Tracking 3D forest density dynamics in a mixed temperate forestusing occlusion-aware voxel transmittance and consistent multitemporalUAV-LiDAR

Abstract

Forest structure and density dynamics drive ecosystem processes including carbon sequestration, microclimate regulation, and photosynthetic capacity. While LiDAR can capture these properties, high-frequency monitoring remains constrained by scarce multi-temporal datasets, inconsistent acquisitions, and occlusion biases confounding true structural change with methodological artifacts. This study investigates spatio-temporal forest structure dynamics through voxel transmittance in a European mixed temperate forest. We acquired an extensive UAV LiDAR dataset of 38 flights with identical sensor settings over more than two years. With a custom ray-tracing framework, we estimated voxel transmittance and mapped occlusion at 0.25m resolution. Multi-temporal comparability was ensured through consistency masking and leaf-off baseline initialization of empty space. Voxel transmittance and attenuation dynamics were analyzed across three areas of interest, covering a mixed plot, a beech and a Douglas fir patch. The results revealed distinct phenological signatures: European beech showed high-amplitude seasonal variation (70% attenuation volume reduction from summer to winter) with rapid spring foliation and stable winter baselines, whereas Douglas fir exhibited 12.7% variance with delayed growth onset and no stable winter baseline. Inter-annual comparisons indicated increasing blocking biomass with upward vertical shifts in transmittance profiles from tree growth and crown expansion. Two-dimensional attenuation mapping enabled the detection of individual tree dynamics and discrete structural disturbances. We demonstrate that consistent high-frequency UAV LiDAR acquisitions enable forest morphology monitoring through relative voxel transmittance. Our occlusion-aware framework differentiates structural changes from sampling biases, providing insights into forest dynamics at high spatio-temporal resolution.