Abstract:

The aim of this study is to spatially assess fire severity and to explore the relationship between the Normalized Vegetation Index (NDVI) derived from WorldView- 2 imagery and the field-based fire severity index “Composite Burn Index” (CBI) on a recent large fire that occurred on the island of Chios, Greece, in 2016. The statistical analysis indicated a relatively strong linear relationship between the NDVI and the CBI (R2 = 0.71). Predictive local thresholds of NDVI values have been determined for accurate thematic classification and mapping of fire severity classes. The overall classification accuracy based on NDVI from WorldView-2 was relatively high (71%), pointing to the potential of using such data for mapping forest fires in the Mediterranean as well as for assessing their severity. The application of the classification thresholds calculated in this study over other forest fire events on similar ecosystem types, could support the rapid assessment of fire severity and hence could enable the decisions needed for proper post-fire management.