Accuracy Assessment of land use maps classification based on remote sensing and GIS techniques

Abstract

The need to classify sentinel-2 satellite images to create land use /land cover (LULC) are essential to analysis the processes of environment problems and to improve living conditions. Hence, this research aims to assess of accuracy classification by Support Vector Machine (SVM) approach to create LULC maps from sentinel-2 satellite images using remote sensing and GIS. The selected study area for this research is Baghdad city because of it has a unique political stability and due to rapid urbanization that lead to rise additional request for natural resources and affected on LULC in Baghdad city. After preprocessing and processing of satellite images, thematic maps were created and classified into five main classes based on visual interpretation and visit the field of the study area containing: urban, vegetation, soil, asphalt roads, and water bodies. The results showed that classification accuracy assessment of SVM algorithm are acceptable because of overall accuracy and Kappa index equal (88%, 0.84) respectively.