在崩塌災害的防治工作上，使用遙測資料，如衛星影像、航空照片等，對於快速擷取災害情資有極大的助益。為快速取得大範圍的崩塌資訊，利用高解析度的遙測影像進行崩塌地判釋為現行的主流的方法，且自動化以及半自動化的快速判釋技術相當有助於進行緊急性的崩塌災情分析。在遙測影像的應用上，光學影像為較為普遍應用之影像資料，但缺點為受天候雲霧影響，有雲霧遮蔽時無法清楚觀測地表，此問題對於強降雨誘發之崩塌(如颱風有雲系覆蓋狀況)的即時判釋工作上有極大的限制。由於合成孔徑雷達感測器屬於主動式微波感測器，相較於光學感測器，其波長較長可以穿透雲層，觀測較不受天候影響，而過去利用雷達影像進行崩塌偵測的應用問題對於崩塌地自身的特性，包含規模、面積、崩塌類型等差異如何反應在雷達迴波訊號上有較少的探究；另外，雷達衛星本身觀測模式(升軌及降軌模式)、極化模式以及觀測波長的不同都將影響觀測結果。 本研究目的為提出一套分析流程來以精進雷達衛星資料對於崩塌地的判釋能力。本研究將針對不同面積、規模之崩塌地，分析其與雷達訊號強度以及影像紋理之關聯性。其次，本研究嘗試整合多觀測模式的雷達資料來弭補單一觀測模式導致的資料缺陷。本研究將針對2015年蘇迪勒颱風於翡翠水庫上游山區造成之崩塌地作為研究對象，使用Sentinel-1 C-band雷達資料進行試驗，針對不同極化模式(VV及VH)以及觀測模式(升軌及降軌)的雷達資料，進行對崩塌偵測適用性的比較分析。本研究認為，應用雷達影像進行崩塌地偵測有其價值及必要，特別在颱風事件及過後期間及過後，在雲霧覆蓋使得可見光影像無法運用在災害監測的狀況下。本計畫期望研究成果能有助於精進現有之山區崩塌災害監測方法，對我國崩塌災害管理與減災措施能有具體貢獻。 ;To detect landslide hazards for a wide region, remotely sensed data has been applied due to its efficiency and low cost. However, the cloudy condition during a typhoon may limit the application of optical data. For an emergent monitoring task, Synthetic Aperture Radar (SAR) is therefore a suitable tool for detecting landslides in cloudy and rainy weather. Previous studies have been applying change detection methods for landslide mapping, but the mechanisms of backscattering of different landslides (inducing, magnitude, size and type) have not been examined yet. In addition, the effects of different sensing modes (ascending and descending mode) on landslide identification over mountainous areas have not been studied as well. This study therefore aims to develop an effective landslide detection method, with applying an algorithm which is expected to combine different sensing modes. In addition, the relationship between landslide size and its backscattering will be considered in the detection method. In this study different polarized Sentinel-1 C-band SAR images (VV and VH) from different sensing modes, before and after Typhoon Soudelor (August 2015) which induced many landslides and caused serious damages in northern Taiwan, will be collected to perform the analysis. An effective and useful landslide detection method, based on SAR data, is expected to serve as a part of rapid response system of landslide hazard, when optical data is not available.
