1. Introduction
The Asian summer monsoon (ASM) covers one of the most populated regions in the world. The precipitation it brings is a major water source for South and Southeast Asia and has a large influence on agriculture and socioeconomic activities in the region. Thus, the future behavior of the ASM under global warming is of great concern.
Many studies have investigated projected changes in the ASM due to global warming using coupled atmosphere–ocean general circulation models (CGCMs). Model experiments have projected an increase in the mean precipitation over the ASM region. The Fifth Assessment Report (AR5) of the Intergovernmental Panel on Climate Change (IPCC), which is based on the results of phase 5 of the Coupled Model Intercomparison Project (CMIP5), showed the same trend. The increase in monsoon precipitation is understood to be the result of the effect of an increase in atmospheric moisture associated with atmospheric warming, which is stronger than the effects of the long-term change in the strength of the monsoon circulation (Christensen et al. 2013). Ogata et al. (2014) reported that the CMIP5 multimodel ensemble (MME) mean projected an increase in summer seasonal precipitation and an intensification of low-level monsoon westerlies. Lee and Wang (2014) suggested that the rainy season over the ASM region might expand in the future. In terms of spatial patterns, precipitation over the current wet regions is projected to increase (i.e., wet-gets-wetter pattern; Held and Soden 2006).
Relative to the number of studies on mean climate change, few studies have examined changes in interannual variation. Real-world climate experiences upswings and downswings, and the extremes of these swings cause floods and droughts. Thus, understanding the long-term changes in interannual variation is important. Model experiments have projected an intensification of the interannual variation in summer seasonal precipitation over the Asian monsoon region under global warming. Kitoh et al. (1997) indicated an increase in interannual variations over South Asia from a single model experiment of a transient increase in CO2 concentrations. Turner and Annamalai (2012) and Lu and Fu (2010) reported the intensification of the interannual variability of summer seasonal precipitation over South Asia and East Asia, respectively, using phase 3 of the Coupled Model Intercomparison Project (CMIP3) models. Relative to the CMIP3 MME, Sperber et al. (2013) reported the improved ability of the CMIP5 MME to reproduce the ASM. Therefore, it is necessary to investigate how CMIP5 models project long-term changes in the interannual variation in ASM precipitation. Based on CMIP5 model projections, Brown et al. (2013) reported the intensification of the interannual variability of seasonal precipitation during the rainy season in the western Pacific monsoon region.
Observational analyses have shown that the interannual variation over the ASM region has regional characteristics. Wang and Fan (1999) reported a poor correlation between intensive convective activity over the Bay of Bengal and in the vicinity of the Philippines. Wang et al. (2001) found a remarkable difference in the interannual variability between the Indian summer monsoon and the western North Pacific summer monsoon in both temporal and spatial structures. In addition, the mechanisms driving interannual variability exhibit regional differences (Chen and Yoon 2000), which affect the long-term changes in interannual variability. These regional differences in the interannual variation among monsoon subsystems may lead to different trends in the fluctuations of interannual variability. Considering that many studies have investigated long-term changes in the interannual variation in summer seasonal precipitation based on specific regional averages, investigating the detailed spatial patterns over the ASM may offer a better understanding of the long-term changes in interannual variation.
Moreover, while intensification of the interannual variation is projected in individual regions of the ASM region based on previous studies, it is not known how the wet and dry extremes of interannual variation will behave. Indices, such as interannual standard deviation, can be used to evaluate the interannual variation to understand changes in intensity, but cannot individually describe the changes in wet and dry extremes. Investigating detailed spatial patterns may reveal an asymmetric nature in the long-term changes in wet and dry seasonal extremes.
This study investigated future changes in the fluctuations of interannual variation in ASM precipitation due to global warming, particularly over the tropical ASM region. We determined whether the interannual variation in ASM precipitation will amplify or decay in the future. In addition, we investigated the long-term changes in extreme wet and dry years. The data and methods used in this study are described in section 2. Sections 3 and 4 confirm the performance of the CMIP5 models and the long-term changes in the mean climate field, respectively. Sections 5 and 6 present the results for the projected long-term trends in the fluctuation of the interannual variation in summer seasonal precipitation and long-term change in its extremes, respectively. Section 7 discusses possible factors responsible for the results. Finally, the conclusions are presented in section 8.
2. Data and assessment methods
a. Model and observational data
The analyses in this study are based on the outputs from 22 CGCMs that participated in the CMIP5 (listed in Table 1). We analyzed the historical run to confirm the performance of the CMIP5 models and analyzed the representative concentration pathway 4.5 (RCP4.5) run to investigate projections for future climate under global warming. The RCP4.5 run is a simulation based on a scenario in which radiative forcing stabilizes at 4.5 W m−2 in 2100 (Taylor et al. 2009). Details of the CMIP5 models and the experiments were documented by Taylor et al. (2009, 2012), Collins et al. (2013), and Flato et al. (2013). Monthly mean precipitation and 850-hPa wind data were calculated for the June–August (JJA) average and transformed to a 5.0° resolution for the periods 1979–2005 and 2007–2100 for the historical and RCP4.5 runs, respectively. We compared the beginning and end of the RCP4.5 run to investigate the long-term changes. Comparing the end of the historical run and the end of RCP runs is common in investigations of future changes for analyses based on CMIP5 experiments. However, such comparisons are not appropriate when focusing on interannual variation because of differences in the experimental designs. For example, the treatment of volcanic eruptions differs between the two runs, and large volcanic eruptions influence the climate. The results of the historical run reflect large volcanic eruptions such as El Chichón and Mount Pinatubo (Flato et al. 2013). However, no volcanic eruptions are simulated in the RCP4.5 scenario. Thus, we used the two periods in the RCP4.5 run for a fairer comparison.
List of CMIP5 models used in this study. The monthly mean data were calculated as the JJA average and transformed into 5.0° resolutions. One ensemble member (r1i1p1) was used from each model. Details of each CMIP5 model are described in Flato et al. (2013, their Table 9.A.1).
In addition, we used observational data to validate the performance of the CMIP5 models for the mean state and interannual variation in summer seasonal precipitation and atmospheric circulations. Precipitation data were obtained from the Climate Prediction Center Merged Analysis of Precipitation (CMAP; Xie and Arkin 1997) data and Global Precipitation Climatology Project (GPCP; Adler et al. 2003) data, and 850-hPa atmospheric circulation data were obtained from the Japanese 55-year Reanalysis (JRA-55; Kobayashi et al. 2015). All of the observational data had 2.5° resolutions and were used to calculate the JJA average for the period 1979–2005.
b. Spatial patterns of the interannual variation in summer seasonal precipitation
Because we were interested in the interannual variation in summer seasonal precipitation, we conducted a composite analysis of the anomalies of precipitation and 850-hPa winds to identify the spatial patterns of the interannual variation in JJA precipitation and atmospheric circulations. Spatial patterns of interannual variation can be found from the composite difference between several wet and dry years. From each CMIP5 model, we selected the five wettest and five driest years based on the summer seasonal precipitation over specific regions such as South Asia (10°–25°N, 70°–100°E), the Indochina Peninsula (10°–20°N, 90°–110°E), and the vicinity of the Philippines (10°–20°N, 115°–140°E) during the 25-yr period from 1981 to 2005. These three areas were chosen because their regional characteristics in terms of the spatial patterns of interannual variation were investigated in the previous studies mentioned in section 1. The five wettest (driest) years for each model were classified as the wet (dry) group. To evaluate the statistical robustness of the composite spatial pattern of the interannual variation, we performed a Welch’s t test on the wet and dry groups at the 95% significance level. We conducted the same composite analysis using observational data, but at the 90% significance level for the Welch’s t test.
c. Long-term trends in the fluctuation of the interannual variation in summer seasonal precipitation
To examine whether the fluctuation of the interannual variation in ASM precipitation is projected to increase or decrease because of global warming, we analyzed the long-term trends in the fluctuation of the interannual variation in JJA precipitation during the RCP4.5 run for the period 2007–2100, based on the study of Kitoh et al. (1997). We used the standard deviation as an index to identify trends in the fluctuation of the interannual variation in ASM precipitation. We also performed the same analysis based on the coefficient of variation (CV), which is a measure of relative variability that is calculated by dividing the standard deviation by the mean (see appendix A).
We evaluated long-term changes in the fluctuation of the interannual variation using the following three steps for each CMIP5 model at each grid. First, we calculated the time series of the interannual standard deviation of JJA precipitation in running 21-yr windows. Second, we performed a Mann–Kendall test for the time series of the standard deviations to examine whether these time series exhibited increasing or decreasing trends. It should be noted that an increasing (decreasing) trend in the standard deviation corresponds to an increasing (decreasing) trend in the fluctuation of the interannual variation in summer seasonal precipitation. In addition, we estimated the degree of freedom for the standard deviation time series due to the dependence of the standard deviations. We estimated the effective number of samples and degree of freedom using the method of Matsuyama and Katasakai (2012) for the Mann–Kendall test (a detailed description is given in appendix B). Finally, we counted the number of CMIP5 models that projected an increasing trend in the standard deviation in each grid.
d. Changes in extremes of summer seasonal precipitation anomalies
We investigated the long-term changes in the wet and dry extremes of JJA precipitation anomalies and evaluated the wet and dry extremes using the following simple method. First, we sorted the JJA precipitation anomalies in descending order for each CMIP5 model at each grid for the two 25-yr periods 2007–31 and 2076–2100. Second, we compared the two 25-yr periods in the wettest and driest JJA precipitation anomalies and counted the number of CMIP5 models that projected JJA precipitation anomalies, which was larger for the period 2076–2100 than for the period 2007–31 at each grid, respectively. We also investigated for the second wettest and driest JJA precipitation anomalies.
3. CMIP5 model performance for ASM precipitation
We evaluated the performance of the CMIP5 models during the historical run to confirm their performance for ASM precipitation in terms of climatological mean states and interannual variation. Figure 1 shows the JJA climatology of precipitation and 850-hPa winds for the observational data and CMIP5 MME. With regard to the ASM region, the CMIP5 MME reproduced the position of the precipitation peaks such as the area west of India, the Indochina Peninsula, and the Philippines. Precipitation over the mei-yu–baiu front in East Asia and the Philippine Sea in the western North Pacific was underestimated, whereas precipitation over the Maritime Continent in the Southern Hemisphere was overestimated. These features were consistent with the results of Sperber et al. (2013). Regarding the 850-hPa winds, the CMIP5 MME was able to reproduce the observational field. The monsoon circulation, characterized by the flow of easterlies over the equatorial south Indian Ocean that continues to the Somali jet and monsoon westerlies over South Asia joining with easterlies from the equatorial Pacific and travels northward along the rim of the Pacific high, was realistically simulated.
JJA climatology of precipitation (mm day−1) and 850-hPa winds (m s−1) from the (a) observational data and (b) CMIP5 MME during 1979–2005. Areas in which the JJA precipitation is lower than one-third of the global mean are shaded gray.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
Furthermore, we evaluated the performance of the CMIP5 models for the interannual variation in summer seasonal precipitation. Figure 2 shows the interannual standard deviation of JJA precipitation during the period 1979–2005 for the observational data and CMIP5 MME. The basic distribution of the standard deviation resembled that of the climatological mean precipitation (Fig. 1). The standard deviation peaks and mean JJA precipitation corresponded over South and Southeast Asia. Although the values in the CMIP5 MME were generally smaller than those of the observational data, the CMIP5 MME was able to reproduce the basic spatial pattern of the standard deviation. However, in the western Pacific, the CMIP5 MME showed a single peak over the equator, while the observational data showed two peaks north of and on the equator (Fig. 2).
Standard deviation for JJA precipitation (mm day−1) from the (a) observational data and (b) CMIP5 MME during 1979–2005.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
The spatial patterns of the interannual variation in summer seasonal precipitation and accompanying low-level atmospheric circulation of the CMIP5 MME were evaluated with a composite analysis. The significant differences in precipitation and 850-hPa winds between several wet and dry years for the observational data and the CMIP5 MME are shown in Figs. 3 and 4, respectively, which is based on the JJA precipitation over South Asia, the Indochina Peninsula, and the vicinity of the Philippines, respectively (see section 2b).
Difference between the wet and dry groups (wet minus dry) for JJA precipitation (mm day−1) and 850-hPa winds (m s−1) from the observational data for the period 1981–2005. The grouping is based on the JJA precipitation over (a) South Asia (10°–25°N, 70°–100°E), (b) the Indochina Peninsula (10°–20°N, 90°–110°E), and (c) the vicinity of the Philippines (10°–20°N, 115°–140°E). Ten and five samples were included in each group for precipitation and 850-hPa winds, respectively. The results shown are significant at the 90% significance level according to a Welch’s t test. The contours represent the streamfunction (×1.0 × 106 m2 s−1).
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
As in Fig. 3, but for CMIP5 MME. There were 110 samples in each group for both precipitation and 850-hPa winds. The results shown are significant at the 95% significance level according to a Welch’s t test.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
The observational data showed that the spatial pattern of the interannual variation in summer seasonal precipitation based on South Asia had intense positive signals for precipitation on the west coast of the Indian subcontinent and the Bay of Bengal, accompanying anomalous westerlies over the southern Indian subcontinent (Fig. 3a). There were anomalous negative signals for precipitation over the equatorial eastern Indian Ocean and positive signals in the equatorial western Indian Ocean. Anomalous monsoon westerlies were observed from the Arabian Sea to the Bay of Bengal. These features were consistent with the spatial patterns of the wet phase of the Indian summer monsoon described by Wang et al. (2001). The CMIP5 MME was able to reproduce the anomalous westerlies over the southern Indian subcontinent (Fig. 4a). However, compared to the observational data, positive precipitation signals appeared over the entire area-averaged region, and the opposite positive and negative signs were observed in the equatorial Indian Ocean. The CMIP5 MME also showed an anomalous strong Pacific high, while there were no clear signals over this area in the observational data.
Regarding the spatial pattern of the interannual variation in JJA precipitation based on the Indochina Peninsula, a zonal band of anomalous positive precipitation appeared from the Bay of Bengal to the equatorial western North Pacific, which corresponded with the monsoon trough, accompanied by an anomalous cyclonic circulation (Fig. 3b). Two peaks in anomalous cyclonic circulation were observed over the Bay of Bengal and the South China Sea. The interannual variation of the summer seasonal precipitation over this area is related to westward-propagating tropical cyclones (TCs) and tropical disturbances from the South China Sea and the western North Pacific (Chen and Weng 1999; Chen and Yoon 2000; Takahashi et al. 2015). Considering that the monsoon trough is generally associated with frequent tracks of westward-propagating disturbances, including TCs, this monsoon trough-like spatial pattern is consistent with Takahashi et al. (2015). The CMIP5 MME was able to reproduce the zonal band of anomalous positive precipitation with cyclonic anomalies, which was wedged between anomalous negative signals in the meridional direction (Fig. 4b). Compared to the observational results, the anomalous negative signal in the south of the zonal band showed a greater extension into the equatorial Indian Ocean and the Middle East.
The spatial pattern of interannual variation based on precipitation in the vicinity of the Philippines (Fig. 3c) was similar to that based on precipitation over the Indochina Peninsula (Fig. 3b). A positive precipitation anomaly zonal band extended from the Bay of Bengal to the western North Pacific in the area of between latitudes 10° and 25°N, accompanied by an anomalous cyclonic circulation. Precipitation anomalies were negative in the vicinity of Japan (i.e., Pacific–Japan pattern) and the area from the Arabian Sea to Borneo. These features were consistent with the spatial patterns of the strong western North Pacific summer monsoon described by Wang et al. (2001). The CMIP5 MME reproduced the zonal band of positive precipitation anomaly with anomalous cyclonic circulation (Fig. 4c).
The results of the composite analyses show that the CMIP5 MME can reproduce the spatial pattern of the interannual variation in JJA precipitation and indicate that the physical phenomena that influence the interannual variation in the ASM precipitation can be reproduced to some degree. Among the three spatial patterns of interannual variation, CMIP5 MME tended to have a negative bias to the west of the anomalous positive area for precipitation and a weak atmospheric circulation compared to the observation-based results.
4. Long-term changes in the mean climate field in the twenty-first century
We confirmed the long-term changes in the mean states of precipitation and low-level circulations. Figure 5 shows the long-term changes in the JJA climatology of precipitation and 850-hPa winds for the period 2076–2100 relative to 2007–31 in the RCP4.5 run. The CMIP5 MME showed an increase in mean precipitation over the ASM region and equatorial Pacific region, particularly from the monsoon trough to the intertropical convergence zone. For low-level circulations, enhancement of the monsoon westerlies was projected. Meanwhile, anomalous westerlies in the west and anomalous easterlies in the east of the Maritime Continent indicated that the Walker circulation was projected to weaken, which has been suggested in previous studies (e.g., Vecchi et al. 2006).
Projected changes in JJA climatology for precipitation (mm day−1) and 850-hPa winds (m s−1) for the period 2076–2100 relative to 2007–31 in the RCP4.5 run from the CMIP5 MME. The results shown are significant at the 95% significance level according to a Welch’s t test. For each model, the climatology for both periods (2007–31 and 2076–2100) was transformed to anomalies relative to the entire RCP4.5 run (2007–2100) before we performed the Welch’s t test.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
Moreover, we investigated the characteristics of the long-term changes in the mean JJA precipitation related to the mean JJA precipitation over the ASM region (0°–50°N, 60°–150°E). A positive relationship was apparent between the long-term changes in the mean JJA precipitation and the mean JJA precipitation over the ASM region (Fig. 6). This relationship was consistent with Held and Soden (2006), which reported a wet-gets-wetter pattern in terms of the long-term changes in the projected spatial pattern of the mean precipitation.
Long-term changes in the mean JJA precipitation (mm day−1) for the period 2076–2100 relative to 2007–31 in the RCP4.5 run from the CMIP5 MME as a function of the JJA climatology for precipitation (mm day−1) during 2007–31 over the ASM region (0°–50°N, 60°–150°E). Areas in which JJA precipitation is lower than one-third of the global mean are excluded.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
5. Long-term trends in the fluctuation of the interannual variation in summer seasonal precipitation
Sections 5 and 6 focus on the projected long-term changes in the fluctuation of the interannual variation in summer seasonal precipitation under global warming. We performed the Mann–Kendall test on the interannual standard deviation of JJA precipitation for each CMIP5 model to examine whether the standard deviation showed an increasing or decreasing trend during the RCP4.5 run (see section 2c). The area where the CMIP5 models projected an increasing (or decreasing) trend for the standard deviation corresponded with the area in which the fluctuation of the interannual variation in the JJA precipitation was projected to amplify (or decay). Figure 7a shows the number of CMIP5 models that projected an increasing standard deviation trend at each grid. Figures 7b and 7c show the number of CMIP5 models that projected statistically significant increasing and decreasing standard deviation trends, respectively. The spatial patterns of the signals in Figs. 7b and 7c were similar to that in Fig. 7a. Thus, we considered the results of Figs. 7b and 7c to be reliable signals, although the maximum number of CMIP5 models that projected statistical significance was less than the majority of the 22 CMIP5 models.
Number of CMIP5 models that projected (a) an increasing trend, (b) a statistically significant increasing trend, and (c) a statistically significant decreasing trend for the running 21-yr standard deviation. Low precipitation grids, where the JJA precipitation of CMIP5 MME was lower than one-third of the global mean, are shaded gray. The trends are significant at the 95% level in (b) and (c), and areas where fewer than five CMIP5 models showed a significant trend are shaded white.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
The CMIP5 models projected an increasing trend for the standard deviation in most of the ASM region (Fig. 7b). Overall, an increasing trend for the standard deviation was projected in 53.6% of the region, whereas a decreasing trend was projected in 10.9% of the region (Figs. 7b and 7c, respectively). The results based on the CV also showed an increasing trend in the index projected over the ASM region (see appendix A). These results indicate that fluctuation of the interannual variation in summer seasonal precipitation is projected to amplify under global warming in the ASM region.
The CMIP5 MME projected an increasing trend for the standard deviation in most parts of the ASM region including areas such as South and Southeast Asia, the Maritime Continent, and the east coast of the Eurasian continent (Fig. 7b). In contrast, there were few land areas where a decreasing trend for the standard deviation was projected, with more signals recognized over ocean areas such as the Indian Ocean and the South China Sea (Fig. 7c).
6. Changes in the extremes of summer seasonal precipitation anomalies
To investigate the specific behavior of long-term changes in the fluctuation of interannual variation in seasonal precipitation, we analyzed the changes in wet and dry JJA extremes, which can show whether long-term changes in interannual variation fluctuations are associated with an increase in extremely wet summers, extremely dry summers, or both. We compared the first and second wettest and driest JJA precipitation anomalies for the periods 2007–31 and 2076–2100 at each grid (see section 2d).
The majority of CMIP5 models projected that the wet extremes of the JJA precipitation anomalies expanded over most of the ASM region during the RCP4.5 run (Fig. 8a). These expansion signals appeared evenly over the entire Asian monsoon region including the Indian subcontinent, Indochina Peninsula, Maritime Continent, and mei-yu–baiu rainband area. The results indicate that these areas are projected to experience more severe floods (more anomalously increased seasonal precipitation) in extremely wet years under global warming.
Number of CMIP5 models that projected the JJA precipitation anomalies for the period 2076–2100 were larger than that for the period 2007–31 for the first and second wettest and driest JJA. The JJA precipitation anomalies were individually sorted at each grid for each 25-yr period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
The dry extremes of the JJA precipitation were also projected to expand over the ASM region (Fig. 8b). These expansion signals were stronger and more organized than those in the wet extremes. This spatial pattern was somewhat similar to that of the changes in mean precipitation (Fig. 5). The expansion signals particularly appeared as a band extending from India to the equatorial western North Pacific, which was consistent with the monsoon trough. An expansion signal was also observed to the east of the midlatitude Eurasian continent. These signals indicate that these areas will experience more severe droughts (more anomalously dry conditions) in extremely dry years under global warming. The second wettest and driest JJA anomalies also showed expansion signals over the ASM region in both wet and dry anomalies, with the spatial patterns of these anomalies being similar to those of the wettest and driest cases, respectively (Figs. 8c,d). Possible mechanisms for these trends are discussed in section 7.
Areas where the majority of the CMIP5 models projected both wet and dry extremes of JJA precipitation anomalies to increase in the latter 25-yr period are likely to experience an increase in interannual variation fluctuations of JJA precipitation. In particular, areas such as South Asia and the Maritime Continent fall under this trend. By contrast, areas such as the south Indian Ocean and the Middle East showed opposite signals, meaning that the fluctuation of the interannual variation in JJA precipitation in these areas is projected to decrease. These results reinforce the conclusion of the former analysis that the fluctuation of the interannual variation in summer seasonal precipitation over the ASM region will increase under global warming.
The asymmetric nature of the spatial distribution of the long-term changes between the wet and dry extremes indicates that specific projected changes in interannual variation differ regionally in the ASM region (Fig. 8). The finding that the spatial structure of the dry extremes was more organized than that of the wet extremes implies that the changes in dry extremes may be more robust and that these areas may experience more severe droughts in the future. However, the broad expansion signals of the wet extremes over the ASM region may be related to the intermodel diversity of the reproducibility of the rainy areas and its projected long-term changes.
Moreover, we investigated whether long-term changes in the seasonal wet and dry extremes exhibited any characteristics in their projected spatial patterns such as a wet-gets-wetter pattern was recognized in terms of projected spatial patterns of the mean JJA precipitation over the ASM region (Fig. 6). We examined whether the long-term changes in the seasonal wet and dry extremes are related to the mean seasonal precipitation or its long-term changes. Although the results for the mean seasonal precipitation and its long-term changes were similar, the latter results were more robust. Thus, we would show the relationship with the long-term changes in the seasonal mean precipitation. Figures 9a–d are scatterplots of the number of CMIP5 models, which projected that the JJA precipitation anomalies for the period 2076–2100 were larger than those for the period 2007–31, compared to the long-term changes in the mean JJA precipitation over the ASM region (0°–50°N, 60°–150°E) in the wet and dry extremes, respectively. We performed a Mann–Kendall test for the median in each scatterplot to investigate the statistical relationship between the extremes and the changes in the mean precipitation.
Number of CMIP5 models that projected the JJA precipitation anomalies for the period 2076–2100 were larger than that for the period 2007–31 in the first and second wettest and driest JJA anomalies as a function of the long-term changes in the mean JJA precipitation (mm day−1) over the ASM region (0°–50°N, 60°–150°E). Areas in which JJA precipitation was lower than one-third of the global mean were excluded. The plus signs represent the value at each grid. The squares and triangles represent the median and mean for each number of CMIP5 models, respectively. The shaded areas show the interquartile range.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
If this region had a tendency for wet and dry extremes to be expanded in places where the mean precipitation increased, the scatterplots in Fig. 9 should show a positive relationship for the wet extreme and a negative relationship for the dry extreme. However, Fig. 9 shows that the relationship with the long-term changes in the mean JJA precipitation differed between the wet and dry extremes for the first and second wettest and driest JJA. The scatterplot of the long-term changes in the seasonal precipitation extremes and that of the mean precipitation showed no clear relationship for the wet extreme but a notable relationship for the dry extreme. For the dry extremes, the spatial pattern of areas the CMIP5 models projected to experience more severe droughts in the dry extremes corresponded to that of the long-term changes in the mean seasonal precipitation over the ASM region. The median in the driest JJA showed a significant negative relationship with the long-term changes in JJA precipitation at the 95% significance level. The second driest JJA also showed a negative relationship but at the 90% significance level. This result indicates more CMIP5 models agreed that dry extremes increased in areas where JJA precipitation was projected to get wetter (Fig. 9b). Conversely, the value of long-term changes in JJA precipitation was roughly constant among the CMIP5 models that projected an increase in the wet extremes (Fig. 9a). As we mentioned at the end of section 4, the wet-gets-wetter pattern can be obtained over the ASM region. Although we can obtain a similar tendency by using the mean JJA precipitation, a statistically robust relationship is shown by using the changes in JJA precipitation.
7. Discussion: Why do CMIP5 models project amplification in fluctuation of the interannual variation in ASM precipitation?
The results from sections 5 and 6 showed that CMIP5 models projected amplification in the interannual variation in summer seasonal precipitation over the Asian monsoon region under global warming. To investigate the cause of the increase in the fluctuation of the interannual variation in summer seasonal precipitation, we analyzed the long-term changes in its spatial patterns by conducting composite analyses for the beginning and end of the RCP4.5 run and investigated the long-term changes in the spatial patterns of the interannual variation in ASM precipitation. This method was the same as that used for the historical run (see section 2b). Considering the results of sections 3 and 5, we focused on the precipitation over the Indochina Peninsula, the area in which the CMIP5 MME was able to reproduce the climatology and interannual variation in JJA precipitation and where the fluctuation of the interannual variation in summer seasonal precipitation was projected to increase during the RCP4.5 run. The spatial pattern of the interannual variation in JJA precipitation showed a spatial representativeness (Fig. 3b). Figure 10 shows the differences in JJA precipitation and 850-hPa winds between the wet and dry groups (wet minus dry), which were chosen based on precipitation over the Indochina Peninsula (10°–20°N, 90°–110°E), for the periods 2007–31 and 2076–2100.
As in Fig. 4b, but for the periods (a) 2007–31 and (b) 2076–2100 in the RCP4.5 run.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
The basic features of the spatial patterns projected by CMIP5 MME during the two periods were generally the same as those of the historical run (Fig. 4b), but the contrast in the spatial pattern of the interannual variation was stronger in 2076–2100 than that in 2007–31. Accompanying an anomalous cyclonic circulation, a band of anomalous positive precipitation that extended from the Bay of Bengal to the equatorial western North Pacific appeared in both periods. The similarity to the historical run indicates that there is no dramatic change in the spatial pattern of the interannual variation in JJA precipitation based on the Indochina Peninsula, implying that the mechanism controlling the interannual variation during each period is the same. The strengthening in contrast to the spatial pattern during the RCP4.5 run indicates that the long-term changes in frequency and/or intensity of the physical phenomena that influence the interannual variation in ASM precipitation drive the increase in fluctuation of the interannual variation in ASM precipitation. As mentioned in section 3, the spatial pattern of the interannual variation in summer seasonal precipitation over the Indochina Peninsula is associated with the monsoon trough (Fig. 3b), and the interannual variation in summer seasonal precipitation over the Indochina Peninsula is controlled by TCs and tropical disturbances that propagate the monsoon trough westward (Takahashi et al. 2015). Thus, with respect to the Indochina Peninsula, the long-term changes in frequency and/or intensity of these westward-propagating TCs and tropical disturbances may contribute to the increase in the fluctuation of the interannual variation in summer seasonal precipitation.
However, it is reported that the intensity of simulated TCs is generally weak in CMIP5 models (Flato et al. 2013). We were neither able to directly evaluate the CMIP5 models’ performance nor show their long-term changes in the westward-propagating TCs and tropical disturbances because we used monthly data outputs. As a substitute, we implied the relationship between the differences in the seasonal streamfunction and activity level of these phenomena by examining perturbation kinetic energy (PKE; see appendix C). Observational data results showed that the activity of westward-propagating tropical disturbances was relatively high (low) during the wet (dry) year over the Indochina Peninsula. This contrast corresponded with the difference of the seasonal mean streamfunction along the monsoon trough between the wet and dry years (Figs. C1, C2; see appendix C). We additionally analyzed the daily data outputs of two CMIP5 models (MIROC5 and MPI-ESM-MR) and confirmed this contrast in the activity of westward-propagating tropical disturbances, and its relationship with the seasonal mean streamfunction were recognized (Figs. C3, C4; see appendix C). From these results, we considered that the CMIP5 models were able to reproduce the difference in the activity level of westward-propagating TCs and tropical disturbances between the wet and dry years over the Indochina Peninsula, and the change in the seasonal mean streamfunction may indicate the related changes in the activity level of tropical disturbances.
Hence, the intensification of the streamfunction presented in Fig. 10 may be related to long-term changes in the frequency and/or intensity of TCs and tropical disturbances. This suggests that long-term changes in westward-propagating TCs and tropical disturbances could have a role in the increase in fluctuation of the interannual variation in summer seasonal precipitation. Regarding the projection of TCs under global warming, Christensen et al. (2013) reported a projected decrease in frequency and increase in intensity; however, uncertainties remain. Because we could not determine whether the frequency or intensity was a more important driver of the increase in interannual variation fluctuation of ASM precipitation from our results, we could not determine if our results were in agreement with previous studies. This issue will be investigated in future studies.
There are other possible factors that may increase the fluctuation of interannual variation in summer seasonal precipitation over the ASM region. An increase in saturation water vapor pressure due to global warming is a possible direct or indirect driver of the long-term changes in the interannual variation in ASM precipitation. An increase in saturation water vapor pressure due to global warming, which is derived from the Clausius–Clapeyron relationship under the condition of unchanged relative humidity, can result in an increase in hourly and daily precipitation (e.g., Meehl et al. 2007). In addition, an increase in intense individual storms and decrease in weak storms is projected at daily to weekly scales in a warmer future (Seneviratne et al. 2012). More investigations are required to understand whether the increase in short-term extreme precipitation influences seasonal precipitation and its interannual variability. Moreover, physical phenomena that influence the interannual variation (TCs and tropical disturbances in the case of the Indochina Peninsula) may also be affected by increased atmospheric moisture and amplify the interannual variation as a result.
Not only an increase in water vapor but changes in dynamic components can also affect the interannual variation of the precipitation, particularly on regional scales. It is reported that the effect of the increase in the atmospheric moisture is superior to that of the weakening of the monsoon circulation, increasing the mean precipitation over the monsoon regions (Christensen et al. 2013). Regarding extreme daily events, the influence of the increase in the atmospheric moisture remains important for the increase in its frequency, but the changes in local vertical winds also contribute to the increase in the intensity of extreme daily events (Freychet et al. 2015). Changes in the monsoon circulation lead to changes in monsoon intensity, area, and timing of the monsoon (Christensen et al. 2013), thus, investigating the influence of the spatially and temporally local variations of the atmospheric circulation is also necessary for further understanding.
Another possible factor is El Niño–Southern Oscillation (ENSO). ENSO is the dominant climate phenomenon that affects climate worldwide and is projected to remain as the dominant mode of interannual variability in the future (Christensen et al. 2013). Cai et al. (2015) reported a projected increase in the frequency of extreme El Niño and La Niña events under global warming, potentially supporting our results. However, intermodel variances remain very large. Moreover, the relationship between interannual variation in the ASM and tropical Pacific sea surface temperature has been changing (e.g., Goswami et al. 1999; Torrence and Webster 1999). Therefore, based on our current understanding, it is difficult to confirm the exact influence of ENSO on the projected amplification of interannual variation fluctuations in summer seasonal precipitation over the Asian monsoon region.
The interannual variation over the ASM region has regional characteristics (Wang and Fan 1999; Chen and Yoon 2000; Wang et al. 2001). Therefore, the dominant factor for the increase in the interannual variation of the ASM precipitation may also have regional characteristics. Moreover, the result that the spatial patterns of the long-term changes in the wet and dry extremes were asymmetric (Fig. 8) indicates that the dominant factor for the expansion in the seasonal extremes may differ between the wet and dry sides.
8. Conclusions
This study analyzed 22 CMIP5 models to investigate long-term changes in the fluctuation of the interannual variation in ASM precipitation and seasonal extremes. The CMIP5 models projected an increase in fluctuation of the interannual variation in ASM precipitation in the RCP4.5 run from the Mann–Kendall test, revealing an increasing trend in the interannual standard deviation of JJA precipitation over the ASM region. The similarity of the spatial patterns of the interannual variation in summer seasonal precipitation in the RCP4.5 suggests that the long-term changes in the physical phenomena that control the interannual variation in summer seasonal precipitation influence the interannual variation in the fluctuation of summer seasonal precipitation. However, other possible factors may contribute to the increase in the fluctuation, and further studies are required to confirm this proposed mechanism.
The spatial distribution of the long-term changes in seasonal precipitation anomalies differed between the wet and dry extremes. While CMIP5 models projected an increase in the wet extreme over the entire ASM region, they showed a stronger agreement for increases in the dry extreme from India to the equatorial western North Pacific. These strong dry signals may be indicative that more severe droughts can be expected in the future. The asymmetric nature of the long-term changes between the wet and dry extremes implies that although intensification of the interannual variation in summer seasonal precipitation is projected over the whole ASM region, the details of the altered distribution of the summer seasonal distribution differ among areas. Some areas may experience more severe droughts than floods, while other areas may experience more floods.
The areas that showed a strong agreement with an increase in the dry extremes among the CMIP5 models corresponded with areas in which the mean JJA precipitation was projected to increase. Conversely, this trend was not apparent for the wet extremes. These results indicate that, although the spatial pattern of long-term changes in the total precipitation is projected to exhibit a wet-gets-wetter pattern (Held and Soden 2006), the spatial patterns of long-term changes in the seasonal extremes show different trends in the wet and dry extremes and do not simply follow a single pattern.
Acknowledgments
We thank Professor Hiroshi Matsuyama for his advice regarding the statistics and the anonymous reviewers for their helpful comments and suggestions that improved our paper. We acknowledge the World Climate Research Programme’s Working Group on Coupled Modelling (WCRP/WGCM), which is responsible for the CMIP, and the climate modeling groups (listed in Table 1) for producing and making their model outputs available. For CMIP, the U.S. Department of Energy’s PCMDI provides coordinating support, which has led to the development of software infrastructure in partnership with the Global Organization for Earth System Science Portals. CMAP and GPCP precipitation data and JRA-55 datasets are provided by the NOAA/OAR/ESRL PSD and by the Japan Meteorological Agency (JMA), respectively. We thank Tomoko Motokado for downloading the CMIP5 datasets. This research was partly supported by the Green Network of Excellence (GRENE) program of the Ministry of Education, Culture, Sports, Science and Technology (MEXT), the 8th Japan Aerospace Exploration Agency (JAXA) Precipitation Measuring Mission (PMM) project number 309, and JSPS KAKENHI Grant JP16K16349.
APPENDIX A
Results of the Mann–Kendall Test Using the CV as an Index
In section 5, we performed the Mann–Kendall test using the standard deviation as an index to investigate the increasing and decreasing trends in the fluctuation of the interannual variation in ASM precipitation. To support this analysis, we performed the Mann–Kendall test using the CV as an index (Fig. A1). The CV is a measure of relative variability that is calculated by dividing the standard deviation by the mean and is therefore a stricter index to project an increasing trend where an increase in the mean precipitation is projected. Using both the standard deviation and CV as an index enables us to identify robust trends in the fluctuation of the interannual variation in ASM precipitation and to avoid the considerable statistical effects of long-term changes in mean ASM precipitation. The spatial patterns of the interannual CV of JJA precipitation during the period 1979–2005 for the observational data and CMIP5 MME are shown in Fig. A2.
As in Fig. 7, but the index used in the Mann–Kendall test is the CV of JJA precipitation in running 21-yr windows, in which the interannual standard deviation was divided by the mean during the 21-yr period.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
Similar to the results shown in Fig. 7, Fig. A1 shows the number of CMIP5 models that showed increasing or decreasing trends based on the Mann–Kendall test, albeit for the CV. Although the signals for the increasing trend were weaker and the decreasing trend were stronger for CV compared to the standard deviation results, the result that more areas were projected to have an increasing trend (40.6%) than a decreasing trend (16.8%) was the same. Regional characteristics were identified. The CMIP5 MME projected an increasing trend for the CV in areas such as southern India, the western side of the Indochina Peninsula, the equatorial islands in the Maritime Continent, and the east coast of the Eurasian continent (Fig. A1b). In contrast, a decreasing trend in the CV was projected in areas such as the South China Sea, Sri Lanka, and the subtropical western North Pacific (Fig. A1c). Figures 7 and A1 show that regardless of whether the standard deviation or CV was used, the CMIP5 models in the RCP4.5 run projected an increase in fluctuation of the interannual variation in ASM precipitation.
APPENDIX B
Calculating the Effective Number of Samples for the Mann–Kendall Test
APPENDIX C
Relationship between the Seasonal Streamfunction and Activity Level of Westward-Propagating TCs and Tropical Disturbances in Observational Data and the CMIP5 Models (MIROC5 and MPI-ESM-MR)
Although we proposed a role of westward-propagating TCs and tropical disturbances for long-term changes in the interannual variation in summer seasonal precipitation over the Indochina Peninsula in section 7, we were unable to directly show the long-term changes in these phenomena because time scales of TCs and tropical disturbances are shorter than months. Therefore, we examined the interannual relationship between the activity of the westward-propagating tropical disturbances and the seasonal mean streamfunction to support the proposed mechanisms. To visualize the longitudinal movement of tropical disturbances, we drew Hovmöller diagrams of the PKE (obtained from JRA-55). Figure C1 presents Hovmöller diagrams showing the PKE averaged over 10°–20°N during summer in 1994 and 1998, years during which the Indochina Peninsula experienced extremely high and low JJA precipitation, respectively.
Hovmöller diagrams of PKE at the 850-hPa level averaged over 10°–20°N (m2 s−2; on a logarithmic scale) in the wettest (1994) and driest (1998) years in the Indochina Peninsula (10°–20°N, 90°–110°E) during 1981–2005, respectively. The PKE was obtained from JRA-55 and was defined as PKE = (u′2 + ε′2)/2, where ( )′ is the deviation from the running 21-day mean. The white dashed line shows the east coast of the Indochina Peninsula.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
Figure C1 shows the activity of westward-propagating tropical disturbances from the South China Sea and western North Pacific to the Indochina Peninsula and the Bay of Bengal. The contrast in the two diagrams indicates that during wet (or dry) years over the Indochina Peninsula, westward-propagating TCs and tropical disturbances have high (or low) activity, consistent with Takahashi et al. (2015). The spatial pattern of the difference in seasonal mean streamfunction between the two years (Fig. C2; wet minus dry) showed an anomalous cyclonic circulation that extends from the Bay of Bengal to the western North Pacific that was similar to the results of the composite analysis based on the summer seasonal precipitation over the Indochina Peninsula (Fig. 3b). The contrast in the Hovmöller diagrams corresponds with the difference in seasonal mean streamfunction between these two years, suggestive of a relationship between the activity of westward-propagating disturbances and the seasonal mean streamfunction.
Difference in the JJA streamfunction (×1.0 × 106 m2 s−1) between the wettest and driest years (wet minus dry) for the observational data.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
In addition, we investigated whether CMIP5 models were able to simulate this contrast of the activity levels of the westward-propagating disturbances and its relationship with the seasonal mean streamfunction by analyzing the daily data outputs for two CMIP5 models (MIROC5 and MPI-ESM-MR). Figure C3 presents Hovmöller diagrams of the PKE averaged over 10°–20°N across the ASM region for the wettest and driest years over the Indochina Peninsula for both of the models. Both models reproduced westward-propagating TCs and tropical disturbances, and the activity levels of these phenomena were high (low) during the wettest (driest) year, consistent with the observational data. The difference in seasonal streamfunction between the wettest and driest years was negative over the Indochina Peninsula, and the signal extended as an anomalous cyclonic zonal band from the Bay of Bengal to the western North Pacific that was similar to the results of the composite analysis based on the summer seasonal precipitation over the Indochina Peninsula for the CMIP5 MME (Fig. 4b), which was also consistent with the observational data results (Fig. C4). These results imply that the relationship between the seasonal streamfunction and activity level of westward-propagating TCs and tropical disturbances were reproduced in the CMIP5 models, at least for these two models. Moreover, changes in the seasonal mean streamfunction along the monsoon trough may indicate the changes of the activity level of westward-propagating tropical disturbances.
As in Fig. C1, but for (a),(b) MIROC5 and (c),(d) MPI-ESM-MR.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
As in Fig. C2, but for (a) MIROC5 and (b) MPI-ESM-MR.
Citation: Journal of Climate 31, 20; 10.1175/JCLI-D-17-0685.1
REFERENCES
Adler, R. F., and Coauthors, 2003: The Version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J. Hydrometeor., 4, 1147–1167, https://doi.org/10.1175/1525-7541(2003)004<1147:TVGPCP>2.0.CO;2.
Brown, J. R., R. A. Colman, A. F. Moise, and I. N. Smith, 2013: The western Pacific monsoon in CMIP5 models: Model evaluation and projections. J. Geophys. Res. Atmos., 118, 12 458–12 475, https://doi.org/10.1002/2013JD020290.
Cai, W., and Coauthors, 2015: ENSO and greenhouse warming. Nat. Climate Change, 5, 849–859, https://doi.org/10.1038/nclimate2743.
Chen, T.-C., and S.-P. Weng, 1999: Interannual and intraseasonal variations in monsoon depressions and their westward-propagating predecessors. Mon. Wea. Rev., 127, 1005–1020, https://doi.org/10.1175/1520-0493(1999)127<1005:IAIVIM>2.0.CO;2.
Chen, T.-C., and J.-H. Yoon, 2000: Interannual variation in Indochina summer monsoon rainfall: Possible mechanism. J. Climate, 13, 1979–1986, https://doi.org/10.1175/1520-0442(2000)013<1979:IVIISM>2.0.CO;2.
Christensen, J. H., and Coauthors, 2013: Climate phenomena and their relevance for future regional climate change. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1217–1308, http://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter14_FINAL.pdf.
Collins, M., and Coauthors, 2013: Long-term climate change: Projections, commitments and irreversibility. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 1029–1136, http://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter12_FINAL.pdf.
Flato, G., and Coauthors, 2013: Evaluation of climate models. Climate Change 2013: The Physical Science Basis, T. F. Stocker et al., Eds., Cambridge University Press, 741–866, http://www.ipcc.ch/pdf/assessment-report/ar5/wg1/WG1AR5_Chapter09_FINAL.pdf.
Freychet, N., H.-H. Hsu, C. Chou, and C.-H. Wu, 2015: Asian summer monsoon in CMIP5 projections: A link between the change in extreme precipitation and monsoon dynamics. J. Climate, 28, 1477–1493, https://doi.org/10.1175/JCLI-D-14-00449.1.
Goswami, B. N., V. Krishnamurthy, and H. Annmalai, 1999: A broad‐scale circulation index for the interannual variability of the Indian summer monsoon. Quart. J. Roy. Meteor. Soc., 125, 611–633, https://doi.org/10.1002/qj.49712555412.
Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 5686–5699, https://doi.org/10.1175/JCLI3990.1; Corrigendum, 24, 1599–1560, https://doi.org/10.1175/2010JCLI4045.1.
Kitoh, A., S. Yukimoto, A. Noda, and T. Motoi, 1997: Simulated changes in the Asian summer monsoon at times of increased atmospheric CO2. J. Meteor. Soc. Japan, 75, 1019–1031, https://doi.org/10.2151/jmsj1965.75.6_1019.
Kobayashi, S., and Coauthors, 2015: The JRA-55 Reanalysis: General specifications and basic characteristics. J. Meteor. Soc. Japan, 93, 5–48, https://doi.org/10.2151/jmsj.2015-001.
Lee, J.-Y., and B. Wang, 2014: Future change of global monsoon in the CMIP5. Climate Dyn., 42, 101–119, https://doi.org/10.1007/s00382-012-1564-0.
Lu, R., and Y. Fu, 2010: Intensification of East Asian summer rainfall interannual variability in the twenty-first century simulated by 12 CMIP3 coupled models. J. Climate, 23, 3316–3331, https://doi.org/10.1175/2009JCLI3130.1.
Matsuyama, H., and H. Katasakai, 2012: A simple but quantitative method to objectively determine the degree of freedom, applicable to serially-correlated meteorological data (in Japanese). Tenki, 59, 1065–1067.
Meehl, G. A., and Coauthors, 2007: Global climate projections. Climate Change 2007: The Physical Science Basis, S. Solomon et al., Eds., Cambridge University Press, 747–846.
Ogata, T., and Coauthors, 2014: Projected future changes in the Asian monsoon: A comparison of CMIP3 and CMIP5 model results. J. Meteor. Soc. Japan, 92, 207–225, https://doi.org/10.2151/jmsj.2014-302.
Seneviratne, S. I., and Coauthors, 2012: Changes in climate extremes and their impacts on the natural physical environment. Managing the Risks of Extreme Events and Disasters to Advance Climate Change Adaptation, C. B. Field et al., Eds., Cambridge University Press, 109–230, http://www.ipcc.ch/pdf/special-reports/srex/SREX-Chap3_FINAL.pdf.
Sperber, K. R., H. Annamalai, I.-S. Kang, A. Kitoh, A. Moise, A. Turner, B. Wang, and T. Zhou, 2013: The Asian summer monsoon: An intercomparison of CMIP5 vs. CMIP3 simulations of the late 20th century. Climate Dyn., 41, 2711–2744, https://doi.org/10.1007/s00382-012-1607-6.
Takahashi, H. G., H. Fujinami, T. Yasunari, J. Matsumoto, and S. Baimoung, 2015: Role of tropical cyclones along the monsoon trough in the 2011Thai flood and interannual variability. J. Climate, 28, 1465–1476, https://doi.org/10.1175/JCLI-D-14-00147.1.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2009: A summary of the CMIP5 experiment design. PCDMI Rep., 33 pp., https://cmip.llnl.gov/cmip5/docs/Taylor_CMIP5_design.pdf.
Taylor, K. E., R. J. Stouffer, and G. A. Meehl, 2012: An overview of CMIP5 and the experiment design. Bull. Amer. Meteor. Soc., 93, 485–498, https://doi.org/10.1175/BAMS-D-11-00094.1.
Torrence, C., and P. J. Webster, 1999: Interdecadal changes in the ENSO–monsoon system. J. Climate, 12, 2679–2690, https://doi.org/10.1175/1520-0442(1999)012<2679:ICITEM>2.0.CO;2.
Turner, A. G., and H. Annamalai, 2012: Climate change and the South Asian summer monsoon. Nat. Climate Change, 2, 587–595, https://doi.org/10.1038/nclimate1495.
Vecchi, G. A., B. J. Soden, A. T. Wittenberg, I. M. Held, A. Leetmaa, and M. J. Harrison, 2006: Weakening of tropical Pacific atmospheric circulation due to anthropogenic forcing. Nature, 441, 73–76, https://doi.org/10.1038/nature04744.
Wang, B., and Z. Fan, 1999: Choice of South Asian summer monsoon indices. Bull. Amer. Meteor. Soc., 80, 629–638, https://doi.org/10.1175/1520-0477(1999)080<0629:COSASM>2.0.CO;2.
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, https://doi.org/10.1175/1520-0442(2001)014<4073:IVOTAS>2.0.CO;2.
Xie, P., and P. A. Arkin, 1997: Global precipitation: A 17-year monthly analysis based on gauge observations, satellite estimates, and numerical model outputs. Bull. Amer. Meteor. Soc., 78, 2539–2558, https://doi.org/10.1175/1520-0477(1997)078<2539:GPAYMA>2.0.CO;2.
