Comparison of Image Processing Techniques to Identify the Land Use/Land Cover Changes in the Indian Semi-arid Region

Abstract

Several parameters are combined to form a system; if one of the parameters is changed, it also influences the other parameters. Both biophysical components and features made by humans are dynamic and constantly changing. Resources are being used to rapidly meet the growing population demand, causing changes in land use and land cover (LULC). It is believed that LULC change is a significant component of global change that impacts climate change. A significant amount of effort has been into creating methods for remotely sensed data change detection. LULC change detection of Hyderabad using remote sensed images has been performed in this project. Landsat 7 ETM+ imagery for 19 May 2000 and Landsat 8 (OLI/TIRS) imagery for 5 May 2015 has been obtained from the USGS Earth Explorer. ERDAS Imagine 9.1 is used for image rectification, layer stacking, and cloud cover correction. A spectral signature training file is prepared using the pixel values of the image and its properties in the ArcGIS 10.1. Then, the supervised classification was done using the maximum likelihood classifier in 4 different classes (Class 1: Vegetation, Class 2: Barren land, Class 3: Waterbody, and Class 4: Built-up). Post-classification comparison of the two maps is made on a pixel-by-pixel basis using a change detection matrix. Moreover, the changes from the before image (19 May 2000) and after image (5 May 2015) have been done in Arc GIS 10.1 using raster calculator with the image difference method (Before and After images) and ERDAS Imagine 9.1 using the Change Detection tool. ERDAS Imagine 9.1 performed better than ArcGIS 10.1 and shown distinct changes. From the LULC change map, great changes in the stretch of water body have been observed from 2000 to 2015. The different classes of the before map have been merged into other classes or changed as new classes in the after map. The area under the water body has been transformed into a built-up area and barren land. LULC change in 15 years also induced two new classes in the study area Class 1 and Class 2. The observed new classes and to distinguish this class from other classes ground survey has been done using Google Earth. In ground survey, the Class 1 has been identified as Nehru zoological park and Class 2 as a planned residential society. LULC change map can be used by the government, decision-makers, and policymakers and can help in proper resource management and policymaking.