1 Introduction

The East Asian summer monsoon (EASM), mainly formed by large land-sea thermal contrast, is characterized by strong southerly winds and abundant rainfall over East Asia, one of the most populous regions in the world (Tao and Chen 1987; Xue et al. 2015; Huang et al. 2017a, b; Li et al. 2017, 2019; Ding et al. 2018; Chen et al. 2019). Anomalous activity of EASM can exert substantial impacts on the agriculture, ecosystem, and socioeconomic development over East Asia and remote regions (Huang et al. 2017a, b; Li et al. 2017; Ding et al. 2018). Investigating the EASM can improve our understanding of monsoon dynamics, its socioeconomic impacts, and short-term climate prediction over many parts of East Asia (Wang et al. 2008; Liu et al. 2019; Ha et al. 2020; Moon and Ha 2020).

A number of studies have examined variations of the EASM’s intensity, which can be measured in different perspectives (Tao and Chen 1987; Nakamura 1992; Wang et al. 2008; Ye and Chen 2019), such as low-level winds (Dai et al. 2000), wind shears (Lau and Yang 2000), and sea level pressure (SLP) gradient (Shi and Zhu 2015). The intensity of EASM exhibits both temporal and spatial variations and the related rainfall anomalies can induce floods and droughts in vast areas of East Asia (Wang 2001; Huang et al. 2007, 2017a, b; Ding et al. 2009; Ding et al. 2018; Li and Leung 2013; Xue et al. 2015; Liu et al. 2019; Yu et al. 2022). In addition to the intensity of EASM, the duration of EASM is also essential to weather and climate anomalies over East Asia. For example, the super Mei-yu in 2020 is characterized by a remarkably early onset (around 1 June), late withdrawal (around 1 August), and persistent precipitation during Mei-yu season, causing 178.96 billion CNY of direct economic losses in the Yangtze–Huaihe River basin (Volonté et al. 2020; Li et al. 2021; Chu et al. 2022). Therefore, detecting the timing of monsoon (i.e. the onset and retreat dates of monsoon) has important implications for society.

As suggested by Bombardi et al. (2020), the main methods used to detect the timing of monsoon can be separated into two types: local-scale methods and regional to large-scale methods. The local-scale methods characterize the timing of monsoon within a small region (an individual grid point or a station). For example, the global gridded dataset of the timing of the rainy seasons presented by Bombardi et al. (2019) is calculated at local scale (grid point). Meanwhile, the regional to large-scale methods usually calculate the spatially averaged variables in the monsoon domain, such as South America (Garcia and Kayano 2013) and India (Carvalho et al. 2016). Based on the time series of these spatially averaged variables, the large-scale evolution of the monsoon system can be revealed. For example, Garcia and Kayano (2013) used the area-averaged OLR anomaly as an index of the rainy season over the central Amazon and central Brazil regions. Both methods are complementary and offer different advantages in detecting the timing of monsoon. The local-scale methods can measure the spatial progression of monsoon system within the monsoon domain, such as the northward advance of the EASM front (Wang et al. 2021). However, due to the noise from small-scale variations, methods based on local-scale features might not fully represent the timing of monsoon, for example the “bogus monsoon onset” associated with synoptic variabilities (Fasullo and Webster 2003; Wang et al. 2008; Noska and Misra 2016). The regional to large scale methods that tailored to a certain monsoon system can reduce the spatial noise and better capture the signal of monsoon (Bombardi et al. 2020).

Additionally, Hu et al. (2019) noted that the timing of monsoon can be depicted from three perspectives, namely, the low-level winds, the warm-humid airmass, and the rainy season. In earlier studies, the low-level winds and the warm-humid airmass were often considered together to reveal the timing of monsoon (Wang 1999; Lv et al. 2011). Recently some studies have depicted the timing of monsoon by considering the perspective of the warm-humid airmass solely, such as the maximum in the tendency of vertically integrated moist static energy (Wang et al. 2021) and the meridional gradient of equivalent potential temperature (Volonté et al. 2021). In addition, Ding and Wang (2008) and Zhan (2016) simply used the precipitation as an indicator to define the timing of EASM. Currently, it is generally accepted that the northward advance of the EASM is stepwise (Huang et al. 2007; Ding and Wang 2008; Zhan et al. 2016; Ding et al. 2018; Volonté et al. 2021; Wang et al. 2021), and can be influenced by both tropical and mid-latitude processes (Huang et al. 2005; Ding and Chan 2005; Ding and Wang 2008; Kong et al. 2017; Kong and Chiang 2020; Volonté et al. 2021). In comparison to the monsoon onset, less attention has been paid to the EASM retreat.

So far, there is no study specialized in analyzing the retreat of EASM, and its characteristics largely remain unknown. A small number of studies suggested that the EASM retreat usually occurs in late-August to early-September, and is significantly faster than the northward advancement of the EASM, without lingering in any areas (Ding and Wang 2008; Lv et al. 2011; Zhan et al. 2016; Ding et al. 2018). However, most of these studies are based on the movement of rainy belt and local-scale methods. For example, Zhan et al. (2016) considered the pentad-averaged precipitation falling below 4.0 mm/day consistently for 2 pentads as the demise time of the EASM. It may be subjective and not sufficient to some degree: (1) the threshold value of 4.0 mm/day and lasting for 2 pentads may be debatable; (2) sometimes the spatial noise, such as heavy rain and relatively dry stage can largely influence the ending date of rainy season in a local area (Trenberth and Kevin 2006; Bombardi et al. 2020). Noting that the Asian monsoon is not only characterized with rapid alternation of dry and wet seasons but also a distinct seasonal reversal of wind (Webster et al. 1998; Ding and Chan 2005; Hu et al. 2019), using the low-level wind as an indicator in defining the EASM retreat may be more objective: due to the naturally existing threshold (0 m/s) for the direction of winds, no subjectively given threshold is required. Also, the East Asian monsoon is a large-scale system and methods based on the local scale may not well capture the rapid retreat of large-scale monsoon phenomena (Trenberth and Kevin 2006; Bombardi et al. 2020). Thus, a large-scale method is needed. In addition, traditional climatic calendar mean data cannot fully represent the features of monsoon retreat as there may be inter-annual and inter-decadal variabilities (Hu et al. 2019).

Therefore, in this paper, based on composited data (i.e. the average of changes around the retreat dates in individual years), we focus on the large-scale phenomenon of winds alteration, and analyze the climatological features of the EASM retreat systematically. In the following, Section "Data and method" describes the datasets and methods. Section "Results" presents the climatological characteristics of the EASM retreat, including the circulation changes and the related synoptic phenomena. Section "Summary and discussion" is the summary and discussion.

2 Data and method

2.1 Data

The datasets used in the present study are as follows: (1) daily (i.e. geopotential height, horizontal winds, vertical velocity, surface temperature, and sea level pressure) and monthly (i.e. horizontal winds and geopotential height) atmospheric data from the Japanese 55-year Reanalysis (JRA-55) project conducted by the Japan Meteorological Agency (Kobayashi et al. 2015) from 1958 to 2019, (2) pentad (5-day) and monthly precipitation analysis data of the Global Precipitation Climatology Project (GPCP; Xie et al. 2003; Alder et al. 2003) from 1979 to 2019, (3) tropical cyclone (TC) best-track data of the China Meteorological Administration (Ming et al. 2014) from 1958 to 2019. The horizontal resolutions are 1.25° × 1.25° for JRA-55; 2.5° × 2.5° for GPCP; and 0.1° × 0.1° for TC best-track from China Meteorological Administration.

2.2 Methods

As mentioned in the Introduction, the EASM retreat tends to occur in late-August to early-September, which is mainly featured by the southward withdrawal of rainy belt and large-scale winds reversal. Figure 1 shows the climatology of 850 hPa winds and rainfall in August, September, and the difference between September and August. In August (Fig. 1a), strong low-level southwesterly winds are dominant over the SCS and the Western North Pacific (WNP). The western Pacific subtropical high (WPSH) is distributed in a belt shape with ridge line extending from the south of Japan to central Pacific, and southeasterly winds are prevalent over its southern flank. In addition, the southerly and southeasterly winds are evident over coastal East Asia (box in Fig. 1). Strong rainfall with value exceeding 4 mm/day appears in most area of the East Asia continent. In September, the ridge line of the WPSH moves southward slightly, but the westernmost point remains stable. It is obvious that the low-level winds shift from southerly to northerly over the coastal and continental East Asia, and the rainy belt moves to the Southeast Asia and tropical WNP. From August to September (Fig. 1b–c), even though the movement of WPSH is subtle, the monsoonal retreat features are very pronounced. The amount of rainfall is significantly decreased over most area of East Asia. In the difference fields (Fig. 1c), an anticyclonic circulation dominates East Asia with distinct meridional winds reversal in eastern China as well as its neighboring ocean, while the southwesterly winds over the SCS and southeasterly winds over the WNP are significantly weakened. Furthermore, since the seasonal cycle of monsoon can be depicted by the winds reversal (i.e., summertime southwesterly and wintertime northeasterly), we further analyse the latitudinal evolution of 850 hPa winds over East Asia. As shown in Fig. 2, in early September, the winds shift rapidly from southerly to northerly, especially to the north of 20° N, indicating the seasonal transition of monsoonal circulation. Therefore, we focus on the transition of lower level circulation and define the largest center (120°–130° E, 20°–40° N) of the V850 difference between September and August as the key region of winds alteration.

Fig. 1
figure 1

Climatology of 850 hPa winds (averaged over 1958–2019, vectors, units in m s–1) and rainfall (averaged over 1979–2019, shading, units in mm day–1) in a August and b September. c The differences between (b) and (a). Only the winds difference significant at 95% confidence level are shown in (c). Red lines in (a) and (b) denote 5875 and 5880 gpm at 500 hPa, respectively. The yellow rectangles indicate the area (120°–130° E, 20°–40° N) over which the 850 hPa meridional winds are averaged (i.e., the key area of the meridional winds reversal)

Full size image
Fig. 2
figure 2

The latitude-time cross-section of climatological 850 hPa winds (vectors, units in m s−1) during 1958–2019 along the latitude belt (coastal East Asia, 120°–130° E) from April to November. Black and red vectors denote the southerly and northerly component of the meridional winds, respectively. The blue shading indicates the southerly component of the winds. The yellow dashed line highlights the reversal time of 850 hPa meridional winds

Full size image

As shown in Fig. 3a, even though the climatology of daily V850 (and its 7-day smoothed series) exhibits notable fluctuations, the withdrawal process in early September is especially transient with the winds rapidly shifting from southerly to northerly. For the yearly series, the daily V850 (and its 7-day smoothed series) shows high-frequency fluctuation, which makes it hard to figure out the specific transitional day. Thus, an objective and unified identification of the transitional day is needed. Previous studies have shown that accumulated series of atmospheric variables has less fluctuation and can be applied to investigate the timing of the SCSSM as well as the rainy seasons over other monsoonal regions (Liebmann and Marengo 2001; Cook and Buckley 2009; Noska and Misra 2016; Luo and Lin 2017; Bombardi 2017, 2019). Therefore, we accumulate the 7-day smoothed V850 starting from January 1 to every calendar day for every year (February 29 for leap years are omitted), and obtain daily cumulative V850 series (Fig. 3b). The equation is provided as follows:

$$V^{\prime}_{y} (d) = \sum\limits_{i = 1}^{d} {V_{y} (i)}$$
(1)
Fig. 3
figure 3

The daily time series of the region-average a 7-day smoothed 850 hPa meridional winds (V850, units in m s–1), b cumulative V850 (units in m s−1) within the region (20°–40° N, 120°–130° E) in 1958–2019, and c V850 and cumulative V850 in 2019. The red thick lines in (a) and (b) represent the climatological mean. The red cross in (c) denotes the maximum of the cumulative V850 in 2019

Full size image

where \(V^{\prime}_{y} (d)\) denotes the daily cumulative V850 for day d of year y, and the \({V}_{y}(i)\) denotes 7-day smoothed V850 for day i of year y. The daily cumulative V850 series for both climatology and each year exhibit less fluctuation than the original V850 series. Similar to Noska and Misra (2016) and Luo and Lin (2017), the retreat day of EASM is defined as the day when cumulative V850 reaches its absolute maximum. This criterion gives rise to objective characterization of the retreat date of EASM. For example, the retreat date of the EASM is September 9 in 2019 (Fig. 3c). On September 9 of 2019, the cumulative V850 reaches it absolute maximum and alters its signal from ascending to descending, suggesting the steady reversal of V850 from southerly to northerly.

The EASM retreat dates are calculated from the period of 1958 to 2019 based on the definition described above. Since the GPCP rainfall data is based on 73-pentad values every year, we transfer the retreat date into pentad series. The 73-pentad mean data are also constructed from the daily data for each year with the leap day omitted. In order to filter out the noises and highlight common climatic characteristic of the EASM retreat, we use climatological composited mean data to analyse the related phenomena in the following. The P–1, P0, and P+1 denote the time before, during, and after one pentad of monsoonal retreat, respectively. In this study, we determine P0 (i.e., the EASM retreat pentad) for every individual year and obtain the features based on average of changes around P0 in individual years.

After the TCs genesises, the activities and landfall of TCs often bring enormous disasters to the surrounding regions such as the Philippines, Vietnam, China, and Korea (Choi et al. 2017; 2019). Studies have shown that TCs genesises and activities are closely related with the EASM circulation (Zhong and Hu 2007; Wang et al. 2012; Chang et al. 2012; Choi et al. 2017; Wang and Wang 2019). The enhancement of the eddy kinetic energy (EKE) can supply more energy for the generation and development of TCs (Huangfu et al. 2017; Hu et al. 2020). Thus, we employ the EKE at 850 hPa, that is, \(\frac{1}{2}\left( {\overline{{u^{{\prime}{2}} }} + \overline{{v^{{\prime}{2}} }} } \right)\), to detect the genesises and activities of TCs. Here, \(u^{\prime}\) and \(v^{\prime}\) denote the 2.5–7-day Butterworth bandpass-filtered 850 hPa zonal and meridional winds, respectively. The overbars represent the pentad mean value of the winds.

Following previous studies, storm tracks can be depicted by various eddy variance and covariance quantities (Blackmon et al. 1977; Chang and Fu 2002; Athanasiadis et al. 2010). In this study, the storm track activity is calculated as the 2.5–7-day Butterworth bandpass-filtered 300 hPa geopotential height variance (referred to as \(\overline{{Z^{\prime}Z^{\prime}}}_{300}\), where the apostrophe indicates the filtered anomalies and the overbar denotes the pentad mean). A two-sided Student’s t-test is used to assess the statistical significance of the composited differences.

3 Results

3.1 Verification of the EASM retreat dates

Since the definition of EASM retreat is characterized as V850 reversal, we use V850 to verify the time series quality. Figure 4 provides the time series of EASM retreat as well as the area-average V850 over the key region (120°–130° E, 20°–40° N; box in Fig. 1) during 1958–2019. Slight changes of the region lead to very similar result. The winds direction changes from southerly to northerly as soon as the monsoon retreats and remains stable afterwards. Even though in some years, there is slight southerly winds disturbance after the monsoon withdrawal, it is transient and disappears quickly. In addition, the climatological retreat date is September 6 and the standard deviation (SD) is approximately 16 days, thus the V850 averaged over August 21 to September 22 (i.e., within one SD) is used to reveal the variation of V850 during the EASM retreat periods. The retreat date and the averaged V850 have a positive correlation of 0.64 (significant at 99% confidence level), indicating that a late retreat year is often associated with anomalously strong southerlies during the climatological monsoon retreat period and vice versa. Above all, it shows that the definition of EASM retreat is reliable and we can make further analysis on the basis of it.

Fig. 4
figure 4

Time series of the EASM retreat date from 1958 to 2019. The solid line denotes the climatological mean of the EASM retreat time, and the dashed lines represent the standard deviation of the EASM retreat time series. The regionally (20°–40° N, 120°–130° E) averaged 850 hPa meridional winds (units in m s–1) are shaded

Full size image

3.2 Changes in atmospheric circulation during the EASM retreat periods

Figure 5 displays the composited 850 hPa winds and rainfall during monsoonal retreat periods. Before the EASM retreats (Fig. 5a), a large anticyclonic circulation dominates the subtropical western Pacific along with strong southerly winds over the neighboring ocean of East China. A cyclonic circulation is located around the SCS with strong southwesterly winds prevailing over the Indochina Peninsula and the southern SCS. The northerlies from mid-high latitudes of Asia intensify and invade equatorward in P0 (Fig. 5b). A strong anticyclonic circulation is established over East Asia in P+1. Winds over the neighboring ocean of East China shift rapidly from southerly to northerly and the strong southwesterly winds over the SCS also weakened and retreated southwards. Since previous studies usually applied the movement of rainy belt to depict monsoonal withdrawal (e.g., Wang and LinHo 2002; Ding and Wang 2008; Zhan et al. 2016), here we provide the rainfall evolution during monsoon retreat periods as well. Corresponding to the circulation evolution, the amount of rainfall decreases observably and the rainy belt moves southward to Southeast Asia after monsoon withdrawal (Fig. 5b–c). It can also be seen from Fig. 5d that the difference between after and before the EASM retreat displays an anticyclonic circulation over East Asia with prevalence of the southerly winds along the coastal area of eastern China. To its east, a cyclone appears with a center located to the south of Japan. Concurrently, the rainfall significantly decreases over coastal East Asia and increases over western Pacific with one center located slightly to the east of the cyclone and the other located over the Mariana Islands, suggesting the inherent coupling between circulation and rainfall, which will be discussed later.

Fig. 5
figure 5

The climatology of composited 850 hPa winds (averaged over 1958–2019, vectors, units in m s–1) and precipitation (averaged over 1979–2019, shading, unit in mm day–1) for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a), including 850 hPa winds (only those significant at 95% confidence level are shown) and precipitation (those significant at 95% confidence level are dotted), which can reveal the changes during the EASM retreat. The yellow rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

At middle troposphere (i.e., 500 hPa), the circulation systems are coherent with those at the lower level. Figure 6a–c displays the climatology of 500 hPa vertical velocity and geopotential height in P–1, P0, and P+1, respectively. The difference between P+1 and P–1 is shown in Fig. 6d. During the monsoon retreat period (Fig. 6a–c), the WPSH migrates eastward, a low trough over Northeast Asia deepens and intrudes southward, and the descending (ascending) motion strengthens in north and northeastern China (subtropical western Pacific). The difference map shows that a large northwest-southeast low trough extends from northeastern China and the Sea of Japan to subtropical western Pacific, along with strengthened ascending motion in the front of the trough and descending motion in the rear of the trough after the EASM retreat (Fig. 6d).

Fig. 6
figure 6

The climatology of the 500 hPa vertical velocity (shading, units in Pa s–1) and geopotential height (contours, unit in gpm) for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a). The significant differences (at 95% confidence level) of the 500 hPa vertical velocity are dotted in (d). Only the 500 hPa geopotential height differences significant at 95% confidence level are shown in (d). The yellow rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

As suggested by Ding (1994), the thermal and vorticity advection by the upper level trough is important in triggering the rising motion. Thus more insight can be gained by exploring the so-called omega equation, which can be decomposed into three terms: the changes of vorticity advection with pressure, the Laplacian of thermal advection, and the Laplacian of atmospheric diabatic heating. The omega equation offers a good qualitative description of the vertical motion field on a large scale (Holton and Hakim 2013). Taking the ascending motion as an example: in the difference fields of the lower troposphere (Fig. 5d), a cyclone appears in the subtropical western Pacific with a small amount of vorticity advection, while positive vorticity advection prevails in the front of the upper-level trough. The vorticity advection increases (decreases) with height (pressure), favoring the upward motion over the subtropical western Pacific. The warm advection in the front of the trough and the released latent heating resulting from increased precipitation are favorable for the upward motion as well. Similarly, the decreasing (increasing) vorticity advection with height (pressure), the cold advection, and the diabatic cooling due to the reduction of precipitation can lead to downward motion in the rear of the trough, which is located in the coastal area of East China.

In the upper troposphere (i.e., 200 hPa), the subtropical westerly jet accelerates and intrudes southward, reaching a speed above 35 m/s after one pentad of monsoon retreat (Fig. 7a–c). For the differences between Fig. 7c and Fig. 7a (i.e., the difference between after and before one pentad of monsoon retreat), despite the strengthening of the subtropical westerly jet, it is interesting to note that to the west of the 500 hPa trough, the difference of the upper troposphere shows a trough as well (Fig. 7d), indicating a structure of baroclinic growth (Holton and Hakim 2013). Also, since the divergent circulation can well reflect the distribution of atmospheric cold and heat sources in the planetary scale (Zhang et al. 1996; Liu et al. 2019), the divergent flow and velocity potential at 200 hPa are provided in Fig. 8. Figure 8a–c displays that the divergent center gradually decreases and moves from the SCS and the Philippines Sea to the Caroline Islands from P–1 to P+1. After the monsoon retreat, the velocity potential over East Asia significantly increases with the largest center located around the southeast of China. In the difference fields (Fig. 8d), the divergent winds are significant from the west Pacific and northwestern China to eastern and northeastern China, and the divergent easterly winds are especially strong over the coastal area of eastern China. Correspondingly, the 850 hPa divergent circulation (not shown) exhibits opposite alteration, with decreased velocity potential and reversed divergent zonal winds over East Asia, indicating the changes of zonal land-sea thermal contrast after monsoonal withdrawal.

Fig. 7
figure 7

The climatology of the 200 hPa geopotential height (contours, units in gpm) and zonal winds (shading, units in m s–1) for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a). The significant differences (at 95% confidence level) of 200 hPa zonal winds are dotted in (d). Only the significant differences (at 95% confidence level) of the 200 hPa geopotential height (black dashed lines, contour interval at 40 gpm with zero lines omitted) are shown in (d). The yellow rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image
Fig. 8
figure 8

The climatology of velocity potential (shading, units in 106 m2 s–1), and divergent winds (vectors, units in m s–1) at 200 hPa for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a). Only the significant differences (at 95% confidence level) of the 200 hPa divergence winds are shown in (d). The significant differences (at 95% confidence level) of the velocity potential are dotted in (d). The yellow rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

After analyzing the circulation in the lower, middle, and upper troposphere separately, it is crucial to investigate its vertical structure. Following previous studies (Chen et al. 2014; Hu et al. 2019), we use divergent winds that are directly associated with vertical motion to depict the vertical circulation. In addition, since both the 500 hPa vertical velocity (Fig. 6d) and divergent winds (Fig. 8d) exhibit zonal difference, we provide meridional averaged (20°–40° N) values. Figure 9a–c displays the climatology of composited divergent zonal winds and vertical velocities in P–1, P0, and P+1, respectively. Figure 9d shows the difference between P+1 and P–1. In P–1 (i.e., before the EASM retreat), distinct ascending motion dominates the East Asia continent, with a center of action near mid-to-upper troposphere over the neighboring ocean of East China (Fig. 9a). Accompanying with the above ascending motion, the divergent flows in both the upper-and-lower level are divided into two branches. In P0 (i.e., the monsoon retreats, Fig. 9b), the primary upward motion disappears immediately. After one pentad of the monsoon withdrawal, the upward motion migrates eastwards, a weak sinking motion occurs over the coastal area of eastern China, and the branches of divergent flow are displaced by a single but intensified divergent flows (Fig. 9c), indicating the formation of the local vertical circulation. The difference map between P+1 and P–1 clearly shows a vertical circulation cell with downward motion over the eastern China as well as its neighboring ocean, upward motion in the subtropical western Pacific, and divergent easterly and westerly winds in upper and lower troposphere, respectively (Fig. 9d).

Fig. 9
figure 9

The climatology of the local zonal circulation cell, depicted by the vertical velocities (units in Pa s–1) and the divergent zonal winds (units in m s–1) averaged in 20°–40° N, for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a). The zonal circulation cell is depicted by the vertical velocities (units in Pa s–1) and the divergent zonal winds (units in m s–1). The shading in (a)–(d) denote the vertical velocities (units in Pa s–1). Only the significant differences (at 95% confidence level) of vertical velocities and the divergent zonal winds are shown in (d)

Full size image

3.3 Precipitation, surface temperature, and synoptic phenomena associated with the EASM retreat

Changes in the large-scale circulation related to the EASM retreat have been examined above. In this subsection, we further investigated precipitation, surface temperature, and synoptic-scale phenomena including the TCs activities and storm tracks activities (STA) during the EASM retreat periods.

After the monsoonal retreat, significant decrease in rainfall is observed over coastal area of East China and marked increase in rainfall is found over subtropical western Pacific as shown in Fig. 5d. The three-dimensional circulation is coherent and is conducive to rainfall difference. Figure 10a–c displays the SLP and surface temperature evolution in P–1, P0 and P+1, respectively. The difference between P+1 and P–1 is shown in Fig. 10d. After the EASM retreat, the East Asia continent is dominated by a cold high, the continental SLP is significantly higher and the difference map shows a low center over the subtropical western Pacific (Fig. 10c–d). Also, it is distinctly colder over the whole East Asia region and the cooling of surface temperature is especially large over the northeastern China (Fig. 10d). The changes of SLP and surface temperature coincide well with the difference of three-dimensional circulation after and before the EASM retreat: The cold advection in the back of 500 hPa trough and the northerly winds in 850 hPa bring colder air from higher latitudes to East Asia, leading to a broadscale decrease in surface temperature. Furthermore, the cold air mass with higher density may enhance both the continental high and the associated zonal land-sea pressure gradient, thus favoring the strengthening of northerly winds along the coastal East Asia.

Fig. 10
figure 10

The climatology of surface temperature (shading, units in ℃) and SLP (contours, units in hPa) for one pentad a before, b during, and c after the EASM retreat. d The difference between (c) and (a). The significant differences (at 95% confidence level) of the surface temperature are dotted in (d). Only the significant differences (at 95% confidence level) of the SLP (yellow lines, contour interval at 1 hPa with zero lines omitted) are shown in (d). The black rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

In addition, studies suggested that the WPSH and the subtropical westerly jet that are essential components of the EASM can alter the vertical wind shear and thereby influence the generation and development of TCs (Wang et al. 2012; Wang and Wang 2019). Also, the EASM circulation can regulate the tracks of TCs via modulating the location of WPSH (Zhong and Hu 2007; Wang et al. 2012; Chang et al. 2012; Choi et al. 2017). For example, when the WPSH shrink eastward, the middle and lower level winds (i.e. southeasterly and southwesterly winds) around the frank of the WPSH will work as steering flows and make the TCs head northward. In our study, after the EASM retreat, the WPSH shrinks eastward and the subtropical westerly jet enhances and intrudes southward significantly. We wonder if there are any changes in TCs genesises and activities during the EASM retreat periods. Figure 11a–b shows the TCs activities before and after the monsoon retreat, and Fig. 11c provides the 850 hPa EKE difference between P+1 and P–1, which can well represent the changes of synoptic-scale eddies activities in the lower level (Huangfu et al. 2021). After monsoonal retreat, due to the steering flows in middle and lower levels, less TCs reach the land of East Asia while much more TCs advance northward. The changes in TCs activities are corresponding to the 850 hPa EKE difference: the EKE is reduced in southeast China and the East China Sea while increased over the subtropical western Pacific (Fig. 11c). The changes in TCs activities can also benefit the rainfall pattern: the upward motion accompanying with cyclonic circulation and abundant moisture brought by TCs can be conducive to precipitation increase and vice versa (Chen et al. 2017).

Fig. 11
figure 11

The TC tracks (from 1958 to 2019) for one pentad a before and b after the EASM retreat. The red curves in (a) and (b) are the contour of 5,880 gpm at 500 hPa. c The difference of 850 hPa EKE (shading, units in m2 s–2) between one pentad after and one pentad before the EASM retreat. Only the significant differences (at 95% confidence level) of the 850 hPa EKE are dotted in (c). The rectangles denote the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

Storm tracks are generally described as the preferred migratory pathways of extratropical cyclone propagation (Blackmon et al. 1977) and changes in STA have a substantial impact on weather and climate by influencing the transport of water vapor, heat, and momentum (Shaw et al. 2016). Studies have shown that the STA are weakest in summer and strongest in winter, and that the STA are strongly coupled with the subtropical westerly jet (Chang 2001; Harnik and Chang 2004). Our study has revealed that the westerly jet accelerates and intrudes southward after the EASM retreat (Fig. 7). Hence, we wonder if the STA will enhance together with the westerly after the EASM retreat, which is considered to occur at the end of summer. Figure 12 presents the changes of STA between P+1 and P–1. Similar to previous studies (Nakamura 1992; Deng and Sun 1994; Chang 2001; Harnik and Chang 2004), the STA are much more active over the Northeast Asia and the North Pacific Ocean, and are slightly poleward toward the jet streams in the difference fields (Fig. 7d), suggesting the intrinsic coupling between the jet and STA (Chang 2001; Harnik and Chang 2004).

Fig. 12
figure 12

The difference of the storm tracks (shading, units in m2) between one pentad after the EASM retreat and one pentad before the EASM retreat. Only the significant differences (at 95% confidence level) are dotted. The rectangle denotes the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E)

Full size image

3.4 Shifts in different atmospheric variables

Figure 13 displays the time series of several atmospheric variables from 10 pentads before to 10 pentads after the EASM retreat, including the V850 over the key region of winds reversal (20°–40° N, 120°–130° E), the 200 hPa zonal winds over the mid-latitudes (30°–45° N, 100°–150° E), which can reveal the strengthening of subtropical westerly jet, and other variables including precipitation, 500 hPa geopotential height and surface temperature over the East Asia continent (20°–50° N, 105°–130° E). All of these variables experience remarkable shift during the EASM retreat. And it is interesting to note that even though we apply the winds reversal at sea to obtain the EASM retreat dates, other variables over the continent also exhibit significant changes during monsoon retreat periods, revealing the spatial aggregation and synchronization of the broadscale monsoon system. In the lower level, the V850 shifts rapidly from southerly to northerly once the monsoon retreat, the divergence of vertically integrated moisture transport (VIMT) shifts from negative to positive, suggesting a less transport of moisture over East Asia continent. Also, as suggested by Zhan et al. (2016), the threshold of 4 mm/day can be used as a criteria for the demise of rainy season. Correspondingly, the precipitation shows an abrupt drop and remains below 4 mm/day after the EASM retreat, indicating the synchronization of rainfall and circulation. In addition, the surface level pressure is becoming higher and the surface temperature is persistently falling. In the middle troposphere, the decreasing 500 hPa geopotential height indicates a deepening of mid-level low trough above the East Asia continent, the pressure vertical velocity exhibits opposite direction and the descending motion is continuously strengthening after the EASM retreat. In the higher level, the subtropical jet changes from easterly to westerly and gradually intensifies after monsoon retreat. The 200 hPa geopotential height is decreasing and the velocity potential is increasing, indicating the alteration of land-sea thermal difference during monsoonal retreat. It can be concluded that the evolution of these variables can not only reveal the coherent three-dimensional circulation but also suggest the inherent coupling between circulation and synoptic phenomena.

Fig. 13
figure 13

Climatology of the composited a the 850 hPa meridional winds (V850, black dashed line) and the 200 hPa zonal winds (U200, blue solid line), b the precipitation (black dashed line) and the divergence of vertically integrated moisture transport (Div WVT, blue solid line), c the surface temperature (Ts, black dashed line) and sea level pressure (SLP, blue solid line), d the 500 hPa geopotential height (Hgt500, black dashed line) and the 500 hPa vertical velocity (Omega500, blue solid line), and e the 200 hPa geopotential height (Hgt200, black dashed line) and 200 hPa velocity potential (Chi200, blue solid line) from 10 pentads before to 10 pentads after the EASM retreat. The V850 is averaged in the key region of 850 hPa meridional winds reversal (20°–40° N, 120°–130° E), the U200 is averaged in (30°–45°N, 100°–150° E), while other elements are averaged in (20°–50° N, 105°–130° E)

Full size image

4 Summary and discussion

Based on the direction of 850 hPa meridional winds over the coastal East Asia, we apply the absolute maximum of cumulative V850 to identify the winds reversal date and obtain the time series of the EASM retreat date from 1958 to 2019. After that, we use climatological composited data to depict the changes of three-dimensional circulation and synoptic phenomena during monsoon retreat periods. Based on the characteristics of the tropospheric circulation and the accompanying weather phenomena, we pay specific attention to the impacts of circulation on the rainfall and surface temperature and the results are as follows:

Figure 14 shows the summary for the difference fields after and before the EASM retreat. In the lower level, the anticyclone (cyclone) over East Asia continent (the western subtropical Pacific) suppresses (favors) the transport of moisture over the SCS and the WNP thus favoring the decrease (increase) of precipitation there. And the shifts of circulation in the lower level also work as steering flows and lead the TCs activities, the upward motion and moisture associated with TCs contributes to the increase of rainfall over the western subtropical Pacific. Furthermore, the northerlies help the southward intrusion of cold air, resulting in the broadscale cooling over East Asia.

Fig. 14
figure 14

The schematic diagram of the three-dimensional atmospheric circulation and the synoptic impacts in association with the EASM retreat. Here, STWJ, WPSH, VIMT, Ts, and TC means subtropical westerly jet, western Pacific subtropical high, vertically integrated moisture transport, surface temperature, and tropical cyclones, respectively

Full size image

In the middle troposphere, the WPSH moves eastward, a low trough from Northeast Asia appears and extends southward. The advection of dry and cold air (moist and warm air) and the negative (positive) vorticity advection in the rear (front) of the trough are favorable for the sinking (ascending) motion, contributing to the decrease (increase) of rainfall in the East Asia continent (the subtropical western Pacific). In the meanwhile, the cold air advection in the rear of the trough will also benefit the cooling over East Asia.

In the 200 hPa field, there exists a trough as well but it is slightly west to the 500 hPa trough, suggesting a typical structure of baroclinic generation. The subtropical westerly jet intensifies continuously and the divergent easterly winds are remarkable over the neighboring ocean of East China.

Climatological circulation characteristics of the EASM retreat as well as its influence on precipitation, surface temperature, and the synoptic phenomena are analyzed in this paper, however, there are still some interesting points to discuss further: Wang and LinHo (2002) noted that the heat source associated with rainfall can drive monsoonal circulation and the circulation is tightly coupled with rainfall. However, in this analysis we only focus on how circulation influences the precipitation and the impacts of precipitation on circulation deserve further discussion. Here, we offer some preliminary hypothesis: based on the Gill model (Gill 1980), the cyclonic circulation over western subtropical Pacific in the difference fields of 850-hPa winds may be regarded as a Rossby wave response to the latent heating caused by increased rainfall in the local area (Fig. 5d). The Rossby wave dynamics may be crucial during the EASM retreat periods. In addition, Qi et al. (2008) pointed out that the precipitation can be conducive to the zonal thermal contrast between Asian continent and the West Pacific that drives the seasonal march of EASM. However, to what extent does the thermal difference that related to rainfall drive the retreat of EASM remains unknown.