A Priori Study of Using Spatial Data Mining Technology with FORMOSAT-2 Imagery for Analyzing Potential Landslide-causing Factors

摘要

Mountainous region, steep slope, broken terrain, and together with the frequent earthquakes and heavy rainfall are easily to trigger severe geological hazards such as large-scale landslides and debris flows in Taiwan. Series property damages and life losses caused by natural hazards can be reduced effectively if modern technology and knowledge are introduced into early warning systems. In this study, the FORMOSAT-2 imagery over Jhuo-shuei River basin acquired from 2006 to 2012 was employed to classify the landslide area. Spatial data mining technology was applied to calculate the weightings of various spatial factors for relevant landslide sites. In addition, spatial auto-correlation analysis and spatial auto-regression analysis were used to achieve a disaster hot spot analysis. The dataset of hot spot concentrated area over Jhuo-shuei River basin can provide to the disaster support system to achieve disaster early warning.