1 Introduction

Abundant rainfall and southerly wind at lower troposphere in summer are the most prominent features of East Asian summer monsoon (EASM). The EASM is characterized by substantial decadal variation in recent decades based on observational record, which redistributes the precipitation and the risk of drought/flood over East Asia at decadal time scales (Zhou et al. 2009; Zhang 2015a, 2015b). The most well-known decadal change of EASM is the weakening of the southerly monsoon circulation around late 1970s associated with a "southern flood and northern drought" precipitation pattern (Zhou et al. 2009; Li et al. 2015, 2018; Zhang and Zhou 2015), which has been attributed to either internal variability of the climate system (Li et al. 2011; Si and Ding 2016) or the external forcing due to increased anthropogenic aerosol concentrations (Song et al. 2014; Li et al. 2015, 2018; Tian et al. 2018a; Luo et al. 2019), but the debate is ongoing.

Aerosols reflect and absorb solar radiation, diming the surface and heating the atmosphere. The relative importance of the "dimming" and "heating" effects are different for scattering aerosols (such as sulfate aerosol) and absorbing aerosols (mainly black carbon). Scattering aerosols over Asian continent act to weaken the land–ocean thermal contrast and EASM since the dimming effect is dominant (Li et al. 2016; Wang et al. 2017; Mu and Wang 2021). Most studies suggested that absorbing aerosols act to strengthen the EASM by its heating effect (Li et al. 2016; Wang et al. 2017; Mu and Wang 2021), but Persad et al. (2017) suggested that the overall effect of absorbing aerosols is also to weaken the EASM dominated by its dimming effect. Forced by the observed decadal increase of total aerosol concentration, coupled general circulation models (CGCMs) simulate a weakening trend of EASM (Song et al. 2014; Li et al. 2015; Tian et al. 2018a). Besides the direct radiative effect within the atmosphere, changes in aerosol concentration also induces a slow adjustment of sea surface temperature (SST), and the climatic effect of aerosol forcing can be decomposed into a "fast response" without the role of change in SST and a "slow response" due to the aerosol-induced change in SST. It is suggested that the total response of EASM to historical aerosol forcing is dominated by the fast response, but partially offset by the slow response (Wang et al. 2017, 2019; Li et al. 2018; Mu and Wang 2021).

The sudden outbreak of the COVID-19 pandemic offers a unique opportunity to examine the responses of the environment and climate to the reduction in anthropogenic emissions (Diffenbaugh et al. 2020; Forster et al. 2020), As a result of emission reduction during the COVID-19 pandemic, aerosol concentration decreased and air quality improved over Asia (Liu et al. 2020; Wang et al. 2021; Xu et al. 2020a, b; Dong et al. 2021; Seo et al. 2020; Sharma et al. 2020), and the change in aerosol has a larger impact on global climate than any other emission changes (e.g., CO2) due to COVID-19 (Gettelman et al. 2021). The aerosol optical depth (AOD) decreases by about 20%-60% in early 2020 soon after the COVID-19 outbreak (Acharya et al. 2021; Ghahremanloo et al. 2021), and the low-AOD condition continued for at least several months (Sanap 2021). Forced by the reduced aerosol concentration during COVID-19 pandemic, the Indian summer monsoon rainfall increased by 4% (Fadnavis et al. 2021), but the possible response of EASM and the mechanism remain unknown. The EASM is strongly modulated by the internal variability of the climate system, such as the tropical SST anomalies and atmospheric internal dynamics at interannual time scale (Cherchi and Navarra 2013; Wang et al. 2015; Xie et al. 2016; Chen et al. 2019). It requires ensemble simulations by multiple climate models to suppress internal variability and stochastic model error before extracting the response of EASM to COVID-19 forcing.

Unified experimental design and forcing dataset have recently been released by the COVID Model Intercomparison Project (COVID-MIP), as a part of Coupled Model Intercomparison Project Phase 6 (CMIP6), to address the climatic effect of the emission reduction during COVID-19 pandemic (Lamboll et al. 2021). Based on these idealized experiments, this study aims to answer the following questions: (1) How does the EASM respond to COVID-19 forcing? (2) What's the mechanism for its response? (3) Is the simulated response of EASM supported by observational evidence? In the reminder of this article, the data, model and method are introduced in Sect. 2, and the above three questions are addressed in Sects. 3–5, respectively, with a summary in Sect. 6.

2 Data, model and method

The observational and reanalysis datasets (referred as "observation" hereafter) spanning from 1979 to 2021 adopted in this study include: (1) The monthly precipitation data from Global Precipitation Climatology Project (GPCP) dataset version 2.3 (Adler et al. 2003); (2) The monthly wind, geopotential height and temperature data from the ERA5 global reanalysis dataset (Hersbach et al. 2020). Eddy geopotential height (He) is calculated by removing the regional averaged geopotential height over 0°–40°N, 180°W–180°E in each month, for reanalysis data and each model, respectively, as He better matches the wind field than the original geopotential height under a global warming background (Huang et al. 2015; Wu and Wang 2015; He et al. 2018). Following the concept of global monsoon (e.g., Lee and Wang 2014; He et al. 2020), the continental EASM region is defined as the land grid points east of 100°E and north of 20°N where the local precipitation from May to September accounts for more than 55% of annual precipitation and is greater than the precipitation from November to March by 2.5 mm/day, based on the 1979–2021 climatology of the GPCP dataset.

Currently, in Jan 2022, there are nine coupled general circulation models (CGCMs) participating in COVID-MIP each offering at least ten realizations for both SSP2-4.5 and SSP2-4.5-COVID experiments online. The monthly outputs for all the available realizations of each model are adopted (see Supplementary Table S1 for the IDs of the realizations). The SSP2-4.5 experiment is a moderate emission pathway (O'Neill et al. 2016) and is used as a baseline. The SSP2-4.5-COVID experiment branches from the SSP2-4.5 experiment on January 2020 and it runs for 5 years. It is parallel with the SSP2-4.5 experiment, but the emissions of aerosol precursors are reduced by 1/3 from January 2020 to December 2021 due to the partial lockdown strategy (Forster et al. 2020; Lamboll et al. 2021), and the emissions gradually rebound during 2022 to an un-perturbed level at the end of 2022 (see Fig. 4 in Forster et al. 2020 for the temporal evolution). The last 150 years of the pre-industrial control (piControl) experiment (Eyring et al. 2016) of these models are also adopted, in order to quantify the amplitude of interannual climate variability under fixed external forcing. All the model data are horizontally interpolated onto a 2.5° × 2.5° grid using bi-linear interpolation before analysis.

As this study mainly focuses on the fast response, the SSP2-4.5-COVID experiment is required to branch exactly from the corresponding realizations of SSP2-4.5 experiment at the beginning of 2020 to minimize the impact of the stochastic internal variability (Lamboll et al. 2021). However, the tropical SST in January 2020 averaged among all the available realizations in SSP2-4.5-COVID experiment deviates obviously from the SSP2-4.5 experiment in four of these nine models (Fig. 1), and the forced response may be masked by the SST variability based on these four model simulations. For example, the difference in the SST in January 2020 averaged among all realizations between SSP2-4.5-COVID and SSP2-4.5 experiments shows a La Niña-like pattern based on MPI-ESM1-2-LR model (Fig. 1h), hence it is hard to distinguish the delayed impact of this La Niña-like SST anomaly on EASM and the fast response of EASM to COVID-19 forcing based on this model. In addition, there are too few models participating in COVID-MIP to suppress the stochastic SST variability by ensemble mean among models. Therefore, only five models with a relatively small root-mean-square deviation of tropical SST (smaller than 0.04 K) are used for further analyses, and they include CanESM5 (10 realizations), E3SM-1–1 (10 realizations), EC-Earth3 (13 realizations), MIROC-ES2L (30 realizations) and MRI-ESM2-0 (10 realizations). For each model, the climate condition averaged within the first two summers (June-July–August, JJA) across all the available realizations are calculated for the SSP2-4.5-COVID and the SSP2-4.5 experiments, respectively, and the multi-model ensemble mean (MME) of the difference between the SSP2-4.5-COVID and the SSP2-4.5 experiments among the five models is calculated and shown.

Fig. 1
figure 1

The difference in the tropical SST in January of 2020 between SSP2-4.5-COVID experiment and SSP2-4.5 experiment for the MME (a) and each model (b–j). For each model in (b–j), the average among all available realizations is shown. The tropical (15°S–15°N) averaged root mean square difference (RMSD) of SST (unit: K) is shown on top of each panel

Full size image

In addition, 16 atmospheric general circulation models (AGCMs) from CMIP6 with available piClim-control and piClim-aer experiments are also adopted (Supplementary Table S2), to verify the fast response of EASM to reduced aerosols. The piClim-control and piClim-aer experiments are both forced by exactly the same SST based on the climatology of the piControl experiment and integrated for 30 years, and the aerosol precursors are set at the level of 2014 A.D. for the piClim-aer experiment but at the pre-industrial level for the piClim-control experiment (Collins et al. 2017). The MME of the difference between piClim-control and piClim-aer experiments indicates the climatic effect of removing all anthropogenic aerosols without any change in SST (Wang et al. 2017; Mu and Wang 2021). The MME of the response is considered to be supported by a high inter-model consensus if at least 70% of the individual models agree in the sign (e.g., Tian et al. 2018b; He et al. 2020; Mu and Wang 2021).

3 EASM response

Figures 2 and 3 show the simulated precipitation and atmospheric circulation responses for boreal summer (all the results in this work are about JJA unless particularly mentioned), respectively. Based on the MME of the 5 selected CGCMs participating in COVID-MIP, the precipitation generally increases over continental East Asian monsoon region (Fig. 2a). The simulated change in the atmospheric circulation at 850 hPa is characterized by decreased He over the Asian continent and increased He over the western North Pacific (WNP), associated with anomalous anticyclone over the subtropical WNP and southerly wind over the East Asian monsoon region (Fig. 3a), suggesting an enhancement of the WNP subtropical high (WNPSH) and the southerly monsoon circulation in the lower troposphere. The change in the atmospheric circulation at 200 hPa is characterized by increased He and anticyclone anomaly over the Asian continent with a center near the Tibetan Plateau (Fig. 3b), suggesting an enhancement of the South Asian High (SAH) in the upper troposphere. The enhancement of the WNPSH in the lower troposphere is dynamically connected to the enhancement of the SAH in the upper troposphere (Jiang et al. 2011; Wei et al. 2019).

Fig. 2
figure 2

a The response of summer (JJA) precipitation to COVID-19 (unit: %) simulated by the MME of the five selected CGCMs participating COVID-MIP. b The response of precipitation to the removal of all anthropogenic aerosols, based on MME of the 16 AGCMs. The MME-simulated changes with the same sign in more than 70% of the individual models are stippled, i.e., at least 4 of the 5 models agree in sign for (a) and at least 12 of the 16 models agree in sign for (b). The purple curve marks the continental EASM region

Full size image
Fig. 3
figure 3

a, b The response of summertime (JJA) eddy geopotential height (He, unit: m) and wind (unit: m/s) at 850 hPa (a) and 200 hPa (b) to COVID-19, based on the MME of the five selected CGCMs participating COVID-MIP. c, d The response of He and wind at 850 hPa (c) and 200 hPa (d) to the removal of all anthropogenic aerosols, based on the MME of the 16 AGCMs. The MME simulated responses of He with the same sign in more than 70% of the individual models are stippled

Full size image

The MME-simulated change in atmospheric circulation is featured by wave train pattern (Fig. 3a,b), which explains the noisy pattern of the simulated change in precipitation (Fig. 2a). In addition, the simulated changes differ a lot among the individual models (Supplementary Figs. S1 and S2), possibly suggesting that the number of models/realizations may not be large enough to effectively suppress the internal variability and stochastic model error. Therefore, the difference in the climatology between piClim-control and piClim-aer experiments based on the MME of 16 AGCMs is also shown in Figs. 2 and 3, since it is less affected by the internal variability and stochastic model error because of stronger forcing (removal of all anthropogenic aerosols), larger number of models and samples (16 models with 30-year integration), and less uncertainty from the SST variability (exactly fixed SST in AGCM simulation).

The large-scale pattern of the AGCM-simulated response over East Asia-western Pacific is consistent with the CGCMs participating in COVID-MIP, despite reversed sign of precipitation response over the Arabian Sea, the Bay of Bengal and the equatorial Pacific (Fig. 2a,b) possibly due to the absence of air-sea interaction. The AGCMs show a stronger increase of precipitation over the entire East Asian monsoon region (Fig. 2b) and enhanced southerly monsoon circulation over East Asia (Fig. 3c), associated with enhanced WNPSH and enhanced SAH (Fig. 3c,d). The AGCM simulation confirms an enhanced EASM as a fast response to reduced aerosols, which is directly caused by the processes within the atmosphere in the absence of aerosol-induced SST change. The enhancement of EASM due to emission reduction could be interpreted as a reversal of the decadal weakening of EASM forced by increased aerosol concentration (Song et al. 2014; Li et al. 2015, 2018; Tian et al. 2018a). Previous studies argued that the EASM response to absorbing and scattering aerosols are different, and the simulated enhancement of EASM during COVID-19 pandemic is consistent with a fast response to reduced scattering aerosol concentration (Wang et al. 2017; Mu and Wang 2021).

The magnitude of responses of EASM are measured by some indices in Table 1. The EASM precipitation index (PEA index), defined as the regional averaged precipitation within the entire EASM region (enclosed by the purple contour in Fig. 2), increases by 2.2% based on the 5 selected CGCMs participating in COVID-MIP. The southerly wind at 850 hPa averaged within 20°–50°N, 110°–130°E (V850 index) increases by 5.4%. The WNPSH intensity index (regional averaged 850 hPa He over 15°–40°N, 130°E–160°W) increases by 4.8%, and the SAH intensity index (regional averaged 200 hPa He over 15°–40°N, 60°–120°E) increases by 2.5% (Table 1). In all, the signs of the response in the above indices are all positive and agreed by at least 4 of the 5 models, suggesting a robust enhancement of EASM during the COVID-19 pandemic. Similarly, the responses of the above indices to the removal of all anthropogenic aerosols based on the AGCMs are also positive (second row in Table 1), and agreed by at least 12 of the 16 models, which confirms a robustly enhanced EASM as a fast response to the reduced anthropogenic aerosols.

Table 1 The forced responses, the amplitudes of interannual variability and the observed anomalies of several indices (The unit is % for all values)
Full size table

The magnitudes of interannual variability for the above indices are also evaluated by the interannual standard deviation, based on the last 150 years of the piControl experiment performed by the five selected CGCMs. To facilitate a direct comparison between the magnitudes of the forced response and the interannual variability, the interannual standard deviation is scaled by the mean value of each index and shown as percentage in Table 1 (also called "coefficient of variation"). Overall, the amplitudes of the responses in PEA and V850 indices to COVID-19 only reach about 1/3 of the amplitudes for interannual variability (2.2% vs. 6.1% for PEA index and 5.4% vs. 15.0% for V850 index), and the responses of the WNPSH and the SAH to COVID-19 are also smaller than the amplitudes of interannual variability. Indeed, the amplitude of the forced response only reaches about 10% to 40% of the standard deviation for interannual variability at most of the grid points over East Asia and WNP, for both precipitation and atmospheric circulation variables (Supplementary Fig. S3).

The year-by-year evolution of the above four indices during the five summers after COVID-19 outbreak, based on the MME of the five selected CGCMs, generally shows a downward tendency despite of evident interannual variation (thick black curve in Fig. 4). This is possibly because of the gradual rebound of the aerosol precursor emission according to the experimental design or the slow response with an SST adjustment. Indeed, it is suggested that the fast response determines the total EASM response to increased aerosol concentration in recent decades, but the slow response associated with aerosol-induced SST change partially dampens the fast response (Wang et al. 2017; Li et al. 2018; Mu and Wang 2021). Our result under the case of decreased aerosol concentration is consistent with the above studies in terms of the critical role of fast response, and suggests that the climatic effect of COVID-19 on EASM dampens quickly after the rebound of aerosol emissions, similar to the time scale of the effect of volcanic eruptions on monsoon (Man et al. 2014; Liu et al. 2016).

Fig. 4
figure 4

The temporal evolution of changes (unit: %) in PEA, V850, WNPSH and SAH indices in the 5-year integration of SSP2-4.5-COVID experiment in comparison to SSP2-4.5 experiment, based on the MME (thick black curve) and the five individual models (dashed color curve). The definition of these four indices are described in Table 1. The horizontal dashed lines show the ± 1 amplitude of interannual variability (unit: %, See Table 1) for each index

Full size image

4 Mechanism for the enhanced EASM

Following an abrupt decrease in the emission of aerosol precursors, the AOD averaged in the first two summers after COVID-19 outbreak decreases by as much as about 10% over most parts of the Asian continent and the surrounding seas, but the AOD changes little over the North Pacific to the east of 150°E (Fig. 5a), based on the MME of the selected models (EC-Earth3 excluded in Fig. 5a because of no AOD output). The surface air temperature (SAT) over Asia in summer does not show obvious increase (Fig. 5b), but the upper-tropoospheric temperature (UTT, averaged within 200–500 hPa) increases over Asia (Fig. 5c), which amplifies the zonal UTT contrast between the Asian continent and WNP. The increased UTT over the Asian continent and increased zonal UTT contrast are more evident in the AGCM simulations with the removal of all anthropogenic aerosols (Fig. 5d). Indeed, the zonal UTT contrast is a much better indicator for the strength of EASM than the SAT contrast (Li et al. 2007; Yu and Zhou 2007; Dai et al. 2013; Mu and Wang 2021), because SAT over Asian continent may decrease (increase) as a passive response to increased/decreased monsoon rainfall (Trenberth and Shea 2005; Hu et al. 2019).

Fig. 5
figure 5

The responses of AOD (a, unit: %), surface air temperature (b, unit: K) and upper-tropospheric temperature (c, unit: K) to COVID-19, simulated by the MME of the five selected models participating in COVID-MIP. d The response of upper-tropospheric temperature to the removal all anthropogenic aerosols, simulated by the MME of the 16 AGCMs under fixed SST. The MME simulated responses with the same sign in more than 70% of the individual models are stippled

Full size image

Based on the vertical profile of the averaged temperature within 20°–40°N, the increase of tropospheric temperature under COVID-19 forcing is seen over the Asian continent through almost the entire troposphere, with a maximum at the upper troposphere (Fig. 6a), and this pattern of response is confirmed by the response of the AGCMs to the removal of all anthropogenic aerosols (Fig. 6b). Associated with the enhanced zonal land–ocean thermal contrast in the troposphere, the isobaric surface over Asian continent rises (drops) at the upper (lower) troposphere relative to WNP (Fig. 6c,d). The above changes in the zonal pressure gradient enhances the southerly monsoon circulation in the lower troposphere blowing from tropical ocean into the EASM domain, and also enhances the WNPSH in the lower troposphere and the SAH in the upper troposphere.

Fig. 6
figure 6

The longitude-height profile for the simulated responses averaged within 20°–40°N. a, b temperature (unit: K), c, d eddy geopotential height (unit: m), e, f apparent heat source (Q1, unit: K/day), and g, h apparent moisture sink (Q2, unit: K/day). The left panel shows the response to COVID-19 forcing based on the MME of the five selected CGCMs, and the right panel shows the response to the removal of all anthropogenic aerosols based on the MME of the 16 AGCMs. The MME simulated responses with the same sign in more than 70% of the individual models are stippled

Full size image

The aerosols are primarily concentrated in the atmospheric boundary layer. To understand the temperature and circulation response through the troposphere, the apparent heat source (Q1) and apparent moisture sink (Q2) are calculated based on Yanai et al. (1973) to examine the changes in total diabatic heating and latent heating. The vertical profiles for the responses in Q1 and Q2 to reduced aerosols are characterized by positive values through the troposphere over subtropical East Asian continent (Fig. 6e-h). Although the pattern is noisy based on the CGCMs participating in COVID-MIP (Fig. 6e,g), the AGCM results clearly show that the enhancement of Q2 shares similar pattern and magnitude with Q1 (Fig. 6f,h), suggesting that increased condensational latent heating accounts for a major fraction of the enhanced total diabatic heating. These evidences suggest that the latent heating feedback (Levermann et al. 2009; Jin et al. 2013) may play a key role in amplifying the response of EASM to reduced aerosols. As the EASM is enhanced by the reduced aerosols, the enhanced latent heating over Asian continent further increases the tropospheric land–ocean thermal contrast, and in turn enhances the EASM circulation and monsoonal rainfall. The increase of cloud amount and precipitation may mask the direct effect of aerosols on the radiative fluxes reaching the surface (Supplementary Fig. S4) and the SAT (Fig. 5b), and the enhanced EASM is more clearly indicated by the zonal land–ocean contrast of tropospheric temperature, particularly at the upper troposphere (Yu and Zhou 2007; Dai et al. 2013; Mu and Wang 2021).

During the COVID-19 pandemic, the CO2 emission is reduced as well as aerosols. It cannot reverse the rising trend of atmospheric CO2 concentration but may slightly mitigate its rising trend (Forster et al. 2020). Previous works suggested that the increase of atmospheric CO2 concentration act to enhance the EASM (Jiang and Tian 2013; He et al. 2019; Liu et al. 2020), primarily through the fast response of EASM to increased CO2 concentration (He and Zhou 2020; Li et al. 2022). It is straightforward to infer that a reduction in CO2 emission may result in a weaker EASM compared with the scenario of no reduction in emission, which cannot explain the simulated enhancement of EASM during COVID-19 pandemic, although reduction in CO2 emission is taken into account in the experimental design of COVID-MIP (Forster et al. 2020; Lamboll et al. 2021). Therefore, the simulated enhancement of EASM is dominated by aerosol reduction. Indeed, Gettelman et al. (2021) suggested that the aerosols has the largest contribution to radiative forcing during COVID-19 pandemic, larger than CO2, ozone and contrail effects.

5 Observational evidence

The observed seasonal anomaly in a summer is strongly affected by the internal climate variability. Indeed, the record-breaking extreme rainfall along the Yangtze River valley in 2020 was primarily attributed to the interannual variability of the climate system associated with tropical SST anomalies (Takaya et al. 2020; Pan et al. 2021; Zheng and Wang 2021; Zhou et al. 2021) and atmospheric internal dynamics (Ding et al. 2021; Liu et al. 2021), particularly the warm SST anomaly over the Indian Ocean associated with a decaying El Niño event (Takaya et al. 2020; Zhou et al. 2021) and the cold SST anomaly over equatorial Pacific associated with a developing La Niña event (Pan et al. 2021). Until now, we have experienced two summers after the COVID-19 outbreak, i.e., 2020 (following an El Niño event in the preceding winter) and 2021 (following a La Niña event in the preceding winter). As averaging the observed anomalies between 2020 and 2021 helps to suppress the interannual variability in the observation, the averaged anomalies of precipitation, atmospheric circulation and UTT based on these two summers are shown in Fig. 7.

Fig. 7
figure 7

Observed averaged anomalies for the summers of 2020 and 2021. a Precipitation anomalies (unit: %), b Eddy geopotential height (shading, unit: m) and wind (vectors, unit: m/s) anomalies at 850 hPa. c Eddy geopotential height and wind anomalies at 200 hPa. d Upper tropospheric temperature anomalies (unit: K). GPCP dataset is adopted in (a) and ERA5 dataset is adopted in (b–d). The anomaly is calculated by removing the observed climatology of 1979–2021

Full size image

The observed precipitation anomalies are characterized by excessive rainfall over almost the entire continental EASM region and its northern flank, and negative precipitation anomaly over the WNP (Fig. 7a). The large-scale pattern of observed precipitation anomaly is the same as the CGCM-simulated response to COVID-19 forcing (Fig. 2a) and the AGCM-simulated response to the removal of all anthropogenic aerosols (Fig. 2b). Indeed, positive precipitation anomalies are seen over almost the entire continental East Asian monsoon region during the summers of both 2020 and 2021 (Supplementary Fig. S5). The averaged precipitation anomaly over the continental EASM region (enclosed by the purple contour in Fig. 7a) reaches 31.0% (Table 1), which is larger than the model-simulated response. The possible cause for this discrepancy may be either the residual of interannual variability or a systematic underestimation of the forced response by the models.

The observed 850 hPa atmospheric circulation anomalies are characterized by an anomalous anticyclone over the WNP with southwesterly wind anomaly over the EASM region (Fig. 7b), suggesting an enhanced WNPSH. The V850 index and the WNPSH index are 20.1% and 31.6% stronger than the climatology (Table 1), which are all positive and greater than the standard deviation for interannual variability. The observed 200 hPa atmospheric circulation anomalies are characterized by an anomalous anticyclone centered on the eastern flank of the Tibetan Plateau (Fig. 7c), suggesting an enhanced SAH. The noisy wave train pattern in Fig. 7c suggests that the two-year average is far from fully removing the internal variability. The observed anomalous UTT is characterized by a warm center from the South Asia to East Asian continent, with a relative cold center to the east of the Philippines (Fig. 7d), which is consistent with the simulated response to decreased aerosols although the observed pattern is noisier. Given all these evidences, we may have already experienced the strengthening effect of COVID-19 pandemic on EASM during 2020 and 2021 in addition to the natural internal variability.

Based on observation, recent studies suggested that both scattering aerosols and absorbing aerosols show an overall decrease over East Asia during COVID-19 pandemic (Xu et al. 2020a, b; Dong et al. 2021; Jia et al. 2021; Wang et al. 2021). According to previous studies, reduced scattering aerosols act to enhance EASM (Li et al. 2016; Wang et al. 2017; Mu and Wang 2021), while the effect of reduced absorbing aerosols on EASM is controversy (e.g., Li et al. 2016; Persad et al. 2017). The observed anomalously stronger EASM consecutively occurred in 2020 and 2021 is consistent with the climatic effect of reduced scattering aerosols, and it supplies observational evidence that the bulk effect of increased (decreased) aerosols is to weaken (enhance) the EASM (Song and Zhou 2014; Li et al.2015, 2018; Tian et al. 2018a; Luo et al. 2019; Mu and Wang 2021).

Although the overall pattern of enhanced EASM is consistent between observation and the model simulation, there still exist obvious difference in the detailed patterns of the precipitation and atmospheric circulation anomalies (cf. Figs. 2, 3 with Fig. 7). An evident difference is also seen among the individual models (Supplementary Figs. S2 and S3) and among the ensemble members of each model (figure not shown). The above discrepancy may arise from the strong internal variability of the climate system, the relatively weak forced response, and the model bias. Although the enhanced EASM is a robust signal during COVID-19 pandemic and the flood risk over East Asia is expected to be higher, it may not be feasible to predict where and when extreme flood events occur based only on aerosol reduction, Indeed, record-breaking flood occurred along the Yangtze River Valley to Japan in the summer of 2020 but over North China particularly in Henan province in the summer of 2021 (Ding et al. 2021; Wang et al. 2022).

6 Summary

Anthropogenic aerosol precursor emission decreases due to the sudden attack of COVID-19 pandemic, but its effect on EASM remains unknown. Based on five selected CGCMs participating in COVID-MIP, the EASM is enhanced because of the fast response to COVID-19 forcing, characterized by increased precipitation over the continental EASM region, enhanced southerly monsoon circulation associated with an enhanced WNPSH in the lower troposphere, and enhanced SAH in the upper troposphere. The amplitudes of the forced enhancement are 2.2% for continental monsoon rainfall and 5.4% for southerly monsoon circulation, and they reaches about 1/3 of the standard deviation for interannual variability. As the numbers of models and realizations may not be large enough to remove the internal variability and the stochastic model error, the pattern of response is somewhat noisy, but the above response is highly consistent with the simulated response of EASM to the removal of all anthropogenic aerosols based on the MME of 16 AGCMs, suggesting that enhanced EASM is a robust fast response to the decreased aerosol concentration.

As summarized in Fig. 8, the EASM is enhanced by COVID-19 via zonal land–ocean thermal contrast due to reduced aerosol concentration, and the response to aerosol forcing may probably be amplified by the positive feedback of monsoonal latent heating. Besides the warming effect of reduced aerosols on the atmosphere, the increased latent heating over East Asia associated with enhanced EASM further warms the troposphere over the continental East Asia and enhances the tropospheric zonal land–ocean thermal contrast. The increased tropospheric zonal thermal contrast results in a drop (rise) of the isobaric surface at the lower (upper) troposphere over the East Asian continent in comparison to the WNP, and the changed zonal pressure gradient in turn enhances the southerly monsoonal circulation and monsoon rainfall over East Asia. The WNPSH and SAH are also enhanced by the changes in the isobaric surfaces at lower and upper troposphere. The above mechanism primarily works as a fast response, and the effect of COVID-19 on EASM is expected to dampen soon after the rebound of aerosol emission.

Fig. 8
figure 8

Schematic diagram showing how the EASM is enhanced by the aerosol reduction during COVID-19 pandemic and the associated latent heating feedback. Forced by the aerosol reduction, the enhanced monsoonal precipitation heats the troposphere over Asian continent by latent heating, and the increase of the tropospheric zonal thermal contrast between Asia and WNP leads to a drop (rise) of the isobaric surface at lower (upper) troposphere over Asian continent relative to the WNP, which in turn enhances the EASM circulation and continental monsoonal rainfall. The WNPSH and the SAH are also enhanced by the stronger tropospheric zonal land–ocean thermal contrast. See Sect. 6 for details

Full size image

Based on the average of observed anomalies in the summers of 2020 and 2021, excessive rainfall occurred over almost the entire continental EASM region, with anomalously strong WNPSH and southerly monsoon circulation at the lower troposphere and anomalously strong SAH in the upper troposphere. All these observed features are consistent with the simulated responses in terms of the sign and the spatial pattern, suggesting that the simulated enhancement of EASM during COVID-19 is supported by the observational evidence. This study only focuses on the response of the seasonal mean monsoon circulation and rainfall to COVID-19 forcing and the mechanisms therein, and it may need further studies in future to attribute how much the COVID-19 pandemic has increased the probability of the observed extreme rainfall events over East Asia in recent two years, such as the great flood along Yangtze River valley in 2020 (Ye and Qian 2021; Yang et al. 2022) and the flood over North China particularly in Henan Province in 2021 (Wang et al. 2022).