As the European Forest Disturbance Atlas suggested a bark beetle outbreak in the Bavarian Forest, I aimed to classify changes in the land cover between 2020-2023 in that area. For this, I created three summer composite images and trained the Random Forest algorithm on the identified stable and transition classes. After validation, I found that stable coniferous forests, continuously infested forests and forests changing from healthy to infested were less likely to be misclassified than other classes. The classification showed that 8% of the study area had become newly attacked by bark beetles between 2020 and 2023.
False colour composites can be created by combining different bands. Here Near Infrared (NIR) is added since vegetation have a very high NIR reflectance. A true colour composite is shown in comparison in a) and b). Based on such a remote sensing image one can conduct a land cover classification.Â
The active fire progression in Ljusdal, Sweden was derived from Sentinel-3a images using SNAP. Even though the satellite image has relatively low resolution, it can be used to quickly get an indication about direction and extent of fire progression.