Remote sensing of the Earth data and the results of their thematic interpretation are of great importance in monitoring forest biodiversity. Using the example of the Kologrivsky Forest Nature Reserve, a thematic interpretation of Landsat 5 images was carried out. The best quality classification of natural objects was achieved using the Random Forest and CatBoost algorithms. For the period from 1984 to 2011, thematic maps of prevailing tree species were obtained, according to which the main directions of changes in the vegetation cover were identified.