Change detection of land use /land cover using object oriented classification of satellite images (Case study: Ghare Sou basin, Ardabil province)

The accuracy of land use changes map obtained from remote sensing data depends on the accuracy of each of the land use maps during the time period studied. In this study, TM and OLI images in 1989 and 2018 and an object-oriented classification method were used to investigate the land use/ land cover change trends with an emphasis on agricultural land use in the Ghare Sou basin. After the pre-processing, the object-oriented processing using the multiresolution segmentation method was applied. In addition to the spectral bands, some additional information such as a normalized difference vegetation index (NDVI), band means, the standard deviation of bands and geometry characteristics were used to extract land use in order to obtain more accurate results. Of these non-spectral data used, 15 characteristics were selected by Feature space optimization (FSO) method to be used in the nearest neighborhood algorithm. The kappa coefficient of the land use maps for 1989 and 2018 was 85% and 96%, respectively, indicating the reliability of the object-oriented classification results. In the next step, the map of the changes was produced comparing the classified maps. According to the results of the change detection, the agricultural land use during the studied period has an increase of 73849 hectares, mainly due to the destruction of rangelands and its conversion to the agricultural land.