1. Introduction
Climate variability and change present enormous global challenges to the sustainability and security of societies around the world, and China, with its vast population and rapidly developing economy, will have a critical role to play in the coming decades. In many respects, access to sufficient water of appropriate quality will be critical for the sustainable development of China since water influences so many facets of life, from food security and human well-being to the responses of ecosystems and the need for infrastructures to protect human lives and property against floods and droughts. Yet, as the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (Solomon et al. 2007) highlighted, projected changes in rainfall for China are hugely uncertain, not just spatially but also in the temporal characteristics of rainfall (i.e., frequency and intensity).
Summer rainfall is vital for water resources throughout the country; however, the spatial distribution of summer rainfall is very uneven over China and its within-season and year-to-year variability can be substantial. The intraseasonal variability of the summer monsoon rainband, associated with the northward propagation of the East Asian summer monsoon (EASM), is most important over China (e.g., Ding 2004), and its unusual behavior often causes local flooding and drought during summer. For example, persistent heavy rainfall over the Yangtze–Huaihe River basins during the 1991 mei-yu season caused floods, and a similar event occurred over the Yangtze River valley in 1998 (Li and Zhao 1998). In the summer of 1997 northern China suffered from reduced rainfall and serious drought conditions. It is clear that the intraseasonal variability of summer rainfall over China varies from year to year. Numerous studies have focused on understanding the mechanisms for both intraseasonal and interannual variability (e.g., Lau et al. 2000, 2004). The boundary forcing from the underlying oceans and land has particularly received attention (e.g., Wang et al. 2001; Lau and Weng 2001), in addition to the internal dynamics of the atmosphere.
It is becoming increasingly apparent that summer rainfall over China also varies substantially on longer decadal time scales. Various studies have suggested that summer rainfall has shown significant interdecadal variability during the past 50 years (Chang et al. 2000; Ding and Sun 2003; Ding et al. 2008). The major features include summer rainfall shifting southward in the late 1970s, associated with the weakening EASM circulation from the relatively cold period of the 1960s–70s to the relatively warm period of the 1980s–90s (e.g., Guo et al. 2003; Yu et al. 2004; Yu and Zhou 2007; Zhao et al. 2010). In addition to changes in the total rainfall, general increases in intensity over China and reduction in the number of rain days over northern China have been evident in the past 50 years (Zhai et al. 1999, 2005).
Various mechanisms have been proposed to provide physical explanations for these changes (Zhou et al. 2009). There are two different perspectives: the focus of one view being the slow boundary forcing and the other considering internal atmospheric dynamics. The most important slow boundary forcing includes regional and basin-scale SST anomalies, snowpack changes, and global warming of surface temperatures. The impact of regional and basin-scale SST anomalies on interdecadal variability in Chinese summer rainfall is apparent in both observational and numerical studies. Chang et al. (2000) and Xu et al. (2006) have attributed the weakening of the EASM to a reduced thermal contrast due to a greater increase of sea surface air temperature over the South China Sea and western North Pacific than the land surface air temperature across terrestrial China. Li et al. (2010), using climate models, have provided evidence that the historical SST forcing, especially the tropical interdecadal variability centered over the central and eastern Pacific, induces most of the observed weakening of the EASM during 1950–2000. The contribution of excessive snow cover and cooling trends over the Tibetan Plateau to the reduction of the thermal contrast has been emphasized (e.g., Qian et al. 2003; Zhang et al. 2004; Peng et al. 2005). However, numerical simulations indicate that greenhouse gas and aerosol forcing increases the EASM circulation by enhancing the land–sea thermal contrast (e.g., Li et al. 2010), in contrast with observations. It is still an open question how Chinese summer rainfall and the EASM react to a warming planet.
The aim of this study is to provide a comprehensive description of the decadal variability and recent trends in Chinese summer rainfall over the last 51 years, using the best available surface-based observations. Compared with previous studies, this paper will place greater emphasis on using daily observations to investigate, in depth, changes in the seasonal evolution and in temporal characteristics of summer rainfall.
A brief description of the data and methods is given in section 2. The mean climate and decadal variability of summer rainfall over China are documented in section 3, focusing on the seasonal evolution and daily rainfall characteristics (such as intensity, frequency, light rainfall, and heavy rainfall). The implications of the results in terms of the interplay between natural variability and anthropogenic climate change are discussed in section 4, which also introduces the motivation for the modeling studies presented in the subsequent paper.
2. Data and methods
A number of data sources have been used: rain gauge observations provided by the China Meteorological Administration (CMA), gridded rainfall data provided by the Climatic Research Unit (CRU), and reanalysis data from 1958 to 2002 provided by the European Centre for Medium-Range Weather Forecasts. CMA and CRU data as observational rainfall records are described in detail. The 40-yr ECMWF Re-Analysis (ERA-40) offers large-scale variables, including surface temperature, 850-hPa wind, and precipitation.
The rain gauge data comprise daily observations at 756 stations, with some missing data in the early 1950s. During the period from 1958 to 2008, 404 stations have continuous records. Therefore, the homogeneity of observed rainfall values used in this study is guaranteed. Station observations collected by the National Meteorological Center of CMA have been subjected to quality control procedures (Zhai et al. 2005). Currently they constitute the best dataset available for analyzing regional rainfall variations over China.
The gridded rainfall data (cru_ts_3.00 version) provides monthly values for the period 1901–2006 with 0.5° × 0.5° resolution based on various available observations over land (Hulme et al. 1998). Records inside the domain of China have been investigated to add information about rainfall variability before the late 1950s and to assess recent trends.
The main foci of summer rainfall characteristics in this paper include the total amount, the seasonal evolution, the daily intensity and frequency, together with the light and heavy rainfall frequency. The total rainfall in June–August has been the major focus in previous studies (e.g., Zhai et al. 1999, 2005; Ding et al. 2008). Associated with the long-term variability of total rainfall, possible changes in the seasonal evolution described in terms of daily changes, the intensity and frequency defined as the average rain rate for the days with rain (0.1 mm day−1 threshold), and the number of rain days are investigated. Because the instrument precision for observed rainfall values is 0.1 mm day−1, days with trace rainfall (<0.1 mm day−1) are excluded from rain days. Light rainfall and heavy rainfall are defined according to the cumulative distribution function (CDF) of precipitation during the period 1958–2008 at the 30th and 90th percentile, respectively.
A low-pass Lanczos filtering (Duchon 1979) is used to highlight decadal and longer time-scale variations in rainfall observations. Figure 1 shows response functions of the low-pass Lanczos filtering, using 3, 5, 7, 9, 11, 13, and 15 points. With increasing points, the Gibbs phenomenon is reduced. Here, taking into account both the focus on decadal variability and the relatively short dataset, the 9-point filter is used. With it a low-frequency rainfall dataset is constructed and then analyzed to identify the dominant patterns of long-term variability using EOF analysis. Further statistical analysis tools are also used, such as combined EOF analysis, the area-averaged mean, and regression, seeking physical consistency. The method used here does not take account of the uneven spatial distribution of observational stations in China.
Response functions of low-pass Lanczos filtering using weights of 3, 5, 7, 9, 11, 13, and 15 points and having response of 0.5 at a period of 10 units (here years).
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
3. Results
a. Climatology of summer rainfall over China
The mean climate of summer rainfall has been analyzed using the 404 rain gauge data over China in the period 1958–2008, and summer rainfall characteristics are shown in Fig. 2, including the total amount (mm), percentage relative to the annual mean, intensity (mm day−1), and frequency (days). The total summer rainfall shows substantial spatial variation over China with a maximum in the southeast and a gradual decrease toward the north and west of the country. In contrast, despite the large total rainfall amount over the south, summer rainfall contributes only about 40% of the annual mean there, whereas the percentage is greater than 60% over the northeast and some of the west, indicating that summer rainfall is of vital importance over these semiarid regions. The distribution of rainfall intensity shows a generally similar structure to that of the total rainfall, suggestive of its importance in giving the pattern of total rainfall in summer. However, frequency also plays an important role in giving the large total rainfall. The high frequency tends to be affected by topography, such as around the Tibetan Plateau, and by exposure to the warm, moist air of the Asian summer monsoon, for example, along the southern coast. Inland the frequency is relatively small and about 30–40 rain days occur over the Yellow River valley and the northwest of China.
Climatology of Chinese summer rainfall based on rain gauge observations from 1958 to 2008: (a) the total rainfall (mm), (b) percentage contribution of summer rainfall to the annual mean (%), (c) the intensity in rainy days (mm day−1), and (d) the frequency (the number of rainy days).
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
The seasonal evolution of daily rainfall over eastern China (Fig. 3a) highlights the different rainfall regimes and the importance of the mei-yu front for bringing rainfall to the Yangtze and Yellow River basins. There is generally heavy rainfall south of 25°N throughout the summer season, but in June there is a clear northward propagation to 35°N. This distinctive stage describes the establishment of the mei-yu front over the Yangtze River valley. In July the heavy rainfall declines at these latitudes but increases in the north (the Yellow River valley and farther north), reaching its peak value in late July. This second transition is associated with the sudden shift of the western North Pacific subtropical high, also seen as the demise of the baiu front over Japan in mid to late summer. As in Suzuki and Hoskins (2009), the northward progression of the EASM can be described by the penetration of moist tropical air in the lower troposphere. Figure 3b shows the isolines of the monthly-mean 335-K moist potential temperature at 850 hPa and emphasizes the major jump northward beyond 40°N in July and commencement of the retreat in August. By September the EASM has retreated rapidly to south of the Yangtze River. Also shown in Fig. 3b are the three regions used in this study to describe the rainfall variations over the distinctive climatic zones of southeast China and the Yangtze and Yellow River basins.
(a) Seasonal evolution of rainfall (mm day−1) over eastern China (105°–122°E), based on daily rainfall where the rain gauge observations have been averaged over 1° latitude bands, and (b) location of the climatological monthly-mean 335-K moist potential temperature isotherm at 850 hPa, showing the northward penetration of the moist subtropical and tropical air. The three regions used to define rainfall behavior over southeast China (SE) and over the Yangtze (YA) and Yellow (YE) River basin are also shown for reference.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
Figure 4 shows the CDF for observed daily rainfall over China, southeast China, and the Yangtze and Yellow River basins (defined in Fig. 3b) during summers from 1958 to 2008. Observed precipitation at all stations throughout China, as well as over the three regions, is sampled in the calculation. Results show that the distribution of heavy rain events is proportionally greater for southeast China and the Yangtze River valley than that over the Yellow River valley. More than 1% of rain events over southeast China have extreme heavy rainfall defined as greater than 100 mm day−1. Based on the CDF over China, the 30th percentile of rain events has a rainfall intensity of 1.1 mm day−1. Heavy rainfall, in excess of 27.45 mm day−1, contributes to the top 10 percentile of rain events. Therefore, the thresholds used for light rainfall and heavy rainfall are determined by 1.1 and 27.45 mm day−1, respectively.
Cumulative distribution functions of daily rainfall (mm day−1) during 1958–2008 at each station over China and over the three regions of southeast China and the Yangtze and Yellow River basins.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
b. Trends and decadal variability
A notable feature of summer rainfall over China during the last half century has been the well-documented trend toward drier conditions in northern China and much wetter conditions in the south (e.g., Zhai et al. 2005), as shown in Fig. 5. This means that the already uneven spatial distribution of rainfall across China (Fig. 1a) has been exacerbated in recent decades, producing major concerns for future management of water resources across the country. There has been considerable debate about whether this trend is a signature of anthropogenic climate change associated with increases in both greenhouse gases and aerosol loadings over China (e.g., Menon et al. 2002; Xu et al. 2006).
Linear trend in total rainfall (mm yr−1) at each station over China in summer during the period 1958–2008.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
However, it is becoming increasingly clear that this simple trend masks considerably more complex regional changes with decadal time-scale variations (e.g., Ding et al. 2008). These decadal changes have been investigated in this study using EOF analysis of the low-pass filtered station data. In Fig. 6, the dominant pattern (EOF1) shows homogeneous rainfall anomalies over a large region south of 30°N, which includes part of the Yangtze River valley and southeast China. The second pattern (EOF2) is characterized by a tripole pattern of anomalies over east China with opposite signs occurring over the Yangtze River valley, and over the Yellow River valley and the southeast, but with similar magnitudes of rainfall anomalies in the three regions.
Leading two EOFs of low-pass filtered summer rainfall for the period 1958–2008 based on station data: (left) the patterns and (right) corresponding principal components (black lines). The two patterns describe ~37% and 17% of the variance, respectively. Also shown are the regressions of the unfiltered summer rainfall (red lines) on the dominant decadal patterns.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
The corresponding principal component (PC) time series show that the loading of these patterns varies on decadal and interdecadal time scales (Fig. 6, right panels). For the southern China rainfall pattern (EOF1), a large shift is evident between the 1980s and 1990s, implying much wetter conditions in the last decade, though this is followed by a substantial decreasing trend after the mid-1990s. To check the robustness of the decadal variations seen in the low-pass filtered data, regressions of the unsmoothed rainfall data onto EOF1 have been computed and are shown with a red line in Fig. 6. The large decadal variability in the recent two decades is still evident. The rapid transitions in the early 1990s and at the start of the twenty-first century are also seen to occur on interannual time scales. In particular, the rapid drop in 2002–03 is suggestive of another possible transition towards drier conditions over southern China.
For the tripole rainfall pattern (EOF2), the time series again show interdecadal variability with a marked transition at the end of the 1970s, consistent with a move toward drier conditions over the Yellow River region and wetter conditions over the Yangtze River region. Changes in the tripole rainfall pattern are suggestive of an in-phase relationship with the large-scale climate shift in 1976–77, according to Trenberth and Hurrell (1994). The regression of unsmoothed annual summer rainfall on EOF2 shows large interannual variability around the time of the transition and suggests that it actually happened in 1978–79. Also, interannual variability tends to increase after the late 1970s, which is reflected in the fluctuations on decadal time scales in PC2. From the twenty-first century, interannual variability is suggestive of a possible reverse to the negative phase of the tripole pattern.
Using the gridded CRU rainfall data, which covers the longer period 1901–2006, Fig. 7 shows the leading two EOFs and PCs (solid black line) of the low-pass filtered summer rainfall over China. The first two rainfall patterns have similar structures to those in Fig. 6, and the corresponding time series indicate that interdecadal variability in the 1990s and the late 1970s is comparable with that seen in the rain gauge data. Similar large decadal variability is seen to occur before the 1950s and, as with the rain gauge data, large decadal transitions are always associated with periods of high interannual variability. Furthermore, these large changes throughout the 100-yr record not only support the decadal variability found in station data but also suggest that natural variability plays an important role.
As in Fig. 6 but for the leading two EOFs of the low-pass filtered summer rainfall for the period 1901–2006 based on the CRU gridded data. The leading patterns describe ~21% and 17% of the variance, respectively.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
The dominant decadal rainfall patterns (Figs. 6 and 7) highlight rainfall variability over southern China (south of 30°N and east of 100°E), over the Yangtze River valley (28°–33°N and east of 105°E), and over the Yellow River valley (35°–43°N and east of 110°E). Using CRU data and gauge data over China, the area-averaged low-pass filtered summer rainfall anomalies over these three regions are shown in Fig. 8. The similarity is clear during the overlapped period between the two datasets. The observed decadal variability in regional rainfall is not only in the last 51 years but also in the entire 100 years. The variation during the first half century is comparable to those in the second half century. In the last decade both datasets suggest that rainfall anomalies over the three regions have continued to undergo large changes and that a new regime may have started in the early twenty-first century. Particularly noteworthy is the decline in rainfall over southern China and the reversal of the anomalies over the Yangtze and Yellow River basins. The substantial decadal variability in the last decades, and in the earlier part of the twentieth century in the CRU data, implies that the linear trend seen in Fig. 5 cannot be attributed necessarily to anthropogenic climate change. Clearly, natural variability plays an important role in determining regional rainfall distributions and is critical for water resource management and sustainability.
The low-pass filtered time series of area-averaged total summer rainfall anomalies (mm) over the Yellow River valley (35°–43°N, east of 110°E), the Yangtze River valley (28°–33°N, east of 105°E), and southern China (south of 30°N, east of 100°E). The dashed (solid) lines indicate the results from in situ measurement (the CRU dataset).
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
c. Seasonality in the decadal changes in summer rainfall
The previous section has shown that the dominant decadal variability in the total summer rainfall is characterized by contrasting behavior over the three regions: southern China and the Yangtze River and the Yellow River valleys. As discussed in the introduction, the EASM has a complex seasonal cycle in which the northward penetration of moist tropical and subtropical air is crucial for determining the timing and intensity of summer rainfall. It is to be expected, therefore, that the decadal changes in the distribution of summer rainfall across China may also display contrasting seasonality and that investigating such changes may help to understand the source of the total rainfall changes.
Figure 9 shows regressions of monthly-mean rainfall in June–August on the two dominant decadal time series (PCs) in Fig. 6. For the southern China rainfall pattern (EOF1), the main distinguishing pattern persists throughout the summer season, suggesting that there is little connection with the seasonal evolution of the EASM and that the source of the anomaly probably relates to changes in the deep tropics and in the moisture supply from the surrounding oceans. In contrast, the tripole rainfall pattern (EOF2) shows large variations throughout the summer with out-of-phase behavior between southeast China and the Yangtze River basin in early summer and between the Yangtze and Yellow River basins later in the season. It is clear that the decadal variability pattern EOF2 is associated with changes in the northward progression and extension of the mei-yu front.
Regressions of monthly rainfall from June to August on the two leading patterns of decadal variability given in Fig. 6.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
Using the daily rain gauge data for the entire period, 1958–2008, Fig. 10 shows the seasonal evolution of the area-averaged daily rainfall (smoothed by 7-day running mean) over the Yellow and Yangtze River basins and southeast China, defined in Fig. 3b. A 9-yr running mean has been applied to the data to highlight the decadal variability. It is evident from Fig. 10 that the large rainfall value shifts from the southeast to the Yangtze and Yellow River valleys from May to August, consistent with the seasonal evolution of monsoon activities (Fig. 3a). In addition, Fig. 10 shows that the seasonality of rainfall over the three regions varies on decadal time scales. Over the Yellow River region, the decadal variations in the peak rainfall during late July and early August are clearly linked to the strength of the meiyu front as depicted in the rainfall variation over the Yangtze River region. When the rainfall over the Yangtze River region is high during late June and early July, then rainfall over the Yellow River region tends to be high in late July and early August. At the same time, the Yangtze River region tends to become drier than normal, indicative of a northwards extension of the meiyu front. Decadal changes in the seasonal evolution are apparent over southeast China, with peak rainfall in late May during the 1970s and 1980s, and in June in the decades both before and after the mid-1970s to 1990. In southeast China, being part of southern China, decadal variability in the seasonal evolution not only shows the change in the late 1970s but also reflects the change in the late 1980s, determined by the decadal southern China rainfall variability described by EOF1 in Fig. 6. Consistent with Fig. 9, enhanced daily rainfall is stable throughout the summer season in the 1990s compared with that in the 1980s.
Seasonal evolution from April to September of the area-averaged daily rainfall (mm day−1) for each year from 1958 to 2008, over the regions of (top) the Yellow River, (middle) the Yangtze River, and (bottom) southeast China, as defined in Fig. 2. A 9-yr running mean has been applied to the data to highlight the decadal variability.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
d. Interaction of decadal variability with intensity and frequency changes
So far the analysis has focused on rainfall amounts but, for many of the impacts of climate variability and change, the characteristics of the rainfall distribution such as intensity and frequency of rain days can be just as important. As the planet warms, changes in both intensity and frequency have been discussed in numerous papers (e.g., Allen and Ingram 2002; Trenberth et al. 2003; Sun et al. 2007). Theoretically, these rainfall characteristics are expected to change because global warming effectively increases the moisture-holding capacity of the atmosphere in accordance with the Clausius–Clapeyron equation. Although changes in global mean precipitation are primarily controlled by the energy budget of the troposphere, increases in atmospheric moisture can lead directly to changes in rainfall intensity, especially over ascending regions. For the period 1951–2000, Zhai et al. (2005) have already documented increases in rainfall intensity and decreases in the frequency of rain days associated with precipitation trends over China. The question discussed here is how much of the decadal variability in summer precipitation (Fig. 6) is attributable to changes in the intensity versus those in frequency.
Figures 11 and 12 show the leading two EOFs and the corresponding time series for, respectively, the low-pass filtered summer rainfall intensity and frequency over China. Regressions of unsmoothed rainfall intensity and frequency on the two dominant patterns are also shown. The leading patterns of rainfall intensity and frequency show trendlike features in contrast to the second patterns that exhibit predominant decadal variability.
Leading two EOFs of low-pass filtered summer rainfall intensity (mm day−1) for the period 1958–2008 based on data from 404 stations: (left) the patterns and (right) the corresponding PCs(black lines). The two paterns describe ~32% and ~15% of the variance, respectively. Also shown on the right panels are the regressions of the unfiltered summer rainfall intensity (red lines) on the dominant decadal patterns.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
As in Fig. 11 but for rainfall frequency: The two patterns describe ~42% and ~21% of the variance, respectively.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
EOF1 of rainfall intensity, explaining 32% of the variance, describes a generally homogeneous anomaly pattern over China, and the time series show a trend toward increasing intensity during the 1980s and early 1990s. The consistency of increased intensity over most of China suggests that it could be a response to global warming, through enhanced moisture in the atmosphere and, consequently, the intensification of precipitation. Taking account of the EOF1 and PC1 values, a typical increase in rainfall intensity corresponds to about 1 mm day−1, and therefore to about 10% of the seasonal mean rainfall intensity of 10 mm day−1. According to Wang and Gong (2000), the annual-mean surface temperature trend over China from 1951 is mainly contributed by the stronger upward trend during the period 1979–98 of ~1.04°C. Therefore, the Clausius–Clapeyron relation gives an ~7% increase in the saturated specific humidity during the last two decades, which is a similar magnitude to the observed change in rainfall intensity. It is noteworthy that the warming trend over China is much higher than the global mean trend of 0.38°C during the last two decades (Wang and Gong 2000), suggestive that other factors may be involved in increasing the annual temperature over China. Moreover, the PC1 of rainfall intensity also shows a flat phase during the 1960s and 1970s and a slow decrease in the recent decade, suggesting that other processes may be involved, such as decadal variability and the influence of aerosols.
The leading pattern of rainfall frequency, explaining 42% of the variance, simply describes a trend in the 51-yr period—decreasing over northeast and southwest China and increasing over the Yangtze River valley and the northwest. The spatial correlation of EOF1 with the linear trend pattern of rainfall frequency during 1958–2008 (figure not shown) is 0.95. The magnitudes of change in both EOF1 and the linear trends are relatively large over northeast China.
The second patterns of rainfall intensity and frequency show decadal time-scale variability in the period 1958–2008, explaining 15% and 21% of the variance, respectively. The pattern of rainfall intensity indicate opposite signs between the upper Yangtze River basin (negative) and southeast China (positive). Rainfall frequency shows a north–south pattern with relative large changes in southern China (positive), reminiscent of the leading pattern of total rainfall (Fig. 6). The decadal variability of the two variables is coincident with positive phases in the 1970s and 1990s and negative phases in the 1980s, and the transitions seem to be at about the same time. Regression of the unsmoothed data suggests that the large swings in the decadal variability in rainfall intensity and frequency are always associated with large interannual variability, as also evident in Fig. 6.
Combined EOF analysis for normalized rainfall amount, intensity, and frequency has been performed, and the results are shown in Fig. 13. It provides coherent pictures of how changes in these precipitation features fit together. The dominant patterns show similarity to those in Figs. 11 and 12, with trendlike and decadal time-scale variability, respectively. The trendlike pattern of EOF1, explaining 29% of the variance, confirms that increases in both rainfall intensity and frequency contribute to the enhancement of rainfall amount over the Yangtze River valley and surrounding areas from the 1960s to 1990s. Decreases in rainfall amount over the Yellow River valley are mainly contributed by reduced rainfall frequency. The second pattern of the combined EOF analysis, explaining 19% of the variance, highlights the decadal variability in rainfall amount, intensity, and frequency over southern China.
The first two patterns of combined EOF analysis for normalized low-pass filtered rainfall amount, intensity, and frequency during the period 1958–2008.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
The analysis so far has emphasized the presence of decadal variability but also has given indications of behavior that could be influenced by anthropogenic activity. To give further information that may be relevant to this, a more detailed analysis of changes in rain rates has been conducted, using the climatological 30th percentile for light rain (1.1 mm day−1) and 90th percentile for heavy rain (27.45 mm day−1).
Figure 14 shows the dominant light rainfall frequency pattern and its time series for the period 1958–2008. Explaining 56% of the variance, the first pattern dominates the variability of light rainfall frequency. Decreases in light rainfall frequency are similar to the trendlike pattern of the frequency in Fig. 12 but are more widespread throughout China. It is interesting to speculate that the rapid transition toward fewer light rain days in the late 1970s may have coincided with the start of China’s industrialization. Increased aerosol loading could reduce average drop size (Levin and Cotton 2007), which further reduces the incidence of light rainfall. Also, global warming could reduce light rainfall frequency due to an increase in atmospheric stability (e.g., Allen and Ingram 2002). More observations and simulations are needed to clarify these two hypotheses. Nevertheless, the reduction in rainfall frequency, particular light rainfall frequency, is suggestive of anthropogenic influences.
First EOF and PC of the light rainfall frequency, defined as the 30th percentile. This pattern describes ~56% of the variance and hence dominates the signal.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
Figure 15 shows the dominant heavy rainfall frequency patterns and corresponding time series. Variances (34% and 16%) and spatial and temporal features of the leading two heavy rainfall frequency patterns are strikingly similar to those of the total rainfall given in Fig. 6. Spatial correlations of the two heavy rainfall frequency patterns with the rainfall amount patterns are 0.92 and 0.85, respectively. Correlations of the decadal time series are 0.997 and 0.97, while correlations on the regressed interannual time series are 0.98 and 0.96, respectively. This strongly suggests that decadal variability in Chinese summer rainfall is dominated by decadal variability in heavy rainfall frequency. Because the 90th percentile heavy rainfall accounts for about 50%–60% of the total summer rainfall over east China, these two patterns of decadal variability in the frequency of heavy rainfall tend to contribute the most to decadal variability in summer rainfall.
As in Fig. 12 but for the heavy rainfall frequency, defined as the 90th percentile: the first two patterns describe ~34% and ~16% of the variance, respectively.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
4. Discussion and conclusions
This study has confirmed the existence, over the last 51 years, of strong decadal variability in Chinese summer rainfall, which is not unique to this period having been recognized in the earlier decades of the twentieth century. It was shown that these variations are characterized by two dominant patterns: one in which rainfall anomalies over southern China persist throughout summer and the other in which southeast China and the Yangtze and Yellow River basins show variations in seasonality, indicative of changes in the northward penetration of the EASM. Both patterns have contributed to the trend in recent decades toward wetter conditions over southern China and much drier conditions to the north, with increasing desertification and substantial impacts on water resources.
An important result of this study is that the decadal variability in the total rainfall is composed of more complex changes in the intensity and frequency, potentially driven by different factors. Over southern China where summer rainfall shows decadal time-scale variability in the period from 1958 to 2008, rainfall intensity, frequency, and heavy rainfall frequency appear to be dominated by decadal variations. However, the overall increase in rainfall intensity, which started in the late 1970s and peaked in the middle 1990s, may also contain discernible anthropogenic signals. Whether the slow downward trend in the early twenty-first century is due to natural decadal variations or the interplay of anthropogenic effects (global warming and aerosol loading) requires further research. The decrease of the light rainfall frequency may be mainly the result of anthropogenic climate changes. Associated with the reducing light rainfall frequency, the decreasing tendency of total rainfall frequency and the drying conditions over northeast China suggest that global warming signals are most prominent here.
It is of interest to investigate large-scale surface temperature and circulation anomalies associated with the observed long-term variability of summer rainfall over China. Figure 16 shows the regressed surface temperature in the preceding spring and the precipitation and 850-hPa wind in the summer on time series of the leading two EOF patterns (Fig. 6). Corresponding to the interdecadal variability in southern China summer rainfall (EOF1), surface temperature in the preceding spring shows significant warming over the northern Eurasian continent. Compared with the magnitudes and extent of land signals, oceanic warming signals in the equatorial Indian Ocean and South China Sea appear to be secondary. Regressions of 850-hPa wind in summer indicate that an anticyclone in the East China Sea strengthens the western portion of the North Pacific subtropical high, consistent with a warmer land than ocean in spring. Over southern China, a cyclonic circulation anomaly in the lower troposphere is consistent with a positive rainfall anomaly.
Regressions of surface temperature in the preceding spring and precipitation and 850-hPa wind in summer on the time series of the leading two EOF patterns of observed summer rainfall over China. Hatching areas indicate significant regressions of surface temperature, with the significance level at 10%.
Citation: Journal of Climate 24, 17; 10.1175/2010JCLI3794.1
For the second interdecadal pattern of a tripole rainfall anomaly over east China (EOF2), regressed surface temperature emphasizes oceanic signals, with significant warmth in the Indian Ocean, the western Pacific, and eastern Pacific during spring. Observed surface temperature anomalies are stable in the following summer and are similar to the pattern of global warming in the last half century (Solomon et al. 2007). Regressed 850-hPa wind shows westerly wind anomalies from the eastern tropical Indian Ocean to the western Pacific Ocean, consistent with the positive rainfall anomalies over the Indian and western Pacific Oceans. Over East China, the tripole structure of summer rainfall anomalies appears to be related to a cyclonic circulation anomaly over the western North Pacific (the weak EASM).
In summary, the interplay between anthropogenic climate change and natural decadal variability appears to have played an important role in shaping Chinese summer rainfall in the period from 1958 to 2008. Natural decadal variability appears to dominate in general but in the cases of rainfall intensity and the frequency of rain days, particularly light rain days, then the dominant EOFs have a rather different character, being of one sign over most of China and having PCs that appear more trend like. Anthropogenic climate changes, especially global warming signals, are most discernible in the drier region of northeast China and are reflected in the reduction of light rainfall frequency. The overall increase of rainfall intensity can also be attributable to global warming. However, the intensity and frequency of total rainfall probably contain both global warming and decadal variability signals, with the latter most discernible in the heavy rain region of southeast China. The observed change of heavy rainfall frequency is mainly attributed to the decadal variations. A major challenge is how this interplay between natural decadal variations and anthropogenic climate change will shape the future distribution of rainfall together with possible changes in frequency and intensity. While the exact causes of the decadal variations remain uncertain, if they are potentially predictable, then the benefits would be enormous in terms of planning and decision making. In a subsequent paper we will explore these questions further using multicentury climate model simulations for the current climate and for a future climate with enhanced greenhouse gas concentrations.
Acknowledgments
We are grateful for valuable comments and discussions on this work with C.P. Chang and two anonymous reviewers. Yonghui Lei was funded through a NERC Dorothy Hodgkin Award. Julia Slingo acknowledges the support of the NERC National Centre for Atmospheric Science.
REFERENCES
Allen, M., and W. Ingram, 2002: Constraints on future changes in climate and the hydrologic cycle. Nature, 429, 224–232.
Chang, C., Y. Zhang, and T. Li, 2000: Interannual and interdecadal variations of the East Asian summer monsoon and the tropical Pacific SSTs. Part I: Roles of the subtropical ridge. J. Climate, 13, 4310–4325.
Ding, Y., 2004: Seasonal march of the East Asian summer monsoon. East Asian Monsoon, C. P. Chang, Ed., World Scientific Publishing, 3–53.
Ding, Y., and Y. Sun, 2003: Interdecadal variation of the temperature and precipitation pattern in the East Asian monsoon region. Proc. Int. Symp. on Climate Change, Beijing, China, China National Climate Committee and Cosponsors, 66–71.
Ding, Y., Z. Wang, and Y. Sun, 2008: Inter-decadal variation of the summer precipitation in east China and its association with decreasing Asian summer monsoon. Part I: Observed evidences. Int. J. Climate, 28, 1139–1161.
Duchon, C., 1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor., 18, 1016–1022.
Guo, Q., J. Cai, and X. Shao, 2003: Interdecadal variability of East-Asia summer monsoon and its impact on the climate of China. Acta Geogr. Sin., 4, 569–576.
Hulme, M., T. Osborn, and T. Johns, 1998: Precipitation sensitivity to global warming: Comparison of observations with HadCM2 simulations. Geophys. Res. Lett., 25, 3379–3382.
Lau, K.-M., and H. Weng, 2001: Coherent modes of global SST and summer rainfall over China: An assessment of the regional impacts of the 1997–98 El Niño. J. Climate, 14, 1294–1308.
Lau, K.-M., K.-M. Kim, and S. Yang, 2000: Dynamical and boundary forcing characteristics of regional components of the Asian summer monsoon. J. Climate, 13, 2461–2482.
Lau, N.-C., M. J. Nath, and H. Wang, 2004: Simulations by a GFDL GCM of ENSO-related variability of the coupled atmosphere-ocean system in the East Asian monsoon region. East Asian Monsoon, C. P. Chang, Eds., Series on Meteorology of East Asia, Vol. 2, World Scientific, 271–300.
Levin, Z., and W. R. Cotton, 2007: Aerosol Pollution Impact on Precipitation: A Scientific Review. Springer, 482 pp.
Li, H., A. Dai, T. Zhou, and J. Lu, 2010: Response of East Asian summer monsoon to historical SST and atmospheric forcing during 1950-2000. Climate Dyn., 34, 501–514.
Li, W., and Z. Zhao, 1998: Analysis of storms and floods. Heavy Flooding and Climate Anomalies in China in 1998 (in Chinese). China Meteorological Press, 62–63.
Menon, S., J. Hansen, L. Nazarenko, and Y. Luo, 2002: Climate effects of black carbon aerosols in China and India. Science, 297, 2250–2253.
Peng, J., L. Chen, and Q. Zhang, 2005: Multi-scale variations of snow cover over the Qinhai-Xizang Plateau (Tibet) and tropical Pacific SST and their influences on summer rainfall in China. J. Plateau Meteor., 24, 366–377.
Qian, Y., Y. Zheng, Y. Zhang, and M. Miao, 2003: Response of China’s summer monsoon climate to snow anomaly over the Tibetan Plateau. Int. J. Climate, 23, 593–613.
Solomon, S., D. Qin, M. Manning, M. Marquis, K. Averyt, M. M. B. Tignor, H. L. Miller Jr., and Z. Chen, Eds., 2007: Climate Change 2007: The Physical Science Basis. Cambridge University Press, 996 pp.
Sun, Y., S. Solomon, A. Dai, and R. W. Portmann, 2007: How often will it rain? J. Climate, 20, 4801–4818.
Suzuki, S., and B. Hoskins, 2009: The large-scale circulation at the end of the baiu season in Japan as seen in ERA40 data. J. Meteor. Soc. Japan, 87, 83–99.
Trenberth, K. E., and J. W. Hurrell, 1994: Decadal atmosphere-ocean variations in the Pacific. Climate Dyn., 9, 303–319.
Trenberth, K. E., A. Dai, R. M. Rasmussen, and D. B. Parsons, 2003: The changing character of precipitation. Bull. Amer. Meteor. Soc., 84, 1205–1217.
Wang, B., R. Wu, and K.-M. Lau, 2001: Interannual variability of the Asian summer monsoon: Contrasts between the Indian and the western North Pacific–East Asian monsoons. J. Climate, 14, 4073–4090.
Wang, S., and D. Gong, 2000: Enhancement of the warming trend in China. Geophys. Res. Lett., 27, 2581–2584.
Xu, M., C.-P. Chang, C. Fu, Y. Qi, A. Robock, D. Robinson, and H. Zhang, 2006: Steady decline of East Asian monsoon winds, 1969–2000: Evidence from direct ground measurements of wind speed. J. Geophys. Res., 111, D24111, doi:10.1029/2006JD007337.
Yu, R., and T. Zhou, 2007: Seasonality and three-dimensional structure of interdecadal change in the East Asian monsoon. J. Climate, 20, 5344–5355.
Yu, R., B. Wang, and T. Zhou, 2004: Tropospheric cooling and summer monsoon weakening trend over East Asia. Geophys. Res. Lett., 31, L22212, doi:10.1029/2004GRL021270.
Zhai, P., A. Sun, F. Ren, X. Liu, and Q. Zhang, 1999: Changes of climate extremes in China. Climatic Change, 42, 203–218.
Zhai, P., X. Zhang, H. Wan, and X. Pan, 2005: Trends in total precipitation and frequency of daily precipitation extremes over China. J. Climate, 18, 1096–1108.
Zhang, Y., T. Li, and B. Wang, 2004: Decadal change of the spring snow depth over the Tibetan Plateau: The associated circulation and influence on the East Asian summer monsoon. J. Climate, 17, 2780–2793.
Zhao, P., S. Yang, and R. Yu, 2010: Long-term changes in rainfall over eastern China and large-scale atmospheric circulation associated with recent global warming. J. Climate, 23, 1544–1562.
Zhou, T., D. Gong, J. Li, and B. Li, 2009: Detecting and understanding the multi-decadal variability of the East Asian summer monsoon - Recent progress and state of affairs. Meteor. Z., 18, 455–467.
