| Summary: | 碩士 === 國立成功大學 === 資源工程學系 === 105 === Recently, the technology of Aerial LiDAR was developed quickly, so we could produce a more accuracy digital elevation model (DEM) with higher accuracy. Based on this high resolution DEM, an automatic and objective process to uncover the distribution of gullies via the following methods : “wavelet analyst”, “Canny edge detector”, “define the gully-shape” and “perform non-linear noises removing”. The 140 〖km〗^2 DEM used in this study is at the northern part of Tseng-Wen Reservoir Catchment in southern Taiwan. The points cloud which formed the DEM was captured at Dec. 2015. After appling the automatic searching model for the DEM, we inspected the results on site at Jan, 2017. Compared the results of the automatic model and the site inspection by Confusion Matrix, the total accuracy of automatic model is 68.88%.
The 18th Taiwan Province highway, we also call Alishan-Motorway, is the trunk major traffic road to access the study area. It is very important for residents and tourists for the traffic mission, but Alishan-Motorway also suffered sever disasters, such as landslide and subgrade-emptieded, during heavy rain almost occurred every year. This study considered that some of the disasters that could related to the growth of gully. Therefore, another purpose in this study is to assess the growth potential of gully. Because the development of gully is triggered by many factors, this study used seven topographic factors as the attributes and the gully-distrubution from automatic gully detection model as the target attribute to map the growth potential of gully. The accuracy of the potential map is 66.07%. Finally, this study search the the area which could be threatened by the vagorous grouth of gully. If the dangered locations are spreated at the upper slope, the landslide might happened, else if the dangered locations should scattered at the lower slope, the the subgrade-emptieded may happen.
|
|---|
