Semi-Supervised Forest Type Mapping in Europe on Satellite Data

Rapid climate change, humanity's need for wood, and other objective and subjective factors strongly affect the forest system of our planet. An analysis of existing global forest cover map products showed that the last map with a division into forest types (deciduous and coniferous) with a spatial resolution of 10 meters was created in 2018. Forest health monitoring requires a recent forest map. Therefore, it is important to be able to cheaply build such cards for the current year. Therefore, the actualization and creation of a modern map of European forests with a division into types of forests using machine learning is an urgent issue. This work describes the results of a forest map creation within HORIZON Europe project “Satellites for Wilderness Inspection and Forest Threat Tracking”. (SWIFTT) for the territory of Europe for 2022 based on Sentinel-l,2 satellite data of the Copernicus program with a high spatial resolution of 10 meters. The resulting map created in this work will be a reliable source of information for further analysis, management decision-making, as well as assessment and prevention of land degradation, in particular for the assessment of indicator 15.3.1 “Proportion of land that is degraded over total land area”, where deforestation is also considered under degraded land.