Street and campus trees are crucial components of green infrastructure to provide various ecosystem services, including mitigation of extreme urban heat. Constructing high-fidelity 3D digital twins of urban vegetation can contribute to scientifically quantifying these services, as well as optimizing green space management. Existing models and software, such as i-Tree, simulate ecosystem services of tress with physical tree parameters, including tree height and crown related metrics. While Airborne LiDAR (ALS) efficiently captures large-scale, topdown canopy distribution, it often lacks fine under-canopy geometry. Conversely, handheld terrestrial LiDAR (TLS) captures high-density trunk and canopy details.
The ultimate goal of our research is to construct a campus tree digital twin by fusing airborne and handheld LiDAR datasets with quantifying ecosystem services. As a step toward this goal, this study aimed to 1) develop a methodology to estimate the missing crown rate based on the i-Tree guideline using terrestrial LiDAR, and 2) to present its application through a case study at the Komaba Campus of The University of Tokyo.
TLS point cloud data were collected in the university campus, and individual tree point clouds were isolated for analysis. First, tree point clouds were classified into trunk, branches without leaves, and canopy using PointNet++ deep learning model. Next, the missing crwon rate was calculated for the extracted canopy point clouds following i-Tree guideline. This was achieved by fitting Minimum Volume Enclosing Ellipsoid (MVEE) for the outer canopy boundary and Alpha Shapes for the actual canopy geometry across three 2D projection planes. ALS point cloud data were obtained from the Tokyo Metropolitan open dataset to support future multi-platform integration.
Future work will focus on validating the proposed TLS methodology across a larger sample size and diverse tree species, as well as constructing a robust ALS-TLS fused tree digital twin.