Methods for fire severity assessments with remotely piloted aircraft systems

Abstract

Fire severity assessments reveal fire’s lasting damage to grassland soil. New methods in remotely piloted aircraft systems (RPASs) and structure-from-motion processing allow for high-resolution models capable of characterising fine details in soil fire severity. This research used a low-flying RPAS to capture pre- and post-fire images and generate models of soil depth-of-burn (DoB) at Ya Ha Tinda Ranch, Alberta, Canada. Results show that RPAS-derived DoB measurements are highly accurate compared to those in-situ when vegetation is absent in pre-fire imagery (RMSE = 1.48 cm) and when it is intact (RMSE = 2.57 cm). The technique developed here for filtering ground surface beneath vegetation is applicable anywhere a bare-earth model is required and should be further studied in variable terrain with multiple flying altitudes. This fire severity assessment method can be used to augment and corroborate established multispectral analysis techniques while replacing subjective and time-consuming severity ground-truthing protocols.