Forest covers more than 40% of the alpine region and provide a range of ecosystem services from protecting settlements, infrastructure and slopes from avalanches, debris flows and erosion to capturing and storing carbon to preserving biodiversity as well as providing timber and social benefits.
Forest cover changes in the Alps are driven by factors related to climate, land-use and natural disturbances. Mapping forest extent changes is not only essential for forest services guiding appropriate management interventions and planning activities in these complex environments, but also to allow a better understanding of patterns and drivers of changes as well as the interactions among them.
High resolution forest cover change maps are derived using Sentinel-2 images employing bi-temporal change detection approaches. Sentinel-2 level 1C data are atmospherically corrected using the sen2cor processor and annual cloud-free composites are generated by considering acquisitions from the period of June to August. To represent initial woodland cover extent, a forest/non-forest cover map is generated for 2015 by employing a machine learning classifier and training samples from the Copernicus High Resolution Forest Layer. Forest cover change maps are then derived by comparing annual cloud-free composites using a change vector analysis. Forest change maps are subsequently postprocessed to filter out small-scaled patches and to reduce false negatives. Validation is performed by combining visual interpretation with a random sampling strategy.