Modeling of forest cover change to identify suitable areas for project REDD (case study: Lordegan county)

Extreme land use/ land cover change, especially deforestation, is one of the environmental challenges in developing countries. Reducing Emissions from Deforestation and forest Degradation (REDD) projects as a strategy will prevent of deforestation trend and greenhouse gas emissions into the atmosphere. In recent decades, there has been a significant reduction in the Zagros forest. In this regard, in the present study, the forest cover changes of Lordegan county located in Chaharmahal Bakhtiari province were examined using forest cover maps produced by Landsat 5 and 8 satellites belonging to 1998, 2008 and 2018. Then, transition potential modeling was performed using two models of artificial neural network and logistic regression, and for validation the ROC and figure of merit were applied. Finally, using overlapping the maps of deforestation probability and carbon resources, the suitable areas for REDD projects were identified. The results of change detection showed that during 1998-2008 and 2008-2018, 17,256 ha and 20,553 ha of forest cover were degraded, respectively. The validation results showed that the logistic regression gained the ROC equal to 0.9752 and the figure of merit equal to 19.01%, and had a better performance than the artificial neural network. Also, based on overlap map, areas with high deforestation probability carbon content above 70 tons per hectare were proposed for REDD. The findings of this study show that using the presented methodology can be identified the areas with deforestation and can be prevented the release of greenhouse gas into the atmosphere.