The present study demonstrates the application of remote sensing for the estimation of areas corresponding to crop and forest lands covered in the district of Aurangabad, (Maharashtra), India. The data acquired by IRS-P6 Advanced Wide Field Sensors (AWiFS) having 56m spatial resolution for the months of October & December 2012 which covered complete study areas with its swath of 740Km has been used for the study. The Maximum Likelihood Classification (MLC) and Knowledge Classification (KC) techniques based on Decision Tree approach were applied. It has basically two elements, knowledge engineering and knowledge classifier. Knowledge engineering provides an interface to build up decision tree which defines the rules and variables represented by three parameters, i. e. Normalized Difference Vegetation Index (NDVI), Soil Adjust Vegetation Index (SAVI), and Normalized Difference Water Index (NDWI) threshold value of each class. Knowledge classifier generates the required output classification. The objective of this research work is to perform classification of crop and forest acreage estimation from the AWiFS data and comparing it to the supervised classification techniques, MLC and KC. The result shows that values of overall classification accuracy were 82% and 84% for the months of December and October 2012 respectively using MLC, whereas corresponding values of accuracy were found to be 85% and 87% based on KC. Thus the classification results based on KC provide better results than corresponding results based on the MLC.

IRS-P6 Advanced Wide Field Sensors Knowledge Classification Maximum Likelihood Classification Crop Forest Accuracy Assessment.