The viability of LiDAR methods to derive urban tree health parameters

By Jesse Brand

Abstract
This thesis explores the use of LiDAR data to derive tree health parameters in an urban context. The purpose is to establish whether LiDAR data can effectively be used in the development of a key performance indicator for urban tree vitality.

Tree point clouds were collected through terrestrial LiDAR scanning, making use of the RIEGL VZ-400. Two tree species were studied: Tilia cordata and Gleditsia triascanthos. For both species a selection was made of stressed trees, located on the pavement, and healthy trees, located in grass. The point clouds were processed in RiScan Pro and CloudCompare. R was used for further analysis.

From the tree point clouds the metrics diameter at breast height, height, canopy area and tree volume were derived. Furthermore, a voxelisation method also delivered a voxel-based volume, as well as crown density visualisations. The parameters were statistically tested for differences between grass and pavement trees. Overall, grass trees tended to have higher values in most metrics, corresponding to scientific consensus, but no significant differences between grass and pavement were found. The study was limited by its limited sample size and an unsuccessful wood/leaf separation.

It can be concluded that LiDAR can be of use in an urban tree setting and can be an effective addition to the data needs of the KPI project. However, it is important to keep in mind the time needed for data collection and processing. Furthermore, a successful method for wood/leaf separation is highly beneficial.

Keywords: Light detection and ranging (LiDAR); Terestrial LiDAR scanning (TLS); Urban green; Voxels