Drone-based photogrammetry for the construction of high-resolution models of individual trees

Abstract

Key message: UAV-based photogrammetry can provide detailed, accurate, and consistent structural models of open-grown trees across different species and growth forms. Environmental and ambient light conditions impact the results. Abstract: Our empirical knowledge of the complicated and highly dynamic three-dimensional structure of trees is limited. Drone-based photogrammetry offers a powerful and effective tool for capturing, analyzing, and monitoring adult trees, particularly individuals growing outside of the closed canopy forest (subsequently termed “open-grown”) in both natural and built environments. Here, we test the accuracy and consistency of high-resolution models of individual open-grown trees of three structurally distinct species. To validate model accuracy, we compared model estimates with direct measurements of branch diameter and internode length. We also examined the consistency of models of the same individual given different ambient light conditions by comparing corresponding measurements from two models of each tree. We use readily available equipment and software, so our protocol can be rapidly adopted by professional and citizen scientists alike. We found that the models captured diameter and interior internode length well (r2 = 0.87 and 0.98, respectively). Difference between measurements from different models of the same tree was less than 3 cm for diameters and interior internode lengths of most size classes. Thin distal branches were not captured well: measurements of terminal internodes in most size classes were underestimated by 50–60 cm. Models from overcast days were more accurate than models from sunny days (p = 0.0056), and a high-contrast background helps capture thin branches. Our approach constructs accurate and consistent three-dimensional models of individual open-grown trees that can provide the foundation for a wide range of physiological, behavioral, and growth studies.