南投縣仁愛鄉崩塌特性分析及崩塌風險模式建構

Analysis of the Landslide Characteristic and Building the Landslide Risk Model for Renai Township, Nantou

吳俊鋐(Chun-Hung Wu) ； 黃均臺(Jun-Tai Hunag) ； 吳亭燁(Tingyeh Wu)

《中華水土保持學報》 49卷3期 (2018/09) Pp. 154-166

https://doi.org/10.29417/JCSWC.201809_49(3).0003

摘要

本研究以南投縣仁愛鄉在2009年莫拉克颱風後崩塌目錄，建構仁愛鄉內之崩塌潛勢評估模式及風險模式。仁愛鄉在2009年莫拉克颱風後之崩塌案例，有57.0%位於下邊坡及20.5%位於上邊坡，且崩塌規模以小規模崩塌為主。崩塌潛勢評估模式建構分別以邏輯式迴歸法及證據權重法進行，兩種方法之正確率分別為74.47%及61.92%，因此選定邏輯式迴歸法建構之崩塌潛勢評估模式，經現場比對後，該模式能預測出多數淺層崩塌，但難以適用於地下水誘發的地滑型崩塌。最後將崩塌潛勢結果結合研究區域內之建物與道路成本，套疊不同頻率年24小時累積降雨，獲得仁愛鄉在50年、100年及200年24小時累積降雨事件下，可能產生的損失風險值分別為11.8億、24.0億及38.7億元。

關鍵字

崩塌特性 ； 崩塌潛勢模式 ； 崩塌風險模式 ； 崩塌重現機率

並列摘要

The study developed landslide susceptibility and risk models based on landslide inventory collected after 2009 Typhoon Morakot in Renai Twonship, Nantou County. Approximately 57.0% of landslide cases induced by Typhoon Morakot in Renai Township were located in downslope locations, whereas 20.5% were located in upslope locations. The study created landslide susceptibility models by using logistic regression and the weight of evidence method. The accuracy of these two landslide models were 74.47% and 61.92%, respectively. The logistic regression-based landslide susceptibility model can predict most shallow landslide cases; however, it cannot predict groundwater-induced landslides. Considering the landslide susceptibility and cost of buildings and roads, landslide loss caused by 24-h accumulated rainfall with the return period of 50, 100, and 200 years were estimated to be 1.18 × 10^9, 2.40 × 10^9, and 3.87 × 10^9 dollars, respectively.

並列關鍵字

Landslide characteristic ； landslide susceptibility model ； landslide risk model ； landslide recurrence probability