The increasing impact of tropical cyclone (TC) rainfall underscores the need to understand regional variations in its spatial distribution. Using high-resolution satellite precipitation data from 1998–2023, this study investigates changes in the spatial inhomogeneity of TC rainfall across China. Results show that TC rainfall inhomogeneity decreases significantly by about 65%, indicating a reduced rainfall in historically high-TC-rainfall areas and an increase in low-TC-rainfall areas. Regionally, this decline is mainly governed by the intra‐regional component in East China, where opposing trends between high- and low-rainfall areas have substantially reduced spatial disparities. In contrast, North China shows a marked increase in TC rainfall, accompanied by increased intra‐regional but decreased inter‐regional inhomogeneity, largely offsetting each other. Other regions exhibit minor or insignificant contributions. These spatial reorganizations of TC rainfall in East and North China are closely linked to the northward migration of TC activity in recent decades, driven by reduced landfalling TCs over Taiwan and Fujian and increased inland penetration of TCs into East and North China. The findings reveal a restructuring of TC rainfall patterns and emphasize the growing regional contrasts in TC‐related hydroclimatic impacts, providing new insights for disaster risk assessments and regional climate adaptation strategies in China.
24 May 2026 in Journal of Geophysical Research: Machine Learning and Computation
Accurate tropical cyclone (TC) intensity (TCI) prediction is critical for effective disaster preparedness. Although machine learning‐based weather prediction models demonstrate strong performance in simulating TC tracks, their TCI predictions show systematic biases. To overcome this limitation, this study proposes TC‐FANet, a deep learning framework built upon a temporal convolutional network (TCN) backbone, incorporating a feature‐aware self‐attention (FSA) mechanism that dynamically reweights atmospheric features according to TCI changes. TC‐FANet achieves an overall mean MAE of 5.90 m s −1 across all lead times (24–120 hr), representing a 46.02% reduction compared with all evaluated baseline categories, and delivers a 40.23% improvement over deep learning baselines during rapid intensification (RI) phases. SHAP analyses revealed that the FSA module consistently elevates RI‐critical controls (tropical cyclone fullness) while down‐weighting redundant or persistence‐like signals, revealing nonlinear, interaction‐driven determinants of TCI. These findings highlight the potential of DL architectures to provide interpretable insights for operational TCI forecasting.
The increasing impact of tropical cyclone (TC) rainfall underscores the need to understand regional variations in its spatial distribution. Using high-resolution satellite precipitation data from 1998-2023, this study investigates changes in the spatial inhomogeneity of TC rainfall across China. Results show that TC rainfall inhomogeneity decreases significantly by about 65%, indicating a reduced rainfall in historically high-TC-rainfall areas and an increase in low-TC-rainfall areas. Regionally, this decline is mainly governed by the intra‐regional component in East China, where opposing trends between high- and low-rainfall areas have substantially reduced spatial disparities. In contrast, North China shows a marked increase in TC rainfall, accompanied by increased intra‐regional but decreased inter‐regional inhomogeneity, largely offsetting each other. Other regions exhibit minor or insignificant contributions. These spatial reorganizations of TC rainfall in East and North China are closely linked to the northward migration of TC activity in recent decades, driven by reduced landfalling TCs over Taiwan and Fujian and increased inland penetration of TCs into East and North China. The findings reveal a restructuring of TC rainfall patterns and emphasize the growing regional contrasts in TC‐related hydroclimatic impacts, providing new insights for disaster risk assessments and regional climate adaptation strategies in China.
East China is a global hotspot for landfalling tropical cyclones (TCs). While previous studies have extensively examined the influence of El Niño–Southern Oscillation (ENSO) on TC activity, whether and how ENSO alters the spatial organization of precipitation in landfalling TCs remains insufficiently understood. Using high-resolution satellite precipitation observations and ERA5 reanalysis data, this study investigates the heterogeneous responses of landfalling TC precipitation over East China to different ENSO phases. Composite analyses indicate ENSO-associated differences in TC precipitation at both regional and storm-relative scales. At the regional scale, positive composite precipitation differences occur over Fujian, southern Zhejiang, and Taiwan during El Niño years, whereas negative differences occur in parts of northern East China relative to La Niña years. At the storm-relative scale, mean outer-rainband precipitation was estimated to be 26.02% higher during El Niño years; however, the storm-level bootstrap 95% confidence interval included zero, indicating uncertainty associated with the limited number of independent TC events. Precipitation during La Niña years was relatively more concentrated within the inner-core region. Mechanistic diagnostics show that the El Niño composite is associated with differences in thermodynamic and dynamic conditions, including higher low-level cyclonic vorticity, upper-level divergence, ascent, and anomalous steering flows over southern East China and adjacent seas. TC-centric analysis further indicates that higher moisture supply and dynamic lifting coincide with the outer-rainband precipitation contrast during El Niño years. These findings suggest that ENSO is associated with spatially heterogeneous and structurally distinct responses of landfalling TC precipitation over East China, providing a diagnostic basis for understanding ENSO-related differences in regional TC precipitation risk.
South China is one of the most frequently affected regions by landfalling tropical cyclones (TCs) worldwide. Owing to its complex topography and environmental conditions, TC precipitation in this region typically exhibits pronounced asymmetry, with substantially stronger rainfall on the southern‐side of the TC center. Here, we show that precipitation asymmetry has decreased by 29% during 1998–2024. Specifically, a significant increase in the area‐averaged precipitation rate is observed on the northern‐side of TC center, accompanied by a decrease on the southern‐side. Diagnostic analyses indicate that this change is closely associated with the weakening of the vertical wind shear (VWS) at an average rate of 0.046 m s −1 per year, along with concurrent changes in upper‐ and lower‐level divergence and vertical motion. These dynamical changes collectively contribute to enhanced precipitation on the northern‐side of TC center, suggesting an increasing potential risk of rainfall impacts in this region.
The South China Sea (SCS) is one of the most active marginal seas for tropical cyclones (TCs) worldwide and a key region for TC hazards along the coasts of southern China, Philippines and Vietnam. However, the drivers of long-term changes in intense TC development in this region remain unclear. Here, we show that the number of TCs reaching Category 3-5 intensity has increased significantly, with more occurring in the northern SCS. This change is primarily linked to an increasing contribution of western North Pacific TCs entering through the Luzon Strait, which experience weaker land-induced weakening than those crossing the Philippine archipelago and retain greater intensity for subsequent development. After entering the SCS, reduced ocean cooling and enhanced oceanic and atmospheric conditions have increased the environmental favorability for intense TC development in the northern SCS. These results highlight the role of regional ocean-land-atmosphere interactions in reshaping where TCs reach high intensities.