Abstract Accurate biomass estimation in large, mature trees is essential for understanding their critical roles in carbon sequestration, biodiversity conservation, and ecosystem function. However, destructive sampling of large and ecologically valuable trees is rarely feasible. Here, we took advantage of an opportunity provided by scheduled forestry operations, in which 35 mature Mediterranean tree individuals (Pinus halepensis, Pinus pinaster, and Quercus pyrenaica) were already planned for removal independently of this study. Using this rare destructive reference dataset, we developed and compared detailed workflows for biomass estimation based on terrestrial laser scanning (TLS) and four quantitative structure modelling (QSM) algorithms: TreeQSM, AdQSM, aRchi, and SimpleForest. For each algorithm, we established rigorous, species-specific protocols for calibration and parameter tuning, and systematically evaluated their relative strengths through novel comparative metrics. Our results showed strong correlations between TLS-derived estimates and actual biomass measurements. They also highlight clear method-specific algorithmic advantages. TreeQSM reliably estimated stem volumes (particularly in conifers), while AdQSM and aRchi better captured complex crown structures, with aRchi being particularly effective for broadleaf species. In contrast, SimpleForest consistently produced the poorest predictions across species and variables, underscoring its limited suitability under the conditions evaluated. Beyond validating algorithm performance, our methodological approach provides robust guidelines and objective metrics for future comparative assessments of QSM-based biomass estimation workflows. Ultimately, this study demonstrates that precise, reproducible, and non-destructive biomass quantification of mature trees is achievable, offering new pathways for accurate forest carbon accounting, and ecosystem monitoring without compromising irreplaceable large trees.
Abstract Improving forest biomass and carbon estimates is essential for sustaining the mitigation of climate change efforts in the forestry sector. Allometric biomass models, which relate biomass to easily measured tree attributes such as diameter at breast height (D) and height, are the primary tool used to estimate forest carbon stocks. When developing allometric biomass models, the process of selecting trees to calibrate the model is an important step that affects the model’s predictive performance. Typically, the D-range of the species is divided into D-classes, followed by random selection of sample trees from within the D-classes; the number of trees in each D-class defines the sample tree D-distribution. Here, using a simulation study, we compared eight types of sample tree D-distributions with respect to the precision of estimates of mean population biomass per hectare that the models produced. The results showed that, for a given sample size, when the proportion of the sample to be randomly selected from each D-class matches the proportion of the basal area in each D-class in the population, the D-distribution was optimal with respect to minimizing the standard error of the estimate of the population mean per hectare. The Neyman allocation approach and the D-distribution based on biomass instead of basal area produced very similar results to those of the optimal strategy. When a-priori information about the distribution of tree attributes in the population is unknown, a uniform D-distribution represents a good alternative to the optimal sample D-distribution. Although producing the greatest precision of the estimate, the optimal D-distribution of sample trees did not produce models with the most accurate model fit (greatest coefficient of determination), suggesting that quality of fit metrics alone are not sufficient indicators to evaluate the effects of model prediction uncertainty on the precision of a subsequent population parameter estimate. The sample tree D-distribution had considerable effects on the precision of estimates of mean population biomass that the models produced. To optimize the tree selection, and, therefore, to develop the models that for a given sample size produce precise estimators of the population mean, we recommend using an optimal D-distribution of the sample trees.
Abstract Wood products store substantial amounts of carbon transferred from forests and are an important component of the global carbon cycle. Reliable and accessible tools for tracking wood product carbon flow are therefore needed for forest management, national greenhouse gas inventories, carbon budget assessment, and climate policy development. In this study, we developed the Wood Products Carbon (WPsC) Tracker, an interactive web-based application for quantifying carbon fluxes and stocks across the full service life of different wood products. Built with Pyodide and JavaScript, the tool runs entirely in a standard web browser and requires no software installation, thereby lowering the barrier to use for forest managers, researchers, and policy analysts. WPsC Tracker contains five modules: (i) Industrial Logs, which tracks carbon flows from individual logs; (ii) Primary Products, which tracks carbon flow from primary products such as lumber and panels; (iii) End-use Products, which tracks carbon flow across eight end-use wood product categories; (iv) Global WPs Carbon, which provides wood product carbon fluxes and storage at country, continental, and global scales using FAO wood production statistics under different boundary approaches; and (v) Sensitivity Analysis, which provides a parameter generator and three parameter sensitivity analysis methods.
Abstract Remote sensing methodologies are widely used for detecting and mapping tree mortality across large spatial extents. High-resolution imagery and lidar data can generate tree mortality and damage maps at the spatial resolution of tree crowns; however, these high-resolution products are usually of limited spatial and temporal extent. Hence, methodologies based on time series analysis of moderate-resolution satellite imagery, which provides global coverage at sub-monthly or finer time scales, are commonly used for detecting and mapping forest damage. The downside of the latter approach is the reduced ability to detect low severity tree mortality, i.e. tree mortality or damage that covers a small percentage of a 30-m pixel, as is common for biotic forest disturbances. Here we evaluate how well four Landsat time series algorithms— the Landsat-based detection of trends in disturbance and recovery (LandTrendr), continuous change detection and classification (CCDC), Meddens, and Landscape Change Monitoring System (LCMS) algorithms —detect various levels of tree damage. Tree-level damage maps, derived from field observations and high-resolution remote sensing imagery, were created across six study extents where biotic disturbance agents (mountain pine beetles, spruce beetles, Western spruce budworms, Armillaria root disease) had caused prevalent tree mortality and damage of varying severity (0%–100% dead or damaged tree cover within a 30 m pixel). We evaluated the four Landsat time series algorithms across these six study areas and found that the accuracy of Landsat time series algorithms decreased as the percentage of tree damage cover decreased in 30-m pixels, with accuracies of <50% when tree damage cover in a 30-m pixel was <25%. In general, we found that the LCMS and Meddens algorithms performed best, with the LCMS algorithm having the advantage of being nationally available and not needing to be trained locally. Our results indicate that Landsat time series analysis at broader scales can be used to capture higher-severity damage as well as identify areas in need of additional investigation with higher-resolution remotely sensed data or field observations.
Abstract Close-range remote sensing systems are driving a transition towards precision forestry, increasing the demand for accurate in situ tree-level measurements. The presented study investigates the potential of directly measuring tree stems from raw Mobile Laser Scanning (MLS) data using 2D projections and 2D Convolutional Neural Networks to improve both processing speed and accuracy. We present a comprehensive workflow, including instance segmentation on projected MLS data, which is evaluated using 48 manually labelled raw scans. The proposed method achieved a diameter at breast height estimation root mean square error of 2.0 cm, comparable with approaches based on simultaneous localization and mapping (SLAM). Diameter measurements at different stem heights were successfully obtained for 70 out of 84 explored trees, with a mean absolute error of 1.8 cm. These findings demonstrate that leveraging 2D projections of raw MLS data enables efficient, SLAM-free, and truly real-time measurement of tree stems, advancing decision-making in managed forests.
