Aided by the Modern-Era Retrospective analysis for Research and Applications-Version 2 (MERRA-2) reanalysis, in combination with the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Tropical Rainfall Measuring Mission (TRMM) satellite data, we conducted comprehensive assessments of aerosols and precipitation across Southeast Asia (the region encompassing the southern part of China) as shown in Fig. 1 (Fig. 2). For the investigation employing reanalysis data, we examined a time span of 39 years, ranging from 1980 to 2018 (“MERRA analysis”). As for the analyses utilizing the MODIS and TRMM data (“satellite analysis”), we focused on a period of 19 years, from 2000 to 2018, limited by the availability of TRMM data.
Using MODIS data from 2002 to 2012, Chang et al. (2015) have shown that the monthly-averaged aerosol optical depth (AOD) in Southeast Asia (the southern part of China) peaks in September and October (March and April), when wildfires in that region are most frequent and intense. These events in September and October (March and April) are referred to as “the SO fire (MA fire)” in this paper and result in AOD peaks in Sumatra and Kalimantan (the southern part of China) (Chang et al., 2015). Henceforth, Sumatra and Kalimantan areas (the southern part of China), as marked by red rectangles in Fig. 1 (Fig. 2), are referred to as the SK (SC) area.
In order to examine the effects of biomass burning aerosol on clouds and precipitation, we categorize September and October days for the SO fire and March and April days for the MA fire into two groups: polluted and clean days. For each of SO and MA fires, this categorization is performed separately for the MERRA analysis, utilizing AOD data from the MERRA-2, and for the satellite analysis, utilizing AOD data from the MODIS dataset. Here, AOD serves as a proxy for aerosol concentration, per numerous previous studies39–42. Days for which the domain-averaged AOD belongs to the lowest 20% percentile are defined as clean days, while days for which the averaged AOD belongs to the highest 20% percentile are defined to as polluted days. Subsequently, a comparison is made between the clean and polluted days in terms of precipitation, for each analysis and for each of SO and MA fires. This comparison allows us to observe apparent relations between aerosol concentration and precipitation.
Figures 1a, 1b, 1d and 1e show the AOD spatial distributions for the SO fire, which are averaged over each of the clean and polluted days and for each of the MERRA- and satellite-analyses, and Figs. 2a, 2b, 2d and 2e show those distributions for the MA fire. Differences in the average AOD between the polluted and clean days are greatest in the SK area for the SO fire and in the SC area for the MA fire as marked by red rectangles where most of fire spots are located for both the satellite- and MERRA-analyses (Figs. 1c,1f, 2c and 2f). These differences represent aerosol perturbations and here, concentrated in the SK area for the SO fire and in the SC area for the MA fire. Fire spots in the SK area are associated with oil palm and pulpwood plantations in corporate areas, as well as forest concession areas in non-corporate areas43. The SC area, which is China’s (as well as the subtropics’ overall) most densely forested region, exhibits a high frequency of fire occurrences and accompanying high aerosol emissions44–45. Figures 3a, 3b, 3d and 3e for the SO fire and Figs. 4a, 4b, 4d and 4e for the MA fire show the spatial distributions of precipitation that is averaged over each of the polluted and clean days and for each of the MERRA- and satellite-analyses. For the MERRA-analysis, precipitation data come from the MERRA-2, while for the satellite-analysis, precipitation data come from the TRMM.
a. SO fire
Most of precipitation occurs north of 10ºS for both the polluted and clean days (Figs. 3a, 3b, 3d and 3e). Upon examining Figs. 3c and 3f, which display the disparity in precipitation distributions between polluted and clean days, it becomes apparent that most of increases in precipitation intensity and amount on polluted days as compared to that on clean days occur over the ocean areas, specifically to the east of the Philippines (red rectangles in Figs. 3c and 3f). Conversely, decreases in precipitation intensity and amount on polluted days are evident over and around the island area, which include Sumatra, Java, Kalimantan, Sulawesi and Iran Java, and Indochinese Peninsula (blue rectangles in Figs. 3c and 3f). For simplicity, henceforth, the combined region encompassing the island area and the Indochinese Peninsula will be referred to as "the Indochinese-island area". Hence, it is observed that associated with aerosol perturbations concentrated in the SK area, precipitation changes throughout the domain, with differing responses between the Indochinese-island and ocean areas. This means that differences in precipitation between the clean and polluted days extend beyond areas where aerosol perturbations mainly occur and exhibit distinct features between the Indochinese-island and ocean areas.
For the SO fire, at the level of 500 hPa where generally updrafts in convective clouds above the planetary boundary layer reach the mature stage, we looked at vertical motions (i.e., updrafts and downdrafts) in the MERRA-2 data (Figure S1). Figures S1a and S1b illustrate that generally in the northern part of the domain between ~ 5ºS and ~ 20ºN, updrafts are developing. Conversely, in the southern part between ~ 20ºS and ~ 10ºS, downdrafts are observed. This is attributed to the characteristic pattern of the Hadley circulation during September and October, based on the analysis of the wind field over the last 40 years46. It is found that in September and October, in general, the updraft branch of the Hadley circulation occurs between ~ 10ºS and 20ºN and its downdraft branch occurs between ~ 30ºS and ~ 10ºS46. This circulation pattern results in overall northward airflow near the surface, and upper-level southward airflow at the 200 hPa level between ~ 20ºS and ~ 20ºN (Figures S2a, S2b, S2d and S2e).
Differences in updrafts between the polluted and clean days show that in the updraft branch of
the Hadley circulation, there is suppression of updrafts over polluted days as compared to those over clean days particularly in the Indochinese-island area (Figure S1c). Conversely, in the ocean area, specifically to the east of the Philippines, there is enhancement of updrafts over polluted days in the updraft branch of the Hadley circulation (Figure S1c). This consequently results in a suppression of precipitation in the Indochinese-island area while enhancing precipitation in ocean areas of the updraft branch. However, differences in vertical motion and precipitation in the downdraft branch of the Hadley circulation between clean and polluted days appears to be negligible.
Looking at differences in the horizontal wind around the surface between the polluted and clean days, there is intensified northeastward wind over polluted days that blows primarily from the Indochinese-island area to the ocean area, which is especially prominent to the east of the Philippines, as marked by a red circle (Figure S2c). Simultaneously, in the upper atmosphere (200 hPa), strengthened southwestward or westward wind is observed over polluted days that predominantly blows from the ocean area, which is to the east of the Philippines, towards the Indochinese-island area, as marked by a red circle (Figure S2f). This indicates that associated with aerosol perturbation, there is the formation of a perturbed circulation which is composed of weakened updrafts in the Indochinese-island area, a modified horizontal airflow between the Indochinese-island and ocean areas, and strengthened updrafts in ocean areas. In other words, the sea-breeze circulation between the Indochinese-island and ocean areas intensifies and this leads to suppressed precipitation in the Indochinese-island area and enhanced precipitation in the ocean area over polluted days.
b. MA fire
For both of the polluted and clean days, the primary precipitation band, as encased in red rectangles, is contained in the domain between ~ 20°S and ~ 5°N (Fig. 4a, 4b, 4d and 4e). A secondary precipitation band forms between ~ 20°N and ~ 40°N as encased in blue rectangles. We see that the primary band is in a region far south of the SC area with the primary aerosol perturbation, while most of the secondary band is to the north and east of the SC area. Hence, the SC area does not overlap with the primary precipitation band and most of the region containing the secondary precipitation band. Difference maps (Figs. 4c and 4f) show that an enhancement of precipitation occurs in locations between 20°S and the equator, in the primary band which is far south of the SC area, over the polluted days as compared to the clean days. In the secondary precipitation band, a suppression of precipitation mostly occurs, which is confined in a region to the north and east of the SC area over the polluted days. Keep in mind that aerosol perturbation is concentrated in the SC area; outside the SC area, there are negligible differences in AOD between the clean and polluted days. Hence, similar to what was seen for the SO fire, regarding the MA fire, differences in precipitation between the clean and polluted days extend beyond the SC area where the primary aerosol perturbation is present and show distinctly different features between the primary and secondary bands of precipitation.
For the MA fire, we similarly looked at updraft and downdraft spatial distributions in the MERRA analysis at the 500 hPa level (Figures S3a and S3b). Updrafts are dominant in the latitudes between ~ 20°S and ~ 5°N. However, in the continental and ocean areas north of ~ 20°N, neither updrafts nor downdrafts prevail as updraft and downdraft patterns are mixed. In Figure S3c, we observe that with the transition from the clean condition to the polluted condition, which is mostly due to the SC-area-concentrated aerosol perturbation, the following changes occur: Updrafts between ~ 20°S and the equator intensify; conversely, in locations north of ~ 20°N, other than some small localized spots where updrafts intensify, updrafts generally weaken, while downdrafts strengthen, as marked by a red circle. Note that locations north of ~ 20°N with weakened updrafts and strengthened downdrafts overlap with those exhibiting suppressed precipitation, while locations between ~ 20°S and the equator with intensified updrafts overlap with those with the enhanced precipitation (Figs. 4c, 4f and S3c). Hence, it can be concluded that overall, the transition from the clean condition to the polluted condition causes intensified updrafts in the area between ~ 20°S and the equator leading to precipitation enhancement in the primary precipitation band. On the flipside, the same transition causes weakened updrafts and strengthened downdrafts in the area north of ~ 20°N which lead to precipitation suppression in the secondary precipitation band.
Regarding wind close to the surface, we see westward trade winds between ~ 20°S and ~ 20°N in both of the clean and polluted conditions (Figures S4a and S4b). Outside the trade-wind belt, particularly north of ~ 20°N, eastward winds prevail (Figures S4a and S4b). Figure S4c shows that associated with the transition from the clean days to the polluted days, there is a predominant southward-wind tendency forming around the surface in the region marked by a red circle. Moreover, the transition also contributes to an overall intensification of northward winds in the upper atmosphere that is at the 200 hPa level (Figure S5), with the most significant intensification occurring in the Philippine Sea and East China Sea as marked by a red circle (Figure S5). Differences in wind flow around the surface and in the upper atmosphere, as marked by red circles, between the clean and polluted conditions represent a typical baroclinic vertical structure.
In summary, the southward-wind tendency close to the surface, the northward-wind tendency in the upper atmosphere, the general tendency of weakened upward air motion and strengthened downward air motion in areas north of ~ 20°N, and the tendency of strengthened upward air motion in areas between ~ 20°S and the equator constitute as a whole a perturbed circulation induced by the transition from the clean days to the polluted days. This anomalous circulation causes enhanced precipitation in the primary precipitation band and suppressed precipitation in the secondary precipitation band.
The updraft region between ~ 20°S and ~ 5°N is climatologically associated with the upward branch of the local Hadley circulation, while the downward branch of the Hadley circulation is observed between ~ 5°N and ~ 30°N in March and April46. However, in our analysis as shown in Figures S3a and S3b, the downward branch between ~ 20°N and ~ 30°N is no longer clearly discernible, since as mentioned above, updraft and downdraft patterns become mixed. This can be attributed to significant topographic and land-use variations in the continental area as well as the land-sea contrast, which create a downstream Rossby wave and a spatial inhomogeneity of convective energy, and disrupt the downdraft branch of the Hadley circulation in the latitude band between ~ 20°N and ~ 30°N. Hence, in general terms, it can be concluded that overall, the perturbed circulation tends to intensify the upward motion in the updraft branch of the Hadley circulation, weaken the upward motion and strengthen the downward motion in the disrupted downdraft branch of the Hadley circulation. In short, the transition from the clean days to the polluted enhances the overall strength of the Hadley circulation.