European Wide Forest Classification Based on Sentinel-1 Data

The constellation of two Sentinel-1 satellites provides an unprecedented coverage Synthetic Aperture Radar (SAR) data at high spatial (20 m) and temporal (2 to 6 days over Europe) resolution. availability dense time series enables the analysis SAR signatures exploitation these for classification purposes. Frequent backscatter observations allow derivation temporally filtered that reinforce effect changes in vegetation phenology by limiting influence short-term related environmental conditions. Recent studies have already shown potential multitemporal forest mapping, type (coniferous or broadleaved forest) as well phenological variables local national scales. In present study, we tested viability a recently published multi-temporal method continental scale mapping applying it Europe evaluating derived tree cover density maps against European-wide Copernicus High Resolution Layers (HRL) datasets national-scale from twelve countries. comparison with HRL revealed correspondence majority European continent overall accuracies 86.1% 73.2% forest/non-forest maps, respectively, Pearson correlation coefficient 0.83 map. Moreover, evaluation both showed obtained are almost within range datasets. average 88.2% 82.7%, compared 90.0% 87.2% This result is especially promising due facts can be produced degree automation only single year required opposed updated every three years.