Short-term assessment of burn severity using the inversion of PROSPECT and GeoSail models

Accurate estimations of burn severity and its distribution in post fire scenarios are critical for short-term mitigation and rehabilitation treatments. The use of remote sensing techniques, coupled with radiative transfer models (RTMs) can improve the accuracy, precision (in terms of number of classes) and cost-effectiveness of burn severity assessment. In this paper, an improved simulation model that combines PROSPECT and GeoSail to estimate burn severity from satellite data was tested in three Mediterranean forest fires. The determination of burn severity was based on a new version of the CBI index (named GeoCBI), that takes into account the vegetation fraction cover (FCOV) to compute burn severity of the total plot. Model inversion results showed accurate estimations of GeoCBI values (RMSE between 0.18 and 0.21) and a uniform performance in all three sites (107 field plots in total) throughout the full GeoCBI range (0-3).