Under-Canopy Biomass Sensing using UAS-Mounted Radar: a Numerical Feasibility Analysis

Accurate forest biomass estimation is crucial for effective wildfire risk assessment and management as well as in agricultural applications. However, existing remote sensing methods cannot estimate under-canopy biomass. This paper investigates the feasibility of a novel approach that leverages ultra-wideband radars mounted on uncrewed aerial systems in conjunction with reference ground reflectors to estimate the under-canopy biomass profile. Through extensive electromagnetic wave propagation simulations encompassing diverse material configurations, we investigate the sensing capabilities of our proposed technique. We leverage deep learning to effectively extract biomass information from radar signals, enabling the prediction of material properties and geometries. The developed deep learning system was successfully trained with an R
²
metric of 98% in testing and produced dielectric permittivity distributions that visually match the target configurations. The results underscore the potential of our approach in advancing wildfire management strategies, bolstering carbon sequestration efforts, preserving forests, and aiding agricultural applications by facilitating accurate biomass characterization and estimations of crop yield and soil moisture.