Analysis, Classification, and Estimation of Pattern for Land of Aurangabad Region Using High-Resolution Satellite Image

Land use land cover (LULC) information extraction is a crucial exercise for agricultural land. The present study highlights the advantages of remote sensing, GIS, and GPS techniques for LULC mapping from high-resolution remote sensing data. High spatial resolution (5.8 m) satellite imagery of IRS-P6 Resourcesat-II LISS-IV having three spectral bands was utilized for LULC classification and for data processing ENVI 4.4 tool and Arc GIS10 software were used. Eight training samples for LULC classes have been selected from the image. Supervised classification using maximum likelihood (ML), Mahalanobis distance (MD), and minimum distance to means (MDM) were applied. The performances of above classifiers were evaluated in terms of the classification accuracy with respect to the collected real-time ground truth information. The evaluation result shows that the overall accuracies of LULC classifications are approximately 84.40, 77.98, and 74.31 % with Kappa coefficients 0.82, 0.74, and 0.70 for the ML, MD, and MDM, respectively. It is noticed that ML has a better accuracy than the MD and MDM classifiers and it is a more effective method for complex and noisy remote sensing data because of its unified approach for estimation of parameters.