Swedish Study Unveils Advanced National Models Predicting Forest Vulnerability to Wind and Snow Damage
A team of researchers has published detailed national-scale vulnerability models for wind and snow damage in Swedish forests. The work draws on airborne laser scanning data combined with extensive National Forest Inventory records. Lead author Inka Bohlin of the Swedish University of Agricultural Sciences collaborated with Emanuele Papucci, Victor Manabe, Sven Adler, Olivia Fors, Bertil Westerlund and Susanne Suvanto. Their paper appears in the 2026 volume of Forest Ecology and Management.
The models enable assessment of damage risk from individual forest stands up to the entire country. They support forest owners and managers adapting practices to rising climate pressures.
Context of Increasing Forest Damage Risks in Sweden
Sweden's forests face growing threats from extreme weather. Windstorms and heavy snow loads have caused significant economic losses and ecological disruption in recent decades. Climate projections indicate more frequent intense events. Accurate vulnerability mapping helps prioritize preventive measures such as adjusted thinning regimes or species selection.
Traditional damage assessments relied heavily on field observations alone. The new approach integrates high-resolution remote sensing to scale predictions efficiently across large areas.
Core Data Sources Powering the Models
Researchers trained the models on approximately 42,000 field plots from the Swedish National Forest Inventory. These plots recorded damage occurrences between 2010 and 2022. Airborne laser scanning datasets spanning 2009 to 2023 supplied detailed three-dimensional forest structure information.
Additional layers included mapped forest attributes, soil characteristics, terrain features, neighborhood spatial variables and weather records. Separate models addressed southern and northern Sweden to account for differences in forest composition and climate.
Methodology: Logistic Regression with Direct ALS Metrics
The team applied logistic regression to predict damage probability. Unlike earlier large-scale efforts, the models incorporated direct airborne laser scanning metrics rather than derived summaries. This preserved fine-scale structural details such as canopy height variation and density profiles.
Performance evaluation used the area under the receiver operating characteristic curve metric. Field-data models achieved AUC values of 0.80 in the north and 0.73 in the south. Remote-sensing versions reached 0.77 and 0.69 respectively. The strongest predictors combined structural variables, dominant tree species, terrain attributes and local weather conditions.
Photo by Oleh Holodyshyn on Unsplash
Key Findings on Model Performance and Geographic Variation
Both model types demonstrated solid predictive power. Field-based versions edged out remote-sensing counterparts slightly. The gap narrows when ALS data quality is high and coverage complete. Northern models generally outperformed southern ones, reflecting clearer structural signals in boreal forests.
Maps generated from the models highlight hotspots of elevated vulnerability. These outputs operate effectively at multiple scales, from single properties to national planning.
Practical Applications for Forest Management and Policy
Forest owners can now identify stands requiring targeted interventions before damage occurs. Examples include promoting wind-firm species mixes or modifying harvest timing. National agencies gain tools for risk zoning and resource allocation during extreme weather seasons.
The approach aligns with broader European efforts to build climate-resilient forestry. It demonstrates how remote sensing can complement traditional inventories without replacing them.
Comparison with Prior Damage Modeling Efforts
Earlier studies often depended on coarser satellite imagery or limited field samples. This work advances the field by fusing national ALS coverage with the dense, repeated measurements of the National Forest Inventory. The result is higher spatial resolution and improved transferability across regions.
Integration of direct structural metrics from laser scanning marks a methodological step forward over index-based proxies used previously.
Implications for Climate Adaptation and Research Directions
As climate change intensifies weather extremes, proactive vulnerability assessment becomes essential. The models provide a replicable framework other countries with ALS and inventory programs can adapt. Future refinements may incorporate dynamic weather forecasts or emerging sensor technologies.
Continued collaboration between remote-sensing specialists and forest ecologists will strengthen these tools. The publication underscores the value of open data sharing from national monitoring programs.
Photo by Oleh Holodyshyn on Unsplash
Access the Original Research Publication
The full study is available at the ScienceDirect link: https://www.sciencedirect.com/science/article/pii/S0378112726005384. It carries the DOI 10.1016/j.foreco.2026.124040 and appears in Forest Ecology and Management Volume 618.
Researchers and practitioners interested in the underlying datasets or code can explore related resources through the Swedish University of Agricultural Sciences and collaborating institutions.
