ANALYSIS OF LAND USE LAND COVER FEATURES IN SEMI URBAN LANDSCAPE USING RANDOM FOREST

Abstract

  Remote sensing offers a unique environmental monitoring capability that covers extensive geographical areas, while capturing information on the Earth’s atmosphere, land and oceans. The images taken are in form of pixel and process of changing it into digital images is known as Image classification. It is an effective technique to automatically discriminate classes of images with some known information. The Land Use Land Cover (LULC) describe the vegetation, water, natural surface, and cultural features on land surface. This study focuses on mapping and monitoring of LULC pattern of Virajpet Taluk, Kodagu District.  The LANDSAT-8 of 2017 and 2020 data were used to extract LC map features. Random Forest is considered for classification as it is flexible for classification. From the result analysis, it is observed that OCA for the multispectral data is 51% and for fused data is 63%. Fused image gives the better classification results of LULU features than the Multispectral image. Random forest proved that is suitable for classification and produces a great accuracy most of the time.