宜蘭縣清水溪集水區崩塌量變化之原因探討

近年氣候變遷加劇與臺灣特殊的地理環境，造成了不少的重大土砂災害事件。位居東北區域的宜蘭地區，在一年四季皆有雨的情形下，山區崩塌地更是易受降雨事件的影響，產生崩塌現象。根據相關計畫調查，每年蘭陽溪流域皆有著可觀的崩塌量體從坡面傾瀉至河道中，其中蘭陽溪每年生產的土砂，近四分之一的量體來自清水溪集水區坡面的崩塌地，進而對河道下游地帶一些居民、財產等保全對象上，恐有安全疑慮之虞。有鑑於崩塌量體對於集水區的危害，現今許多防災、學術研究等單位皆著手致力於精準推估崩塌量體可能產生的數量多寡，以及推測量體產生之區域分布，遂發展許多分析模型等方法。故本研究亦針對蘭陽溪流域主要崩塌量體來源的主要河川──清水溪，以該集水區為研究主軸，使用兩期年度的數值高程模型及眾多地文因子圖層，搭配崩塌深度研究方法的理論，套用一個適用於本研究區域之迴歸式模型，並由每一個網格崩塌深度與面積的乘積，推估清水溪集水區內的崩塌量體體積。根據本研究成果顯示，本研究所使用之崩塌深度迴歸式模型，修正自由度的迴歸判定係數可達0.97，可有效解釋清水溪集水區內的崩塌量體體積推估，並從相關地文圖層與崩塌量體的空間分布位置，得知清水溪集水區的崩塌深度會受坡度、地形濕度指數等地文因子影響；在地層方面，也會受到表土岩層特性、崩塌型態的影響，使崩塌量體多發生在頁岩層、砂岩層的淺層崩塌區域內。希冀本研究對於該區域未來有關之土砂治理計畫，可供參考與提供實質建議，以達到實際減災效果。

關鍵字

集水區；崩塌深度；崩塌量體推估

並列摘要

Given the deteriorating climate change and the unique geographic context of Taiwan, many major landslides crisis occurred in the recent years. Locating at north-eastern area of Taiwan, mountain areas in Yilan is subjects to landslides due to the year-round precipitation events. According to relevant survey, a significant amount of landslide volume was transported from the slope to the channel in the drainage basin of Lanyang River. Among the sediments produced by Lanang river annually, around a quarter of the volume are contributed by the landslides from the slope of Chingshuei River, which threatens the safety of residences and properties downstream of the river. Due to the threat of landslides to the drainage, many disaster prevention and academic institutions have been working on to precisely predict the amount of potential landslide volume outcome, and to estimate the distribution of the volumes occurring. Hence, many analytic models are developed. This research targeted the main sources of volume in the drainage basin of Lanyang River, Chingsuei River. Using digital elevation model of two years and several geomorphic factors in combined with the theory of landslides depth research, this research develops a regression model suitable for this research area. From the multiplication of the landslides depth and area, the volume of the landslides in the catchment of Chingshuei River can be estimated. The results show that with the regression model of landslides depth, the adjusted coefficient of determination is 0.97, meaning that the model can effectively explain the estimation of landslides volume in the catchment of Chingshuei River. From the distribution of geomorphic factor layers and volume, we know that landslides depth would be affected by geomorphic factors such as slope and terrain wetness index. As for the stratum, due to the influence of rock formation in the topsoil and the form of landslides, shallow layer area with shale bed and sandstone bed are more subject to landslides. It is expected that this research will contribute practical suggestions to the future sediment mitigation project of this area and serve as the reference in order to achieve disaster mitigation.