Land-atmosphere coupling and summer climate variability over East Asia

Authors


Abstract

[1] Two long-term simulations with the weather research and forecasting model are conducted to assess the contribution of land-atmosphere coupling to interannual variability of summer climate over East Asia. The control experiment (CTL) uses a fully coupled land surface model, while an additional experiment replaces soil moisture evolution at each time step with the climatology of CTL and thus removes the interannual variability of soil moisture. CTL is able to reproduce relatively well climatic means and interannual variability of summer climate over East Asia though some biases exist. It is found that land-atmosphere coupling plays a critical role in influencing summer climate variability, in particular over the climatic and ecological transition zones. Interactive soil moisture strongly amplifies daily mean temperature variability over the southern Siberia–northern Mongolia region, the region from northeast China to central China, and the eastern part of South Asia, accounting for half or more of the total variance. Soil moisture is found to exert substantially stronger impacts on daily maximum temperature variability than on daily mean temperature variability but generally has small effects on daily minimum temperature except for the eastern Tibetan Plateau and some other areas. Soil moisture makes a dominant contribution to precipitation variability over the climatic and ecological transition zones of the southern Siberia–northern Mongolia region and northern China and many areas of western China. While soil moisture-temperature coupling is largely determined by the ability of soil moisture to affect surface fluxes, soil moisture–precipitation coupling also depends on other physical processes, particularly moisture convection.

1. Introduction

[2] The interactions of the atmosphere with the land surface particularly involving variable of soil wetness are of major importance for the climate system. Soil moisture plays a critical role in influencing surface energy and water balance components mainly through its effects on evapotranspiration or latent heat flux. In soil moisture–limited region, a soil moisture deficit can damp evapotranspiration and, consequently, more energy is partitioned into sensible heat, thus enhancing surface air temperature. Soil moisture–induced anomalies in surface fluxes may be also transferred into precipitation anomalies over some areas [e.g., Delworth and Manabe, 1989; Koster and Suarez, 1995; Zheng and Eltahir, 1998; Schlosser and Milly, 2002].

[3] The land-atmosphere coupling strengths vary spatially and with season. In the warm season, soil moisture impacts are suggested to be large relative to oceanic impacts over continental midlatitudes [e.g., Koster and Suarez, 1995; Koster et al., 2000; Douville, 2004; Conil et al., 2007]. Strong coupling of soil moisture with precipitation during summertime may mainly occur over the transition zones between dry and wet climates and/or between ecosystems [e.g., Koster et al., 2004; Seneviratne et al., 2006; Zhang et al., 2008b]. Due to its asymmetric effects on daytime and nighttime surface energy budgets, soil moisture affects daily maximum (Tmax) and minimum (Tmin) temperatures differently. In summer, a reduction of soil moisture is generally demonstrated to warm Tmax much more largely than Tmin, and vice versa [e.g., Dai et al., 1999; Durre et al., 2000; Diffenbaugh et al., 2005; Alfaro et al., 2006; Fischer et al., 2007; Zhang et al., 2009; Lorenz et al., 2010].

[4] Due to the scarcity of soil moisture observations, current understanding of land-atmosphere coupling is largely based on atmospheric general circulation model (AGCM) simulations (including many mentioned above). Back to the early 1980s, several pioneering studies [e.g., Shukla and Mintz, 1982; Yeh et al., 1984] detected the strong sensitivity of the climate response with respect to evapotranspiration and soil moisture anomalies. A large body of AGCM modeling studies has grown since then that examined how soil moisture anomalies affect precipitation and temperature on a variety of timescales [e.g., Delworth and Manabe, 1988, 1989; Beljaars et al., 1996; Fennessy and Shukla, 1999; Koster et al., 2000, 2004, 2010; Douville and Chauvin, 2000; Dirmeyer, 2001] (see also an overview by Seneviratne et al. [2010]). However, large uncertainties exist in AGCM studies of land-atmosphere interactions particularly regarding local to regional scales [e.g., Dirmeyer et al., 2006; Pitman et al., 2009; Wei et al., 2010b]. For example, the Global Land-Atmosphere Coupling Experiment (GLACE-1) demonstrated large spread in both strength and positioning of land-atmosphere coupling hot spots among a dozen of participating models while yielding a first global multimodel average distribution of land-atmosphere coupling strength [Koster et al., 2004, 2006; Guo et al., 2006]. Dirmeyer et al. [2006] found that the GLACE-1 participating models have large systematic errors compared to observed relationships between land and the atmosphere on the local to regional scales. It is also suggested that these models may overestimate the coupling strength on a global average basis since they tend to place too much variance at intraseasonal scales except at high latitudes [Wei et al., 2010a].

[5] Compared to AGCMs, regional climate models (RCMs) are more skillful at resolving orographic effects and other small-scale processes, thus better reproducing regional climate characteristics especially when driven by relatively realistic reanalysis data [e.g., Dickinson et al., 1989; Giorgi, 1990; Jones et al., 1995; Christensen et al., 1998; Leung and Ghan, 1998; Giorgi and Mearns, 1999; Hong and Leetmaa, 1999; Fennessy and Shukla, 2000]. Since the 1990s, RCMs have been used intensively to examine how land–surface boundary conditions or the initial state of soil moisture affect weather and climate [e.g., Paegle et al., 1996; Giorgi et al., 1996; Seth and Giorgi, 1998; Schär et al., 1999; Hong and Pan, 2000]. Very recently, RCM assessment of soil moisture influences on interannual climate variability has received more attentions since increased climate variability and associated potentially more extreme events can have a greater impact on both human society and the natural environment than do changes in mean climate [e.g., Seneviratne et al., 2006; Zhang et al., 2008a; Leung et al., 2011; Jaeger and Seneviratne, 2011]. These studies are generally carried out over the United States and Europe. In most cases, soil moisture interactions have been demonstrated to enhance temperature and precipitation variability and associated climate extremes such as heat waves. For example, Seneviratne et al. [2006] used four long-term RCM simulations to isolate the role of land-atmosphere coupling for projected changes in interannual variability of European climate during summer. They found that climate change–induced temperature variability increase is mainly attributed to feedbacks between the land surface and the atmosphere. Zhang et al. [2008a] (hereafter ZWL2008), based on a pair of long-term weather research and forecasting (WRF) model simulations with and without soil moisture interactions over the contiguous United States, identified that the hot spot where soil moisture can strongly amplify summer precipitation variability appears over the northern United States. The location of land-atmosphere coupling hot spot of ZWL2008 agrees to a large extent with that from statistical analysis to observations and observationally based analyses [Zhang et al., 2008b], and has been further supported by recent other studies [Meng and Quiring, 2010; Leung et al., 2011].

[6] Previous statistical and model assessments of East Asian or global land-atmosphere coupling mainly focused on the timescales from days to months [e.g., Koster et al., 2004; Kim and Hong, 2007; Chow et al., 2008; Zhang et al., 2008b; Dirmeyer et al., 2009]. Several previous studies have used AGCM simulations to investigate the contribution of soil moisture feedbacks to interannual climate variability on the global scale [e.g., Koster et al., 2000; Reale and Dirmeyer, 2002; Krakauer et al., 2010]. However, as mentioned above, AGCM studies of land-atmosphere coupling have large uncertainties on the local to regional scales. The objective of the present study is to improve our understanding of the role of land-atmosphere coupling for interannual variability of summer climate over East Asia with a RCM. We perform two long-term WRF integrations with and without soil moisture interactions to isolate the contribution of land-atmosphere coupling to summer climate variability from other factors (e.g., atmospheric circulation, sea surface temperature). Both simulations are driven by the National Centers for Environmental Prediction (NCEP)–Department of Energy (DOE) reanalysis [Kanamitsu et al., 2002] as the lateral boundary conditions that ensures a realistic representation of the large-scale atmospheric circulation. We examine the effects of interactive soil moisture not only on interannual variability of mean temperature (Tmean) and precipitation but also on Tmax and Tmin variability which is closely related with extreme temperature events.

[7] The paper is organized as follows. We start by describing the WRF model and the setup of numerical experiments in section 2. Section 3 presents the methods to quantify land-atmosphere coupling strength, and its contribution to summer climate variability over East Asia. Section 4 is devoted to evaluating model performance and identifying the model biases. The contributions of soil moisture–atmosphere coupling to temperature and precipitation variability are presented in sections 5 and 6, respectively. In section 7, further analyses are conducted to explain our findings. Finally, conclusions and discussion are given in section 8.

2. Model and Experimental Design

[8] We employ the WRF version 3.1.1 with the advanced research WRF (ARW) dynamics solver, which was developed at the National Center for Atmospheric Research (NCAR) [Skamarock et al., 2008]. WRF is a fully compressible, Euler nonhydrostatic model (with a runtime hydrostatic option available) with multiple options for physical parameterization suitable for applications across scales ranging from large-eddy to global simulations. A spatial resolution of 60 km was adopted in many previous RCM studies of East Asian or Asian regional climate [e.g., Leung et al., 1999; Wang et al., 2004; Fu et al., 2005]. Gao et al. [2006] investigated the role of spatial resolution in the RegCM2 simulation of East Asian precipitation, and found that a relatively fine resolution (60 km or 45 km) is needed to accurately simulate the observed precipitation patterns in their study. In this study, we use a 60 km horizontal grid resolution, and 28 terrain-following vertical layers. The model domain is centered at 36°N and 116°E and consists of 136 (west–east) × 96 (south–north) grid points, covering East Asia and the adjacent oceans (Figure 1). Figure 1 also presents topography and the analysis domain utilized in the study. NCEP-DOE reanalysis is used to provide initial and lateral boundary conditions and sea surface temperature for the simulations, updated every 6 h. The model physics packages include microphysics, cumulus parameterizations, surface physics, planetary boundary layer, and atmospheric radiation.

Figure 1.

WRF model domain and topography (in meters) and analysis domain (marked by the rectangle) defined in this study.

[9] In this study, we use WRF single-moment six-class (WSM6) microphysics scheme [Hong and Lim, 2006] and the Kain-Fritsch convective parameterizations [Kain, 2004]. In the WSM6 scheme, six prognostic water substance variables are included: the mixing ratios of water vapor, cloud water, cloud ice, snow, rain and graupel. The Kain-Fritsch scheme utilizes a simple cloud model with moisture updrafts and downdrafts, including the effects of detrainment, entrainment, and relatively simple microphysics. For surface physics, we use the unified Noah land surface model [Chen and Dudhia, 2001]. This land surface model consists of one canopy layer and four soil layers with the thicknesses of 10, 30, 60, and 100 cm from the top to down. It includes root zone in the upper 1 m soil layer, evapotranspiration, soil drainage, and runoff, taking into account vegetation categories, monthly vegetation fraction, and soil texture. The planetary boundary layer is parameterized by the Yonsei University scheme [Hong et al., 2006]. The scheme uses the countergradient terms to represent fluxes due to both local and nonlocal gradients. It includes an enhanced stable boundary layer diffusion algorithm that allows deeper mixing in windier conditions. Atmospheric radiations are calculated every 30 min using the NCAR community atmosphere model (CAM 3.0) spectral band shortwave and longwave radiation schemes [Collins et al., 2006].

[10] Two long-term simulations with the WRF model are performed, using the same approach as, e.g., Koster et al. [2004], Seneviratne et al. [2006], and ZWL2008, to separate the contribution of land-atmosphere coupling to summer climate variability over East Asia from the other factors (e.g., atmospheric circulation, sea surface temperature). A 21 year control run (CTL) uses interactive soil moisture, covering the period 1979–1999. CTL is initialized on 1 January 1979, with atmospheric and land surface conditions from the global reanalysis. In an additional simulation (SoilM), we repeat the integrations for summers of 1980–1999 with the same model configuration but replacing soil moisture evolution at each time step with the climatology of CTL. This removes the interannual variability of soil moisture which exists in CTL, and thus allows us to assess the role of interactive soil moisture in influencing summer climate variability. SoilM restarts on 1 June of each year and integrates to 31 August. The first 17 months (from January 1979 to May 1980) are used as model spin-up period to minimize the initialization effects of soil moisture and soil temperature. We analyze the following 20 summers of simulations after the spin-up period in this study.

[11] To evaluate simulated precipitation of CTL, we use the 0.5° gridded daily precipitation data set from the East Asia gauge-based analysis [Xie et al., 2007]. The data set has been constructed using the optimal interpolation-based technique to gauge observations at over 2200 stations collected from several individual sources. For temperature, we use a station observation-based global land monthly data set at a resolution of 0.5° from the Climate Prediction Center (CPC), NCEP [Fan and van den Dool, 2008]. The data set was developed using a combination of two large individual data sets of station observations collected from the Global Historical Climatology Network version 2 (GHCN) and the Climate Anomaly Monitoring System (CAMS).

3. Measures of Land-Atmosphere Coupling Strength

[12] The variance analyses and the GLACE-type coupling strength parameter are used to objectively quantify the land-atmosphere coupling strength and its contribution to summer climate variability.

[13] The two experiments designed in this study allow us to separate the contribution of interactive soil moisture to the interannual variability of a surface variable V from the other factors. We first calculate the difference (DsdV) in standard deviation of summer (June–July–August) mean V:

equation image

where σV(CTL) and σV(SoilM) represent the interannual standard deviation of summer mean V in the CTL and SoilM simulations, respectively.

[14] A percentage parameter PVv is further computed to measure the relative contribution of land-atmosphere coupling to summer precipitation and temperature variability:

equation image

where σV2(CTL) and σV2(SoilM) represent the interannual variance of summer mean V in the CTL and SoilM simulations, respectively. PVv can be interpreted as the fraction of the variance of a specific variable in CTL that results from soil moisture interactions. The fraction 1 − PVv stems from the effects of other factors (e.g., atmospheric circulation, sea surface temperature).

[15] The GLACE-1 study applied a coupling strength parameter that was proposed by Koster et al. [2000] to measure the contribution of soil moisture to the evolution of synoptic-timescale precipitation. In this study, we employ the revised GLACE-1 coupling strength parameter used by ZWL2008. For each specific variable, we ignore the first 10 days (1–10 June) to avoid the problems associated with initial “shocks” of the modeled atmosphere, and aggregate the following 80 days into 16 pentads (5 day periods). For each pentad, we have 20 year simulations. At each grid cell, we arrange 20 year simulations into one group (20 values in total), and thus get 16-group data. The time series V(t) in each group represent interannual variations of the variable at a specific pentad. We compute ΩV at the grid cell to measure the time series similarity:

equation image

where σequation image2 is the variance of the 16-group average time series (across 20 values in total) for the variable V(t), and σV2 is the variance of pentad mean climate variable computed from all values available in 16-group data (across 16 group by 20 pentads or 320 values in total). The coupling strength parameter for the climate variable is defined as ΩV difference between CTL and SoilM:

equation image

The ΩV(CTL) represents the similarity between pentads in interannual variability of a specific variable induced by all factors, and ΩV(SoilM) represents the similarity induced by everything but the soil moisture. The difference ΔΩV thus provides a measure of the degree to which the land-atmosphere coupling induces the similarity of interannual variations for one specific variable between pentads.

[16] The DsdV reflects the absolute impact of land-atmosphere coupling on summer climate variability, but its value depends on the locations particularly for precipitation. The PVv removes such dependence, and provides a more objective measure of the relative contribution of interactive soil moisture to interannual climate variability. The ΔΩV can be used to test if our results are dependent on the methods we choose.

[17] Several key processes involved in the soil moisture feedbacks including surface fluxes, atmospheric boundary layer processes, and clouds are of small scale, and, thus must be parameterized at the WRF model. The performance of Kain-Fritsch convective parameterization scheme may depend on climate regimes and the locations, and the resolution of 60 km is still too coarse to represent all important dynamical processes of mesoscale convective systems. Unrealistic representations of these processes will cause uncertainties in calculated land-atmosphere coupling strength. Recent several studies have noted the importance of realistic treatment of surface exchange coefficients for moisture and heat in the Noah land surface model which are closely related with the surface coupling strength [LeMone et al., 2008; Chen and Zhang, 2009; Chen et al., 2010]. These studies demonstrated that the Noah land surface model underestimated the heat exchange coefficient for forests, and overestimated it for short vegetation such as grass, crops, and shrub. This implies that the surface coupling is too strong (too weak) for short vegetation (tall vegetation) in the Noah land surface model. The uncertainties caused by parameterized convection, surface parameters and other processes should be kept in mind when the results are analyzed.

4. Model Evaluation

[18] To evaluate the capability of the WRF model to reproduce the observed climate of East Asia, we compare the CTL simulation during 1980–1999 to the observations, with focus on precipitation and surface air temperature. Figure 2 presents summer mean precipitation, standard deviation of summer precipitation, and the coefficient of variation (standard deviation normalized by the mean) in the East Asia gauge-based precipitation analysis [Xie et al., 2007] and the CTL simulation. Spatially, observed summer mean precipitation exhibits a clear southeast to northwest gradient, with maximum precipitation occurring over the tropics, southern China, Korea, and Japan (>4 mm/d), and minimum precipitation over the northwest (<1 mm/d). Generally speaking, the WRF model simulates well climatic means of summer precipitation over most of East Asia, both in the magnitude and geographical distribution. However, it underestimates summer precipitation over some areas of western China, and simulates too much precipitation over Indochina and some other areas. Simulating and predicting interannual variations of East Asian summer monsoon precipitation is still a major challenge in both global and regional climate model communities since floods and droughts often occur over different regions, and many processes involved are not well understood [e.g., Kang et al., 2002; Qian and Leung, 2007]. The precipitation variability expressed in terms of the standard deviation of summer precipitation is largely dependent on the precipitation mean, exhibiting a similar pattern to that of summer mean precipitation in both the observations and the CTL simulation. The summer precipitation variability is generally represented well in the CTL simulation compared to the observations. The model biases mainly occur over the areas that have large mean precipitation deficiencies. To remove the dependency of the standard deviation on the mean precipitation, we further calculate precipitation coefficients of variation which provide a more objective measure of interannual variability [Giorgi et al., 2004]. The observed spatial structure of the coefficients is well characterized by the CTL simulation. Regarding the magnitude, the WRF model tends to overestimate the coefficients to some degree especially over many areas at relatively high latitude (>40°N).

Figure 2.

The 1980–1999 summer mean precipitation, standard deviation of summer precipitation, and the coefficient of variation (standard deviation divided by the mean) in (left) the East Asia gauge-based precipitation analysis [Xie et al., 2007] and (right) CTL: (a, b) mean (in mm/d), (c, d) standard deviation (in mm/d), and (e, f) coefficient of variation. The grid cells with summer mean precipitation of 1–4 mm/d are marked by the circles in the regions (82°E–129°E, 25°N–57°N). The marked areas roughly indicate the climatic and ecological transition zones over East Asia.

[19] Figure 3 presents the summer surface air temperature mean and the standard deviation in the observations [Fan and van den Dool, 2008] and the CTL simulation. A general agreement is found between summer mean temperatures in the observations and the CTL simulation, especially with regard to the geographical distribution. The WRF model exhibits warm biases over many areas relative to the observations. The observed temperature standard deviation, and thus the interannual variability, exhibits a large south-to-north gradient. The standard deviation is smallest in the tropics and subtropics where climate is largely affected by oceanic processes. The WRF model successfully reproduces the south-to-north gradient, and also realistically simulates the magnitude of the temperature variability over most areas south of 45°N. However, over northeast Asia, the variability is largely overestimated by the model.

Figure 3.

The 1980–1999 summer mean surface air temperature and standard deviation of summer temperature in (left) GHCN_CAMS [Fan and van den Dool, 2008] and (right) CTL: (a, b) mean and (c, d) standard deviation.

5. The Role of Interactive Soil Moisture for Summer Temperature Variability

[20] Figure 4 presents differences in standard deviations of summer Tmean, Tmax, and Tmin between CTL and SoilM (DsdT). Since soil moisture interactions are disabled in SoilM, the difference fields reflect the variability induced by the land-atmosphere coupling. Compared to SoilM, interannual variability of summer Tmean in CTL is generally enhanced. Particularly large increases in the standard deviation of Tmean are induced over the southern Siberia–northern Mongolia region, the region from northeast China to central China, and eastern part of South Asia with the magnitude on the order of 0.2°C–0.8°C. Many areas over which soil moisture has significant effects on Tmean variability are located in the climatic and ecological transition zones as roughly indicated in Figure 2a. The land-atmosphere coupling is found to exert a diurnal asymmetric impact on surface air temperature variability. It shows a stronger ability to enhance the variability of Tmax than that of Tmean, with much more grid cells passing the 90% confidence level. In contrast, soil moisture generally plays a small role in influencing the Tmin variability. Soil moisture only exerts a significant positive control on the Tmin standard deviation over the eastern Tibetan Plateau and some other areas. It is interesting to note that the significantly decreased Tmin variability appears over the middle and lower reaches of the Yangtze River basin. The asymmetric impact on Tmax and Tmin variability are not unexpected. The daytime Tmax strongly depends on surface solar heating and the partitioning of net energy into sensible and latent heat fluxes, which are largely affected by soil moisture. In contrast, the nighttime Tmin is largely determined by the greenhouse effect of the atmospheric water vapor, which is closely linked to large-scale processes.

Figure 4.

Difference in standard deviation of summer temperature (in °C) between CTL and SoilM (DsdT): (a) Tmean, (b) Tmax, and (c) Tmin. Grid cells with values significant at the 90% level by F test are marked by the solid circles.

[21] Figure 5 shows soil moisture–temperature coupling strengths measured by the two parameters PVT and ΔΩT. Results obtained from the two measures show similar patterns regarding regions of strong soil moisture–temperature coupling. The percent variance contributed by interactive soil moisture can be measured by PVT.

Figure 5.

(left) Percentage of interannual summer temperature variance due to land-atmosphere coupling (PVT) and (right) the GLACE-type coupling strength parameter (ΔΩT): (a, d) Tmean, (b, e) Tmax, and (c, f) Tmin.

[22] Hot spots of soil moisture–Tmean coupling mainly appear over the southern Siberia–northern Mongolia region, the region from northeast China to central China, and eastern part of South Asia. Over these areas, soil moisture interactions contribute to half or more of the total Tmean variance (Figure 5a). These hot spots largely coincide with the regions of strong soil moisture feedbacks on Tmean identified by Zhang and Dong [2010] by statistical analyses to the Global Land Data Assimilation System soil moisture [Rodell et al., 2004] and observational temperature, and European Centre for Medium-Range Weather Forecasts 40 year reanalysis (ERA-40) soil moisture and temperature [Uppala et al., 2005]. Also, regions of strong soil moisture–Tmean coupling identified by this study (and also ZWL2008) appear to be consistent in many areas with those by the GLACE-1 study, although detailed structures have some disagreement.

[23] The land-atmosphere coupling plays a leading role in influencing summer Tmax variability almost over all land areas outside of western China and Japan. Particularly large effects are seen over the hot spots of soil moisture–Tmean coupling with more than 60% of the total variance explained over many areas (Figure 5b). Strong coupling of soil moisture with Tmax suggests that, as found in previous studies for the United States and Europe [e.g., Diffenbaugh et al., 2005, 2007; Fischer et al., 2007], soil moisture is an important contributor to the occurrence of extreme hot temperatures and heat waves during summer over East Asia.

[24] The land-atmosphere coupling only makes a limited contribution to summer East Asian Tmin variability which is highly constrained by other factors including atmospheric circulation and sea surface temperature. Soil moisture interactions play a role in amplifying the Tmin variability mainly over the zone from the eastern Tibetan Plateau to northern China, and eastern South Asia. The greater effects of soil moisture on Tmax than Tmin also have been demonstrated over other regions in previous studies [e.g., Dai et al., 1999; Durre et al., 2000].

[25] For all three temperature variables, the enhancement of summer variability by the land-atmosphere coupling dominates. Yet there are some exceptions. For example, soil moisture interactions tend to decrease summer Tmin variability over the middle and lower reaches of the Yangtze River basin and some other isolated areas.

6. The Role of Interactive Soil Moisture for Summer Precipitation Variability

[26] Figure 6 presents differences in standard deviations of total precipitation, convective precipitation, and large-scale precipitation between CTL and SoilM (DsdP). The interannual variability of summer precipitation is significantly enhanced by land-atmosphere coupling over the climatic and ecological transition zones of the southern Siberia–northern Mongolia region and northern China, and many areas of western China. Decreased precipitation variability is mainly found over some areas of the Yangtze-Huai River Valley. Over the rest of the analysis domain, changes in the precipitation variability are much heterogeneous; neither positive nor negative effects dominate. Convective precipitation changes show a similar pattern to that of changes in total precipitation, but with the smaller magnitudes. Overall, changes in the variability of large-scale precipitation are more inhomogeneous over the monsoon region, and smaller over other areas than those in total and convective precipitation.

Figure 6.

Difference in standard deviation of precipitation (in mm/d) between CTL and SoilM (DsdP): (a) total precipitation, (b) convective precipitation, and (c) large-scale precipitation. Grid cells with values significant at the 90% level by F test are marked by the solid circles.

[27] To avoid the effects of precipitation heterogeneity, we calculate the percentage of total interannual summer variance induced by the land-atmosphere coupling (PVP) and the GLACE-type coupling strength parameter (ΔΩP) for total precipitation, convective precipitation, and large-scale precipitation using model data smoothed in each direction with a nine-point filter (Figure 7). Two measures consistently show that the land-atmosphere coupling amplifies summer precipitation variability over the climatic and ecological transition zones of the southern Siberia–northern Mongolia region and northern China, and many areas of western China. Over these areas, land-atmosphere coupling accounts for about half of the total variance of the summer precipitation (Figure 7a). Overall, soil moisture control on precipitation variability is small over the monsoon region. While the positive sign dominates the soil moisture–precipitation coupling, the negative sign exists over part of the Yangtze-Huai River Valley and the areas from North Korea to eastern part of northeast China. Some studies have provided evidence for the existence of negative soil moisture feedbacks [e.g., Giorgi et al., 1996; Findell and Eltahir, 2003; Cook et al., 2006]. Except for negative feedbacks, it is also possible that the decreased climate variability is caused by other reasons such as large-scale effects.

Figure 7.

(left) Percentage of interannual summer precipitation variance due to land-atmosphere coupling (PVP) and (right) the GLACE-type coupling strength parameter (ΔΩP): (a, d) total precipitation, (b, e) convective precipitation, and (c, f) large-scale precipitation.

[28] Further analysis shows that convective precipitation is more sensitive to soil moisture variations than large-scale precipitation, dominating the coupling of soil moisture with total precipitation (Figures 7b, 7c, 7e, and 7f). Soil moisture exerts some impacts on large-scale precipitation mainly over the transition zone of northern China and some areas over western China and southern Siberia. The dominance of convective precipitation in the soil moisture's impact on total precipitation also showed up in the GLACE-1 study and ZWL2008.

[29] The GLACE-1 study used ensemble experiments from a dozen of AGCMs to obtain a multimodel average depiction of the global distribution of land-atmosphere coupling strength for the first time. Due to the limited capacity of AGCMs to represent small-scale processes, however, even such a multimodel estimate of soil moisture–precipitation coupling strength may still have many uncertainties on local to regional scales. For example, over the United States, WRF modeling study by ZWL2008 and statistical analysis to land assimilation product, reanalysis data and available observations [Zhang et al., 2008b] consistently identified that hot spot of soil moisture–precipitation coupling during summer appears over the northern United States. This result does not support the GLACE-1 hot spot hypothesis over the central United States, but is largely in agreement with the region where realistic land surface initialization contributes to the skill of subseasonal precipitation in GLACE-2 [Koster et al., 2010]. Over East Asia, the GLACE-1 study found that soil moisture–precipitation coupling strength during summer appears to be stronger over some monsoon areas of eastern China relative to the areas outside of the monsoon region, while our result together with previous studies [e.g., Zhang et al., 2008b; Dirmeyer et al., 2009] demonstrates that the coupling strength is generally weak over the monsoon areas.

[30] Since different studies focus on different time scales and study periods, this type of comparisons should be treated with caution. In addition, the results of this study and ZWL2008 agree qualitatively with those of the GLACE-1 study, although the exact locations of soil moisture–precipitation coupling hot spots disagree. They all find that positive soil moisture feedbacks dominate, and the feedbacks are generally weak over too wet or too dry areas. Also, they consistently agree that the land's control on surface air temperature is generally larger than its control over precipitation.

7. Further Analyses

[31] In summer, evapotranspiration and the associated change in sensible heat, can effectively translate a soil moisture anomaly into temperature and precipitation anomalies. The GLACE-1 study demonstrated that the ability of the soil moisture to affect evapotranspiration can explain most of the differences between the participating models and within a given model. That is, evapotranspiration variations are relatively high and are strongly dependent on soil moisture that resulting in a high land-atmosphere coupling strength. Figure 8 presents the distributions of the products of the GLACE-type coupling strength parameter and standard deviation for latent heat and sensible heat. The product can measure the degree to which the surface flux signal varies strongly and consistently with soil moisture. The products for latent and sensible heat fluxes exhibit a similar spatial pattern. High values mainly appear over the climatic and ecological transition zones. While the low products in dry and wet areas are mainly caused by low soil moisture variability and low sensitivity of surface fluxes to soil moisture, respectively.

Figure 8.

Distribution of the product of the GLACE-type coupling strength parameter and standard deviation: (a) latent heat and (b) sensible heat.

[32] Spatial structures of the products for latent and sensible heat fluxes resemble those of the coupling of soil moisture with Tmean and Tmax. This indicates that soil moisture–temperature coupling strength is largely controlled by the sensitivity of latent and sensible heat fluxes to soil moisture. The regions where strong soil moisture–precipitation coupling occurs generally have high values for the product of latent heat. However, a high value for latent heat product does not necessarily guarantee a high soil moisture feedback on precipitation since precipitation also depends on convection and atmospheric boundary layer formulations. Figure 9 presents the GLACE-type coupling strength parameter for the atmospheric boundary layer height (ΔΩABL). The ΔΩABL exhibits a consistent pattern with those of the products for latent and sensible heat fluxes, suggesting that the sensitivity of the atmospheric boundary layer is largely determined by the sensitivities of surface heat fluxes. The dependence of moisture convection on surface fluxes (Figures 7b and 7e) and the ability of soil moisture to affect surface fluxes (both in term of mean and variability) together determine the locations of soil moisture–precipitation coupling hot spots. Over areas where weak land-atmosphere coupling exists, summer climate variability is highly constrained by other factors including atmospheric circulation and sea surface temperature.

Figure 9.

The GLACE-type coupling strength parameter for the atmospheric boundary layer height (ΔΩABL).

8. Conclusions and Discussion

[33] This study investigates the role of land-atmosphere coupling for summer climate variability over East Asia using two long-term simulations with the WRF model, driven with NCEP-DOE reanalysis data. The control run covering the period 1979–1999 allows soil moisture to freely interact with the atmosphere, and an additional experiment repeats summer integrations for 1980–1999 but with soil moisture–atmosphere interactions disabled by using prescribed climatology of CTL. Comparing the two experiments allows us to isolate the contribution of land-atmosphere coupling to summer climate variability.

[34] Land-atmosphere coupling is found to strongly increase summer Tmean variability over the southern Siberia–northern Mongolia region, the region from northeast China to central China, and eastern part of South Asia, accounting for half or more of the total variance. The effects on Tmax and Tmin differ significantly. Compared to Tmean, the land-atmosphere coupling enhances Tmax variability with larger magnitude and over more areas. In contrast, the amplifying effect of land-atmosphere coupling on Tmin is much smaller, and is limited to the eastern Tibetan Plateau and some other areas. Soil moisture's control on precipitation, though weaker than its control on Tmean and Tmax, is still important over many areas. The land-atmosphere coupling plays a leading role in amplifying summer precipitation variability over the climatic and ecological transition zones of the southern Siberia–northern Mongolia region and northern China, and many areas of western China.

[35] While the land-atmosphere coupling mainly plays a role in amplifying summer climate variability, it may act to damp climate variability over some particular areas. For example, soil moisture interactions may cause decreased summer Tmin variability over the middle and lower reaches of the Yangtze River basin and some other isolated areas, and decreased summer precipitation variability over part of the Yangtze-Huai River Valley and the areas from North Korea to eastern part of northeast China. However, the total areas with negative sign are relatively small. These changes may be caused by other reasons such as large-scale effects.

[36] The amplification of temperature variability through land-atmosphere coupling is largely controlled by the ability of soil moisture to affect surface heat fluxes. Precipitation is produced at higher atmospheric levels, and its coupling with soil moisture also depends on other physical processes, particularly moisture convection.

[37] Overall, summer precipitation and temperature means and interannual variability simulated by the WRF model compare relatively well with the observations. Meanwhile, the model exhibits substantial biases over some areas, which may be associated with model resolution and domain, model physical formulations and large-scale driving fields. Giorgi and Bi [2000] found that some of the effects of the internal RCM variability are comparable in magnitude to those of modifications in physical forcings, and thus cautioned that the internal model variability should be taken into consideration in the design, analysis, and interpretations of RCM experiments. It is possible that, at least in some cases, the internal model variability contributes to low RCM skill in simulating interannual summer climate anomalies [Qian and Leung, 2007]. More studies are clearly needed to understand and clarify these issues. Since our simulations are driven by prescribed sea surface temperature, the possible sea surface temperature feedbacks are thereby neglected. A comparison of simulations from the WRF model with coupled ocean-atmosphere system and driven by observed sea surface temperature would be valuable to further our understanding of soil moisture feedbacks on climate variability.

[38] This study highlights the importance of land-atmosphere coupling in influencing summer climate variability over East Asia. The hot spots for the coupling of soil moisture with temperature and precipitation are largely comparable with those from observations and observationally based analyses [Zhang et al., 2008b; Dirmeyer et al., 2009; Zhang and Dong, 2010] though the focal time scales are different in different studies. However, the hot spots of soil moisture–precipitation coupling do not support those identified by the GLACE-1 study over East Asia. The WRF model simulates relatively well climatic means and interannual variability over East Asia despite some biases, and can be a valuable tool to elucidate feedback mechanism. This agreement with previous studies based on observations and observationally based analyses is encouraging, and provides confidence that the modeling study represents a relatively realistic estimate of the locations of strong land-atmosphere coupling over East Asia. The WRF model with a more realistic description of land-atmosphere interactions particularly over the land-atmosphere coupling hot spots can serve as a useful tool for seasonal-to-interannual temperature and precipitation forecasting.

Acknowledgments

[39] We would like to thank the editor and the three reviewers for their valuable comments, which helped to significantly improve the quality of the manuscript. We thank Pingping Xie for supplying the East Asia gauge-based precipitation analysis. NCEP/DOE reanalysis which is used to drive the WRF was produced with the support of the U.S. National Weather Service and of PCMDI (U.S. Department of Energy). The work was supported by “100-talent program” of the Chinese Academy of Sciences, special fund for President's prize of the Chinese Academy of Sciences, and the National Basic Research Program of China (2009CB421405).