–November (SON)]. The observed runoff patterns resemble the precipitation climatology distributions (Liang et al. 2004), exhibiting high values east of 100°W and over the west coast, with low values in the dry zone along the downstream slopes of the Rockies. In general, runoff peaks in spring because of increasing snowmelt in the north and large precipitation minus evaporation surplus in the south. The summer pattern is rather uniform over most areas with adequate rainfall, except for far north in Canada where snow melting still occurs. Winter runoff is maximized along the upstream slopes of the Cascades and over the southeast United States—both caused by heavy rainfall events. On the other hand, runoff in fall is relatively small as intermediate rain is roughly balanced by evaporation, except for eastern Canada where surplus exists because of smaller evaporation under colder temperatures.

Fig. 6.
Fig. 6.

Observed (GRDC) and simulated (CoLM, CLM3.5, and CSSP) total runoff (mm day−1) climatology for winter (DJF), spring (MAM), summer (JJA), and fall (SON). The modeled runoffs are from average results during 1984–2008.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

The CSSP and CLM3.5 reproduce well the major characteristics of the observed runoff seasonal variations and geographic distributions, while the CoLM shows large discrepancies. Over the Cascades, all three models are quite realistic throughout the year, except for summer underestimation by the CoLM. For other regions and seasons, the CoLM generally underpredicts runoff, especially in spring and summer. Over the Rocky Mountain region, the CSSP simulates larger runoff than the CLM3.5 or GRDC in winter and spring. This enhancement may likely result from its incorporation of the effects of subgrid topographic control and shallow bedrock constraint on surface and subsurface water movements. The differences, however, may be well within observational uncertainties because of the scarcity and poor quality of the actual data driving the GRDC analysis—including discharge, precipitation, and temperature—in the mountainous region. Over the Sierra Madre Occidental Mountains, the CSSP correctly simulates summer and fall runoff distributions, which are totally missed by the CLM3.5. This indicates the advantage of considering subgrid topographic effects. Along the central Great Plains, the CSSP is also more realistic as a result of the lateral flow and bottom drainage from the eastern transition zone where the bedrock depths are shallower. The CSSP successfully simulates the runoff extremes in winter and spring along the lower Mississippi River, and in spring over southeastern Canada and the northeastern United States, both of which are attributable to its explicit treatment of overland and channel flow routing. These extremes are substantially underestimated by the CLM3.5. On the other hand, the CSSP tends to overpredict fall runoff over the southeast United States, which is not fully understood.

Table 4 lists the interannual CC and RMSE statistics of monthly mean model discharges as compared with observations at 18 USGS streamflow gages, while Fig. 7 illustrates the corresponding annual hydrographic evolutions. The diagnostic river transport model from the CLM3.5 (Oleson et al. 2004) is adopted to route the CoLM and CSSP runoff toward the USGS gages for a direct comparison. Both Table 4 and Fig. 7 indicate that the CSSP best represents observations at 11 out of the 18 gages, with higher CC and lower RMSE, or better annual hydrographic evolution. These include New England, the mid-Atlantic, the Great Lakes, Ohio, Tennessee, upper and lower Mississippi, Missouri, the Texas Gulf, lower Colorado, and the Pacific Northwest (Figs. 7a,b,d–h,j,l,o,q). The bulk CC and RMSE averaged among all 18 gauges for the CoLM, CLM3.5, and CSSP are 0.36, 0.54, and 0.67, and 742, 852, and 524 m3 s−1, respectively. This indicates the overall CSSP superior skill and poorest CoLM performance. The CSSP also has a clear advantage in capturing the extreme floods, including the Brazos River in 1992 (Fig. 7l), the Mississippi and Verde Rivers in 1993 (Figs. 7g,o), the Altamaha River in 1998 (Fig. 7c), and the Tallahatchie River in 2002 (Fig. 7h). Note that the relative RMSE divided by the observed mean discharge shows large values (>1) for all models at Souris–Red–Rainy, the Texas Gulf, the Rio Grande, upper and lower Colorado, the Great Basin, and California. This suggests that the streamflow prediction is less credible over the arid and semiarid regions than humid and semihumid areas.

Table 4.

CC and RMSE of monthly discharge (m3 s−1) for 18 selected USGS gauges. Bold numbers show the best CC/RMSE of the three models.

Table 4.
Fig. 7.
Fig. 7.

Observed and simulated annual discharge (m3 s−1) for the selected USGS gauges.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

d. Snow depth and water equivalent

The undercatch of solid precipitation (Adam and Lettenmaier 2003; Yang et al. 2005) at high latitudes and the underestimation of precipitation at high elevations (Adam et al. 2006) may have a nontrivial impact on snow accumulation (Tian et al. 2007). Thus, this study focuses on snow interannual variations. Figure 8 compares interannual anomalies during 1984–97 of winter (DJF) snow depth and water equivalent averaged over 35°–40°, 40°–45°, and 45°–52°N from the CMC observations and model simulations. Interestingly, the CMC data showed many more years of negative than positive anomalies. This resulted from fewer but much stronger positive anomalies, including severe snowstorms in 1984, 1985, and 1993 over 35°–40°N, 1984 and 1993 over 40°–45°N, and 1997 over 45°–52°N. As driven by the observational precipitation analysis, the LSMs reproduce well the CMC snow depth and water equivalent interannual anomalies, having correlations all higher than 0.87 (and CSSP is the best) and being statistically significant at the 99% confidence level.

Fig. 8.
Fig. 8.

Interannual variations of winter (DJF) snow depth (m) and snow water equivalent (mm) averaged over (a),(d) 35°–40°, (b),(e) 40°–45°, and (c),(f) 45°–52°N during 1984–97 from CMC observations and model results.

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

The winter snow depth interannual variability in the three LSMs is all substantially underestimated between 35° and 45°N and slightly overestimated between 45° and 52°N. The ratios of standard deviation simulated over observed for CoLM, CLM3.5, and CSSP are respectively 0.49–0.52, 0.68–0.70, and 0.56–0.69 between 35° and 45°N, and 1.19, 1.24, and 1.14 between 45° and 52°N. Since the snow prediction module is identical in the conceptual design and key numerical formulations, their differences among the LSMs may be related to subgrid vegetation types, snow interception, and albedo. On the other hand, the common model discrepancies may result from forcing errors in the NARR, especially in winter precipitation. Further investigation on these aspects is needed.

e. Water table depth

Figure 9 compares the water table depths derived from the USGS observations (Fan et al. 2007; Miguez-Macho et al. 2008) and the 1984–2008 climatological mean results simulated by the CoLM, CLM3.5, and CSSP; Fig. 10 illustrates their frequency distributions. Note that the modeled water table depths are less than 8 m because 1) 5 yr is insufficient to spin up deep aquifers, which would take at least 30 yr in arid regions (Oleson et al. 2008); 2) the lack of lateral groundwater flow mechanism (Fan et al. 2007) may prevent the water table from dropping quickly over mountainous areas; and 3) the shallow bedrock is specified less than 6 m (Liang et al. 2005b), which is also used to constrain deep aquifers. As discussed in section 4, we therefore focus only on the shallow groundwater.

Fig. 9.
Fig. 9.

Climatological water table depths (m) from USGS observations and model results (1984–2008).

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

Fig. 10.
Fig. 10.

Frequency distributions of mean values and biases for the climatological water table depths (m).

Citation: Journal of Hydrometeorology 12, 4; 10.1175/2010JHM1302.1

The CSSP simulates water table depths that exhibit stronger heterogeneity and more realistic geographic distribution than both the CoLM and CLM3.5. It reproduces the observed deep (shallow) water table over the Rocky and Appalachian Mountains (river valleys and coasts), which resembles the results of Yu et al. (2006) and Fan et al. (2007). As compared to the USGS data, the CSSP makes a significant improvement from the other two LSMs over the northeast United States, Gulf states, central Great Plains, lower Colorado, and Wisconsin. Among the 5508 grid cells with available USGS data, 66% are observed to have the water table at about 4–6 m, while the CoLM, CLM3.5, and CSSP simulate 37%, 38%, and 58%, respectively. The average water table depths for the USGS data, and the CoLM, CLM3.5, and CSSP simulations, are 4.8, 3.1, 3.2, and 4.4 m. Thus, the CSSP reduces the systematic mean biases and absolute errors by 1.2 and 0.4 m.

7. Conclusions and discussion

This study presents a comprehensive evaluation of the CSSP at regional–local scales over the contiguous United States during 1979–2008 as driven by atmospheric conditions from the NARR observational reanalysis. The CSSP has been developed most recently as the core LSM of the CWRF to predict soil temperature–moisture distributions, terrestrial hydrology variations, and land–atmosphere flux exchanges. The full CSSP coupling with the CWRF and the resulting climate effect will be documented in an upcoming paper (X.-Z. Liang et al. 2010, unpublished manuscript). Since the CSSP is rooted in the CoLM with a few updates from the CLM3.5, the evaluation is elaborated here in parallel comparison with these two LSMs against various in situ measurements and synthetic observations, including surface sensible and latent heat fluxes, soil moisture and temperature, runoff and streamflow, snow depth and water equivalent, and water table depth.

It has been demonstrated that the CSSP performance is overall superior to both the CoLM and CLM3.5. In particular, the CSSP 1) reduces surface heat flux errors by 10 W m−2 on average and largely increases correlations with observations for daily latent heat variations by up to 0.3 from the CoLM, while outperforming the CLM3.5 over the crop, grass, and shrub sites in depicting the latent heat annual cycles; 2) substantially improves the CoLM and CLM3.5 in representing the upper-layer soil moisture dynamics and thus greatly alleviates the seasonal–interannual variability underestimations in mid- to high latitudes by the CLM3.5 and in the western United States by the CoLM; 3) reduces soil temperature errors from the CLM3.5 (CoLM) by 0.2 (0.7) K at 0.1 m and 0.3 (0.6) K at 1 m; 4) captures seasonal–interannual runoff and streamflow variations and extreme events more realistically than the CoLM and CLM3.5 over most regions because of advanced parameterizations of subgrid topographic control, shallow bedrock constraint, and conjunctive surface–subsurface flow interaction; 5) produces better snow depth anomalies in high latitude (45°–52°N) than the CoLM and CLM3.5; and 6) reduces climatological water table depth systematic bias (absolute error) from the CoLM and CLM3.5 by about 1.2 (0.4) m.

The CSSP still contains several common (to the CoLM and CLM3.5) though reduced deficiencies. These include the overestimation of sensible heat over shrublands (Fig. 1d), low amplitude for the annual cycle of surface soil moisture (Fig. 2d), less credible soil moisture anomaly over semihumid regions (Fig. 3), warm biases in deep soil temperature (Fig. 5b), underestimation of runoff over humid and high-latitude cold areas during winter and spring (Fig. 6), large relative errors in streamflow over arid and semiarid areas, and small interannual variability for winter snow depth between 35° and 45°N. It is not trivial to overcome these shortcomings separately since the water and energy cycles are highly coupled in the climate system, where modifications in one process will have positive or negative impacts on others. Nevertheless, we will further improve the CSSP in the following aspects:

  1. Develop high-quality and fine-resolution data for soil and vegetation characteristics. Figures 4 and 9 show that the CoLM reproduces stronger heterogeneities for soil moisture and water table depth than the CLM3.5, although the former has simpler parameterizations. One possible reason is that the CoLM uses the more comprehensive soil and vegetation data of the CSSP at 30-km grid spacing (Liang et al. 2005a,b), while the CLM3.5 interpolates these from the less detailed 0.5° global distributions.

  2. Adopt advanced statistical techniques for optimizing model parameters to reduce overall model errors due to uncertainties in numerical representations. Rosero et al. (2009) showed that once properly calibrated, a physically enhanced model is not necessarily better than the original one. On the other hand, they also indicated that the multiobjective calibration (Gupta et al. 1998) is useful to strengthen the model’s general performance if the relevant observations are available.

  3. Incorporate more-complete physical processes and/or improved parameterizations. The current CSSP predicts surface water depths and considers conjunctive surface–subsurface flow interaction through infiltration flux. The lateral groundwater flow (Fan et al. 2007; Miguez-Macho et al. 2007; Maxwell and Kollet 2008; Xie and Yuan 2010) is needed to i) provide the basis for coupling groundwater to streams, ii) establish the deep infiltration flux (recharge) from soil to groundwater, and iii) control the lower boundary for the soil moisture, which is critical to land surface energy fluxes. In addition, the 3D radiation model (Dickinson et al. 2008) may improve sensible heat flux over shrublands; the hydraulic redistribution (Amenu and Kumar 2008) may increase surface soil moisture variability; the organic thermal conductivity effect (Lawrence and Slater 2008) may decrease soil temperature warm biases over regions with high soil carbon contents; the subgrid precipitation and canopy water storage effects (Wang et al. 2007) may reduce snow interception and increase snow depth on the ground, generating larger interannual variability; and the interplay between vertical structure and photosynthetic pathway and the role of acclimation under elevated CO2 may significantly determine the ecohydrological responses of dense canopies to environmental variability (Drewry et al. 2010a,b). Furthermore, integrations of human activities, such as crop growth and irrigation practice, dam operation and management (Hossain et al. 2009), and biofuel plant production (Grassini et al. 2009), although challenging, will provide essential impacts on the LSMs’ skill in representing reality. These will be our ongoing and future refinements to the CSSP.

Acknowledgments

We thank Ross Brown for sharing snow observations and Bob Scott for providing Illinois soil moisture data. We greatly appreciate constructive discussions with Hyun Il Choi on the CSSP formulation. We acknowledge NOAA/ESRL and UIUC/NCSA for their supercomputing support. The research is supported by the NOAA Climate Prediction Program for the Americas (CPPA) Grants NA08OAR4310575 and NA08OAR4310875, the Grant NASA NNX08AL94G, and the NOAA Education Partnership Program (EPP) COM Howard 631017. The views expressed are those of the authors and do not necessarily reflect those of the sponsoring agencies or the Illinois State Water Survey.

REFERENCES

  • Adam, J. C., and Lettenmaier D. P. , 2003: Adjustment of global gridded precipitation for systematic bias. J. Geophys. Res., 108, 4257, doi:10.1029/2002JD002499.

    • Search Google Scholar
    • Export Citation
  • Adam, J. C., Clark E. A. , Lettenmaier D. P. , and Wood E. F. , 2006: Correction of global precipitation products for orographic effects. J. Climate, 19, 1538.

    • Search Google Scholar
    • Export Citation
  • Amenu, G. G., and Kumar P. , 2008: A model for hydraulic redistribution incorporating coupled soil-root moisture transport. Hydrol. Earth Syst. Sci., 12, 5574.

    • Search Google Scholar
    • Export Citation
  • Avissar, R., and Pielke R. , 1989: A parameterization of heterogeneous land surfaces for atmospheric numerical models and its impact on regional meteorology. Mon. Wea. Rev., 117, 21132136.

    • Search Google Scholar
    • Export Citation
  • Baldocchi, D., and Coauthors, 2001: FLUXNET: A new tool to study the temporal and spatial variability of ecosystem-scale carbon dioxide, water vapor, and energy flux densities. Bull. Amer. Meteor. Soc., 82, 24152433.

    • Search Google Scholar
    • Export Citation
  • Beltrami, H., 2002: Earth’s long-term memory. Science, 297, 206207.

  • Betts, A. K., 2004: Understanding hydrometeorology using global models. Bull. Amer. Meteor. Soc., 85, 16731688.

  • 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 Tech. Note NCAR/TN-417+STR, 165 pp.

    • Search Google Scholar
    • Export Citation
  • Bonan, G. B., 1998: The land surface climatology of the NCAR Land Surface Model coupled to the NCAR Community Climate Model. J. Climate, 11, 13071326.

    • Search Google Scholar
    • Export Citation
  • Brown, R. D., Brasnett B. , and Robinson D. , 2003: Gridded North American monthly snow depth and snow water equivalent for GCM evaluation. Atmos.–Ocean, 41, 114.

    • Search Google Scholar
    • Export Citation
  • Chen, F., and Dudhia J. , 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569585.

    • Search Google Scholar
    • Export Citation
  • Chen, J., and Kumar P. , 2001: Topographic influence of the seasonal and interannual variation of water and energy balance of basins in North America. J. Climate, 14, 19892014.

    • Search Google Scholar
    • Export Citation
  • Chen, M., Wang W. , and Kumar A. , 2010: Prediction of monthly-mean temperature: The roles of atmospheric and land initial conditions and sea surface temperature. J. Climate, 23, 717725.

    • Search Google Scholar
    • Export Citation
  • Choi, H. I., 2006: 3D volume averaged soil-moisture transport model: A scalable scheme for representing subgrid topographic control in land–atmosphere interactions. Ph.D. dissertation, University of Illinois at Urbana–Champaign, 189 pp.

  • Choi, H. I., and Liang X.-Z. , 2010: Improved terrestrial hydrologic representation in mesoscale land surface models. J. Hydrometeor., 11, 797809.

    • Search Google Scholar
    • Export Citation
  • Choi, H. I., Kumar P. , and Liang X.-Z. , 2007: Three-dimensional volume-averaged soil moisture transport model with a scalable parameterization of subgrid topographic variability. Water Resour. Res., 43, W04414, doi:10.1029/2006WR005134.

    • Search Google Scholar
    • Export Citation
  • Dai, Y., and Coauthors, 2003: The Common Land Model. Bull. Amer. Meteor. Soc., 84, 10131023.

  • Dai, Y., Dickinson R. E. , and Wang Y.-P. , 2004: A two-big-leaf model for canopy temperature, photosynthesis, and stomatal conductance. J. Climate, 17, 22812299.

    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., Henderson-Sellers A. , Kennedy P. J. , and Wilson M. F. , 1986: Biosphere–Atmosphere Transfer Scheme (BATS) for the NCAR Community Climate Model. NCAR Tech. Note NCAR/TN-275+STR 72 pp.

    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., Oleson K. W. , Bonan G. B. , Hoffman F. , Thornton P. , Vertenstein M. , Yang Z.-L. , and Zeng X. , 2006: The Community Land Model and its climate statistics as a component of the Community Climate System Model. J. Climate, 19, 23022324.

    • Search Google Scholar
    • Export Citation
  • Dickinson, R. E., Zhou L. , Tian Y. , Liu Q. , Lavergne T. , Pinty B. , Schaaf C. B. , and Knyazikhin Y. , 2008: A three-dimensional analytic model for the scattering of a spherical bush. J. Geophys. Res., 113, D20113, doi:10.1029/2007JD009564.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., 2005: The land surface contribution to the potential predictability of boreal summer season climate. J. Hydrometeor., 6, 618632.

    • Search Google Scholar
    • Export Citation
  • Dirmeyer, P. A., Gao X. , Zhao M. , Guo Z. , Oki T. , and Hanasaki N. , 2006: GSWP-2: Multimodel analysis and implications for our perception of the land surface. Bull. Amer. Meteor. Soc., 87, 13811397.

    • Search Google Scholar
    • Export Citation
  • Drewry, D. T., Kumar P. , Long S. , Bernacchi C. , Liang X.-Z. , and Sivapalan M. , 2010a: Ecohydrological responses of dense canopies to environmental variability: 1. Interplay between vertical structure and photosynthetic pathway. J. Geophys. Res., 115, G04022, doi:10.1029/2010JG001340.

    • Search Google Scholar
    • Export Citation
  • Drewry, D. T., Kumar P. , Long S. , Bernacchi C. , Liang X.-Z. , and Sivapalan M. , 2010b: Ecohydrological responses of dense canopies to environmental variability: 2. Role of acclimation under elevated CO2. J. Geophys. Res., 115, G04023, doi:10.1029/2010JG001341.

    • Search Google Scholar
    • Export Citation
  • Entin, J. K., Robock A. , Vinnikov K. Y. , Zabelin V. , Liu S. , Namkhai A. , and Adyasuren T. , 1999: Evaluation of global soil wetness project soil moisture simulations. J. Meteor. Soc. Japan, 77, 183198.

    • Search Google Scholar
    • Export Citation
  • Famiglietti, J. S., and Wood E. F. , 1994: Multiscale modeling of spatially variable water and energy balance processes. Water Resour. Res., 30, 30613078.

    • Search Google Scholar
    • Export Citation
  • Fan, Y., Miguez-Macho G. , Weaver C. P. , Walko R. , and Robock A. , 2007: Incorporating water table dynamics in climate modeling: 1. Water table observations and equilibrium water table simulations. J. Geophys. Res., 112, D10125, doi:10.1029/2006JD008111.

    • Search Google Scholar
    • Export Citation
  • Fekete, B. M., Vörösmarty C. J. , and Grabs W. , 2000: Global composite runoff fields based on observed river discharge and simulated water balances (version 1.0). GRDC Rep. 22, 108 pp. [Available online at http://www.grdc.sr.unh.edu.]

    • Search Google Scholar
    • Export Citation
  • Fekete, B. M., Vörösmarty C. J. , and Grabs W. , 2002: High-resolution fields of global runoff combining observed river discharge and simulated water balances. Global Biogeochem. Cycles, 16, 1042, doi:10.1029/1999GB001254.

    • Search Google Scholar
    • Export Citation
  • Foley, J. A., Prentice I. C. , Ramankutty N. , Levis S. , Pollard D. , Sitch S. , and Haxeltine A. , 1996: An integrated biosphere model of land surface processes, terrestrial carbon balance, and vegetation dynamics. Global Biogeochem. Cycles, 10, 603628.

    • Search Google Scholar
    • Export Citation
  • Freer, J., McDonnell J. J. , Beven K. J. , Peters N. E. , Burns D. A. , Hooper R. P. , Aulenbach B. , and Kendall C. , 2002: The role of bedrock topography on subsurface storm flow. Water Resour. Res., 38, 1269, doi:10.1029/2001WR000872.

    • Search Google Scholar
    • Export Citation
  • Friend, A. D., and Coauthors, 2007: FLUXNET and modelling the global carbon cycle. Global Change Biol., 13, 610633.

  • Govindaraju, R. S., and Kavvas M. L. , 1991: Dynamics of moving boundary overland flows over infiltrating surfaces at hillslopes. Water Resour. Res., 27, 18851898.

    • Search Google Scholar
    • Export Citation
  • Grassini, P., Hunt E. , Mitchell R. B. , and Weiss A. , 2009: Simulating switchgrass growth and development under potential and water-limiting conditions. Agron. J., 101, 564571.

    • Search Google Scholar
    • Export Citation
  • Guo, Z., and Dirmeyer P. A. , 2006: Evaluation of the Second Global Soil Wetness Project soil moisture simulations: 1. Intermodel comparison. J. Geophys. Res., 111, D22S02, doi:10.1029/2006JD007233.

    • Search Google Scholar
    • Export Citation
  • Gupta, H. V., Sorooshian S. , and Yapo P. O. , 1998: Toward improved calibration of hydrologic models: Multiple and noncommensurable measures of information. Water Resour. Res., 34, 751763.

    • Search Google Scholar
    • Export Citation
  • Hollinger, S. E., and Isard S. A. , 1994: A soil moisture climatology of Illinois. J. Climate, 7, 822833.

  • Hossain, F., Jeyachandran I. , and Pielke R. , 2009: Have large dams altered extreme precipitation patterns? Eos, Trans. Amer. Geophys. Union, 90, 453454.

    • Search Google Scholar
    • Export Citation
  • Hu, Q., and Feng S. , 2003: A daily soil temperature dataset and soil temperature climatology of the contiguous United States. J. Appl. Meteor., 42, 11391156.

    • Search Google Scholar
    • Export Citation
  • Katsuyama, M., Ohte N. , and Kabeya N. , 2005: Effects of bedrock permeability on hillslope and riparian groundwater dynamics in a weathered granite catchment. Water Resour. Res., 41, W01010, doi:10.1029/2004WR003275.

    • Search Google Scholar
    • Export Citation
  • Kollet, S. J., and Maxwell R. M. , 2006: Integrated surface–groundwater flow modeling: A free-surface overland flow boundary condition in a parallel groundwater flow model. Adv. Water Resour., 29, 945958.

    • Search Google Scholar
    • Export Citation
  • Koster, R. D., and Coauthors, 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 11381140.

  • Kumar, P., 2004: Layer averaged Richard’s equation with lateral flow. Adv. Water Resour., 27, 521531.

  • LaBolle, E. M., Ahmed A. A. , and Fogg G. E. , 2003: Review of the Integrated Groundwater and Surface-Water Model (IGSM). Ground Water, 41, 238246.

    • Search Google Scholar
    • Export Citation
  • Lawrence, D. M., and Slater A. G. , 2008: Incorporating organic soil into a global climate model. Climate Dyn., 30, 145160.

  • Lawrence, P. J., and Chase T. N. , 2007: Representing a new MODIS consistent land surface in the Community Land Model (CLM 3.0). J. Geophys. Res., 112, G01023, doi:10.1029/2006JG000168.

    • Search Google Scholar
    • Export Citation
  • Liang, X., Xie Z. , and Huang M. , 2003: A new parameterization for surface and groundwater interactions and its impact on water budgets with the Variable Infiltration Capacity (VIC) land surface model. J. Geophys. Res., 108, 8613, doi:10.1029/2002JD003090.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., Li L. , Kunkel K. E. , Ting M. , and Wang J. X. L. , 2004: Regional climate model simulation of U.S. precipitation during 1982–2002. Part I: Annual cycle. J. Climate, 17, 35103528.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., Choi H. , Kunkel K. E. , Dai Y. , Joseph E. , Wang J. X. L. , and Kumar P. , 2005a: Development of the regional Climate-Weather Research and Forecasting (CWRF) model: Surface boundary conditions. Illinois State Water Survey Scientific Rep. 2005-01, 32 pp.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., Choi H. , Kunkel K. E. , Dai Y. , Joseph E. , Wang J. X. L. , and Kumar P. , 2005b: Surface boundary conditions for mesoscale regional climate models. Earth Interact., 9, [Available online at http://EarthInteractions.org.]

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., and Coauthors, 2005c: Development of land surface albedo parameterization based on Moderate Resolution Imaging Spectroradiometer (MODIS) data. J. Geophys. Res., 110, D11107, doi:10.1029/2004JD005579.

    • Search Google Scholar
    • Export Citation
  • Liang, X.-Z., Xu M. , Zhu J. , Kunkel K. E. , and Wang J. X. L. , 2005d: Development of the regional Climate-Weather Research and Forecasting Model (CWRF): Treatment of topography. Proc. Joint WRF/MM5 User’s Workshop, Boulder, CO, NCAR, 9.3.

    • Search Google Scholar
    • Export Citation
  • Lo, M. H., Yeh P. J. F. , and Famiglietti J. S. , 2008: Constraining water table depth simulations in a land surface model using estimated baseflow. Adv. Water Resour., 31, 15521564.

    • Search Google Scholar
    • Export Citation
  • Manabe, S., 1969: Climate and the ocean circulation: I. The atmospheric circulation and the hydrology of the earth’s surface. Mon. Wea. Rev., 97, 739774.

    • Search Google Scholar
    • Export Citation
  • Maxwell, R. M., and Miller N. L. , 2005: Development of a coupled land surface and groundwater model. J. Hydrometeor., 6, 233247.

  • Maxwell, R. M., and Kollet S. J. , 2008: Interdependence of groundwater dynamics and land–energy feedbacks under climate change. Nat. Geosci., 1, 665669, doi:10.1038/ngeo315.

    • Search Google Scholar
    • Export Citation
  • Mesinger, F., and Coauthors, 2006: North American Regional Reanalysis. Bull. Amer. Meteor. Soc., 87, 343360.

  • Miguez-Macho, G., Fan Y. , Weaver C. P. , Walko R. , and Robock A. , 2007: Incorporating water table dynamics in climate modeling: 2. Formulation, validation, and soil moisture simulation. J. Geophys. Res., 112, D13108, doi:10.1029/2006JD008112.

    • Search Google Scholar
    • Export Citation
  • Miguez-Macho, G., Li H. , and Fan Y. , 2008: Simulated water table and soil moisture climatology over North America. Bull. Amer. Meteor. Soc., 89, 663672.

    • Search Google Scholar
    • Export Citation
  • Montaldo, N., and Albertson J. D. , 2003: Temporal dynamics of soil moisture variability: 2. Implications for land surface models. Water Resour. Res., 39, 1275, doi:10.1029/2002WR001618.

    • Search Google Scholar
    • Export Citation
  • Morita, M., and Yen B. C. , 2002: Modeling of conjunctive two-dimensional surface-three-dimensional subsurface flows. J. Hydraul. Eng., 128, 184200.

    • Search Google Scholar
    • Export Citation
  • Niu, G.-Y., and Yang Z.-L. , 2006: Effects of frozen soil on snowmelt runoff and soil water storage at a continental scale. J. Hydrometeor., 7, 937952.

    • Search Google Scholar
    • Export Citation
  • Niu, G.-Y., Yang Z.-L. , Dickinson R. E. , and Gulden L. E. , 2005: A simple TOPMODEL-based runoff parameterization (SIMTOP) for use in global climate models. J. Geophys. Res., 110, D21106, doi:10.1029/2005JD006111.

    • Search Google Scholar
    • Export Citation
  • Niu, G.-Y., Yang Z.-L. , Dickinson R. E. , Gulden L. E. , and Su H. , 2007: Development of a simple groundwater model for use in climate models and evaluation with Gravity Recovery and Climate Experiment data. J. Geophys. Res., 112, D07103, doi:10.1029/2006JD007522.

    • Search Google Scholar
    • Export Citation
  • Oleson, K. W., and Coauthors, 2004: Technical description of the Community Land Model (CLM). NCAR Tech. Note NCAR/TN-461+STR, 173 pp.

  • Oleson, K. W., and Coauthors, 2008: Improvements to the Community Land Model and their impact on the hydrological cycle. J. Geophys. Res., 113, G01021, doi:10.1029/2007JG000563.

    • Search Google Scholar
    • Export Citation
  • Pielke, R. A., Liston G. E. , Eastman J. L. , Lu L. , and Coughenour M. , 1999: Seasonal weather prediction as an initial value problem. J. Geophys. Res., 104, 19 46319 479.

    • Search Google Scholar
    • Export Citation
  • Pollard, D., and Thompson S. L. , 1995: Use of a land surface transfer scheme (LSX) in a global climate model: The response to doubling stomatal resistance. Global Planet. Change, 10, 129161.

    • Search Google Scholar
    • Export Citation
  • Qian, T., Dai A. , Trenberth K. E. , and Oleson K. W. , 2006: Simulation of global land surface conditions from 1948–2004. Part I: Forcing data and evaluation. J. Hydrometeor., 7, 953975.

    • Search Google Scholar
    • Export Citation
  • Rosero, E., Yang Z.-L. , Gulden L. E. , Niu G.-Y. , and Gochis D. J. , 2009: Evaluating enhanced hydrological representations in Noah-LSM over transition zones: Implications for model development. J. Hydrometeor., 10, 600622.

    • Search Google Scholar
    • Export Citation
  • Schaefer, G. L., Cosh M. H. , and Jackson T. J. , 2007: The USDA Natural Resources Conservation Service Soil Climate Analysis Network (SCAN). J. Atmos. Oceanic Technol., 24, 20732077.

    • Search Google Scholar
    • Export Citation
  • Schenk, H. J., and Jackson R. B. , 2002: The global biogeography of roots. Ecol. Monogr., 72, 311328.

  • Schlosser, C. A., and Milly P. C. D. , 2002: A model-based investigation of soil moisture predictability and associated climate predictability. J. Hydrometeor., 3, 483501.

    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., Mintz Y. , Sud Y. C. , and Dalcher A. , 1986: A Simple Biosphere Model (SiB) for use within general circulation models. J. Atmos. Sci., 43, 505531.

    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., Heiser M. D. , and Hall F. G. , 1992: Relations between surface conductance and spectral vegetation indices at intermediate (100 m2 to 15 km2) length scales. J. Geophys. Res., 97, 19 03319 059.

    • Search Google Scholar
    • Export Citation
  • Sellers, P. J., and Coauthors, 1996: A revised land surface parameterization (SiB2) for atmospheric GCMs. Part I: Model formulation. J. Climate, 9, 676705.

    • Search Google Scholar
    • Export Citation
  • Seyfried, M. S., Grant L. E. , Du E. , and Humes K. S. , 2005: Dielectric loss and calibration of the Hydra Probe soil water sensor. Vadose Zone J., 4, 10701079.

    • Search Google Scholar
    • Export Citation
  • Spitters, C. J. T., Toussaint H. A. J. M. , and Goudriann J. , 1986: Separating the diffuse and direct component of global radiation and its implications for modeling canopy photosynthesis. Part I: Components of incoming radiation. Agric. For. Meteor., 38, 217229.

    • Search Google Scholar
    • Export Citation
  • Stieglitz, M., Rind D. , Famiglietti J. , and Rosenzweig C. , 1997: An efficient approach to modeling the topographic control of surface hydrology for regional and global climate modeling. J. Climate, 10, 118137.

    • Search Google Scholar
    • Export Citation
  • Stöckli, R., and Coauthors, 2008: Use of FLUXNET in the Community Land Model development. J. Geophys. Res., 113, G01025, doi:10.1029/2007JG000562.

    • Search Google Scholar
    • Export Citation
  • Taylor, K. E., 2001: Summarizing multiple aspects of model performance in a single diagram. J. Geophys. Res., 106, 71837192.

  • Thornton, P. E., and Zimmermann N. E. , 2007: An improved canopy integration scheme for a land surface model with prognostic canopy structure. J. Climate, 20, 39023923.

    • Search Google Scholar
    • Export Citation
  • Tian, X., Dai A. , Yang D. , and Xie Z. , 2007: Effects of precipitation-bias corrections on surface hydrology over northern latitudes. J. Geophys. Res., 112, D14101, doi:10.1029/2007JD008420.

    • Search Google Scholar
    • Export Citation
  • Wallach, R., Grigorin G. , and Rivlin J. , 1997: The errors in surface runoff prediction by neglecting the relationship between infiltration rate and overland flow depth. J. Hydrol., 200, 243259.

    • Search Google Scholar
    • Export Citation
  • Wang, D., Wang G. , and Anagnostou E. N. , 2007: Evaluation of canopy interception schemes in land surface models. J. Hydrol., 347, 308318, doi:10.1016/j.jhydrol.2007.09.041.

    • Search Google Scholar
    • Export Citation
  • Wang, Y.-P., and Leuning R. , 1998: A two-leaf model for canopy conductance, photosynthesis, and portioning of available energy. Part I: Model description and comparison with a multilayered model. Agric. For. Meteor., 91, 89111.

    • Search Google Scholar
    • Export Citation
  • Xie, Z., and Yuan X. , 2010: Prediction of water table under stream–aquifer interactions over an arid region. Hydrol. Proc., 24, 160169, doi:10.1002/hyp.7434.

    • Search Google Scholar
    • Export Citation
  • Xue, Y., Sellers P. , Kinter J. , and Shukla J. , 1991: A simplified biosphere model for global climate studies. J. Climate, 4, 345364.

  • Xue, Y., Fennessy M. J. , and Sellers P. J. , 1996: Impact of vegetation properties on U.S. summer weather prediction. J. Geophys. Res., 101, 74197430.

    • Search Google Scholar
    • Export Citation
  • Xue, Y., Zeng F. J. , Mitchell K. E. , Janjic Z. , and Rogers E. , 2001: The impact of land surface processes on simulations of the U.S. hydrological cycle: A case study of the 1993 flood using the SSiB land surface model in the NCEP Eta regional model. Mon. Wea. Rev., 129, 28332860.

    • Search Google Scholar
    • Export Citation
  • Yang, D., Kane D. , Zhang Z. , Legates D. , and Goodison B. , 2005: Bias corrections of long-term (1973–2004) daily precipitation data over the northern regions. Geophys. Res. Lett., 32, L19501, doi:10.1029/2005GL024057.

    • Search Google Scholar
    • Export Citation
  • Yang, Z.-L., Dickinson R. E. , Henderson-Sellers A. , and Pitman A. J. , 1995: Preliminary study of spin-up processes in land surface models with the first stage data of Project for Intercomparison of Land Surface Parameterization Schemes Phase 1(a). J. Geophys. Res., 100, 16 55316 578.

    • Search Google Scholar
    • Export Citation
  • Yeh, P. J.-F., and Eltahir E. A. B. , 2005: Representation of water table dynamics in a land surface scheme. Part I: Model development. J. Climate, 18, 18611880.

    • Search Google Scholar
    • Export Citation
  • Yu, Z., Pollard D. , and Cheng L. , 2006: On continental-scale hydrologic simulations with a coupled hydrological model. J. Hydrol., 331, 110124.