Design and evaluation of a cloud-oriented procedure based on SAR and Multispectral data to detect burnt areas

Abstract This article presents the design and an experimental evaluation study of a novel procedure based on the synergistic use of SAR and Multispectral data to detect burnt area over vegetated natural surfaces. The procedure is designed to take advantages of open satellite datasets and cloud computing services. The underpinning data processing workflow exploits the variation of the backscattering coefficients and the spectral signature of surfaces affected by fire damages by applying a threshold-based technique optimized for different land cover classes. The presented experimental study focuses on three large wildfires occurred over Europe during the last four years. By comparing the burnt areas detected by the procedure with data from the European Forest Fire Information Service, we obtained an overall accuracy higher than 0.88 for all the considered test cases. The presented data also include various metrics that allow to compare the results achievable by using in synergy SAR and Multispectral data with respect to the individual use of them. Overall, the results of our study show that the presented procedure, and more in general the exploited design approach, can be of interest for researchers and practitioners for the development of efficient automated solutions for the detection of burnt areas.