Studies of the Usability of Satellite Images in the Identification of Forest Stands in Slovakia

This article was created on the basis of the Silvanus (H2020) and Hysped (ERDF) projects. Its aim was to examine the possibilities of using satellite images for the identification of forest types in Slovakia as an alternative to hyperspectral images. 40.6% of the total area of Slovakia is formed by forest stands. The paper describes the methodology of creating the data set based on the FORESTRY INFORMATION SYSTEM of Slovakia and Sentinel 2 image on selected parcels in Poľana protected landscape area of Slovak country. We discuss the pre-processing of the obtained data in order to achieve optimal results. Several algorithmic approaches were used for the purpose of classification (namely, SVM, RF, and CNN) in four distinct approaches. Based on the presented results, the best combination is the pixel-based classification approach with binary classification and SVM or RF models.