HTML
-
Received 2 January 2015; revised 22 May 2015; accepted 4 June 2015
Hot extremes, which can cause severe societal, economic and ecological losses, have been shown to have had an increasing trend in the past several decades (Easterling et al., 2000; Meehl and Tebaldi, 2004; Alexander et al., 2006). A World Meteorological Organization (WMO) report showed that heat waves led to a 2300% increase in the loss of life in 2001-10 compared to 1991-2000 (WMO, 2013). In particular, the mega heatwaves that hit Europe in 2003 and the Russian Federation in 2010 caused over 66 000 and 55 000 deaths, respectively. Hot extremes over China have also become more frequent and severe in recent decades (Ren and Zhai, 1998; Zhai et al., 1999; Yan et al., 2002; Gong et al., 2004; Qian and Lin, 2004; Ren et al., 2010; Wang et al., 2012). For example, extraordinary heat waves struck many areas of China in 2013, leading to dramatic impacts on human life, property, agriculture, and water resources. The frequency and intensity of hot extremes have been projected to increase in the future over China and other regions of the globe (IPCC, 2012).
It is still a scientific challenge to accurately predict cli-
Figure 1. Time series of regionally averaged original (a) spring soil moisture and (c) summer hot days over North China, and linearly detrended time series of (b) spring soil moisture and (d) summer hot days.
mate extremes such as heat waves one season or more in advance, limiting our ability to avoid or reduce their adverse impacts (e.g., Shukla, 1998; Zhang and Wu, 2014). Seasonal climate prediction must take advantage of the memories of slowing evolving earth system components. In this regard, the ocean has been considered as a critical predictor of seasonal climate anomalies (e.g., Cane et al., 1986). However, the ocean forcing is relatively weak in the midlatitude land areas in summer (e.g., Koster and Suarez, 1995; Dirmeyer et al., 2003; Douville, 2004). Similar to sea surface temperature, soil moisture also has a long memory, and thus can offer the potential for improving seasonal climate prediction, particularly regarding the summer season (e.g., Koster et al., 2004a; Seneviratne et al., 2006a; Zhang et al., 2008; Dirmeyer et al., 2009; Seneviratne et al., 2010; Wu and Zhang, 2013a). The importance of soil moisture to hot extremes has been highlighted in both observational and modeling studies (Durre et al., 2000; Diffenbaugh et al., 2005; Fischer et al., 2007; Teuling et al., 2010; Hirschi et al., 2011; Quesada et al., 2012; Chen and Zhou, 2013; Liu et al., 2013). For example, Zhang and Wu (2011) demonstrated that soil moisture feedbacks significantly enhance the occurrence of hot extremes over the areas outside of the arid and semi-arid regions in China. However, it is still unclear if antecedent soil moisture condition provide useful predictors for summer hot extremes.
In this study, we focus on North China (37°-42°N, 108°-118°E), and investigate the relationship between spring (March-April-May) soil moisture and summer (June-July-August) hot extremes for the period 1981-2008. This region is a "hot spot" of soil moisture-temperature coupling, as established by the Global Land-Atmosphere Coupling Experiment (Koster et al., 2006) and previous statistical and regional climate modeling studies (Zhang and Dong, 2010; Zhang et al., 2011). Since soil moisture measurements in China have many missing data, we use soil moisture from the Global Land Data Assimilation System (GLDAS) (Rodell et al., 2004).
The paper is organized as follows. In section 2, we briefly describe the data and method used in this study. Section 3 examines the relationship of spring soil moisture to summer hot days and heat waves over North China. In section 4, the physical mechanism involved is discussed. A summary of the main findings is given in section 5.
-
In this study, we use monthly soil moisture data from GLDAS version 1, at a resolution of 1°, for the period 1981-2008 (Rodell et al., 2004). The GLDAS soil moisture data have been evaluated extensively against observations (Berg et al., 2005; Kato et al., 2007; Syed et al., 2008; Zaitchik et al., 2010), and have been widely used to study the hydrological cycle, weather and subseasonal forecasting, and land-atmosphere interactions (e.g., Koster et al., 2004b; de Goncalves et al., 2006; Zhang et al., 2008; Wu and Zhang, 2013b). (Berg et al., 2005) demonstrated that GLDAS soil moisture data agree well with observations over eastern China. Since surface soil moisture is immediately affected by precipitation, we use subsurface soil moisture from three land surface models including the Common Land Model (CLM; Dai et al., 2003), Mosaic (Chen et al., 1996) and Noah (Ek et al., 2003) with subsurface layer thicknesses of 9-138, 2-150 and 10-100 cm, respectively. We also use latent heat and sensible heat flux data from GLDAS to explore the physical processes that explain our findings. To test the robustness of our results, we use limited observed soil moisture data from the China Meteorological Administration (
http://cdc.nmic.cn ). We use observed soil moisture at a depth of 50 cm from two stations——Nangong (37.37°N, 115.38°E) and Fenyang (37.25°N, 111.76°E)——that have relatively complete data available for the period 1996-2008.For this analysis, we use the number of hot days (NHD) and number of heat waves (NHW) to measure hot extremes. The two indices are defined based on daily maximum temperature at a resolution of 0.5° (CN05), developed by (Xu et al., 2009). The CN05 dataset has been constructed by interpolating the data from 751 observation stations over mainland China. NHD is defined as the number of days with daily maximum temperature meeting or exceeding the mean 90th percentile for all summer days over the period 1981-2008. NHW is defined as the frequency of the occurrence of at least two consecutive hot days. In other words, one NHW event denotes two consecutive hot days or more than two consecutive hot days.
In this study, all data are linearly detrended before statistical analyses are performed. As a sample, Fig. 1 shows the original time series of spring soil moisture for CLM and summer hot days regionally averaged over North China, and the linearly detrended time series.
Figure 2. Correlations of summer hot days averaged over North China (enclosed by the box) to spring soil moisture in (a) CLM, (b) Mosaic and (c) Noah. All data are linearly detrended before the correlation coefficient is calculated. Correlations of 0.32, 0.37 and 0.48 are significant at the 90%, 95% and 99% confidence levels, respectively.
Figure 3. Correlations of summer heat waves averaged over North China (enclosed by the box) to spring soil moisture in (a) CLM, (b) Mosaic and (c) Noah. All data are linearly detrended before the correlation coefficient is calculated. Correlations of 0.32, 0.37 and 0.48 are significant at the 90%, 95% and 99% confidence levels, respectively.
Figure 4. Autocorrelations of spring soil moisture to summer soil moisture in (a) CLM, (b) Mosaic and (c) Noah. All data are linearly detrended before the correlation coefficient is calculated. Correlations of 0.32, 0.37 and 0.48 are significant at the 90%, 95% and 99% confidence levels, respectively.
-
Figure 2 shows the correlation patterns of spring soil moisture with summer hot days averaged over North China (37°-42°N, 108°-118°E) for the period 1981-2008. The correlations from CLM, Mosaic and Noah exhibit a similar pattern, albeit with some differences. Over most areas of North China, the correlations between spring soil moisture and summer hot days are negative and significant at the 95% confidence level.
The correlation patterns of spring soil moisture with summer heat waves averaged over North China are presented in Fig. 3. The correlation patterns with heat waves are similar to those with summer hot days. For all three models, spring soil moisture anomalies mainly have negative correlations with the following-summer heat waves, with high correlations mainly appearing over North China.
Table 1 lists the correlation coefficients of regionally averaged spring soil moisture with regionally averaged hot days and heat waves over North China for the period 1981-2008. All correlation coefficients are negative and significant at the 95% confidence level. Among the three models, the spring soil moisture condition in CLM have the strongest correlations with summer hot days and heat waves.
We further use the observed soil moisture to test the robustness of the results from the GLDAS soil moisture. Since the observed soil moisture measurements over North China generally have many missing data, we use averaged soil moisture at a depth of 50 cm from two stations that have relatively complete data available. The correlations of observed spring soil moisture to summer heat waves and hot days regionally averaged over North China are -0.49 and -0.40 during the period 1996-2008, significant at the 90% and 80% confidence levels, respectively.
The above results show that the local spring soil moisture condition is closely linked to, and can provide a useful predictor for, summer hot extremes over North China. In the following section, we further examine if a clear physical linkage between spring soil moisture and hot extremes over North China exists in our statistical analysis.
-
Figure 4 shows the autocorrelations of spring soil moisture to summer soil moisture over North China for the period 1981-2008. The autocorrelation coefficients of spring soil moisture to summer soil moisture are generally significant at the 90% confidence level over North China for the three models. In particular, they are significant at the 99% confidence level in most grids in CLM and Noah. We further estimate soil moisture memory using the decorrelation time,
\(t_\rm d=\dfrac1+\alpha1-\alpha ,\) \endcenter
where α is the autocorrelation between spring and summer soil moisture (von Storch and Zwiers, 1999). Soil moisture generally has memory of one season or more in most grids over North China for the three models (Fig. 5). These results indicate that spring soil moisture anomalies can persist to the summer season over North China. Compared to previous studies, which have generally demonstrated that soil moisture has a memory of several months (e.g., Wu and Dickinson, 2004; Seneviratne et al., 2006b), the decorrelation time method used in this study may overestimate soil moisture memory length over some areas. Soil moisture affects, and is affected by, precipitation. We further calculate the correlations of regionally averaged spring precipitation to regionally averaged spring and summer subsurface GLDAS soil moisture averaged over North China for the period 1981-2008. The precipitation data are obtained from Climatic Research Unit (CRU) at a resolution of 0.5° [
http://www.cru.uea.ac.uk/cru/data/hrg/ , CRU TS v. 3.22 (Harris et al., 2014)]. The correlation coefficients are all positive with values ranging from 0.33 to 0.57. Plus, they are all significant at the 90% confidence level.Figure 5. Decorrelation time (in seasons) of soil moisture in (a) CLM, (b) Mosaic and (c) Noah. All data are linearly detrended before the decorrelation time is calculated.
Figure 6. Correlations of summer soil moisture to summer latent (left) and sensible heat (right) fluxes in (a, b) CLM, (c, d) Mosaic and (e, f) Noah. All data are linearly detrended before the correlation coefficient is calculated. Correlations of 0.32, 0.37 and 0.48 are significant at the 90%, 95% and 99% confidence levels, respectively.
Figure 7. Correlations of summer sensible heat flux to summer hot days (left) and heat waves (right) in (a, b) CLM, (c, d) Mosaic and (e, f) Noah. All data are linearly detrended before the correlation coefficient is calculated. Correlations of 0.32, 0.37 and 0.48 are significant at the 90%, 95% and 99% confidence levels, respectively.
Previous global and regional climate modeling studies have demonstrated that summer soil moisture can strongly affect summer mean temperature and warm temperature extremes mainly through its effects on the partitioning of available energy into latent and sensible heat fluxes (Koster et al., 2006; Zhang and Wu, 2011). Our previous regional climate model simulation demonstrated that summer soil moisture affects hot extremes over North China and also many other areas of China, mainly through its modifications to latent heat and sensible heat fluxes (Zhang et al., 2011; Zhang and Wu, 2011). Here, we statistically examine the physical linkage between summer soil moisture and hot extremes over North China.
The first part of the linkage is from soil moisture anomalies to latent heat and sensible heat anomalies. Figure 6 shows that the correlations of summer soil moisture to summer latent and sensible heat fluxes are generally significant at the 95% confidence level in all three models, indicating that summer soil moisture anomalies are closely related with summer latent and sensible heat fluxes.
We further look for the second part of the linkage from summer sensible heat flux to hot extremes. Figure 7 shows the correlations of summer sensible heat flux to summer hot days and heat waves over North China for the period 1981-2008. In all three models, summer sensible heat flux have strong and positive correlations with both hot days and heat waves, indicating that sensible heat anomalies induced by soil moisture anomalies can subsequently affect hot extremes over North China. It should be noted that soil moisture anomalies may also affect summer hot extremes over North China through its effects on atmospheric circulations (e.g., Taylor et al., 2011; Koster et al., 2014; Zhang et al., 2015).
-
Temperature extremes have much more important effects on natural and human systems than mean temperature. Over China and other regions of the globe, hot extremes have become more frequent and severe in recent decades, and have been projected to continually increase throughout the remainder of the present century. However, our current ability to predict hot extremes is largely limited by poor understanding of slowly evolving earth system components such as soil moisture, and their interactions with climate. In this study, we investigate the relationship of the spring soil moisture condition to summer hot extremes over North China, which is a "hot spot" of soil moisture-temperature coupling (e.g., Koster et al., 2006; Zhang et al., 2011).
The results show that the local spring soil moisture condition is closely related with hot extremes over North China. The correlation coefficients of regionally averaged spring soil moisture condition with regionally averaged hot days and heat waves range from -0.44 to -0.58, accounting for 19%-34% of the total variances. Spring soil moisture anomalies can persist to the summer season. Furthermore, it is found that summer soil moisture anomalies have a strong ability to affect summer latent and sensible heat flux, thus exerting substantial effects on summer hot extremes. Our findings indicate that the spring soil moisture condition is closely associated with summer hot extremes over North China, with a clear physical linkage. Since the spring soil moisture condition can be monitored easily, we recommend that spring soil moisture anomalies should be implemented into practical predictions of hot extremes over North China.