The advantage of point cloud derived tree modelling on urban greenery maintenance: Shortlisting dangerous trees, assessing ecosystem services

Abstract

Trees and greenery are the bedrock of a liveable, healthy municipality. Trees need continuous monitoring and maintenance to fit in and nourish in urban environments, to maximize the benefits they provide to their surroundings through provisioning-, cultural-, supporting- and regulating ecosystem services, and to minimize the damages they could potentially cause through falling or branches breaking off. Digital tree inventories are necessary to be able to retain and track the ever-changing condition of the trees. Information for these inventories is collected through regular manual or digital field surveys. This thesis compared the speed, cost, accuracy and usability of these methods through an empirical example of the measurement and digitization process of 134 selected trees in a canyon in Budapest, Hungary. The gathered differing information was used as an input in the i-Tree Eco software to attain and highlight the differences in the monetary valuation of certain regulating ecosystem services provided by the trees. The thesis found that a terrestrial LiDAR created point cloud could bear the necessary information with otherwise unmatched accuracy, in centimetres, yet to acquire the tree semantics manually from these robust files is rather time consuming. GreeHill extracts tree-semantics automatically from point clouds through their machine learning algorithm, drastically reducing the needed time and resources for tree measurements. Point cloud derived data can offer digital Voxel models of the trees, which are necessary elements of semantic 3D models of cities. These representations could allow the simulation of surface-plant-air interactions to enable to create sustainable living conditions in a constantly changing environment and curtail the dangers through data driven, pre-emptive decision making.