Evaluation of changes in the forest environment in Guilan province using a combination of remote sensing data

چکیده مقاله

Awareness of the area and coverage of the forest can be one of the important indicators for assessing the sustainability of the forest. Replacing up-to-date methods with traditional methods in classifying and analyzing satellite images is one of the basic requirements for improving the accuracy of remote sensing results. Therefore, the aim of this study is to extract Hyrcanian forest coverage map using a combination of normalized differential vegetation index (NDVI) on a monthly and annual basis from MODIS vegetation continuous field (VCF), satellite images of Sentinel 2, Landsat 5 and Landsat 8. Supervised classification algorithms, support vector machine and random forest were used for processing. In 2010 and 2017, the support vector machine algorithm showed an overall accuracy of 71.49% and 84.56% and Kappa coefficient of 65.87% and 89%, respectively. Also, the random forest algorithm showed an overall accuracy of 82.73% and 75.04% and Kappa coefficient was 80.98% and 68% in 2010 and 2017, respectively. The results of this study showed that in the 7-year period (2010 to 2017) with the algorithm of support vector machine and random forest, respectively, 22,196 and 34,469 hectares of the forest coverage of Guilan province have been destroyed. Also, the total area of Guilan forests in 2017 based on the support vector machine algorithm equals to 562,618 hectares and with the random forest algorithm, 531,260 hectares was estimated. Considering the difference in the type of images in 2010 (Landsat-5) and 2017 (Landsat-8), the algorithms used also have different results. Therefore, the appropriate algorithm for selecting images in 2010 was RF with the Kappa coefficient of 80.98% and for 2017 was Support Vector Machine (SVM) with the Kappa coefficient of 89%. In General, the results of this research can be used in forest planning and management, improving tourism and ecotourism programs and natural resource management.