Burned Area and Fire Severity Prediction of a Forest Fire Using a Sentinel 2-Derived Spectral Index in Çanakkale, Turkey

Gökkaya K.

Turkish Journal of Bioscience and Collections, vol.6, no.2, pp.37-44, 2022 (Peer-Reviewed Journal)


Objective: The objective of this study was to determine the extent and degree of severity of a burned area resulting from a forest fire using Sentinel 2 remote sensing data in Çanakkale, Turkey within the Mediterranean Basin, an area of the world where forest fire occurrence and severity are increasing.
Materials and Methods: Pre and postfire Sentinel images were obtained. The Normalized Burn Ratio (NBR) index was calculated for each scene. Then the difference NBR (dNBR) was calculated by subtracting the postfire NBR from the prefire NBR. dNBR ranges were classified into fire severity categories. A map with 20 m spatial resolution displaying the burned area and fire severity was generated from the classified dNBR image. Finally, a forest stand map of the burn area was laid over the fire severity map to examine the relationship between fire severity and stand and cover types.
Results: Approximately 1400 ha of area was predicted to have been burned. Twenty nine, 21, 42, and 8% of the burned area was identified as low, moderate low, moderate high, and severely burned using the dNBR index, respectively.
Conclusions: The overlay of the stand map on the burn severity map revealed that the forested areas were more severely burned compared to the agricultural sections. dNBR is an effective index to delineate fire area extent and identify fire severity. Sentinel 2 data provide a fast and accurate means to monitor forest fire extent and severity due to its improved spatial and temporal resolution.