雨型為降雨在某特定延時的時間之分佈，為從事水文分析與設計之重要參數。總降雨量相同但其時間上之分佈可能不同，以致其尖峰降雨對應之尖峰流量即有可能改變，如何由集水區雨量測站中選用適合之雨型為重要之課題。本研究使用南投縣國立臺灣大學實驗林管理處轄區鄰近之11個測站1960至2013年延時大於24小時之降雨事件，以尺度不變高斯馬可夫（Scale-Invirant Gaussian Markov）雨型進行分析，並與平均法進行比較外，另因應氣候變遷可能形成降雨型態之改變，將分析事件分為1960至1990年、1991至1999年與2000年以後三時段進行分析。研究結果顯示11測站以24等分點之雨型有所差異，其中發現尖峰降雨等分點所佔百分比例有下降之趨勢，1991年至1999年之尖峰降雨等分點有延後之趨勢。本研究結果顯示水文演算採用雨型時，必需考量該測站雨型變化之趨勢，以免採用過於保守或不足之設計標準，維護相關水工結構物之壽限與安全。
Rainfall hyetograph, temporal distribution at a certain period, is an essential parameter for hydrologic design and analysis. Hyetograph may be quite different even though total rainfall amount is the same. For hydrologic routing, corresponding peak discharge with respect to the peak of hyetograph may be different; thus, selection of suitable rainfall patterns hydrologic design become an important issue for watershed management. This study analyzed rainfall events having duration longer than 24 hours recorded by 11 stations in the National Taiwan University Experimental Forest (NTUEF) from 1960 to 2013 using Scale-Invariant Gaussian Markov (SIGM) hyetograph and comparing it with Averaging Method. To assess possible climate impact on hyetograph, we grouped rainfall events into three time intervals-between 1960 and 1990, between 1991 and 1999, and after 2000. The results showed that (1) rainfall hyetograph among the eleven stations was different, (2) the proportion of peak hyetograph was decreasing, and (3) the interval of peak hyetograph was delayed for the period from 1991 to 1999. It suggested that variable trend in hyetograph should be considered for hydrologic routing to prevent overestimating or underestimating standard level so that safety and life of hydrologic structures can be maintained.
