宜蘭縣大同鄉土地利用與邊坡崩塌之災害潛勢分析

Abstract

邊坡崩塌為臺灣常見的塊體運動現象，若影響到人類的生命或財產，則為坡地災害。在特定時空下所產生的災害，其潛勢值可被計算，本研究以國土測繪中心的全國土地利用類型資料做為人類成災的條件，而自然條件為一地環境的背景要素，兩者交互作用下便可能產生災害，而其中的潛勢值便可計算出來，此為災害潛勢的概念。研究方法主要以現今在邊坡災害研究上普遍使用的羅吉斯迴歸，而研究區選定為土地利用類型多元和自然環境脆弱的宜蘭縣大同鄉。 本研究以地理學中的人地互動觀點去做串聯，先以大同鄉土地利用對邊坡崩塌的影響，是為「人」對「地」的影響，進行因子的篩選，得出土地利用類型、農作距離、道路距離此三項具影響力的因子；接著為「地」對「地」的作用，即自然條件對邊坡崩塌的影響，篩選出以下七項因子，為高程、坡度、坡向、地質、河道距離、一日最大降雨分級和年總雨量。前兩個步驟皆著重於「因子」對於「地」的影響評估，目的在於找出可以解釋造成邊坡崩塌的影響因子，然而若要評估其影響力的大小以得出災害潛勢值，需將篩選得出的土地利用因子與自然條件因子結合，最後將此十項因子合併進行單因子檢驗和羅吉斯迴歸的統計分析，刪除坡向和年總雨量此兩者不具解釋力的因子後，得出一解釋力68.3%的模型。本研究藉由統計方法來計算大同鄉土地利用與邊坡崩塌的潛勢值後，繪製出大同鄉的災害潛勢地圖，但此潛勢值與潛勢地圖需與實際現況配合，才能有效達到防災應用，因此最後以野外考察和居民訪談做一防災管理的探討，並希冀未來能持續投入研究以實際達到減災效果。Landslide is a common phenomenon in Taiwan. If it affected people’s life or properties, that has become slope hazard. According to the hazard in specific time and space, its susceptibility could be calculated. This study defines landuse type data which is from National Land Surveying and Mapping Center as one of the condition of hazard. When landuse types interact with the physical factors of any environment backgroungs could yield hazard, and the susceptibility value could be generated. This is the concept of susceptibility. As for study method, this study applied Logistic regression which is commonly used in recent landslide studies. The study area must be provide with plentiful landuse types and vulnerable physical environment, so Tatong town was choosed. Human-land relationship is the structure in this study. To begin with, the impact of “human to land” was evaluated by the landuse types to landslide in Tatong town. In this procedure, this study selected factors by collinearity diagnosis which help to discover and eliminate multicollinearities. 3 influential factors was obtained, that were landuse types, crop distance and road distance. Sencondly, the impact of “land to land” was assessed by physical factors to landslides, and this study acquired 7 influential factors, that were DTM, slope, aspect, geology, river distance, 24 hours maximum precipitation and annual rainfall by collinearity diagnosis as the same. This two steps focused on the evaluation of “factors to land”, and the aim was to find the causal factors which could precisely explain the reason why makes landslide. Moreover, this study have to combine the landuse factors and physical factors so that could generate hazard susceptibility. And then put this 10 factors into univariate tests of association and Logistic regression, after deleted the aspect and annual rainfall factors which were uninterpretable, the 68.3% explanation model was obtained. This study use statistics methods to demostrate the susceptibility of landuse type and landslide, by this we could create hazard suceptibility maps in Tatong town. In order to utilize the susceptibility value and susceptibility maps practically, we carried out field works and interview with residents to revise the study to achieve hazard prevention.