Spatial forest disturbance modeling using landscape metrics in Golestan national park of Iran

The aim of this study is to determine the status of forest spatial disturbance in Golestan National Park using remote sensing and landscape metrics. Detailed forest/non-forest map was prepared by on-screen digitizing method using Landsat 8 data. After determining the optimal area of gridding, the studied area was divided into 550 hexagonal zones with the area of 100 hectares each one. Spatial distribution, connectivity and composition of non-forest patches in each zone were measured using 12 spatial metrics (MESH, NP, PD, ENN-MN, LPI, FRAC-MN, CONTIG_MN, SPLIT, AREA-MN, DIVISION, PLAND and SHAPE-MN). After standardization of metrics values, principal component analysis was also performed in order to determine the effective metrics in disturbance and their associate weight in the spatial disturbance model. Afterwards, based on the prepared model, the disturbance index map was prepared and classified in four classes of disturbance including 1) without disturbance, 2) low disturbance 3) medium disturbance and 4) high disturbance. The results showed that the mentioned classes cover 48.53, 16.15, 19.90 and 15.41 percentage of the studied area, respectively. With the approach that used in this study, the disturbances map of the under management forest areas can be prepared periodically and used as one of its monitoring tools.