Remote sensing for classification of different forest types in Google Earth Engine

Habitat encompasses the environments where living organisms sustain their lives and carry out various activities. Habitats, vital for the existence and continuity of biological species, play a crucial role in the healthy functioning of ecosystems. This study is conducted with the aim of developing an advanced methodology for mapping tree species in forest ecosystems. Employing the Random Forest algorithm, this approach focuses on classifying different tree species, a significant component of biodiversity within the forest ecosystem of North Macedonia, by combining Sentinel-1 SAR and Sentinel-2 Optical satellite images. In this study, optical and radar data are integrated using NDVI, LAI, EVI, MSAVI, NDWI, LCI, PRI, ChlCI indices. The results obtained in the examination of five different forest habitat classes yield a general accuracy rate of 90.1% and a kappa value of 0.874. Additionally, the SSI index calculated for each band and class on the Sentinel-2 satellite image allows for a more detailed evaluation of the spectral similarity of the results obtained in the classification process. By introducing a novel method for monitoring and managing biodiversity in forest ecosystems, this study emphasizes how remote sensing techniques can effectively contribute to forestry applications.