Harvester-mounted mobile laser scanning (HarvMLS) is a novel approach to quantify forest structure in situ during harvesting operations. However, its reliability relative to established aerial and ground-based laser scanning systems remains unknown. In this study, we benchmarked HarvMLS against drone-based laser scanning (DLS) and backpack mobile laser scanning (BpMLS) for forest structure assessment. Point cloud data were acquired using these laser scanning systems in a boreal forest in southern Finland during thinning operations in 2024. HarvMLS collected data during thinning, while BpMLS and DLS collected data immediately before and after thinning. A grid-level approach was used for point cloud processing. Sixty-two 10 m × 10 m plots were analyzed, each with pre- and post-thinning data. Forest structure was characterised using estimates of structural attributes and complexity. The structural attributes included canopy height, canopy cover (CC), and filled voxel proportion (FVP). Structural complexity was quantified using the coefficient of variation (CV) of the structural attributes. Furthermore, FVP was analyzed across various canopy height classes to evaluate the scanning system’s capacity to capture vegetation density at different canopy layers. The scanning systems consistently estimated canopy height, CC, and the CV of these attributes, including the CV of FVP. Significant differences were detected for FVP and FVP at lower canopy layers (<15 m). However, post hoc tests showed no significant difference between HarvMLS and BpMLS. These findings indicate that HarvMLS provides reliable estimates of forest structural attributes and their variability during active forest operations.
Fires are major disturbances in boreal forests, and forest fire risk is projected to increase in Finland. Despite this, comprehensive large-scale analyses evaluating multiple potential drivers of fire ignition and size remain scarce. The aim of this study was to assess how selected environmental and human-related factors influence the relative ignition probability and fire size in Finnish boreal forests. We utilized data on forest fire occurrence and size, the Finnish Forest Fire Index (FFI), and ignition cause, compiled by rescue authorities across mainland Finland during 2014–2023. We supplemented it with spatial data describing environmental factors (site trafficability, slope, canopy cover, and dominant tree species) and human-related factors (human population density and distance to nearest road). The dataset contained 10 284 forest fire observations and an equal number of pseudo-absence observations. We fitted one generalized additive model (GAM) each to explain relative ignition probability and fire size. The relative ignition probability model explained 51.6% of the deviance and achieved a classification accuracy of 83.7%, while the fire size model explained 21.5% of the deviance. Human-related variables were among the most influential in both models, with human presence increasing relative ignition probability but reducing fire size. Seasonality and weather were also significant, suggesting a major influence of weather and vegetation dynamics on fuel moisture, whereas dominant tree species affected only fire size and slope only relative ignition probability. The contrasting effects of humans on forest fires suggest potential for reducing future fire risk through targeted fire management strategies.
Some locally rare tree species – like European aspen (Populus tremula L.) – contribute disproportionately to forest biodiversity, making their accurate monitoring important for conservation. However, most remote sensing studies of rare tree species ignore their low prevalence/base rate, necessary for model transferability. We aimed to assess European aspen classification under realistic landscape prevalence and find methods to improve classification performance. Across 253 plots from field data in Våler, Norway we detected and segmented 5455 living trees, including 55 European aspens (ca. 1% prevalence), from high-density airborne laser scanning (ALS) data merged with multispectral aerial imagery. Individual segments were then classified as European aspen or non-aspen using random forest and synthetic minority oversampling (SMOTE). We evaluated classification based on tree counts and summed tree volumes. To underpin the importance of prevalence consideration and compare our results, we also re-analysed published Fennoscandian European aspen classification studies. Contrary to a perfect balance, we found an optimal SMOTE-ratio at ca. 0.07 (European aspen–non-aspen ratio). Our count-based classification achieved 32% (16–52) false discovery rate (FDR) and 73% (60–83) false negative rate (FNR), whereas volume-based classification achieved 31% (11–51) FDR and 54% (38–71) FNR. Re-analysed studies averaged 66% FDR and 33% FNR at a 1% prevalence. Our results demonstrate that rare tree species classification benefits from optimising majority–minority ratios and, given the ecological value of larger trees, volume-based approaches are warranted. Finally, we urge reporting landscape prevalence and adjust classification metrics to reflect these to enable transferability to real-world applications in rare tree species classification.
Logging trails are a critical component of modern forest harvesting operations, particularly in thinning and partial cutting, where trail layout strongly influences harvesting efficiency, soil disturbance, future stand growth, and carbon balance. Despite their importance, existing regulations and commonly used metrics – such as mean trail spacing and trail density – primarily address quantitative properties of logging trail networks and do not capture the qualitative aspects of network design. In this study, we introduce a novel concept for assessing the spatial arrangement performance of logging trail networks based on two complementary spatial configuration metrics: a Coverage Coefficient, describing the proportion of forest area reachable by harvesting machinery, and a Structural Efficiency Coefficient, comparing the actual trail length to an idealized network layout. Together, these metrics provide a geometrically meaningful and operationally relevant evaluation of trail network spatial arrangement beyond conventional density-based measures. We also introduce a method for computing a geometrically meaningful mean spacing between adjacent trails. The methodology is demonstrated using ten thinning sites from Central Finland. The results reveal clear differences in the spatial pattern of trail layouts among sites that are not detectable using traditional metrics alone. The proposed approach offers a practical tool for evaluating spatial arrangement performance of logging trail networks after harvesting, supporting operator feedback, regulatory assessment, and the development of automated trail network design and planning procedures. More broadly, the metrics provide a foundation for improving harvesting efficiency, reducing unnecessary trail density, and supporting environmentally sustainable forest management.
