降雨誘發山崩之逆分析及分析成果之評估指標探討

Study of Assessment Indices for Back Analysis of Rainfall Induced Landslide

蔡至航(Jhih-Hang Tasi)

指導教授 ： 劉家男

國立暨南國際大學/科技學院/土木工程學系/碩士(2014年)

https://doi.org/10.6837/NCNU.2014.00198

摘要

台灣本島四面環海位於熱帶與亞熱帶之間，常發生豪雨及引發邊坡災害事件，對於邊坡災害分析不容小覷。山崩分析中，因物理模型之土壤分析參數不易取得，由逆分析方法取得參數為理想方法之一，然而逆分析常有非唯一性問題，難以客觀決定單一最佳參數，故解決非唯一性問題及找尋最佳強度參數為本研究的重要課題。
本研究先建立兩組5x5及30x30網格做分析，利用山崩目錄進行有效凝聚力與有效摩擦角及水力傳導係數之定值式逆分析，並設定三種參數的範圍，由三種土壤強度參數之組合套入TRIGRS軟體進行邊坡穩定分析，得到各組之網格安全係數(FS)及山崩機率(PF)，這些分析結果產生的預測山崩網格及預測非山崩網格與山崩目錄之真實山崩網格及真實非山崩網格比較，比較之結果分別以五種指標下進行準確度之探討，分析結果發現不同指標下，非唯一解之現象皆普遍存在。
利用莫拉克颱風發生於南投縣仁愛鄉地區之資料，建立一組虛擬山崩網格進行分析，接著利用航照圖山崩判釋所產生之山崩網格進行分析，發現土壤強度參數為非均質性之山崩網格難以決定最佳土壤參數強度，而利用網格分布較多之山崩目錄進行分析，則可有效地提升逆分析解非唯一解之問題。

關鍵字

淺層山崩 ； TRIGRS ； 羅吉斯函數 ； Likelihood ； 逆分析 ； ROC曲線

並列摘要

Taiwan island surrounded by the sea located between the tropics and subtropics, often occurs torrential rain and trigger slope disasters, for slope hazard analysis should not be underestimated. Landslide analysis, because the soil physical model of analysis parameters easy to obtain, obtained by the inverse analysis method parameters for one of the ideal methods, however, the non-uniqueness of the inverse problem analysis often difficult to objectively determine the single best argument, it is to solve the problem of non-uniqueness and find the best strength parameters for this study an important issue.
In this study, first create two 5x5 and 30x30 grid for analysis, the use of landslides directories given effective cohesion and effective friction angle and hydraulic conductivity of the value of the type inverse analysis, and set the scope of the three parameters, a combination of three sets of soil strength parameters into TRIGRS slope stability analysis software were obtained for each group of grid security factor (FS) and landslide probability (PF), to predict landslides grid and forecast the results of these analyzes generated by the grid and non-landslide landslide landslide directory of real grid and true non landslides grid comparison result of the comparison of the following five indicators were discussed accuracy of the analyzes found that different indicators, the phenomenon of non-unique solution for all common.
Typhoon Morakot use occurred in Renai Township, Nantou County area of information, the establishment of a set of virtual landslide grid analysis, then use a landslide landslides sentence grid aerial image interpretation arising from the analysis, found that soil strength parameters for non-homogeneity of landslides difficult to determine the optimal grid soil strength parameters, and use more of the landslide catalog distribution grid for analysis, can effectively improve the inverse analysis of the problem of non-unique solution of the solution.

並列關鍵字

shallow landslides ； TRIGRS ； logistic functions ； Likelihood ； back analysis ； ROC curve

參考文獻

1.Guimaraes, R.F., et al.,“Parameterization of soil properties for a model of topographic controls on shallow landsliding: application to Rio de Janeiro”,Engineering Geology,Vol. 69, pp. 99-108, 2003.

2.Salciarini,D.,et al.,“Modeling regional initiation of rainfall-induced shallow landslides in the eastern Umbria Region of central Italy”, Landslides,Vol.3,pp.181-194, 2006.

3.Deb,S.K.and El-Kadi,A.I., “Susceptibility assessment of shallow landslides on Oahu, Hawaii, under extreme-rainfall events”, Geomorphology, Vol. 108, pp. 219-233, 2009.

4.Baum, R. L., Savage, W. Z., Godt, J. W., “TRIGRS – A Fortran program for transient rainfall infiltration and grid-based regional slope-stability analysis”, USGS Open-file Report 02-424, 2002.

5.Kim, D., et al., “Predicting the rainfall-triggered landslides in a forested mountain region using TRIGRS model”, Journal of Mountain Science, Vol. 7, pp. 83-91, 2010.