標題: 氣候變遷下水資源長期調配與高濁度短期操作風險分析
The Risk Analysis of Long-Term and Short-Term Water Supply Considering Climate Change
作者: 何智超
Ho, Chih-Chao
張良正
Chang, Liang-Cheng
土木工程學系
關鍵字: 全球氣候變遷;蒙地卡羅分析;條件氣候繁衍降尺度模式;拉丁超立方取樣法;單位特性歷線;Global Climate Change;Monte Carlo Simulation;Conditional Weather Generation Downscaling Model;Latin Hypercube Sampling;Unit Characteristic Hydrograph
公開日期: 2009
摘要: 在全球氣候變遷影響下,台灣地區發生乾旱和洪水等極端水文事件機率較以往大幅增加,加上因地震頻繁造成多數集水區上游土石鬆軟,汛期時易發生高濁度供水問題,然而以往水資源供水分析多將枯水期供水問題和高濁度供水問題個別考量,且工程實務上多採用定率分析。但實際上在臺灣地區,因上述水文及地質之特殊情況,必需同時考量枯水期水源不足及汛期高濁度水質不良等所引起之缺水問題。除此之外,降雨量本身具有高度之不確定性,而氣候變遷影響將更增加此不確性,因此缺水問題本質上必須考量因此不確定所帶來的風險問題。故本研究以蒙地卡羅分析方法為基礎,發展考量包括區域水資源長期調配與高濁度缺水緊急應變之整體供水風險分析方法,並以石門水庫供水區為應用實例。 蒙地卡羅分析需要大量資料進行重複模擬,本研究以自行發展之條件氣候繁衍降尺度模式合成多組長期雨量,再以設計降雨概念配合拉丁超立方取樣法(LHS)合成多組高濁度雨量事件,接著透過降雨逕流模式計算各組雨量對應之逕流量,經供水調配模式分析後求得各組流量下之缺水率,並以韋伯點繪法統計缺水風險。高濁度供水調配部分,本研究發展單位特性歷線濁度推估模式,配合波爾斯法進行洪水操作,建立考量濁度及水庫防洪操作之供水調配模式。 本研究分析結果乃基於ECHAM5大氣環流模式之預測雨量,根據石門水庫地區缺水風險分析結果顯示,枯水期缺水風險在氣候變遷影響下有降低之趨勢,但氣候變遷卻造成高濁度缺水風險提高,在綜合考量下,石門水庫整年度缺水風險在氣候變遷影響下有稍為降低之趨勢。此外,由模式驗證結果顯示,氣候繁衍降尺度模式能透過大尺度氣候因子有效反映石門水庫集水區降雨量長期變化趨勢,單位特性曲線法可透過水庫入流量與出流量有效預測暴雨事件之濁度變化,顯示本研究提出之風險分析架構與評估方法,除兼顧整體供水考量外,模式預測亦有一定之可靠度,可推廣應用於其他地區。
Under global climate changes, extreme hydrological events, such as drought and flood, occurred more frequently in Taiwan, which both events can induce water supply problems. During flood events, a flood carries loose stones to a reservoir and increase water turbidity. The high turbidity then reduces the water treatment efficiency and causes water deficit problems. Moreover, the uncertainties of hydrological conditions make water allocation more difficult. It is essentially a stochastic problem and both the drought and flood conditions should be considered simultaneously. However, most of the previous studies investigate the water deficit problems caused by those two conditions separately by applying deterministic analysis only. Hence, this study uses the Monte Carlo Simulation to analyze water deficit problem considering both the drought and flood situations. The proposed method was applied to Shihmen Reservoir to demonstrate its feasibility. Monte Carlo Simulation requires large number of data for repeated analysis. This study proposes Conditional Weather Generation Downscaling Model (CWGDM) to synthesize long-term rainfall data and uses Latin Hypercube Sampling (LHS) to generate short-term (flood event) rainfall data. The rainfall data are transformed into runoff data by rainfall-runoff model, those runoff data are the input of long-term and short-term water allocation models. The short-term water allocation model must consider the high water turbidity treatment and reservoir flood mitigation operation simultaneously. The study proposes Unit Characteristic Hydrograph (UCH) to predict reservoir turbidity and uses Puls Mehtod to estimate the reservoir overflow. Through simulation of water allocation models, we can obtain serial water shortage data that can be used to estimate the water deficit risk by Weibull Plotting Method. The simulation results are based on the predictive rainfall data by ECHAM5. The risk analysis results indicate that climate changes decrease the water deficit risk caused by drought condition but increase the risk caused by high turbidity. In summary, the annual water deficit risk has decreased slightly under climate change. The rainfall synthesis results demonstrate the proposed CWGDM model can well predict the long-term rainfall. The turbidity simulation results demonstrate that the UCH well predict reservoir turbidity based on reservoir inflow and outflow. Although the study is applied to Shihmen Reservoir, the proposed concepts and procedure can be easily applied to other areas.
URI: http://140.113.39.130/cdrfb3/record/nctu/#GT079116815
http://hdl.handle.net/11536/40298
Appears in Collections:Thesis


Files in This Item:

  1. 681501.pdf
  2. 681502.pdf
  3. 681503.pdf