Estimation of Forest Aboveground Biomass from Derivatives of Vegetation-Structure Profiles

• Published in: IEEE International Geoscience and Remote Sensing Symposium proceedings pp. 2442 - 2445
• Main Authors: Treuhaft, Robert, Cushman, K.C., Hensley, Scott, Pinto, Naiara, Stocker, Olivier, Hawkins, Brian, Lavalle, Marco, Chen, Richard
• Format: Conference Proceeding
• Language: English
• Published: IEEE 07.07.2024
• Subjects: Biomass; Estimation; Forest vegetation structure; Forestry; Fourier transforms; Laser radar; lidar remote sensing; radar remote sensing; Tomography; Vegetation mapping
• ISSN: 2153-7003
• DOI: 10.1109/IGARSS53475.2024.10641220

Several studies have found that the vertical Fourier transform of lidar, interferometric Synthetic Aperture Radar (SAR), and stereo photogrammetric profiles at empirically-determined spatial frequencies enables high-performance forest aboveground biomass (AGB) estimation. Linear combinations of real and imaginary parts of Fourier transforms of Tomographic (multi-baseline) SAR (TomoSAR) profiles, from Uninhabited Aerial Vehicle Synthetic Aperture Radar (UAVSAR) airborne data, generate ~20%-precision estimates of AGB in the Saskatchewan area of Canada. We found that this 20% precision can be improved to ~15%, a factor of 30% improvement in root mean square error (RMSE) if, in addition to using Fourier transforms of the profile itself, we use Fourier transforms of the spatial, vertical derivative of the profile. The formulation of this "derivative" algorithm is the subject of this paper.