1 Introduction

The Indochina Peninsula (ICP), located between the Indian Ocean (IO) and the western Pacific, is strongly affected by the tropical monsoon regime, which dominates weather and climate (Matsumoto 1997; Wang and Fan 1999). The monsoonal precipitation over the ICP has the distinct characteristic of a dry season from October to the subsequent April followed by a rainy season from May to September on which local water resources, food security and crop production are particularly dependent (Wu and Wang 2000; Takahashi and Yasunari 2006; Zhu et al. 2020a). Furthermore, extreme precipitation events in the tropics are also affected by climate change and, therefore, more likely to occur under rapid global warming with increasing frequency of weather and climate extremes disrupting agricultural industries, imposing the heavy drought/flood risks and leading to huge social stresses (Allen and Ingram 2002; Knutson et al. 2010; Zhang et al. 2014; Ge et al. 2021). Thus, the investigation of interannual variability of precipitation is fundamental to understanding the physical mechanisms, which potentially determine predicting signals several months in advance, supporting countries in and around the ICP to implement effective adaptations.

Plenty of efforts in the past several decades have improved the understanding of the interannual and interdecadal variability of climate in Southeast Asia (He et al. 1987; Matsumoto 1992; Webster et al 1998; Lau et al. 2000; Wang et al. 2001; McBride et al. 2003; Takahashi and Yasunari 2008; Fan et al. 2018; Zhu 2018). Many studies have documented that the rainy season precipitation over the ICP has a significant relationship with ENSO (Zhou et al. 2011; Hsu et al. 2014; Villafuerte and Matsumoto 2015; Ge et al. 2017; Fan et al. 2019). Individual countries, e.g. Vietnam, Thailand and Myanmar, are prone to experiencing temperature and precipitation extremes during the decaying phase of the ENSO life cycle (Manton et al. 2001; Nicholls et al. 2005; Caesar et al. 2011; Nguyen et al.2014; Sein and Zhi 2016). These results indicate that ENSO modulating the interannual variability of summer monsoon precipitation remains an important scientific challenge, and how the spatial scales of South East Asia, the main land at large and the individual countries of the ICP, are affected.

It is generally recognized that the onset of the Asian monsoon initially occurs over the Bay of Bengal (BoB) from late April to early May, establishing strong convective activities and the change of direction of the prevailing wind. A number of studies have shown a connection between increased convection and patchy cells of convective storms over the ICP during the monsoon transition over the Asian continent (Lau and Yang 1997; Wu and Zhang 1998; Zhang et al. 2002; Wang and Chan 2002; Li and Zhang 2009). On the other hand, the rainy season precipitation variability over the ICP is closely related to the early/late BoB summer monsoon (BoBSM) onset (Wang and LinHo 2002; Mao and Wu 2007; He and Zhu 2015; Liu et al 2015). The early/late BoBSM onset is more likely to occur during the La Niña/El Niño years (Wang et al. 2013; Xing et al. 2016; Li et al. 2018). Following these studies, we will refer to the role of convection, represented here by anomalous patterns of outgoing longwave radiation and apparent heat source, in Sect. 3.2.

The rainy season precipitation over the ICP is generally divided into three periods: onset/transition (April–May), peak (June–August) and withdrawal (September–October), which are associated with the summer monsoon activities. However, the interannual variability in precipitation during the rainy season over the ICP has yet not been comprehensively revealed. Many studies have focused on the peak and post-monsoon period, suggesting that precipitation is mainly contributed by tropical cyclones crossing the ICP (Fudeyasu et al. 2006; Takahashi and Yasunari 2008; Takahashi et al. 2015). However, the monsoon transition related precipitation and its variability have seldom been investigated with latest high-resolution observational datasets for the Southeast Asian region. The remote SST forcing over the tropical Pacific associated atmospheric processes may influence year-to-year variabilities of AM precipitation over the ICP. The identification of impact factors triggering the monsoon onset and the assessment of associated precipitation variabilities appear to be crucial for understanding the early intense floods in the ICP region. Thus, there are several issues still open: (1) How are interannual variabilities of AM precipitation linked to ENSO events? (2) How does the atmosphere respond to the tropical SSTAs? In the following sections, we use the latest observations and an ensemble of sensitivity experiments to investigate the relationship between the tropical SST forcing and AM precipitation over the ICP.

This paper is organized as follows. Datasets, methods of analysis and the adopted atmospheric general circulation model (GCM) are introduced in Sect. 2. The interannual variability of precipitation during the monsoon transition over the ICP, referring to April and May (AM), and the possible teleconnections with the tropical Pacific and IO are presented in Sect. 3. Results of the corresponding numerical experiments are shown in Sect. 4. Conclusions and discussions follow in Sect. 5.

2 Data and methods of analysis

2.1 Observations

In this study, the observed monthly mean precipitation and surface air temperatures (SATs) over the ICP (6°N–23°N, 96°E–110°E) for the time period 1981–2016 with a resolution on 0.5° × 0.5° are obtained from the Southeast Asian Climate Assessment and Dataset (SACA&D). This new high- resolution (land-only) dataset, called SA-OBSv2.0, has undergone strict quality control procedures, including data homogenization and time consistency, to improve reliability. Detailed descriptions of this dataset could be found in van den Besselaar et al. (2017).

The National Centers for Environment Prediction/National Center for Atmospheric Research (NCEP/NCAR) reanalysis dataset (Kalnay et al. 1996) are also used in this study. Upper-air variables (e.g.: wind fields, geopotential height and relative humidity) have a spatial resolution of 2.5° × 2.5° with 17 vertical pressure levels. The interpolated Outgoing Longwave Radiation (OLR) data (Liebmann and Smith 1996) are derived from the National Oceanic and Atmospheric Administration (NOAA) with a horizontal resolution of 2.5° × 2.5°. In addition, global monthly mean SST data is from the Extended Reconstructed Sea Surface Temperature dataset (ERSST; Huang et al. 2015; 2017) with a horizontal resolution of 2° × 2°. The NCEP/NCAR reanalysis, OLR and ERSST datasets cover the same period as the observations (1981–2016).

The Oceanic Niño Index (ONI), the primary indicator for monitoring El Niño and La Niña events by Climate Prediction Center of the National Oceanic and Atmospheric Administration (NOAA), is considered to represent ENSO variability in this study, which is computed as the 3-month running mean SST anomaly (SSTA) over the Niño 3.4 region (5°N-5°S, 120°-70°W) (Huang et al. 2017). A warm (cold) ENSO event is defined when the winter ONI value is greater (less) than 0.6 °C (– 0.6 °C). The winter and spring in this study are defined as December–January–February (DJF) and March–April–May (MAM), respectively. Note that the number -1 denotes the previous year but not to ENSO cycle.

The empirical orthogonal function (EOF) analysis is employed to capture the dominant modes of AM precipitation variations over the ICP. The composite and linear correlation analyses are applied to investigate the spatial and temporal variability of AM precipitation anomalies and its association with the large-scale SSTAs. The onset dates of the BoBSM (Li et al. 2018) are defined as the areal mean zonal wind at 850 hPa (U850) over 90°–100°E, 5°–15°N, satisfying the following two criteria: (1) the day where U850 exceeds 0 m/s and (2) when the mean U850 is larger than 2 m/s in the subsequent 10 days (including the onset day).

According to Yanai et al. (1973, 1992) and Yanai and Li (1994), the apparent heat source \(Q_{1}\) is calculated by

$$ Q_{1} = C_{p} \left( {\frac{p}{{p_{0} }}} \right)^{k} \left[ {\frac{\partial \theta }{{\partial t}} + \overrightarrow {V} \cdot \nabla \theta + \omega \frac{\partial \theta }{{\partial p}}} \right] = Q_{R} + L\left( {c - e} \right) - \frac{{\partial \left( {\overline{{s^{\prime}\omega^{\prime}}} } \right)}}{\partial p}\begin{array}{*{20}c} {} & {} & {\left( 1 \right)} \\ \end{array} $$

where \(\theta\) is the potential temperature, \(V\) the horizontal wind, \(\omega\) the vertical \(p -\) velocity, \(k = R/C_{p}\),\(R\) and \(C_{p}\) are the gas constant and the specific heat at constant pressure of dry air, \(L\) the latent heat, \(p_{0} = 1000hPa\),\(Q_{R}\) the radiative heating rate, \(s = c_{p} T + gz\) the dry static energy, \(c\) the rate of condensation per unit mass of air and \(e\) the rate of re-evaporation of cloud and rain water. Statistical significance is examined by a two-tailed Student’s t-test. Note that the long-term linear trends in all datasets have been removed.

2.2 Model and model experiments

The intermediate complexity climate model, Planet Simulator (PlaSim; Fraedrich et al. 2005a, b), is used to conduct the numerical experiments to investigate atmospheric circulation and moisture transport responding on different SSTAs forcing during the monsoon transition. PlaSim is a spectral general circulation model (GCM), whose dynamical core is based on Portable University Model of the Atmosphere (PUMA), an atmospheric GCM that solves the moist primitive equations conserving momentum, mass and energy in the Earth’s atmosphere. The atmospheric module is coupled to a 5-layer land surface model with biosphere. The ocean is represented by a 50 m depth mixed layer-ocean, which includes a thermodynamic sea-ice model. The horizontal resolution used in this study is T42 (2.8° × 2.8°) with 10 vertical sigma levels (for more details see Lunkeit et al. (2011).

The Planet Simulator (PlaSim) is freely available for download at https://www.mi.uni-hamburg.de/en/arbeitsgruppen/theoretische-meteorologie/modelle/plasim.html. It is known in the world-wide user community (including the Open University in England, National Research Council of Italy, Siberian Branch of the Russian Academy of Sciences, and Peking University etc.), this model, being one of the few freely available GCMs, has also been constructed for reproducibility, a demand required for physical experiments in general (Holden et al. 2014). Previous studies have demonstrated that PlaSim can be successfully employed to analyze the evolution of the climate system under different physical scenarios (Bordi et al. 2012, 2013; Holden et al. 2016). It also shows a high skill in reproducing the global impacts of the tropical SST anomalies in an idealized numerical modelling environment (Dahms et al. 2011; Schmittner et al. 2011) and under conditions prescribed by observations (Zhang et al 2015, 2017).

3 Results

3.1 Interannual variations of the April–May precipitation over the ICP

3.1.1 Precipitation in April and May

Figure 1a presents the monthly averaged mean surface air temperature and precipitation during the period of 1981–2016. The results show that the warmest months is in May, reaching over 27.2 °C. The monthly mean precipitation is strongly associated with the summer monsoon variability, which reveals that nearly 80% of total annual precipitation over the ICP occurs in the rainy season (May–October). Furthermore, it is also clearly shown in Fig. 1b that the monthly mean precipitation in April and May is highly correlated with the Oceanic Niño Index (ONI) in winter [D (-1) JF] and spring (MAM).

Fig. 1
figure 1

a Monthly surface air temperature (SAT, red line, °C) and precipitation (blue bars, mm month −1) during 1981–2016 over the ICP. b Correlation coefficients between monthly precipitation and the Oceanic Niño Index (ONI) in D (-1) JF (blue line) and MAM (cyan line). The black dash lines represent the 95% significant level

Full size image

The temporal and spatial features of April and May (AM) precipitation over the ICP is explored by EOF analysis on the SA-OBS dataset. The first leading mode shows a uniform monopole pattern, accounting for 60% of the total variance (Fig. 2a). The linear correlation coefficients between the ONI and the normalized principal component of the first EOF mode (PC-1) are − 0.60 (Fig. 2b) and − 0.65 (Fig. 2c) in winter and spring, respectively, both passing the 95% significance level. In contrast, its correlation with the Indian Ocean dipole (IOD) index is only − 0.16 (not shown), which suggests that the AM precipitation is more closely related to SST anomalies of the Niño-3.4 region than to the IOD. The annual cycle of precipitation over Southeast Asia varies significantly due to the complex land-sea interaction associated with the reversed surface monsoonal winds. The southern part of ICP (e.g. Vietnam) and Malay Peninsula are influenced by the winter monsoon regime during boreal autumn and winter, and these rainfall events are also associated with the ENSO cycle (Chang et al. 2004, 2005; Zhou et al. 2008; Feng et al. 2010). In this study, we mainly focus on the precipitation anomalies over the ICP during the summer monsoon transition period represented by the wetter/drier than normal years of AM precipitation.

Fig. 2
figure 2

a The first EOF mode of April–May (AM) precipitation during 1981–2016 over the ICP. b Scatterplots of the Oceanic Niño Index (ONI) and the normalized principal component of PC-1 in D (-1) JF. c Same as (b), but for MAM

Full size image

3.1.2 Relationship with SSTA

To investigate the year-to-year variability of AM precipitation over the ICP and its association with the SSTA patterns over tropical ocean, we categorize the normalized PC-1 from 1981 to 2016 with ± 1 std as the threshold, and obtain four wet years (1999, 2000, 2009, and 2011) and seven dry years (1983, 1987, 1992, 2005, 2014, 2015 and 2016), which are used for composite studies (Fig. 3a, b).

Fig. 3
figure 3

Composites of precipitation anomalies (mm month −1) in April–May (AM) during the dry (a) and wet (b) years. The white crosses indicate statistical significance at the 90% confidence level. c Time series of the BoBSM onset day anomalies defined by U850, the red and blue stars represent the dry and wet years, respectively

Full size image

The BoBSM also plays an important role for the interannual variability of AM precipitation over the ICP. The dry years are usually concurrent with a late onset and a wet year with an early onset of the BoBSM. The selected dry and wet years are identical to the late and early BoBSM onset years (Fig. 3c). Note that SSTA composites in winter and spring closely resemble the El Niño (La Niña) pattern during dry (wet) years (Fig. 4), which are consistent with the results of previous studies (Liu et al. 2013; Nguyen-Le et al 2015; Xing et al. 2016; Li et al. 2018). They found that the composite SSTA over the western and central-eastern Pacific for early /late monsoon onset shows a typical ENSO pattern during the preceding winter and spring. In summarizing, ENSO events appear to modulate the AM precipitation anomalies, which are closely linked with the early /late BoBSM onset years. This modulating mechanism needs to be more deeply elucidated by analyzing the large-scale atmospheric circulation response to tropical SSTA and moisture transport during the dry and wet years in the following subsection.

Fig. 4
figure 4

Composites of SST anomalies (SSTA) in D (− 1) JF and MAM during dry (a, c) and wet (b, d) years. The white crosses indicate statistical significance at the 90% confidence level. The red rectangles represent three SSTA key areas

Full size image

3.2 Anomalous atmospheric circulation response to tropical SST forcing

3.2.1 Tropical atmospheric circulation

Wind field composites (Fig. 5a, b, arrows) of dry (wet) years are characterized by two anti-cyclonic (cyclonic) circulation anomalies over the BoB and WNP. The low-level atmospheric circulations are similar to those of early (late) BoBSM onset years over the ICP. Compared with its climatological mean conditions, the ICP is dominated by enhanced moisture divergence (convergence) anomalies (Fig. 5c, d, shaded areas) associated with the anomalous anti-cyclonic (cyclonic) flow during the dry (wet) years.

Fig. 5
figure 5

Composites 850 hPa wind field anomalies (arrows, m/s) and vertical integral of divergence of moisture flux anomalies (shaded, 10−5 kg m−2 s−1) in April–May (AM) during dry (a, c) and wet (b, d) years. Only anomalies statistically significant at the 90% confidence level for wind are shown in (a) and (b). Divergences of moisture flux are significantly stippled in purple crosses in (c) and (d). The marks of “A” and “C” indicate the centers of anomalous anti-cyclones and cyclones, respectively

Full size image

Actually, the direct atmospheric response to the equatorial central Pacific warming can be interpreted by the classical Matsuno-Gill response theory (Matsuno 1966; Gill 1980). That is, a pair of anomalous cyclonic circulations located along both sides of the equator and associated with the equatorial westerly anomalies. Formation and persistence of anti-cyclonic (cyclonic) circulation anomalies over the WNP, in terms of the Matsuno-Gill dynamics, is a Rossby wave response of the suppressed (enhanced) convective heating, which is induced by the local WNP SSTA cooling (warming) associated with the central-eastern Pacific warming (cooling). This Pacific-East Asian teleconnection could significantly influence the interannual variations of precipitation over the East and Southeast Asian (Wang et al 2000) and our observational results indeed support the theory.

During the dry years (Fig. 5a), the cold SSTA of the Maritime Continent (MC) favors the anomalous anti-cyclone. The returning southwesterly along the western flank of the WNP anti-cyclone dominates the ICP, thereby cutting Pacific moisture transports from the WNP into this region. On the other hand, during the wet years (Fig. 5b), the situation is reversed due to La Niña conditions and the anomalous WNP cyclone being maintained.

The anti-cyclonic (cyclonic) circulation anomalies over the BoB, however, cannot be satisfactorily explained by a direct SST heating response mechanism as in the Matsuno-Gill model. It is noteworthy that the weaker positive SSTA has been found over the BoB during the dry year (Fig. 4c). That means, the local SST is warmer than the climatological mean and the anti-cyclone over the BoB (Fig. 5a) is not consistent with the simultaneous SSTA. During the wet years, negative SSTA occurs over most of the tropical IO but the SST over the BoB is still warmer and the anomalous cyclonic circulation appears in this region. This result implies that the local SSTA probably is not the clue which establishes the anti-cyclone (cyclone) over the BoB nor account for its variability.

Atmospheric diabatic heating appears as another predominant contributor to induce circulation anomalies. This is clearly indicated by the weakened convective activity (as documented by the OLR anomalies) spreading from the BoB to South China Sea during the dry years (Fig. 6a), while enhanced convective activities occur over the BoB and ICP during the wet years (Fig. 6b). In addition, the vertically integrated apparent heat source < Q1 > features appear to support our notion. The negative < Q1 > anomalies cover the BoB and ICP during the dry years (Fig. 6c), while the positive < Q1 > anomalies primarily dominate from the BoB via central ICP to the South China Sea (SCS) during the wet years (Fig. 6d). The anti-cyclonic (cyclonic) circulation anomalies are related to anomalous diabatic heating associated with convective activities over the BoB during the dry (wet) years. Xing et al (2016) also found that the anomalous atmospheric heating associated with the late (early) BoBSM onset contributes to the generation of an anomalous anti-cyclone (cyclone) over the BoB, which further suppress (favor) the precipitation in May over the southern Indian peninsula, the ICP and the SCS.

Fig. 6
figure 6

Composites of OLR (a, b, W/m2) and vertically integrated apparent heat source < Q1 > (c, d, W/m2) anomalies in April -May (AM) during dry and wet years. The white crosses indicate statistical significance at the 90% confidence level

Full size image

Moisture transport is mainly dominated by winds at the low level such as the 850 hPa, which is shown in Fig. 5. Therefore, drought conditions over the ICP can be attributed to influences of the westerlies along the northern flank of an enhanced anti-cyclone over the BoB and the southerlies along the northwestern flank of an extended anti-cyclone over the WNP. The two anomalous anti-cyclonic circulations transport the moisture from the SCS to the southern China, inducing the suppressed convection and decreased precipitation over the ICP. On the other hand, during the wet years, the reversed pattern occurs over the ICP. The abnormal AM precipitation over the ICP can be explained by the anomalous northeasterlies along the northern flank of an extended WNP cyclone and the enhanced westerlies along the southern flank of BoB cyclone. The two cyclonic circulations steer the moisture transport from the WNP and BoB to the ICP in April and May and subsequently increase the convergence inducing the larger than normal precipitation.

3.2.2 The role of Walker and Hadley circulation

Figure 7a–d shows the composite anomalies of the 850 and 200 hPa velocity potential and divergent wind during the dry and wet years. During the dry years, significant lower level convergences are observed over the tropical central-eastern Pacific, while the enhanced divergences appear over the WNP (Fig. 7a). Strong upper-level divergences predominately occur over the tropical central-eastern Pacific, whereas the anomalous convergences are located over the WNP (Fig. 7c). This implies that the eastward shifted/extended Walker circulation suppresses the convective activities over the MC and the anomalous heat sink excites the descending Rossby waves, which reinforce the anomalous anti-cyclone over the WNP during the dry years. In contrast, the wet years show the western Pacific being dominated by a lower-level center of convergence while strong divergence is observed over the tropical central-eastern Pacific (Fig. 7b). The reversed features can be found in the upper-level (Fig. 7d). These baroclinic patterns being linked with the SST anomalies over the tropical Pacific (Fig. 4), suggest that the SST anomalies associated with El Niño (La Niña) events appear to be driving the Walker circulation anomalies of opposite sign during dry and wet years.

Fig. 7
figure 7

850 and 200 hPa velocity potential (a–d, contours, 10−6 m2/s), divergence wind (a–d, vectors, m/s) anomalies and cross sections of the mass streamfunction anomalies (e, f, 1010 kg/s) in April–May (AM) during dry and wet years, averaged over 115°–125°E. Shaded areas are statistically significant at the 90% confidence level

Full size image

The composite of mass streamfunction (MSF) anomalies over 115°–125°E are presented in Fig. 7e, f (Buja 1994; Nguyen et al 2013). This MSF streamfunction \(\left(\Psi \left(p,y\right)\right)\) is computed by vertically integrating the monthly meridional wind

$$\Psi \left(p,y\right)=\frac{2\pi a\mathrm{cos}(y)}{g}{\int }_{p}^{{P}_{s}}V\left(p,y\right)dp$$

where \(p\) is the pressure, \(y\) is the latitude, \(a\) is the average radius of Earth, \(g\) is the gravitational acceleration and \({P}_{s}\) is the surface pressure. In the dry years, a weaker than normal local Hadley circulation is observed, while a stronger than normal local Hadley circulation appears in the wet years. This is consistent with the changes of Walker circulation, as revealed by the composite of lower–upper level velocity potential and divergence wind. The weaker local Hadley circulation in dry years, which is concurrent with the overturning Walker circulation, implies a weakened tropical convective activity (as documented by the OLR in Fig. 6a, b) over the MC and the ICP, which leads to a further decrease of precipitation in this region. The opposite holds for the wet years. These results suggest that changes of local Hadley circulation and Walker circulation appear to be responsible for regulating the AM precipitation variability over the ICP. Note, that the Walker circulation links to ENSO SST anomalies, whereas the local Hadley circulation is attached to local meridional SST anomalies over the MC region.

In conclusion, the atmospheric circulation response to the tropical SSTA forcing is crucial for modulating the interannual variabilities of AM precipitation over the ICP. To verify the observed teleconnections between the tropical SSTAs and AM precipitation over the ICP, global circulation model experiments are performed with PlaSim-GCM in the following subsection.

4 Model experiments

4.1 Upscaled SSTA forcing

The climate mean control (CTL) case is compared with two sensitivity experiments with prescribed forcing by the observed tropical SST fields (Table1). The SST forcing of the CTL run is provided by the SST with annual cycle from the climatological global Atmospheric Model Intercomparison Project (AMIP). The control case is obtained from the last 20 years of the 30-year PlaSim control run. Analogously, the PlaSim sensitivity experiments are also integrated for 30 years with the results of last 20 years being used to construct a 20-ensemble member for investigating the simulated atmospheric response to the SSTA forcing. The SSTA are derived from the composites of seasonal mean SSTA over the key tropical areas in dry and wet years (Fig. 4, red rectangle). In order to obtain better atmospheric responses, we use the three area-averaged SSTAs, each for one of the three key tropical areas. It takes better into account the different spatial variances of the SSTA forcing areas than only using the spatial means since the three key tropical regions are highly related in the dry and wet year cases. These are upscaled based on the formula:

Table 1 Design of the PlaSim experiments
Full size table
$${\mathrm{SSTA}}_{upscaled}=\mathrm{SSTA}\times \frac{1}{{\mathrm{SSTA}}_{max}}$$

with \({\mathrm{SSTA}}_{max}\) as the maximum within the forcing area (red rectangle). The averaged SSTA of the three key tropical areas and its corresponding upscaled SSTAs are listed in Table 2.

Table 2 The averaged SSTA (°C) of the three key tropical areas and its corresponding upscaled SSTA (°C) imposed in the sensitivity experiments. The upscaled SSTA (°C) is highlighted in bold
Full size table

4.2 GCM responses

The simulated atmospheric circulation and moisture transport responses to the abnormal SST forcing of the tropical ocean during dry and wet years are shown in Fig. 8a and b (after subtracting CTL runs from sensitivity experiments). Compared with the observed circulation composites (Fig. 5a, b) the 850 hPa wind and moisture responses are well reproduced by PlaSim. In the DRY-experiment, the anomalous lower-level anti-cyclone occurs over the WNP with a slight eastward extension, while the abnormal anti-cyclone appears over the BoB with reduced meridional extent. The southwesterlies caused by the anti-cyclone over the WNP transports abundant moisture to South China and the northwesterlies associated with the BoB anti-cyclone brings more dry air from the Indian subcontinent, leading to drought conditions over the ICP (Fig. 8a). By contrast, in the WET-experiment, two anomalous low-level cyclonic circulation centers occur over the BoB and WNP, with enhanced moisture convergence over the ICP (Fig. 8b).

Fig. 8
figure 8

Differences of 850 hPa wind field (arrows, m/s) and vertical integral of divergence of moisture flux (shaded, 10−5 kg m−2 s−1) between two experiments in April–May (AM). a, c DRY-CTL, (b, d) WET-CTL. Only anomalies statistically significant at the 90% confidence level for wind are shown in (a) and (b). Divergences of moisture flux are significantly stippled in purple crosses in (c) and (d). The marks of “A” and “C” indicate the centers of anomalous anti-cyclones and cyclones, respectively

Full size image

In summarizing, the SSTA forcing experiments confirm the modulating mechanisms analyzed in Sect. 3, which explain the observed remote impacts of the tropical SSTAs on the interannual variability of AM precipitation observed over the ICP.

5 Conclusions and discussion

The interannual variability of April–May (AM) precipitation over the Indochina Peninsula (ICP) has been investigated for the time period 1981–2016 using observed monthly precipitation data. The analysis of the large-scale atmospheric response to tropical SSTA is based on reanalysis data and results from experiments with Planet Simulator (PlaSim), a climate model of intermediate complexity. The results are comprised in a schematic diagram (Fig. 9) and illustrated as follows.

  1. (i)

    The climatological monthly mean precipitation in AM over the ICP are highly correlated with the preceding winter [D (-1) JF] and spring (MAM) SSTA over the central-eastern Pacific. There is a close relationship between the dry/wet conditions over the ICP and the ENSO phenomenon. The below (above) normal AM precipitation over the ICP coincides with El Niño (La Niña) years, which also corresponds well to the late (early) BoBSM onset years. The atmospheric circulation response to the tropical SSTA forcing plays an important role on interannual variability of AM precipitation over the ICP.

  2. (ii)

    During the dry years, the anomalous WNP anti-cyclone induced by El Niño, transports abundant moisture to South China by enhanced southwesterlies, while the anti-cyclone over the BoB brings cold and dry air from Indian subcontinent to the ICP. The reduced tropical convective activities occur over the MC and the ICP due to the weakened local Hadley circulation concurrent with the overturning Walker circulation, which favor drier than normal conditions over the ICP (Fig. 9a).

  3. (iii)

    During the wet years, the northeasterly anomalies along the northwestern flank of WNP cyclone bring more moisture from Pacific to the ICP, while cyclonic circulation over the BoB also transports abundant warm and humid air to the ICP. The strengthened local Hadley circulation concurrent with the enhanced Walker circulation indicate more tropical convective activities over the MC and the ICP, which leads to a wetter than normal conditions (Fig. 9b).

Fig. 9
figure 9

Schematic diagram of atmospheric circulation anomalies in lower level (850 hPa) and upper level (200 hPa) during dry and wet years. Red/blue shaded areas indicate warm/cold SST anomalies. Light blue/yellow shaded areas over the BoB represent the diabatic heating < Q1 > anomalies. AC/C represents the anomalous anti-cyclone/cyclone. Black and grey arrows indicate anomalous zonal and meridional circulation branches, respectively. CON and DIV indicate convergence and divergence at the upper level. HC represents the anomalous Hadley Circulation

Full size image

The sensitivity experiments conducted with PlaSim successfully reproduce the teleconnections between the AM precipitation over the ICP and the tropical SSTA forcing, which further confirms the modulating mechanism explaining the prolonged atmospheric responses to ENSO events.

Located in Southeast Asia, the ICP is presently attracting increasing concern due to its high climate risk exposure to global warming. A comprehensive understanding of the year-to-year variability of temperature and precipitation is of critical importance to mitigate the potential threat for the developing countries of the ICP. More work remains necessary to be conducted on comparisons of the moisture supply in present and future climate over this region (Li et al. 2013; Li and Zhou 2016). It is evident that the ongoing global warming has enhanced severe extremes during the past two decades, which makes the ecosystem vulnerable and becoming more fragile in this region (IPCC 2018; Eckstein et al. 2019). This is a condition, which is very likely to continue into the not too distant future. Thirumalai et al. (2017) have shown that the robust relationship between ENSO and the SATs in Southeast Asia, especially all April temperature extremes over the ICP, have occurred during El Niño years. Yang and Wu (2019) found that the variations of March–April precipitation over the ICP are significantly influenced by the increasing thermal contrast between the neighboring SST and atmospheric column, which is modulated by the surface heat fluxes changes.

In addition, the SCS and IO also play important roles in precipitation anomalies over the surrounding areas of the ICP (e.g. southern China and MC). The eastward WNP anti-cyclone induced by canonical El Niño enhances meridional moisture transport from the SCS, which results in an increased precipitation over Southern China. On the other hand, the westward shift of WNP anti-cyclone associated with the El Niño Modoki suppresses the meridional moisture transport from the SCS, leading to a decreased precipitation over Southern China (Wang et al. 2018, 2019). Abnormal moisture transport can generate higher sea surface salinity over the SCS, which further influences the summer precipitation over the middle and lower Yangtze River Valley of China (Zeng et al. 2019). Positive (negative) SSTAs over the tropical western IO could induce a suppressed (enhanced) Walker circulation over the tropical IO and WNP, leading to a weaker (stronger) monsoon trough in the SCS, which further reduces (increases) precipitation in June over the ICP (Leung et al. 2020). The abnormal positive IOD event during 2015/2016 could establish the El Niño‐SCS connection in August, which leads to an early anomalous anticyclone (AAC) and further suppresses the rainfall over the MC and ICP (Xiao et al. 2020).

More recently, Ge et al. (2019) and Zhu et al. (2020b) have pointed out the impacts of global warming on climatic stresses in Southeast Asia especially in the densely populated large coastal regions of the ICP and have further revealed the necessity of restricting the warming trends. In any case, the record-breaking droughts and floods, super ENSO events, and intensified tropical cyclones make the detection and attribution of climate change even more challenging in the tropics. In addition to the underlying impacts of the tropical SSTAs shown in this study, these results suggest to further analyze the climate conditions of Southeast Asia mainland and to initiate wider attention to this region, in order to support sustainable future development.