Publications
NIBIOs employees contribute to several hundred scientific articles and research reports every year. You can browse or search in our collection which contains references and links to these publications as well as other research and dissemination activities. The collection is continously updated with new and historical material.
2024
Authors
Fride Høistad ScheiAbstract
No abstract has been registered
Authors
Binbin Xiang Maciej Wielgosz Theodora Kontogianni Torben Peters Stefano Puliti Rasmus Astrup Konrad SchindlerAbstract
Detailed forest inventories are critical for sustainable and flexible management of forest resources, to conserve various ecosystem services. Modern airborne laser scanners deliver high-density point clouds with great potential for fine-scale forest inventory and analysis, but automatically partitioning those point clouds into meaningful entities like individual trees or tree components remains a challenge. The present study aims to fill this gap and introduces a deep learning framework, termed ForAINet, that is able to perform such a segmentation across diverse forest types and geographic regions. From the segmented data, we then derive relevant biophysical parameters of individual trees as well as stands. The system has been tested on FOR-Instance, a dataset of point clouds that have been acquired in five different countries using surveying drones. The segmentation back-end achieves over 85% F-score for individual trees, respectively over 73% mean IoU across five semantic categories: ground, low vegetation, stems, live branches and dead branches. Building on the segmentation results our pipeline then densely calculates biophysical features of each individual tree (height, crown diameter, crown volume, DBH, and location) and properties per stand (digital terrain model and stand density). Especially crown-related features are in most cases retrieved with high accuracy, whereas the estimates for DBH and location are less reliable, due to the airborne scanning setup.
Abstract
This study focuses on advancing individual tree crown (ITC) segmentation in lidar data, developing a sensor- and platform-agnostic deep learning model transferable across a spectrum of dense laser scanning datasets from drone (ULS), to terrestrial (TLS), and mobile (MLS) laser scanning data. In a field where transferability across different data characteristics has been a longstanding challenge, this research marks a step towards versatile, efficient, and comprehensive 3D forest scene analysis. Central to this study is model performance evaluation based on platform type (ULS vs. MLS) and data density. This involved five distinct scenarios, each integrating different combinations of input training data, including ULS, MLS, and their augmented versions through random subsampling, to assess the model's transferability to varying resolutions and efficacy across different canopy layers. The core of the model, inspired by the PointGroup architecture, is a 3D convolutional neural network (CNN) with dedicated prediction heads for semantic and instance segmentation. The model underwent comprehensive validation on publicly available, machine learning-ready point cloud datasets. Additional analyses assessed model adaptability to different resolutions and performance across canopy layers. Our results reveal that point cloud random subsampling is an effective augmentation strategy and improves model performance and transferability. The model trained using the most aggressive augmentation, including point clouds as sparse as 10 points m−2, showed best performance and was found to be transferable to sparse lidar data and boosts detection and segmentation of codominant and dominated trees. Notably, the model showed consistent performance for point clouds with densities >50 points m−2 but exhibited a drop in performance at the sparsest level (10 points m−2), mainly due to increased omission rates. Benchmarking against current state-of-the-art methods revealed boosts of up to 20% in the detection rates, indicating the model's superior performance on multiple open benchmark datasets. Further, our experiments also set new performance baselines for the other public datasets. The comparison highlights the model's superior segmentation skill, mainly due to better detection and segmentation of understory trees below the canopy, with reduced computational demands compared to other recent methods. In conclusion, the present study demonstrates that it is indeed feasible to train a sensor-agnostic model that can handle diverse laser scanning data, going beyond current sensor-specific methodologies. Further, our study sets a new baseline for tree segmentation, especially in complex forest structures. By advancing the state-of-the-art in forest lidar analysis, our work also lays the foundation for future innovations in ecological modeling and forest management.
Abstract
Mapping individual tree quality parameters from high-density LiDAR point clouds is an important step towards improved forest inventories. We present a novel machine learning-based workflow that uses individual tree point clouds from drone laser scanning to predict wood quality indicators in standing trees. Unlike object reconstruction methods, our approach is based on simple metrics computed on vertical slices that summarize information on point distances, angles, and geometric attributes of the space between and around the points. Our models use these slice metrics as predictors and achieve high accuracy for predicting the diameter of the largest branch per log (DLBs) and stem diameter at different heights (DS) from survey-grade drone laser scans. We show that our models are also robust and accurate when tested on suboptimal versions of the data generated by reductions in the number of points or emulations of suboptimal single-tree segmentation scenarios. Our approach provides a simple, clear, and scalable solution that can be adapted to different situations both for research and more operational mapping.
Abstract
Introduction: Plantations located outside the species distribution area represent natural experiments to assess tree tolerance to climate variability. Climate change amplifies warming-related drought stress but also leads to more climate extremes. Methods: We studied plantations of the European larch (Larix decidua), a conifer native to central and eastern Europe, in northern Spain. We used climate, drought and tree-ring data from four larch plantations including wet (Valgañón, site V; Santurde, site S), intermediate (Ribavellosa, site R) and dry (Santa Marina, site M) sites. We aimed to benchmark the larch tolerance to climate and drought stress by analysing the relationships between radial growth increment (hereafter growth), climate data (temperature, precipitation, radiation) and a drought index. Results: Basal area increment (BAI) was the lowest in the driest site M (5.2 cm2 yr-1; period 1988–2022), followed by site R (7.5 cm2 yr-1), with the youngest and oldest and trees being planted in M (35 years) and R (150 years) sites. BAI peaked in the wettest sites (V; 10.4 cm2 yr-1; S, 10.8 cm2 yr-1). We detected a sharp BAI reduction (30% of the regional mean) in 2001 when springto-summer conditions were very dry. In the wettest V and S sites, larch growth positively responded to current March and June-July radiation, but negatively to March precipitation. In the R site, high April precipitation enhanced growth. In the driest M site, warm conditions in the late prior winter and current spring improved growth, but warm-sunny conditions in July and dry-sunny conditions in August reduced it. Larch growth positively responded to spring-summer wet conditions considering short (1-6 months) and long (9-24 months) time scales in dry (site M) and wet-intermediate (sites S and R) sites, respectively. Discussion: Larch growth is vulnerable to drought stress in dry slow-growing plantations, but also to extreme spring wet-cloudy events followed by dry-hot conditions in wet fast-growing plantations.
Abstract
Potential climate change impacts on water resources have been extensively assessed in Norway due to substantial changes in climate in the recent decades. However, the combined and isolated effects of forest and forest management have been rarely considered in the climate impact studies in Norway although about 38% of the land area is covered by forest. This study aims to improve hydrological impact projections in forest dominant catchments by considering the effects of forest growth and management and to attribute hydrological changes to climate and forest changes. The eco-hydrological model SWIM (Soil and Water Integrated Model) was applied to simulate hydrological processes and extremes for two micro-scale, two meso-scale and two macro-scale catchments, accounting for the effects of spatial scale. The climate projections were generated by three EURO-CORDEX (Coordinated Downscaling Experiment for the European domain) regional climate models (RCMs) for two RCPs (Representative Concentration Pathways, RCP2.6 and RCP4.5) and were bias corrected using the quantile-mapping method. Forest development over time was simulated as a function of climate determining growth and SSP-dependent harvest levels determining wood outtake. The simulations were initialized with the forest status of the year 2020 and different forest types are distinguished according to structural characteristics represented by three key parameters: leaf area index, mean tree height and surface albedo. Preliminary simulation results show that there are minor changes (within ±5%) in hydrological processes under the combinations of the climate and forest scenarios for these catchments. Climate change is the major driver of hydrological change at the catchment scale whereas forest development mainly influences the spatial distribution of the hydrological fluxes. The results further indicate that forest growth under a warming climate helps to reduce the risk of the floods and drought slightly by reducing surface runoff in wet periods and increasing base flow in dry periods, respectively.
Abstract
No abstract has been registered
Abstract
Forest restoration and improved forest management are seen as options to enhance terrestrial carbon dioxide removal in many regions, yet concerns surrounding their potentially adverse surface albedo impacts exist, particularly in high latitude and altitude regions. Such concerns are often based on generalized conclusions rooted in analyses carried out over broad spatial extents at coarse resolutions. The impacts of surface albedo change are highly sensitive to local environmental factors governing both the surface albedo and solar radiation budgets, and many previous assessments either do not sufficiently deal with such sensitivities or do not qualify the conditions under which they are relevant. Using the country of Norway with its diverse gradients in topography and climate as an ideal case study region, we seek clarity to the question of whether surface albedo is relevant to consider in forestry planning, and if so, what are the important factors determining it. We find that the adverse impact of a forest's albedo outweighs its carbon cycle benefit on only ∼4% of Norway's total forested area, reducing to <∼1% when future climate changes are considered. Our findings challenge the common perception that surface albedo concerns are highly relevant to forestry planning at high latitudes and emphasize the importance of carrying out albedo impact assessments at spatial scales aligning with those of local forestry planning.
Abstract
Forest management planning often relies on Airborne Laser Scanning (ALS)-based Forest Management Inventories (FMIs) for sustainable and efficient decision-making. Employing the area-based (ABA) approach, these inventories estimate forest characteristics for grid cell areas (pixels), which are then usually summarized at the stand level. Using the ALS-based high-resolution Norwegian Forest Resource Maps (16 m × 16 m pixel resolution) alongside with stand-level growth and yield models, this study explores the impact of three levels of pixel aggregation (stand-level, stand-level with species strata, and pixel-level) on projected stand development. The results indicate significant differences in the projected outputs based on the aggregation level. Notably, the most substantial difference in estimated volume occurred between stand-level and pixel-level aggregation, ranging from −301 to +253 m3⋅ha−1 for single stands. The differences were, on average, higher for broadleaves than for spruce and pine dominated stands, and for mixed stands and stands with higher variability than for pure and homogenous stands. In conclusion, this research underscores the critical role of input data resolution in forest planning and management, emphasizing the need for improved data collection practices to ensure sustainable forest management.
Abstract
There is currently no quality sorting of harvested hardwood timber in Norway on a national scale. Medium- and high-quality logs including those from birch (Betula pubescens Ehrh., B. pendula Roth) are thus not utilized according to their potential monetary value. Increased domestic utilization of quality birch timber requires that the quality of harvested logs be properly assessed for potential end uses. A preferred sorting procedure would use visually detectable external log defects to grade roundwood timber. Knots are an important feature of inner log quality. Thus, the aim of this study was to evaluate whether correlations between branch scar size and knot features could be found in Norwegian birch. Using 168 knots from seven unpruned birch trees, external bark attributes often showed strong correlations with internal wood quality. Both length of the mustache and length of the seal performed well as predictors of stem radius at the time of knot occlusion. The presence of a broken off branch stub as part of an occluded knot significantly increased the knot-effected stem radius, proving that the practice of removing branches and branch stubs along the lower trunk is a crucial measure if quality timber production is the primary management goal.