台灣於夏季屬熱帶氣旋活躍之地區，每年會有多個颱風入侵台灣，尤其在2009年莫拉克颱風為台灣創下重大的災害。鑒於颱風對台灣造成的災損如此嚴重，因此實有必要對颱風的特性作進一步的探討。本研究為了解颱風降雨時空分布之特性，採用具有處理訊號非線性、非穩態性及完全自適性的希爾伯特－黃轉換(HHT)過程中之經驗模態分解(EMD)方法，搭配在大氣科學上常用之經驗正交函數(EOF)分析方法，來處理降雨在時間及空間上的變異性，並以荖濃溪流域之颱風降雨為例，探討其時空分布特性。 本研究根據莫拉克等十個颱風在荖濃溪流域之降雨特性分析結果顯示，在時間變動所展現之特性方面，與EOF相應之伸展係數可充分反映颱風雨量之時間變化趨勢；而在空間模態所展現之特性方面，透過前三個正交之空間函數(即EOF)在空間上的變異性分布及趨勢，與土石流潛勢溪流及坡地崩塌潛勢圖之套疊結果，可鑑別及確認該次颱風降雨在哪些地方屬於高災害潛勢區域。此分析成果可以當作一種防災指標，據以作為擬定山崩、地滑與土石流等災害之防災策略的參考。 Taiwan is among the most active tropical cyclone regions in the world. There are several typhoons hit Taiwan every year. Especially in 2009, typhoon Morakot brought a significant disaster in Taiwan. Due to the serious damages caused by typhoons, it is necessary to explore the characteristics of typhoons. In order to understand the characteristics of rainfall of typhoon, the method of empirical mode decomposition (EMD) in Hilbert-Huang transform (HHT), able to consider the non-linear and non-stationary nature of signal, is adopted in this study. On the other hand, the method of empirical orthogonal function (EOF) analysis, usually used in atmospheric sciences, is also employed to analyze the spatiotemporal characteristics of typhoon rainfall. In addition, the data of Laonong River Basin is taken for example to investigate the temporal and spatial distribution of typhoon rainfall. According to the analysis results of 10 typhoons in Laonong River Basin, it shows that the expansion coefficients associated with the EOF can reflect the variation of typhoon rainfall with time appropriately. In addition, the spatial distribution of rainfall can be depicted clearly by the first few spatial modes of EOF. By overlapping the map of EOF isopleth and the map of potential debris flow torrent, or the landslide potential map, we can easily identify where the high-risk hazard areas are. The results in this study can be used as an indicator for risk management and as a tool to develop practical strategies for disaster prevention, such as mitigation of landslides and debris flow.