Applying a GIS to the modelling of slope stability in Phewa Tal Watershed, Nepal.

摘要

Mountain hazards are an inherent, but dangerous and costly element of mountainous environments. Most of the wide range of hazardous geophysical processes operating in mountain areas may be referred to as slope instabilities. Numerous models have been developed for application to individual slope failures, however these models are data intensive, requiring detailed geotechnical measurements at the site. Recent advancements in the application of digital elevation models (DEMs) and geographical information systems (GISs) have begun to explore the possibility of modelling slope instabilities on a regional scale. The work to date serves to emphasize the need for diverse geographical databases containing information not only on the geomorphic system, but also on its interaction with human activity.;This research compiles a diverse geographical database containing information on geological structure, soils, land use, and drainage for Phewa Tal watershed in western Nepal. The information was attached to geomorphometrically significant terrain units extracted from a DEM. In the context of this research, this represented the main advantage of applying a GIS, i.e. the ability to geo-reference the diverse data to a common reference scheme. The database was then analyzed using a variety of statistical techniques, with logistic regression and discriminant analysis producing the most useful results. Logistic regression was particularly successful, predicting terrain units as being either stable or unstable at a rate of almost 90% concordance. The predicted probabilities of terrain units being unstable were used to produce maps of predicted slope stability that compare well with a map produced through conventional means. However, continued research is needed to remove the subjective decisions involved in classifying the predicted probabilities.;The results of this research support the findings of the limited other work which has been done to date that DEMs and GISs may provide the facilities to develop regional slope stability modelling. In addition, this research suggests that the use of geomorphometrically significant terrain extracted from a DEM is a more efficient approach to slope stability modelling. The terrain units both facilitate the use of logistic regression and significantly decrease the amount of computing resources required. And finally, the findings of this research suggest that much of the information needed may be gathered from readily available information sources, such maps, aerial photographs, and written documents. Based on this finding, other information sources such as air-borne radar and satellite imagery should be examined.