半分布並聯型線性水庫降雨-逕流模式結合率定曲線於洪水位預報之研究

Applications of the Semi-distributed Parallel-type Linear Reservoir Runoff Model and Rating Curve on Flood Stage Forecasting

陳毅學(Yi-Shiue Chen)

指導教授 ： 朱子偉 ； 謝龍生

國立臺北科技大學/工程學院/土木與防災研究所/碩士(2012年)

摘要

台灣位於副熱帶季風區，每年因梅雨及颱風侵襲常帶來豪雨，加上地形陡峻，河川短促，經常引發嚴重水災，因此若能即時提供可靠的洪水資訊則將有助於災害應變單位擁有更充足的應變時間，以使洪災損失降至最小。故本研究之主要目的係藉由降雨-逕流模式與即時更新逕流量-水位率定曲線建立一套快速河川洪水位預報方法。
研究中首先將定量估計降水及預報資料，輸入半分布並聯型線性水庫降雨-逕流模式模擬及預報逕流量，其次由模擬之逕流量與觀測水位以迴歸方法建立線性方程式、二次多項式及冪次型三種不同的率定曲線關係式，最後則透過率定曲線關係式將預報逕流量轉換成洪水位預報，同時經由即時觀測水位資訊的更新，對率定曲線關係式進行即時修正。
本研究以蘭陽溪流域為研究區域，並挑選辛樂克(Sinlaku)颱風事件及梅姬(Megi)颱風事件於蘭陽大橋進行水位預報，預報結果顯示冪次型率定曲線關係式於蘭陽大橋預報1-3小時洪水位有較佳之結果，辛樂克颱風與梅姬颱風之平均相對誤差(RAE)及均方根誤差(RMSE)平均值分別依序為6.06 %、0.38 m與6.87%、0.42 m。研究中又利用比值法進行水位預報誤差修正，經修正後辛樂克颱風與梅姬颱風之平均相對誤差(RAE)及均方根誤差(RMSE)平均值分別依序為3.96 %、0.27 m與3.35 %、0.22 m。經由誤差修正後之預報結果可有效提升預報的準確性且更趨近於觀測水位，故本研究所建立之洪水位預報模式，將可提供應變單位決策時之參考依據。

關鍵字

洪水位預報 ； 半分布並聯型線性水庫降雨-逕流模式 ； 率定曲線

並列摘要

Taiwan is threatened by disastrous storms that result from subtropical cyclones or typhoons every year. The torrential rainfall of storms which combined the geographical features such as steep land slope and short rivers course often result in disastrous damages per year. Therefore, if readily available and reliable information will help the flood management agency will take proper for response and disaster mitigation before the disaster happens. The main purpose of this study is by rainfall-runoff model and real-time updates of runoff-stage rating curve to the establishment of a fast flood stage forecasting methods.
In this study, the quantitative precipitation estimation and forecast were used as input data to the Semi-distributed Parallel-type Linear Reservoir Runoff Model for streamflow simulation and forecast. Next, three types of rating curve obtained by regression model in linear, quadratic equations and power law forms developed based on the runoff simulation and flood stage. Finally, the Forecast runoff converted to the flood stage prediction by rating curve relationship, in the meantime, immediate modified the rating curve by update information via real-time observation flood stage.
In this study, selection of typhoon Sinlaku and Megi events to proceed flood stage forecast in the Lanyang Bridge at Lanyang River Basin. The results show that using power law rating curve in 1-3hrs of flood stage forecasting the better results at the Lan-Yang Bridge. Sinlaku and Megi typhoon event RAE and RMSE are 6.06%, 0.38 m and 6.87%, 0.42 m. respectively. Also use the ratio method to modify the error of flood stage forecasting, Sinlaku and Megi typhoon event RAE and RMSE are 3.96%, 0.27 m and 3.35%, 0.22 m. respectively. Through the forecast results after the error correction can effectively improve the forecasting accuracy is close to the observation flood stage, so the flood forecasting model established in this study will provide the response units and decision-making reference.

並列關鍵字

Flood Stage Forecasting ； Semi-distributed Parallel-type Linear Reservoir Rainfall-Runoff Model ； Rating Curve

參考文獻

15.吳建均，河川洪水位演算模式之研究，碩士論文，國立臺灣大學生物環境系統工程學系暨研究所，台北，2005。

32.蔡孟原，雷達定量降水估計應用在河川洪水預報之研究，碩士論文，國立臺灣大學生物環境系統工程學系暨研究所，台北，2009。

37.戴嘉慧，氣候變遷對翡翠水庫供水、發電與防洪功能之衝擊評估，碩士論文，國立臺灣大學生物環境系統工程學系暨研究所，台北，2010。

40鍾世豐，類神經網路在洪水演算及預報之應用，碩士論文，國立臺灣大學生

1.Booij, M.J., " Impact of climate change on river flooding