Trend and variability of China's summer precipitation during 1955–2008
State Key Laboratory of Grassland and Agro-Ecosystems, Institute of Arid Agroecology, School of Life Sciences, Lanzhou University, PR China
Correspondence to: J-S Ye, State Key Laboratory of Grassland and Agro-Ecosystems, Institute of Arid Agroecology, School of Life Science, Lanzhou University, Lanzhou 730000, PR China. E-mail: firstname.lastname@example.org
Precipitation varies greatly in different seasons and regions across China; it ranges from less than 100 mm per year in the arid northwest to more than 1000 mm per year in the southeast, and the most events and amounts occur in the summer monsoon season (Liu et al., 2005). A previous study has shown that annual precipitation has increased at more than half of the weather stations in China during the period 1971–2000; the increasing trends mainly occurred in summer and winter, and the decreasing trends occurred in spring and autumn (Wu et al., 2006). While summer precipitation averaged over China did not exhibit a statistically significant trend, it has significantly increased in East China especially in the middle and lower reaches of Yangtze River basin (Liu et al., 2005; Xu et al., 2006; Lei et al., 2011; Ye et al., 2013a). Summer precipitation general has decreased at most weather stations in North China Plain, although most of these trends were found not to be statistically significant (Liu et al., 2005; Xu et al., 2006; Ye et al., 2013a).
The average position of the summer monsoon rain belt over China lies northward of 35°N, however, a significant southward trend of the rain belt has been observed during recent decades (Xu et al., 2006; Lei et al., 2011; Ye et al., 2013a). Xu (2001) has attributed such a shift to the acceleration of industrialization in East China, which has caused the formation of a large amount of sulphate aerosols reducing the sunlight reaching the surface, which, in turn, alters the development of the mid-summer monsoon rain belt. On the basis of simulations of a global climate model, Menon et al. (2002) suggested that precipitation trends in China over the past several decades, along with increased rainfall in the south and drought in the north, might be related to increased black carbon aerosols. Others argued that both industrial aerosol emission (Streets et al., 2008) and combustion of biofuel in rural households and open crop residue (Xue et al., 2012; Tang et al., 2013) resulted in increased aerosol optical depth (AOD) over South-central China (Ye et al., 2013a). Xu (2001) and Ye et al. (2010, 2013a) suggested that rising concentration of aerosols over South-central China caused the land surface to cool in the region which, along with the steady warming over nearby ocean surface, resulted in decreased temperature and pressure contrasts between the land and the sea, which in turn, reduced the strength of the summer monsoon circulation. Wang and Zhou (2005) related the observed precipitation trends in China to trends towards a stronger continental high over Eurasia and a weaker subtropical high over western Pacific.
In addition to changes in precipitation regimes, increases in extreme precipitation events are also evident in recent decades. Examples are increased drought in North China Plain since 1980 and the great flood in the mid-lower reaches of Yangtze River basin in 1998. On the basis of daily precipitation observations from meteorological stations over China (1961–2001), Wang and Zhou (2005) found increasing trends in extreme events of summer precipitation in East China and Northwest China in 1961–2001. The increase in flood and drought extremes should be closely related to the changes of the interannual variability of summer precipitation in these regions. The year-to-year variation of the East Asia Summer Monsoon (EASM) has been generally recognized to be associated with the El Niño-Southern Oscillation (ENSO) (Wang et al., 2008; Wu et al., 2009; Wu et al., 2012). There has been some evidence that spring North Atlantic Oscillation could also exert some effects on EASM (Wu et al., 2009; Wu et al., 2012).
On the basis of long records of precipitation (1955–2008) over China from a large number of stations (323), this study aims to characterize the trend and variability of China's summer precipitation with an emphasis on year-to-year variability. We examine the spatial patterns of trends in summer precipitation across China, and their association with China's summer air temperature and the Pacific sea surface temperature (SST) using singular value decomposition (SVD). We will also show evidence of intensified summer precipitation variability over China and its close links to the increased Pacific SST variation in the recent three decades (by frequency density distribution analysis).
2. Data and methods
The data used in this study consist of China's summer precipitation and air temperature (measured at 1.5 m height), and Pacific SST. A large portion of meteorological stations were established after the founding of People's Republic of China (October 1949). Considering both the numbers of stations and duration of measurement at each station, we focus on a 54-year time series spanning from 1955 to 2008. A total of 323 stations are included in the study (Figure 1). Data of meteorological stations were obtained from the China Meteorological Administration Data Sharing Service System (Ye et al., 2010). The SST data used are the NOAA Extended Reconstructed data (ERSST v3) (Smith et al., 2008). The latitudinal ranges of Pacific SST used in the study are 110°E–90°W and 40°S–70°N. Summer seasonal mean precipitation, air temperatures and SST are taken to be an average of June, July, and August (JJA) monthly mean values.
To assist the analysis and discussion, we divide the whole nation into three climatic sub-regions, i.e. West, South, and North China (Figure 1). The reasons for such a dividing are given in the following section (Section 'Trends in China's summer precipitation'). Trends at individual station and in each sub-region are evaluated with the nonparametric Mann–Kendall test (Mann, 1945), which reliably identifies monotonic linear and nonlinear trends in non-normal datasets containing outliers and has been commonly used to assess the trend in meteorological parameters. The relationship between China's summer precipitation and the Pacific SST is examined by using the SVD of the covariance matrix between the two data fields (Bretherton et al., 1992). This statistical technique is very powerful in objectively detecting the coupled patterns between two data fields such as precipitation and SST (Ting and Wang, 1997; Shabbar and Skinner, 2004; Wang et al., 2010).
3.1. Trends in China's summer precipitation
The majority of stations in the South show positive trends of summer precipitation while stations in the North generally show negative trends (Figure 1, top). Unlike the East, there is no obvious pattern for stations in the West, although trends at most stations are positive in the northwest corner. On the basis of the eight climatic zones of China defined by Liu et al. (2005), the whole China is divided into three climatic sub-regions (i.e., South, North, and West) since summer precipitation measurements within the South and North show homogeneous trends (Figure 1). All stations west of 110°E are grouped as one sub-region because stations here are relatively sparser compared to the east of 110°E.
Mean air temperatures show a different spatial pattern compared to that of precipitation. Negative trends are observed at most stations in the middle and lower reaches of Yangtze River basin (i.e., South-central China) and positive trends over the rest of China (Figure 1, bottom).
The sub-regional trends are also presented for the three sub-regions (Figure 2). The trends are significantly positive with 4.6 mm per decade in the South (p < 0.05) and significantly negative with −3.9 mm per decade in the North (p < 0.05) during the whole time series of 1955–2008. There is no overall trend (the Mann–Kendall trend is nearly zero) in the West during the entire period. The sub-regional values are weighted by their respective areas to derive a national average. As the decrease in the North and the increase in the South are nearly equal, the national trend is not significant with −0.6 mm per decade.
3.2. Intensification of summer precipitation variability
To examine summer precipitation variability in China, a precipitation index is constructed by normalizing precipitation anomalies by standard deviation for each sub-region (Figure 3, left side). We define wet and dry extremes as positive or negative summer precipitation anomalies exceeding one standard deviation, i.e. normalized anomalies exceeding one unit. The whole time series was divided into two bins, i.e. before and after 1977 since the global climate shift has happened in 1976/1977 (Miller et al., 1994; Giese et al., 2002). Compared with the first bin (1955–1976), wet extremes in the second bin (1977–2008) have increased in the West, South and whole nation, while decreased in the North (Table 1). Dry extremes have increased in all the three sub-regions. In general, the precipitation indices display higher interannual variability with more wet and dry extremes all over China in the second bin. Increases in extremes are most obvious in the South and West. Increases in dry extremes are more evident compared with those of wet extremes in the second bin, even in the South with increasing trend of precipitation.
Table 1. Changes in summer precipitation variability during the two bins: first bin (1955–1976) and second bin (1977–2008)a
Wet (dry) extremes are positive (negative) summer precipitation anomalies exceeding one standard deviation; percentages of wet (dry) extremes in their corresponding bins are shown in brackets.
The variances are calculated for the whole time series and each bin. The summer precipitation in the second bin contributes to 68% the total precipitation variance in China overall, in contrast to the first bin contributing to 32%. These changes are also evident in the South (the second bin 75% in contrast to the first bin 25%) and West (the second bin 55% in contrast to the first bin 45%). The two bins contribute to general equal variances in the North. The corresponding frequency density distributions of precipitation indices generally conform to normal distributions (Figure 3, right side). The standard deviations of the normal distributions have increased in the second bin in the West, South, and overall China compared with those in the first bin, i.e. distributions in the second bin are more dispersed. The frequency distribution in the first bin is wider in the North than in other sub-regions of China, and the difference of frequency distributions between the two bins is not as obvious as in other sub-regions. This analysis supports the hypothesis that summer precipitation variability over China has intensified in the last three decades.
3.3. Relation between SST and precipitation
We conducted SVD analysis to explore possible links between China's precipitation and Pacific SST. The covarying regions between Pacific SST and precipitation are shown in homogenous correlation maps (Figure 4). The first two leading modes explain 35% and 23% of the squared covariance, respectively. The first mode clearly shows a tropical Pacific SST that is characterized by the ENSO-like SST pattern (Figure 4, top left). Tropical and eastern Pacific areas display positive SST anomalies; central and western Pacific areas generally show negative anomalies.
Furthermore, the positive correlation coefficients are more significant than negative ones. Previous studies on the covariability between summertime US precipitation and Pacific SST (Ting and Wang, 1997; Wang et al., 2010) also reported an interannual ENSO pattern as the first leading mode, with a similar structure as the first mode in this study. Shabbar and Skinner (2004) found a similar pattern in the Pacific basin as the second mode while examining variations in the Canadian summer Palmer Drought Severity Index and winter global SST. The associated precipitation pattern displays negative anomalies in the North and slightly positive anomalies in the South and West (Figure 4, top right).
The second mode features generally positive SST anomalies over the western Pacific and negative anomalies over the tropical and eastern north Pacific (Figure 4, bottom left). In the study on the links between summertime US precipitation and Pacific SSTs, Ting and Wang (1997) also found a generally similar pattern as their second mode while the pattern was confined to the north Pacific. Shabbar and Skinner (2004) reported extratropical SST fluctuations in the central north Pacific and weaker centre in the eastern tropical Pacific as the third leading mode in their Canadian study. Furthermore, Shabbar and Skinner (2004) suggested that the coupled ocean–atmosphere mode known as the Pacific decadal oscillation (PDO) is prominent in this mode. The corresponding summer precipitation in China shows positive anomalies in the South and West and slightly negative anomalies in the North (Figure 4, bottom right). We propose that SVD mode 2 corresponds to the PDO, or PDO-like ENSO variability.
The time series for SVD modes 1 and 2 are shown in Figure 5 (left side). The correlation coefficients (R) between each pair of SST and precipitation time series are 0.75 (p < 0.001) for the first mode and 0.70 (p < 0.001) for the second mode. The first mode, characterized by ENSO, captures the recent major events (such as 1982–1983 and 1997–1998); the 1997–1998 event, the strongest ever recorded, is also evident from the time series. Similar to previous study of Shabbar and Skinner (2004), some warm (negative) and cold (positive) years of the ENSO phenomenon are identified in the time series corresponding to the second mode. The percentages of SST and precipitation variance explained by each mode during the two bins suggest that increased variability in the second mode in the second bin is consistent with the intensified precipitation variability in the same period (Table 2).
Table 2. Percentages of Pacific SST and China's precipitation variance explained by the first two SVD modes (refer to Figure 5) during two bins: first bin (1955–1976) and second bin (1977–2008)a
The percentage is defined as the ratio of variance in each half time series in variance of the whole time series.
Furthermore, we also show the corresponding frequency density distributions of the two bins (Figure 5, right side). The frequency density distributions generally conform to normal distributions. For the first mode, the second bins of precipitation and SST (1977–2008) have larger means than those of the first bin (1955–1976) (Figure 5, top right). The two bins of the precipitation time series have nearly equal dispersal (or standard deviation). The distribution for the second bin of the SST time series is narrower than that of the first bin (Figure 5, top right). For the second mode, precipitation and SST have the same changes: (1) distributions of China's precipitation and the Pacific SST in the second bin are more dispersed than those in the first bin; (2) the means for both SST and precipitation distributions shift to the right in the second bin (i.e. warmer and wetter) (Figure 5, bottom right). This analysis strongly suggests a direct relationship between the intensification of China's summer precipitation variability, especially in the South, with higher Pacific SST variability over the past three decades.
3.4. Relation with air temperature
Previous studies have suggested that aerosols might have affected precipitation through their effects on air temperature in China (Xu, 2001; Xu et al., 2006; Ye et al., 2013a). The association between air temperature and precipitation can be examined by the simultaneous correlation between mean air temperature at each grid point and the precipitation SVD time series (Ting and Wang, 1997). Maps of the correlation between air temperature and SVD precipitation modes 1 and 2 are shown in Figure 6. Both the correlation maps resemble the spatial pattern of air temperature trends at individual stations shown in Figure 1. A negative correlation centre is found over the South-central China, corresponding to the negative temperature trends at most stations in this region (cool zone) (Figure 1). The negative correlations between mean air temperature and the first SVD mode of precipitation are significant over the cool zone. General positive correlations are found over the rest regions of China, while significant correlations are most noted over large areas of North and West for the second SVD mode.
Observation shows a weakened East Asia Summer Monsoon (EASM) that precipitation generally decreased at most northern stations in China and increased in the middle and lower reaches of the Yangtze River in recent decades (Lei et al., 2011; Ye et al., 2013a). Ye et al. (2013a) have suggested that the weakened EASM is associated with increased aerosol loading and AOD over the South-central China (cool zone) and continuous warming over nearby ocean (SST). This is further confirmed by the significantly negative correlations between the first SVD mode of precipitation and mean air temperature over the cool zone (Figure 6, left side).
Previous analysis of precipitation time series has suggested that summer precipitation in the North China decreased in El Niño years although the correlation between precipitation and ENSO is not strong (Gong and Wang, 1999). In this study, SVD analysis suggests that decreased summer precipitation in North China is closely related to ENSO (the first mode). The observed decline may also be associated with the PDO (SVD mode 2), although the covariability is not statistically significant. The upward trend in summer precipitation in the South may relate to either PDO or PDO covariability with ENSO.
The main source of summer monsoon precipitation in China is the southerly transport of moisture from the western Pacific and South China Sea (Xu et al., 2006). The interannual variability of East Asian summer precipitation is closely related to the position, shape, and strength of the Western North Pacific Subtropical High (WNPSH). In particular, East Asia experiences floods (droughts) when the WNPSH extends westwards (retreats eastwards) (Lu, 2001). The interannual variability of WNPSH is closely related to SST anomalies over the western north Pacific (Sui et al., 2007). The percentages of SST and precipitation variance explained by each mode during the two bins indicate that increased variability in the second mode (SST fluctuations in the western and central Pacific) in the second bin is consistent with the increased precipitation variability in the same period. The first mode (ENSO), however, shows almost the same variability during the two bins. Wang et al. (2008) also found intensified summer precipitation in the southeastern United States in recent decades. Their SVD analysis concluded that the increased Atlantic SST variability, instead of ENSO, in the recent 30 years was consistent with the increased precipitation variability in the same period.
The EASM has great economic and climatic importance for a quarter of the world's population (Wu et al., 2009; Lu and Fu, 2010). The intensification of summer precipitation variability is closely related to the potential occurrence of floods or droughts which affects the populations that are highly vulnerable to these extreme events. Modelling simulation also suggested that intensified variability of precipitation would increase variability of ecosystem productivity and reduce ecosystem resilience (Ye et al., 2013b). Furthermore, out of 19 models used by the Intergovernmental Panel on Climate Chang Fourth Assessment Report (IPCC AR4), 13 models showed significant increases in interannual variability of East Asian summer rainfall under the radiative forcing of doubling CO2 scenario (Kripalani et al., 2007). Results of 12 models in the World Climate Research Programmer's Coupled Model Intercomparison Project phase 3, also projected an enhanced interannual summer rainfall variability over East Asia in the 21st century relative to the 20th century, under scenarios A1B and A2 (Lu and Fu, 2010). And an increase of approximately 30% in the interannual variability of East Asian summer rainfall is projected in the 2070s relative to the early 21st century (Fu, 2012). Kripalani et al. (2007) suggested that these changes in summer rainfall could be stabilized by controlling the CO2 emissions.
Changes in China's summer precipitation and its variability are examined for the recent five decades. The ‘north drying and south wetting’ summer precipitation pattern is both associated with the cooling over South-central China (cool zone) and fluctuations of Pacific SST. The interannual variability is greatly intensified all over China with more wet and dry extremes in the second bin (1977–2008) than in the first bin (1955–1976). The analysis suggests that the increase in summer precipitation variation is closely related to the SST variation in the Pacific.
This work is supported by the Natural Science Foundation of China #31200373. The author are grateful to Dr Wal Anderson and two anonymous reviewers for their valuable comments on the manuscript.