Context-aware UAV LiDAR reveals forest structure and improves tree diameter estimates in subalpine forest

Jaime C Revenga , Stefan Oehmcke, Mana Gharun, Flurin Sutter, Katerina Trepekli, Fabian Gieseke, Nina Buchmann, Alexander Damm

Forest structure, tree diameter, and aboveground biomass (AGB) are central variables in trait‑based ecology and forest management, and recent advances in Unmanned Aerial Vehicle (UAV) and LiDAR surveys have substantially improved tree‑level phenotyping of these structural attributes. Building on these developments, machine‑learning (ML) applications are increasingly used to refine tree‑diameter estimates and, by extension, improve AGB predictions derived from allometric relationships. Here, we evaluated the capacity of shallow learning methods to leverage local information from the surrounding context of the tree of interest to improve predictions of stem diameter and tree-level biomass, over 33 ha of a Norway spruce forest (Davos, CH). Our objectives were to (i) characterise gradients in tree height, (ii) examine group-level morphology of tree assemblages as an indicator of forest structural organisation, and (iii) assess whether these patterns can be leveraged to improve tree diameter and AGB predictions. We segmented the point cloud data scene into individual canopies and focused on LiDAR-derived tree canopy features. We then used local indicators of spatial association of tree heights to characterize local context and identified tree assemblages within the forest. Assemblage-level metrics were first analysed to characterise forest spatial structure and ecological similarity, and subsequently evaluated as additional predictors in ML regression experiments for tree diameter. The focus was on comparing performance of tree diameter predictions between twin regression methods that either consider assemblage metrics (i.e. context-aware), or not. Then, the improvements provided by context awareness were assessed in terms of accuracy gained in estimating tree diameter and AGB. We obtained results of three different shallow learning methods and evaluated these based on nested cross-validation. We considered two datasets within the same site: one being scattered in sparse measurement plots, the other spatially continuous. In both sparse and continuous datasets, we found enhanced prediction performance in context-aware regressions, where RMSE on tree diameter estimation was reduced by 4.1% and by 0.8%, respectively, suggesting that an heterogeneous context supports enhanced estimates. These findings indicate that gradients in tree height can reflect underlying ecological drivers of forest structure, and that this structural information may be leveraged to enhance predictions of tree diameter and AGB. The method proposed is fully native to UAV LiDAR data.