Application of the BNDVI index as proxy for burn severity assessment

Abstract

Burn severity assessment is of paramount importance to evaluate the likelihood of post-fire debris flows and landslides. These processes are more likely to occur in steep landscapes burned with high severity, where large amounts of white ash may result from the complete combustion of surface fuels. Ash type and cover (extent) are among the most important indicators to discriminate burn severity degrees, if timely and accurately mapped. In this work, the capability of the Blue Normalized Difference Vegetation Index (BNDVI) calculated from PlanetScope imagery is tested for the rapid mapping of white and black ash at fine spatial scale, and for using this information as proxy for burn severity assessment. The test is performed in Sicily, southern Italy, where a wildfire burned an area of about 1300 hectares on 25 July 2023. Four BNDVI maps are developed in the time span from 4 August 2023 to 17 February 2024. Two independent regression analyses are conducted to assess statistical relationships between the BNDVI index and burn severity represented by the widely used dNBR index. Results suggest a strong relationship between BNDVI and dNBR, as highlighted by Pearson correlation coefficients ranging between 0.76 and 0.88, p-value < 0.001, and a fair agreement between areas truly covered by white ash (i.e., low BNDVI) and those burned with high severity (i.e., high dNBR). In addition, the comparison of multi-temporal BNDVI maps reveals both spatial and temporal evolution of the ash cover and vegetation regrowth in the analyzed time interval.