Different methodologies for calculating crown volume of Platanus hispanica trees by terrestial laser scanner and comparison with classical dendrometric measurements

Abstract

Terrestrial laser scanners (TLSs) are used in forestry and fruit culture applications to perform a threedimensional geometrical characterization of trees and so make it easier to develop management systems based on that information. In addition, this data can improve the accuracy of dendrometric variable estimations, such as crown volume, obtained by standard methods. The main objective of this paper is to compare classical methods for crown volume estimation with the volumes obtained from the processing of point clouds obtained using a terrestrial laser scanner (TLS) on urban Platanus hispanica trees. This will allow faster quantification of residual biomass from pruning and therefore an improved management in future. The methods applied using TLS data were also evaluated in terms of processing speed. A set of 30 specimens were selected and their main dendrometric parameters (such as diameter breast height, crown diameter, total height, and distance from the crown base to the soil) were manually measured using classical methods. From these dendrometric parameters, the apparent crown volumes were calculated using three geometric models: cone, hemisphere, and paraboloid. Simultaneously, these trees were scanned with a Leica ScanStation2. A laser point cloud was registered for each tree and processed to obtain the crown volumes. Four processing methods were analyzed: (a) convex hull (an irregular polyhedral surface formed by triangles that surround the crown) applied to the whole point cloud that forms the crown; (b) convex hull using slices of 10 cm in height from the top to the base of the crown; (c) XY triangulation in horizontal sections; and (d) voxel discretization. All the obtained volumes (derived from classical methods and TLS) were assessed and compared. The regression equations that compare the volumes obtained by dendrometry and those derived from TLS data showed coefficients of determination (R2) greater than 0.78. The highest R2 (0.89) was obtained in the comparison between the volume calculated using a paraboloid and flat sections, which was also the fastest method. These results show the potential of TLS for predicting the crown volumes of urban trees, such as P. hispanica, to help improve their management, especially the quantification of residual biomass.