PEMETAAN BAHAYA LONGSORAN BERDASARKAN KLASIFIKASI STATISTIK PEUBAH TUNGGAL MENGGUNAKAN SIG: STUDI KASUS DAERAH CIAWI-PUNCAK-PACET, JAWA-BARAT

Abstract

Regarding environmental degradation in Puncak and its surrounding area due to the rapid landuse changes during 1981-1994, some mitigation schemes for soil erosion have been implemented but landslides. Data on landslide and its causative factors (landuse, soil, geology, slope, climate, and terrain mapping unit) and their relationships in the area are still not available. The objective of this research is to develop the method and procedures to map landslide hazards by using GIS (Geographical Information System) univariate-statistical analysis applied to the area as a case study. For this purpose, three methods to classify and to map landslide hazards were evaluated. They were respectively developed by considering: (a) unweighted density total number of the landslide, Method#1, (b) density total number, weighting value, and age of landslide, Method#2, and (c) density total number, weighting value, age, and activity level of the landslide, Method#3. The density number is counted by overlying· each of the landslide maps and each of the causative factors maps used, and from this step, the weighting value is derived. The resulting density number of landslide given as a cumulative percentage and the corresponding weighting value were then plotted on an X-Y graph. From the graph, the level of landslide hazard is classified by applying: (1) standard classification procedure as the default statistical analysis given by the software used, and (2) natural classification procedure as it based on the nature of the curve slope of the cummulative graphic. The result given by the three methods were varied but in general, they gave landslide hazard maps with a similar pattern in which the very high and very low hazard levels in the study area increased during the period of 1981-1994. Of the three methods, applying the natural classification gave a better result than that of the standard classification procedure. Method#2 and Method#3 were better than Method#1 in predicting the future landslide occurrence. Apparently, Method#3 should show the best result but the effect of conversion from raster to vector data in GIS significantly reduced the quality of the resulting map.