Accurate, sub-canopy forest inventory is essential for sustainable forest management, carbon accounting, and automated mechanical thinning prescriptions, yet traditional field surveys remain labor-intensive and remote sensing (e.g., airborne LiDAR, satellite imagery) struggles to resolve detailed understory structure beneath closed canopies. The CUTMAP (Comprehensive Utility for Thinning Mechanically via Automated Prescription) project addresses this bottleneck by deploying an edge-computed, multi-sensor mobile mapping payload—integrating non-repetitive scanning LiDAR (Livox Mid-360), high-rate inertial odometry (FAST-LIO2), global shutter stereo vision, and GNSS georeferencing on an NVIDIA Jetson edge platform.
This research at Harvard Forest aims to evaluate and benchmark the system’s sub-canopy localization, 3D point cloud accumulation, and automated tree mensuration algorithms across representative temperate forest stands with varying species compositions, stem densities, and canopy closures. Field activities involve non-destructive pedestrian traversals along established trails and transects to generate dense, georeferenced 3D point clouds. Automated algorithms normalize digital terrain models, segment individual tree stems, and perform RANSAC circle fitting to extract Diameter at Breast Height (DBH), tree height, lean, presence of defects, and geospatial mapping. System performance will be validated against manual forestry tape measurements and permanent plot records. This work advances non-destructive, rapid digital forest inventory and provides scalable ground-truth data for forest ecology and precision management.