Abstract: Shrub volume is a key ecological indicator in rangeland ecosystems. Volume is important as it closely relates to biomass and other indicators such as fuel loading, wildlife habitat, site productivity, and ecosystem structure. Biomass has been traditionally derived in the field using three main techniques: destructive sampling, ocular estimates, or allometric techniques, which estimate volume using shrub height and canopy widths and then translate volume to biomass with species-specific allometric equations. These techniques each have their own set of challenges, making neither of them ideal, especially due to their tedious and time intensive field requirement and limitations for re-sampling. We examined the suitability of using structure-from-motion derived point cloud generated from drone-captured images to estimate canopy volume for seven dominant shrub species within mountain big sagebrush (Artemisia tridentata subsp. vaseyana) plant communities in southern ID, USA. Canopy height and two perpendicular width measurements were made on a sample of 103 shrubs representing the size range present at the study area, and canopy volume was estimated using a traditional allometric equation. Then, overlapping aerial images collected with a DJI Mavic 2 Professional drone were used to create a 3-dimensional (3D) representation of the study area using structure-from-motion photogrammetry. Each individual shrub was extracted from the 3D photogrammetric point cloud, and canopy volume was estimated in CloudCompare using both allometric and volumetric methods. For the allometric method, shrub height and canopy widths were directly measured by an observer using CloudCompare’s measurement tool. For the volumetric method, volume was calculated by converting the point cloud to a raster canopy height model with 2.5 and 5 cm grid cells. Volumetric methods to estimate canopy volume performed better than allometric methods (R2 > 0.7) and were overall easier to reproduce and more robust to user-related variability. Drone-estimated volume best matched field-estimated volume (R2 > 0.9) for three larger species: A. tridentata subsp. tridentata, A. tridentata subsp. vaseyana, and Purshia tridentata. Volume of smaller shrubs (canopy widths < 1 m) was slightly but consistently overestimated from drone-based models. These results demonstrate the viability of using drone-collected images to assess shrub canopy volume for at least five upland sagebrush steppe shrub species and support efforts to integrate drone data-collection into rangeland vegetation monitoring. Further, this process could be automated by employing crown detection algorithms and programmed volume estimation workflows.
