Spatiotemporal Patterns of Typhoon Rainfall in Taiwan Associated with Track Clusters

Typhoon rainfall is one of the primary meteorological hazards in Taiwan, with its spatial distribution strongly modulated by storm tracks and complex topography. In this study, landfalling typhoons affecting Taiwan from 1950 to 2024 are analyzed using gridded precipitation data. Hierarchical clustering is first applied to typhoon tracks to derive interpretable track types, with cluster numbers objectively determined from within-cluster variance. For each track type, a data-driven pattern extraction framework is subsequently applied to the corresponding rainfall fields, enabling the identification of dominant spatial features and representative canonical rainfall patterns. We further focus on typhoon events associated with pronounced rainfall impacts, systematically examining the correspondence between their rainfall characteristics and the identified canonical patterns, and quantitatively assessing the relationship between track clusters and high-risk rainfall spatial features. In addition, detailed grid-point analyses are conducted for extreme rainfall cases, including Typhoons Gaemi (2024) and Danas (2025), to evaluate whether their rainfall spatial distributions conform to the identified canonical patterns. By quantifying the linkage between typhoon track clusters and rainfall spatial patterns, this study provides a physically grounded reference framework for subseasonal-to-seasonal typhoon rainfall prediction and scenario-oriented analysis.