Methodology for burned areas delimitation and fire severity assessment using Sentinel-2 data. A case study of forest fires occurred in Spain between 2018 and 2023

Abstract

The objective of this article was to develop a methodology for the burned areas delimitation and fire severity assessment in forest fires occurred in Spain between 2018 and 2022. As input data, this study was based on the use of Sentinel-2 spectral indices, which are characterized by having spectral bands in the near-infrared (NIR) and short-wave infrared (SWIR) spectral regions, allowing a high distinction between burned and unburned areas, and between different fire severity degrees too. All possible combinations between Sentinel-2 bands applied to a spectral normalized difference index (SP) were analyzed, along with the most commonly used burn spectral indices in remote sensing as the Burned Area Index (BAI), the Burned Area Index for Sentinel-2 (BAIS2), the Mid-Infrared Burn Index (MIRBI), the Normalized Burn Ratio (NBR), the Relativized Burn Ratio (RBR) and the relative differential Normalized Burn Ratio (RdNBR). In addition, in order to delete confusions between burned area and the presence of other land cover areas, the Sentinel-2 Global Land Cover (S2GLC 2017) and the temporal differences between pre-fire and post-fire dates were obtained for each spectral index (dSP). The results were compared by: in the case of burned areas, the Emergency Mapping Service (EMS) and the Galicia forest service; in the case of fire severity, using field plots classified as in Ruiz-Gallardo et al. (2004) study (null, low, moderate and high severity). The final statistic results obtained showed that the dNBR2 spectral index (using B11 and B12 Sentinel-2 spectral bands) provided the highest results of burned area delimitation (7% of commission error and 3% omission error, respectively) whereas, the combination of the BAIS2, the NBR and the modified Normalized Burn Ratio (NBR2, using B7 and B12 Sentinel-2 spectral bands), used in areas with low, mix and full vegetation respectively, provided the highest results in fire severity assessment (kappa statistic, F1-score and Balanced Accuracy equal to 0.87, 0.86 and 0.92, respectively). The methodology developed in this work allows obtaining accurate maps of burned area and fire severity in Spain, contributing to the reinforcement of national forest fires statistics.