Change detection of Oak forests using object-based classification of multitemporal Landsat imageries (Case study: forests of the northern province of Ilam)

Oak is the dominant forest species in the Zagros mountains, it is seen as a semi dense forest in western provinces. Multiple droughts, development of cities and villages, land use change from forest to farmland, are the main causes of forest destruction in Ilam. The purpose of this research is to determine the change detection of forest land to other land uses, using multi temporal TM sensor images (April 1986) and OLI (April 2015). In object-based classification, the data segmentation as initial phenomena, with appropriate scale and compression parameters for each image, was performed. Two classifying phenomena, nearest neighbor method and fuzzy functions were used. Land use classes were prepared by specifying training points and using the normalized difference vegetation index (NDVI). The Overall accuracy of the TM image was 88% and OLI image 94%, and the kappa coefficient for the TM and OLI image was calculated 84% and 91%, respectively. By post classification comparison method, the change matrix for each districts and changes forest to other uses was formed. The results showed that all districts faced with the phenomenon of deforestation and forest wide change (-42%) have been made.