A new dynamic fire and forest model for diagnosing the causes and consequences of increased burning in the western United States

Winslow D. Hansen, Cary Institute of Ecosystem Studies, Millbrook, NY, Meg A. Krawchuk, Forest Ecosystems and Society, Oregon State University, Corvallis, OR and A. Park Williams, Department of Geography, University of California Los Angeles, Los Angeles, CA

Background/Question/Methods
Wildfire activity has increased markedly in forests of the western United States since the 1970s; trends that are expected to continue. Fires are a powerful agent of social and ecological change, and increased burning has already altered disturbance-succession cycles, strained suppression resources and threatened life and property. Much of the growth in burned area can be attributed to human-caused warming. However, other drivers such as forest structure, past fire suppression, and changing patterns of human settlement also likely contribute to recent trends. Thus, new models are needed that can account for multiple drivers of fire and their complex interactions and feedbacks. To address this gap, we have developed a simulation model that dynamically represents fire and forests of the western US. The model pairs a state-of-the-art statistical fire module that accounts for both lightning and people as ignition sources with a new hybrid vegetation module that combines mechanistic and deterministic approaches for representing forest succession and the key processes through which fire affects forests. To benchmark model performance, we simulated five large regions of diverse forest types across the western US. We compared modeled forest and fire characteristics with ~20,000 forest-inventory plots and 850 observed fires greater than 100 ha.
Results/Conclusions
We dynamically simulated fire and forests ecosystems for 300 years in the Southwest, Sierra-Nevada mountains, Cascades, Idaho, and Greater Yellowstone forced with 1950-2019 climate (repeated sequentially). The fire-forest model reasonably recreated patterns of forest type, stand structure, and live and dead biomass when compared to forest-inventory plots. The model also generated fire frequencies, sizes, shapes, total annual area burned, and area burned at high severity that corresponded well to observed fires. Modeled tree regeneration was generally robust following fires. However, simulated forests did occasionally convert to grassland/shrubland in hot-dry areas of the Southwest and Sierra-Nevada mountains and in the middle of large burned patches. Increased fire activity in the western US is a growing crisis with profound consequences for forests and people. Scientists, managers, and decision makers need new tools to help determine whether/how we can bend trajectories toward more sustainable social-ecological outcomes. By combining mechanistic, statistical, and deterministic approaches, this fire-forest model will allow us for the first time to diagnose how changes in vegetation, climate, and human settlement have shaped recent trends in fire activity. These are critical insights for constraining future projections of burning and for informing more effective forest stewardship and community adaptation strategies.