1. Introduction
Southwest China (SWC), characterized by complex topography, receives considerable rainfall during summer. It contains the headwaters of many rivers, including the Yangtze, Langcang, and Nujiang, which provide as much as 46% of China’s available water resources. Because of large interannual variability of summer rainfall over SWC, droughts and floods hit SWC frequently, especially over the last decade, resulting in severe economic losses (Wang et al. 2015a). It is, therefore, imperative to understand the mechanisms responsible for variability of the rainfall over SWC and its predictability.
Before the severe summer drought in SWC in 2006, variability of rainfall over SWC received little attention. In the last decade, progress in understanding characteristics of the droughts and floods in SWC and their possible causes has been made, especially for the dry season (Wang et al. 2015a). Warm sea surface temperature (SST) anomalies in the western North Pacific and the central equatorial Pacific favor a reduction in autumn rainfall over SWC (Zhang et al. 2013; Wang et al. 2015b), while suppressed winter rainfall is mostly caused by negative phase of the North Atlantic Oscillation (Jiang and Li 2011; Xu et al. 2012). Warm SST anomalies in the Indian Ocean (IO) may favor below-normal spring rainfall over SWC (Huang et al. 2012).
As for summer rainfall, recent studies focused on characteristics of rainfall anomalies and associated atmospheric circulation anomalies in 2006 and 2011 (e.g., Li et al. 2011; Li et al. 2014), when severe droughts occurred. It is found that patterns of rainfall anomalies are different for the two summers, as are the anomalies of the western North Pacific high, a key atmospheric circulation system for SWC rainfall variability (Li et al. 2014). Therefore, it is important to identify different modes of rainfall anomalies and associated circulation anomalies. In addition, although some features of atmospheric circulation anomalies associated with summer rainfall anomalies over SWC, especially eastern SWC, have been reported, mechanisms responsible for the associated atmospheric circulation anomalies remain unclear.
Current prediction skill of season mean rainfall is largely dependent on the prediction skill of SST anomalies in the tropics (e.g., Kumar et al. 2013), which can affect prediction of extratropical climate by tropical convection anomalies excited by teleconnections (Wang et al. 2016). Indeed, the anomalies in circulation and rainfall in East Asia are attributed partially to tropical convection anomalies (e.g., Nitta 1987; Nitta and Hu 1996; Jiang et al. 2016). Interannual variability of the summer rainfall in East Asia is significantly modulated by variability of the convections around the Philippine Sea (PS; Nitta and Hu 1996). Interannual variability of the midsummer rainfall in the eastern edge of the Tibetan Plateau (TP) is affected by the convection anomalies over the Maritime Continent (Jiang et al. 2015). Li et al. (2016) also found that the rainfall anomalies over eastern SWC are different during different phases of summer Madden–Julian oscillation activities because of different convection anomalies over the PS and the eastern IO. All these studies indicated that rainfall over SWC is affected by tropical convection anomalies. However, the relationship of tropical convection with the rainfall over SWC has not been well understood.
In this study, we focus on the relationship between tropical convection and the rainfall over SWC on the interannual time scale, as well as their links to tropical SST anomalies. The remainder of this paper is organized as follows. A description of the data and methods is presented in section 2. The dominant modes of summer rainfall over SWC and associated circulation anomalies are presented in section 3. The roles of tropical convection anomalies over different regions in different modes of rainfall anomalies over SWC are investigated in section 4. The relationship between tropical convection and simultaneous and previous SST, as well as the relationship between tropical SST anomalies and rainfall anomalies over SWC, are discussed in section 5. Finally, a discussion and summary are presented in section 6.
2. Data and methods
The datasets used in this study include ERA-Interim (Dee et al. 2011); monthly mean SST from the HadISST dataset from 1979 to 2015 with a horizontal resolution of 1° in both latitude and longitude (Rayner et al. 2003); monthly mean rainfall from the Global Precipitation Climatology Project (GPCP), version 2.2, monthly rainfall analysis dataset from 1979 to 2015 (Adler et al. 2003); and the rain gauge data (1979–2015) from the latest version [version 3 (V3)] of surface climatological data compiled by the China National Meteorological Information Center.
To investigate different roles of convection over various regions, partial correlations (e.g., Behera and Yamagata 2003) and partial regression are used in this study. The partial correlation coefficient between two variables A1 and A2, after removing the influence of variable A3, is expressed as, 
To highlight interannual variability, interdecadal variability of some variables has been removed by a nine-point Gaussian-type filter before they are used to calculate correlation and regression. The empirical orthogonal function (EOF) analysis is also used to extract dominant modes of variability of rainfall over SWC. All statistical significance tests for correlations and regressions are performed using the two-tailed Student’s t test.
3. Dominant modes of summer rainfall in southwest China
According to administrative divisions of China, SWC includes the municipality of Chongqing; the provinces of Sichuan, Yunnan, and Guizhou; and the Tibet Autonomous Region. In this study, SWC is defined as the part of China in the rectangular region 21°–35°N, 90°–110°E. This region does not include western Tibet, where only a few rain gauge stations exist. The positions of the rain gauge stations are denoted by closed circles in Fig. 1a. As shown in Fig. 1, SWC features complex topography, including plateau and basin. Terrain height ranges from a few hundred meters to several thousand meters. The summer rainfall over SWC generally decreases from southeast to northwest, exceeding 6 mm day−1 in most of SWC.
(a) Terrain height (km; shading) and summer mean rainfall (mm day−1; contours) over China and positions of rain gauge stations in SWC used in this study. (b) Summer mean rainfall over SWC and administrative divisions of China in this region.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Terrain height (km; shading) and summer mean rainfall (mm day−1; contours) over China and positions of rain gauge stations in SWC used in this study. (b) Summer mean rainfall over SWC and administrative divisions of China in this region.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Terrain height (km; shading) and summer mean rainfall (mm day−1; contours) over China and positions of rain gauge stations in SWC used in this study. (b) Summer mean rainfall over SWC and administrative divisions of China in this region.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Figure 2 shows the first two modes of an EOF analysis applied to the temporal correlation matrix of summer rainfall, as well as corresponding principal components (PCs). The first four EOFs account for 19%, 14%, 6%, and 5% of the total variance, respectively. Here, we only discuss the first two modes, which have larger variance and are statistically significant according to the rule suggested by North et al. (1982). Both the first two EOFs generally exhibit a north–south dipole pattern. The first EOF shows increases in rainfall over most of SWC except the northern edge and central Sichuan, with maxima in eastern Tibet and Guizhou. The second EOF presents dry conditions in the south but wet conditions in the north. Both the first two PCs exhibit large variability on interannual and interdecadal time scales. The generally dry conditions in SWC over the last decade and the three severe summer droughts (2006, 2011, and 2013) are well reflected by the first mode. The drought in 2006 is also strongly projected to the second mode. As indicated by the variance explained by the second mode, the out-of-phase rainfall anomalies between southern SWC and northern SWC play an important role in modulating the spatial variation of droughts over SWC. For example, because of the combination of the first two modes, the 2006 drought in SWC occurred mostly in northern SWC.
(a) Spatial pattern of the first mode of an EOF analysis applied to temporal correlation matrix of JJA rainfall and (c) corresponding PC. (b),(d) As in (a),(c), but for the second mode. The thick lines in (c) and (d) are nine-point Gaussian-type filtered values of the corresponding PC.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Spatial pattern of the first mode of an EOF analysis applied to temporal correlation matrix of JJA rainfall and (c) corresponding PC. (b),(d) As in (a),(c), but for the second mode. The thick lines in (c) and (d) are nine-point Gaussian-type filtered values of the corresponding PC.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Spatial pattern of the first mode of an EOF analysis applied to temporal correlation matrix of JJA rainfall and (c) corresponding PC. (b),(d) As in (a),(c), but for the second mode. The thick lines in (c) and (d) are nine-point Gaussian-type filtered values of the corresponding PC.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
To illustrate anomalies of circulation, tropical convection, and water vapor transport related to the first two modes, we calculate regressions of winds and vertically integrated water vapor transport against the detrended first two PCs (Fig. 3). The rainfall anomalies related to the first principal component (PC1) indicate that the rainfall over SWC is positively correlated with that in the middle and lower reaches of the Yangtze River and Japan, as well as the western Maritime Continent (WMC), but negatively correlated with rainfall over the northwestern Indian subcontinent and the PS. In the lower troposphere, a positive PC1 is associated with anomalous anticyclones over the western North Pacific and South Asia and an anomalous cyclone over northeastern Asia. This anomalous wind pattern causes anomalous convergence of water vapor flux from the southeastern TP eastward to southern Japan, where above-normal rainfall can be seen.
Patterns of regression of 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on detrended (a) PC1 and (d) PC2, and patterns of regression of vertically integrated water vapor transport (kg m−1 s−1; vectors) on detrended (c) PC1 and (d) PC2 and its divergence (10−4 kg m−2 s−1; shading). Only values exceeding the 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on detrended (a) PC1 and (d) PC2, and patterns of regression of vertically integrated water vapor transport (kg m−1 s−1; vectors) on detrended (c) PC1 and (d) PC2 and its divergence (10−4 kg m−2 s−1; shading). Only values exceeding the 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on detrended (a) PC1 and (d) PC2, and patterns of regression of vertically integrated water vapor transport (kg m−1 s−1; vectors) on detrended (c) PC1 and (d) PC2 and its divergence (10−4 kg m−2 s−1; shading). Only values exceeding the 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
A positive second principal component (PC2) is associated with above-normal rainfall from the central TP to central China, to the south and east of Japan, and over the WMC, but below-normal rainfall over southeast China, to the south of the TP, and around the PS. It is also associated with anomalous anticyclones from the central Indian subcontinent to the western North Pacific and an anomalous cyclone over the tropical IO. Compared with PC1, the anomalous anticyclones from the northern Bay of Bengal to southeast China shift northward, leading to a northward shift of the convergent zone of water vapor on the northern flank of the anomalous anticyclones and the divergent zone on the center. Consequently, the rainfall over SWC increases in the north but decreases in the south. The rainfall anomalies associated with PC2 over the tropics indicate that a positive PC2 is accompanied by enhanced local Hadley circulation, which further affects rainfall in the extratropics through the anomalous anticyclones (Figs. 3b,d).
A positive PC1 is associated with significant anomalous northeasterlies to the north of the areas with above-normal rainfall in eastern SWC, while no significant anomalous winds are found in this region for PC2. Given climatological southerly water vapor transport over East Asia, the PC1-related anomalous northerly water vapor transport indicates a weaker-than-normal southerly water vapor transport. Although both PC1 and PC2 are associated with significant convection anomalies over both the WMC and the PS, PC1 (PC2) has a closer relationship with convection over the PS (WMC). The anomalies of water vapor flux also exhibit similar features. The different roles of the convection activities over the WMC and the PS for rainfall anomalies over SWC will be discussed in the following section.
4. Different roles of convection over the western Maritime Continent and the Philippine Sea
To investigate the possible impacts of the convection over the WMC and the PS on rainfall over SWC, we use regional average rainfall over 10°S–5°N, 90°–115°E (10°–20°N, 120°–140°E) to represent convection intensity over the WMC (PS). The two regional average rainfall indices are presented in Fig. 4. They exhibit strong interannual variability, with no apparent long-term trend. However, there is a weak interdecadal change in the PS rainfall around 1999, and after that rainfall anomalies are mostly positive. Convection over the two regions varies out of phase, with a correlation coefficient of −0.49. The correlations of monthly mean rainfall between the two regions indicate that the highest correlation occurs in July. Because PC1 is negatively correlated with convection over the PS, we use the negative regional average rainfall over the PS to calculate correlation and the regression in this section for easier comparison with those related to PC1.
Time series of average rainfall over the PS (10°S–5°N, 90°–115°E; black curve with plus signs) and the WMC (10°–20°N, 120°–140°E; red curve with open circles).
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Time series of average rainfall over the PS (10°S–5°N, 90°–115°E; black curve with plus signs) and the WMC (10°–20°N, 120°–140°E; red curve with open circles).
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Time series of average rainfall over the PS (10°S–5°N, 90°–115°E; black curve with plus signs) and the WMC (10°–20°N, 120°–140°E; red curve with open circles).
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Figures 5a and 5b show regression patterns of detrended GPCP rainfall and 850-hPa winds onto the detrended rainfall indices for the WMC and the PS. Suppressed convection over the PS is accompanied by above-normal rainfall around the WMC and from the southeastern TP to the east of southern Japan, and below-normal rainfall over central India and the northern Bay of Bengal. It is associated with anomalous anticyclones over the western North Pacific and the northern Bay of Bengal and an anomalous cyclone over northeastern Asia. These features are considered as conditions for a strong East Asian summer monsoon (Wang et al. 2008). On the other hand, the enhanced convection over the WMC is associated with significant below-normal rainfall over the western North Pacific and the northern Bay of Bengal and above-normal rainfall from the south-central TP northeastward to the Korean Peninsula. It is accompanied by anomalous anticyclones over the western North Pacific and the northern Bay of Bengal and a cyclone over the tropical Indian Ocean. The features of anomalous rainfall and wind patterns related to the suppressed convection over the PS and the enhanced convection over the WMC are similar to those related to PC1 and PC2, respectively, but the decreases in rainfall related to PC2 over southern SWC are not significant and the magnitude of anomalous rainfall associated with the convection anomalies is weaker than that related to the PCs.
Patterns of regression of detrended 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial regression. Only values exceeding 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of detrended 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial regression. Only values exceeding 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of detrended 850-hPa winds (m s−1; vectors) and GPCP rainfall (mm day−1; shading) on (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial regression. Only values exceeding 90% confidence level are shown for rainfall anomaly; thick arrows indicate values exceeding 90% confidence level in either the zonal or the meridional component, and the black dashed lines denote the topographic height of 1500 m.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Western North Pacific subtropical high is one of the important circulation systems affecting summer rainfall over East Asia (Lu 2002; Jiang et al. 2011). The suppressed convection over the PS is associated with a dipole pattern of geopotential height anomalies over East Asia, indicating a strong and southward-shifted western North Pacific subtropical high (Fig. 6a). The midtropospheric geopotential height anomalies are consistent with the low-level wind anomalies. On the other hand, the enhanced convection over the WMC is accompanied by significant increases in 500-hPa geopotential height from the southern TP eastward to the western North Pacific, where an anomalous anticyclone is found in the lower troposphere. The geopotential height anomalies indicate that the enhanced convection over the WMC is accompanied by an intensified and westward-shifted western North Pacific high.
As in Fig. 5, but for 500-hPa geopotential height (gpm). Stippling indicates values exceed 90% confidence level, and the contour interval is 1 gpm.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
As in Fig. 5, but for 500-hPa geopotential height (gpm). Stippling indicates values exceed 90% confidence level, and the contour interval is 1 gpm.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
As in Fig. 5, but for 500-hPa geopotential height (gpm). Stippling indicates values exceed 90% confidence level, and the contour interval is 1 gpm.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Since water vapor is concentrated in the lower troposphere, anomalies of water vapor transport associated with convection anomalies over the PS and WMC are basically similar to the wind anomalies in the lower troposphere. Thus, the suppressed convection over the PS favors more water vapor transported to SWC by the two anomalous anticyclones over the western North Pacific and the northern Bay of Bengal (Figs. 5a,b). As descending motion prevails in the center of the subtropical high, rainfall generally decreases in the center of the area with increases in 500-hPa geopotential height but increases on its northern flank. These anomalies associated with the suppressed convection over the PS favor significant rainfall increases in eastern Tibet, west-central Sichuan, Guizhou, and Chongqing (Fig. 7a). Compared with the anomalies associated with the suppressed convection over the PS, the enhanced convection over the WMC-related anomalous anticyclone over the western North Pacific shifts northward, especially the western part (Fig. 5b); the region with increased geopotential height over the western North Pacific also shifts northward (Fig. 6b). The northward shift of the anticyclone could cause the northward shift of the significant above-normal rainfall on the northern flank of the anomalous anticyclone (Fig. 7b). On the other hand, southern SWC is located at the center of the region with increased geopotential height and thus receives below-normal rainfall (Figs. 6b, 7b).
Correlations of detrended rainfall with (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial correlation. Correlations of detrended rainfall with detrended (e) PC1 and (f) PC2. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Correlations of detrended rainfall with (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial correlation. Correlations of detrended rainfall with detrended (e) PC1 and (f) PC2. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Correlations of detrended rainfall with (a) negative detrended average rainfall over the PS (−PSR) and (b) detrended average rainfall over the WMC (WMCR). (c),(d) As in (a),(b), but for partial correlation. Correlations of detrended rainfall with detrended (e) PC1 and (f) PC2. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Because of the high correlation of convection between the PS and the WMC, we also calculate partial regressions of rainfall, 850-hPa winds, and 500-hPa geopotential height on the two average rainfall indices to highlight their different roles in anomalies of regional rainfall and atmospheric circulation. The anomalous features in East Asia associated with the enhanced convection over the PS do not change significantly after removing the effect of convection over the WMC (Figs. 5c, 6c).
However, the enhanced WMC convection-related anomalies change significantly after removing the effect of the convection over the PS. The weakening of the anticyclone over the western North Pacific is apparent (Fig. 5d). There is an anomalous anticyclone around Japan, which is accompanied by below-normal rainfall on the center but above-normal rainfall on the northwestern flank (Fig. 6d). Geopotential height increases from eastern SWC to the east of Japan, but decreases in the tropical Indian Ocean (Fig. 6d). There are no significant geopotential height anomalies around the southern TP. Rainfall increases significantly in the northwestern and northern flanks of the increased geopotential height (Figs. 5d, 6d).
Partial correlations of rainfall over SWC with the two average rainfall indices are presented in Figs. 7c and 7d. The correlations of rainfall with convection over both the PS and the WMC decrease over eastern Tibet after removing the effect from each other, suggesting both the suppressed convection over the PS and the enhanced convection over the WMC tend to enhance rainfall over eastern Tibet. The different impacts of convection anomalies over the PS and the WMC on rainfall over central Sichuan get clearer in the patterns of partial correlation. Comparisons among the patterns of partial correlation and patterns of correlation of rainfall with the PCs indicate that the convection anomalies over the PS can generate the first EOF-like pattern of rainfall anomalies, without significant rainfall anomalies over Yunnan. However, the convection anomalies over the WMC only contribute to the north–south dipole pattern over eastern SWC, with significant rainfall anomalies only over central Sichuan and southern Gansu.
It is worth noting that the suppressed convection over the PS is associated with significant geopotential height increases over the tropics. Previous studies reported that geopotential height increases uniformly in the tropics during El Niño–Southern Oscillation (ENSO) decaying summer and convection is suppressed over the PS (e.g., Wang et al. 2000; Xie et al. 2009). Thus, parts of the anomalies associated with the suppressed convection over the PS cannot be regarded as a response to the diabatic cooling associated with the suppressed convection over the PS. A linear baroclinic model simulation indicated that suppressed convection over the PS can excite the anomalous pattern of 850-hPa winds and 500-hPa geopotential height over East Asia, as well as the anomalous anticyclone to the south of the TP [Fig. 13 in Jiang et al. (2016)].
Simulation of a linear baroclinic model [Fig. 10 in Jiang et al. (2015)] also indicates that enhanced convection over the Maritime Continent can cause decreases in geopotential height to the west of the heating center but increases in geopotential height to the north and the northeast, which are similar to the anomalies in Fig. 6b rather than the anomalies in Fig. 6d. Given that the suppressed convection over the PS is associated with significant increases in geopotential height to the northwest, the removal of the effect of convection over the PS also removes anomalies around the South China Sea associated with both the convection over the WMC and the PS. The model simulation also shows that the increase in 500-hPa geopotential height over the South China Sea and the PS is accompanied by divergence in the lower troposphere, which could suppress convection there. Results from the partial correlations and regressions are also consistent with previous results from numerical simulations that local convection–circulation feedback is important for the enhanced convection over the WMC exciting the anomalous anticyclone and the increase in geopotential height over southeast China and the western North Pacific (Jiang et al. 2015).
To further highlight the different roles of convection over the PS and the WMC in interannual variability of rainfall over SWC, we composite rainfall anomalies based on various patterns of summer convection anomalies over the PS and the WMC. Because of the limited sample, we only consider six types of convection anomalies patterns, and the details are listed in Table 1. Because of the negative correlation of convection anomalies between the WMC and the PS, the most frequently occurring pattern of convection anomalies is positive or negative convection anomalies over the PS accompanied by opposite anomalies over the WMC (Fig. 8a). These opposite convection anomalies are often associated with in-phase rainfall anomalies over SWC (Figs. 8a,d). When convection anomalies over the PS are not accompanied by significant convection anomalies over the WMC, rainfall increases over most of Tibet, western Sichuan, Chongqing, and Guizhou but decreases over central Sichuan. These results are consistent with the results from partial correlation (Figs. 7c, 8b,e). We also try to composite rainfall anomalies for the years when significant convection anomalies over the WMC are not accompanied by significant convection anomalies over the PS. However, there is only one case where significant positive or negative anomalies over the WMC occurred with normal convection over the PS. Thus, we composite rainfall for the years when significant convection anomalies over the WMC are not accompanied by opposite convection anomalies over the PS. The results show that the convection anomalies over WMC are accompanied by convection anomalies over the eastern Maritime Continent, the southern South China Sea, and the southern PS. Consequently, significant rainfall anomalies occur not only over central Sichuan, but also over the eastern SWC (Figs. 8c,f). Comparisons among the patterns of rainfall anomalies indicate that the region with significant rainfall anomalies over eastern SWC in Fig. 8f shifts southward compared with that in Fig. 8e, consistent with the southward shift of significant rainfall anomalies over the PS (Figs. 8b,c). These composite analyses further indicate that rainfall anomalies over SWC are significantly modulated by convection anomalies over both the WMC and the PS.
Years with convection anomalies over the PS and the WMC according to various categories of normalized summer rainfall averaged over the Philippine Sea (PSR) and the western Maritime Continent (WMCR).
(a) Composite normalized detrended GPCP summer rainfall anomalies for years with significant summer rainfall anomalies over the PS and opposite significant summer rainfall anomalies over the WMC. (d) As in (a), but for rain gauge data over SWC. (b),(e) As in (a),(d), respectively, but for significant summer rainfall anomalies over the PS that occurred without significant summer rainfall anomalies over the WMC. (c),(f) As in (a),(d), respectively, but for significant summer rainfall anomalies over the WMC that occurred without opposite summer rainfall anomalies over the PS. Years for composite are listed in Table 1. Composite values are differences between years with opposite convection anomalies. For example, values in (a) are differences in summer rainfall between the years with PSR < −0.5 and 0.2 > WMCR > −0.2 and the years with PSR > 0.5 and 0.2 > WMCR > −0.2. Stippling indicates values exceeding 90% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Composite normalized detrended GPCP summer rainfall anomalies for years with significant summer rainfall anomalies over the PS and opposite significant summer rainfall anomalies over the WMC. (d) As in (a), but for rain gauge data over SWC. (b),(e) As in (a),(d), respectively, but for significant summer rainfall anomalies over the PS that occurred without significant summer rainfall anomalies over the WMC. (c),(f) As in (a),(d), respectively, but for significant summer rainfall anomalies over the WMC that occurred without opposite summer rainfall anomalies over the PS. Years for composite are listed in Table 1. Composite values are differences between years with opposite convection anomalies. For example, values in (a) are differences in summer rainfall between the years with PSR < −0.5 and 0.2 > WMCR > −0.2 and the years with PSR > 0.5 and 0.2 > WMCR > −0.2. Stippling indicates values exceeding 90% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
(a) Composite normalized detrended GPCP summer rainfall anomalies for years with significant summer rainfall anomalies over the PS and opposite significant summer rainfall anomalies over the WMC. (d) As in (a), but for rain gauge data over SWC. (b),(e) As in (a),(d), respectively, but for significant summer rainfall anomalies over the PS that occurred without significant summer rainfall anomalies over the WMC. (c),(f) As in (a),(d), respectively, but for significant summer rainfall anomalies over the WMC that occurred without opposite summer rainfall anomalies over the PS. Years for composite are listed in Table 1. Composite values are differences between years with opposite convection anomalies. For example, values in (a) are differences in summer rainfall between the years with PSR < −0.5 and 0.2 > WMCR > −0.2 and the years with PSR > 0.5 and 0.2 > WMCR > −0.2. Stippling indicates values exceeding 90% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
5. Role of tropical SST
As discussed before, the convection over the PS is affected by ENSO. Thus, understanding of relationship between convection over the two regions and tropical SST may enhance our skill in predicting rainfall anomalies over SWC. Here, we discuss the SST anomalies related to the convection anomalies over the PS and the WMC and their relationships with rainfall anomalies over SWC.
Figure 9 shows regression patterns of SST from December–February (DJF) to June–August (JJA) against the JJA-averaged rainfall indices for the PS and the WMC. The suppressed convection over the PS is associated with warm SST anomalies in most of the tropical Indian Ocean and the South China Sea simultaneously. Previous studies reported that ENSO can cause such a pattern of SST anomalies during its decaying phase (e.g., Xie et al. 2009; Ding and Li 2012). As expected, the suppressed convection over the PS is associated with significant cold SST anomalies in the equatorial central and eastern Pacific, and the eastern Indian Ocean in the previous winter and spring, with stronger anomalies in the Pacific in winter. On the other hand, the enhanced convection over the WMC is accompanied by significant local warm SST anomalies, as well as cold SST anomalies in the equatorial central Pacific, simultaneously. It is also associated with significant warm SST anomalies in the tropical eastern Pacific and the tropical western Indian Ocean in the previous winter, but the magnitude of anomalous SST is weaker compared with that associated with the suppressed convection over the PS. The warm SST anomalies in winter weaken rapidly in spring.
Patterns of regression of detrended SST from DJF to JJA against negative detrended average JJA rainfall over the PS and detrended average JJA rainfall over the WMC. Stippling indicates values exceed 99% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of detrended SST from DJF to JJA against negative detrended average JJA rainfall over the PS and detrended average JJA rainfall over the WMC. Stippling indicates values exceed 99% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Patterns of regression of detrended SST from DJF to JJA against negative detrended average JJA rainfall over the PS and detrended average JJA rainfall over the WMC. Stippling indicates values exceed 99% confidence level.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Comparisons among the anomalous SST patterns indicate that convection over the PS is suppressed significantly during El Niño decaying summers, but enhanced convection over the WMC occurs not only during El Niño decaying summers but also La Niña developing summers. Possible mechanisms responsible for the relationship of convection over the PS with ENSO are well reported by previous studies (Xie et al. 2009). ENSO can suppress simultaneous convection over the WMC by changes in the Walker circulation but enhance convection during its decaying summers by inducing basinwide warming in the Indian Ocean (e.g., Ding and Li 2012). Because of the close relationship of ENSO and convection over the PS and the WMC, ENSO may be used to predict rainfall anomalies over SWC. It should be noted that significant SST anomalies in the tropical southeastern Indian Ocean are not always accompanied by ENSO (e.g., Ding and Li 2012). Thus, the independent variability of SST in the southeastern Indian Ocean should also be considered in prediction of rainfall over SWC.
ENSO and SST anomalies in the tropical southeastern Indian Ocean have close connections to convection over the PS and WMC, and thus they may affect rainfall over SWC. To reveal the relationships of the SST anomalies and the rainfall anomalies over SWC, we use Niño-3.4 to measure ENSO and SST averaged in 15°S–0°, 95°–120°E to represent the variability of SST in the tropical southeastern Indian Ocean (SEIOSST). Correlations between rainfall over SWC and the SST indices are shown in Fig. 10. During ENSO decaying summers, rainfall increases over most of SWC except Yunnan (Fig. 10a) because it is associated with enhanced convection over the WMC but suppressed convection over the PS during its decaying phase (Figs. 9c,d), consistent with the composite rainfall anomalies for the years when significant convection anomalies over the PS are accompanied by opposite convection anomalies over the WMC (Fig. 8d). The summer warm SST anomaly in the tropical southeastern Indian Ocean is accompanied by a north–south dipole pattern of rainfall anomalies over SWC, with significant correlation only over central Sichuan, because it mostly affects convection over the WMC (Figs. 9b, 10b). During ENSO developing summers, rainfall decreases significantly only over central Sichuan because it suppresses the convection over the WMC (Figs. 9b, 10c).
Correlations of JJA rainfall with (a) previous DJF Niño-3.4, (b) JJA SEIOSST (15°S–0°, 95°–120°E), and (c) JJA Niño-3.4. (d) Partial correlations of JJA rainfall with previous DJF Niño-3.4 without the effect of JJA SEIOSST, (e) partial correlation of JJA rainfall with JJA SEIOSST without the effect of previous DJF Niño-3.4, and (f) partial correlation of JJA rainfall with simultaneous Niño-3.4 without the effect of JJA SEIOSST. Data used to compute the correlations have removed nine-point Gaussian-type filtered values. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to values of 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Correlations of JJA rainfall with (a) previous DJF Niño-3.4, (b) JJA SEIOSST (15°S–0°, 95°–120°E), and (c) JJA Niño-3.4. (d) Partial correlations of JJA rainfall with previous DJF Niño-3.4 without the effect of JJA SEIOSST, (e) partial correlation of JJA rainfall with JJA SEIOSST without the effect of previous DJF Niño-3.4, and (f) partial correlation of JJA rainfall with simultaneous Niño-3.4 without the effect of JJA SEIOSST. Data used to compute the correlations have removed nine-point Gaussian-type filtered values. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to values of 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
Correlations of JJA rainfall with (a) previous DJF Niño-3.4, (b) JJA SEIOSST (15°S–0°, 95°–120°E), and (c) JJA Niño-3.4. (d) Partial correlations of JJA rainfall with previous DJF Niño-3.4 without the effect of JJA SEIOSST, (e) partial correlation of JJA rainfall with JJA SEIOSST without the effect of previous DJF Niño-3.4, and (f) partial correlation of JJA rainfall with simultaneous Niño-3.4 without the effect of JJA SEIOSST. Data used to compute the correlations have removed nine-point Gaussian-type filtered values. Correlation coefficients of 0.28, 0.42, and 0.52 correspond to values of 90%, 99%, and 99.9% confidence levels, respectively.
Citation: Journal of Hydrometeorology 18, 7; 10.1175/JHM-D-16-0292.1
As variability of SST in the tropical southeastern Indian Ocean is partly dependent on evolution of ENSO, we also compute partial correlations to illustrate independent impact of SST in different regions on rainfall over SWC. Partial correlations indicate that rainfall does not increase significantly over eastern Sichuan, but increases significantly over southeastern SWC during ENSO decaying summers after removing the effect of SST in the tropical southeastern Indian Ocean (Fig. 10b). The pattern of summer rainfall anomalies over SWC associated with simultaneous SST in the tropical southeastern Indian Ocean without ENSO during the previous winter is similar to that without ENSO during the simultaneous summer, which exhibits a north–south dipole, but significant correlations are only over central Sichuan, southern Gansu, and other small areas (Fig. 10e). The correlations of rainfall over SWC with simultaneous Niño-3.4 do not change apparently after removing the effect of SST in the tropical southeastern Indian Ocean (Fig. 10f). Overall, the relationships of the rainfall anomalies over SWC with tropical SST anomalies can be well understood by the relationships of the rainfall anomalies over SWC and the convection over the PS and the WMC.
6. Summary and discussion
The rainfall over southwest China (SWC) exhibits strong interannual variability, resulting in frequent floods and droughts. However, rainfall variability over SWC has received little attention. Previous studies reported that the convection anomalies over both the Philippine Sea (PS) and the western Maritime Continent (WMC) are accompanied by rainfall anomalies over parts of SWC on both interannual and intraseasonal time scales. In this study, we investigate the different roles of tropical convection over the PS and the WMC in interannual variability of summer rainfall over SWC and their possible links to tropical SST.
The summer rainfall over SWC has two dominant modes on interannual time scales. The first mode features increases in rainfall over most of SWC except central Sichuan and the northern edge. The second mode exhibits wet conditions in the north but dry conditions in the south. The suppressed convection over the PS can affect the first mode by inducing anomalous anticyclones over the western North Pacific and to the south of the Tibetan Plateau, which transport more water vapor to western Tibet and eastern SWC and thus favor above-normal rainfall there. However, it does not significantly affect rainfall over Yunnan. The enhanced convection over the WMC could excite an anomalous anticyclone over the western North Pacific, which shifts northward compared with that associated with the suppressed convection over the PS. This northward shift of the anomalous anticyclone leads to significant increases in rainfall over northeastern SWC but weak decreases in rainfall over southeastern SWC.
The interannual variability of the convection over the PS and the WMC has a close relationship with ENSO. The convection over the WMC is suppressed during ENSO developing summers but enhanced during its decaying summers. The convection over the PS is suppressed during ENSO decaying phase. Because of the close relationship between the convection over the PS and the WMC and the rainfall over SWC, rainfall varies in phase over most of SWC during the ENSO decaying phase when convection over the WMC and the PS tends to vary out of phase. The warm SST anomalies in the southeastern tropical Indian Ocean tend to generate a dipole pattern of rainfall anomalies over SWC when it occurs either in absence of an ENSO event during previous winter or with a La Niña event simultaneously.
The interannual variability of rainfall over SWC exhibits strong spatial inhomogeneity. Rainfall is even not significantly correlated with the PCs over some areas of SWC, especially for PC2 (Figs. 7e,f). Thus, it is not easy to find a factor that is significantly correlated with rainfall over most of SWC. A better prediction of rainfall over SWC may be achieved when two or more factors are used as predictors. The fraction of variance of rainfall explained by convection anomalies over the WMC and PS together is higher than 0.1 over most of SWC except Yunnan, with the maximum over Chongqing reaching 0.3. However, it is mostly in northeastern SWC that the fraction of variance of rainfall explained by combinations of two of the SST indices is higher than 0.1, with the maximum higher than 0.2 over central Sichuan (figures not shown). Rainfall anomalies over a large part of SWC are hard to predict only based on the tropical factors found in this study, in particular southern SWC. Thus, the interannual variability of rainfall over SWC and its predictability deserve further study.
Because the convection over the WMC during the ENSO decaying phase is also affected by simultaneous SST anomalies over the equatorial central and eastern Pacific, timing of ENSO decaying and developing should be considered when it is used as a predictor for rainfall anomalies over SWC. While winter SST anomalies in the equatorial central and eastern Pacific can be used to predict the following summer rainfall several months in advance, summer SST anomalies over both the tropical southeastern Indian Ocean and the equatorial central and eastern Pacific can be well predicted generally with a lead of only 1–2 months because of the spring predictability barrier of ENSO and complicated air–sea interactions in the tropical Indian Ocean (e.g., Jiang et al. 2013; Yang and Jiang 2014). Thus, a skillful prediction of the summer rainfall over northeastern SWC may be achieved 1–2 months in advance based on both observed SST anomalies and predicted SST anomalies by dynamical models.
Acknowledgments
This study was jointly supported by the National Natural Science Foundation of China (Grants 41661144019, 91337107, and 41375081), the operational prediction development program of CMA (CMAHX20160504), key project of basic applied research plan of Sichuan Province (2016JY0046), and the Basic Research and Operation Program of the CMA Institute of Plateau Meteorology (BROP 201514).
REFERENCES
Adler, R. F., and Coauthors, 2003: The version 2 Global Rainfall Climatology Project (GPCP) Monthly Precipitation Analysis (1979–present). J. Hydrometeor., 4, 1147–1167, doi:10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Behera, S. K., and T. Yamagata, 2003: Influence of the Indian Ocean dipole on the Southern Oscillation. J. Meteor. Soc. Japan, 81, 169–177, doi:10.2151/jmsj.81.169.
Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553–597, doi:10.1002/qj.828.
Ding, R. Q., and J. P. Li, 2012: Influences of ENSO teleconnection on the persistence of sea surface temperature in the tropical Indian Ocean. J. Climate, 25, 8177–8195, doi:10.1175/JCLI-D-11-00739.1.
Huang, R., Y. Liu, L. Wang, and L. Wang, 2012: Analyses of the causes of severe drought occurring in southwest China from the fall of 2009 to the spring of 2010 (in Chinese). Chin. J. Atmos. Sci., 36, 443–457, doi:10.3878/j.issn.1006-9895.2011.11101.
Jiang, X., and Y. Li, 2011: Spatio-temporal variability of winter temperature and precipitation in southwestern China and possible causes. J. Geogr. Sci., 21, 250–262, doi:10.1007/s11442-011-0842-9.
Jiang, X., Y. Li, S. Yang, and R. Wu, 2011: Interannual and interdecadal variations of the South Asian and western Pacific subtropical highs and their relationships with Asian-Pacific summer climate. Meteor. Atmos. Phys., 113, 171–180, doi:10.1007/s00703-011-0146-8.
Jiang, X., S. Yang, J. Li, Y. Li, H. Hu, and Y. Lian, 2013: Variability of the Indian Ocean SST and its possible impact on summer western North Pacific anticyclone in the NCEP Climate Forecast System. Climate Dyn., 41, 2199–2212, doi:10.1007/s00382-013-1934-2.
Jiang, X., Y. Li, S. Yang, J. Shu, and G. He, 2015: Interannual variation of mid-summer heavy rainfall in the eastern edge of the Tibetan Plateau. Climate Dyn., 45, 3091–3102, doi:10.1007/s00382-015-2526-0.
Jiang, X., Y. Li, S. Yang, K. Yang, and J. Chen, 2016: Interannual variation of summer atmospheric heat source over the Tibetan Plateau and the role of convection around the western Maritime Continent. J. Climate, 29, 121–138, doi:10.1175/JCLI-D-15-0181.1.
Kumar, A., M. Chen, and W. Wang, 2013: Understanding prediction skill of seasonal mean precipitation over the tropics. J. Climate, 26, 5674–5681, doi:10.1175/JCLI-D-12-00731.1.
Li, Y., H. Xu, and D. Liu, 2011: Features of the extremely severe drought in the east of southwest China and anomalies of atmospheric circulation in summer 2006. Acta Meteor. Sin., 25, 176–187, doi:10.1007/s13351-011-0025-8.
Li, Y., B. Xiang, C. Lu, J. Ju, and N. Wang, 2016: Impact of Madden–Julian oscillation activities on precipitation in summer over the east of southwest China and its possible mechanism (in Chinese). Chin. J. Atmos. Sci., 40, 437–450, doi:10.3878/j.issn.1006-9895.1506.15134.
Li, Z., J. Chen, and X. Dong, 2014: Comparison analysis for summer heavy drought and its circulation features in 2011/2006 in Chongqing (in Chinese). J. Southwest Univ. Nat. Sci. Edit., 36, 1–10.
Lu, R., 2002: Indices of the summertime western North Pacific subtropical high. Adv. Atmos. Sci., 19, 1004–1028, doi:10.1007/s00376-002-0061-5.
Nitta, T., 1987: Convective activities in the tropical western Pacific and their impacts on the Northern Hemisphere summer circulation. J. Meteor. Soc. Japan, 65, 373–390, doi:10.2151/jmsj1965.65.3_373.
Nitta, T., and Z. Z. Hu, 1996: Summer climate variability in China and its association with 500 hPa height and tropical convection. J. Meteor. Soc. Japan, 74, 425–445, doi:10.2151/jmsj1965.74.4_425.
North, G., T. L. Bell, and R. F. Cahalan, 1982: Sampling errors in the estimation of empirical orthogonal functions. Mon. Wea. Rev., 110, 699–706, doi:10.1175/1520-0493(1982)110<0699:SEITEO>2.0.CO;2.
Rayner, N. A., and Coauthors, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, doi:10.1029/2002JD002670.
Wang, B., R. G. Wu, and X. H. Fu, 2000: Pacific–East Asian teleconnection: How does ENSO affect East Asian climate? J. Climate, 13, 1517–1536, doi:10.1175/1520-0442(2000)013<1517:PEATHD>2.0.CO;2.
Wang, B., Z. Wu, J. Li, J. Liu, C. Chang, Y. Ding, and G. Wu, 2008: How to measure the strength of the East Asia summer monsoon. J. Climate, 21, 4449–4463, doi:10.1175/2008JCLI2183.1.
Wang, L., W. Chen, W. Zhou, and G. Huang, 2015a: Drought in southwest China: A review. Atmos. Ocean. Sci. Lett., 8, 339–344, doi:10.3878/AOSL20150043.
Wang, L., W. Chen, W. Zhou, and G. Huang, 2015b: Teleconnected influence of tropical northwest Pacific sea surface temperature on interannual variability of autumn precipitation in southwest China. Climate Dyn., 45, 2527–2539, doi:10.1007/s00382-015-2490-8.
Wang, L., M. Ting, D. Chapman, D. E. Lee, N. Henderson, and X. Yuan, 2016: Prediction of northern summer low-frequency circulation using a high-order vector auto-regressive model. Climate Dyn., 46, 693–709, doi:10.1007/s00382-015-2607-0.
Xie, S.-P., K. Hu, J. Hafner, H. Tokinaga, Y. Du, G. Huang, and T. Sampe, 2009: Indian Ocean capacitor effect on Indo-western Pacific climate during the summer following El Niño. J. Climate, 22, 730–747, doi:10.1175/2008JCLI2544.1.
Xu, H., J. Li, J. Feng, and J. Mao, 2012: The asymmetric relationship between the winter NAO and the precipitation in southwest China (in Chinese). Acta Meteor. Sin., 70, 1276–1291, doi:10.11676/qxxb2012.107.
Yang, S., and X. Jiang, 2014: Prediction of eastern and central Pacific ENSO events and their impacts on East Asian climate by the NCEP Climate Forecast System. J. Climate, 27, 4451–4472, doi:10.1175/JCLI-D-13-00471.1.
Zhang, W., F. Jin, J. Zhao, L. Qi, and H. Ren, 2013: The possible influence of a nonconventional El Niño on the severe autumn drought of 2009 in southwest China. J. Climate, 26, 8392–8405, doi:10.1175/JCLI-D-12-00851.1.
