Mapping Soil Burn Severity at Very High Spatial Resolution from Unmanned Aerial Vehicles

The evaluation of the effect burn severity on forest soils is essential to determine impact wildfires a range key ecological processes, such as nutrient cycling and vegetation recovery. main objective this study was assess potentiality different spectral products derived from RGB multispectral imagery collected by unmanned aerial vehicles (UAVs) at very high spatial resolution for discriminating variations in soil after heterogeneous wildfire. In case study, we chose mixed-severity fire that occurred northwest (NW) Iberian Peninsula (Spain) 2019 affected 82.74 ha covered three types forests, each dominated Pinus pinaster, sylvestris, Quercus pyrenaica. We evaluated field 1 month using Composite Burn Soil Index (CBSI), well pool five individual indicators (ash depth, ash cover, fine debris coarse unstructured depth) easy interpretation. Simultaneously, operated an vehicle obtain postfire images, allowing deriving six indices. Then, explored relationship between indices metrics means univariate proportional odds regression models. These models were used predict CBSI categories, classifications validated through confusion matrices. Results indicated outperformed when assessing severity, being more strongly related than indicators. Normalized Difference Water (NDWI) best-performing index modelling (R2cv = 0.69), showing best ability categories (overall accuracy 0.83). Among depth one achieved results, specifically it modelled NDWI 0.53). This work provides useful background design quick accurate assessments be implemented immediately fire, which factor identify priority areas emergency actions fires.