請用此 Handle URI 來引用此文件:
http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/54180
標題: | 集水區逕流因子對土石流發生之影響 - 以陳有蘭溪集水區為例 Effects of Watershed Runoff Factors on Debris Flow Occurrence–Using the Watershed of Chenyoulan Stream as An Example |
作者: | Ching-Fu Chang 張景富 |
指導教授: | 范正成(Jen-Chen Fan) |
關鍵字: | 土石流,邏輯斯迴歸,崩塌率,降雨逕流演算, debris flow,logistic regression,landslide ratio,rainfall runoff routing, |
出版年 : | 2015 |
學位: | 碩士 |
摘要: | 從過去至今,已有許多研究以渠槽實驗或數值模擬之方式,解釋土石流與地表逕流之力學現象;然而,在集水區尺度利用地表逕流來預測土石流的發生,卻較少研究探討。本研究旨在探討集水區逕流因子對土石流發生之影響,以陳有蘭溪集水區為例,建立使用降雨及逕流等共六種不同水文因子的土石流警戒模式,並比較模式優劣。
首先,以統計檢定之方式篩選出有效集水區面積(EWA)、溪床最大坡度(MSS)以及崩塌率(LR)等三個彼此獨立且與土石流發生顯著相關的地文因子,並以兩種不同的方法(即使用原始值與使用隸屬度)來量化地文因子。接著,本研究以單位歷線法進行陳有蘭溪集水區之降雨逕流演算,並以內茅埔流量站之實測流量驗證之,最後獲得陳有蘭溪集水區內土石流潛勢溪流之集水區逕流因子;同時亦參考前人作法建立降雨因子。而後,利用邏輯斯迴歸結合地文因子及不同的水文因子建立土石流警戒模式。 研究結果顯示,在兩種地文因子量化法、六種水文因子共十二種組合之中,使用隸屬度之土石流警戒模式較佳;而在其中又屬使用有效累積雨量、單位面積尖峰逕流以及Shields Stress Parameter推估值的五組模式之準確率較佳,均達81%以上且在伯仲之間。同時,本研究以土石流發生機率等高線圖來呈現土石流警戒參考指標,可以提供防災與整治管理兩個方向之應用價值。防災單位依照可接受之土石流發生機率,配合即時更新之崩塌狀況,訂定水文因子之警戒值以利警戒發布。另一方面,整治管理單位亦可給定一設計暴雨之水文條件,根據可接受之土石流發生機率制定崩塌地降低或管控之目標崩塌率,以利土石流之防災。 The mechanical relationship between debris flow and surface runoff has been studied and described through channel experiments and numerical modeling. However, research on using surface runoff to predict the occurrence of debris flow at watershed scales is relatively rare. The objective of this study is thus to investigate the effects of surface runoff in watersheds on debris flow occurrence. With the watershed of Chenyoulan stream as an example, this study uses six different hydrologic factors, including one rainfall factor and five runoff factors, to establish debris flow warning models, whose performances and accuracy are then compared and analyzed. To start with, based on the results from statistical tests, effective watershed area(EWA), maximum streambed slop steepness(MSS), and landslide ratio(LR) are selected for model establishment, because of their mutual independence and their significant relation to debris flow occurrence. These physiographic factors are then quantified by two methods, namely their genuine values and their degree of membership(DOM). Secondly, through synthetic unit hydrograph and validation by observed data, Chenyoulan stream’s runoff is simulated, from which five runoff factors are obtained. The effective accumulated rainfall is also obtained for the purpose of comparison. Finally, twelve debris flow warning models are established using logistic regression; each of them comprises one hydrologic factor and two or three physiographic factors quantified by either their genuine values or their DOMs. The results show that using DOM to represent physiographic factors makes the performances of models better. Furthermore, there are five hydrologic factors that yield significantly better model accuracy, which is higher than 81%. These factors include effective accumulated rainfall(EAR), unit area one-hour peak direct runoff(q1), unit area three-hour peak direct runoff(q3), unit area five-hour peak direct runoff(q5), and estimated Shields stress parameter(SS). The findings of this study and the consequential debris flow occurrence contour maps can be applied for disaster prevention and site rehabilitation purposes. Based on acceptable debris flow occurrence probabilities and updated landslide ratios, the authorities can establish threshold values of hydrologic factors for issuance of debris flow warning. On the other hand, with the help of designed storms and acceptable debris flow occurrence probabilities, the authorities can also set up site rehabilitation goals of landslide area reduction for debris flow prevention. |
URI: | http://tdr.lib.ntu.edu.tw/jspui/handle/123456789/54180 |
全文授權: | 有償授權 |
顯示於系所單位: | 生物環境系統工程學系 |
文件中的檔案:
檔案 | 大小 | 格式 | |
---|---|---|---|
ntu-104-1.pdf 目前未授權公開取用 | 3.57 MB | Adobe PDF |
系統中的文件,除了特別指名其著作權條款之外,均受到著作權保護,並且保留所有的權利。