Application of unmanned aerial system structure from motion point cloud detected tree heights and stem diameters to model missing stem diameters

Monitoring of tree spatial arrangement is increasingly essential for restoration of dry conifer forests. The presented method was developed for high-density point clouds, like those from unmanned aerial system imagery, to extract and model individual tree location, height, and diameter at breast height (DBH). Extraction of tree locations and heights uses a variable window function searching point cloud-derived canopy height models. Tree DBH is extracted for a subset of point cloud trees using a slice at 1.32-1.42 m and a least-squares circle fitting algorithm. Extracted heights and DBHs are spatially matched and filtered against each tree's expected DBH predicted using a regional National Forest Inventory height to DBH relationship. Values remaining after filtering are used to create a site-specific height to DBH relationship for predicting missing DBH values. Applying the method in a ponderosa pine-dominated forest found that extracted height values exceeded the precision of field height measurement approaches, while the accuracy of extracted and modeled DBH values had a mean error of 0.79 cm. • Leveraging National Forest Inventory to filter DBH values eliminates the need for in situ observations. • Produces tree list for all extractable stems in the point cloud. • Transferable to high-density point clouds in open-canopy forests.