執行成果摘要
In 2014, this project had selected 2 experimental areas, Maolin and ThatMaxia, from the public and private forest-lands in southern Taiwan. Using FORMOSAT Satellite 2 images in 2007, 2009 and 2011 three dates of the two experimental areas to conduct the research. An image preprocessing procedure including radiance normalization and geometry correction had been developed to derive the correct NDVI index for different time images. After the preprocessing, the research completed NDVI images extraction from forest-use land covers, NDVI image classification procedure, and un-vegetation land extraction form the results of NDVI image classifications. After the classification procedure, an object-oriented and decision tree procedure integrated with landform geospatial information applied to extract landslide lands from un-vegetation lands. After that, three date landslide land and un-vegetation land change analyses had been finished. According to the eight kinds of changes from the three date GIS spatial overlay analysis, a 6 grades classification GIS layer for the evaluation of landslide crisis had been derived with a soil conservation rules created by author. The results show that in the two experimental areas judged from the rate of 3 kind of landslide areas changed from 2007 to 2009, the average impact of typhoon Morakot is 5.12 times the influence from 2009 to 2011. The 6 grades classification GIS layers for the evaluation of landslide crisis can be overlay on the Google Earth to see the landslide distribution impacts in different 3D views to each mountain village and provide the alarm information for the landslide management for the government institutes.