2004 Volume 57 Issue 2 Pages 27-38
Landslide hazard prediction can provide very useful information or technical supports to regional landslide mitigation and control. Now there are many methods and examples of landslide prediction, but statistics methods are usually used methods, in which, quantification II approach is the most used. On the other hand, being an effective tool, GIS has been used in many regions now, including landslide prediction. In this paper, we present a combination of quantification II approach and GIS techniques to predict landslide hazard in HAYACHINE Mountain ranges, Japan. The details are showed as follows: (1) Based on the recognition, delineation and analysis of existing and past landslides (derived from aerial photograph interpretation) and factors impacting on regional landslide (termed “causal factors”, such as slope gradient, slope aspect, geology, soil, precipitation, slope type, vegetation and elevation, they were extracted and analyzed with the help of GIS techniques), landslide and 8 causal factors grid maps were produced respectively, each of them has the total of 1024*1266 pixels and pixel size is 9*9m ; (2) 372 samples (pixels) were selected from each grid map. Based on these samples, the regional landslide prediction model was obtained by quantification II approach. Through testing with the overall pixels in research area, the precision of landslide prediction showed 80.49%; (3) the landslide hazard zones were classified by the probability of landslide occurrence in the research area.