區域化流量延時曲線及其不確定性研究

Abstract

流量延時曲線(FDC)為一累積頻率曲線，其原理雖簡單，但實務上卻具重要性。由於河川流量本身具隨機性，且其分析結果常受紀錄長短之影響，而因所選用日、旬或月流量之不同，均會使FDC具不確定性。傳統上習慣將其製作多年一條之曲線(n-FDC)，本研究則建立每年一條之流量延時曲線，即以n條FDC來代替一條n-FDC，據以探討其統計特性，並透過群集分析針對FDC之特性劃分水文均勻區。在水文工程實務上，常常面臨集水區缺乏逕流資料，或是紀錄年期過短之問題，致使無法推求未設站或記錄過短處之流量延時曲線，於此情況下，進行水文區域化分析，遂為常用的解決方法。一般來說，迴歸分析為最被廣泛使用者, 而本研究則分別以回歸分析和平均無因次流量延時曲線來代表該均一群區的流量延時特性，進而發展出各水文均勻區之區域流量延時曲線，並進一步利用自迴歸積分移動平均（ARIMA）模式，推估後龍溪流域未來年之FDC。

Although it has simple theoretical basis, the flow duration curve (FDC) is important in many practical engineering applications. Because of the inherent natural randomness of stream flows, traditional end-of-period FDCs show variation when the selected record lengths are different. It is used to produce a multi-year curve in tradition; however, it is to product annual flow duration curve in this work. In other words, the n-yr-FDC curve is substituted by n annual FDC curves to discuss the statistical characteristics and to divide ally homegeneous hydrologic areas by the cluster analysis. In hydrologic engineering practice, one usually faces the problem of lacking flow discharge data or short record period. Hence, the FDC at the location of interest cannot be derived. Under such circumstances, the regional hydrologic analysis then becomes the commonly adopted methodology. In general, regression analysis is the most common used approach in the regional analysis. In this study, the regional FDCs in each homegenous hydrologic area are derived with their characteristics spcified by the averaged dimensionless form of the FDCs. Furthermone, the ARIMA(Autoregressive Integated Moving Average) model is adopted to estimate future FDCs of the Ho-Long Creek basin.