首页|期刊导航|Tropical Cyclone Research and Review|AI-Enhanced Detection and Comparative Analysis of Tropical Cyclones Across Global Ocean Basins

AI-Enhanced Detection and Comparative Analysis of Tropical Cyclones Across Global Ocean BasinsOA

中文摘要

This study develops a deep-learning framework for detecting tropical cyclone(TC)life-cycle stages from infrared satellite imagery across six global ocean basins.You Only Look Once(YOLO)v8 models were trained separately to detect genesis,intensification,and decay stages within individual basins as well as in generalized settings.Detection accuracy was consistently highest during intensification,with F1 scores exceeding 0.80 in several basins,while genesis and decay remained more challenging due to weaker or more disorganized storm structures.Cross-basin experiments revealed substantial performance degradation when Northern Hemisphere models were applied to Southern Hemisphere storms,highlighting the influence of hemispheric rotation and storm asymmetry.Rotation-based augmentation(180°rotation via image flipping)improved detection in the South Pacific and South Indian basins,particularly for genesis and intensification.Nonetheless,cross-basin trans-ferability was constrained even within the same hemisphere,with notable weaknesses in the North Atlantic and variability across other basins.To interpret these disparities,three structural indices were evaluated;symmetry,texture,and shape complexity and Detected versus Non-detected distributions were statistically tested.Detected cases were generally associated with stronger texture and more symmetric structure,whereas shape-complexity contrasts were more basin-and stage-dependent.Spatial localization was evaluated using a YOLO box-midpoint center proxy,which can be biased for asymmetric systems.Overall,the results indicate that detection performance depends on life cycle stage,hemispheric orientation,and storm structural organization,underscoring the need for regionally adapted,structure-aware-specific generalization strategies for robust global TC detection.

G.M.W.L.Gunawardena;Hironori Fudeyasu;Daisuke Matsuoka;Ryuji Yoshida;Nobuhiro Yugami;Yuiko Ohta;Takashi Yanase

Department of Town and Country Planning,Faculty of Architecture,University of Moratuwa,Sri Lanka Typhoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,JapanTyphoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,JapanTyphoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,Japan Data Science Research Program,Research Institute for Earth and Information Sciences,Japan Agency for Marine-Earth Science and Technology(JAMSTEC),JapanTyphoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,Japan Graduate School of Environment and Information Sciences,Yokohama National University,Japan College of Urban Sciences,Yokohama National University,JapanTyphoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,Japan Artificial Intelligence Laboratory,Fujitsu Research,Fujitsu Limited,JapanTyphoon Science and Technology Research Center,Institute for Multidisciplinary Sciences,Yokohama National University,Japan Artificial Intelligence Laboratory,Fujitsu Research,Fujitsu Limited,JapanSpace Data Frontiers Research Center,Fujitsu Research,Fujitsu Limited,Japan

海洋科学

Deep learningInfrared satellite imageryLife-cycle detectionStructural analysisTropical cyclonesYOLO v8

《Tropical Cyclone Research and Review》 2026 (2)

P.274-295,22

supported by the collaborative efforts of the Fujitsu-Yokohama National University Typhoon Science and Technology Research Center Collaborative Research Laboratory,Japanpartially supported by JSPS KAKENHI Grant Number JP22H01316.

10.1016/j.tcrr.2026.04.007

评论