Weixin Xu

Mesoscale convective systems (MCSs) are the major contributors of warm-season rainfall and severe weather over southern China, which can develop under various large-scale circulation types. Using radar observations of > 1000 MCS lifecycles, this study investigates the characteristics and triggering mechanisms of MCSs associated with six major synoptic situations, including the frontal (FT), tropical low (TL), shear line due to terrain (SLT), shear line associated with vortex (SLV), southwest vortex (SWV), and weak synoptic forcing (WSF). Results show that the initiation of MCSs in SC mainly peaks at nighttime, except WSF-MCSs which initiate in the afternoon. Most MCSs reach maturation 4–5h after the initiation and exhibit a bimodal diurnal pattern with peaks in the afternoon and early morning. Specifically, FT-MCSs own the fastest propagation speed and longest lifecycle, but they have the lowest rain rate and weakest convective intensity (lightning frequency) due to a relatively dry and stable atmospheric conditions. In contrast, TL-MCSs have the slowest movement, longest duration, and intense precipitation, which may impose greater flooding risk. Interestingly, MCSs under weak synoptic forcing (WSF) are the strongest in terms of precipitation and convective intensity likely owing to the humid and unstable storm environments. As a result, WSF-MCSs contribute the most (~50%) extreme hourly rainfall events in SC. These findings highlight the critical role of large-scale environments in shaping MCS lifecycle, structure and intensity, providing valuable insights for improving high-impact weather forecasting and understanding regional climate variability.

Accurate radar quantitative precipitation estimation (QPE) remains challenging in regions with complex terrain and diverse precipitation regimes, such as South China, because traditional methods rely on empirical Z-R relations derived from raindrop size distributions. Using dense rain gauge network and ground-based radar measurements, this study fully investigates the potential of deep learning (DL) methods in improving radar QPE over South China. Firstly, multi-layer perceptron (MLP), Convolutional Neural Networks (CNNs), and graph neural networks (GNNs) with similar architectures are constructed and compared to traditional Z-R relationships considering precipitation types. DL methods outperform traditional Z-R relationships, and GNNs perform the best. More importantly, this study provides a systematic evaluation of precipitation intensity, type, and organization, and the GNN achieves outstanding results. For extreme precipitation ($>$30 mm/h) and regional total rainfall, GNN achieves the smallest MAE. It also maintains low bias and dispersion for stratiform and organized cases. By contrast, CNN better handles isolated rainfall. Last but not least, the Z-R relationship shows systematic spatial biases, generally overestimating precipitation in plains and underestimating it in some inland high-altitude regions. DL methods alleviate these biases by incorporating spatial information. Overall, this study highlights the advantages of DL methods across different precipitation scenarios and demonstrates their ability to mitigate systematic biases from complex terrain.

While diurnal pulses (outward propagation) of clouds and precipitation in tropical cyclones (TCs) over the ocean have been well recognized, the diurnal propagation of precipitation (DPP) in landfalling TCs and its impacts on overland precipitation are less known. To fill this gap, this study utilizes the 30‐min Integrated Multi‐satellite Retrievals for GPM (IMERG) data set to identify overland DPP events in landfalling TCs. Overall, overland DPP events account for approximately 30% of all landfalling TC days globally. In these overland DPP events, a positive precipitation anomaly is observed to propagate radially outward from the inner‐core to the outer‐rainband region of TCs, with enhanced precipitation occurring along the propagation path. The propagation signal in overland DPP events initiates later (06–09 LST) than open‐ocean DPP events, which typically initiate at 03–06 LST. This difference is attributed to the presence of two distinct initiation times for the overland DPP signal: early morning (00–06 LST) and near noon (09–15 LST). The early morning initiation is primarily driven by convection‐induced inertial gravity waves, which result from the nocturnal long‐wave radiative cooling enhancement of inner‐core convection in strong TCs. In contrast, overland DPPs initiated near noon are typically produced by weak TCs where the inner‐core convection peaks at noon in response to the short‐wave radiative heating. As a result, TCs during overland DPP events tend to produce wider rainfall distribution and more extreme rainfall, even in landfalling weak TCs.

Extreme precipitation events (EPEs) without severe convective weather signatures such as frequent lightning may be more difficult to forecast and potentially more dangerous. This study investigates the differences in the macro‐ and micro‐structures between EPEs with and without lightning over eastern and southern China, as well as their underlying environmental conditions. EPEs are defined as convective features with maximum hourly rain rate reaching the gauge‐based climatological extreme precipitation threshold (99.9%). Results show that EPEs with lightning (EPE_LIG) account for 51%, and the other 49% EPEs have no lightning (EPE_NoLIG), whose fractional maxima are located along the coast. Most EPE_NoLIGs are embedded in large organized precipitation systems, although their convective cores are smaller than EPE_LIGs. The parent systems of both EPE types are multicellular in nature, but those of EPE_NoLIGs are more likely to produce multiple extreme precipitation centers. EPE_LIGs have the most intense convection, which is stronger than regular thunderstorms (NonEPEs with lightning), while the convective intensity of EPE_NoLIGs is just close to NonEPEs. Microphysical processes of the two types of EPEs differ significantly. The downward increasing radar reflectivity profiles and drop size distribution analyses suggest that warm‐rain processes highly dominate (94.1%) in the formation of extreme precipitation in EPE_NoLIGs. Even for EPEs with very active ice‐based processes (EPE_LIGs), warm‐rain processes still contribute significantly (83.31%). The large‐scale environments of EPE_NoLIGs are featured by a relatively convectively stable but deep moist troposphere and enhanced low‐level southwesterly winds, highlighting the importance of excessive moisture transport and convergence in these events.

Most tropical cyclones (TCs) landfalling Southern China originated from the Northwest Pacific (NWP) and tracked over the South China Sea (SCS) before landfall. The internal structures such as convective characteristics of the tropical cyclones may change as the TCs translate from the open ocean (NWP) to the enclosed sea (SCS) due likely to the impacts from nearby landmass. This study compares the lightning activity and convective structures, as well as the large-scale environments, of TCs over the NWP and SCS to better understand the structural changes and underlying physical mechanisms. It is interesting that TCs over SCS are much more electrically active than NWP TCs (especially in the outer rainbands), even though the NWP TCs precipitate heavier. Multi-satellite observations suggest that the NWP TCs have a deeper layer of ice particles, producing heavier surface rainfall; however, the SCS TCs own more large ice particles or supercool liquid in the mixed-phase region, which is essential for charge separation thus lightning production. It is surprising that the thermodynamic conditions (e.g., SST and atmospheric instability) of the NWP are more favorable for convective development than SCS. A few factors may contribute to higher lightning activity in SCS TCs, including stronger vertical wind shear, thinner warm cloud depth and higher aerosol optical depth, all may help to produce asymmetric intense convection and active mixed-phase processes. Furthermore, SCS TCs display a marked lightning maximum in the front quadrants of the moving direction, but NWP TCs are less so, likely because the thermodynamic and aerosol impacts from land are stronger in the SCS. Lightning in the SCS TCs is also more asymmetric relative to the vertical wind shear than the NWP TCs, which is featured by a maximum in the right of the downshear region of the outer rainband (opposite to the precipitation pattern).

Mesoscale convective processes and related dynamic/thermodynamic responses may play an important role in the Tropical Cyclone (TC) genesis, in addition to the favorable environmental conditions. The objective of this study is to determine whether and how unique are mesoscale organizations and convective properties of tropical disturbances prior to TC formation in the Northwest Pacific. Previous studies with a similar goal are based on either a small sample size or limited observational source (e.g., only Infrared). This study identifies over 3000 episodes of developing (Dev) and nondeveloping (Nondev) tropical disturbances and utilizes a large amount of multi-source satellite observations (precipitation, infrared, microwave, spaceborne radar, etc.) to comprehensively compare their convective structures. The Dev and Nondev disturbances were borne in similar large-scale environments such as SST, low-level vorticity, vertical wind shear, except that the Dev tropospheric conditions are slightly moister. However, the frequency, organization, intensities, and ensemble microphysics of the convection are significantly different between Dev (48-96h prior to TC formation) and Nondev. Both Dev and Nondev show very asymmetric distributions of convection with maxima in the down-shear quadrants, but Dev systems have greater areas of precipitation and cold clouds. The embedded individual convective systems of Dev are also more organized, i.e., greater areas and higher stratiform rain fraction. Furthermore, Dev convection is stronger and present greater ice-phase content as indicated by both the passive microwave and spaceborne measurements. Interestingly, Dev disturbances also have markedly higher frequency of shallow warm convection, especially in the up-shear regions, which may help moistening the lower-to-middle troposphere and beneficial for further deep convective developing. In most of the Dev storms, convection rapidly become more organized and deeper during 24-48 prior to the TC genesis. This study further compares the organization and convective properties among Dev systems generated under different types of large-scale flow pattern such as monsoon trough and easterly wave.

This study uses satellite passive microwave (PMW) and spaceborne precipitation radar (PR) observations to investigate changes in the cloud and precipitation structures of tropical cyclones (TCs) experiencing diurnal pulses (DPs). A total of 4677 PMW and 3074 PR snapshots are matched with DP and non‐DP events of TCs objectively identified from infrared satellite images. The PMW observations suggest that compared to non‐DP events, the inner‐core (0–100 km) cold clouds containing both ice and liquid particles become more frequent and widespread when DP occurs. The greatest DP‐associated enhancement of cold clouds occurs in the downshear left quadrant of weak TCs (tropical storms) and in the upshear region of moderate‐to‐strong (≥ CAT 1) hurricanes, respectively. Also, the inner‐core cold clouds of DP events still maintain a relatively symmetric distribution under high‐shear (>10 ms⁻¹) environments compared to non‐DP counterparts. The inner‐core warm cloud, dominated by liquid hydrometeors, increases in regions right of the shear but decreases left of the shear during DP events. Similarly, PR echo‐top height statistics show significant increases of moderate‐to‐deep convection in the TC inner region (0–300 km) on DP events especially for weak TCs. Particularly, the upshear‐right quadrant has the greatest increase in moderate‐to‐deep convection during DP events. These changes lead to a deeper and more symmetric convective structure of TCs' inner region with the occurrence of DPs.

This study investigates whether the initiation time and intensification rate (INTRATE) of intensifying tropical cyclones (TCs) vary diurnally and how they are related to deep convection. TC intensifying events are identified and classified into slowly intensifying (SI) and rapidly intensifying (RI) events. RI events last ∼42 hr on average, much longer than SI events. More importantly, the onset of the TC intensification, especially RI, markedly peaks at 00–06 local time. However, the INTRATEs of both RI and SI events show very weak diurnal variability. The INTRATE continues to increase after RI initiation and slightly peaks in the late afternoon (at 90% significance level). Inner‐core convection of all intensifying events maximizes in the early morning, in phase with the peak initiation time. In short, our results suggest that the nocturnally enhanced inner‐core convection may play a role in triggering TC intensification (e.g., RI), but not maximizing the INTRATE.

The radially outward propagating, cloud top cooling, diurnal pulse (DP) is a prominent feature in tropical cyclones (TCs) that has important implications for changes in TC structure and intensity. By using an objective identification algorithm, this study characterizes DPs over various ocean basins and examines their environmental conditions and convective structures. DPs occur on 52% of TC days globally and the occurrence frequency exhibits significant regional variability. The Northwest Pacific (NWP) has the highest DP frequency (60%) and shares the largest fraction of DPs worldwide (34%). The median duration and propagation distance of DPs are 12–15 hr and 500–600 km, respectively. Although the mean propagation speed of DPs is 11–13 m s⁻¹, persistent DPs (lasting >15 hr) mostly propagate at speeds similar to internal inertial gravity waves (5–10 m s⁻¹). Additionally, the longer the pulse duration, the stronger the pulse amplitude. Further, most DPs initiate in the inner core overnight, in phase with inner‐core deep convection. Inner‐core cold clouds, precipitation, and lightning are all markedly enhanced on DP days compared to non‐DP days. Interestingly, the DP signal significantly weakens and becomes slower while propagating through the 200–400‐km annulus during 09–12 local time (LT). Finally, DPs are more likely to occur over warm sea surface temperatures (SSTs), in low shear, and with a moist mid‐to‐upper troposphere. SST plays an important role in DP development over all basins, while shear and humidity are less important in the Northeast Pacific (NEP) and North Atlantic (NA) basins.

The large‐scale flows and rainfall pattern change markedly with the progression of the summer monsoon, but how the convective and microphysical characteristics of monsoon extreme precipitation vary subseasonally remains elusive. This study compares 16,844 extreme precipitation features (EPFs) among the pre‐monsoon, active‐monsoon, post‐monsoon periods, and tropical cyclone (TC) days using multisource data including dual‐polarimetric radar observations over the Pearl River Delta in South China. Results indicate that the environmental dynamics and thermodynamics significantly influence the EPFs' convective and microphysical properties. Despite the similar environmental conditions such as abundant moisture and moderate to large convective available potential energy (CAPE) under which the EPFs occur, differences among the four periods are noticed. During the pre‐monsoon period, precipitation systems are the largest in area but their EPFs are the least frequent and have the lowest raindrop concentration, likely due to the colder and drier environment with large vertical wind shear (VWS). The onset of the summer monsoon increases both the frequency and the convective intensity of EPFs, leading to an increase in raindrop size, which is consistent with the substantial increases of CAPE and moisture during the active‐monsoon period. The EPFs share similar convective intensity and raindrop size distribution (RSD) between the post‐monsoon and the active‐monsoon periods, although the post‐monsoon EPFs are slightly less frequent and have a smaller horizontal scale related to the reduced 0–6‐km VWS. Interestingly, EPFs associated with TCs have the weakest convective intensity but the most active warm‐rain processes with the RSD being closer to the maritime regime.