過去之地形地貌研究中,皆是以長時間尺度探討其變化情形,使用之數值模型資料也較粗糙。本研究主要以莫拉克颱風單一事件為基礎,研究區涵蓋南化水庫、旗山溪流域及荖濃溪流域之集水區,並從該集水區中挑選樣本區域,使用2米之高精度數值高程模型(DEM)資料以及高解析度之航空照片,做為樣本研究區中,地形指標之研究分析,本研究採用的地貌指標有坡度(Slope)、地表粗糙度(RI)及曲率(Curvature)。 本研究利用40米之DTM搭配地理資訊系統(GIS)外掛模組Hydrology進行集水區自動劃分,並使用綠度指數(Greeness)進行航空照片裸露地判釋。挑選莫拉克颱風事件後,崩塌比率較高之集水區,並從該集水區中,挑選出樣本研究區域,將其樣本研究區域資料輸入自行研發之MATLAB自動化視窗程式,進行地貌指標分析,以瞭解單一颱風事件期間之地貌變化情形。 研究成果顯示,坡度與地表粗糙度皆反應出在莫拉克颱風事件後擾動區域之平均變異百分率變化較顯著,擾動區之變異百分率在70%以上。且地表粗糙度程度,在擾動區域中可發現,在莫拉克颱風事件前,平均值及標準差皆偏大;而未擾動區域中,平均值及標準差偏小,而曲率部分,採用15m x 15m以及10m x 10m視窗網格基礎進行最大曲率以及平均曲率之運算,發現在莫拉克颱風事件前後,擾動區域與未擾動區域之平均變異百分率中,同樣可看出擾動區域其數值變化較顯著,而未擾動區域其數值變化較不顯著。
In the past the research of topography and landscape evolution, applied the Digital Elevation Model (DEM) and the scale was 40mX40m, focus on the long-term variation processes. The study areas including Chi-Shan River Basin、 Lao-Nong River Basin and Nanhwa Reservoir Watershed. The study takes Typhoon Morakot for example and uses 2 m Digital Elevation Model and Aerial-photos to analyze topographic change. A topographic index that is a composition of three different indexes to evaluate the model’s performance, that is Slope、Roughness index and Curvature. The Hydrology tools of Arc-GIS are used to perform watershed delineation on a digital terrain model and applied Greeness to recognize the landslides on aerial-photos. The sampling data are extracting from catchment that has higher slump ratio. In order to evaluate the topographic evolution during typhoon events an automatic window program based on MATLAB is proposed to analyze topographic change. The results indicate that slope and surface roughness has significant variation and the percent of variance up to 70 percent on disturbance area after event of Typhoon Morakot. According to surface roughness variation, the average and standard deviation of surface roughness in disturbance area are larger than the undisturbed area. In the Curvature analysis, using computational mesh of 15m x 15m and 10m x 10m to perform maximum and average of Curvature. The result was coincident with surface roughness analysis.