應用最小二乘技術於河川水文因子動態分析

Dynamic Analysis of Hydrology Factors Based on Least-Squares Technique

當豪雨及颱風過後，洪氾淹水、邊坡崩塌等災害遍佈於台灣山林集水區之中，相關水文觀測資訊或流量水位觀測站，易受到災害影響而損毀、缺測，造成相關洪水預報、淹水模擬分析或水工構造設計無法獲得穩定的實際觀測資料進行驗證，故如何利用較穩定之水文觀測資訊，有效建立相互轉換關係模型進行推演，是目前資料補遺與水文監測迫切所需。其中，雨量資料為水文觀測資料中較穩定之觀測，亦能直接反映降水於集水區之量體多寡，因此可作為模式中可靠的觀測資料來源。本研究利用集水區中多個雨量站觀測資料，依徐昇多邊形法進行雨量觀測加權計算出降水水體量，再由降水水體量與流量關係建構水文因子轉換模型，最後以平差技術擬合計算出水文因子之參數值及其代表特性。研究成果顯示，利用雨量觀測搭配水體轉換流量模型，不僅可以由平差擬合方式得到模式內相關水文參數，也可適度修正流量觀測資訊，因此可提供集水區長期水文資料補遺與監測等相關分析之建立來源。 After torrential rains and typhoons, flood flooding, landslides and other disasters have frequently damaged the montane basin in Taiwan. Moreover, those disasters usually damaged hydrological stations. As a result, one cannot obtain sufficient observations to support predication, flood simulation analysis or hydraulic structure design. Therefore, how to effectively establish the conversion relationship model by steady hydrological observation information is an essential requirement in data addendums and hydrological monitoring. Particularly, rainfall data is not only regular observation, it also directly reflects the amount of precipitation in a catchment providing are liable observational source for model development mode. In this study, the precipitation observations from multiple rainfall stations were collected and the total amount of precipitation was estimated based on the Thiessen polygons method. Next, a hydrological model describing the relation between the precipitation and discharge was established and associated factors were then estimated by the least-squares approach. The test results revealed that the conversion of water flow model not only provides the relevant hydrological parameters by the least-squares approach the flow observation information can also be corrected accordingly. In other words, the proposed approach together with the rainfall information can facilitate a long-term data addendum and dynamic monitoring for the hydrological factors in a catchment.

After torrential rains and typhoons, flood flooding, landslides and other disasters have frequently damaged the montane basin in Taiwan. Moreover, those disasters usually damaged hydrological stations. As a result, one cannot obtain sufficient observations to support predication, flood simulation analysis or hydraulic structure design. Therefore, how to effectively establish the conversion relationship model by steady hydrological observation information is an essential requirement in data addendums and hydrological monitoring. Particularly, rainfall data is not only regular observation, it also directly reflects the amount of precipitation in a catchment providing are liable observational source for model development mode. In this study, the precipitation observations from multiple rainfall stations were collected and the total amount of precipitation was estimated based on the Thiessen polygons method. Next, a hydrological model describing the relation between the precipitation and discharge was established and associated factors were then estimated by the least-squares approach. The test results revealed that the conversion of water flow model not only provides the relevant hydrological parameters by the least-squares approach the flow observation information can also be corrected accordingly. In other words, the proposed approach together with the rainfall information can facilitate a long-term data addendum and dynamic monitoring for the hydrological factors in a catchment.