A model for landslide prediction is developed using land cover information obtained from a satellite multi-spectral scanner data and geographical information such as the lay of the land, the nature of the soil and so on.
In this study, the applicability of remote sensing data for this model, the position of the remote sensing data on this model, and the significance of vegetation index and reflection characteristics are investigeted.
The conclusion of this study is as follows.
(1)The effective bands of TM-data for the landslide prediction are band-1, 4, 5, 6 selected with the Quantification Method Type I. The usage of the infrared bands is assumed to be particularly effective for the landslide prediction.
(2)The clustering of reflection characteristics is defined and evaluated with the Quantification Method Type II using the four bands described above. The selecting method of the effective bands from TM-data is found resonable.
(3)Although vegetation index in deduced from some different data with Quantification Method Type II, the most effective vegetation index of the four indices cannot be decided for the landslide prediction. Therefore another study from different viewpoint is necessary for the selection of the most effective vegetation index.
(4)The landslide prediction map thus made up is successfully compared with the existing field condition. This map can be used practically for the landslide prediction.
View full abstract