1 Introduction

Interdecadal climate change is an important reference point for evaluating interannual climate change, and can also be a source of disturbance in long-term climate change. It has become a hot topic in climate change studies since the 1980s, with substantial attention paid to the decadal and interdecadal variability of climate in China and East Asia during recent years (Xu and Chan, 2002; Li et al., 2004; Gong et al., 2011; Chen et al., 2014; Gao et al., 2014; Yim et al., 2014; Zhang et al., 2014; Gong et al., 2016). An increasing trend in precipitation was observed during the second half of the 20th century in western China, where the 20- and 40-yr climate periodicities dominated trends in precipitation variability (Wang et al., 2004). The variation in regional rainfall was related to a change in the global mean surface temperature from the relatively cold period of the 1960s–70s to the relatively warm period of the 1980s–90s (Zhao et al., 2010).

Eastern China experienced a major regime shift in summer precipitation during the late 1970s, while another regime shift occurred in South China in 1992 (Ding et al., 2008). The spatial pattern of the interdecadal variabi-lity of summer precipitation in China is mainly structured with two meridional modes: the dipole pattern and the positive–negative–positive pattern. High Tibetan Plateau (TP) snow and oceanic forcing factors have a positive correlation with subsequent-summer precipitation in the Yangtze River basin and most of South China, and a negative correlation with summer precipitation in North China (Ding et al., 2009). The summer precipitation over eastern China experienced another notable interdecadal change in the late 1990s (Zhu et al., 2011; Huang et al., 2013). With this interdecadal shift, the dominant mode of summer precipitation switched from a meridional tripole pattern to a dipole pattern. Using a high-resolution global atmospheric dataset, Chang et al. (2014) noted that precipitation over southern China reflected a decadal shift in the mid 1990s. As seen from the above studies, the interdecadal shifts of summer precipitation for the past 60 years in eastern China mainly occurred in the late 1970s, early 1990s, and late 1990s, but the change points in different regions were not the same.

It has been concluded that summer climate in East Asia is closely correlated with the East Asian summer monsoon (EASM) and atmospheric circulation (Qian and Qin, 2008; Wang et al., 2015). Meanwhile, other external climate forces include the Pacific Decadal Oscillation (PDO) (Duan et al., 2013; Fu and Lin, 2013), North Atlantic Oscillation (Li and Li, 2000; Gu et al., 2009), Arctic Oscillation (Ju et al., 2005; Yan et al., 2005), and snow cover over the TP (Si and Ding, 2013) and Eurasia (Wu et al., 2009), all of which significantly impact upon precipitation in eastern China by influencing the EASM and atmospheric circulation. The EASM’s northward propagation has exhibited prominent interdecadal variation, with a rapid northward advance prior to the late 1970s and a slow northward movement after the late 1970s (Jiang et al., 2008). During the southward shift of the high-precipitation zone, the two abrupt regime shifts observed in the late 1970s and early 1990s were likely responses to the increased winter and spring snow over the TP, which correspond to two major warming events of the SST in the tropical central and eastern Pacific at the same time (Ding et al., 2009). In strong EASM periods, there is abundant rainfall in North China (Li et al., 2004). Previous work has demonstrated that the variation in summer rainfall over China is due to the interaction of multiple factors (Yan et al., 2005).

Southern China (20°–28°N in eastern China) experienced less-than-average precipitation between the mid-1970s and 1980s, but there was an abrupt change in the early 1990s and rainfall continued to increase into the late 2000s (Fig. 1). However, rainfall in southern China decreased after the late 2000s, while rainfall in northern China (35°–42°N in eastern China) increased. It is particularly important to investigate this change in eastern China in light of global warming because, during the last decade, northern China suffered the most severe drought ever recorded (Zhu et al., 2011). With the goal of investigating the change in summer precipitation in eastern China and the underlying causes, the present work investigated the change in summer precipitation and the associated atmospheric circulation during the late 2000s using long-term observational precipitation datasets and monthly NCEP–NCAR reanalysis datasets. Following this introduction, the data and methods used are described in Section 2. Section 3 presents the interdecadal change and dominant modes of summer precipitation over eastern China. Section 4 addresses the interdecadal change of atmospheric circulation. Section 5 discusses the possible causes of the summer precipitation change. A summary and conclusions are given in Section 6.

2 Data and methods

A daily precipitation dataset of 2400 stations in China between 1961 and 2015 from the National Meteorological Center of the China Meteorological Administration was used in this study. Quality control of this dataset was performed by using the cumulative deviations test (Feng et al., 2004), which left 1840 stations with continuous time series during the study period for analysis.

NCEP–NCAR reanalysis data, including horizontal and vertical winds, specific humidity, geopotential height, and air temperature, as well as NOAA ERSST (Extended Reconstructed Sea Surface Temperature) data, for the same period (1961–2015), were used to investigate the changes in large-scale circulation and SST features. The PDO indices were obtained from NCEP–NCAR (see ).

Because of the uneven distribution of sites in China, which are dense in the east but sparse in the west, results calculated via the arithmetic mean are largely driven by the eastern region. Therefore, the area-weighted averaging method was used to calculate the precipitation in China. Referring to the Jones grid-area weighted-average method (Jones and Hulme, 1996), we divided China into a grid of 71 × 41 sections, each with 1° longitude × 1° latitude.

To compare changes in the spatial patterns of summer precipitation, empirical orthogonal function (EOF) analysis was applied to the summer precipitation in China over the 55-yr period (1961–2015). We also used BP-canonical correlation analysis (BP: Barnett–Preisendorfer) to investigate the relationship between SST and 500-hPa geopotential height by their main component, as described in Jia et al. (2010). The Fourier filtering method was used to extract decadal precipitation features, with the climatological mean defined as the average of 1981–2010.

3 Features of the change in summer precipi-tation patterns

Large positive precipitation anomalies were observed mainly in northern China before the end of the 1970s, as shown in Fig. 1. Then, these anomalies moved southward to the Yangtze River and finally, further into southern China in the 1990s and 2000s. This southward shift of the rainbelt shows the interdecadal variability of summer precipitation because the variability on the time-scales of less than 11 yr has been filtered out in Fig. 1.

Based on Fig. 1, a shift in rainfall patterns in eastern China occurred in the late 1970s, and followed by a second transition period in the early 1990s, which is rela-tively consistent with previous findings (e.g., Ding et al., 2008). The precipitation in northern China underwent significant variability in the late 1990s and was signifi-cantly reduced afterwards. Therefore, the mechanism underlying the precipitation changes in northern China that occurred in 1998 should be considered (Zhu et al., 2011; Huang et al., 2013). The regime shift in precipitation was completed between 1998 and 2008. There was a shift from the pre-1977 tripolar pattern of high precipitation in the north and south regions, with low precipitation between, to a dipole pattern of higher precipitation in the north than that in the south from 1978 to 1992, and again after 2008.

To compare summer precipitation patterns in eastern China (longitudinal range: 105°–125°E), four sub-regions were examined for the period 1961–2015. The year-to-year precipitation anomaly percentages in eastern China are shown in Fig. 2. Significantly less-than-normal precipitation occurred in the 1970s and 2000s for the eastnorthern region, while more-than-normal precipitation was observed in the 1960s and 1980s–1990s. With the ex-ception of the 2000s, precipitation during other decades demonstrated positive anomalies in northern China. It was only in the 1990s that there was significantly higher-than-normal precipitation over the Yangtze River. Two remarkable precipitation anomalies occurred in the early 1990s and early 2000s. In southern China, there was significantly higher-than-normal precipitation in the 1990s and 2000s, with less-than-normal precipitation from the 1960s to 1980s. These trends indicate that summer rainfall underwent periodic changes in eastern China, although there were clear differences in the transition points within the four sub-regions. Since the late 2000s, the northeastern region and northern China have experienced increased precipitation, while the Yangtze River and southern China have seen decreased rainfall (Fig. 2).

Because 1998 was the last transition point between summer rainfall regimes in eastern China (Zhu et al., 2011), we chose 1999 as the starting year for the following comparison. We calculated differences in summer rainfall between 1999–2008 and 2009–15. Figure 3 clearly shows the varied patterns of summer precipitation in eastern China. Above-normal rainfall occurred in southern China, while below-normal rainfall occurred in the north of northern China and south of the Yangtze River during 1999–2008 (Fig. 3a). Beginning in 2009, the rainbelt clearly moved northward, and the rainfall in the whole of the northern China region was above-normal (Fig. 3b).

To compare the spatial features of summer precipitation in China, EOF analysis was applied for the period (1961–2015). Figure 4 shows the first two dominant components of summer precipitation in China, which obtained the interdecadal variability processed by Fourier filtering. Therefore, these principal components mainly indicate the interdecadal variability. The first EOF component displays a dipole pattern (Han and Zhang, 2009), corresponding to the increased rainfall over southern China and the Yangtze River and the decreased rainfall over northern China and northeastern region (Fig. 4a). It is clear that the dipole pattern is approximately opposite to the spatial pattern of summer rainfall for 2009–15, as shown in Fig. 3b. This was the dominant mode of summer precipitation in eastern China, accounting for 33.3% of the total variance. The positive time coefficient indi-cates above-normal precipitation over the Yangtze River and southern China in 1992, with an abrupt change after 2008.

The second EOF component (EOF2) demonstrates a pre-cipitation pattern of positive (northern China)–negative (Yangtze River)–positive (southern China) (Fig. 4b) in the meridional direction, which is consistent with previous findings (e.g., Ding et al., 2008). The EOF2 mode in part resembles the meridional pattern of summer rainfall during 1999–2008, as shown in Fig. 3a. The time coefficient of EOF2 (Fig. 4d) indicates a major regime shift during the late 1970s, characterized by a positive–negative–positive pattern before the late 1970s and by a negative–positive–negative pattern afterwards. The EOF2 mode can be viewed as another dominant mode of summer precipitation in China, which accounts for 21.5% of the total variance. This comparison shows that the observed shift in summer precipitation around the late 2000s corresponded to the transition from the negative–positive–negative tripolar mode to the positive–negative dipole mode of the meridional precipitation.

4 Changes in atmospheric circulation

Figure 5 illustrates the variation in horizontal winds at 700 and 200 hPa between 1999–2008 and 2009–15. Figure 5a shows an anomalous cyclone over Lake Baikal at 700 hPa, which forced the cold air southward. Southerly wind anomalies were maintained in eastern China, which moved water vapor northward. The circulation anoma-lies at 200 hPa shown in Fig. 5b were responsible for an enhancement of the westerly jet in the upper levels. Precipitation in the export area of the jet on the right-hand side should have been increased—a region that corresponded well to North China. This configuration of high- and low-level circulation could have caused increased summer precipitation in northeastern region and northern China, and decreased precipitation over the Yangtze River and southern China during 2009–15, as compared with 1999–2008.

The differences in vertically integrated specific humidity from 300 to 1000 hPa between 1999–2008 and 2009–15 are shown in Fig. 6a. The more-normal specific humidity increased the rainfall in northern China and northeastern region during 2009–15, as compared with 1999–2008, while the less-normal humidity caused decreased rainfall over the Yangtze River and southern China regions. The moisture transport changes illustrated in Fig. 6b further explain the differences in the water vapor content. Moisture from the Pacific Ocean was transported northward to northern China and northeastern region, increasing the water vapor content in these regions. However, the significant southward anomalies within 100°–120°E decreased moisture transport into southern China and the Yangtze River, decreasing the humidity in these regions. The vertical wind and moisture fluxes partly explain the precipitation changes in these four sub-regions. However, to determine if these anomalies were due to decadal shifts or stochastic noise, the background circulation should be investigated (Zhu et al., 2011).

The western Pacific subtropical high (WPSH) signifi-cantly influences the summer climate over eastern China (Hu, 1997; Zhou et al., 2009); thus, we examined the differences of the 500-hPa geopotential height between 1999–2008 and 2009–15. Figure 7a shows strong nega-tive height anomalies over Lake Baikal, which agree with the southern moisture transport to northern China and northeastern region shown in Fig. 6b. The height anoma-lies over Lake Baikal may be related to the phase shift of the PDO via dynamic interaction between the oceanic circulation and the atmosphere (Kushnir, 1994). The WPSH moved westward during 2009–15, as compared with 1999–2008 (Fig. 7b), which strengthened the southerly moisture transport into northern China and northeastern region, as shown in Fig. 6a. Additionally, sea surface temperature (SST) forcing is an essential factor that modu-lates the interdecadal variation of the WPSH (Wu and Zhou, 2008).

The subtropical jet stream can explain the interaction between climate systems at low and high latitudes. The blocking effect is enhanced with stronger jets, and the mixing of warm and cold air decreases, and vice versa (Zhu et al., 2011). Figure 7c shows the jet stream located over northeastern China at 200 hPa, which is indicative of a weak interaction near the jet. But why did the jet stream become stronger during 2009–15? The negative height anomalies over Lake Baikal are linked to the increased air temperature gradient averaged over 30°–50°N, 110°–120°E at 200 hPa (Fig. 7d). This increased meridional temperature gradient may have been responsible for the strengthened jet stream due to the thermal wind relationship.

We also analyzed the 11-yr filtered zonal wind ano-malies at 200 hPa, averaged for 100°–140°E. As shown in Fig. 8a, the westerly jet at 200 hPa revealed interdecadal changes accompanied by varied summer rainfall (Huang et al., 2013). The zonal wind anomalies over eastern China showed a positive–negative–positive tripole structure in the meridional direction before the mid 1970s. Since the early 1990s, however, a negative–positive dipole pattern of zonal wind anomalies has been observed, which is opposite to the pattern observed during the late 1970s and 1980s. The interdecadal variation of the subtropical westerly jet over eastern China was thus responsible for the switch in the zonal wind from a tripole to a dipole pattern (Kwon et al., 2007). The zonal wind anomalies at 500 hPa are shown in Fig. 8b, and are consistent with those at 200 hPa.

The zonal wind anomalies in the meridional direction at 200 and 500 hPa are consistent with the summer precipitation patterns in eastern China depicted in Figs. 1, 2. Both zonal wind and summer rainfall demonstrated a change during the late 2000s. This suggests that changes in the mid and upper tropospheric flow during the late 2000s affected the low-level EASM circulation, and further changed summer rainfall patterns in eastern China (Huang et al., 2013).

5 Possible causes of the changes in precipitation pattern

To explore the connection between various atmospheric circulations, we correlated the summer mean PDO index and 500-hPa geopotential height during 1979–2015 (Fig. 9). Negative correlation was observed from Lake Baikal to the central North Pacific, which suggests that the phase shift of the PDO might have been responsible for the height anomalies over Lake Baikal. Thus, the PDO likely impacts upon summer rainfall in eastern China by modulating the atmospheric circulation over Lake Baikal (Zhu et al., 2011).

To further investigate the mechanism of how the PDO affects interdecadal changes in summer precipitation in eastern China, we used canonical correlation analysis between 500-hPa geopotential height in summer with a 9-yr moving average and the Pacific SST during the same period (Fig. 10). As can be seen, when the SSTs in high and mid latitudes over Northwest Pacific are warm, but SSTs in the central and eastern equatorial Pacific are cool, the Pacific Ocean is in the cold phase of the PDO (Mantua et al., 1997). The 500-hPa geopotential height indicates a positive Pacific–Japan (PJ) teleconnection, in which the negative–positive–negative–positive–negative PJ decadal teleconnection wave train was in the warm pool near western North America (Fig. 10a).

Summer climate over East Asia is greatly influenced by the thermal state of the warm pool over the tropical western Pacific and corresponding convective activities. Previous findings have indicated that the 500-hPa geopotential height would create a positive PJ decadal teleconnection wave train when the PDO is in a cold phase (Huang and Sun, 1994). The cold phase of the PDO is responsible for strong convective activities above the Indochina Peninsula around the South China Sea from the Philippines and northward of the WPSH, corresponding to drought in the middle and lower reaches of the Yangtze River to Japan, and increased rainfall in northern China. Alternatively, a negative PJ teleconnection wave train would bring increased rainfall in the middle and lower reaches of the Yangtze River to Japan, and decreased rainfall in northern China. Meanwhile, convective activity around the Philippines is weak, and the location of the WPSH is southward. Thus, the phase shift of the PDO during the late 2000s may have been caused by a negative PJ teleconnection wave train over the Pacific to Japan, resulting in summer precipitation variation in eastern China.

6 Summary and conclusions

According to this study, there were observable changes in summer precipitation patterns in eastern China during the late 2000s: more rainfall in the northern China and northeastern sub-regions, and less rainfall in the Yangtze River and southern China sub-regions. Additionally, atmospheric circulation changes corres-ponded to the NCAR monthly reanalysis datasets. During 1999–2008 and 2009–15, atmospheric circulation ano-malies over eastern China, including the lower- and upper-level winds and their divergence patterns, corresponded well with the changes in summer precipitation patterns. We also determined that changes in the water vapor content in the different sub-regions (increased in northern China and northeastern region, and decreased in the Yangtze River and southern China) could account for the precipitation changes in eastern China.

Preliminary analyses also revealed that the interdecadal changes in summer precipitation in eastern China may have been related to prominent large-scale circulation features. Negative height anomalies over Lake Baikal during 2009–15, as compared with 1999–2008, the shift in the meridional temperature gradient southward, the intensification of the westerly jet (via the thermal wind relation), and increased moisture flux out of the northern China and the northeastern region, all caused increased specific humidity in northern China and the northeastern region. The WPSH shifted westward during 2009–15, which led to increased moisture flux into northern China and the northeastern region, and increased water vapor content. There is also observational evidence of the association between the PDO and summer rainfall in eastern China on the interdecadal timescale. The shift of the PDO phase in the late 2000s led to a PJ-type teleconnection wave train over the Pacific to Japan, resulting in summer precipitation variation in eastern China.

Because of time limitations, the present study focused on the mid-/high-level circulation anomalies and potential causes of changes in summer precipitation over eastern China during the late 2000s. However, several key questions remain: How long will the rainfall over northern China and the northeastern region remain as a positive anomaly? What is the connection between the regime shift of summer rainfall over eastern China and global warming? More attention should be paid to answering these questions in future work, which will require longer time rainfall datasets.