Comparative performance of dNBR and dNDVI for burn severity mapping via large wildfires of 2021 in Türkiye

Background The large wildfires that occurred in 2021 in the southwestern coastlines of Türkiye, characterized by typical Mediterranean ecosystems, caused significant damage to forests and shrublands. Purpose Studies on burn severity classes using remote sensing data have been increasingly conducted in recent years. Methods Maps related to burn severity classes were produced using differenced normalized burn ratio (dNBR) and differenced normalized difference vegetation index (dNDVI) derived from Landsat 8 and Sentinel-2 satellite images for the mega wildfires that happened in Marmaris and Manavgat regions of Türkiye in 2021. Additionally, Overall Accuracy (OA) and Kappa Value (KV) metrics were calculated for the burn severity maps. Key results The OA (%) and KV for the dNBR index derived from Landsat 8 and Sentinel-2 images for the Manavgat fire was calculated as 93%–0.91 and 91%–0.88, and the dNDVI index was 89%–0.85 and 88%–0.84, respectively. For the Marmaris fire, OA was calculated as 92%–0.89 and 91%–0.88 for the dNBR index and 89%–0.85 and 86%–0.81 for the dNDVI index, respectively. Conclusions The dNBR index gave better results than the dNDVI index in accurately determining burn severity classes. Implications Field-validated dNBR maps can provide forest managers with a reliable and cost-effective basis for rapidly identifying burn severity patterns.