ESTIMATING LEAF AREA DENSITY (LAD) ACROSS URBAN TREE GENERA USING LIDAR DATA

Abstract

Urban greenery provides natural processes that benefit human well-being, called ecosystem services, such as climate change mitigation and is an important factor for establishing sustainable cities. Some ecosystem services, such as carbon fixation, providing space for Particulate Matter (PM) deposition, and temperature cooling because of transpiration are partly dependant on the structure of the canopy, which can be investigated by the estimation of leaf area density (LAD (m2 /m3 )). To facilitate urban planning and evaluation of possible effectiveness of ecosystem services, allometric relationships are used to establish correlation between morphological variables such as crown diameter, tree height, diameter at breast height (DBH) and age. The aim of this study was to estimate LAD from LiDAR (Light Detection and Ranging) scanned urban trees situated in the City of Malmö and additionally investigate whether differences between tree genera could be found and if LAD could be correlated to allometric variables (crown diameter, tree age, height and DBH). Investigation of whether allometric relationships for the growth of urban trees could be established was also conducted. Around 2 100 urban trees, scanned by a terrestrial scanner, were analysed. The point clouds for each individual tree were filtered and divided in halves to account for trees only being scanned from one direction, hence keeping the scanned side. The side that had been scanned directly was used in the analysis. The four most prominent genera (Tilia spp., Quercus spp., Sorbus spp., and Acer spp.) were analysed separately. The estimation of LAD was done through the Voxel-based Canopy Profiling (VCP) method. Estimated LAD was correlated against allometric variables, with crown diameter, tree height, DBH, and tree age as the explanatory variables. The allometric variables were also plotted against each other, with DBH and tree age as the explanatory variables. The results indicated that LAD was strongly, negatively and non-linearly correlated to crown diameter (Radj 2 > 0.68) and moderately correlated to tree height (Radj 2 > 0.54). A multiple regression analysis between crown diameter, tree height and LAD resulted in a statistically significant model with Radj 2 = 0.70, indicating that an additional 2 percentage points could be added by tree height in addition to crown diameter to explain LAD variability. No correlation was found between LAD and tree age or DBH. Comparison of LAD between different tree genera showed a slightly higher LAD for Sorbus spp. and slightly lower for Acer spp., however, variation of crown size seemed to be much more important for LAD estimation than the genus of the tree. This might be the result of the pruning of urban trees, resulting in crown diameter not being dependant on age or DBH for its size. Correlation between tree height and crown diameter was found (Radj 2 = 0.60), but no correlation between DBH and crown diameter or tree height could be established. The results indicate that the tree height and crown diameter are more defining parameters than tree genus when estimating LAD, and for future modelling of LAD, establishing a function based on crown diameter and tree height might be useful for estimating LAD without the need for LiDAR scans of the canopy.