區域化貯蓄函數法之發展及其不確定性分析

Abstract

本研究利用優選方法求出單場及多場暴雨之參數最佳值，依據雨型集群分析的結果，運用多變量迴歸方法，建立貯蓄函數法中相關參數之區域化公式，以利於無水文站或資料短缺之集水區推估逕流量之用。此外，亦運用哈爾點估計法（Harr□s Point Estimation Method）及蒙地卡羅法（Monte Carlo Simulation）分析參數之不確定性。 本研究以高屏溪6個流量站以及基隆河2個流量站為應用案例。由驗證及其參數不確定性分析結果中可知，該貯蓄函數法參數之區域化公式可應用於未設站之流域。此外，雨型分群之區域化公式所得逕流歷線之精確性較不分群為佳。

In this study, the parameters of the storage function for the single and multiple storms are optimized first. Then, based on the cluster analysis on storm patterns, the regional equations for the parameters of storage function are developed by using the multivariate regression technique. It is expected that the proposed regional equations for the parameters of storage function can be used to estimate the discharge in ungauged watershed. Furthermore, the uncertainty of the parameters of storage function is assessed by Harr’s Point Estimation Method and Monte Carlo simulation. As an example case, six flow stations in Kaoping Rier catchment and two flow stations in Keelung River are selected in developing and assessing the regional equations. From the results of verification and uncertainty analysis, the developed regional equations are able to appropriately apply in ungauged watershed. In addition, it is found that discharges estimated by regional equations with rainfall patterns cluster are more precise than those without clustering.