Patil, S. S.HARSHAVARDANA, R.2020-10-052020-10-052019-08-24https://krishikosh.egranth.ac.in/handle/1/5810152359This study presents a land use, land cover classification of satellite imagery. Visualization of feature space allows exploration of patterns in the imagery data. The Machine learning algorithms are utilized for pattern classifications. Test imagery was obtained through Sentinel-2 Satellite on February 2018 and February 2019 for Madikeri Taluk, Kodagu District. The second image was taken to measure the changes due to disaster happened in the study area in the month of August 2018.The supervised classifier is identifying the classes using trained set. The statistical significance of satellite image classifiers into constituent classes is of greater importance in remote sensing pattern recognition methods. Maximum Likelihood Classification, Minimum Distance to means Classification, Mahalanobis Distance Classification and Spectral Correlation Mapper Classification were performed using ERDAS imagine and ArcGIS 10.5.1 image processing Algorithms. Accuracy of the classification of data set and classifier were expressed using confusion matrix. F-measure value and Kappa coefficients were used to measure the overall accuracy, user’s accuracy, producer’s accuracy. The test of significance of the Kappa coefficient was performed using Z- test. Maximum likelihood classification was out performing with highest overall accuracy by 75.56 per cent followed by Spectral correlation mapper 71.02 per cent, Minimum distance 67.61 per cent and Mahalanobis distance 61.93 per cent were observed. The post disaster image accuracy by the Maximum likelihood classifier is 73.29 per cent. The changes in the total area among the feature after the disaster noticed was 1184 ha. This study guides the Researchers and policymakers to study the pre and post disaster.EnglishThesis