台灣地區年最大一日雨量區域頻率分析之研究：(Ⅰ)理論部分

Study on Regional Frequency Analysis for Annual Maximum Daily Rainfall in Taiwan

陳儒賢(Lu-Hsien Chen) ； 陳清田(Yu-Ting Hong) ； 洪毓婷(Ching-Tien Chen)

《中華水土保持學報》 41卷1期 (2010/03) Pp. 61-78

https://doi.org/10.29417/JCSWC.201003_41(1).0005

摘要

本研究主要目的是應用主成分分析（principal component analysis; 簡稱PCA）、自組織映射圖（self-organizing map; 簡稱SOM）網路及線性動差(L-moment)進行年最大一日雨量之區域頻率分析(regional frequency analysis)。首先，本研究應用PCA擷取最大一日降雨量資料之各項主要成分，並將主成分與雨量測站之地文因子作爲SOM網路之輸入項，根據SOM網路之二維映射圖可目測得知雨量測站可被分群之數目，進而可劃分成數個均一性區域。再者，經由線性動差之不一致估量及異質性估量，可評估同一區域內之測站是否具一致性及均一性。最後，再以適合度估量選取各均一性區域內之最佳區域機率分佈函數，即可推估各均一性區域於不同重現期距下之頻率降雨量。關於區域頻率分析之實際應用例，將於下篇文章敘述之。

關鍵字

區域頻率分析 ； 主成分分析 ； 自組織映射圖 ； 線性動差

並列摘要

The purpose of the study aims to investigate the regional frequency for annual maximum daily rainfall using principal component analysis (PCA), self-organizing map (SOM) and L-moment. First, PCA is applied to obtain the principal components from the annual maximum daily rainfall data. Based on the transformed data resulting from PCA and the geographic characters of the gauges, the SOM is used to group the rain gauges into specific clusters. According to two-dimensional feature map, one can determine the number of clusters of each cluster directly by eyes, and can then delineate the homogeneous regions for regional frequency analysis. Moreover, the L-moment based discordancy and heterogeneity measures are used to test whether regions may be acceptable as being homogeneous. Finally, goodness-of-fit measure is applied to select the regional probability distributions of rainfalls, and then the amounts of daily rainfall with various return periods for each region are estimated. The application to regional frequency analysis is stated in the following paper.

並列關鍵字

regional frequency analysis ； principal component analysis ； self-organizing map ； L-moment