1 Introduction

The Asian summer monsoon features a prominent seasonal transition of prevailing winds and an abrupt change from dry to wet climate. Usually the Asian summer monsoon includes the tropical summer monsoon and the subtropical East Asian summer monsoon (EASM). The EASM is a distinctive and prominent regional component of the grand Asian summer monsoon system (Tao and Chen 1987; Chen et al. 1991; Ding 1994; Chang et al. 2000; Lau et al. 2000; Wang and LinHo 2002; Ding and Chan 2005). Unlike the tropical summer monsoon such as the Indian summer monsoon that is characterized by low-level westerly wind over Indian Peninsula, the EASM is characterized by humid low-level southerly wind that prevails over eastern China, Korea and Japan in summer. The strong monsoon flow brings abundant water vapor into East Asian continent that converges and forms a large-scale rain belt (Meiyu/Baiu/Changma) in East Asia. The EASM undergoes substantial variabilities on various timescales that cause severe droughts and floods that draw much attention in the region (Huang et al. 2003; Yang et al. 2005; Xu et al. 2005; Zhu and Yang 2003; Huang et al. 2007; Ding 2007; Yang and Zhu 2008).

A large amount of literatures have been contributed to the understanding of formation mechanisms responsible for the Asian summer monsoon. Two types of the mechanisms are widely considered to be fundamental: the seasonal variation of solar radiation and the inhomogeneity of the Earth’s surface. The seasonal variation of solar radiation is always viewed as a primary driver of the monsoon (Tao and Chen 1987; Ding 1994; Lau et al. 2000). Zeng and Li (2002) believed that because of the angle between the equatorial plane and the ecliptic plane, the solar radiation varies annually, serving to the seasonal transition of the general circulation and the formation of cross-equatorial current. And the cross-equatorial current from the Southern Hemisphere constitutes the origins of the Asian summer monsoon.

The inhomogeneity of the Earth’s surface, on the other hand, is responsible for the regionality of the Asian summer monsoon. It is widely recognized that the drastic change of thermal contrast between the land and the ocean has a causal relationship to the onset of EASM and Indian summer monsoon (Murakami and Ding 1982; Luo and Yanai 1983; Chen et al. 1991). Qian et al. (2004) demonstrated that the continental warming plays a more important role than the oceanic warming in the onset of tropical monsoon. Liang et al. (2006) conducted a series of idealized experiments to investigate the importance of land-sea contrast, and found that the monsoon disappears in an aqua-planet case and that the extension of the subtropical continent into the tropics intensifies summer convective activities over the subtropical land.

Another mechanism exclusively for the Asian summer monsoon circulation formation is associated with the direct influence of the dynamical forcing as well as the thermal forcing by the huge elevated sensible heating of the Tibetan Plateau (Ye 1981, 1982; Ye and Wu 1998; Wu and Zhang 1998; Ueda and Yasunari 1998). This was explored by the early study of Hahn and Manabe (1975), in which the mountains were excluded. Their model showed that without mountains Asian summer monsoon was not found, and humid southerly flow does not extend as far as into northern and central India, but locates near the equator.

As ones place emphases on the role of the land-surface sensible heat in triggering monsoon onset (Ye and Wu 1998; Wu et al. 2007), there are increasing evidences showing that the atmospheric internal moisture process feedback, i.e., the latent heat release or condensational heating due to monsoonal rainfall, could also make contribution to the Asian summer monsoon formation. However, most of previous studies examined this impact only on the tropical monsoon. Early studies showed that the moisture process feedback could enhance the South Asian High, an important member of the Asian summer monsoon system (Li and Luo 1988; He et al. 1989). Recent studies have proposed that the condensational heating can excite a low-level cyclone and an upper-level anticyclone to the west of the heating center, and opposite circulation to the east (Wu and Liu 1998; Wu et al. 1999; Liu et al. 2002). The condensational heating over the tropical monsoon region including the Bay of Bengal and Indochina Peninsula was found to serve to the seasonal transition of the meridional temperature gradient over South China Sea and the weakening of the subtropical anticyclone in the lower troposphere through the Rossby wave chain which are both in favor of onset of the South China Sea summer monsoon (Liu et al. 2002; Wen et al. 2004).

The EASM brings out a large amount of rainfall over the subtropical East Asian region and consequently generates substantial latent heat release that could heat the atmosphere and feedback to the EASM itself. However this feedback effect has not been well explored and isolated with atmospheric General Circulation Model (GCM) experiments before. To what extent and how the condensational heating can feedback to the EASM are required to elucidate. This study tries to identify the impact of condensational heating caused by the EASM rainfall on the EASM circulation itself by contrasting two ensembles of atmospheric GCM experiments respectively with and without feedback of the condensational heating in East Asia. The paper is organized as follows. The model and experimental design are described in Sect. 2. Since the condensational heating is largely determined by the rainfall, a verification of the model performance in simulating EASM rainfall is presented in Sect. 3. The impacts of condensational heating on the land-sea thermal contrast determining the EASM, and further on the EASM circulation and dynamical structure inferred from these numerical experiments are depicted in Sects. 4–6, respectively. Final section is devoted to conclusions and discussion.

2 Model and experimental design

The model used in this study is the Community Climate Model version 3 (CCM3) developed at the National Center for Atmospheric Research (NCAR). It is a global spectral model with a horizontal resolution of 2.8° latitude by 2.8° longitude (a T42 horizontal spectral resolution), 18 vertical levels and a time step of 20 min. A detailed description of the model dynamical core and physical process parameterizations can be found in Kiehl et al. (1996) and Kiehl et al. (1998). The model is capable of producing a stable and realistic climate, especially for large-scale features of the Asian monsoon climate (Liu et al. 2002).

First, a 10-member ensemble of experiments with CCM3 was conducted with identical climatological seasonally-varying daily sea surface temperature forcing as a control run (hereinafter referred to as the CTRL run). Initial conditions for the ensemble were taken from simulated atmospheric states for 1 December through 10 December, spaced 1 day apart, with climatological time-varying SST forcing. Each ensemble member was integrated for roughly 400 days from its initial date through 31 December next year. Then, to identify the effect of condensational heating, a parallel ensemble of experiments was conducted as a sensitivity run (hereinafter referred to as the exDiab run) in which everything is exactly the same as the CTRL run except for the feedback on the atmosphere of the condensational heating that was switched off from 1 January through 31 December over the East Asian domain (90°E–125°E, 20°N–50°N). We choose this domain because the EASM rainfall occurs mostly over the region. To turn off the feedback of condensational heating in the exDiab experiment, the diabatic heating rate caused by all of the condensational processes over the domain is set to be zero at the right hand side of the thermodynamical equation.

The difference between the exDiab run and the CTRL run indicates the effect of condensational heating. To eliminate the impact of initial conditions, the ensemble mean for each run was used for all of the following analyses and the output of the first month was excluded with focus on the feedback of the condensational heating from January through December (pentads 1–73), especially for summer (June–July–August, JJA) season (pentads 31–48).

3 Verification of simulated EASM rainfall

Since the condensational heating is directly associated with the rainfall, it is necessary to verify the model performance in simulating rainfall over East Asia. An observational precipitation dataset for 1979–2007 that was taken from the NOAA CPC Merged Analysis of Precipitation (CMAP) is used for the verification. Figure 1a shows the observed climatological raining rate averaged within 110°E–120°E, as a latitude-time (pentad) plot. There is a persistent rain starting from very early year through pentad 28 that occurs over the subtropical East Asia. This phenomenon is the so-called South China spring rain (Tian and Yasunari 1998; Wan and Wu 2007, 2009; LinHo et al. 2008). With the South China Sea summer monsoon onset around pentad 28 (Wang and LinHo 2002; Ding and Chan 2005; Ding 2007), an increasing rain occurs over southern China with a rate of 8–9 mm per day and then progresses northward, reaching the Yangtze River valley or central eastern China around the period from late June to early July (usually referred to as the Meiyu period) and afterward reaching northern China around the period from late July to early August. Most of Chinese researches considered the rain from pentad 28 throughout whole summer as the subtropical EASM rain. Besides the northward progressive EASM rain, a strong tropical monsoon rain over South China Sea persists from pentad 28 through entire summer as well. Obviously, the subtropical EASM rain appears to separate from the tropical monsoon rain from late summer.

Fig. 1
figure 1

Latitude-time (pentad) section of raining rate (mm/day) averaged within 110°E–120°E for a climatological CMAP analysis, b the CTRL run, and c the exDiab run. The period between two red dashed lines at pentad 31 and 48, respectively, roughly represents summer (June–July–August) time

Full size image

In comparison with the observation shown in Fig. 1a, Fig. 1b exhibits the latitude-time distribution of simulated raining rate along 110°E–120°E for the CTRL run. Generally, the model simulates a subtropical East Asian rain pattern similar to the observation. Simulated South China spring rain that occurred before pentad 28 coincides well with the observation in either timing or intensity. The model reproduces well a northward progression of the EASM rain from southern China to the Yangtze River valley and then to northern China, successively, starting from pentad 28 through following summer (pentads 31–48), despite the fact that the simulated EASM rain during summer appears to be 2–3 mm per day weaker than the observed. Certainly, the model also exhibits some imperfect performances that lie in the following two aspects. One is that the simulated EASM rain extends northward too far so that the simulated rain over northern China is more than the observed. This is a common drawback of most of the atmospheric GCMs in simulating East Asian summer climate. The other is that the tropical monsoon rain is simulated with a southward shift so that it remains too apart from the subtropical EASM rain.

4 Impact on the land-sea thermal contrast

Persistent and abundant rainfall over the subtropical East Asia can generate substantial latent heat release to contribute to diabatic heating over the region. Here we calculate diabatic heating by integrating heating rate vertically throughout the troposphere, that is,

$$ \left\langle Q \right\rangle = \frac{{c_{p} }}{g}\int\limits_{{p_{t} }}^{{p_{s} }} {Qdp} , $$
(1)

where Q is the diabatic heating rate, g is the gravity acceleration, p s the surface pressure, and p t the pressure at the top of troposphere (here 100 hPa is used). Figure 2 displays the latitude-time distribution of vertically-integrated diabatic heating averaged within 90°E–125°E. As shown in Fig. 2a, for the CTRL run, striking positive diabatic heating with amplitude larger than 100 wm−2 can be observed to persist up to two seasons long over most of the subtropical East Asian regions. The heating starts from very early year and then increases in amplitude with time. The positive heating center also appears to expand northward, which is found to be significantly associated with the northward progression of the EASM rain as shown in Fig. 1b.

Fig. 2
figure 2

Latitude-time (pentad) section of vertically-integrated diabatic heating (w/m2) averaged within 90°E–125°E for a the CTRL run and b the exDiab run

Full size image

However, as the feedback of condensational heating was switched off over the East Asian domain (90°E–125°E, 20°N–50°N) in the exDiab run, the positive diabatic heating disappears over most of the East Asian regions, as shown in Fig. 2b. Instead, negative diabatic heating (i.e., cooling) can be observed therein. This happens especially after the EASM startup which is conventionally indicated by the onset of South China Sea summer monsoon around pentad 28 (Wang and LinHo 2002; Ding and Chan 2005; Ding 2007). Contrasting the diabatic heating between the two runs suggests that the monsoon rain-induced condensational heating absolutely dominates positive diabatic heating over the subtropical EASM region. In addition, it is worth while to note that during the South China spring rain period (say, pentad 8–28), there is a weak positive heating around 30°N, as seen in Fig. 2b, that is, the switch-off of condensation heating does not completely reverse the sign of diabatic heating. This implies that besides the condensational heating, other diabatic processes (say, the land surface sensible heat flux) can also make contribution to the diabatic heating during that period.

The EASM is substantially driven by the land-sea thermal contrast that is largely determined by the warming over land (Murakami and Ding 1982; Chen et al. 1991; Qi et al. 2007). Figure 3 shows the horizontal distribution of 500–200 hPa vertically-integrated temperature deviations respectively from its zonal mean and meridional mean during summer (JJA). For the CTRL run, the model captures well zonal and meridional land-sea thermal contrast in the East Asian sector, as illustrated in Fig. 3a, c. Due to the different thermal properties, the air column over the East Asian continent warms up more rapidly than the surrounding oceans, accompanying with the annual variation of solar radiation and the heating by Tibetan Plateau in spring (Wu and Zhang 1998; Ueda and Yasunari 1998). By summer, the tropospheric atmosphere over the East Asian continent is considerably (up to 5 K) warmer than the subtropical North Pacific and the tropical Indian Ocean. The zonal thermal contrast is characterized by the warmest center over Tibetan Plateau and the coolest center over the subtropical North Pacific, as shown in Fig. 3a, while the meridional thermal contrast is characterized by the warmest center over Tibetan Plateau and the coolest center over the equatorial Indian Ocean, as shown in Fig. 3c. As mentioned in the introduction, the warmest center over Tibetan Plateau is attributed to the huge elevated sensible heating (Ye 1981, 1982; Ueda and Yasunari 1998; Wu and Zhang 1998).

Fig. 3
figure 3

Horizontal distribution of 500–200 hPa vertically-integrated temperature deviations (K) respectively from its zonal mean (left panels) and meridional mean (right panels) during summer (JJA) for the CTRL run (upper panels) and the exDiab run (lower panels)

Full size image

Since condensational heating can considerably determine diabatic heating of the atmosphere over East Asia, as inferred from Fig. 2, the question now is to what extent such a heating can affect the land-sea thermal contrast in the EASM region and eventually affect the EASM itself. As the feedback of condensational heating associated with the EASM rainfall is switched off in the exDiab run, the land-sea thermal contrast is found to be altered significantly, as seen in Fig. 3b, d. Although the spatial pattern of zonal temperature deviations (Fig. 3b) remains similar to that in the CTRL run (Fig. 3a), the warmer region exhibits to move westward while the cooler region intrudes to the East Asian continent. As a result, the zonal thermal contrast is greatly reduced by 2–3 K.

The most significant change induced by the switch-off of condensational heating feedback lies in the meridional thermal contrast (Fig. 3c, d). While major positive meridional temperature deviations locate over the subtropical East Asia (~30°N) in the CTRL run (Fig. 3c), they almost disappear, in particular over Tibetan Plateau, in the exDiab run (Fig. 3d) and instead the negative meridional temperature deviations occur north of 30°N. As a result, the spatial pattern of meridional temperature deviations is characterized by a warmer zone along 20°N and a cooler zone north of it and south of it, respectively. Hence the meridional thermal contrast along 30°N is totally reversed, while it is greatly reduced along 10°N. In terms of thermal wind relation, these changes in the land-sea thermal contrast determine associated atmospheric circulation change.

5 Impact on the EASM circulation

5.1 Horizontal circulation

One of unique features of the EASM circulation is that the prevailing low-level winds reverse from winter northerlies to summer southerlies, whereas the upper-level winds transit primarily from winter southerlies to summer northerlies (Wang and LinHo 2002). The distribution of simulated 850 hPa winds and geopotential height in summer as shown in Fig. 4a displays basic aspects of the Asian summer monsoon circulation at lower levels. The cross-equatorial Somalia jet along 45°E–60°E forms intensive low-level westerly flows along 10°N–15°N around Indian Peninsula which extend to the Bay of Bengal, Indochina Peninsula, South China Sea, and even to northwestern tropical Pacific. Along the low-level westerly, three monsoon troughs are formed near the Bay of Bengal, South China Sea and northwestern tropical Pacific, respectively. The cross-equatorial flows along 100°E–130°E join in above low-level westerly flows contributing to the latter two monsoon troughs. This flow pattern constitutes a low-level component of the tropical summer monsoon circulation. Meanwhile, the subtropical high located at western Pacific extends westward to eastern China and forms strong low-level southerly flow, a low-level key component of the EASM circulation.

Fig. 4
figure 4

Horizontal distribution of summertime 850 hPa winds (vector, m/s) and geopotential height (shaded, 10 gpm) for a the CTRL run and b the exDiab run. The differences between the exDiab run and the CTRL run are displayed in c

Full size image

When compared with the CTRL run, the exDiab run (Fig. 4b) seems not to greatly alter the low-level flow pattern either for the tropical summer monsoon or for the EASM. However, there is an obvious change over Indochina Peninsula where the westerly flow in the CTRL run (Fig. 4a) was replaced by the easterly flow in the exDiab run (Fig. 4b). This change tends to separate the South China Sea monsoon trough from the monsoon trough near the Bay of Bengal, and give rise to a slight southward shift of the westerly flow over South China Sea. Furthermore, as seen in Fig. 4c, the difference between the exDiab run and the CTRL run exhibits a large-scale anti-cyclonic perturbation over from tropical subcontinents to subtropical East Asian continent. This perturbation induces a strong tropical easterly anomaly at 10°N–20°N and a strong northerly anomaly along the East Asian coast. Besides, the difference map also shows a cyclonic perturbation over the western North Pacific. All of these features are out of phase with the story in the CTRL run, and indicative of the weakening of the low-level Asian summer monsoon circulation systems such as the western Pacific subtropical high, the southerly flow over East Asia, the tropical monsoon trough and the tropical westerly flows. These results suggest that the exclusion of the feedback of condensational heating over East Asia not only causes locally a weakening of the low-level EASM circulation but also remotely a weakening of the low-level tropical summer monsoon circulation.

Figure 5a illustrates simulated 200 hPa winds and geopotential height in summer that displays the Asian summer monsoon circulation at high levels. A unique feature is the prominent and strong South Asian High (SAH), a key member unifying all of the Asian summer monsoon subsystems. A strong divergence at high levels with SAH as a compensation of the low-level convergence over different monsoon regions induces a tropical easterly flow over South Asia and a northerly flow over East Asia. These flows as the high-level component of the Asian summer monsoon circulation are completely opposite in direction to those at low-levels as shown in Fig. 4a.

Fig. 5
figure 5

As in Fig. 4, but for 200 hPa

Full size image

The SAH formation is closely associated with the monsoon-related extraordinary heating over South Asia, East Asia and Tibetan Plateau. As seen in Fig. 5b, the exDiab run exhibits a dramatic change of SAH both in location and in intensity. Without the feedback of condensational heating over East Asia, the SAH intensity is largely reduced by more than 6 gpm especially over the East Asian sector, and its ridge line undergoes a 5° southward shift from 30°N in the CTRL run to 25°N in the exDiab run. Another important change appearing in the exDiab run is that due to the larger reduction of the SAH intensity over the East Asian sector, the SAH splits into two centers, one of which locates near 50°E and the other near 130°E. And the transition belt between the two centers occurs just over eastern China. Under this circumstance, there are southerly flows prevailing in the upper troposphere over East Asia, in contrast to the northerly flows in the CTRL run. It is worthy to note that as the low-level wind change only lies in its intensity (Fig. 4b), the reversal of the high-level wind direction (Fig. 5b) would alter the EASM vertical circulation structure. Above changes can be further verified with the difference map shown in Fig. 5c. The removal of condensational heating role causes a strong cyclonic perturbation at 200 hPa over the East Asian sector, whereas a strong anti-cyclonic perturbation occurred at 850 hPa as described before. Such perturbations characterized by a baroclinic vertical structure can be interpreted as the response of subtropical atmosphere to a prescribed convective heating (Liu et al. 2001).

5.2 Vertical circulation

Associated with the passage of the dry phase of a Madden-Julian Oscillation (MJO) over the Indonesian-Australian sector, a slowdown of the inter-hemispheric local Hadley cell in East Asia occurs in mid-March (LinHo et al. 2008). Hence, Hadley cell is weakened so much that it reverses direction to an anti-Hadley cell which is also called typical monsoon cell (Tao and Chen 1987; Chen et al. 1991). In mid-summer, the subtropical region of the Northern Hemisphere is under the control of a local monsoon cell, and that is manifested in the CTRL run.

Figure 6a displays the latitude-altitude distribution of winds and diabatic heating rate averaged over 100°E–125°E for the CTRL run. It can be observed that a grand monsoon cell was simulated over the East Asian sector. There are three major ascending flows centered roughly at 5°N, 15°N and 35°N, respectively, that contribute to the grand monsoon cell. These ascending flows are clearly associated with positive convective diabatic heating rate. The low-level cross-equatorial currents converge first in the tropical monsoon trough region forming ascending flows in the tropics, and then the southerly winds converge in the subtropical region forming another significant ascending flow at around 35°N. Interestingly, although all of ascending flows can join in the high-level divergent flows backing to the Southern Hemisphere, part of the flows originating from the ascending flow centered at 35°N sink in a narrow region located over 22°N–25°N, thus establishing a closed local monsoon cell over the subtropical East Asia. We define such a local cell as the subtropical EASM cell. Obviously the subtropical EASM cell is a unique feature of the EASM. The tropical monsoon cell plus the subtropical EASM cell constitute the grand monsoon cell in this specific sector.

Fig. 6
figure 6

Latitude-altitude section of summertime streamline and diabatic heating rate (K/day) averaged within 100°–125°E for a the CTRL run and b the exDiab run

Full size image

However, it can be clearly seen in Fig. 6b that switching off the feedback of condensational heating induces a complete disappearance of the local subtropical EASM cell existing in Fig. 6a. Instead of strong ascending flows around 35°N as in the CTRL run, extensive descending motions are evident at 20°N–50°N where large negative diabatic heating (i.e., radiative cooling) occupies. It is the disappearance of the subtropical EASM cell that causes a discernible decreasing of rainfall over the area 30°–50°N from pentad 36–48, as shown in Fig. 1c. In this case, not only the low troposphere is under control of southerly, but also the southerly winds prevail in the upper troposphere. These changes are consistent with the aforementioned changes of horizontal flows including SAH. Apparently, despite the existence of the zonal land-sea thermal contrast (as shown in Fig. 3b), the local subtropical EASM cell is unable to maintain if the feedback of condensational heating is removed. This indicates that the condensational heating is a dominant factor for forming the subtropical EASM cell.

6 Impact on the EASM dynamical structure

To further understand the role of condensational heating in the EASM formation, we have investigated the dynamical relationship between the condensational heating and the EASM circulation structure. Figure 7 illustrates the longitude-altitude distribution of the summer geopotential height and temperature deviations from their zonal means together with the diabatic heating rate averaged over 30°N–40°N, as calculated with National Center for Environmental Prediction (NCEP)/NCAR reanalysis data for 1948–2007. The distribution of the zonal temperature deviations (shading in Fig. 7b) exhibits a much warmer tropospheric atmosphere over land than over ocean, that is, a strong zonal land-sea thermal contrast. This vertical structure of zonal temperature deviations are determined substantially by the considerable diabatic heating (shading in Fig. 7a) that can penetrate into entire troposphere over Tibetan Plateau (Luo and Yanai 1983) as well as over East Asia. Corresponding to the vertical thermal structure, the zonal geopoential height deviations are characterized by a baroclinic vertical structure with an upper-level low and a lower-level high over ocean and an upper-level high and a lower-level low over land (contours in Fig. 7a, b). The lower-level high over ocean reflects the western Pacific subtropical high, while the upper-level high indicates the SAH.

Fig. 7
figure 7

Longitude-altitude sections of the observed geopotential height (gpm) and temperature (K) deviations from their respective zonal means and the diabatic heating rate (K/day) averaged within 30°–40°N in summer. The geopotential height deviation is indicated by contours, while the diabatic heating rate is shaded in a and the temperature deviation is shaded in b

Full size image

Figure 8 shows the longitude-altitude distribution of the simulated summer geopotential height and temperature deviations from their zonal means together with the diabatic heating rate averaged over 30°N–40°N for two runs. In general, the CTRL run reasonably captures the zonal land-sea thermal contrast in the tropospheric temperature (Fig. 8c) as illustrated in Fig. 3a and the baroclinic vertical structure in the tropospheric geopotential height (Fig. 8a), although there are some biases in the diabatic heating such as weaker heating over East Asia and lack of heating over the subtropical western North Pacific.

Fig. 8
figure 8

Longitude-altitude sections of the geopotential height (gpm) and temperature (K) deviations from their respective zonal means and the diabatic heating rate (K/day) averaged within 30°–40°N in summer for the CTRL run (upper panels) and the exDiab run (lower panels). The geopotential height deviation is indicated by contours, while the diabatic heating rate is shaded in the left panels and the temperature deviation is shaded in the right panels

Full size image

Since the geopotential height deviations are hydrostatically balanced with the temperature deviations, the geopotential deviations with baroclinic vertical structure are dynamically balanced by the diabatic heating. In terms of a quasi-geostrophic potential vorticity (QGPV) balance relation suitable for the subtropical region where the background zonal flow is close to be zero, the vorticity produced by the vertical gradient of diabatic heating is primarily balanced by the meridional advection of ambient vorticity, which can be expressed in the pressure (p) coordinate as

$$ \beta v = - f\frac{\partial }{\partial p}\left[ {\frac{QR}{{\sigma_{p} p}}} \right], $$
(2)

where Q is the diabatic heating rate and σ p is an alternate form of static stability. Thus a convective heating centered at mid-troposphere can excite a baroclinic atmospheric response similar to the simulated geopotential height structure (Liu et al. 2001). When a diabatic heating occurs in the mid-troposphere over East Asia, its vertical gradient is positive at low levels, but negative at high levels. The positive gradient tends to balance the northward advection of ambient vorticity, thus yielding a low-level southerly wind over East Asia, whereas the negative gradient tends to balance the southward advection of ambient vorticity, thus yielding a high-level northerly wind over East Asia. This explains why a local meridional subtropical EASM cell can sustain. Thus, the diabatic heating due to convective latent heat release can dynamically account for many aspects of the EASM circulation structure.

The role of diabatic heating in forming the EASM circulation dynamical structure can be further verified in the exDiab run. Once the feedback of condensational heating is switched off in the exDiab run, we can see that the typical baroclinic vertical structure of geopotential height deviations (contours, lower panels of Fig. 8) does not sustain. In this case, the diabatic heating (shading in Fig. 8b) in the atmosphere disappears over land, and instead the diabatic cooling occurs in the low-level atmosphere over East Asia. Although the tropospheric atmosphere over land is still warmer than over ocean (shading in Fig. 8d), the zonal land-sea thermal contrast is greatly reduced, as displayed in Fig. 3b. Corresponding to these thermal distributions, the geopotential height deviations are characterized by a barotropic vertical structure with a low over land and a high over ocean. Such a vertical structure for geopotential height yields a low-level horizontal circulation similar to that in the CTRL run, but a high-level horizontal circulation dramatically different. This is why the SAH splits, the high-level wind direction reverses, and the local subtropical EASM cell disappears over East Asia, as aforementioned. The barotropic vertical structure of the geopotential height deviations can also be explained in terms of the QGPV balance relation. When a diabatic cooling occurs near the surface of East Asia, its vertical gradient is always positive throughout the troposphere. The positive gradient tends to balance the northward advection of ambient vorticity, thus yielding a southerly wind throughout the troposphere over East Asia with a barotropic low over land and a barotropic high over ocean.

7 Conclusions and discussion

The formation mechanism responsible for the Asian summer monsoon is a classical issue in which the seasonal variation of solar radiation and the inhomogeneity of the Earth’s surface (land-sea contrast and orography) are widely believed to be two fundamental factors. Recently, there are increasing evidences recognizing that the feedback of atmospheric moisture processes could also make contribution to the Asian summer monsoon formation.

The EASM is a distinctive and prominent regional component of the grand Asian summer monsoon system. It is characterized by humid low-level southerly wind that prevails over eastern China, Korea and Japan in summer. The strong monsoon flow brings out abundant water vapor into East Asian continent and forms a large amount of rainfall over the subtropical East Asia, consequently generating substantial latent heat release or condensational heating that could feedback on the EASM itself. However such a feedback effect has not been well explored with atmospheric GCM experiments before. This study identified to what extent and how condensational heating can affect the EASM circulation by conducting and contrasting two 10-member ensembles of atmospheric GCM experiments with CCM3/NCAR respectively with and without feedback of the condensational heating over East Asian domain (90°E–125°E, 20°N–50°N). The major conclusions inferred from the numerical experiments are as follows.

The EASM is substantially driven by the land-sea thermal contrast that is largely determined by the warming over land. The condensational heating induced by the EASM rain is found to absolutely dominate diabatic heating during summer over the subtropical East Asia. It is the heating that warms up the air column aloft. Without the feedback of condensational heating, the magnitude of the warming over the continent including Tibetan Plateau is significantly weakened and the land-sea thermal contrast between entire Asian continent and surrounding oceans is greatly reduced. This reduction is especially significant along the meridional direction over the East Asian sector.

The reduction of the land-sea thermal contrast does not alter the low-level monsoon flow pattern for either the tropical summer monsoon or the EASM, however it does weaken most of the low-level Asian summer monsoon circulation systems such as the western Pacific subtropical high, the southerly flow over East Asia, the tropical monsoon trough and the tropical westerly flows. Therefore, the exclusion of the feedback of condensational heating over East Asia not only causes locally a weakening of the low-level EASM circulation but also remotely a weakening of the low-level tropical summer monsoon circulation.

Most significant change induced by the switch-off of condensational heating feedback lies in the upper troposphere circulation. Without the condensational heating over East Asia, the SAH intensity at 200 hPa is largely reduced especially over the East Asian sector and its ridge line undergoes a southward shift from 30°N to 25°N. Also, the SAH breaks over East Asia and splits into two centers with one locating near 50°E and the other near 130°E. Under this circumstance, the southerly flows are found to prevail in the upper troposphere over East Asia, in contrast to the northerly flows in reality.

The EASM circulation is characterized by a baroclinic vertical structure with an upper-level low and a lower-level high over ocean and an upper-level high and a lower-level low over land. Such a structure is dynamically determined by convective condensational heating over East Asia. Corresponding to such a structure, the lower troposphere is under control of southerly flows while the upper troposphere features northerly flows. Therefore, there exists a local meridional subtropical EASM cell. However, without condensational heating feedback, the EASM circulation is altered to be characterized by a barotropic vertical structure with a low over land and a high over ocean. The southerly flows not only prevail in the lower troposphere but in the upper troposphere. Consequently, the local meridional subtropical EASM cell is unable to maintain, despite the existence of a weaker zonal land-sea thermal contrast, suggesting that the condensational heating is a dominant factor for forming the subtropical EASM cell.

Therefore, we conclude that the feedback of condensational heating induced by monsoon rain acts to largely enhance low-level flows of the EASM and essentially determine its baroclinic vertical structure and meridional cell, once the solar radiation and inhomogeneity of the Earth’s surface form low-level monsoon flows in East Asia by enhancing land-sea thermal contrast. Apparently, this study based on numerical experiments provides a basic knowledge on the importance of the feedback of condensational heating in forming and maintaining the EASM. The importance of condensational heating was also proved in the EASM variability from an anomalous perspective (Lu and Lin 2009).

It should be noted that this study only focused on the response of monsoon circulation to the condensational heating over a specified region. However, we found that switching off the condensational heating over the East Asian domain can affect the circulation over a much broader area, as described in Sect. 5. It is interesting that the circulation response is significant over outside the East Asian domain, particularly over the tropical monsoon regions. This is likely related to the feedback of rainfall change over these regions. Figure 9 shows the horizontal distribution of the simulated summer rainfall for both experiments and their difference. It can be seen in Fig. 9b, c that the rainfall in the exDiab run is reduced over most of the tropical monsoon regions where the monsoon is largely weakened as described before (Fig. 4c), but increased over the South China Sea region where a cyclonic circulation is enhanced (Fig. 4c). This result suggests that the response of atmospheric circulation to the exclusion of condensational heating over the East Asian domain involves a positive feedback between circulation and rainfall-induced heating over outside the East Asian domain, particularly over the tropical monsoon regions. Therefore, the role of condensational heating in forming and maintaining monsoon circulation over the EASM region could be interactive with that over the tropical monsoon region. Obviously, such an interaction needs to be identified in the future study. Also, this study only focused on the impact of condensational heating on summertime mean state of EASM, without insight into its impact on the onset process of EASM. Examining such an impact would be of interest to understand the transient processes of the EASM setup.

Fig. 9
figure 9

Horizontal distribution of precipitation (mm/day) for a the CTRL run and b the exDiab run. The differences between the exDiab run and the CTRL run are displayed in c

Full size image

Another issue is that model has some deficiencies in simulating the EASM rainfall and whether these deficiencies could influence the conclusions obtained. As pointed out in Sect. 3, despite the success of the model in simulating the northward progression of the EASM rain from southern China to the Yangtze River valley and then to northern China, successively, as illustrated in Fig. 1, the model has deficiencies that lie in two aspects. One is that the simulated EASM rain extends northward too far so that the simulated rain over northern China is more than the observed. The other is that the tropical monsoon rain is simulated with a southward shift so that it remains too apart from the subtropical EASM rain. Although these deficiencies are common drawbacks of most of the atmospheric GCMs in the world, as systematic biases these deficiencies would affect the conclusions inferred from the model. However, we believe that the biases would not qualitatively change the main conclusions obtained in this study since the model reasonably captures the main EASM rain belt with seasonal northward migrations and reproduces well the EASM dynamical structure.