Land-atmosphere coupling amplifies hot extremes over China

Climate extremes, such as extreme hot temperatures and heat waves, can have dramatic societal, economic, and ecological consequences. China has experienced remarkable interannual and decadal changes in hot extremes during the last several decades. However, the underlying mechanisms responsible for changes in the hot extremes over China have not been clearly identified. In this study, we investigate the role of land-atmosphere coupling for hot days and heat waves during summer over China using two long-term Weather Research and Forecasting model simulations with and without interactive soil moisture. Results indicate that land-atmosphere coupling mainly amplifies hot extremes over China. In particular, significant amplifying effects appear over most of eastern and southwestern China. Over these areas, land-atmosphere coupling generally accounts for 30%–70% of the numbers of hot days and heat waves. This study highlights the critical importance of land-atmosphere interactions for the occurrence of hot extremes over China.

In recent years, hot extremes have received increasing concerns due to their tremendous effects on human and natural systems [1,2]. A well-documented example is the European heat wave of 2003, which resulted in around 35000 heat-related deaths and major financial losses [3]. In the same summer, South China also experienced an unprecedented heat wave [4]. In the summers of 2006 and 2007, hot extremes occurred over many areas of China [5,6]. These hot extremes severely affected human health and life, energy supply and demand, water resources, and agricultural production.
Over China, hot days and heat waves have experienced remarkable interannual and decadal variations during the last several decades [7][8][9][10][11][12][13][14]. It is very likely that hot extremes will be of longer duration, more intense, and more frequent over East Asia and the globe in the future [15]. Previous studies have emphasized the importance of large-scale circulation anomalies and sea surface temperature (SST) anomalies for hot extremes over China [16,17].
It is also noted that global warming and the effects of urbanization may contribute to enhance the occurrence of hot extremes [18,19]. Soil moisture, as a critical component of the land surface, has been recognized to play an important role in influencing surface climate [20][21][22]. In particular, recent studies highlighted the critical role of soil moisture for daily mean and maximum temperatures over China [23,24]. However, how land-atmosphere coupling affects hot extremes over China is not yet well understood.
Regional climate models have been widely used as a valuable tool to understand regional climate characteristics and change, and elucidate the mechanisms involved [25][26][27][28][29][30]. Zhang et al. [24,31] demonstrated that the Weather Research and Forecasting (WRF) model can simulate well local and regional land-atmosphere coupling compared with statistical analysis to observations and observationally based analysis data [23,32,33]. The objective of this study is to evaluate the role of land-atmosphere coupling for hot extremes over China by analyzing two long-term WRF model simulations with and without interactive soil moisture.

Data and method
We use two long-term experiments with the WRF ARW version 3.1.1 [34] described by Zhang et al. [24] to isolate the role of land-atmosphere coupling for the occurrence of summer hot extremes over China. The model is a fully compressible, Euler nonhydrostatic model. The physical schemes used include the WRF single-moment six-class (WSM6) microphysics scheme [35], the Kain-Fritsch convective parameterizations [36], the Yonsei University planetary boundary layer scheme [37], the NCAR Community Atmosphere Model (CAM 3.0) spectral-band shortwave and longwave radiation schemes [38], and the unified Noah land model [39]. We use a horizontal grid resolution of 60 km, and 28 vertical layers. The model domain centered at 36N and 116E covers the whole of China and surrounding areas.
A 21-year-long control simulation (CTL) with the fully coupled Noah land model allows the soil moisture to interact with the atmosphere. An additional experiment (SoilM) uses the same model configuration but disables interactive soil moisture by using prescribed climatology derived from CTL at each time step. CTL covers the period of 1 January 1979 to 31 December 1999, driven with initial and boundary conditions and SST from the National Centers for Environmental Prediction(NCEP)-Department of Energy (DOE) reanalysis [40]. Boundary conditions and SST are updated every 6 h. For SoilM, we restart CTL on 1 June and integrate to 31 August for each year of 1980-1999. The first 17 months are discarded, and we analyze the simulations for the following summers. SoilM effectively removes the effects of interactive soil moisture, and its difference with CTL can be used to quantify the contribution of land-atmosphere coupling to hot extremes over China.
In this study, we use two indices to measure hot extremes: (1) number of hot days (NHD), and (2) number of heat waves (NHW). NHD is defined as the number of the days for each grid cell in which the daily maximum temperature meets or exceeds the long-term mean 90th percentile of daily maximum temperatures. NHW is defined as the frequency of the occurrence of two or more consecutive hot days. For each of the analyzed 20 years (1980)(1981)(1982)(1983)(1984)(1985)(1986)(1987)(1988)(1989)(1990)(1991)(1992)(1993)(1994)(1995)(1996)(1997)(1998)(1999), the 90th percentile is based on the 92-day period of June-August. The long-term mean 90th percentile is calculated as the mean of the 90th percentile daily maximum temperatures over the 20 years. The long-term 90th percentiles for the simulations and the observations are based on CTL and the CN05 dataset [41], respectively.
The model set-up has been evaluated with respect to its mean climate and interannual summer climate variability by Zhang et al. [24]. To evaluate simulated hot extremes of CTL, we use the 0.5 gridded daily maximum temperature dataset (CN05) developed by Xu et al. [41]. The dataset is based on the interpolation of data from 751 observing stations in China, and it is primarily developed for the validation of climate models.

Results
To evaluate the capability of the WRF model to reproduce the observed NHD and NHW over China during 1980-1999, we compare the CTL simulation with the observations (Figure 1). The observed NHD and NHW show a similar pattern. The WRF model generally simulates NHD and NHW relatively well over eastern and southwestern China, both in the magnitude and spatial pattern. The model biases mainly occur over the middle and lower reaches of the Yangtze River Valley and northern part of northeastern China where both NHD and NHW are overestimated. Over northwestern China, the model reproduces well the spatial distribution of hot extremes, but underestimates the magnitudes.
The difference between CTL and SoilM represents the change induced by land-atmosphere coupling since soil moisture interactions are disabled in SoilM. Figure 2 shows that changes in NHD and NHW exhibit a very similar spatial distribution. Positive changes appear over most of China, indicating that land-atmosphere coupling mainly amplifies hot extremes. In the meanwhile, the amplifying effects depend on climate regimes. Significant increases in NHD and NHW appear over most of humid and semi-humid eastern and southwestern China, with the magnitudes of 3-7 days per year and 0.75-1.5 times per year, respectively. In contrast, over dry northwestern China, soil moisture effects are generally insignificant and much smaller though positive signs still dominate. These results are not unexpected. Soil moisture modifies surface air temperature mainly through its effects on evapotranspiration and the associated change in sensible heat flux. Zhang et al. [24] demonstrated that the coupling of soil moisture with daily mean and maximum temperatures is largely determined by the sensitivity of surface fluxes to soil moisture. Hot extremes are closely associated with daily maximum temperature. Similarly, soil moisture effects on hot extremes also depend on the ability of soil moisture to affect surface fluxes. Over regions where land-atmosphere coupling significantly amplifies hot extremes, soil moisture generally exhibits a strong ability to affect surface fluxes, in particular sensible heat (see Figure  8 in [24]). In contrast, over dry northwestern China, soil moisture anomalies are too small to play an important role in influencing surface fluxes, and thus hot extremes.
We further compute the ratio of the difference in hot extremes between CTL and SoilM to the value in CTL, which reflects the relative contribution of land-atmosphere coupling to hot extremes ( Figure 3). Again, a very similar spatial distribution is seen for NHD and NHW. Over most of eastern and southwestern China, where significant changes occur, land-atmosphere coupling generally accounts for 30%-70% of NHD and NHW. In contrast, it makes a relatively limited contribution in dry northwestern China, explaining 30% or less of hot extremes. In addition to land-atmosphere coupling, hot extremes over China can also be affected by other factors such as SST anomalies [16][17][18][19]42]. In particular over dry northwestern China, other factors may play a more important role in the occurrence of hot extremes than land-atmosphere coupling does.

Conclusions
The occurrence of hot extremes over China has been previously attributed to large-scale atmospheric circulation anomalies, SST anomalies, global warming and urbanization effects [16][17][18][19]. However, the role of land-atmosphere interactions for the occurrence of hot extremes over China is not yet well understood. In this study, we take the first attempt to investigate the contribution of land-atmosphere coupling to hot extremes over China. We make two long-term WRF simulations driven with the NCEP-DOE reanalysis data: a control run covering the period 1979-1999 with fully interactive soil moisture, and an additional experiment repeating summer integrations for 1980-1999 but replacing soil moisture evolution at each time step with the climatology of the control run. The difference between the two experiments allows us to quantify the contribution of land-atmosphere coupling to the occurrence of hot extremes over China.
Over most of eastern and southwestern China, landatmosphere coupling significantly enhances hot extremes, accounting for 30%-70% of NHD and NHW. Over dry northwestern China, soil moisture is too small to play an important role in influencing surface fluxes, and thus makes a relatively limited contribution to hot extremes. The findings suggest that land-atmosphere coupling is a key player for hot extremes over most areas outside of dry northwestern China.
Generally speaking, the WRF model reproduces relatively well hot extremes over China. In the meanwhile, the model has the biases that should be recognized. For example, the model underestimates NHD and NHW over dry northwestern China, and overestimates them over the middle and lower reaches of the Yangtze River Valley and northern part of northeastern China. In the future, more studies are clearly needed to evaluate how these biases may affect simulated soil moisture feedbacks on hot extremes.