Abstract
The purpose of the study is to evaluate the suitability of moderate-resolution imaging spectroradiometer (MODIS) data to study the land use land cover over India. The study is based on secondary data sets pertaining to forest, cropland, pasture, and barrenland obtained from Directorate of Economics and Statistics (DES) and MODIS (Terra) global land use land cover data yearly composite from 2002 to 2005. A family of statistical and mathematical techniques is adopted here in order to compare the MODIS data with DES statistics. The comparison at the country level shows estimated forest cover has least uncertainty compared to pasture and barrenland. Comparison at the state level, on the other hand, shows high degree of association between the data sets in cropland (R 2 =0.9 ), followed by forest cover and pastureland. Barrenland shows weakest association between DES and MODIS. The computed average accuracy in cropland shows a level of 84% and has been chosen as the best fitted land cover category among all land cover classes selected for the study. Hierarchical clustering of the MODIS cropland at the state level based on the estimated accuracy shows that, except for Andhra Pradesh, Tamilnadu, Haryana, West Bengal, Chhattisgarh, and Orissa, which are far off from the true estimate, the rest of the states are in closer correspondence of the cropland statistics reported by Ministry of Agriculture.
