The wood properties of standing trees are usually measured through destructive sampling, which is laborious and limited in terms of the number of observations that can be collected across a range of forest structures. In this thesis, the potential of bi-temporal laser scanning (LS) was explored to address these limitations by establishing a link between wood properties and the development of external tree characteristics. This thesis is an amalgamation of Studies I–III, in which all the experiments were conducted at the Evo study site in Southern Finland, encompassing diverse boreal forests.
Study I assessed the feasibility of detecting increments in crown metrics using bi-temporal airborne LS (ALS) acquired over a 5-year time interval. Significant increments were obtained across different crown metrics, the most prominent being recorded for crown volume and crown surface area. Differences were also noted between tree species in relative increments of the crown metrics, with Scots pine (Pinus sylvestris L.) differing significantly from Norway spruce (Picea abies [L.] H. Karst.) and birch (Betula spp.), though species still accounted for a small portion of variability.
The increments in tree height and crown metrics observed over a 7-year monitoring period, in addition to their initial state, were then used to explain stem volume growth (ΔV) in Study II. To avoid point cloud occlusion, which typically occurs when the data is acquired using either aerial or terrestrial platforms, a combination of helicopter-borne ALS and terrestrial LS point clouds was used for the tree observations. Scots pine showed the highest associations between ΔV and tree height, crown projection area, and crown perimeter. By contrast, increments in crown volume and crown surface area emerged as highly important metrics for predicting the ΔV of Norway spruce and birch using random forest regression.
Building on these findings, Study III addressed the use of bi-temporal ALS for assessing wood properties and their variations between the trees and stands represented by the sample plots. Wood properties were measured using X-ray microdensitometry over 15 growing seasons corresponding with ALS acquisitions. It was demonstrated that the mean annual increment in tree height was moderately associated with mean ring width across all species at both levels of the tree (RWmean-tree) and sample plot (RWmean-plot). In turn, basal area weighted mean wood density showed limited associations with the growth metrics, with only Scots pine yielding significant models at both levels of the tree (WDmean-tree) and sample plot (WDmean-plot). However, accounting for plot-level variability in linear mixed-effect regression improved the explanatory power of both the WDmean-tree and RWmean-tree models at the tree level.
Overall, this thesis contributes to the current knowledge by demonstrating the feasibility of utilizing bi-temporal point clouds to characterize increments in tree and crown metrics. It provides insights into methodologies for assessing growth allocation and highlights the potential of tree and crown metrics to explain wood properties and their variations non-destructively and repeatedly.