Key points are not available for this paper at this time.
Urban forests provide essential ecosystem services and are key components of urban policy-making. Both accounting for these ecosystem services and effective urban forest management planning, however, requires an up-to-date and detailed 3D tree inventory. Conducting and maintaining up-to-date tree inventories often involves field visits and manual recordings. In this study, we demonstrate the applicability of close-range remote sensing in measuring and monitoring tree characteristics over time. A mobile LiDAR system was deployed to collect dense point clouds (1789.92 points/m²) in the city of Hasselt, Belgium. The proposed individual tree detection algorithm obtained a recall score of 0.84, a precision of 0.82 and an F-score of 0.83. The average estimated characteristics were: tree height (H) of 9.84 ± 2.68 m, diameter at breast height (DBH) of 0.48 ± 0.74 m, crown projection (PA) of 14.87 ± 11.60 m² and crown volume (AV) of 73.59 ± 98.42 m³ . For a subset of 47 manually measured trees (r)RMSE values were 0.58 m (259 %) and 0.96 m (10 %) for DBH and H, respectively. Excluding incorrectly segmented trees mainly improved (r)RMSE values of DBH (0.05 m, 21 %) and only had a minor effect on H (0.95 m, 11 %). Two case studies based on multi-temporal MLS-datasets allowed to estimate the volume of pruning biomass and to detect a nest of Asian hornet ( Vespa velutina ). The obtained results can support both policy-making and operational planning by integrating 3D tree inventories and assessing the effect of management activities. • The proposed method identified a total of 464 trees, including 379 true positives. • Stem diameter estimates are more error-prone than tree height estimates. • More advanced segmentation can improve diameter estimation. • MLS time series can support urban tree management and monitoring.
Ottoy et al. (Tue,) studied this question.