Voxel-based Urban Vegetation Volume Analysis with LiDAR Point Cloud
The 3D volume and spatial distribution of urban vegetation are highly related to the delivery of multiple ecosystem services. However, due to the intricate vegetation structure, little research has been conducted to visualize and model the 3D spatial structure of urban vegetation. This study proposes an automated voxel-based modeling method to visualize and quantify the urban vegetation volume with LiDAR point cloud and performs a case study of the No.6 Middle School campus in Hengyang City, Hunan Province, China. The PointCNN model is used to perform semantic segmentation of the LiDAR data to extract the tree points. Then the points are voxelized into a 3D volume model with 1m×1m×1m cells. The result shows that the total vegetation volume of the area is 61,192m³, accounting for 37.28% of the total voxelized study area. The green space in front of the north teaching buildings has the largest proportion of vegetation volume, 19,366m³, accounting for 68.37% of the vegetation volume of the whole campus, due to the diverse vegetation and complex structure. The automated segmentation voxel modeling process could provide an efficient way to represent the spatial distribution of urban greenery. With an adjustable voxel size, the model could be adapted to various scales from regional to neighborhood. The model could also be used to analyze the green space structure at the human scale, as well as the interactions between green space and the surrounding environment, and to provide spatial data for the evaluation of multiple ecosystem services.
Zhang, Wei; He, Ziqi; and Li, Xin
"Voxel-based Urban Vegetation Volume Analysis with LiDAR Point Cloud,"
Proceedings of the Fábos Conference on Landscape and Greenway Planning: Vol. 7:
1, Article 67.
Available at: https://scholarworks.umass.edu/fabos/vol7/iss1/67