Spatio-Temporal Assessment of Fire Severity and Vegetation Recovery Utilising Sentinel-2 Imagery in New South Wales, Australia

Abstract

Fire severity and vegetation indices are widely used to understand fire disturbance and vegetation recovery in heterogenous landscapes. Multispectral sensors are used to understand fire severity and post-fire vegetation dynamics. We utilised four fire severity (dNBR, RBR, RdNBR and MIRBI) and four vegetation indices (Chlorophyll Index of Red-Edge, dNDVI, EVI and NDRE) derived from Sentinel-2 images to understand the spatio-temporal variability of fire severity and vegetation recovery. The Masonite Road fire burned 2321.32 hectares of mixed Eucalyptus and wetland forests in the southwestern Telligerry State Conservation Area of New South Wales, Australia, in January 2018. Spectral and spatial analysis of different fire severity and vegetation indices revealed the blaze affected the landscape with moderate to high level of severity. Fire severity indices showed similar patterns in spatio-temporal analysis for a year, but Chlorophyll Index of Red-Edge and EVI did not show any pattern of fire impacts on vegetation. Among the analysed indices, two combinations of fire severity and vegetation indices (dNBR or RBR or RdNBR with NDVI and MIRBI withNDRE) were significantly correlated and can be effective in assessing fire severity, burned area and vegetation recovery.