Capturing seasonal radial growth of boreal trees with terrestrial laser scanning

Abstract

Detailed observation techniques are needed to reveal the underlying eco-physiological mechanisms driving tree growth processes. Terrestrial laser scanning (TLS) has proven to be a feasible technique for characterizing trees, but it has still remained unclear whether TLS point clouds and the existing point cloud processing methods can be used for capturing even the smallest signs of the growth process of individual trees. The aim of this study was to investigate the capacity of TLS in observing seasonal radial growth of boreal trees. The experimental setup included 91 sample trees from 20 sample plots characterized with multi-scan TLS point clouds pre- and post-growing season. The sample trees were equipped with dendrometers that provided reference measurements for the increment in diameter at the breast height (Δdbh) that varied from −1.4 mm to 4.0 mm with a mean of 1.0 mm. The experiment confirmed challenges related to quantification of millimeter-level increments in dbh using TLS but cautiously highlighted its feasibility for radial tree growth monitoring when the magnitude of Δdbh exceeds several millimeters and when the aim is to characterize sample plot mean rather than individual tree growth. While the capacity of TLS to characterize Δdbh of individual trees remained rather low (r = 0.17, p = 0.07), the TLS-based estimates for sample plot mean Δdbh were slightly better in line with dendrometer measurements (r = 0.46, p = 0.04). At an individual tree level, the capacity of TLS to determine the occurrence of radial tree growth seemed to be dependent on the magnitude of observed Δdbh and benefit from the analysis of paired diameter measurements along the stem for determining individual tree growth. The results showed overall classification accuracies of a) 60.7 % and b) 70.6 % for the use of TLS in determining whether radial growth had occurred or not when the analysis was based on a) Δdbh measurements only or b) statistically significant mean increment in paired diameter measurements along the stem, respectively. Using the Δdbh-based method, the overall accuracy improved from 56.3 % to 73.0 % when the magnitude of observed Δdbh increased from ≤ 1 mm to > 1 mm, as was expected. Altogether, this study contributes by demonstrating that with TLS data acquisition and existing point cloud processing methods, it is possible to observe seasonal increments in tree structures, which emphasizes the feasibility of TLS in regular monitoring of structural changes even in boreal forest ecosystems.