Influence of voxel size and point cloud density on crown cover estimation in poplar plantations using terrestrial laser scanning

Abstract

Accurate estimates of crown cover (CC) are central for a wide range of forestry studies. As direct measurements do not exist to retrieve this variable in the field, CC is conventionally determined from optical measurements as the complement of gap fraction close to the zenith. As an alternative to passive optical measurements, active sensors like Light Detection and Ranging (LiDAR) allows for characterizing in situ the 3D canopy structure with unprecedented detail. We evaluated the reliability of terrestrial laser scanning (TLS) to estimate CC using a voxel-based approach. Specifically, we tested how different voxel sizes (5-20 cm) and voxel densities (1-9 points/dm 3 ) influenced the retrieval of CC. Results were compared against benchmark values obtained from DCP. The trial was performed in hybrid poplar plantations in Northern Italy.  Results indicate that TLS can be used for obtaining accurate estimates of CC, but the choice of voxel size and point density is critical for achieving such accuracy. in hybrid poplars, the best performance was obtained using voxel size of 10 cm and point density of 8 points/dm 3 . The combined ability of measuring and mapping CC also holds great potential to use TLS for calibrating and upscaling results using coarser-scale remotely sensed products combined ability of measuring and mapping crown cover also holds great potential to use TLS data for calibrating and upscaling results using coarser-scale remotely sensed products.