Spatial structure of cloud effective radius is associated with rainfall intensification in eastward propagating convective systems

Spatial structure of cloud effective radius is associated with rainfall intensification in eastward propagating convective systems | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Spatial structure of cloud effective radius is associated with rainfall intensification in eastward propagating convective systems Jing Sun, Yunying Li, Xiong Hu, Zitong Chen, Xiang Lin, Fan Li This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-9477802/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Organized convective systems contribute substantially to warm season rainfall, yet their cloud microphysical structure is often analyzed as horizontally homogeneous. Here we introduce a Voronoi based framework to characterize sector resolved cloud effective radius (CER)-temperature structure from FY-4A geostationary satellite observations and apply it to 162 eastward-propagating convective systems over middle eastern China during 2021–2022. Targeted evaluation against three independent aircraft in situ cases shows that, relative to a homogeneous cluster approach, the proposed framework more consistently reproduces observed CER-temperature profile tendencies, with the mean Trend Consistency index increasing from 0.47 to 0.60. In both case and composite analyses, the retrieved CER-temperature structure evolves from near monotonic growth to a more clearly unimodal pattern as rainfall intensifies. Sectoral contrasts emerge before the defined rainfall intensification time. The frontal sector exhibits the deepest warm rain growth layer, whereas the rear sector contains larger particles and a stronger unimodal structure during mature stages. These results indicate a composite level lead-lag association between the internal spatial organization of CER structure and subsequent rainfall intensification. More broadly, the framework provides an observationally constrained way to examine horizontally heterogeneous CER structure and its relation to precipitation evolution in propagating convective systems. Earth and environmental sciences/Climate sciences Earth and environmental sciences/Hydrology Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Organized convective systems account for a disproportionate share of heavy rainfall, but the microphysical pathways that regulate where and when precipitation intensifies within these systems remain incompletely constrained 1 – 3 . Vertical profiles of cloud effective radius (CER) provide a powerful observational proxy for particle growth 4 – 5 , capturing transitions from condensation and collision-coalescence to mixed-phase conversion and glaciation 6 – 8 . Therefore, satellite-based analysis of CER profiles has become an important tool for diagnosing precipitation formation in deep convective clouds 9 – 13 . Most existing retrieval frameworks, however, infer CER evolution by treating an entire cloud cluster as horizontally homogeneous 14 – 15 . This approximation is unlikely to hold in propagating convective systems, where cloud growth, hydrometeor transport and precipitation development vary systematically along the direction of motion 16 . As a result, conventional approaches are unable to resolve whether microphysical contrasts between the leading, middle and trailing parts of a system emerge only after rainfall intensifies, or whether they develop earlier as part of the microphysical reorganization associated with subsequent rainfall intensification 7 , 12 , 17 – 18 . Eastward-propagating convective systems (EPCSs) over East Asia provide a particularly suitable setting in which to examine this problem. These systems evolve in an environment of sustained low-level moisture transport, frontal baroclinicity and upper-level dynamical support, allowing warm-rain, mixed-phase and ice processes to coexist within deep, rapidly evolving cloud clusters 19 – 22 . This makes them a useful setting for examining whether the internal spatial organization of retrieved CER structure is associated with precipitation intensification. This problem also has direct implications for model development. Cloud microphysics schemes are known to exhibit systematic biases in warm-rain auto-conversion thresholds, mixed-phase partitioning and ice growth rates 23 – 24 . Yet these biases are commonly evaluated using cloud-system means, which may obscure process-level contrasts across dynamically distinct parts of organized convection 25 – 27 . For propagating systems with pronounced front-to-rear gradients, observational constraints on the spatial organization of microphysics are therefore needed both to improve physical understanding and to provide more discriminating benchmarks for model evaluation. Here we develop a Voronoi-based framework to construct sector-resolved CER-temperature profiles from FY-4A Advanced Geostationary Radiation Imager (AGRI) observations. We apply this framework to 162 eastward-propagating convective systems over middle-eastern China during the summers of 2021 and 2022, and evaluate the retrieval against independent aircraft in situ measurements. We do not attempt to reconstruct the full three-dimensional in-cloud microphysical field. Instead, we use retrieved CER-temperature structure as an observational proxy for the vertical evolution of cloud particle size near the cloud top and examine how its spatial organization varies along the propagation axis of eastward-propagating convective systems. The main objectives of this study are to: (1) develop a Voronoi-based framework for describing horizontally heterogeneous CER-temperature structure within organized convective systems; (2) evaluate whether this framework captures observed sector-dependent profile tendencies more consistently than a homogeneous approach; and (3) examine whether systematic sectoral differences in retrieved CER structure are associated with the timing of rainfall intensification at the composite level. Results Validation of the Voronoi-CER approach We first evaluated the Voronoi-based CER retrieval using three independent aircraft in situ cases (Supplementary Fig. S1 and Table 1). Compared with the conventional homogeneous-cluster approach 6 , 9 , the Voronoi framework more consistently reproduced the observed vertical tendency and turning points of CER in all three cases (Fig. 1 a-f). The mean Trend Consistency Index (TC) increased from 0.47 to 0.60, indicating improved agreement in profile tendency relative to the conventional method. Layer-wise error statistics further support this result (Supplementary Fig. S2). In the warm-cloud layer (5 to 0°C), the Voronoi retrieval showed smaller gradient-error dispersion and lower root-mean-square error (RMSE) relative to the aircraft observations than the conventional Region method. In the mixed-phase layer (0 to -20°C), the two methods performed comparably overall, although the Voronoi framework generally exhibited a similar or slightly narrower error spread. In the cold upper layer (< -20°C), the Voronoi retrieval reduced the magnitude of negative bias and showed lower RMSE than the conventional method, indicating improved representation of CER structure under colder cloud-top conditions. These results suggest that the benefit of the Voronoi framework is not limited to noise reduction. Rather, explicitly partitioning horizontally heterogeneous cloud structure before constructing vertical retrieved CER-temperature profiles yields better agreement with independently observed profile tendencies and reduces layer-dependent bias in key parts of the cloud column. Although the TC improvement is modest in absolute terms, it is consistent across all three cases, suggesting a systematic rather than incidental benefit of the Voronoi partitioning approach. Complementing this point-scale evaluation, the statistical robustness of the sector-resolved structures is further examined using system-scale GPM/DPR precipitation diagnostics (Ka and Ku band reflectivity, mass-weighted mean diameter (D m ) and normalized intercept parameter (N w ) profiles) across all EPCSs (see Composite sector-resolved CER structure across EPCSs). Together, these analyses support the interpretation that the Voronoi-based method better captures local vertical CER tendencies while providing a physically interpretable view of broader microphysical organization in organized convection. Evolution of CER structure during convective development We next examined the temporal evolution of a representative eastward-propagating convective system observed over the middle reaches of the Yangtze River on 1 June 2022 (Fig. 2 ). Between 00:00 and 04:00 UTC, the cloud-cluster area increased from 7,528 km² to 18,768 km², accompanied by a progressive deepening and increasing vertical differentiation of the retrieved CER–temperature profile. During the initiation stage (00:00–01:00 UTC), the convective cluster developed rapidly but produced only sporadic surface rainfall (Supplementary Fig. S3a). The retrieved CER-temperature profile extended only to temperatures slightly below 0°C (Fig. 2 f, g), reflecting the limited horizontal extent of the young cloud cluster and the insufficient number of pixels available at colder cloud-top temperatures to construct a statistically robust profile 9 , 28 . At this stage, the profile exhibited no clear sign of vertical differentiation beyond early condensational growth. By 02:00 UTC, during the developing stage, the retrieved CER-temperature profile extended from about 10°C to -30°C and displayed a clear increase-then-decrease structure (Fig. 2 h). CER increased from about 11 µm near 5°C to 38–42 µm near − 15 to -20°C, and then decreased towards cloud top. The widening interquartile range through the mixed-phase layer indicates increasing spatial heterogeneity as collision-coalescence and early ice-phase processes coexist within the system. The reduction of CER above about − 25°C is consistent with the possible sedimentation or redistribution of large particles as precipitation intensified at the surface (Supplementary Fig. S3b, c). At maturity (03:00–04:00 UTC), the retrieved CER–temperature profile underwent a more pronounced structural transition (Fig. 2 i, j). A near-vertical rainout zone emerged in the warm-cloud layer, where median CER stabilized at about 14–17 µm, indicating saturation of droplet growth and sustained precipitation fallout 29 – 30 . Above this layer, CER increased steeply through the mixed-phase region to a maximum near − 15°C and then decreased rapidly towards cloud top. This unimodal structure is consistent with vertically differentiated particle growth, in which warm-rain production, mixed-phase growth and upper-level glaciation become increasingly separated in altitude 31 . The transition from a monotonic to a unimodal profile therefore marks a change in vertical microphysical structure during convective development and occurs as rainfall intensifies at the surface. Spatial heterogeneity of cloud microphysics within an EPCS To determine whether this evolution was spatially organized, we partitioned the cloud cluster into frontal (see Methods), middle and rear sectors using Voronoi sub-regions projected onto the principal axis of propagation (Supplementary Figs. S4). During initiation, retrieved CER-temperature profiles in the three sectors were broadly similar (Fig. 3 b, d). All sectors showed quasi-linear growth from cloud base to around − 30°C, with no pronounced frontal-to-rear gradient in profile shape or magnitude. This indicates that during the earliest stage of development, particle growth remained comparatively homogeneous across the cloud cluster and was dominated by condensation and early collision-coalescence. By the developing stage, however, clear sector-dependent contrasts had emerged (Fig. 3 f). CER increased from the frontal to the rear sector in the warm-cloud and lower mixed-phase layers. Frontal CER values ranged from about 10 to 26 µm, middle-sector values from 12 to 34 µm, and rear-sector values from 12 to 44 µm. This front-to-rear increase is consistent with a propagating system in which younger convective elements are concentrated near the leading edge, while larger hydrometeors accumulate in the trailing region through sedimentation and mesoscale transport 32 – 33 . At the same time, the middle-sector profile extended to lower temperatures than either the frontal or rear-sector profiles, suggesting that the deepest convective development occurred in the system core. At maturity, all three sectors extended below − 40°C, but their vertical structures diverged further (Fig. 3 h, j). The frontal and middle sectors exhibited rapid quasi-linear CER growth from 10°C to about − 15°C, followed by a marked reduction in growth rate at colder levels. By contrast, the rear sector maintained larger CER values below − 20°C and exhibited the most pronounced unimodal profile, with a peak near − 25 to -30°C before decreasing towards cloud top. This behavior is consistent with enhanced accumulation and redistribution of large ice-phase hydrometeors in the rear portion of the system 33 – 35 . The observed sectoral contrast therefore cannot be explained as purely local variability; rather, it indicates a dynamically organized spatial structure of cloud microphysics along the propagation axis. Composite sector-resolved CER structure across EPCSs To assess whether the sector-dependent CER organization identified in the case study is a recurring feature, we composited the retrieved CER–temperature structures from all selected events and compared them with collocated GPM/DPR precipitation diagnostics. These pooled composites are intended to describe common structural tendencies across the sample rather than to imply pixel-level independence or deterministic behavior in individual events. All three sectors exhibited a similar background CER structure, with quasi-linear growth from approximately 10–12 µm to 20–22 µm through the warm-cloud and lower mixed-phase layers (Fig. 4 a), consistent with the canonical rainout regime reported in satellite-based microphysical profiling studies 8 , 28 . Superimposed on this common structure, however, the depth of the warm-rain growth layer differed systematically among sectors. The temperature interval between the onset of precipitation-sized droplets (T 14 ) and the termination of quasi-linear growth (T L ) was approximately 15°C in the frontal sector, compared with about 13°C in the middle and rear sectors (Fig. 4 a). These results suggest that the leading part of EPCSs sustains a deeper layer of efficient droplet growth before CER departs from quasi-linear behavior, indicating a systematically organized microphysical structure along the propagation axis 36 . The corresponding vertical gradient profiles further highlight differences in particle evolution among sectors (Fig. 4 b-d). In the warm-cloud layer, the frontal sector exhibited weaker positive gradients than the middle and rear sectors, indicating a more gradual increase in CER with height. In the mixed-phase layer, the middle and rear sectors showed stronger negative gradients, implying more rapid particle-size adjustment as clouds extended into colder levels 31 . Above this layer, gradient magnitudes weakened in all sectors, but the rear sector maintained comparatively large CER values below about − 20°C, consistent with a greater persistence of larger hydrometeors in the trailing part of the system. To further quantify the temporal evolution of this microphysical organization, we introduced two CER-based indices: CER Peak , defined as the maximum CER within the retrieved CER-temperature profile, and \(\:{T}_{peak}\) , defined as the temperature at which the maximum CER occurs. CER Peak characterizes the amplitude of particle growth, whereas \(\:{T}_{peak}\) indicates the vertical position of peak particle growth within the cloud. The temporal composites show that both indices begin to evolve before the rainfall intensification time (Fig. 4 e, f). In all three sectors, CER Peak increased prior to the onset of intensified surface rainfall, with the composite increase beginning about 75 min before \(\:\tau\:=0\) , indicating that the production of larger particles starts before the rainfall maximum is reached (Fig. 4 e). This earlier increase is most pronounced in the frontal and middle sectors, consistent with the sector-dependent enhancement of warm rain growth inferred from the vertical retrieved CER-temperature profiles. Meanwhile, \(\:{T}_{peak}\) shifted toward colder temperatures around the intensification time (Fig. 4 f), indicating that the vertical level of maximum particle growth was reorganized as the system developed. Notably, the frontal sector exhibited a weaker cooling tendency in \(\:{T}_{peak}\) than the middle and rear sectors. This more gradual downward evolution is consistent with the deeper rainout growth layer identified in the frontal-sector CER profiles, suggesting that efficient warm-rain growth is sustained over a greater temperature depth before the peak-growth level shifts upward into colder layers. Taken together, the composite statistics indicate that the particle-size reorganization observed in the case study is not an isolated feature. Instead, the selected 162 EPCSs exhibit a consistent composite signal of microphysical reorganization on timescales leading surface-rainfall intensification. These results support a composite-level lead-lag association between changes in CER structure and subsequent precipitation enhancement, but do not by themselves establish predictive skill or a deterministic causal relationship. They occur within a broader large-scale environment favorable for organized propagating convection, including upper-level divergence, mid-level ascent, and strong low-level moisture supply (Supplementary Fig. S5), although those composites provide background context rather than sector-resolved process attribution. These optical retrievals are broadly consistent with the collocated radar-based precipitation structure. The GPM/DPR contoured frequency-by-altitude diagrams showed a vertically continuous precipitation column, with moderate raindrop diameters extending from the surface to about 6 km altitude, close to the − 5°C isotherm and to the upper limit of the rainout layer inferred from the retrieved CER-temperature profiles (Fig. 5 c). In addition, neither the Ku nor Ka-band reflectivity composites showed a pronounced bright-band signature near the melting level (Fig. 5 a, b), indicating that the sampled systems were still dominated by warm-rain and early mixed-phase processes rather than by a fully developed stratiform melting layer. Across the 162 EPCSs, the frontal, middle, and rear sectors therefore form a coherent propagation-relative sequence in retrieved CER structure and precipitation organization. Discussion The results suggest that the internal microphysics of EPCSs becomes increasingly spatially organized as rainfall intensification develops. First, retrieved CER–temperature profiles tend to evolve from monotonic growth to a unimodal structure during convective development. Second, distinct frontal, middle, and rear-sector contrasts emerge before the systems reach full maturity (Fig. 6 a). Together, these findings indicate increasing spatial differentiation of particle growth across the propagation axis as propagating convection evolves. From an observational perspective, the transition from a monotonic to a unimodal retrieved CER-temperature profile can therefore be interpreted as a signature of increasingly vertically differentiated microphysical structure in organized convection 6 , 28 – 30 . An important result is that this organization becomes evident before intense rainfall is fully established at the surface. The temporal evolution of CER Peak and \(\:{T}_{peak}\) shows that both the amplitude of particle growth and the vertical position of peak particle growth begin to adjust systematically before the defined rainfall-intensification time. This behavior is consistent with a stage of microphysical reorganization before precipitation enhancement, in which larger particles develop and the level of maximum particle growth shifts within the cloud column. In particular, the CER Peak increases before rainfall intensification, while \(\:{T}_{peak}\) shifts toward colder temperatures, indicating that the vertical location of peak particle growth becomes reorganized as the system develops. The frontal sector exhibits a weaker cooling tendency in \(\:{T}_{peak}\) than the middle and rear sectors, which is consistent with the deeper rainout growth layer identified in the frontal-sector CER profiles. The agreement between these CER-based indices and the sector-resolved profile evolution further suggests that intracloud microphysical structure is statistically associated with the transition toward stronger rainfall. However, the present analysis is based on composite lead-lag behavior and should not be interpreted as demonstrating event-by-event predictive skill or a deterministic precursor relationship. The mature-stage sectoral pattern is broadly consistent with the canonical leading-convective/trailing-stratiform framework of mesoscale convective systems 32 , 37 , which supports the physical interpretability of the Voronoi-CER framework. At the same time, the frontal sector exhibits the deepest warm rain growth layer, whereas the rear sector contains the largest particles and the strongest unimodal signatures at maturity. This pattern suggests that warm-rain efficiency may depend not only on convective maturity, but also on the thermodynamic and microphysical characteristics of the inflow feeding the leading edge 32 . One physically plausible interpretation is that sustained low-level inflow into the frontal sector may help maintain a deeper layer of efficient warm rain growth 6 , 12 , while the rear sector may become increasingly influenced by hydrometeor loading, sedimentation, and rear-inflow dynamics 38 – 40 (Fig. 6 b, c). However, the environmental analysis presented here is composited at system scale rather than sector resolved (Supplementary Fig. S5), so these process-level interpretations should be regarded as physically motivated hypotheses rather than demonstrated mechanisms. Direct sector-resolved dynamical evidence would be required to test them more rigorously. This study also highlights a broader methodological point. Conventional CER profiling methods assume horizontal homogeneity at the cloud-cluster scale and therefore emphasize the mean state of the system. The Voronoi framework relaxes this assumption and instead treats cloud organization as part of the retrieval problem, making it better suited to propagating systems in which microphysical evolution is coupled to advection, differential cloud stage, and mesoscale circulation 32 . Several limitations should nevertheless be noted. First, validation of the retrieval is based on only three independent aircraft cases. Although these cases do not sample the full diversity of cloud regimes represented in the 162-event composite, they provide targeted evidence that the Voronoi-based method better captures the tendency of CER evolution than the conventional homogeneous-cluster approach. Second, the current environmental analysis is not sector resolved and therefore cannot directly attribute the frontal-to-rear microphysical contrasts to specific dynamical or thermodynamic mechanisms; the process interpretations discussed above should accordingly be viewed as physically motivated hypotheses. Third, the final 162-event sample represents a filtered subset of summer daytime eastward-propagating systems over middle-eastern China, selected further by the availability of collocated GPM/DPR observations and a lifetime threshold of 6 h. The results should therefore be interpreted within the scope of this observational subset rather than as universally representative of all propagating convective systems. Fourth, the composite analysis is based on pooled pixels and is intended to describe recurring structural tendencies across the sample rather than pixel-level independence or deterministic behavior in individual events. Despite these limitations, the main conclusions remain robust within the scope of the present evidence. EPCSs exhibit a spatially organized microphysical structure prior to rainfall intensification at the composite level, and this organization can be detected in sector-resolved retrieved CER-temperature profiles. These findings provide an observational constraint on precipitation formation in organized convection and offer a new benchmark for evaluating how models represent the early microphysical evolution of convective systems. Methods Observational datasets The satellite data used in this study consist of FY-4A/AGRI full-disk observations during June-August of 2021–2022 over middle-eastern China (105–120 ºE, 24–39 ºN). FY-4A provides geostationary observations of the Earth-atmosphere system at 15 min temporal resolution. Cloud effective radius (CER) fields were derived following the algorithm of Sun et al. 41 , which reported a correlation of 0.91 against the MODIS Collection 6.1 CER product. For consistency in object identification and sector-resolved analysis, all satellite fields were mapped to a common analysis grid prior to processing. Quality-controlled precipitation observations from automatic weather stations, available every 5 min, were interpolated onto a 0.04 º grid using a radial basis function method. For each EPCS time step, the precipitation field nearest in time to the corresponding FY-4A/AGRI scan was used, with a maximum temporal mismatch of 5 min. ERA5 hourly reanalysis fields 42 , including horizontal wind components, vertical velocity, specific humidity, relative humidity, and column-integrated moisture flux, were extracted at 0.25 º resolution and composited over the 162 events in a common system-relative framework to characterize the large-scale dynamical and thermodynamic background of the analyzed EPCS sample. GPM/DPR Ku- and Ka-band reflectivity profiles and retrieved microphysical parameters were considered collocated when they intersected the EPCS boundary within 30 min of the corresponding AGRI observation, and were then composited into propagation-relative CFADs. Three independent aircraft cloud-penetration missions conducted by the KA350 research aircraft were used to evaluate the FY-4A/AGRI CER retrieval framework (Supplementary Fig. S1 ). The aircraft primarily employed a combination of horizontal level legs and stepwise vertical penetrations. After takeoff, the aircraft ascended to cloud top or to a maximum altitude of approximately 7300 m and circled over the target area. It then descended in steps of about 300 m, maintaining stable horizontal orbits at each altitude level to collect cloud-particle and precipitation-particle measurements. The minimum flight altitude was approximately 1500 m, after which the aircraft ascended again to complete a closed vertical profile. This stepwise sampling strategy provided vertically resolved in situ measurements of cloud droplet spectra and precipitation particle spectra 43 . For aircraft–satellite evaluation, the in-situ observations were collocated with the nearest FY-4A/AGRI CER retrieval in both time and space. Specifically, aircraft samples were matched to the closest AGRI scan within a temporal window of ± 7.5 min and within a 4 km spatial window centered on the aircraft position, consistent with the FY-4A resolution used in this study. Because aircraft observations sample along a moving flight path at much finer scales than satellite retrievals, the in-situ measurements within each collocation window were aggregated to the AGRI sampling scale before comparison. Aircraft CER values were then grouped by ambient temperature and compared with the satellite-derived retrieved CER-temperature profile using the same temperature-bin framework as in the satellite analysis. Given the inherent scale mismatch between aircraft in situ sampling and satellite pixel-scale retrievals, this comparison is intended as a targeted evaluation of vertical CER structure rather than a point-to-point validation of instantaneous pixel values. Starting from all mobile convective cloud clusters identified over China during June-August of 2021–2022, we focused on eastward-propagating systems because they represent the dominant propagation class in the study region and provide a more dynamically homogeneous population for process-oriented composite analysis (Supplementary Fig. S6). A total of 9481 mobile convective cloud clusters were identified during the study period, of which eastward-propagating systems accounted for 33.1% and westward-propagating systems for 17.0%. Because FY-4A/AGRI CER retrievals rely on daytime solar-reflectance information, only daytime eastward-propagating events were retained. To further ensure sufficient lifecycle sampling and an independent constraint on precipitation vertical structure, we additionally required collocated GPM/DPR overpasses and a system lifetime longer than 6 h. These filtering steps yielded the final sample of 162 EPCSs used in this study. This dataset should therefore be regarded as a restricted observational subset rather than as the full population of EPCSs over China, and the conclusions are intended to apply to this selected sample of summer daytime eastward-propagating systems over middle-eastern China. Identification and tracking of convective cloud clusters Convective cloud clusters are identified and tracked using the FLEXTRKR algorithm 44 , with criteria following Chen et al. 45 . Convective cores were defined by 10.8 µm brightness temperatures below 263 K, and cluster boundaries by contiguous regions below 280 K. A minimum cluster area of 256 km 2 was required, together with a contiguous precipitation region having intensity ≥ 0.5 mm h − 1 and a longest axis ≥ 8 km. Propagating convective clusters were identified from centroid displacement, with eastward-propagating systems defined by movement azimuths between 22.5º and 157.5º. Construction of sector-resolved Voronoi-retrieved CER–temperature profiles The Voronoi-CER method constructs sector-resolved CER-temperature profiles through the following steps. First, the FLEXTRKR-derived cloud-cluster boundary defines the analysis domain. Second, K-means clustering is applied to CER fields within the domain to identify coherent subregions. Rather than prescribing a fixed number of subregions, the optimal cluster number (K) is adaptively determined for each cloud system using a composite index combining the Silhouette, Calinski-Harabasz, and Davies-Bouldin scores (Supplementary Fig. S7). Third, the centroids of the resulting clusters are used as seed points to generate Voronoi tessellations, partitioning the cloud system into K polygonal subregions. Fourth, the principal axis of the cloud cluster, derived from principal component analysis and aligned with the propagation direction, is used to group these subregions into front, middle, and rear sectors of equal length along the propagation axis (Supplementary Fig. S4). Finally, CER values within each sector are composited as a function of cloud-top brightness temperature using 1.75 K temperature bins. The 25th, 50th, and 75th percentiles are then calculated, with the median defining the retrieved CER-temperature profile and the interquartile range representing sectoral variability. Trend Consistency Index To evaluate the reliability of the Voronoi-CER method, three aircraft in-situ cases are used to compare satellite-retrieved and aircraft-observed CER–temperature profiles. A trend consistency index (TC) is defined as: $$\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:\:TC=\frac{{N}_{consistent}}{{N}_{total}}$$ 1 where N total is the number of adjacent-layer comparisons, N consistent is the number of cases in which the direction of CER change (increase or decrease) is consistent between satellite and aircraft observations. A value of TC = 1 indicates perfect agreement, while TC = 0.5 corresponds to random agreement. Definition of rainfall intensification Rainfall intensification is defined based on the cloud averaged precipitation within the cloud boundary at each time step. The mean 5 min accumulated precipitation is the first converted to rainfall rate \(\:R\left(t\right)\) . The fractional coverage of effective precipitation is defined as: $$\:\:\:{A}_{0.05}\left(t\right)=\frac{N\left(P\ge\:0.05\right)}{{N}_{total}}$$ 2 where \(\:N\left(P\ge\:0.05\right)\) is the number of grid points within the cloud with precipitation no less than 0.05 mm per 5min, and \(\:{N}_{total}\) is the total number of grid points within the cloud. The rainfall intensification time \(\:{t}_{int}\) is identified as the first time when the following conditions are satisfied for at least two consecutive time steps: $$\:{A}_{0.05}\left(t\right)\ge\:0.1$$ 3 $$\:R\left(t\right)\ge\:0.5$$ 4 The relative time is then defined as $$\:\tau\:=t-{t}_{int}$$ 5 CER Index To characterize the intensity of cloud microphysical development, the peak value of CER is used. In this study, CER Peak and CER topbase are used as descriptive indices of microphysical evolution relative to the rainfall-intensification time. They are not intended here as event-level predictive metrics, and no formal forecast-skill evaluation is performed. $$\:{\:\:\:\:\:\:\:\:CER}_{\_Peak}=\text{m}\text{a}\text{x}\left(CER\right)$$ 6 To further summarize the vertical structure of cloud microphysical properties, a second CER-based index is introduced. CER topbase is defined as the difference between the maximum and minimum CER values along the cloud vertical profile and is used here as a simple descriptive measure of vertical CER contrast. Because this index depends on profile extrema, it can be sensitive to local variability; accordingly, it is interpreted together with the full sector-resolved retrieved CER-temperature profile composites rather than as a standalone robust metric. $$\:{\:\:\:\:\:\:\:\:CER}_{topbase}={\text{C}\text{E}\text{R}}_{max}-{CER}_{min}$$ 7 where CER_max and CER_min represent the maximum and minimum CER values along the cloud vertical profile, respectively. Declarations Competing interests The authors have no conflicts of interest to disclose. Author Contribution J.S. and Y.Y.L. conceived the study. J.S., X.H., and Z.T.C. processed and organized the data. J.S. wrote the original draft. J.S. and Y.Y.L. developed the methodology, designed the algorithms, and revised the manuscript. X.H., Z.T.C., X.L., and F.L. contributed to the interpretation of the results and to manuscript revision. Acknowledgments This work was supported by the National Natural Science Foundation of China(U2242201), the Hubei Provincial Natural Science Foundation of China(2025AFD423), and the Open Fund Project for Heavy Rain (BYKJ2024M07). Data Availability FY-4A/AGRI Level-1 radiance data (2021-2022) were obtained from the Fengyun Satellite Data Center (https://satellite.nsmc.org.cn/DataPortal/cn/data/order.html). Cloud effective radius (CER) retrievals were derived using the algorithm described in Sun et al. (2025). The PyFLEXTRKR algorithm can be obtained from https://github.com/FlexTRKR/PyFLEXTRKR. The GPM 2ADPR precipitation data used in this study were collected from the Precipitation Measurement Mission website (https://pmm.nasa.gov). ERA5 reanalysis data (hourly, 0.25º resolution) were obtained from the Copernicus Climate Data Store (https://cds.climate.copernicus.eu/) to characterize the large-scale dynamical and thermodynamic environment of the EPCSs. The custom code used in this study for Voronoi partitioning, adaptive selection of the optimal cluster number K, is available from the corresponding author during peer review and will be deposited in a public repository upon publication. References Schumacher, C. & Houze, R. A. Jr. Stratiform rain in the tropics as seen by the TRMM precipitation radar. J. Clim. 16, 1739–1756 (2003). Kukulies, J., et al. Mesoscale convective systems in the third pole region: Characteristics, mechanisms and impact on precipitation. Front. Earth Sci. 11, 1143380 (2023). Zhao, Y., et al. Distinct structure, radiative effects, and precipitation characteristics of deep convection systems in the Tibetan Plateau compared to the tropical Indian Ocean, Atmos. Chem. Phys. 24, 9435–9457 (2024). Shang, H., et al. Remote sensing of liquid cloud profiles based on an analytical cloud profiling model. Sci. China Earth Sci. 68, 998–1012 (2025). Wang, Y. T., et al. Characterization of liquid cloud profiles using global collocated active radar and passive polarimetric cloud measurements. Atmos. Chem. Phys., 25, 16167–16187 (2025). Rosenfeld, D., & Lensky, I. M. Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc. 79, 2457–2476 (1998). Kokhanovsky, A.A., Painemal, D., & Rozanov, V. V. The Intercomparison of Satellite-Derived and In Situ Profiles of Droplet Effective Radii in Marine Stratocumulus Clouds. IEEE Geosci. Remote Sens. Lett. 10(5), 1147–1151 (2012). Rosenfeld, D. Cloud-aerosol-precipitation interactions based of satellite retrieved vertical profiles of cloud microstructure. In Remote sensing of aerosols, clouds, and precipitation. 6, 129–152 (2018). Chen, Y. L., et al. Retrieval of the vertical evolution of the cloud effective radius from the Chinese FY-4 (Feng Yun 4) next-generation geostationary satellites. Atmos. Chem. Phys. 20, 1131–1145 (2020a). Chen, Y. L., et al. Linkage between the vertical evolution of clouds and droplet growth modes as seen from FY-4A AGRI and GPM DPR. Geophys. Res. Lett. 47, e2020GL088312 (2020b). Efraim, A., et al. Satellite-based detection of secondary droplet activation in convective clouds. J. Geophys. Res. Atmos. 127, e2022JD036519 (2022). Freud, E., & Rosenfeld, D. Linear relation between convective cloud drop number concentration and depth for rain initiation. J. Geophys. Res. 117, D02207 (2012). Rosenfeld, D., et al. High-resolution (375 m) cloud microstructure as seen from the NPP/VIIRS satellite imager. Atmos. Chem. Phys. 14, 2479–2496 (2014). Letu, H., et al. High-resolution retrieval of cloud microphysical properties and surface solar radiation using Himawari-8/AHI next-generation geostationary satellite. Remote Sens. Environ. 239, 111583 (2020). Liu, C., et al. A cloud optical and microphysical property product for the advanced geosynchronous radiation imager onboard China’s Fengyun-4 satellites: The first version. Atmos. Ocean. Sci. Lett. 16, 100337–100343 (2023). Song, F. F., et al. Crucial roles of eastward propagating environments in the summer MCS initiation over the U.S. Great Plains. J. Geophys. Res. Atmos. 126, e2021JD034991(2021). Sokol, A. B., & Storelvmo, T. The spatial heterogeneity of cloud phase observed by satellite. J. Geophys. Res. Atmos. 129, e2023JD039751 (2024). Huang, Y., & Dewitt, B. Differentiating Thick Clouds from Thin Clouds by Using Intensity Inhomogeneity. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 18, 5782–5792 (2025). Guo, Y., et al. Trends in warm season mesoscale convective systems over Asia in 2001–2020. J. Geophys. Res. Atmos. 128, e2023JD038969 (2023). Yang, R., et al. The comparison of statistical features and synoptic circulations between the eastward-propagating and quasi-stationary MCSs during the warm season around the second-step terrain along the middle reaches of the Yangtze River. Sci. China Earth Sci, 63, 1209–1222(2020). Li, P., et al. Sensitivity of simulated mesoscale convective systems over East Asia to the treatment of convection in a high-resolution GCM. Clim. Dyn. 60, 2783–2801 (2023). Fu, Y., et al. Initiations of mesoscale convective systems in the middle reaches of the Yangtze River Basin based on FY-4A satellite data: Statistical characteristics and environmental conditions. J. Geophys. Res. Atmos. 128, e2023JD038630 (2020). Song, X., & Zhang, G. J. Microphysics parameterization for convective clouds in a global climate model: Description and single-column model tests, J. Geophys. Res. 116, D02201, (2011). Wang, J.Y., et al. Impact of a new cloud microphysics parameterization on the simulations of mesoscale convective systems in E3SM. J. Adv. Model. Earth Syst. 13, e2021MS002628 (2021). Li, S. S., et al. Statistical characteristics and synoptic patterns of convection initiation over the middle reaches of the Yangtze River Basin as observed using the Fengyun-4A satellite. J. Hydrometeorol. 25(3), 445–463 (2024). Wei, Q., et al. Convection initiation over middle–eastern China: Statistics show higher frequencies over mountains with less stringent atmospheric conditions and more heterogeneous surface characteristics. Atmos. Res. 328, 108438–108454 (2026). Kumar, S., & Srivastava, S. Vertical characteristics of precipitating cloud systems during different phases of life cycle of cloud systems using satellite-based radar over tropical oceanic areas. J. Appl. Nat. Sci. 14(4), 1272–1285 (2022). Rosenfeld, D., et al. Satellite detection of severe convective storms by their retrieved vertical profiles of cloud particle effective radius and thermodynamic phase. J. Geophys. Res. Atmos. 113, D04208 (2008). Braga, R. C., et al. Linear relationship between effective radius and precipitation water content near the top of convective clouds: measurement results from ACRIDICON–CHUVA campaign, Atmos. Chem. Phys., 21, 14079–14088 (2021). Rosenfeld, D., et al., Aerosol-driven droplet concentrations dominate coverage and water of oceanic low level clouds, Science, 363, 6427 (2019). Suzuki, K., Nagao, T. M., & Murai, A. Satellite-based diagnostics of precipitation process in mixed‐phase clouds: Extension from warm rain process statistics. Geophys. Res. Lett. 51, e2024GL110573 (2024). Houze, R. A. Jr. Mesoscale convective systems. Rev. Geophys., 42, RG4003 (2004). Short, E, Lane, T., & Vincent, C., Objectively Diagnosing Characteristics of Mesoscale Organization from Radar Reflectivity and Ambient Winds. Mon. Weather Rev. 151(3) (2023). Han, C., et al. Sensitivity of cloud-phase distribution to cloud microphysics and thermodynamics in simulated deep convective clouds and SEVIRI retrievals, Atmos. Chem. Phys., 23, 14077–14095 (2023). Korolev, A., et al. High ice water content in tropical mesoscale convective systems (a conceptual model), Atmos. Chem. Phys., 24, 11849–11881 (2024). Wang, Z, & Yuan, J. Observing convective activities in complex convective organizations and their contributions to precipitation and anvil cloud amounts. Atmos. Chem. Phys. 24(23):13811–13831 (2024). Skow, A., et al. A Multi-Platform, In-Situ Kinematic and Microphysical Analysis of a Hybrid Parallel/Trailing Stratiform Mesoscale Convective System. Mon. Weather Rev. 150(4):927–948 (2022). Weisman, M. The Role of Convectively Generated Rear-Inflow Jets in the Evolution of Long-Lived Mesoconvective Systems. J. Atmos. Sci. 49: 1826–1847 (1992). Nizar, S., Thomas, J., & Jainet P.J, K. P. S. A Novel Technique for Nowcasting Extreme Rainfall Events using Early Microphysical Signatures of Cloud Development. PLOS Clim 4(5), e0000497 (2025). Smull, B. F., & Houze, R. A., Jr. Rear inflow in squall lines with trailing stratiform precipitation. Mon, Wea, Rev. 115(12), 2869–2889. (1987). Sun, J., et al. Impact of Cloud Vertical Structure Perturbations on the Retrieval of Cloud Optical Thickness and Effective Radius from FY4A/AGRI. Atmos. Chem. Phys. 25(22),16347–16361 (2025). Hersbach, H., et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc., 146(730), 1999–2049 (2020). Cui, C., et al. Phase Two of the Integrative Monsoon Frontal Rainfall Experiment (IMFRE-II) over the middle and lower reaches of the Yangtze River in 2020. Adv. Atmos. Sci.,38,346–356 (2021). Feng, Z., et al. PyFLEXTRKR: a flexible feature tracking Python software for convective cloud analysis. Geosci. Model Dev. 16(10), 2753–2776 (2023). Chen, Z., et al. Summer surface rainfall deviations from convective cold cloud shields over China. Geophys. Res. Lett, 52, e2025GL117889. (2025). Additional Declarations No competing interests reported. Supplementary Files Supplementarymaterial.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-9477802","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":634012576,"identity":"c9e42c62-5ae5-44d7-ae4f-07c085540716","order_by":0,"name":"Jing Sun","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Sun","suffix":""},{"id":634012583,"identity":"a7e371dc-d830-4d5c-b8f5-46a75a120a69","order_by":1,"name":"Yunying Li","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA40lEQVRIiWNgGAWjYBACPmYwJcHAwN7AwMADYh8goIUNroUHqJQngRgtcJZEArFa2HnMpAvbLPLkI98ek3j7g0GO70YC4+cCvA4DapnZJlFseDsvTXJOAoOx5I0EZukZhLTwtkkkbpydYyYNdFjihhsJQEGitMw8A9ZST7yW+RI8YC0JBoS1sBVb85yTSNzAk2NsOSdNwnDmmYfN0vi08PMf3nibp6wucX77GcMbb2xs5PmOJx/8jE8LAwOHAQMjMHYMDoB5wDhlYGzAqwGYUB4wMPxhYJAnpG4UjIJRMApGLgAAzXA+uJKXMD4AAAAASUVORK5CYII=","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":true,"prefix":"","firstName":"Yunying","middleName":"","lastName":"Li","suffix":""},{"id":634012595,"identity":"5c196d11-1cd4-4d9a-9f0a-c33194dbf2f8","order_by":2,"name":"Xiong Hu","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiong","middleName":"","lastName":"Hu","suffix":""},{"id":634012596,"identity":"542e13f3-37f9-49b5-8301-e71c1aaa5b7d","order_by":3,"name":"Zitong Chen","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Zitong","middleName":"","lastName":"Chen","suffix":""},{"id":634012604,"identity":"6332571b-9dc5-42b4-bb17-1e2ec9921fdf","order_by":4,"name":"Xiang Lin","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Xiang","middleName":"","lastName":"Lin","suffix":""},{"id":634012605,"identity":"0cfba6c2-aef5-466d-bdcc-5aba7a00442b","order_by":5,"name":"Fan Li","email":"","orcid":"","institution":"National University of Defense Technology","correspondingAuthor":false,"prefix":"","firstName":"Fan","middleName":"","lastName":"Li","suffix":""}],"badges":[],"createdAt":"2026-04-21 02:53:34","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-9477802/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-9477802/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":108805065,"identity":"91a29445-f60f-4fe5-b1a9-2b289fccf4b9","added_by":"auto","created_at":"2026-05-08 15:24:39","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":532507,"visible":true,"origin":"","legend":"\u003cp\u003eValidation of the Voronoi-CER retrieval against aircraft in situ observations. Panels show CER vertical-profile tendencies derived from sliding-window linear regression, expressed as the slope of the CER–temperature relationship. Solid lines indicate median slopes and shaded areas denote 95% confidence intervals. Positive slope indicates CER increases with temperature, whereas negative slope indicates CER increases with decreasing temperature. Grey dashed lines mark the zero-slope reference and the 0 °C isotherm. Panels (a–c) show the Voronoi-based retrieval and panels (d–f) the conventional homogeneous-cluster method. Trend Consistency Index (TC) values are shown for each case.\u003c/p\u003e","description":"","filename":"floatimage1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/bc1dde01aadee9176a252784.jpeg"},{"id":108632649,"identity":"581d88eb-1ef8-4d1e-aafa-44363330a1e7","added_by":"auto","created_at":"2026-05-06 17:06:26","extension":"jpeg","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":1844531,"visible":true,"origin":"","legend":"\u003cp\u003eTemporal evolution of cloud microphysical structure in a propagating convective system on 1 June 2022, derived from the Voronoi-CER method. (a–e) Spatial distribution of 10.8 μm brightness temperature (K) at five successive times (00:00–04:00 UTC). Black contours delineate cloud cluster boundaries (brightness temperature \u0026lt; 280 K). (f–j) Corresponding Voronoi-CER vertical profiles representing the microphysical state of the entire cloud cluster at each time step. Red dots indicate median CER; blue vertical bars represent the IQR. The grey dashed line marks the 0 °C isotherm.\u003c/p\u003e","description":"","filename":"floatimage2.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/c6d58272d97362e5aad787e4.jpeg"},{"id":108632647,"identity":"61062e0b-1174-47da-98b1-152f85a71722","added_by":"auto","created_at":"2026-05-06 17:06:26","extension":"jpeg","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":1744803,"visible":true,"origin":"","legend":"\u003cp\u003eSpatial segmentation and sector-resolved retrieved CER–temperature profiles of a propagating convective system from 00:00 to 04:00 UTC on 1 June 2022. Left panels (a, c, e, g, i) show Voronoi-based segmentation of the cloud cluster into K sub-regions, where K is determined by a composite clustering metric (see Fig. S6); colors indicate median CER. Right panels (b, d, f, h, j) show vertical retrieved CER–temperature profiles for the front (blue), middle (yellow), and rear (red) sectors, defined by dividing the system into equal thirds along the principal propagation axis. Red dots denote median CER, and shaded envelopes represent the interquartile range. The grey dashed line indicates the 0 °C isotherm.\u003c/p\u003e","description":"","filename":"floatimage3.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/e9f6af1068978b0825cf562d.jpeg"},{"id":108632653,"identity":"7a0713b1-1552-489a-a6e6-4bfab98b71b6","added_by":"auto","created_at":"2026-05-06 17:06:26","extension":"jpeg","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":1380465,"visible":true,"origin":"","legend":"\u003cp\u003eStatistical organization of sector-resolved CER structure across 162 EPCSs. (a) Composite CER vertical profiles for the front (blue), middle (yellow), and rear (red) sectors. Solid lines denote median CER; shading indicates the interquartile range. Horizontal dashed lines mark T₁₄ (onset of precipitation-sized droplets at CER = 14 μm), T\u003csub\u003eL\u003c/sub\u003e (termination of quasi-linear growth), and T_g (onset of glaciation), following Rosenfeld and Lensky⁸. (b–d) Vertical CER temperature gradients (dCER/dT; μm °C⁻¹) for the front, middle, and rear sectors; negative values indicate CER increasing with decreasing temperature. (e–f) Temporal evolution of CER\u003csub\u003ePeak\u003c/sub\u003e and relative to rainfall intensification (; light grey dashed line). CER\u003csub\u003ePeak\u003c/sub\u003e is the maximum CER within the vertical profile, and is the temperature at which the maximum CER occurs. Solid lines denote sector means. In panel (e), darker dashed lines mark the composite onset time of systematic CER increase in each sector. In panels (e–f), dotted lines of the same color indicate the sampling uncertainty range.\u003c/p\u003e","description":"","filename":"floatimage4.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/8827076d3be8574b5d72eabe.jpeg"},{"id":108632650,"identity":"42555a64-17a4-4b2a-a6d4-a19ad051643e","added_by":"auto","created_at":"2026-05-06 17:06:26","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":694916,"visible":true,"origin":"","legend":"\u003cp\u003eGPM/DPR CFAD composited over 162 propagating convective events collocated with the Voronoi-CER dataset (within ±30 min and the cloud cluster boundary). (a) Ka-band reflectivity (dBZ) as a function of altitude. (b) Ku-band reflectivity (dBZ) as a function of altitude. (c) D\u003csub\u003em\u003c/sub\u003e (mm) as a function of altitude. (d) N\u003csub\u003ew\u003c/sub\u003e (m⁻³ mm⁻¹) as a function of altitude. Colour shading indicates probability density. The absence of a pronounced bright-band signature near the melting layer in panels (a-b), together with the vertically continuous D\u003csub\u003em\u003c/sub\u003e and N\u003csub\u003ew\u003c/sub\u003e profiles in panels (c-d), is broadly consistent with precipitation growth dominated by warm-rain and early mixed-phase processes rather than by a fully developed stratiform melting layer.\u003c/p\u003e","description":"","filename":"floatimage5.png","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/a1dcbc9877edfdf8c94835cc.png"},{"id":108632651,"identity":"9dbd3c9f-b668-4582-beb1-c9d5a52e32d7","added_by":"auto","created_at":"2026-05-06 17:06:26","extension":"jpeg","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":1980534,"visible":true,"origin":"","legend":"\u003cp\u003eConceptual model of the spatial organization of cloud microphysical structure in propagating convective systems (EPCSs). (a) Schematic CER vertical profiles for the front (blue), middle (yellow), and rear (red) sectors, with four characteristic zones following Rosenfeld and Lensky\u003csup\u003e6\u003c/sup\u003e: C (condensation and coalescence), R (rainout), M (mixed-phase growth), and G (glaciation). (b) Early-stage schematic showing developing updrafts with limited inter-sector microphysical differentiation. (c) Mature stage schematic illustrating a spatially organized microphysical structure, with warm-rain growth in the frontal sector, deep particle growth in the middle sector, and larger hydrometeors in the rear sector. Arrows indicate schematic airflow patterns, including low-level convergence and rear inflow.\u003c/p\u003e","description":"","filename":"floatimage6.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/6fc14543f286d60d6533b4dd.jpeg"},{"id":109422149,"identity":"6adb156b-4ed6-4832-8810-a15f647aa22f","added_by":"auto","created_at":"2026-05-18 00:39:37","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8731084,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/3d0059cd-6712-4b1f-b645-1ba7c03d0387.pdf"},{"id":108632646,"identity":"2e9067ac-ba78-44e8-9018-0c81d2f718e9","added_by":"auto","created_at":"2026-05-06 17:06:25","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"supplement","size":4511913,"visible":true,"origin":"","legend":"","description":"","filename":"Supplementarymaterial.docx","url":"https://assets-eu.researchsquare.com/files/rs-9477802/v1/be5b87300226b5993a137921.docx"}],"financialInterests":"No competing interests reported.","formattedTitle":"Spatial structure of cloud effective radius is associated with rainfall intensification in eastward propagating convective systems","fulltext":[{"header":"Introduction","content":"\u003cp\u003eOrganized convective systems account for a disproportionate share of heavy rainfall, but the microphysical pathways that regulate where and when precipitation intensifies within these systems remain incompletely constrained\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. Vertical profiles of cloud effective radius (CER) provide a powerful observational proxy for particle growth\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e, capturing transitions from condensation and collision-coalescence to mixed-phase conversion and glaciation\u003csup\u003e\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Therefore, satellite-based analysis of CER profiles has become an important tool for diagnosing precipitation formation in deep convective clouds\u003csup\u003e\u003cspan additionalcitationids=\"CR10 CR11 CR12\" citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eMost existing retrieval frameworks, however, infer CER evolution by treating an entire cloud cluster as horizontally homogeneous\u003csup\u003e\u003cspan citationid=\"CR14\" class=\"CitationRef\"\u003e14\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR15\" class=\"CitationRef\"\u003e15\u003c/span\u003e\u003c/sup\u003e. This approximation is unlikely to hold in propagating convective systems, where cloud growth, hydrometeor transport and precipitation development vary systematically along the direction of motion\u003csup\u003e\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. As a result, conventional approaches are unable to resolve whether microphysical contrasts between the leading, middle and trailing parts of a system emerge only after rainfall intensifies, or whether they develop earlier as part of the microphysical reorganization associated with subsequent rainfall intensification\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e,\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eEastward-propagating convective systems (EPCSs) over East Asia provide a particularly suitable setting in which to examine this problem. These systems evolve in an environment of sustained low-level moisture transport, frontal baroclinicity and upper-level dynamical support, allowing warm-rain, mixed-phase and ice processes to coexist within deep, rapidly evolving cloud clusters\u003csup\u003e\u003cspan additionalcitationids=\"CR20 CR21\" citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. This makes them a useful setting for examining whether the internal spatial organization of retrieved CER structure is associated with precipitation intensification. This problem also has direct implications for model development. Cloud microphysics schemes are known to exhibit systematic biases in warm-rain auto-conversion thresholds, mixed-phase partitioning and ice growth rates\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Yet these biases are commonly evaluated using cloud-system means, which may obscure process-level contrasts across dynamically distinct parts of organized convection\u003csup\u003e\u003cspan additionalcitationids=\"CR26\" citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. For propagating systems with pronounced front-to-rear gradients, observational constraints on the spatial organization of microphysics are therefore needed both to improve physical understanding and to provide more discriminating benchmarks for model evaluation.\u003c/p\u003e \u003cp\u003eHere we develop a Voronoi-based framework to construct sector-resolved CER-temperature profiles from FY-4A Advanced Geostationary Radiation Imager (AGRI) observations. We apply this framework to 162 eastward-propagating convective systems over middle-eastern China during the summers of 2021 and 2022, and evaluate the retrieval against independent aircraft in situ measurements. We do not attempt to reconstruct the full three-dimensional in-cloud microphysical field. Instead, we use retrieved CER-temperature structure as an observational proxy for the vertical evolution of cloud particle size near the cloud top and examine how its spatial organization varies along the propagation axis of eastward-propagating convective systems. The main objectives of this study are to: (1) develop a Voronoi-based framework for describing horizontally heterogeneous CER-temperature structure within organized convective systems; (2) evaluate whether this framework captures observed sector-dependent profile tendencies more consistently than a homogeneous approach; and (3) examine whether systematic sectoral differences in retrieved CER structure are associated with the timing of rainfall intensification at the composite level.\u003c/p\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eValidation of the Voronoi-CER approach\u003c/h2\u003e \u003cp\u003eWe first evaluated the Voronoi-based CER retrieval using three independent aircraft in situ cases (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e and Table\u0026nbsp;1). Compared with the conventional homogeneous-cluster approach\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e, the Voronoi framework more consistently reproduced the observed vertical tendency and turning points of CER in all three cases (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ea-f). The mean Trend Consistency Index (TC) increased from 0.47 to 0.60, indicating improved agreement in profile tendency relative to the conventional method.\u003c/p\u003e \u003cp\u003eLayer-wise error statistics further support this result (Supplementary Fig. S2). In the warm-cloud layer (5 to 0\u0026deg;C), the Voronoi retrieval showed smaller gradient-error dispersion and lower root-mean-square error (RMSE) relative to the aircraft observations than the conventional Region method. In the mixed-phase layer (0 to -20\u0026deg;C), the two methods performed comparably overall, although the Voronoi framework generally exhibited a similar or slightly narrower error spread. In the cold upper layer (\u0026lt; -20\u0026deg;C), the Voronoi retrieval reduced the magnitude of negative bias and showed lower RMSE than the conventional method, indicating improved representation of CER structure under colder cloud-top conditions.\u003c/p\u003e \u003cp\u003eThese results suggest that the benefit of the Voronoi framework is not limited to noise reduction. Rather, explicitly partitioning horizontally heterogeneous cloud structure before constructing vertical retrieved CER-temperature profiles yields better agreement with independently observed profile tendencies and reduces layer-dependent bias in key parts of the cloud column.\u003c/p\u003e \u003cp\u003eAlthough the TC improvement is modest in absolute terms, it is consistent across all three cases, suggesting a systematic rather than incidental benefit of the Voronoi partitioning approach. Complementing this point-scale evaluation, the statistical robustness of the sector-resolved structures is further examined using system-scale GPM/DPR precipitation diagnostics (Ka and Ku band reflectivity, mass-weighted mean diameter (D\u003csub\u003em\u003c/sub\u003e) and normalized intercept parameter (N\u003csub\u003ew\u003c/sub\u003e) profiles) across all EPCSs (see Composite sector-resolved CER structure across EPCSs). Together, these analyses support the interpretation that the Voronoi-based method better captures local vertical CER tendencies while providing a physically interpretable view of broader microphysical organization in organized convection.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003c/div\u003e\n\u003ch3\u003eEvolution of CER structure during convective development\u003c/h3\u003e\n\u003cp\u003eWe next examined the temporal evolution of a representative eastward-propagating convective system observed over the middle reaches of the Yangtze River on 1 June 2022 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Between 00:00 and 04:00 UTC, the cloud-cluster area increased from 7,528 km\u0026sup2; to 18,768 km\u0026sup2;, accompanied by a progressive deepening and increasing vertical differentiation of the retrieved CER\u0026ndash;temperature profile.\u003c/p\u003e \u003cp\u003eDuring the initiation stage (00:00\u0026ndash;01:00 UTC), the convective cluster developed rapidly but produced only sporadic surface rainfall (Supplementary Fig. S3a). The retrieved CER-temperature profile extended only to temperatures slightly below 0\u0026deg;C (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ef, g), reflecting the limited horizontal extent of the young cloud cluster and the insufficient number of pixels available at colder cloud-top temperatures to construct a statistically robust profile\u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. At this stage, the profile exhibited no clear sign of vertical differentiation beyond early condensational growth.\u003c/p\u003e \u003cp\u003eBy 02:00 UTC, during the developing stage, the retrieved CER-temperature profile extended from about 10\u0026deg;C to -30\u0026deg;C and displayed a clear increase-then-decrease structure (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eh). CER increased from about 11 \u0026micro;m near 5\u0026deg;C to 38\u0026ndash;42 \u0026micro;m near \u0026minus;\u0026thinsp;15 to -20\u0026deg;C, and then decreased towards cloud top. The widening interquartile range through the mixed-phase layer indicates increasing spatial heterogeneity as collision-coalescence and early ice-phase processes coexist within the system. The reduction of CER above about\u0026thinsp;\u0026minus;\u0026thinsp;25\u0026deg;C is consistent with the possible sedimentation or redistribution of large particles as precipitation intensified at the surface (Supplementary Fig. S3b, c).\u003c/p\u003e \u003cp\u003eAt maturity (03:00\u0026ndash;04:00 UTC), the retrieved CER\u0026ndash;temperature profile underwent a more pronounced structural transition (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ei, j). A near-vertical rainout zone emerged in the warm-cloud layer, where median CER stabilized at about 14\u0026ndash;17 \u0026micro;m, indicating saturation of droplet growth and sustained precipitation fallout\u003csup\u003e\u003cspan citationid=\"CR29\" class=\"CitationRef\"\u003e29\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e. Above this layer, CER increased steeply through the mixed-phase region to a maximum near \u0026minus;\u0026thinsp;15\u0026deg;C and then decreased rapidly towards cloud top. This unimodal structure is consistent with vertically differentiated particle growth, in which warm-rain production, mixed-phase growth and upper-level glaciation become increasingly separated in altitude\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. The transition from a monotonic to a unimodal profile therefore marks a change in vertical microphysical structure during convective development and occurs as rainfall intensifies at the surface.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eSpatial heterogeneity of cloud microphysics within an EPCS\u003c/h3\u003e\n\u003cp\u003eTo determine whether this evolution was spatially organized, we partitioned the cloud cluster into frontal (see Methods), middle and rear sectors using Voronoi sub-regions projected onto the principal axis of propagation (Supplementary Figs. S4).\u003c/p\u003e \u003cp\u003eDuring initiation, retrieved CER-temperature profiles in the three sectors were broadly similar (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eb, d). All sectors showed quasi-linear growth from cloud base to around \u0026minus;\u0026thinsp;30\u0026deg;C, with no pronounced frontal-to-rear gradient in profile shape or magnitude. This indicates that during the earliest stage of development, particle growth remained comparatively homogeneous across the cloud cluster and was dominated by condensation and early collision-coalescence.\u003c/p\u003e \u003cp\u003eBy the developing stage, however, clear sector-dependent contrasts had emerged (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003ef). CER increased from the frontal to the rear sector in the warm-cloud and lower mixed-phase layers. Frontal CER values ranged from about 10 to 26 \u0026micro;m, middle-sector values from 12 to 34 \u0026micro;m, and rear-sector values from 12 to 44 \u0026micro;m. This front-to-rear increase is consistent with a propagating system in which younger convective elements are concentrated near the leading edge, while larger hydrometeors accumulate in the trailing region through sedimentation and mesoscale transport\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. At the same time, the middle-sector profile extended to lower temperatures than either the frontal or rear-sector profiles, suggesting that the deepest convective development occurred in the system core.\u003c/p\u003e \u003cp\u003eAt maturity, all three sectors extended below \u0026minus;\u0026thinsp;40\u0026deg;C, but their vertical structures diverged further (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003eh, j). The frontal and middle sectors exhibited rapid quasi-linear CER growth from 10\u0026deg;C to about\u0026thinsp;\u0026minus;\u0026thinsp;15\u0026deg;C, followed by a marked reduction in growth rate at colder levels. By contrast, the rear sector maintained larger CER values below \u0026minus;\u0026thinsp;20\u0026deg;C and exhibited the most pronounced unimodal profile, with a peak near \u0026minus;\u0026thinsp;25 to -30\u0026deg;C before decreasing towards cloud top. This behavior is consistent with enhanced accumulation and redistribution of large ice-phase hydrometeors in the rear portion of the system\u003csup\u003e\u003cspan additionalcitationids=\"CR34\" citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u003c/sup\u003e. The observed sectoral contrast therefore cannot be explained as purely local variability; rather, it indicates a dynamically organized spatial structure of cloud microphysics along the propagation axis.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e\n\u003ch3\u003eComposite sector-resolved CER structure across EPCSs\u003c/h3\u003e\n\u003cp\u003eTo assess whether the sector-dependent CER organization identified in the case study is a recurring feature, we composited the retrieved CER\u0026ndash;temperature structures from all selected events and compared them with collocated GPM/DPR precipitation diagnostics. These pooled composites are intended to describe common structural tendencies across the sample rather than to imply pixel-level independence or deterministic behavior in individual events.\u003c/p\u003e \u003cp\u003eAll three sectors exhibited a similar background CER structure, with quasi-linear growth from approximately 10\u0026ndash;12 \u0026micro;m to 20\u0026ndash;22 \u0026micro;m through the warm-cloud and lower mixed-phase layers (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea), consistent with the canonical rainout regime reported in satellite-based microphysical profiling studies\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e,\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. Superimposed on this common structure, however, the depth of the warm-rain growth layer differed systematically among sectors. The temperature interval between the onset of precipitation-sized droplets (T\u003csub\u003e14\u003c/sub\u003e) and the termination of quasi-linear growth (T\u003csub\u003eL\u003c/sub\u003e) was approximately 15\u0026deg;C in the frontal sector, compared with about 13\u0026deg;C in the middle and rear sectors (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ea). These results suggest that the leading part of EPCSs sustains a deeper layer of efficient droplet growth before CER departs from quasi-linear behavior, indicating a systematically organized microphysical structure along the propagation axis\u003csup\u003e\u003cspan citationid=\"CR36\" class=\"CitationRef\"\u003e36\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eThe corresponding vertical gradient profiles further highlight differences in particle evolution among sectors (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003eb-d). In the warm-cloud layer, the frontal sector exhibited weaker positive gradients than the middle and rear sectors, indicating a more gradual increase in CER with height. In the mixed-phase layer, the middle and rear sectors showed stronger negative gradients, implying more rapid particle-size adjustment as clouds extended into colder levels\u003csup\u003e\u003cspan citationid=\"CR31\" class=\"CitationRef\"\u003e31\u003c/span\u003e\u003c/sup\u003e. Above this layer, gradient magnitudes weakened in all sectors, but the rear sector maintained comparatively large CER values below about\u0026thinsp;\u0026minus;\u0026thinsp;20\u0026deg;C, consistent with a greater persistence of larger hydrometeors in the trailing part of the system.\u003c/p\u003e \u003cp\u003eTo further quantify the temporal evolution of this microphysical organization, we introduced two CER-based indices: CER\u003csub\u003ePeak\u003c/sub\u003e, defined as the maximum CER within the retrieved CER-temperature profile, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003e, defined as the temperature at which the maximum CER occurs. CER\u003csub\u003ePeak\u003c/sub\u003e characterizes the amplitude of particle growth, whereas \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003eindicates the vertical position of peak particle growth within the cloud. The temporal composites show that both indices begin to evolve before the rainfall intensification time (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee, f). In all three sectors, CER\u003csub\u003ePeak\u003c/sub\u003e increased prior to the onset of intensified surface rainfall, with the composite increase beginning about 75 min before \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\tau\\:=0\\)\u003c/span\u003e\u003c/span\u003e, indicating that the production of larger particles starts before the rainfall maximum is reached (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ee). This earlier increase is most pronounced in the frontal and middle sectors, consistent with the sector-dependent enhancement of warm rain growth inferred from the vertical retrieved CER-temperature profiles. Meanwhile, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003eshifted toward colder temperatures around the intensification time (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003ef), indicating that the vertical level of maximum particle growth was reorganized as the system developed. Notably, the frontal sector exhibited a weaker cooling tendency in \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003ethan the middle and rear sectors. This more gradual downward evolution is consistent with the deeper rainout growth layer identified in the frontal-sector CER profiles, suggesting that efficient warm-rain growth is sustained over a greater temperature depth before the peak-growth level shifts upward into colder layers. Taken together, the composite statistics indicate that the particle-size reorganization observed in the case study is not an isolated feature.\u003c/p\u003e \u003cp\u003eInstead, the selected 162 EPCSs exhibit a consistent composite signal of microphysical reorganization on timescales leading surface-rainfall intensification. These results support a composite-level lead-lag association between changes in CER structure and subsequent precipitation enhancement, but do not by themselves establish predictive skill or a deterministic causal relationship. They occur within a broader large-scale environment favorable for organized propagating convection, including upper-level divergence, mid-level ascent, and strong low-level moisture supply (Supplementary Fig. S5), although those composites provide background context rather than sector-resolved process attribution.\u003c/p\u003e \u003cp\u003eThese optical retrievals are broadly consistent with the collocated radar-based precipitation structure. The GPM/DPR contoured frequency-by-altitude diagrams showed a vertically continuous precipitation column, with moderate raindrop diameters extending from the surface to about 6 km altitude, close to the \u0026minus;\u0026thinsp;5\u0026deg;C isotherm and to the upper limit of the rainout layer inferred from the retrieved CER-temperature profiles (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ec). In addition, neither the Ku nor Ka-band reflectivity composites showed a pronounced bright-band signature near the melting level (Fig.\u0026nbsp;\u003cspan refid=\"Fig5\" class=\"InternalRef\"\u003e5\u003c/span\u003ea, b), indicating that the sampled systems were still dominated by warm-rain and early mixed-phase processes rather than by a fully developed stratiform melting layer. Across the 162 EPCSs, the frontal, middle, and rear sectors therefore form a coherent propagation-relative sequence in retrieved CER structure and precipitation organization.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cp\u003e \u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThe results suggest that the internal microphysics of EPCSs becomes increasingly spatially organized as rainfall intensification develops. First, retrieved CER\u0026ndash;temperature profiles tend to evolve from monotonic growth to a unimodal structure during convective development. Second, distinct frontal, middle, and rear-sector contrasts emerge before the systems reach full maturity (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003ea). Together, these findings indicate increasing spatial differentiation of particle growth across the propagation axis as propagating convection evolves. From an observational perspective, the transition from a monotonic to a unimodal retrieved CER-temperature profile can therefore be interpreted as a signature of increasingly vertically differentiated microphysical structure in organized convection\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan additionalcitationids=\"CR29\" citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR30\" class=\"CitationRef\"\u003e30\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eAn important result is that this organization becomes evident before intense rainfall is fully established at the surface. The temporal evolution of CER\u003csub\u003ePeak\u003c/sub\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003eshows that both the amplitude of particle growth and the vertical position of peak particle growth begin to adjust systematically before the defined rainfall-intensification time. This behavior is consistent with a stage of microphysical reorganization before precipitation enhancement, in which larger particles develop and the level of maximum particle growth shifts within the cloud column. In particular, the CER\u003csub\u003ePeak\u003c/sub\u003e increases before rainfall intensification, while \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003eshifts toward colder temperatures, indicating that the vertical location of peak particle growth becomes reorganized as the system develops. The frontal sector exhibits a weaker cooling tendency in \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{T}_{peak}\\)\u003c/span\u003e\u003c/span\u003ethan the middle and rear sectors, which is consistent with the deeper rainout growth layer identified in the frontal-sector CER profiles. The agreement between these CER-based indices and the sector-resolved profile evolution further suggests that intracloud microphysical structure is statistically associated with the transition toward stronger rainfall. However, the present analysis is based on composite lead-lag behavior and should not be interpreted as demonstrating event-by-event predictive skill or a deterministic precursor relationship.\u003c/p\u003e \u003cp\u003eThe mature-stage sectoral pattern is broadly consistent with the canonical leading-convective/trailing-stratiform framework of mesoscale convective systems\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e,\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e, which supports the physical interpretability of the Voronoi-CER framework. At the same time, the frontal sector exhibits the deepest warm rain growth layer, whereas the rear sector contains the largest particles and the strongest unimodal signatures at maturity. This pattern suggests that warm-rain efficiency may depend not only on convective maturity, but also on the thermodynamic and microphysical characteristics of the inflow feeding the leading edge\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e. One physically plausible interpretation is that sustained low-level inflow into the frontal sector may help maintain a deeper layer of efficient warm rain growth\u003csup\u003e\u003cspan citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, while the rear sector may become increasingly influenced by hydrometeor loading, sedimentation, and rear-inflow dynamics\u003csup\u003e\u003cspan additionalcitationids=\"CR39\" citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e (Fig.\u0026nbsp;\u003cspan refid=\"Fig6\" class=\"InternalRef\"\u003e6\u003c/span\u003eb, c). However, the environmental analysis presented here is composited at system scale rather than sector resolved (Supplementary Fig. S5), so these process-level interpretations should be regarded as physically motivated hypotheses rather than demonstrated mechanisms. Direct sector-resolved dynamical evidence would be required to test them more rigorously.\u003c/p\u003e \u003cp\u003eThis study also highlights a broader methodological point. Conventional CER profiling methods assume horizontal homogeneity at the cloud-cluster scale and therefore emphasize the mean state of the system. The Voronoi framework relaxes this assumption and instead treats cloud organization as part of the retrieval problem, making it better suited to propagating systems in which microphysical evolution is coupled to advection, differential cloud stage, and mesoscale circulation\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eSeveral limitations should nevertheless be noted. First, validation of the retrieval is based on only three independent aircraft cases. Although these cases do not sample the full diversity of cloud regimes represented in the 162-event composite, they provide targeted evidence that the Voronoi-based method better captures the tendency of CER evolution than the conventional homogeneous-cluster approach. Second, the current environmental analysis is not sector resolved and therefore cannot directly attribute the frontal-to-rear microphysical contrasts to specific dynamical or thermodynamic mechanisms; the process interpretations discussed above should accordingly be viewed as physically motivated hypotheses. Third, the final 162-event sample represents a filtered subset of summer daytime eastward-propagating systems over middle-eastern China, selected further by the availability of collocated GPM/DPR observations and a lifetime threshold of 6 h. The results should therefore be interpreted within the scope of this observational subset rather than as universally representative of all propagating convective systems. Fourth, the composite analysis is based on pooled pixels and is intended to describe recurring structural tendencies across the sample rather than pixel-level independence or deterministic behavior in individual events.\u003c/p\u003e \u003cp\u003eDespite these limitations, the main conclusions remain robust within the scope of the present evidence. EPCSs exhibit a spatially organized microphysical structure prior to rainfall intensification at the composite level, and this organization can be detected in sector-resolved retrieved CER-temperature profiles. These findings provide an observational constraint on precipitation formation in organized convection and offer a new benchmark for evaluating how models represent the early microphysical evolution of convective systems.\u003c/p\u003e \u003cp\u003e \u003c/p\u003e \u003cdiv id=\"Sec8\" class=\"Section2\"\u003e \u003ch2\u003eMethods\u003c/h2\u003e \u003cdiv id=\"Sec9\" class=\"Section3\"\u003e \u003ch2\u003eObservational datasets\u003c/h2\u003e \u003cp\u003eThe satellite data used in this study consist of FY-4A/AGRI full-disk observations during June-August of 2021\u0026ndash;2022 over middle-eastern China (105\u0026ndash;120 \u0026ordm;E, 24\u0026ndash;39 \u0026ordm;N). FY-4A provides geostationary observations of the Earth-atmosphere system at 15 min temporal resolution. Cloud effective radius (CER) fields were derived following the algorithm of Sun et al.\u003csup\u003e41\u003c/sup\u003e, which reported a correlation of 0.91 against the MODIS Collection 6.1 CER product. For consistency in object identification and sector-resolved analysis, all satellite fields were mapped to a common analysis grid prior to processing.\u003c/p\u003e \u003cp\u003eQuality-controlled precipitation observations from automatic weather stations, available every 5 min, were interpolated onto a 0.04\u003csup\u003e\u0026ordm;\u003c/sup\u003egrid using a radial basis function method. For each EPCS time step, the precipitation field nearest in time to the corresponding FY-4A/AGRI scan was used, with a maximum temporal mismatch of 5 min. ERA5 hourly reanalysis fields\u003csup\u003e\u003cspan citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u003c/sup\u003e, including horizontal wind components, vertical velocity, specific humidity, relative humidity, and column-integrated moisture flux, were extracted at 0.25\u003csup\u003e\u0026ordm;\u003c/sup\u003e resolution and composited over the 162 events in a common system-relative framework to characterize the large-scale dynamical and thermodynamic background of the analyzed EPCS sample. GPM/DPR Ku- and Ka-band reflectivity profiles and retrieved microphysical parameters were considered collocated when they intersected the EPCS boundary within 30 min of the corresponding AGRI observation, and were then composited into propagation-relative CFADs.\u003c/p\u003e \u003cp\u003eThree independent aircraft cloud-penetration missions conducted by the KA350 research aircraft were used to evaluate the FY-4A/AGRI CER retrieval framework (Supplementary Fig. \u003cspan refid=\"MOESM1\" class=\"InternalRef\"\u003eS1\u003c/span\u003e). The aircraft primarily employed a combination of horizontal level legs and stepwise vertical penetrations. After takeoff, the aircraft ascended to cloud top or to a maximum altitude of approximately 7300 m and circled over the target area. It then descended in steps of about 300 m, maintaining stable horizontal orbits at each altitude level to collect cloud-particle and precipitation-particle measurements. The minimum flight altitude was approximately 1500 m, after which the aircraft ascended again to complete a closed vertical profile. This stepwise sampling strategy provided vertically resolved in situ measurements of cloud droplet spectra and precipitation particle spectra\u003csup\u003e\u003cspan citationid=\"CR43\" class=\"CitationRef\"\u003e43\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eFor aircraft\u0026ndash;satellite evaluation, the in-situ observations were collocated with the nearest FY-4A/AGRI CER retrieval in both time and space. Specifically, aircraft samples were matched to the closest AGRI scan within a temporal window of \u0026plusmn;\u0026thinsp;7.5 min and within a 4 km spatial window centered on the aircraft position, consistent with the FY-4A resolution used in this study. Because aircraft observations sample along a moving flight path at much finer scales than satellite retrievals, the in-situ measurements within each collocation window were aggregated to the AGRI sampling scale before comparison. Aircraft CER values were then grouped by ambient temperature and compared with the satellite-derived retrieved CER-temperature profile using the same temperature-bin framework as in the satellite analysis. Given the inherent scale mismatch between aircraft in situ sampling and satellite pixel-scale retrievals, this comparison is intended as a targeted evaluation of vertical CER structure rather than a point-to-point validation of instantaneous pixel values.\u003c/p\u003e \u003cp\u003eStarting from all mobile convective cloud clusters identified over China during June-August of 2021\u0026ndash;2022, we focused on eastward-propagating systems because they represent the dominant propagation class in the study region and provide a more dynamically homogeneous population for process-oriented composite analysis (Supplementary Fig. S6). A total of 9481 mobile convective cloud clusters were identified during the study period, of which eastward-propagating systems accounted for 33.1% and westward-propagating systems for 17.0%. Because FY-4A/AGRI CER retrievals rely on daytime solar-reflectance information, only daytime eastward-propagating events were retained. To further ensure sufficient lifecycle sampling and an independent constraint on precipitation vertical structure, we additionally required collocated GPM/DPR overpasses and a system lifetime longer than 6 h. These filtering steps yielded the final sample of 162 EPCSs used in this study. This dataset should therefore be regarded as a restricted observational subset rather than as the full population of EPCSs over China, and the conclusions are intended to apply to this selected sample of summer daytime eastward-propagating systems over middle-eastern China.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e\n\u003ch3\u003eIdentification and tracking of convective cloud clusters\u003c/h3\u003e\n\u003cp\u003eConvective cloud clusters are identified and tracked using the FLEXTRKR algorithm\u003csup\u003e\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e, with criteria following Chen et al.\u003csup\u003e45\u003c/sup\u003e. Convective cores were defined by 10.8 \u0026micro;m brightness temperatures below 263 K, and cluster boundaries by contiguous regions below 280 K. A minimum cluster area of 256 km\u003csup\u003e2\u003c/sup\u003e was required, together with a contiguous precipitation region having intensity\u0026thinsp;\u0026ge;\u0026thinsp;0.5 mm h\u003csup\u003e\u0026minus;\u0026thinsp;1\u003c/sup\u003e and a longest axis\u0026thinsp;\u0026ge;\u0026thinsp;8 km. Propagating convective clusters were identified from centroid displacement, with eastward-propagating systems defined by movement azimuths between 22.5\u0026ordm; and 157.5\u0026ordm;.\u003c/p\u003e \u003cdiv id=\"Sec11\" class=\"Section2\"\u003e \u003ch2\u003eConstruction of sector-resolved Voronoi-retrieved CER\u0026ndash;temperature profiles\u003c/h2\u003e \u003cp\u003eThe Voronoi-CER method constructs sector-resolved CER-temperature profiles through the following steps. First, the FLEXTRKR-derived cloud-cluster boundary defines the analysis domain. Second, K-means clustering is applied to CER fields within the domain to identify coherent subregions. Rather than prescribing a fixed number of subregions, the optimal cluster number (K) is adaptively determined for each cloud system using a composite index combining the Silhouette, Calinski-Harabasz, and Davies-Bouldin scores (Supplementary Fig. S7). Third, the centroids of the resulting clusters are used as seed points to generate Voronoi tessellations, partitioning the cloud system into K polygonal subregions. Fourth, the principal axis of the cloud cluster, derived from principal component analysis and aligned with the propagation direction, is used to group these subregions into front, middle, and rear sectors of equal length along the propagation axis (Supplementary Fig. S4). Finally, CER values within each sector are composited as a function of cloud-top brightness temperature using 1.75 K temperature bins. The 25th, 50th, and 75th percentiles are then calculated, with the median defining the retrieved CER-temperature profile and the interquartile range representing sectoral variability.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec12\" class=\"Section2\"\u003e \u003ch2\u003eTrend Consistency Index\u003c/h2\u003e \u003cp\u003eTo evaluate the reliability of the Voronoi-CER method, three aircraft in-situ cases are used to compare satellite-retrieved and aircraft-observed CER\u0026ndash;temperature profiles. A trend consistency index (TC) is defined as:\u003cdiv id=\"Equ1\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ1\" name=\"EquationSource\"\u003e\n$$\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:\\:TC=\\frac{{N}_{consistent}}{{N}_{total}}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e1\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere N\u003csub\u003etotal\u003c/sub\u003e is the number of adjacent-layer comparisons, N\u003csub\u003econsistent\u003c/sub\u003e is the number of cases in which the direction of CER change (increase or decrease) is consistent between satellite and aircraft observations. A value of TC\u0026thinsp;=\u0026thinsp;1 indicates perfect agreement, while TC\u0026thinsp;=\u0026thinsp;0.5 corresponds to random agreement.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec13\" class=\"Section2\"\u003e \u003ch2\u003eDefinition of rainfall intensification\u003c/h2\u003e \u003cp\u003eRainfall intensification is defined based on the cloud averaged precipitation within the cloud boundary at each time step. The mean 5 min accumulated precipitation is the first converted to rainfall rate \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:R\\left(t\\right)\\)\u003c/span\u003e\u003c/span\u003e. The fractional coverage of effective precipitation is defined as:\u003cdiv id=\"Equ2\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ2\" name=\"EquationSource\"\u003e\n$$\\:\\:\\:{A}_{0.05}\\left(t\\right)=\\frac{N\\left(P\\ge\\:0.05\\right)}{{N}_{total}}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e2\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:N\\left(P\\ge\\:0.05\\right)\\)\u003c/span\u003e\u003c/span\u003eis the number of grid points within the cloud with precipitation no less than 0.05 mm per 5min, and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{N}_{total}\\)\u003c/span\u003e\u003c/span\u003e is the total number of grid points within the cloud.\u003c/p\u003e \u003cp\u003eThe rainfall intensification time \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{t}_{int}\\)\u003c/span\u003e\u003c/span\u003e is identified as the first time when the following conditions are satisfied for at least two consecutive time steps:\u003cdiv id=\"Equ3\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ3\" name=\"EquationSource\"\u003e\n$$\\:{A}_{0.05}\\left(t\\right)\\ge\\:0.1$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e3\u003c/div\u003e\u003c/div\u003e\u003cdiv id=\"Equ4\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ4\" name=\"EquationSource\"\u003e\n$$\\:R\\left(t\\right)\\ge\\:0.5$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e4\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eThe relative time is then defined as\u003cdiv id=\"Equ5\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ5\" name=\"EquationSource\"\u003e\n$$\\:\\tau\\:=t-{t}_{int}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e5\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cdiv id=\"Sec14\" class=\"Section3\"\u003e \u003ch2\u003eCER Index\u003c/h2\u003e \u003cp\u003eTo characterize the intensity of cloud microphysical development, the peak value of CER is used. In this study, CER\u003csub\u003ePeak\u003c/sub\u003e and CER\u003csub\u003etopbase\u003c/sub\u003e are used as descriptive indices of microphysical evolution relative to the rainfall-intensification time. They are not intended here as event-level predictive metrics, and no formal forecast-skill evaluation is performed.\u003cdiv id=\"Equ6\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ6\" name=\"EquationSource\"\u003e\n$$\\:{\\:\\:\\:\\:\\:\\:\\:\\:CER}_{\\_Peak}=\\text{m}\\text{a}\\text{x}\\left(CER\\right)$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e6\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003eTo further summarize the vertical structure of cloud microphysical properties, a second CER-based index is introduced. CER\u003csub\u003etopbase\u003c/sub\u003e is defined as the difference between the maximum and minimum CER values along the cloud vertical profile and is used here as a simple descriptive measure of vertical CER contrast. Because this index depends on profile extrema, it can be sensitive to local variability; accordingly, it is interpreted together with the full sector-resolved retrieved CER-temperature profile composites rather than as a standalone robust metric.\u003cdiv id=\"Equ7\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equ7\" name=\"EquationSource\"\u003e\n$$\\:{\\:\\:\\:\\:\\:\\:\\:\\:CER}_{topbase}={\\text{C}\\text{E}\\text{R}}_{max}-{CER}_{min}$$\u003c/div\u003e\u003cdiv class=\"EquationNumber\"\u003e7\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e \u003cp\u003ewhere CER_max and CER_min represent the maximum and minimum CER values along the cloud vertical profile, respectively.\u003c/p\u003e \u003c/div\u003e \u003c/div\u003e "},{"header":"Declarations","content":"\u003cp\u003e \u003ch2\u003eCompeting interests\u003c/h2\u003e \u003cp\u003eThe authors have no conflicts of interest to disclose.\u003c/p\u003e \u003c/p\u003e\u003ch2\u003eAuthor Contribution\u003c/h2\u003e\u003cp\u003eJ.S. and Y.Y.L. conceived the study. J.S., X.H., and Z.T.C. processed and organized the data. J.S. wrote the original draft. J.S. and Y.Y.L. developed the methodology, designed the algorithms, and revised the manuscript. X.H., Z.T.C., X.L., and F.L. contributed to the interpretation of the results and to manuscript revision.\u003c/p\u003e\u003ch2\u003eAcknowledgments\u003c/h2\u003e \u003cp\u003eThis work was supported by the National Natural Science Foundation of China(U2242201), the Hubei Provincial Natural Science Foundation of China(2025AFD423), and the Open Fund Project for Heavy Rain (BYKJ2024M07).\u003c/p\u003e\u003ch2\u003eData Availability\u003c/h2\u003e\u003cp\u003eFY-4A/AGRI Level-1 radiance data (2021-2022) were obtained from the Fengyun Satellite Data Center (https://satellite.nsmc.org.cn/DataPortal/cn/data/order.html). Cloud effective radius (CER) retrievals were derived using the algorithm described in Sun et al. (2025). The PyFLEXTRKR algorithm can be obtained from https://github.com/FlexTRKR/PyFLEXTRKR. The GPM 2ADPR precipitation data used in this study were collected from the Precipitation Measurement Mission website (https://pmm.nasa.gov). ERA5 reanalysis data (hourly, 0.25\u0026ordm; resolution) were obtained from the Copernicus Climate Data Store (https://cds.climate.copernicus.eu/) to characterize the large-scale dynamical and thermodynamic environment of the EPCSs. The custom code used in this study for Voronoi partitioning, adaptive selection of the optimal cluster number K, is available from the corresponding author during peer review and will be deposited in a public repository upon publication.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eSchumacher, C. \u0026amp; Houze, R. A. Jr. Stratiform rain in the tropics as seen by the TRMM precipitation radar. \u003cem\u003eJ. Clim.\u003c/em\u003e 16, 1739\u0026ndash;1756 (2003).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKukulies, J., et al. Mesoscale convective systems in the third pole region: Characteristics, mechanisms and impact on precipitation. Front. Earth Sci. 11, 1143380 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eZhao, Y., et al. Distinct structure, radiative effects, and precipitation characteristics of deep convection systems in the Tibetan Plateau compared to the tropical Indian Ocean, Atmos. Chem. Phys. 24, 9435\u0026ndash;9457 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShang, H., et al. Remote sensing of liquid cloud profiles based on an analytical cloud profiling model. Sci. China Earth Sci. 68, 998\u0026ndash;1012 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, Y. T., et al. Characterization of liquid cloud profiles using global collocated active radar and passive polarimetric cloud measurements. Atmos. Chem. Phys., 25, 16167\u0026ndash;16187 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenfeld, D., \u0026amp; Lensky, I. M. Satellite-based insights into precipitation formation processes in continental and maritime convective clouds. Bull. Amer. Meteor. Soc. 79, 2457\u0026ndash;2476 (1998).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKokhanovsky, A.A., Painemal, D., \u0026amp; Rozanov, V. V. The Intercomparison of Satellite-Derived and In Situ Profiles of Droplet Effective Radii in Marine Stratocumulus Clouds. IEEE Geosci. Remote Sens. Lett. 10(5), 1147\u0026ndash;1151 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenfeld, D. Cloud-aerosol-precipitation interactions based of satellite retrieved vertical profiles of cloud microstructure. In Remote sensing of aerosols, clouds, and precipitation. 6, 129\u0026ndash;152 (2018).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, Y. L., et al. Retrieval of the vertical evolution of the cloud effective radius from the Chinese FY-4 (Feng Yun 4) next-generation geostationary satellites. Atmos. Chem. Phys. 20, 1131\u0026ndash;1145 (2020a).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, Y. L., et al. Linkage between the vertical evolution of clouds and droplet growth modes as seen from FY-4A AGRI and GPM DPR. Geophys. Res. Lett. 47, e2020GL088312 (2020b).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eEfraim, A., et al. Satellite-based detection of secondary droplet activation in convective clouds. J. Geophys. Res. Atmos. 127, e2022JD036519 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFreud, E., \u0026amp; Rosenfeld, D. Linear relation between convective cloud drop number concentration and depth for rain initiation. J. Geophys. Res. 117, D02207 (2012).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenfeld, D., et al. High-resolution (375 m) cloud microstructure as seen from the NPP/VIIRS satellite imager. Atmos. Chem. Phys. 14, 2479\u0026ndash;2496 (2014).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLetu, H., et al. High-resolution retrieval of cloud microphysical properties and surface solar radiation using Himawari-8/AHI next-generation geostationary satellite. Remote Sens. Environ. 239, 111583 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLiu, C., et al. A cloud optical and microphysical property product for the advanced geosynchronous radiation imager onboard China\u0026rsquo;s Fengyun-4 satellites: The first version. Atmos. Ocean. Sci. Lett. 16, 100337\u0026ndash;100343 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong, F. F., et al. Crucial roles of eastward propagating environments in the summer MCS initiation over the U.S. Great Plains. J. Geophys. Res. Atmos. 126, e2021JD034991(2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSokol, A. B., \u0026amp; Storelvmo, T. The spatial heterogeneity of cloud phase observed by satellite. J. Geophys. Res. Atmos. 129, e2023JD039751 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHuang, Y., \u0026amp; Dewitt, B. Differentiating Thick Clouds from Thin Clouds by Using Intensity Inhomogeneity. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens. 18, 5782\u0026ndash;5792 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eGuo, Y., et al. Trends in warm season mesoscale convective systems over Asia in 2001\u0026ndash;2020. J. Geophys. Res. Atmos. 128, e2023JD038969 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eYang, R., et al. The comparison of statistical features and synoptic circulations between the eastward-propagating and quasi-stationary MCSs during the warm season around the second-step terrain along the middle reaches of the Yangtze River. Sci. China Earth Sci, 63, 1209\u0026ndash;1222(2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, P., et al. Sensitivity of simulated mesoscale convective systems over East Asia to the treatment of convection in a high-resolution GCM. Clim. Dyn. 60, 2783\u0026ndash;2801 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFu, Y., et al. Initiations of mesoscale convective systems in the middle reaches of the Yangtze River Basin based on FY-4A satellite data: Statistical characteristics and environmental conditions. J. Geophys. Res. Atmos. 128, e2023JD038630 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSong, X., \u0026amp; Zhang, G. J. Microphysics parameterization for convective clouds in a global climate model: Description and single-column model tests, J. Geophys. Res. 116, D02201, (2011).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, J.Y., et al. Impact of a new cloud microphysics parameterization on the simulations of mesoscale convective systems in E3SM. J. Adv. Model. Earth Syst. 13, e2021MS002628 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eLi, S. S., et al. Statistical characteristics and synoptic patterns of convection initiation over the middle reaches of the Yangtze River Basin as observed using the Fengyun-4A satellite. J. Hydrometeorol. 25(3), 445\u0026ndash;463 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWei, Q., et al. Convection initiation over middle\u0026ndash;eastern China: Statistics show higher frequencies over mountains with less stringent atmospheric conditions and more heterogeneous surface characteristics. Atmos. Res. 328, 108438\u0026ndash;108454 (2026).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKumar, S., \u0026amp; Srivastava, S. Vertical characteristics of precipitating cloud systems during different phases of life cycle of cloud systems using satellite-based radar over tropical oceanic areas. J. Appl. Nat. Sci. 14(4), 1272\u0026ndash;1285 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenfeld, D., et al. Satellite detection of severe convective storms by their retrieved vertical profiles of cloud particle effective radius and thermodynamic phase. \u003cem\u003eJ. Geophys. Res. Atmos.\u003c/em\u003e 113, D04208 (2008).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eBraga, R. C., et al. Linear relationship between effective radius and precipitation water content near the top of convective clouds: measurement results from ACRIDICON\u0026ndash;CHUVA campaign, Atmos. Chem. Phys., 21, 14079\u0026ndash;14088 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eRosenfeld, D., et al., Aerosol-driven droplet concentrations dominate coverage and water of oceanic low level clouds, Science, 363, 6427 (2019).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSuzuki, K., Nagao, T. M., \u0026amp; Murai, A. Satellite-based diagnostics of precipitation process in mixed‐phase clouds: Extension from warm rain process statistics. Geophys. Res. Lett. 51, e2024GL110573 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHouze, R. A. Jr. Mesoscale convective systems. Rev. Geophys., 42, RG4003 (2004).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eShort, E, Lane, T., \u0026amp; Vincent, C., Objectively Diagnosing Characteristics of Mesoscale Organization from Radar Reflectivity and Ambient Winds. Mon. Weather Rev. 151(3) (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHan, C., et al. Sensitivity of cloud-phase distribution to cloud microphysics and thermodynamics in simulated deep convective clouds and SEVIRI retrievals, Atmos. Chem. Phys., 23, 14077\u0026ndash;14095 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eKorolev, A., et al. High ice water content in tropical mesoscale convective systems (a conceptual model), Atmos. Chem. Phys., 24, 11849\u0026ndash;11881 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWang, Z, \u0026amp; Yuan, J. Observing convective activities in complex convective organizations and their contributions to precipitation and anvil cloud amounts. Atmos. Chem. Phys. 24(23):13811\u0026ndash;13831 (2024).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSkow, A., et al. A Multi-Platform, In-Situ Kinematic and Microphysical Analysis of a Hybrid Parallel/Trailing Stratiform Mesoscale Convective System. Mon. Weather Rev. 150(4):927\u0026ndash;948 (2022).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eWeisman, M. The Role of Convectively Generated Rear-Inflow Jets in the Evolution of Long-Lived Mesoconvective Systems. J. Atmos. Sci. 49: 1826\u0026ndash;1847 (1992).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eNizar, S., Thomas, J., \u0026amp; Jainet P.J, K. P. S. A Novel Technique for Nowcasting Extreme Rainfall Events using Early Microphysical Signatures of Cloud Development. PLOS Clim 4(5), e0000497 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSmull, B. F., \u0026amp; Houze, R. A., Jr. Rear inflow in squall lines with trailing stratiform precipitation. Mon, Wea, Rev. 115(12), 2869\u0026ndash;2889. (1987).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eSun, J., et al. Impact of Cloud Vertical Structure Perturbations on the Retrieval of Cloud Optical Thickness and Effective Radius from FY4A/AGRI. Atmos. Chem. Phys. 25(22),16347\u0026ndash;16361 (2025).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eHersbach, H., et al. The ERA5 global reanalysis. Q. J. R. Meteorol. Soc., 146(730), 1999\u0026ndash;2049 (2020).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eCui, C., et al. Phase Two of the Integrative Monsoon Frontal Rainfall Experiment (IMFRE-II) over the middle and lower reaches of the Yangtze River in 2020. Adv. Atmos. Sci.,38,346\u0026ndash;356 (2021).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eFeng, Z., et al. PyFLEXTRKR: a flexible feature tracking Python software for convective cloud analysis. Geosci. Model Dev. 16(10), 2753\u0026ndash;2776 (2023).\u003c/span\u003e\u003c/li\u003e \u003cli\u003e\u003cspan\u003eChen, Z., et al. Summer surface rainfall deviations from convective cold cloud shields over China. Geophys. Res. Lett, 52, e2025GL117889. (2025).\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-9477802/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-9477802/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eOrganized convective systems contribute substantially to warm season rainfall, yet their cloud microphysical structure is often analyzed as horizontally homogeneous. Here we introduce a Voronoi based framework to characterize sector resolved cloud effective radius (CER)-temperature structure from FY-4A geostationary satellite observations and apply it to 162 eastward-propagating convective systems over middle eastern China during 2021\u0026ndash;2022. Targeted evaluation against three independent aircraft in situ cases shows that, relative to a homogeneous cluster approach, the proposed framework more consistently reproduces observed CER-temperature profile tendencies, with the mean Trend Consistency index increasing from 0.47 to 0.60. In both case and composite analyses, the retrieved CER-temperature structure evolves from near monotonic growth to a more clearly unimodal pattern as rainfall intensifies. Sectoral contrasts emerge before the defined rainfall intensification time. The frontal sector exhibits the deepest warm rain growth layer, whereas the rear sector contains larger particles and a stronger unimodal structure during mature stages. These results indicate a composite level lead-lag association between the internal spatial organization of CER structure and subsequent rainfall intensification. More broadly, the framework provides an observationally constrained way to examine horizontally heterogeneous CER structure and its relation to precipitation evolution in propagating convective systems.\u003c/p\u003e","manuscriptTitle":"Spatial structure of cloud effective radius is associated with rainfall intensification in eastward propagating convective systems","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2026-05-06 17:06:20","doi":"10.21203/rs.3.rs-9477802/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"49390c7e-2a3b-4a0e-b6c9-8db2a3183c5e","owner":[],"postedDate":"May 6th, 2026","published":true,"recentEditorialEvents":[{"type":"decision","content":"Rejected","date":"2026-05-18T00:30:36+00:00","index":"","fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-14T20:15:58+00:00","index":37,"fulltext":""},{"type":"editorInvitedReview","content":"","date":"2026-05-09T07:49:53+00:00","index":36,"fulltext":""},{"type":"reviewerAgreed","content":"61199493244533441891665033342821809221","date":"2026-05-02T00:02:48+00:00","index":33,"fulltext":""},{"type":"reviewerAgreed","content":"286305223187174788139141775748095732493","date":"2026-05-01T11:50:28+00:00","index":32,"fulltext":""},{"type":"reviewerAgreed","content":"76209740531852237748196147267688161684","date":"2026-05-01T07:40:38+00:00","index":31,"fulltext":""}],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":67468689,"name":"Earth and environmental sciences/Climate sciences"},{"id":67468690,"name":"Earth and environmental sciences/Hydrology"}],"tags":[],"updatedAt":"2026-05-18T00:39:28+00:00","versionOfRecord":[],"versionCreatedAt":"2026-05-06 17:06:20","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-9477802","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-9477802","identity":"rs-9477802","version":["v1"]},"buildId":"XKTyCvWXoU3ODBz1xrDgd","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}