| Abstract: | 本文之研究目的在建立台灣地區北部、中部、南部及東部區域之年最大洪水頻率分佈，並將之應用於未設站之地區以推估其不同頻率之洪水。本文所用於區域洪水頻率分析的方法為指數洪水法並配合以線性動差法推估參數，包括以線性動差為基礎之不一致（discordancy）、異質性（heterogeneity）及適合度（goodness-of-fit）估量衡量同一區域內不同站之年最大洪水是否具有一致性及均勻性，並選取一最佳之區域洪水頻率模式。本文以台灣地區記錄年限超過18年且完整未有間斷之62個流量站的日流量紀錄為基礎，經選取年最大日流量分區建立區域年最大洪水頻率分佈，在北、中、南、東各區域之最佳無因次年最大洪水機率分佈分別為通用極端值分佈（generalized extreme-value distribution, GEV）、通用帕雷托分佈（generalized Pareto distribution, GPA）、皮爾遜第Ⅲ型分佈（Pearson type Ⅲ distribution, PE3）及皮爾遜第Ⅲ型分佈（Pearson type Ⅲ distribution, PE3）。之後即可推估各區域各站發生不同迴歸期之年最大洪水，各站歷史洪水事件之發生頻率亦可據以推估。至於未設站地點之洪水頻率分析，本文則建議先建立區域內具流量記錄之各站集水面積及平均年最大洪水之關係，即利用未設站地點之集水面積推估該地之年平均最大洪水，再利用指數洪水法之概念即可推估該區域未設站之洪水頻率。
This study aims to establish the annual maximum flood frequency distributions for the North, Central, South, and East regions in Taiwan, and apply the resulting regional models to estimate flood frequencies of the ungauged sites. The methodology adopted in this study is the index flood method associated with the L-moments method used to estimate parameters of frequency distributions. The L-moment based discordancy, heterogeneity, and goodness-of-fit measures are employed to detect discordancy and heterogeneity of flood data and to select the best data-fitted regional flood frequency distributions. The daily flow data with exceeding 18-year record are the bases for regional flood frequency analysis. The results show that the best annual maximum flood frequency distributions for North, Central, South, and East regions are generalized extreme-value distribution(GEV), generalized Pareto distribution(GPA), Pearson type III distribution(PE3), and Pearson type III distribution(PE3), respectively. Frequencies of some historical flood events for each region are also explored. Since high correlation existed between mean annual maximum flood and the drainage area, the mean annual maximum flood of the ungauged sites can be estimated by the drainage area for each region. Associated with the derived regional flood frequency models, the estimation of flood frequencies for the ungauged sites is made possible. |
