Few large or many small fires: Using spatial scaling of severe fire to quantify effects of fire-size distribution shifts

As wildfire activity increases and fire-size distributions potentially shift in many forested regions worldwide, anticipating the spatial patterns of burn severity expected with future fire activity is critical for ecological understanding and informing management and policy. Because spatial patterns of burn severity are influenced by a complex mixture of drivers, they remain difficult to predict for any given burned landscape. At broader extents, however, spatial scaling relationships relating high-severity patch size and shape to overall fire size, when combined with scenarios regarding regional area burned and fire-size distributions, offer a means to anticipate the spatial configuration of burn severity in future fires. Here, leveraging a satellite burn-severity dataset for 1615 fire events occurring across the northwest United States between 1985 and 2020, we present an approach for simulating expected patch-level burn-severity patterns at the scale of a region or fire regime of interest. We demonstrate this approach in a historically climate-limited fire regime within the Pacific Northwest, USA, where relatively infrequent but large and severe fires shape biomass-rich forests, and where fire potential is projected to increase as summer fire seasons become warmer and drier. We quantify how, for a given total burned area, the range of cumulative burn-severity patterns is expected to vary with the size distributions of fire events. Our results illustrate how shifts in fire-size distributions toward larger fire events will lead to increasingly large high-severity burn patches with interior areas that are increasingly far from unburned seed sources following fire. In contrast, the same total area burned in more numerous but smaller fire events will result in qualitatively different cumulative patterns of burn severity, characterized by smaller high-severity patches and closer proximity to postfire seed sources across burned landscapes. These results have important implications in forested regions, informing management actions ranging from prefire planning (e.g., fire response preparedness) to real-time decision-making (e.g., fire suppression vs. managed wildfire use) and postfire responses (e.g., replanting to restore tree cover and/or promoting early-seral habitat). The approach we present is generalizable and can be applied across regions and fire regimes to anticipate potential future fire effects.