Imaging Burned Areas and Fire Severity in Mediterranean Fragmented Ecosystems Using Sentinel-1 and Sentinel-2: The Case Study of TortoliCOgliastra Fire (Sardinia)

The study aims to explore the added value of the joint use of Sentinel-1 (S1) and Sentinel-2 (S2) data for assessing burn severity in heterogeneous, fragmented ecosystems. The importance of this aim lies in the fact that for both S2 and S1 (as for all the synthetic aperture radar (SAR) C-bands), the impact of fire was found to cause ambiguous effects in complex and fragmented ecosystems. For our investigation, the effectiveness of S1 and S2 fire metrics was first statistically analyzed using ISODATA coupled with field surveys conducted for a fire that occurred on 13 July 2019 in Sardinia. Later, to automatically map burn areas and categorize fire severity, S1 and S2 fire metrics were integrated through a multilevel classification performed at a pixel and feature level. Results were successful (accuracy higher than 94%) compared with independent data sets and in situ investigations.