Assessment of Spatio-Temporal Changes in Land Use/Land Cover Over a Decade (2000-2014) Using Earth Observation Datasets: A Case Study of Varanasi District, India

Land use/land cover (LULC) changes have emerged as a major concern on global as well as on the local stage because of its considerable impact on climate and environment, especially in rapidly developing areas. Therefore, accurate mapping of LULC and ongoing changes over a time period have drawn a lot of attention in recent years. Remote sensing images from Landsat series satellites are a major information source for LULC change analysis. The present study mainly focuses on the evaluation of three classification techniques, namely maximum likelihood classifier (MLC), artificial neural network (ANN) and support vector machine (SVM) using multi-temporal Landsat images in order to choose the best method among them. The overall analysis based on accuracy measures indicates that the SVM is superior to ANN and MLC. The classification results achieved by the best recognized technique (SVM) were applied to assess the spatio-temporal changes in LULC that occurred in a fast growing Varanasi district of India over a period of 14years (2001-2014). A paired samples t test was also carried out to determine the statistical significance of changes in LULC between different studied time periods. The results reveal the rapid expansion in built-up area resulted in substantial decrease in agricultural land and other LULC classes. This study also highlights the importance of Landsat images to provide accurate and timely LULC maps that can be used as inputs in a number of land management and planning activities.