On Discovery of permanent land cover changes using time series segmentation approach

Abstract

Sustainable land management is one of the crucial aspects that need to be considered in order to protect the resources for future generations. Understanding of land cover changes that occurred during the past decade is necessary to formulate policies and actions for sustainable land management. Land cover changes affect ecosystem and global climate. These changes transform natural habitats of plant and animal species and also modify air temperature and near surface moisture content which leads to many drastic changes in climate. Land cover change study helps climate and ecosystem scientists in understanding role of land cover changes in bringing climate and ecosystem changes. This paper has used segmentation-based data mining approach on MODIS NDVI (Normalized Difference Vegetation Index) data for understanding of land cover changes in states of Gujarat and Rajasthan. Data smoothing using Savitzky-Golay filtering method has been performed before applying algorithm for land cover change detection. Algorithm is able to identify the time point of change along with type of land cover change. The findings show a lot of industrialization and urbanization in Surat and its satellite towns. Other major cities such as Ahmedabad and Jaipur have shown urban growth towards periphery over a period of time. Although agricultural area has got reduced due to urban growth but barren land has got converted into agricultural area as irrigation facility has improved over time due to emergence of Narmada Canal Network.