Introduction

Deep convective systems (DCS) are pivotal in driving the climate system, propelling global circulation, and producing precipitation, especially over the tropics1,2,3. Deep convective storms and associated floods, as predominant drivers of extreme weather events, are responsible for global economic losses exceeding $140 billion annually4. Further, DCS currently become more frequent and extreme due to the moistening world with rapid global warming. Aerosols can exert a substantial impact on the formation and evolution of DCS, by acting as cloud condensation nuclei (CCN), modifying the global energy and water cycle5,6,7. This process, known as aerosol-cloud interaction, currently represents the most significant source of uncertainty in evaluating climate change8,9. Understanding the aerosol-DCS interactions is crucial in accurately assessing the impact of human activities, like air pollution from industry and burning fossil fuel10,11, on extreme weather events and climate change12,13,14.

Previous studies have proposed that aerosols have significant potential to adjust the development of DCS through multiple lines of evidence15,16,17. Adding aerosols to such clouds allows more droplet activation and condensation, enhancing the latent heat release and invigorating convection. This process is named the ‘condensational convective invigoration’15,18. Another hypothesis is the ‘mixed-phase invigoration’, positing that increased aerosol produces more cloud particles impeding warm rain formation, and promoting cloud water transport to higher altitudes and its freezing there, enhancing latent heat release and resulting in invigoration16,19. Furthermore, recent studies have shown that aerosol invigoration can have positive meteorological feedback because the invigorated convection leads to more low-level water vapor convergence, moistening atmosphere, and further invigorating convection17,20.

Although various hypotheses regarding aerosol effects on DCS have been proposed, the magnitude of their influence and even the existence of invigoration remains intensely debated21,22. Romps et al.23 suggested that aerosol invigoration may be of no practical significance due to the actual supersaturation does not meet the threshold for aerosol invigoration. Most of the studies based on observations report aerosol invigoration, but with variable magnitude. Zhang et al.24 reported that the area of convective clouds increases by a factor of 1.3 with the increase of aerosol optical depth (AOD) in East Asia. Koren et al.15 indicated that added aerosol enhances the rainfall rate from warm cumulus clouds by averaging 0.2 mm/h from clean to polluted conditions. Similarly, Han et al.25 recorded aerosol-enhanced convective rainfall rates by a factor of 2 over land based on polar-orbiting satellites. Most previous studies missed this dynamic evolution of convective cloud clusters due to the limitation of polar-orbiting satellites, resulting in the incomplete quantification of aerosol convective invigoration.

The complete lifecycle of DCS starts from initial convection, and its development to the tropopause, often followed by horizontal growth that includes splitting and merging, finally dissipating to anvils and cirrus. Several studies utilized high temporal resolution of geostationary observation to capture the convective evolution. Chakraborty et al.26 quantified an aerosol-increased rainfall amount of mesoscale convective systems by a factor up to 4 (from 3 km³ to over 12 km³) at total lifecycle using 3-h resolution data of International Satellite Cloud Climatology Project, while the daily AOD is used from Aqua satellite. Based on geostationary satellite observation, Pan et al.27 revealed that the aerosol-invigorated convective activity by enhanced factors of 1.3 and 2.6 in lifetime and rainfall amount over tropical land, respectively. However, that study possibly missed the part of DCS information when the most enlarged and long-lived DCSs spilled over the confines of the 5° × 10° domain27,28. Furthermore, although observation of geostationary satellites enables tracking the DCS lifecycle, previous studies still overlook the dynamical evolution of DCS, such as the splitting and merging process29,30.

In summary, the above-mentioned limitations likely caused a significant underestimation of the aerosol invigoration on DCS. To address the issues of splitting, merging, and spillover, we have developed a tracking method that incorporates the DCS evolution, including its merging and splitting. This new methodology enabled us to track the entire lifecycle and all components, from the genesis of the convective core to the cirrus dissipation31. We found that a DCS splits, on average, into eight sub-clouds throughout its whole lifetime. Therefore, integrating all the DCS’s splits and mergers increases the DCS’s rainfall amount by an order of magnitude compared to the previous methods of mainly tracking a single convective core31. In addition, it was found that cirrus detrained from the convective core persisted up to 10 hours after the dissipation of the core, contributing 16% total coverage of the DCS. Consequently, the tracking methodology of Yin et al.31 enables us to quantify the invigoration effects of aerosol on the total lifecycle of DCS comprehensively.

Here, benefiting from the previously developed full-tracking algorithm, we are now able to track the entire lifecycle of DCS accurately over both ocean and land. We found that the increase of fine aerosol (FA) concentrations can result in a ×5 increase of both the total area and rainfall amount of DCS over land. The combined influences of FAs and coarse sea salt aerosols (CSA) can amplify the rainfall amount of DCS by more than a factor of ten. We highlight that the previous studies have underestimated aerosol convective invigoration due to the limitation of data and method. This study comprehensively quantifies the substantial impact of aerosols on DCS, providing new insights and significantly advancing the understanding of aerosol-cloud interactions and their role in the climate system.

Results

Aerosol-driven effect on DCSs over land

Figure 1 illustrates the relationship between the BT of core, lifetime, total area, and total rainfall amount with FA in core areas over land during the total lifetime. The initial three bins represent the top 1%, 3%, and 5% segments of the clean aerosol environment, followed by 5% intervals for a total of 22 bins. In Fig. 1a, there is a notable decrease for BT of core by 4.0 K (from 216.4 K to 212.4 K) with the increase of FA over land, which implies FA drive DCS has a higher development consistent with previous study16,32,33, and further validates the aerosol-invigoration hypothesis. During the attenuation stage, the BT of the core increased by 3.0 K compared to the development stage. Notably, in the dissipation phase, only cirrus remain, therefore, this stage is not included in Fig. 1a.

Fig. 1: Variations of DCS properties with the increase in fine aerosol over land.
Fig. 1: Variations of DCS properties with the increase in fine aerosol over land.
Full size image

The a–d shows the brightness temperature of core, lifetime, total area, and total rainfall amount change with the increase of fine aerosol, respectively. The initial three bins of lines represent the top 1%, 3%, and 5% of fine aerosol concentration, illustrating the rapid change of DCS under a clean environment. The subsequent bins are spaced at 5% intervals and comprise a total of 22 bins. Each bin is accompanied by the standard error of the I-type shape. The red line represents the average value during the whole lifetime. The deep, moderate, and light blue color lines indicate the averaged value during the development, attenuation, and dissipation stages, respectively.

Figure 1b reveals that the lifetime of DCS exhibits a non-monotonic response to varying FA concentrations. Specifically, the lifetime of DCS demonstrates a discernible increase of 1.6 times, transitioning from 6.3 hours to 10.0 hours, as the system transitions from a state characterized by minimal FA loading to optimal FA loading, around 5 μg/m³. This marks a critical juncture wherein the lifetime of DCS undergoes a shift from an increasing to a decreasing trend with rising aerosol concentration. The lifecycle reduction caused by high aerosol loads may be due to three reasons. Firstly, an excessive concentration of FAs directly absorbs and reflects solar radiation, thus blocking surface heating. This diminishes surface heating, thereby reducing the energy available for convective processes34,35. Secondly, a heavy aerosol load suppresses the precipitation-forming processes beyond the point of optimal invigoration16,19,36. In addition, heavy rain scavenges the aerosols. Therefore, one should expect a negative correlation between rainfall and aerosol amounts. The increasing rainfall up to the optimal aerosol loading counters this expectation, thus related to the aerosol convective invigoration effect.

In Fig. 1c, d, the sensitivity of total area and rainfall to FA variations is pronounced. Figure 2 delves deeper, comparing total area and rainfall amount at different stages under both pristine and optimal FA conditions. The total area of DCS achieves a quintupling, expanding from 0.37 to 1.74 × 106 km2. In the scenario of optimal concentration, the area during the attenuation stage is nearly twice as large as in the development stage (0.63 × 106 km2 and 1.02 × 106 km3) for a comparable duration, with the anvil covering a significantly broader area during dissipation. Figure 2b illustrates a 5-times increase in rainfall amount, from 0.16 to 0.79 km3, with the majority of this increase stemming from the core. Notably, the rainfall amount remains consistent across both the development and attenuation phases (0.33 km3 vs 0.31 km3). The factors for the increase in area and rainfall amounts of different components are displayed in Table S2. The increase in FAs has led to the upward transport of more cloud water and enhanced condensation efficiency, resulting in the formation of more and smaller ice crystals, thereby enlarging the convective area, and boosting precipitation16,26,37.

Fig. 2: DCS properties from different cloud components under pristine and optimal fine aerosol concentrations over land.
Fig. 2: DCS properties from different cloud components under pristine and optimal fine aerosol concentrations over land.
Full size image

Total area (a) and rainfall amount (b) of DCS under pristine and optimal fine aerosol concentration over land, respectively. Each bar represents the total area or rainfall amount of the three components within DCS, with red, yellow, and blue corresponding to the core, anvil, and cirrus, respectively. The numbers indicate the average values of these DCS properties under pristine and optimal fine aerosol concentration conditions. The complete lifecycle and its various stages are separated by gray dashed lines.

To further reveal the invigoration effect of the FA on the convective system, Fig. 3 presents the probability density function (PDF) and cumulative distribution function (CDF) of BT of core, lifetime, total area, and rainfall amount across pristine, optimal, and highly polluted aerosol conditions, respectively. Figure 3a demonstrates the increasing trend in DCS’s height from clean to optimal concentration scenarios through to the most polluted conditions. Figure 3b demonstrates that under pristine conditions, DCS exhibits narrow lifecycle distribution, with nearly 90% under 10 hours. Conversely, at optimal FA concentration, the lifetime of DCS can extend to ~35 hours. Similarly, Fig. 3c, d show that while pristine conditions confine DCS within 3 × 106 km² and 1 km3 for area and rainfall, respectively, optimal conditions broaden these ranges to 50 × 106 km2 and over 10 km3. These findings underscore that convection can only develop into a more vigorous state under polluted conditions leading to an increase in DCS’s area and rainfall amount by an order of magnitude over land.

Fig. 3: Probability density functions (PDF) and cumulative distribution function (CDF) of DCS properties over land under different concentrations of fine aerosol.
Fig. 3: Probability density functions (PDF) and cumulative distribution function (CDF) of DCS properties over land under different concentrations of fine aerosol.
Full size image

The panels indicate the information of a brightness temperature of core, b lifetime, c total area, and d rainfall amount, respectively. The blue and red bars and lines indicate the PDF and CDF on pristine and optimal fine aerosol concentration. The yellow bar and line in a represent the PDF and CDF of the brightness temperature of core in the most polluted fine aerosol concentration.

Size-depended aerosol convective invigoration over ocean

The humid atmospheric conditions over the ocean, coupled with the abundance of CSA, impart a distinct influence on the aerosol-driven effects in oceanic DCS, distinct from those observed over land. Figure 4a illustrates that DCS over the ocean display higher BT and a weaker sensitivity to increasing FA compared to those over land. The lifetime of DCS increases from 8.2 h to 10.3 h when from pristine to optimal FA concentration and stabilizes after 1.0 μg/m3 (Fig. 4b). Both the total area and rainfall amount reached the highest point around 3 μg/m3, which increased 4 and 5 times compared with the pristine state, respectively (Fig. 4c, d). Generally, DCS over the ocean exhibits a weaker response to FA compared to those over land, characterized by less intense convective development and smaller spatial extents. However, they tend to have similar lifetimes and produce larger precipitation at optimal FA concentration.

Fig. 4: Variations of DCS properties with the increase in fine aerosol (top) and coarse sea salt aerosol (bottom) over the ocean.
Fig. 4: Variations of DCS properties with the increase in fine aerosol (top) and coarse sea salt aerosol (bottom) over the ocean.
Full size image

a–d and e–h show the brightness temperature of core, lifetime, total area, and total rainfall amount change with the increase of fine aerosol and coarse sea salt aerosol over ocean, respectively. The initial three bins of lines represent the top 1%, 3%, and 5% of fine aerosol concentration, illustrating the rapid change of DCS under a clean environment. The subsequent bins are spaced at 5% intervals and comprise a total of 22 bins. Each bin is accompanied by the standard error of the I-type shape. The red line represents the average value during the whole lifetime. The deep, moderate, and light blue color lines indicate the averaged value during the development, attenuation, and dissipation stages, respectively.

Contrasting with the effects of FAs, the addition of CSA increases the BT of core, with Fig. 4e showing a rise from 218.4 K to 220.9 K. Figure 4f illustrates that the lifetime of DCS over the ocean extended from 7.8 hours to 10.5 hours around 70 μg/m³. Notably, from pristine to optimal CSA levels, the area expands 4-times from 0.3 to 1.4 ×106  km2, similar to the effect of FAs, but with a more pronounced increase in rainfall from 0.20 to 1.1 km3, as shown in Fig. 4g, h. These counteracting trends observed between BT of core and other parameters suggest that the formation of lighter rainfall by CSA might delay atmospheric energy dissipation, thereby extending the DCS’s lifetime, area, and precipitation. CSA promotes warm rain by acting as giant CCN, facilitating the formation of larger cloud droplets38. These larger droplets enhance the efficiency of collision-coalescence processes, leading to earlier warm rain development39,40. This would reduce the vertical transport of water to higher freezing altitudes, resulting in weak and moderate convection. As a result, more convective energy remains in the atmosphere, aiding the maintenance of moderate convective systems. The delayed dissipation extends their lifetime, enlarges the area, and increases the total precipitation of DCS.

Considering the coexistence of FAs and CSA, it is necessary to further clarify the competitive effects of these two types of aerosols on DCS over the ocean. Figure S1 illustrates that under the same CSA concentration, DCS can develop to higher altitudes, larger sizes, longer durations, and exhibit increased rainfall in the presence of elevated FA concentrations. Furthermore, Fig. 5 illustrates the function among various DCS parameters in response to different levels of FAs and CSA over the ocean. Consistent with Fig. 4a, e, Fig. 5a depicts a linear downward trend in the BT of the Core, transitioning from a high concentration of CSA and a low concentration of FA, to a low concentration of CSA and a high concentration of FA. Figure 5b demonstrates an extension in the lifecycle duration from 8.2 hours under pristine conditions, delineated by the cleanest 1% for FAs, to 11.6 hours under conditions of dual optimal concentrations, around 5 μg/m³ for FAs and 80 μg/m³ for CSA. Figure 5c, d reveal that the area of DCS expanded from 0.40 to 2.1 × 106 km2, and precipitation increased from 0.25 to 1.7 km3 under this combined optimal concentration. These findings suggest that a more marked enhancement effect is observed at the optimal concentration combination of the two aerosol types, compared to the influence of a single aerosol type.

Fig. 5: Co-variations of DCS characteristics with the increase of fine aerosol and coarse sea salt aerosol (CSA) concentration over the ocean.
Fig. 5: Co-variations of DCS characteristics with the increase of fine aerosol and coarse sea salt aerosol (CSA) concentration over the ocean.
Full size image

The a–d illustrate the function of BT of Core, lifetime, total area, and rainfall amount with distinct bins of fine aerosol and CSA over the ocean. The figure shows bins with a minimum sample size of five.

Identify the influence of meteorological factors on aerosol effect

Aerosols and DCS are simultaneously influenced by a range of meteorological factors. The observation that DCS evolution driven by aerosols may be attributed to changes in meteorological elements highlights the necessity of distinguishing the specific impact of these elements on DCS. Therefore, clarifying the distinct influence of aerosols on DCS becomes imperative. This study selected five key meteorological parameters that have shown a robust correlation with convective development in prior research: precipitable water (PW), convective available potential energy (CAPE), vertical velocity at 450 hPa (ω), wind shear of 850–200 hPa (WS), and temperature at 850 hPa (T). These parameters are divided into three levels of bins with comparable sample sizes. Within each bin, it is hypothesized that meteorological parameters exert a consistent influence on the interaction between aerosols and DCS.

Figure 6 illustrates the relationships between the BT of core, lifetime, total area, rainfall amount, and FA concentration over land, categorized within different bins of meteorological cloud-controlling factors. A decrease in the BT of the core, extended lifetime, and an increase in both area and rainfall were observed with added FA. It is noteworthy that in conditions unfavorable for DCS’s development, characterized by low PW, CAPE, and high wind shear, the effect of FA on decreasing BT of the core becomes more pronounced. Under conditions of moderate CAPE, strong ω, moderate wind shear, and moderate lower tropospheric temperatures, the FAs have the strongest effect on DCS’s total area and rainfall amount. Similar meteorological perturbations applied to the impact of FA on DCS have also been observed over the ocean as shown in Figure S2 and Fig. 7. Moreover, Fig. 7 reveals that, under the most favorable meteorological conditions, the synergistic impact of the two aerosol types is greater than the effect of FAs in isolation. Under conditions of high PW, strong ω, moderate wind shear, and moderate lower tropospheric temperatures, a pronounced enhancement is observed, particularly for precipitation, which can experience up to a 10-times increase.

Fig. 6: Effect of fine aerosol on tracked deep convection systems at different fixed meteorological bins over land.
Fig. 6: Effect of fine aerosol on tracked deep convection systems at different fixed meteorological bins over land.
Full size image

The a–d indicate BT of core, lifetimes, total areas, and rainfall amounts of DCS over land. The light blue, medium blue, and dark blue represent the value of DCS’s parameters under pristine fine aerosol (FA) conditions (cleanest 1% of fine aerosols), in low, middle, and high values of meteorological factors bins, respectively. Green marks in Fig. 6b–d (Fig. 6a) represent the value of DCS’s parameters under optimal (most polluted 5%) fine aerosol conditions. The number on the green marks represents the specific changes in value in a and the multiple variations in b–d. The ranges for the bins of the five meteorological factors: precipitable water (PW), convective available potential energy (CAPE), vertical velocity at 450 hPa (ω), wind shear of 850–200 hPa (WS), and temperature at 850 hPa (T) are noted in parentheses.

Fig. 7: Same as Fig. 6, but for the variations of DCS with fine aerosol and the optimal combinations of fine aerosol and CSA over the ocean.
Fig. 7: Same as Fig. 6, but for the variations of DCS with fine aerosol and the optimal combinations of fine aerosol and CSA over the ocean.
Full size image

The green marks represent the value of DCS’s parameters in optimal fine aerosol conditions (around 5 μg/m³ of FA). The red bars represent the values of DCS parameters in the combined optimal concentration of fine aerosols and CSA (around 5 μg/m³ of FA and 80 μg/m³ of CSA), and the number on the red bar represents the factor of enhancement with respect to the pristine FA.

This large enhancement factor is too large to be caused by the primary effect of the aerosols alone and, therefore, must involve positive feedback processes of the adjustment of the circulation and moisture convergence to the primary effect as shown by Zang et al.17. They noted that consecutive convection events increase atmospheric PW, thereby enhancing moisture supply and intensifying subsequent convective events in a positive feedback loop. Part of this feedback may affect the cloud-controlling factors, which may cause some covariation with the aerosols.

Discussion

Our study indicates that prior to reaching an optimal concentration, aerosols can significantly enhance DCS, resulting in elevated cloud tops, extended lifespans, expanded impact areas, and increased precipitation up to 5 times over land. With added coarse sea spray the enhancement factor can reach up to 10 times. This result indicates the strong potential of aerosol to adjust the regional distribution of convective precipitation. Nonetheless, the total precipitation must balance the total evaporation, which is in turn determined by the solar energy reaching the surface. Even global warming results only in a slight increase in precipitation41. This suggests that aerosols predominantly influence convective precipitation through a spatial redistribution mechanism, transferring rainfall from relatively unpolluted regions to those with higher pollution levels, without augmenting the overall precipitation volume.

The driving mechanism is the invigorated convection in polluted regions releases more latent heat, leading to moisture convergence and further enhancement of the convection in a positive feedback loop20. This intensified circulation leads to stronger subsidence and drier conditions in nearby clean areas, thereby suppressing convective activities in these regions17,42. Such notable shifts in rainfall intensity and spatial distribution may emerge as one of the primary risks associated with climate change, such as extreme floods and drought.

Although aerosols have been found to significantly influence DCS development, there is still an intense debate over the magnitude and even the very existence of aerosol effects on DCS. Methodological limitations of quantifying the invigoration contributed to the uncertainties. To address this, we have developed and applied a DCS tracking methodology that enables us to comprehensively track all DCS components while accounting for splitting and merging events31. Based on this methodology, we quantified the invigoration effects of FAs and CSA on DCS. The key findings from this study can be summarized in the following points:

  1. (1)

    The invigoration effect of FA on DCS has been quantified throughout its complete lifecycle over tropical land, resulting in a 5-times increase in total DCS area and rainfall amount, and up to an order of magnitude increases in extreme convective events, from pristine to optimal FA concentration.

  2. (2)

    Coarse sea spray aerosols can impede the upward development of convection, diminishing the invigorating effect of FAs on oceanic DCS, resulting in moderately strong DCS. Nonetheless, added CSA concentration still led to a 4-times expansion in total area and a 5-times increase in rainfall amount.

  3. (3)

    The resultant DCS invigoration of the combination of optimal concentrations of FAs and CSA is greater than can be caused by either aerosol type alone. The related aerosol-driven invigoration on total rainfall amount of DCS can reach up to 10 times with aerosols changing from clean to optimal concentrations.

The present results show much greater invigoration than previous studies for the following reasons:

  1. (1)

    Previous tracking methods did not account for the full growth of the DCS, because they split off in time and space as they grow. Our methodology integrates the entire evolution of the DCS’ splits in time and space.

  2. (2)

    Previous methods using fixed domains, although effective in examining the integrated aerosol invigoration effect on DCSs within a region, may fail to capture the most active convective systems as they can spillover the domain as they grow, thus clipping the invigoration effect.

  3. (3)

    Previous studies missed the incorporation of the combined effects of fine and coarse sea spray aerosols that led to additional enhancement of the rainfall from tropical marine DCS.

These results show that interactions of aerosol and DCS have crucial potential to redistribute global precipitation and alter atmospheric circulation. Especially, aerosol can intensify extreme weather events, such as floods and droughts. Therefore, this study underlines that detailed aerosol processes must be considered in order to improve the overall accuracy of numerical weather prediction models and climate models.

Methodology

Observation and reanalysis data

Here, the multi-source observation and reanalysis data are used to track the full lifecycle of DCSs and quantify the aerosol effect on DCS from 2004 to 2012, spanning −50°W to 50°E and −25°S to 25°N, as shown in Fig. 8. The GPM IMERG product with 30-minute intervals and 0.1° × 0.1° spatial resolution was chosen to allocate precipitation to each DCS. MERRA-2 aerosol component data are employed to represent aerosol concentrations in convective regions. Also, multiple meteorological parameters from NCEP-fnl data are used to investigate the influence of meteorological conditions on the interaction between aerosols and DCS. The detail is provided in Table S1.

Fig. 8: Map for region of interest.
Fig. 8: Map for region of interest.
Full size image

The colors represent the frequency of DCSs’ occurrence, the number of observed DCS divided by total observations at each pixel.

The METEOSAT Second Generation (MSG) geostationary satellite is used to track the full lifecycle of DCS. The primary sensors equipped on the MSG satellite include the Spinning Enhanced Visible and Infrared Imager (SEVIRI) and the Geostationary Earth Radiation Budget (GERB) instruments. SEVIRI offers 12 spectral channels with a 15-minute temporal and 3 km spatial resolution at substellar point43. GERB measures the radiation budget at the atmosphere zenith by recording the energy of long-wave and short-wave radiation at the top of the atmosphere. This data will be utilized in subsequent studies to quantify the radiative forcing attributable to aerosol-convection interactions. Here, the brightness temperature (BT) of the 12 μm band is used to identify different components of DCS, as shown in Yin et al.31.

The Global Satellite Precipitation Program (GPM) is an international initiative for satellite precipitation observation, spearheaded by the National Aeronautics and Space Administration and Japan Aerospace Exploration Agency44. The project aims to intercalibrate, merge, and interpolate observations from multiple satellites to cater to global scale considerations with high spatiotemporal coverage. Before the 2014 launch of core GPM, the GPM Global Precipitation Mission data primarily relied on precipitation observations from the Tropical Rainfall Measurement Mission (TRMM), which has demonstrated commendable observational accuracy45. The average absolute monthly deviation of rainfall estimates from TRMM is approximately 9% when compared to rain gauge datasets46.

The National Centers for Environment Prediction final operational global reanalysis (NCEP-fnl) data offers reanalysis of meteorological data with a spatial grid resolution of 1° × 1° and a time resolution of 6 hours. It encompasses 26 pressure layers in the vertical direction, ranging from 1000 hPa to 10 hPa. Modern-Era Retrospective Analysis for Research and Application Version 2 (MERRA-2) represents a comprehensive, long-term global atmospheric reanalysis dataset developed based on the Goddard Earth Observation System Data Assimilation System. MERRA-2 represents the first global reanalysis of aerosol data that incorporates space-based observations along with the internal physical processes of weather systems, with a spatial resolution of 0.5° × 0.625° and a time resolution of 3 hours. MERRA-2 assimilates six distinct aerosol concentration categories from various instruments, including Moderate Resolution Imaging Spectroradiometer and Aerosol Robotic Network, to reduce model biases47,48. Previous studies have validated that MERRA-2 AOD shows significant spatial consistency with AERONET ground measurements globally, with a correlation coefficient of 0.85, and up to 0.93 over oceans48,49.

Data collection and matching

The multi-source observation and reanalysis data are aligned to the GERB 9 × 9 km grid using the nearest neighbor interpolation method, facilitating subsequent aerosol-convection-radiation research. To enable a more accurate assessment of DCS rainfall, this study opted for a 30-minute resolution precipitation data of GPM. For the reanalysis data with a resolution lower than 30 minutes (MERRA-2 and NCEP), linear interpolation was employed to resample these data to a consistent 30-minute resolution31. Specifically, sulfates, organic carbon, black carbon, and fine sea salt are chosen to depict the concentration of FAs (with a radius smaller than 1 μm), which have been confirmed to serve as an excellent representative of CCN27. Coarse sea salt is employed to represent the presence of giant-sized aerosols, which have been found to promote warm rain precipitation over the ocean38. All these aerosol types are sampled from the near-surface layer.

Methodology of DCS full-lifecycle tracking

Here, the full lifecycle and related clouds of DCS are tracked based on our previous method, named as Full-Tracking Algorithm for Convective Thunderstorm System (FACTS)31. The main steps are outlined as follows:

  1. 1.

    FACTS uses a BT threshold of 235 K and a rainfall rate of 1 mm/h to identify the core and 253 K to identify the anvil and cirrus because of the typical glaciation temperature of deep tropical convective clouds. The difference between anvil and cirrus is whether they are connected to the core. Cloud pixels with higher BT are identified as supercooled water clouds or water cloud pixels.

  2. 2.

    Based on the principle that BT increases from the core to the edge, with BT peaking at boundaries between adjacent clusters, FACTS identifies these boundaries using the BT gradients method.

  3. 3.

    FACTS continuously tracks each cloud from its formation to its dissipation. When a cloud splits, it continues to track all sub-clouds until each one dissipates. If small clouds merge into larger clouds, the lifecycle of the smaller clouds is terminated. In this process, every cloud and its evolution are recorded.

  4. 4.

    FACTS systematically organizes convection clouds splitting from the same parent convective core into the same convective system and merges smaller surrounding convective clouds. Consequently, the properties of the system throughout its lifecycle are derived by aggregating the properties of all constituent clouds, including sub-clouds, from splitting and merging at each time.

These process features ensure that FACTS can track the complete lifecycle of a convective system from the birth of the core to the dissipation of the cirrus, maintaining the integrity of all components of an individual system. It should be noted that the current study does not consider low-level water clouds as part of the system. They will be considered in calculations of the radiative forcing of DCS in our future work.

While this comprehensive approach is effective, FACTS ignores convective activity before the core reaches 235K. However, since the DCC quickly reaches this altitude, it contributes little to the overall precipitation and area. Thus, this methodological limitation does not significantly impact the assessment of aerosol effects, as the primary climatic impacts of DCS are reflected in precipitation driven by the core, while the radiation is dominated by area and duration of anvil.

Figure 9 shows a 10-hour tracking case on June 10th, 2011. Anvil is formed when the core expands horizontally upon reaching the tropopause. When part of the anvil detaches from the convective core, it transitions into cirrus. The complete dissipation of cirrus signifies the end of the DCS lifecycle. Figure 10 illustrates the evolution of area and rainfall flux of different components, as initially introduced in Fig. 9. To investigate the variations in aerosol-DCS interaction across different developmental stages, we categorize the entire lifecycle of DCS into three stages. The development stage begins with the formation of a convective core and lasts until its rainfall flux peaks. Following this peak, the phase until the core dissipates is called the attenuation stage. The dissipation phase starts with the core’s disappearance, marked by only cirrus remains, and concludes at the end of the DCS lifecycle.

Fig. 9: Tracked the full lifecycle of a DCS case through the FACTS method on 10th June 2011.
Fig. 9: Tracked the full lifecycle of a DCS case through the FACTS method on 10th June 2011.
Full size image

Red, yellow, and blue represent the core, anvil, and cirrus pixels, respectively, while the boundary of the cloud is depicted in black.

Fig. 10: Variations of area and rainfall flux of a DCS throughout its lifetime, which is the same DCS as shown in Fig. 9.
Fig. 10: Variations of area and rainfall flux of a DCS throughout its lifetime, which is the same DCS as shown in Fig. 9.
Full size image

The rainfall flux is equal to the rainfall rate by the area, and the rainfall amount is determined by accumulating this flux throughout the total lifecycle. In this case, the period from 18:30 to 20:30 marks the development stage, the period from 21:00 to 2:00 is the attenuation stage, and the period from 2:30 to 4:30 marks the dissipation stage. The values in parentheses represent the average value over the lifecycle.

To accurately characterize the aerosol invigoration effects on the entire lifecycle of DCS, we excluded three types of convective events: those with a core duration shorter than 2 hours, those with incomplete development, and those occurring consecutively. The lifecycles of the first two types of DCS are incomplete. Continuous convection, where multiple convective events occur consecutively at the same location, can falsely prolong the lifecycle of DCS, leading to a biased and inaccurate representation of aerosol impact. Ultimately, we obtain a DCS with a full lifecycle that exhibits a single rainfall flux (area multiplied by the rate of rainfall at each moment) peak. Following a nine-year tracking period from 2004 to 2012, we have identified a total of 80264 land systems and 33422 ocean systems. These are classified based on the criterion that 70% of the cloud area during its lifecycle is predominantly located over either land or ocean.