Modelling the effect of soil moisture variability on summer precipitation variability over East Asia

Authors

  • Zhongxian Li,

    Corresponding author
    1. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, China
    2. Key Laboratory of Meteorological Disaster of Ministry of Education (KLME), Nanjing University of Information Science and Technology, China
    • Correspondence to: Z. Li, Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, Nanjing 210044, China. E–mail: lizhongxian@nuist.edu.cn

    Search for more papers by this author
  • Tianjun Zhou,

    1. State Laboratory of Numerical Modeling for Atmospheric Sciences and Geophysical Fluid Dynamics, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
    Search for more papers by this author
  • Haishan Chen,

    1. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, China
    2. Key Laboratory of Meteorological Disaster of Ministry of Education (KLME), Nanjing University of Information Science and Technology, China
    Search for more papers by this author
  • Donghong Ni,

    1. Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters, Nanjing University of Information Science and Technology, China
    2. Key Laboratory of Meteorological Disaster of Ministry of Education (KLME), Nanjing University of Information Science and Technology, China
    Search for more papers by this author
  • Rong-Hua Zhang

    1. Key Laboratory of Ocean Circulation and Waves, Institute of Oceanology, Chinese Academy of Sciences, Qingdao, China
    Search for more papers by this author

ABSTRACT

The potential role of surface soil moisture (SSM) in improving the simulation of interannual East Asian summer precipitation is investigated using a coupled land-atmosphere climate model, the Community Atmosphere Model version 3 (CAM3). Forced by specified observational sea surface temperature (SST), two ensemble simulations for 22 boreal-summer seasons (1979–2000) are conducted, one with interannually varying SSM (SSMinter) forcing and another with climatological SSM (SSMclim) forcing. Results show that, relative to the SSMclim run, interannual variability of East Asian summer precipitation is better simulated in the SSMinter run over the mid- and high-latitudes of East Asia, especially in northwest China, where the correlation coefficient between precipitation simulated and observed from the Climate Prediction Center Merged Analysis of Precipitation (CMAP) is increased from 0.12 to 0.56 during 1979–2000. Meanwhile, positive relationships between anomalies of local soil evaporation and summer precipitation in northwest China can be better reproduced in the SSMinter run than in the SSMclim run. Possible mechanisms for the improved simulations of the interannual summer precipitation variability in northwest China are analysed; the improvement is resulted from the reasonable reproduction of anomalous atmospheric circulation in Siberia and Iranian plateau.

1 Introduction

East Asia is dominated by a typical monsoon climate. Both the economy and society of the region rely heavily on the onset and retreat of monsoon precipitation. The East Asian summer monsoon exhibits obviously interannual variability. The East Asian summer precipitation prediction is of crucial importance (Chou and Xu, 2001). At present, remarkable challenges exist in understanding mechanisms of the East Asian precipitation variability and its accurate prediction (Zhou and Zou, 2010). For example, in many previous studies, atmospheric general circulation models (AGCMs), when forced by prescribed SSTs, have been used to study monsoon-related precipitation variability in the region (Zhou and Li, 2002; Lang and Wang, 2005; Shi et al., 2008; Zhou et al., 2009a; Zhou and Zou, 2010). However, many AGCMs fail in simulating precipitation variations in East Asia (Zhou et al., 2008, 2009b; Li et al., 2010).

In the past three decades, considerable attention has been paid to soil moisture memory and its role in coupled land–climate variations (Walker and Rowntree, 1977; Shukla and Mintz, 1982; Yeh et al., 1984; Dirmeyer, 2000, 2011; Dirmeyer et al., 2009). It is found that the potential predictability of the atmosphere is very sensitive to initial soil moisture conditions (Koster et al., 2000; Reale and Dirmeyer, 2002). Through positive feedback onto the atmosphere, land surface conditions (e.g. soil moisture) apparently make a contribution to the regional climate anomalies (Dirmeyer, 2005). Thus, realistic soil moisture boundary conditions are important for simulating seasonal precipitation anomalies (Koster et al., 2004, 2006; Guo et al., 2006; Kanae et al., 2006; Conil et al., 2007).

Indeed, many previous studies have shown that land surface process plays an important role in the East Asian monsoon region, where its impact on the monsoon system is significant. For example, it is shown that soil moisture in spring and summer is very important for regional climate simulation and prediction over Huaihe River basin (Lin et al., 2001). As identified by Yang et al. (2001), when comprehensive land surface processes are incorporated into a climate model, the model's capability in reproducing the monsoon circulation and the related precipitation is enhanced. Xue et al. (2004) explored the influence of soil and vegetation biophysical processes on intraseasonal monsoon development. Zhang et al. (2012) suggested that the soil moisture may serve as a useful forcing factor in improving seasonal forecast skills over China. But little is known about the influence of summer soil moisture on monsoon precipitation in the East Asian region. The motivation of this study is to assess and understand the influence of surface soil moisture (SSM) on interannual variability of East Asian summer precipitation by conducting numerical experiments. We show evidences that interannual SSM variation plays an important role in simulating the precipitation rate anomaly over northwest China.

The remainder of the paper is organized as follows. Experimental design and observed data used to evaluate the model are described in Section 'Experimental design and data'. Section 'Simulation of summer precipitation over East Asia' presents the simulated interannual variability of East Asian summer precipitation. Section 'Anomalies of soil evaporation and 500 hPa atmospheric circulation associated with precipitation variations in northwest China' discusses the mechanisms responsible for the effects of SSM on the East Asian summer precipitation. A conclusion is given in Section 'Conclusions'.

2 Experimental design and data

2.1 Model and experimental design

The climate model used in this study is the Community Atmosphere Model version 3 (CAM3), the fifth generation of the National Center for Atmospheric Research (NCAR) AGCM (Collins et al., 2004), which employs an Eulerian dynamic core on a T42 (∼2.8°) horizontal resolution and 26 vertical levels. The performance of CAM3 model in the simulation of East Asian summer monsoon has been systematically evaluated by Chen et al. (2010).

CAM3 is coupled with the Community Land Model version 3 (CLM3), which represents land surface processes within the context of global climate simulation (Oleson et al., 2004). CLM3 has energy and water biases resulting from deficiencies in some of its canopy and soil parameterizations related to hydrological processes (Oleson et al., 2008). Although the coupling of surface fluxes to the atmosphere in CAM3 is strong, the coupling of soil moisture to surface fluxes in CLM3 is weak (Guo et al., 2006).

Two ensemble runs are preformed from May to August for each year during 1979–2000 with the specified observational SSTs. The SST data set is a blended version of the HadISST and Reynolds data set (Hurrell et al., 2008). In each year, eight ensemble members are initialized at 0000 UTC on early 8 days of May, from National Center for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) reanalysis products (Kalnay et al., 1996). Two ensemble simulations are performed using the NCAR CAM3, which is coupled with CLM3. One is a climatological SSM (SSMclim) run, in which SSM data set is derived from mean CLM3 climatology during 1979–2000. Another is an interannual SSM (SSMinter) run, in which SSM is taken from the NCEP-NCAR reanalysis six-hourly 0–10 cm soil moisture data during 1979–2000 (Kalnay et al., 1996). In the SSMinter simulations, the 0–10 cm soil moisture field in the NCEP-NCAR reanalysis is interpolated to the model grid on the top four CLM3 layers (about 0.7, 2.8, 6.2, and 11.9 cm) at every time step. The analysis is based on the ensemble mean of eight realizations.

2.2 Data

The data used as observational evidences to gauging model performance include the MERRA-Land soil evaporation from January 1980 to December 2000 (Reichle et al., 2011), and 500 hPa geopotential height field obtained from the European Centre for Medium-Range Weather Forecasts (ECMWF) Re-analysis (ERA-40) from January 1979 to December 2000 (Uppala et al., 2005), and the Climate Prediction Center Merged Analysis of Precipitation (CMAP) data from January 1979 to December 2000 (Xie and Arkin, 1996). The horizontal resolution of soil evaporation is 1/2° × 2/3°, and that of geopotential height and precipitation is 2.5° × 2.5°. The analysis period spans the boreal summer season, including June, July, and August (JJA) from 1979 to 2000, respectively.

We use the MERRA-Land soil evaporation instead of ERA-40 in this analysis. ECMWF uses soil moisture nudging to correct near-surface temperature and humidity errors via the evaporation pathway, regardless of the actual source of the errors (Mahfouf 1991; Douville et al., 2001; Uppala et al., 2005). Thus, the evaporation is as much controlled by systematic errors in reanalysis screen-level quantities as by precipitation in many locations. Modern-Era Retrospective Analysis for Research and Applications (MERRA) is a reanalysis product generated by the National Aeronautics and Space Administration (NASA) Global Modeling and Assimilation Office (GMAO) using the Goddard Earth Observing System (GEOS) version 5.2.0. The system incorporates information from in situ and remote sensing observations of the atmosphere (Rienecker et al., 2011). MERRA land surface estimates, however, utilize no directly assimilated land surface observations. MERRA-Land is the postprocessed data set of the MERRA land fields. The postprocessing of the MERRA land fields involves the reintegration of the land surface model with more realistic precipitation forcing and with a parameterization change designed to counteract certain known problems with MERRA's diurnal rainfall and radiation cycles (Reichle et al., 2011). Like all model products today including all reanalyses, the evaporation is still a model product. MERRA-Land is partially constrained by observations because observed precipitation is specified, but there are no observations of evaporation involved in MERRA-Land. This makes the MERRA-Land evaporation better than that of ERA-40, but it still has limitations.

3 Simulation of summer precipitation over East Asia

The geographical distributions of climatological mean and standard deviation of East Asian summer precipitation rate are shown in Figure 1. In the CMAP data, the climatological mean shows large values over the Bay of Bengal, northwest Pacific, and eastern Arabian sea, with the maximum value of 15 mm day−1 over the Bay of Bengal, and small values over the mid- and high-latitudes of Asia, i.e. northwest China (Figure 1(a)). The two simulations well reproduce the pattern of East Asian summer precipitation (Figure 1(c) and (e)). But compared with the CMAP data, the two simulations exhibit some obvious discrepancies over East Asia, especially over the Bay of Bengal and northwest Pacific. As shown in Table 1, the climatological mean of summer precipitation over the Bay of Bengal estimated from the SSMclim (SSMinter) run is only 36% (48%) of that from the CMAP data. In addition to model bias, uncertainties in the observations should also be noted (Sperber et al., 2013).

Figure 1.

(a,c,e) Climatological mean and (b,d,f) standard deviation of East Asian summer precipitation rate (mm day−1): (a) and (b) are derived from the CMAP data; (c) and (d) are derived from the SSMclim simulations; (e) and (f) are derived from the SSMinter simulations.

Table 1. Climatological mean and standard deviation of summer precipitation rate over the Bay of Bengal, northwest Pacific and northwest China (mm day−1).
  Climatological meanStandard deviation
Bay of Bengal (15°–25°N, 85°–95°E)CMAP13.972.84
SSMclim simulations4.981.06
SSMinter simulations6.651.26
Northwest Pacific (15°–30°N, 120°–135°E)CMAP8.021.83
SSMclim simulations4.280.89
SSMinter simulations4.560.86
Northwest China (38°–45°N, 80°–100°E)CMAP0.560.19
SSMclim simulations0.240.04
SSMinter simulations0.840.11

In the CMAP data, the maximum value standard deviation of 3 mm day−1 is found over the Bay of Bengal, and the minimum value of 0.2 mm day−1 over northwest East Asia (Figure 1(b)). The geographical distribution of standard deviation is reasonably depicted by the two simulations (Figure 1(d) and (f)), but the standard deviation of summer precipitation rate at the low latitudes is underestimated. For example, the standard deviation of summer precipitation over the Bay of Bengal calculated from the SSMclim (SSMinter) run is only 37% (44%) of that from the CMAP data (Table 1).

Figure 2 shows the correlation coefficients between CMAP data and the two simulations for East Asian precipitation rate anomaly in JJA during 1979–2000. It is shown that interannual variability of East Asian precipitation in the SSMinter simulations is closer to the CMAP data than that in the SSMclim simulations, especially in the western and central Tibetan Plateau, western China, and Changjiang-Huaihe River basin. When comparing Figure 2(a) with Figure 2(b), it is apparent that the anomalous precipitation in the western-central Tibetan Plateau and western China in the SSMclim (SSMinter) simulations is negatively (positively) correlated with the CMAP data. It is indicated that the SSM variability can significantly affect the simulation of interannual variability of East Asian summer precipitation.

Figure 2.

Correlation coefficients between the CMAP data and model simulations for the East Asian precipitation rate in JJA during 1979–2000: (a) the SSMclim simulations; (b) the SSMinter simulations. Only regions with correlation exceeding a 90% confidence level are shaded.

Figure 3 depicts the interannual variability of precipitation rate anomaly in northwest China (38°–45°N, 80°–100°E) in JJA during 1979–2000. It is seen that the amplitude of precipitation rate anomaly is underestimated in the SSMclim simulations, whereas it is well reproduced in the SSMinter simulations. The standard deviation of summer precipitation rate in northwest China is 0.19 mm day−1 in the CMAP data, 0.04 mm day−1 in the SSMclim simulations and 0.11 mm day−1 in the SSMinter simulations (Table 1), respectively. The correlation coefficient between the SSMinter simulations and the CMAP data in northwest China is 0.56, which is statistically significant at a 98% confidence level. But the correlation coefficient between the SSMclim simulations and the CMAP data in northwest China is only 0.12. These results indicate that interannual SSM variation plays a key role in simulating the precipitation anomaly over northwest China, where the land-atmosphere feedbacks is significant (Dirmeyer, 2011; Dirmeyer et al., 2012).

Figure 3.

Interannual variability of precipitation rate anomaly (mm day−1) in northwest China (38°–45°N, 80°–100°E) in JJA during 1979–2000. The bar, dashed and solid lines represent the results from the CMAP data, the SSMclim simulations and the SSMinter simulations, respectively.

4 Anomalies of soil evaporation and 500 hPa atmospheric circulation associated with precipitation variations in northwest China

Figure 4(a) presents the correlation coefficients between the MERRA-Land soil evaporation and the CMAP precipitation rate over northwest China in JJA during 1979–2000. The correlations show that West China is located in moisture-limited regimes, whereas East China is located in energy-limited regimes, following directly from those of Koster et al. (2002). Positive values are found in the mid- and high-latitudes of East Asia, especially in northwest China, with a maximum of 0.70 (at a 98% confidence level). It is indicated that the local soil evaporation anomaly is positively related with the summer precipitation anomaly in northwest China. Also, in the SSMinter run (Figure 4(c)), positive values are seen in northwest China, with a maximum of 0.70 (at a 98% confidence level). However, in the SSMclim run, negative values are found in northwest China, with the minimum correlation coefficient of −0.70 (at a 98% confidence level) (Figure 4(b)).

Figure 4.

Correlation coefficients between soil evaporation and precipitation rate in northwest China in JJA during 1980–2000. (a) the MERRA-Land soil evaporation and CMAP precipitation; (b) the SSMclim-simulated soil evaporation and precipitation; (c) the SSMinter-simulated soil evaporation and precipitation. Only regions with correlation exceeding a 98% confidence level are shaded.

Figure 5 shows the correlation coefficients between the MERRA-Land data and model simulations for East Asian soil evaporation in JJA during 1979–2000. Dirmeyer et al. (2009) suggested that the soil moisture controls the evaporation rates in arid and semiarid regions, such as in northwest China. It is expected that there are obvious differences between the SSMinter run and SSMclim run in simulating the evaporation anomaly in northwest China. In the SSMclim run (Figure 5(a)), small positive values or negative values are seen in northwest China, indicating that the model fails in reproducing the interannual variability of local soil evaporation in northwest China. However, in the SSMinter run (Figure 5(b)), positive values emerge in northwest China, with a maximum of 0.4 (at a 90% confidence level), indicating that interannual soil evaporation variability in northwest China is reasonably reproduced in the SSMinter run.

Figure 5.

Correlation coefficients between the MERRA-Land data and model simulations for the East Asian soil evaporation in JJA during 1980–2000: (a) the SSMclim simulations; (b) the SSMinter simulations. Only regions with correlation exceeding a 90% confidence level are shaded.

These results show that, forced by the observed SST, interannual variability of JJA precipitation in northwest China is better captured in the SSMinter simulations than in the SSMclim simulations. At the same time, the positive relationship between the local soil evaporation anomaly and summer precipitation anomaly in northwest China can be well (poorly) reproduced in the SSMinter (SSMclim) simulations.

Many previous studies have shown that the summer precipitation anomaly in northwest China is dominantly influenced by the 500 hPa atmospheric circulation anomaly in mid- and high-latitudes of Asia (He et al., 2005; Tang and Lin, 2007; Chen and Dai, 2009). For example, Chen and Dai (2009) indicated that the precipitation in northwest China is highly related to 500 hPa geopotential height changes in Siberia and the Iranian plateau.

Figure 6 gives the climatological mean of summer 500 hPa geopotential height. In the ERA-40 data (Figure 6(a)), there are two ridges over Lake Baikal and Caspian Sea and a shallow trough over Lake Balkhash. The two simulations well reproduce the above features (Figure 6(c) and (e)). But the 500 hPa geopotential height is overestimated in the SSMclim run. For example, the 500 hPa geopotential height around Lake Baikal simulated in the SSMclim run is 12 dagpm larger than that of ERA-40 data (Figure 6(b)). When the interannual SSM forcing is included in the model's boundary conditions, the reproduction of the 500 hPa geopotential height is improved (Figure 6(e)). The root-mean-square error (RMSE) of the 500 hPa geopotential height over (40°–70°N, 40°–135°E) is 10.78 dagpm in the SSMclim run and 5.29 dagpm in the SSMinter run, respectively.

Figure 6.

Climatological mean of summer 500 hPa geopotential height (dagpm) derived from (a) the ERA-40 data, (c) the SSMclim simulations and (e) the SSMinter simulations; (b) and (d) are differences between the SSMclim and SSMinter simulations, respectively, with the ERA-40 data; (f) represents differences between the SSMinter simulations and the SSMclim simulations.

Figure 7 shows the correlation coefficients between the ERA-40 500 hPa geopotential height field and the CMAP precipitation rate time series averaged over northwest China in JJA during 1979–2000. Positive values (>0.4) are seen in Siberia (55°–65°N, 50°–90°E), and negative values (<−0.4) in Iranian plateau (35°–42.5°N, 60°–75°E). It shows that the summer precipitation anomaly in northwest China is highly related to 500 hPa geopotential height anomaly in Siberia and Iranian plateau; it is consistent with the result from the work of Chen and Dai (2009).

Figure 7.

Correlation coefficients between the ERA-40 500 hPa geopotential height field and the CMAP precipitation rate time series averaged over northwest China in JJA during 1979–2000. Only regions with correlation exceeding a 98% confidence level are shaded.

To explore the effect of SSM on the simulation of 500 hPa general circulation anomaly over mid- and high-latitudes of Asia, Figure 8 presents the correlation coefficients between the ERA-40 data and model simulations for 500 hPa geopotential height in JJA during 1979–2000. As seen in Figure 8(a), there are generally negative values in Siberia and Iranian plateau, showing that the SSMclim simulations fail to reproduce the general circulation anomaly in Siberia and Iranian plateau. In contrast, the reproduction of general circulation anomaly in Siberia and Iranian plateau is improved in the SSMinter simulations, especially in Siberia (Figure 8(b)). For example, the correlation coefficient is 0.37 in the SSMinter simulations, which is statistically significant at a 90% confidence level, but it is −0.10 in the SSMclim simulations. As a result, the improved simulation of general circulation anomaly over Siberia and Iranian plateau is expected to enhance the predictability of anomalous precipitation in northwest China.

Figure 8.

Correlation coefficients between the ERA-40 data and model simulations for the East Asian 500 hPa geopotential height in JJA during 1979–2000: (a) the SSMclim simulations; (b) the SSMinter simulations. Only regions with correlation exceeding a 90% confidence level are shaded.

5 Conclusions

To investigate the influence of SSM on interannual variability of East Asian summer precipitation, two ensemble simulations are conducted using the NCAR CAM3 with specified observational SST during 1979–2000. The two simulations differ in ways SSM is climatological (SSMclim) or interannually varying (SSMinter) in the AGCM. Modelling results show that interannual variability of JJA precipitation in the SSMinter simulations is better depicted than that in the SSMclim simulations over the mid- and high-latitudes of East Asia, especially in northwest China. The standard deviation of summer precipitation anomaly in northwest China is 0.19 mm day−1 in the CMAP data, 0.04 mm day−1 in the SSMclim simulations, and 0.11 mm day−1 in the SSMinter simulations (Table 1). The correlation coefficient between the SSMinter (SSMclim) simulations and the CMAP data for northwest China precipitation is 0.56 (0.12) in JJA during 1979–2000. Results also show that the anomalous local soil evaporation is positively related with the anomalous summer precipitation in northwest China, which is well (poorly) reproduced in the SSMinter (SSMclim) run.

The mechanisms responsible for the improvement are summarized as follows: The interannual variability of JJA precipitation in northwest China is significantly influenced by the 500 hPa geopotential height anomaly in Siberia and Iranian plateau. The SSMclim simulations fail to reproduce the general circulation anomaly in Siberia and Iranian plateau. The reproduction of 500 hPa geopotential height anomaly in Siberia and Iranian plateau is improved in the SSMinter simulations, especially in Siberia. The correlation coefficient between ERA-40 data and the SSMinter simulations for the 500 hPa geopotential height in Siberia is 0.37 (at a 90% confidence level), but it is −0.10 between ERA-40 data and the SSMclim simulations. As a result, the interannual variability of JJA precipitation in northwest China is better reproduced in the SSMinter simulations.

Limitations of this study should also be acknowledged. The deficiencies are clearly evident in the interannual six-hourly 0–10 cm layer soil moisture from NCEP-NCAR reanalysis products (Mintz and Serafini, 1981; Mintz and Walker, 1993; Kalnay et al., 1996). Also, the soil moisture state simulated by a land surface model is a highly model-dependent quantity, meaning that the direct transfer of one model's soil moisture into another can lead to a fundamental and potentially detrimental inconsistency (Koster et al., 2009). The SSM from NECP-NCAR is inserted directly into the CLM3 in this study, so the quantitative estimate for the influence of SSM on the precipitation in northwest China is necessary to be further verified based on other data in the future.

Acknowledgements

We would like to thank Drs Paul Dirmeyer, Jiangfeng Wei, and Jieshun Zhu for their comments. Zhongxian Li has been supported by the NSF of China under grant 40905045, Qing Lan Project of Jiangsu Province, and the NSF of Jiangsu Higher Education Institutions (13KJB170013). Haishan Chen acknowledges supports from the NSF of China (41230422), the NSF of Jiangsu Province (BK20130047), the Special Fund for Public Welfare Industry (GYHY201206017), and the NCET Program of Ministry of Education. This research has been jointly supported by the Open Project of Key Laboratory of Meteorological Disaster of Ministry of Education (KLME1104) and the Project Funded by the Priority Academic Program Development (PAPD) of Jiangsu Higher Education Institutions.

Ancillary