Enhanced Simulation of Radar Backscatter From Forests Using LiDAR and Optical Data

Focusing on a forest dominated by Poplar Box (Eucalyptus populnea) near Injune in Queensland, Australia, light detection and ranging (LiDAR) and optical remote sensing data are integrated with tree- and stand-level information to parameterize a coherent L-band synthetic aperture radar (SAR) imaging simulation that models microwave penetration and interaction with the canopy, understory, and ground. The approach used LiDAR data to generate a three-dimensional representation of the distribution of tree components (leaves and small branches) by species (based on 1-m³ voxels) and the ground surface. Tree trunks were mapped across a 7.5-ha forest stand using a LiDAR-derived height-scaled crown openness index. Primary and secondary branches were modeled as tapering cylinders and linked the canopy voxels to the LiDAR trunks. The dimensions of vegetation and soil components and their geometric and dielectric properties required by the model were calibrated with field-based measurements. Visual and numerical comparison between NASA JPL Airborne SAR data and the model simulation suggests the effective modeling of SAR imagery at L-band. The study provides a proof-of-concept approach for integrating LiDAR data in the parameterization of coherent SAR simulation models, and the model presents options for better understanding of the information content of SAR data in real forest situations