Investigation of the spatial pattern of forest cover changes using logistic regression in Malekshahi

In order to investigate the spatial pattern of forest cover changes for the county of Malekshahi in Ilam province during 1976 to 2014, logistic regression modeling technique were used. Logistic regression model is able to well determine the relationship between forest cover change (dependent variable) and its influencing factors (independent variables).
To survey the forest cover changes, the images of Landsat 1 satellite from MSS sensor relating to 1976 and the images of Landsat 8 from OLI sensor relating to 2014 were processed and classified. The surveyed images were classified into two classes of forest and non forest. To investigate the causes of degradation, forest cover changes map with physiographic and human spatial variables were integrated into model.
The evaluation of regression model with ROC (0.8210) and Pseudo-R2 (0.2514) indices indicated the ability of the model to describe the changes and to identify the areas prone to change. The results showed that approximately 26491.9 ha of forest areas were reduced during 38 years.
According to the results of modeling, it was found that altitude, soil depth, distance to the forest edge, aspect, slope, distance to drainage, population distribution, distance to the road and village had the greatest impact on the reducing of forest cover in the county of Malekshahi, respectively. The results of study can be considered in the future planning for the area that are related to the vegetation / land use changes.