Wed, Aug 17, 2022: 11:00 AM-11:15 AM
513E
Background/Question/MethodsThe annual forest area burned in the western United States (US) has increased more than 1,100% since the 1980s. Past analyses demonstrate that rapid increases in forest-fire extent are related to warming and reduced warm-season precipitation frequency. However, effects of aridity on fire activity are contingent on many other factors, including fuel availability and ignitions. Further, climate, fuels, and human drivers of fire all vary spatially. Therefore, new approaches that represent the dynamic and interacting causes of fire at fine spatial resolutions are desperately needed. We present a new, spatially explicit forest-fire model for the western US that is coupled to a new dynamic forest model. The forest-fire model has a spatial resolution of 12 km and operates on a monthly timestep. The forest model has a spatial resolution of 1 km, represents 12 western US forest types and grassland, and simulates drivers of postfire forest recovery and succession, including tree-seedling establishment and tree growth and mortality. We assessed the new forest-fire model compared with observations. We then conducted a climate-change attribution exercise where the coupled forest and fire models were run for the period 1980–2021 forced by CMIP6 climate models with and without observed human-caused climate trends.
Results/ConclusionsOur forest-fire model explained approximately 70% of interannual variability in the frequency of large ( >100 ha) forest fires and forest area burned in the western US. The model also accounted for essentially all of the observed doubling in the frequency of large forest fires since the mid-1980s, and for 70% of the increase in annual forest-fire area. Model biases included greater burned area than observed in the southern Rocky Mountains and less than observed burned area in northern California. Simulated increases in forest-fire frequency and area since the mid-1980s were largest and most strongly correlated with drying trends where live forest biomass was higher and where downed fuels were abundant but dry enough to burn. Climate-fire relationships were also pronounced in areas with a Mediterranean climate. Increases in forest-fire activity occurred despite simulated reductions in forest biomass and connectivity as a result of past fire and drying trends. However, some fuel limitation was evident; we estimate that increases in forest-fire frequency and area would have been approximately 10% larger if fire- and drought-driven reductions in fuel abundance and connectivity were not accounted for. These results demonstrate the importance of representing fire and forests as complex coupled and spatially-explicit systems.
Results/ConclusionsOur forest-fire model explained approximately 70% of interannual variability in the frequency of large ( >100 ha) forest fires and forest area burned in the western US. The model also accounted for essentially all of the observed doubling in the frequency of large forest fires since the mid-1980s, and for 70% of the increase in annual forest-fire area. Model biases included greater burned area than observed in the southern Rocky Mountains and less than observed burned area in northern California. Simulated increases in forest-fire frequency and area since the mid-1980s were largest and most strongly correlated with drying trends where live forest biomass was higher and where downed fuels were abundant but dry enough to burn. Climate-fire relationships were also pronounced in areas with a Mediterranean climate. Increases in forest-fire activity occurred despite simulated reductions in forest biomass and connectivity as a result of past fire and drying trends. However, some fuel limitation was evident; we estimate that increases in forest-fire frequency and area would have been approximately 10% larger if fire- and drought-driven reductions in fuel abundance and connectivity were not accounted for. These results demonstrate the importance of representing fire and forests as complex coupled and spatially-explicit systems.