A STUDY ON THE METHOD OF PRECISE GEOMETRIC CORRECTIONS OF LANDSAT MSS DATA

Abstract

Remote sensing analysis of Landsat MSS data has been studied for more than ten years with good results being obtained in some fields. However, it is still insufficient for really practical purposes, for example, in the case of land use classification. That is mainly because of the existence of mixed pixels in the images of Landsat MSS data. The problem of mixed pixels is very difficult but important.
So a new method for geometric corrections over a small area of Landsat MSS image has been proposed in this paper.
The original image data of Landsat MSS of the Kanto plains was analyzed in this paper. This MSS data was very precisely corrected with the new geometric corrections method developed in this paper. Then, the relation between the MSS values of the mixed pixels and mixture rates of land covers in those mixed pixels were statistically examined by sampling mixed pixels in which mixture rates of land cover were various.
The results are as follows:
a) The new method of geometric corrections is proposed and its efficiency is certified.
b) The mixing method seems to hold in the case of the pixels in which two land covers are mixed, paddy field and asphalt pavement.
As a result of this, a more precise analysis of remote sensing will be possible.