EXTRACTION OF SEMI-URBAN LANDSCAPES WITH REMOTELY SENSED DATA USING MAXIMUM LIKELIHOOD CLASSIFICATION TECHNIQUE

Abstract

A geospatial technique called remote sensing includes seeing and interpreting electromagnetic radiation that is emitted or reflected from surface of the Earth without making direct physical contact. It is crucial to identify and keep track of the physical traits of various ecosystems, including those in terrestrial, aerated, and aquatic environments. The categorization of Land Use and Land Cover (LU/LC) using remotely sensed data is an important study field in remote sensing. The Somvarpet taluk in Kodagu District is used for mapping and detecting LU/LC patterns. The LU/LC thematic map classification is performed on LANDSAT-8 satellite images from the years 2017–18 using Maximum Likelihood Classification (MLC) technique. The experiment was performed on five distinct training sample size as 100, 200, 300, 400, and 500 and examined semi-urban features using Panchromatic data. For these training sample sizes, the total classification accuracy achieved was 69.28%, 72.86%, 81.90%, 84.44%, and 90.80%, respectively. Additionally, for these training sample sizes, the corresponding Kappa Statistics were 0.5979, 0.6042, 0.7249, 0.7493, and 0.8510. Future studies can concentrate on enhancing urban feature classification accuracy and in-depth investigation of the variables influencing classification performance.