Study on Land Cover Classification by LANDSAT MSS data

Abstract

The study was made on the (accuracy in) classification of land cover with the effects of the variation of seasons and shadows of landforms based on remotely sensed data. In this study, we were made of data of the isolated mountains at the Ibaraki Pref. and its surrounding plain areas in Japan in autumn and winter in 1983 by the most likelihood method. The result indicates the following features, 1) The category of rice paddy, crop fields and residential areas are not suitable for interpretation of land covers to practical application by only one seasonal data. 2) Accuracy of classification will be up to about 80% by using of data both autumn and winter and overlaid of its data. 3) In generally, an analysis of data at the shadowing areas were difficult but classification of conifer forest and deciduous forest may be available in this area.