Accurately classifying 3-D point clouds into woody and leafy components has been an interest for applications in forestry and ecology including the better understanding of radiation transfer between canopy and atmosphere. The past decade has seen an increase in the methods attempting to classify leaves and wood in point clouds based on radiometric or geometric features. However, classification purely based on radiometric features is sensor-specific, and the method by which the local neighborhood of a point is defined affects the accuracy of classification based on geometric features. Here, we present a leaf-wood classification method combining geometrical features defined by radially bounded nearest neighbors at multiple spatial scales in a machine learning model. We compared the performance of three different machine learning models generated by the random forest (RF), XGBoost, and lightGBM algorithms. Using multiple spatial scales eliminates the need for an optimal neighborhood size selection and defining the local neighborhood by radially bounded nearest neighbors makes the method broadly applicable for point clouds of varying quality. We assessed the model performance at the individual tree- and plot-level on field data from tropical and deciduous forests, as well as on simulated point clouds. The method has an overall average accuracy of 94.2% on our data sets. For other data sets, the presented method outperformed the methods in literature in most cases without the need for additional postprocessing steps that are needed in most of the existing methods. We provide the entire framework as an open-source python package.

Terrestrial laser scanning (TLS) has been applied to estimate forest wood volume based on detailed 3D tree reconstructions from point cloud data. However, sources of uncertainties in the point cloud data (alignment and scattering errors, occlusion, foliage...) and the reconstruction algorithm type and parameterisation are known to affect the reconstruction, especially around finer branches. To better understand the impacts of these uncertainties on the accuracy of TLS-derived woody volume, high-quality TLS scans were collected in leaf-off conditions prior to destructive harvesting of two forest-grown common ash trees ( L.; diameter at breast height ~28 cm, woody volume of 732 and 868 L). We manually measured branch diameters at 265 locations in these trees. Estimates of branch diameters and tree volume from Quantitative Structure Models (QSM) were compared with these manual measurements. The accuracy of QSM branch diameter estimates decreased with smaller branch diameters. Tree woody volume was overestimated (+336 L and +392 L) in both trees. Branches measuring < 5 cm in diameter accounted for 80% and 83% of this overestimation respectively. Filtering for scattering errors or improved coregistration approximately halved the overestimation. Range filtering and modified scanning layouts had mixed effects. The small branch overestimations originated primarily in limitations in scanner characteristics and coregistration errors rather than suboptimal QSM parameterisation. For TLS-derived estimates of tree volume, a higher quality point cloud allows smaller branches to be accurately reconstructed. Additional experiments need to elucidate if these results can be generalised beyond the setup of this study.Fraxinus excelsior

Key message: Stump-to-tip trends in basic wood density complicate the conversion of tree volume into aboveground biomass. We use 3D tree models from terrestrial laser scanning to obtain tree-level volume-weighted wood density. Terrestrial laser scanning (TLS) is used to generate realistic 3D tree models that enable a non-destructive way of quantifying tree volume. An accurate value for basic wood density is required to convert tree volume into aboveground biomass (AGB) for forest carbon assessments. However, basic density is characterised by high inter-, intra-species and within-tree variability and a likely source of error in TLS-derived biomass estimates. Here, 31 adult trees of 4 important European timber species (Fagus sylvatica, Larix decidua, Pinus sylvestris, Fraxinus excelsior) were scanned using TLS and then felled for several basic wood density measurements. We derived a reference volume-weighted basic density (ρ_w) by combining volume from 3D tree models with destructively assessed vertical density profiles. We compared this to basic density retrieved from a single basal disc over bark (ρ_bd), two perpendicular pith-to-bark increment cores at breast height (ρ_ic), and sourcing the best available local basic wood density from publications. Stump-to-tip trends in basic wood density caused site-average woody AGB estimation biases ranging from −3.3 to + 7.8% when using ρ_bd and from −4.1 to + 11.8% when using ρ_ic. Basic wood density from publications was in general a bad predictor for ρ_w as the bias ranged from −3.2 to + 17.2%, with little consistency across different density repositories. Overall, our density-attributed biases were similar to several recently reported biases in TLS-derived tree volume, leading to potentially large compound errors in biomass assessments with TLS if patterns of vertical basic wood density variation are not properly accounted for.

There is mounting empirical evidence that lianas affect the carbon cycle of tropical forests. However, no single vegetation model takes into account this growth form, although such efforts could greatly improve the predictions of carbon dynamics in tropical forests. In this study, we incorporated a novel mechanistic representation of lianas in a dynamic global vegetation model (the Ecosystem Demography Model). We developed a liana‐specific plant functional type and mechanisms representing liana–tree interactions (such as light competition, liana‐specific allometries, and attachment to host trees) and parameterized them according to a comprehensive literature meta‐analysis. We tested the model for an old‐growth forest (Paracou, French Guiana) and a secondary forest (Gigante Peninsula, Panama). The resulting model simulations captured many features of the two forests characterized by different levels of liana infestation as revealed by a systematic comparison of the model outputs with empirical data, including local census data from forest inventories, eddy flux tower data, and terrestrial laser scanner‐derived forest vertical structure. The inclusion of lianas in the simulations reduced the secondary forest net productivity by up to 0.46 t_C ha⁻¹ year⁻¹, which corresponds to a limited relative reduction of 2.6% in comparison with a reference simulation without lianas. However, this resulted in significantly reduced accumulated above‐ground biomass after 70 years of regrowth by up to 20 t_C/ha (19% of the reference simulation). Ultimately, the simulated negative impact of lianas on the total biomass was almost completely cancelled out when the forest reached an old‐growth successional stage. Our findings suggest that lianas negatively influence the forest potential carbon sink strength, especially for young, disturbed, liana‐rich sites. In light of the critical role that lianas play in the profound changes currently experienced by tropical forests, this new model provides a robust numerical tool to forecast the impact of lianas on tropical forest carbon sinks.

Tree allometric models, essential for monitoring and predicting terrestrial carbon stocks, are traditionally built on global databases with forest inventory measurements of stem diameter (D) and tree height (H). However, these databases often combine H measurements obtained through various measurement methods, each with distinct error patterns, affecting the resulting H:D allometries. In recent decades, terrestrial laser scanning (TLS) has emerged as a widely accepted method for accurate, non‐destructive tree structural measurements. This study used TLS data to evaluate the prediction accuracy of forest inventory‐based H:D allometries and to develop more accurate pantropical allometries. We considered 19 tropical rainforest plots across four continents. Eleven plots had forest inventory and RIEGL VZ‐400(i) TLS‐based D and H data, allowing accuracy assessment of local forest inventory‐based H:D allometries. Additionally, TLS‐based data from 1951 trees from all 19 plots were used to create new pantropical H:D allometries for tropical rainforests. Our findings reveal that in most plots, forest inventory‐based H:D allometries underestimated H compared with TLS‐based allometries. For 30‐metre‐tall trees, these underestimations varied from −1.6 m (−5.3%) to −7.5 m (−25.4%). In the Malaysian plot with trees reaching up to 77 m in height, the underestimation was as much as −31.7 m (−41.3%). We propose a TLS‐based pantropical H:D allometry, incorporating maximum climatological water deficit for site effects, with a mean uncertainty of 19.1% and a mean bias of −4.8%. While the mean uncertainty is roughly 2.3% greater than that of the Chave2014 model, this model demonstrates more consistent uncertainties across tree size and delivers less biased estimates of H (with a reduction of 8.23%). In summary, recognizing the errors in H measurements from forest inventory methods is vital, as they can propagate into the allometries they inform. This study underscores the potential of TLS for accurate H and D measurements in tropical rainforests, essential for refining tree allometries.