Abstract
The distribution, size and change of landslides are mainly investigated by aerial photographs and field survey, though it takes much labor if a wide range of landslides is measured. An alternative approach is to use satellite remote sensing. A purpose of this study is to map and monitor landslide areas at Mt. Ichifusa in Kyushu using remote sensing data of Landsat and SPOT images. The results are as follows :
1) The accuracy of classification of the landslide is over about 71% using NDVI from Landsat and SPOT data. The total area of the landslide estimated from TM and HRV XS data is nearly the same as that mapped by using aerial photographs.
2) A different sensor and different day of observation on Landsat and SPOT cause the change of NDVI values for the classification and each time NDVI values change, the ortho photograph of ground truth data is required to esti-mated classification. This is a problem in monitoring the landslide areas by satellite data for a long term.
3) Computing shadow and stream areas from DEM by Geographical Survey Institute data using GIS (Geographic Information System) increases the accuracy of classification of the landslide areas.
4) Using satellite remote sensing data that have the ground resolution appropriately selected to the size of the land-slide leads to high accuracy of classification of the landslide areas. At mountainous areas it is more important to use satellite data in for shadowless period than to use high ground resolution data.
