Interdecadal Variability of the Afro-Asian Summer Monsoon System

Asia and Africa are two well-known regions typical of monsoonal climate. The Afro-Asian monsoon is a major component of the global monsoon system, and its behavior and variation exert great impacts on precipitation over Africa to East Asia. The Afro-Asian monsoon rainbelt generally refers to a narrow zonally elongated belt extending from the Sahel through to the Indian peninsula, China and eastwards onto western Japan. Paleoclimate studies have reported that such an elongated rainbelt have already taken shape during the whole of geological time. The intensity of the Asian summer monsoon and African summer monsoon vary in phase throughout ancient history (Ji et al., 1993). On millennial timescales, active/break phases of the Afro-Asian summer monsoon are manifested in abundant/deficient precipitation over China, India and West Africa (Yan and Petit-Maire, 1995). (Liu et al., 2014b) demonstrated that summertime precipitation in northern China is not only governed by the East Asian summer monsoon (EASM), but is also influenced by the South Asian summer monsoon and African summer monsoon. During the past 21 000 years, the major precipitation pattern exhibits either a northward advance of the Afro-Asian summer monsoon rainbelt with enhanced precipitation, or is characterized by a southward retreating rainbelt accompanied by decreasing rainfall. From the global monsoon perspective, the Afro-Asian monsoon system may also interact with the intertropical convergence zone (ITCZ) in the lower troposphere (Chao, 2000; Gadgil, 2003). (Wang, 2009) reported that the ITCZ is able to reflect the variations of the Afro-Asian summer monsoon on geological timescales. The southward shift of the ITCZ was closely related to the dry phase of the Afro-Asian summer monsoon during Heinrich Event 1 (McManus et al., 2004; An et al., 2015).

Previous studies have conducted analyses of interdecadal variations of rainfall, winds and sea level pressure (SLP) within the Afro-Asian summer monsoon in the past century (Ye et al., 1996; Song et al., 2009). Derived results have mainly revealed the interdecadal abrupt change of the Afro-Asian summer monsoon from strong to weak in the 1960s (Song and Ji, 2001). Various studies have confirmed that the Sahel region also witnessed a sharp reduction in precipitation during the 1960s that lasted until the 1980s (Giannini et al., 2003; Biasutti and Giannini, 2006; Baines and Folland, 2007). Such a consistent reduction in Afro-Asian monsoon precipitation has been widely investigated, and the tropical easterly jet (TEJ) is believed to be a main driver. The rainy areas of the Afro-Asian summer monsoon are beneath the northern flange of the entrance of the TEJ (East Asia), and are located well to the south of the exit region (North Africa) (Koteswaram, 1958; Krishnamurti, 1971; Zeng and Guo, 1982). The reduction in Afro-Asian summer monsoon rainfall could be attributed to a weakened TEJ caused by a decreasing land-sea thermal contrast (Chen et al., 2007; Song et al., 2009). Additionally, a considerable number of studies have ascribed this interdecadal variability of African and East Asian rainfall to the simultaneous abrupt regime shift of the Atlantic Multidecadal Oscillation (AMO) towards its cold phase (Liu et al., 2014a). Model simulations have revealed that cold sea surface temperature (SST) in the extratropical North Atlantic could result in a cooling of continental Eurasia and North Africa, disturbing the moisture cycle, and further dampening the African summer monsoon. Consequently, droughts in the Sahel occur in tandem with cold phases of the AMO (Liu and Chiang, 2012).

Quite a few studies have attributed monsoonal precipitation over East Asia, South Asia and Africa to SST anomalies in key areas. In particular, tropical Atlantic SST anomalies have been reported to influence precipitation over Africa and India (Zhou et al., 2008; Kucharski et al., 2009). Warmer SSTs in the tropical Indian Ocean and tropical Pacific could modulate the low-frequency variations of Sahelian rainfall and drought in North China (Giannini et al., 2003; Yang and Ding, 2007). From the decadal to the centennial scale, North Atlantic SST anomalies also exert significant influences on drought in Africa and Southeast Asia (Feng and Hu, 2008; Zhang et al., 2008; Shanahan et al., 2009). In the present study, we focus on the role of North Atlantic SST anomalies. Corresponding to the cold phase of the AMO, cooling tends to appear in the mid-upper troposphere across the Eurasian continent. Thus, the thermal contrast between land and sea weakens, further leading to a weakening of the Indian summer monsoon and EASM (Lu and Dong, 2008). Some paleoclimate data also indicate that severe droughts in West Africa and a weakening of the Asian summer monsoon might arise from North Atlantic cooling (Gupta et al., 2003; Stager et al., 2011). From another perspective, warm SSTs in the North Atlantic lead to warming in Eurasia and the northern Indian subcontinent, favoring a northward advance of the ITCZ and enhancement of southwesterly flow over the Sahel and Indian subcontinent. Low-level moisture convergences bring more abundant precipitation to the Sahel and central-southern India (Zhang and Delworth, 2006; Wang et al., 2009). North Atlantic SST warming reinforces coupled atmosphere-ocean feedback by exciting SST anomalies in the western Pacific and Indian Ocean, and temperature anomalies in the troposphere, leading to a low-level anticyclonic anomaly over the western North Pacific and more rainfall over East Asia (Lu et al., 2006).

Recently, we noticed that the AMO has transitioned into its warm phase since the 1990s. Meanwhile, in the Sahel, after a relative dry period persisting for 20 years from the 1960s to 1980s, precipitation has been recovering slowly, possibly caused by direct or indirect effects of anthropogenic warming (Dong and Sutton, 2015; Giannini, 2015). During the early 1990s, the recovery of the EASM led to increasing precipitation over the Huai River valley in northern China since the late 1990s (Si et al., 2009; Zhu et al., 2011; Si and Ding, 2013; Zhang, 2015). These phenomena naturally raise the question as to whether the increasing rainfall over the Sahel and the recovery of the EASM are both related to the warm phase of the AMO in the past two decades, and whether changes of the Afro-Asian monsoon are caused by joint influences of anthropogenic emissions and SST anomalies. In this context, we investigate the climatological features and temporal variabilities of the Afro-Asian monsoon system during the past century and the past three decades. Additionally, a mechanism responsible for the change in the Afro-Asian summer monsoon in the warm phase of the AMO is proposed. This work enriches the perspectives and conclusions raised in the research of (Liu and Chiang, 2012).

The rest of the paper is structured as follows: Section 2 describes the data and methodology used. In section 3, we present the climatological features of the Afro-Asian summer monsoon during the past century and the last 36 years. Next, in section 4, we discuss a possible physical mechanism accounting for the change in the Afro-Asian summer monsoon in the warm phase of the AMO. In section 5, we discuss the response of the TEJ to the warm phase of the AMO and its implications. The key findings of the study are summarized in section 6.

Figure 1.  Spatial patterns of the (a) first and (b) second EOF mode of the monthly mean precipitation during 1901-2014 (shading; units: mm d^-1). Panels (c) and (d) are the first and second normalized PCs. The gray lines are the 11-point smoothing of the respective normalized PCs.

The observational data include the following products: The monthly precipitation data for summer (June-August) during 1901-2014 are provided by the Climatic Research Unit of the University of East Anglia, at a horizontal resolution of 0.5^°× 0.5^° (Hulme, 1992; Brohan et al., 2006). Another set of monthly mean precipitation data for summer (June-August) during 1979-2014 is obtained from the Global Precipitation Climatology Project, at a horizontal resolution of 2.5^°× 2.5^° (Adler et al., 2003). We also use reanalysis data (June-August) from the ERA-Interim reanalysis dataset of the ECMWF, including precipitation, relative vorticity, winds, and 2-m surface temperature. All these variables are provided at a resolution of 1^°× 1^°, and their temporal coverage spans from 1979 to 2014 (Dee et al., 2011). The National Centers for Environmental Prediction-National Center for Atmospheric Research (NCEP-NCAR) geopotential height and wind reanalysis, with a resolution of 2.5^°× 2.5^° and covering the period 1979-2014, is also utilized (Kalnay et al., 1996). The monthly-mean SLP data for summer (June-August) are from the Hadley Center (Allan and Ansell, 2006).

In this study, we analyze horizontal wind at 850 hPa because troughs at the planetary and peninsula scale are manifested more obviously in the near-surface layers (Trenberth et al., 2000). The core wind speed in the TEJ usually appears between 150 hPa and 100 hPa (Koteswaram, 1958; Chen et al., 2007), so the TEJ index is constructed by averaging the zonal wind speed within 10^°-15^°N between these two levels.

We conduct analyses via the empirical orthogonal function (EOF) and multi-variable EOF (MV-EOF) methods, as detailed in (Wang, 1992), as well as linear regression and phase composites to reveal basic features and changes in the Afro-Asian monsoon.

Figure 2.  MV-EOF patterns between summer (June-August) precipitation and 850 hPa winds during 1979-2014: (a) first leading pattern of the Afro-Asian region; (b) first leading pattern of the African region; (c) second leading pattern of the Indian region; (d) first leading pattern of the East Asian region. The shading indicates the precipitation patterns regressed on the PCs (units: mm d^-1). The vectors are the 850 hPa wind regressed on the PCs.

Afro-Asian summer monsoon precipitation can intuitively depict the Afro-Asian summer monsoon system. Figure 1 shows the first two leading modes of Afro-Asian monsoon precipitation during 1901-2014. The first leading mode behaves as a zonally extended mode from Africa to Asia (Fig. 1a), with a variance contribution of 7%. Specifically, precipitation over the Sahel, South Asia and East Asia vary in phase. The normalized PC1 shows an interdecadal shift of Afro-Asian rainfall from positive to negative in the late 1960s (Fig. 1c). This corresponds to concurrent reductions in precipitation over the Sahel, South Asia and eastern China. A similar phenomenon has also been reported in some previous studies (Biasutti and Giannini, 2006; Wang and Fan, 2013). However, of particular note is that the rainfall in the Afro-Asian region seems to have increased gradually in the last three decades. The precipitation in the Sahel has been recovering slowly, accompanied by an interdecadal regime shift of EASM precipitation. In eastern China in particular, the spatial pattern of summer precipitation is more typical of a meridional mode, with a positive-negative-positive distribution in the last decade. The second EOF mode accounts for 4.1% of the total variance (Fig. 1b), and exhibits an interruption of the Afro-Asian monsoon rainband, with the break-point located over the Indian Peninsula. Accordingly, Afro-Asian summer monsoon precipitation can be separated into three sectors: the African summer monsoon region, the South Asian summer monsoon region, and the EASM region. The precipitation in the African summer monsoon region and EASM region still vary in phase. As revealed in Fig. 1d, the PC2 is mainly dominated by interannual to decadal oscillations. Generally, these two leading modes show that, during the past century, the Afro-Asian summer monsoon was an integral system on interdecadal timescales. However, on interannual timescales, each subsystem was governed by different factors, leading to out-of-phase variations between them. The reason for the interannual pattern of Afro-Asian summer monsoon is an interesting issue worthy of further exploration, but is beyond the scope of the current study.

The above analyses highlight that, in the last three decades, the Afro-Asian summer monsoon precipitation has begun a recovery. Coincidentally, if we apply the Mann-Kendall test to PC1, a regime shift point is detected around 1975. An in-depth understanding of the recent recovery of the Afro-Asia summer monsoon is desirable, and that is the very purpose of this study. So, in the following sections, the time period for investigating the precipitation, winds and other variables, is 1979 to 2014.

Figure 3.  MV-EOF PCs between precipitation and 850 hPa winds during 1979-2014: (a) first PC of the Afro-Asian region; (b) first PC of the African region; (c) second PC of the Indian region; (d) first PC of the East Asian region.

The above analyses reveal that the Afro-Asian summer monsoon is a planetary-scale system, covering Africa, South Asia and East Asia. Also, the precipitation of the Afro-Asian monsoon has experienced a recovery in the last three decades. Figures 2 and 3 show the spatial patterns and corresponding normalized principal components of the first and second leading MV-EOF modes of precipitation and 850 hPa winds during 1979-2104. As shown in Fig. 2a, the first leading mode accounts for 13.5% of the total variance. In contrast to the first EOF pattern on centennial timescales, the first leading pattern during the past 36 years exhibits a continuous rainbelt in the African and South Asian monsoon regions and a dipole pattern referred to as "south flood-north drought" in East Asia. To guarantee the robustness of the derived leading modes, MV-EOF analysis is also conducted for the African, South Asian and East Asian regions, and the results are shown in Figs. 2b, c and d. The normalized PCs show that, during 1979-2014, the Afro-Asian summer monsoon precipitation in both the entire Afro-Asian region (Fig. 3a) and its sub-regions [the Sahel (Fig. 3b); South Asia (Fig. 3c); East Asia (Fig. 3d)] varied in phase. The PCs are mainly characterized by interdecadal oscillations. A decadal abrupt change of the precipitation in these regions occurred in the late 1990s. The interdecadal shift is in tandem with the evolution of the centennial-scale (1901-2014) PC1 (Fig. 1c) during the past 30 years, and also coincides well with the aforementioned recovery of centennial-scale summer precipitation in the Afro-Asian region. In this context, the variability of Afro-Asian summer monsoon precipitation during 1979-2014 is an important cycle of centennial-scale summer precipitation in the Afro-Asian region. A concrete analysis of the Afro-Asian summer monsoon precipitation during 1979-2014 could reveal more detail.

The leading spatial pattern and PCs together show that rainfall in the Sahel, South Asia, and regions from south of the Yangtze River valley to the Huanghe-Huaihe River valley, has increased since the late 1990s. Several studies have indicated that summer precipitation in China can be categorized into a dipole pattern and a tripole pattern (positive-negative-positive), and that the former one, i.e., the widely-recognized "south flood-north drought" pattern, formed at the beginning of the 1990s (Huang et al., 2004; Ding et al., 2008, 2013). However, it is worth noting that precipitation in the Huanghe-Huaihe River valley increased in the late 1990s (Fig. 2d). Since the late 1990s, the rainbelt in eastern China has marched northwards to the Huanghe-Huaihe River valley. Along with the recovery of precipitation in the Sahel, the integral rainbelt of the Afro-Asian summer monsoon has shifted northwards.

Figure 4.  (a) Normalized temporal variability of the AMO during 1901-2014 [construction method derived from Enfield et al. (2001)]. The solid line denotes the 11-year running average. (b) 11-yr running correlation coefficient between the normalized AMO index and PC1 of the first EOF PC of Afro-Asian summer monsoon precipitation during 1901-2014. The dashed lines denote the significance at the 0.05 level.

In short, based on an analysis of both 100 years and 36 years of data, the climatology of the Afro-Asian summer monsoon can be clearly characterized by its interdecadal variability. From the 1930s to the early 1960s, the Afro-Asian summer monsoon rainfall consistently shifted northwards. After the mid-1960s, the reduction in the rainfall covered from the Sahel in Africa to North China in East Asia. It is also very interesting that, since the late 1990s, the Afro-Asian summer monsoon has revived with strengthened rainfall.

The EOF analyses of summer precipitation (section 3) delineate the northward shift of the Afro-Asian summer monsoon and in-phase changes of precipitation between the Sahel and the Huanghe-Huaihe River valley since the late 1990s. Coincidently, the AMO has turned into one of its warm phases since the 1990s. So, how does the warm phase of the AMO result in the northward advance of the entire Afro-summer monsoon system? To answer this question, we further analyze the upper- and lower-level circulations of the Afro-Asian summer monsoon, and summarize the coupled response of the upper and lower atmosphere to the warm AMO phase.

Figure 5.  Differences in 200 hPa meridional wind between the composite warm phase and cold phase of the AMO (units: m s^-1). During 1979-2014, the AMO cold phase years are 1979-86, 1988-94, 1996-97, 2002, and 2009; the AMO warm phase years are 1987, 1995, 1998-2001, 2003-08, and 2010-14. Values exceeding the 95% confidence level are filled with dots. The black letters "A" and "C" denote the centers of anomalous anticyclones and cyclones in the upper troposphere. The dashed curve with an arrow denotes the teleconnection wave train during the AMO warm phase.

Figure 6.  Differences in (a) surface air temperature (shading; units: ^°C) and (b) SLP (shading; units: Pa) between the composite warm and cold phase of the AMO. The AMO warm phase and cold phase years are the same as in Fig. 5. The black plus and minus signs denote the positive and negative values in the surface air temperature field and SLP field. Values exceeding the 95% confidence level are filled with dots.

Following the definition of the AMO introduced by (Enfield et al., 2001), we construct the AMO index, as shown in Fig. 4a. A cold phase of the AMO persisted from the 1960s to 1980s, followed by a phase transition into a warm phase since the 1990s. Thus, changes in the Afro-Asian summer monsoon occurred in the context of a warm phase of the AMO. During 1901-2014, the correlation coefficient between the normalized AMO index and the centennial-scale (1901-2014) PC1 of Afro-Asian summer monsoon precipitation reaches 0.35, significant at the 0.05 level. With an 11-yr running mean applied, the coefficient between these two smoothed indices can reach 0.68, also significant at the 0.05 level. The significant and strong correlations pinpoint that the AMO exerts significant impacts on Afro-Asian summer monsoon precipitation, especially on the decadal scale and even longer time scales. Figure 4b shows the 11-yr running correlation coefficients between the normalized AMO index and the PC1 during 1901-2014. Clear positive correlations appear during the late 1960s to 1980s, as well as in the regime since the late 1990s. Specifically, the cold AMO phase during the 1960s to 1980s corresponds to reduced Afro-Asian summer monsoon precipitation, while the warm AMO phase is in tandem with the recovery of Afro-Asian summer monsoon precipitation since the late 1990s.

Modulated by the AMO warm phase, a global wave train with wavenumbers 5-6 is apparent at 200 hPa over the Northern Hemisphere (Fig. 5). This zonal wave train propagates along the westerly jet in the midlatitudes. Six prominent centers are situated over the North Atlantic Ocean, Western Europe, Eastern Europe, central Asia, East Asia and the North Pacific, respectively. We note that this wave train closely resembles the circumglobal teleconnection (CGT) (Ding and Wang, 2005) and the AMO-Northern Hemisphere teleconnection over the Eurasian continent (Si and Ding, 2016). (Wu et al., 2016) also proved that the interdecadal CGT pattern in the zonal direction is excited by the forcing of the AMO. In fact, the circumglobal zonal wave train at 200 hPa derived from the meridional wind field reflects the structure of the Rossby wave associated with the alternating alignment of cyclones/anticyclones. As shown in Fig. 5, two prominent positive centers are located over Western Europe and central Asia, and two prominent negative centers are located over Eastern Europe and East Asia. This teleconnection pattern causes both North Africa and East Asia to be under the control of an upper-level anticyclone. Upper-level divergence prevails over the Sahel and eastern China. The above results present the upper-level circulation system. In a baroclinic atmosphere, there should be a coupled circulation pattern manifested in the lower troposphere. Therefore, we next examine the corresponding circulation pattern in the lower troposphere during the AMO warm phase.

The response of surface temperature and the SLP in continental Eurasia to warm SST anomalies in the North Atlantic may be a possible mechanism linking the AMO to the Afro-Asian monsoon. During the AMO warm phase, surface temperature anomalies over the Eurasian continent enhance the meridional land-sea thermal contrast. Our study calculates the differences in surface temperature and SLP between warm phases and cold phases of the AMO (Fig. 6). Figure 6a illustrates that, during warm phases of the AMO, warmer surface temperature can be detected across the Eurasian continent, especially in Europe, North Africa, midlatitude East Asia, and the Maritime Continent. Surface temperature warming favors the formation of a low-pressure zone in the low-level troposphere, which is firmly located over the Sahara and the Sahel. This can be validated by the SLP anomalies (Fig. 6b). Sandwiched between the low-pressure system north of the Sahel and the high-pressure system over south-central Africa, enhanced westerlies/southwesterlies forced by the extra pressure gradient will enhance precipitation locally in the Sahel region. (Lavaysse, 2015) pointed out that a stronger low-pressure system tends to appear in the warmer region of the Sahara, resulting in a stronger West African summer monsoon. The coupling between the low-level thermal low and the upper-level anticyclone in North Africa favors ascending motion in the Sahel. Similar coupling of the atmospheric circulations occurs in the EASM region. A positive-negative-positive meridional pattern is manifested in the SLP anomaly field. Negative SLP anomalies appear in the region between 28^°N and 40^°N in East Asia, and positive SLP anomalies reside over the area south of the Yangtze River. So, in the Huanghe-Huaihe River valley, the extra pressure gradient will enhance precipitation locally, as shown in Fig. 7. Moreover, ascending motion over the Huanghe-Huaihe River valley can be excited by the coupling of the upper-level divergence and surface low pressure. The similarity between the circulation configurations in East Asia and the Sahel may be the reason for the synchronous change in precipitation over the two regions (Fig. 7).

In South Asia, anomalies of surface temperature and SLP are generally consistent with those in the Sahel during warm AMO phases. However, positive anomalies in surface temperature and SLP are small and insignificant in South Asia. This is the reason for the insignificant change in the intensity of the South Asian monsoon. Above all, the response of the meridional wind, surface temperature and SLP to SST variations in the North Atlantic play a key role in modulating the northward shift of the Afro-Asian summer monsoon rainbelt and simultaneous enhancement of precipitation in the Sahel and Huanghe-Huaihe River valley. Indeed, increasing precipitation in the Sahel and Huanghe-Huaihe River valley since the late 1990s is associated with the AMO warm phase.

Figure 7.  Differences in precipitation anomalies (shading, units: mm d^-1) between the composite warm phase and cold phase of the AMO. The AMO warm and cold phase years are the same as in Fig. 5. Values exceeding the 95% confidence level are filled with dots.

Figure 8.  (a) Linear trends of zonal winds averaged between 150 hPa and 100 hPa during 1979-2014 [shading; units: m s^-1 (10 yr)^-1]. The isolines denote zonal winds (units: m s^-1). (b) Longitude-time cross section of zonal wind anomalies (averaged over 10^°-15^°N) between 150 hPa and 100hPa during 1979-2014 (shading; units: m s^-1). The dashed red line denotes the time of the mid-1990s. Values exceeding the 95% confidence level are filled with dots. The arrow in (a) represents the direction of the TEJ axis, and in (b) represents the direction of the TEJ increase.

Figure 9.  (a) Thickness anomaly (height difference between 200 hPa and 500 hPa, representing upper-tropospheric temperature) during the warm phase of the AMO (shading; units: gpm). (b) Temporal variation in the meridional thickness gradient (-∂ Z/∂ y, where z denotes thickness) within the region (0^°-45^°N, 10^°-45^°E). Values exceeding the 95% confidence level are filled with dots. The AMO warm and cold phase years are the same as in Fig. 5.

From the perspective of upper-level winds, the TEJ is an important component of the Afro-Asian monsoon system. The strength and location of the TEJ directly affects the rainfall near the jet entrance (East Asia) and the jet exit (the Sahel region). Linear trends of the TEJ during 1979-2014 can be detected in Fig. 8a. The TEJ significantly intensifies to the right (north) side of the entrance region around (35^°N, 115^°E) (Huanghe-Huaihe River valley in China) and the exit region around (10^°N, 30^°E) (the Sahel in Africa). But, the enhanced trend of the TEJ in the exit region is stronger than that in the entrance region. In the entrance region, the TEJ displaces northwards to 30^°N. Temporal changes in the TEJ from Africa to Asia can be more obviously identified in the longitude-time cross section shown in Fig. 8b. The TEJ has strengthened obviously in Africa during the last two decades. The branch in East Asia accelerated in the 1980s to mid-1990s, but has been weaker than the African branch in the last decade. The branch in Africa has become stronger since the mid-1990s. The above results indicate that the strengthened and northward-extended TEJ is one of the important reasons for the northward advancement of the Afro-Asian summer monsoon rainbelt. Although the changes in the African summer monsoon and EASM are generally consistent, the variation in Africa may be stronger than that in East Asia. The TEJ has changed during the AMO warm phase. The responses of the upper troposphere to the warm AMO phase show some important influences on the TEJ during the last three decades. In our study, the upper-tropospheric temperature anomaly is represented by the thickness anomaly, achieved through the height difference between 200 hPa and 500 hPa. In Eurasia, the upper-tropospheric temperatures show positive anomalies in response to the warm AMO phase, resulting in a strengthening meridional gradient of the upper-tropospheric temperature (Fig. 9). Consequently, both the westerlies to the north of the warm center and easterlies to the south of the warm center accelerate. As shown in Fig. 9a, the significant maximum temperature anomalies lie in the north of the entrance and exit region of the TEJ, reinforcing the easterly in the entrance and exit region. In addition, the upper-level warming during the warm AMO phase facilitates the acceleration of the TEJ in the exit region, i.e., in Africa (0^°-45^°N, 10^°-45^°E), with a weakening regional meridional northward pressure gradient (Fig. 9b). This temporal variation in the meridional height gradient over Africa confirms the significantly increasing trend of the upper-tropospheric temperature in the African branch of the TEJ.

The strengthening of the TEJ enhances vertical cells in both the entrance and exit region, whose ascending branches bring more precipitation (Webster et al., 1998). Figure 10 presents the pressure-latitude cross section of anomalous meridional winds and geopotential height in the entrance and exit regions during warm AMO phases. A secondary circulation is discernable in the entrance region of the TEJ (East Asia), with southerlies prevailing in the lower levels and northerlies in the upper levels. The ascending branch is located to the north of the TEJ axis, where obvious precipitation is also apparent (20^°-38^°N, from South China to the Huanghe-Huaihe River valley). Similarly, another secondary cell resides over the exit region of the TEJ (North Africa). In the northern flange of the exit region, northerlies prevail in the low-level wind, accompanied by southerlies in the upper levels. The opposite situation occurs in the southern flange of the exit region. Thus, low-level convergence promotes strong ascending motion and subsequent precipitation near 10^°N. In warm AMO phases, the precipitation belt associated with the TEJ shifts more northwards and behaves stronger than that during cold phases.

Notably, both Fig. 9 and Fig. 10 show that, during warm AMO phases, pressures to the north of the TEJ tend to strengthen. Accordingly, an intensified pressure gradient favors acceleration of the TEJ. Also intensified are the jet-related secondary cells, ageostrophic motions, and precipitation in both the entrance and exit regions (Cressman, 1984).

The response of the upper-level temperature to AMO warm phases also plays a key role in the strengthened and northward extension of the TEJ, which directly influence the Afro-Asian summer monsoon rainfall through intensification of the meridional pressure gradient from north to south both in the entrance region (East Asia) and the exit region (Africa).

Figure 10.  Pressure-latitude cross section of meridional wind (shading; units: m s^-1) and geopotential height anomalies (contours; units: gpm) in the AMO warm phase during 1979-2014 in (a) North Africa along 20^°E and (b) East Asia along 118^°E. The arrows represent the vertical movement by ω values, which are taken from the NCEP-NCAR reanalysis dataset. Values exceeding the 95% confidence level are filled with dots. "JE" is the position of the TEJ axis. The black bold line on the abscissa is the position of precipitation.

Based on reanalysis data of global precipitation and atmospheric variables, the climatological features and variabilities of the rainbelt and structure of the Afro-Asian summer monsoon have been investigated. The main conclusions can be summarized as follows:

(1) The major mode of the centennial-scale (1901-2014) Afro-Asian summer monsoon precipitation behaves as a large-scale rainbelt extending from Africa via South Asia to East Asia on interdecadal timescales. This zonally planetary-scale system can also be depicted by variables including precipitation and convergence in the low-level wind field during 1979-2014. In the last three decades (1979-2014), a decadal abrupt change of the precipitation in these regions occurred in the late 1990s. The interdecadal shift is in tandem with the evolution of the centennial-scale (1901-2014) PC1 during the past 30 years. Since the late 1990s, the rainbelt in eastern China has marched northwards to the Huanghe-Huaihe River valley. Along with the recovery of precipitation in the Sahel, the entire rainbelt of the Afro-Asian monsoon system advanced northwards in a coordinated way.

(2) The reason for the northward shift of the Afro-Asian summer monsoon rainbelt and the in-phase increase in precipitation over the Sahel and Huanghe-Huaihe River valley since the late 1990s, is the coupled responses of wind, SLP and surface temperature to warm SST anomalies in the North Atlantic. A coupling mechanism for the influence of the AMO on the variability of Afro-Asian summer monsoon rainfall has been summarized, as shown in Fig. 11. During the AMO warm phase, a teleconnection wave train is apparent in the upper troposphere, with alternating cyclone/anticyclone pairs and with a dipole (negative-positive) pattern of SLP in Africa and a tripole (positive-negative-positive) pattern of SLP in eastern Asia. This configuration between circulation anomalies couples the upper-level anticyclone with the lower-level thermal low in northern Africa, forming a structure of upper-level divergence and lower-level convergence that excites ascending motion in the Sahel. Simultaneously, with a positive-negative-positive pattern of SLP in East Asia, a similar coupled circulation pattern, with upper-level divergence and surface low pressure, triggers ascending motion in the Huanghe-Huaihe River valley. Such coupling between low-level and upper-level systems induces synchronous ascending motions in the Sahel and Huanghe-Huaihe River valley.

Figure 11.  Schematic diagram illustrating the effect of the AMO warm phase on Afro-Asian summer monsoon rainfall. The red "L" and blue "H" denote an anomalous warm cyclone and cold anticyclone in the surface system, respectively. The purple "A" and "C" denote an anomalous anticyclone and cyclone in the upper troposphere. The dashed curve with an arrow denotes the teleconnection wave train during the AMO warm phase.

(3) Warm AMO phases also result in warming in the upper troposphere, and such warming contributes to intensification of the TEJ both in the entrance region (East Asian branch) and the exit region (African branch). This in-phase change in the entrance branch and exit branch of the upper-level TEJ could intensify precipitation in the Sahel and Huanghe-Huaihe River valley through exciting ageostrophic secondary circulation cells.

The results of this study indicate that the Afro-Asian summer monsoon has assumed a consistent and holistic interdecadal change with the in-phase increasing precipitation.