Fire has played an important role in dictating vegetation patterns in the western U.S. since the start of the Holocene. Across the U.S. total burned area has increased in the last 30 years and fire severity, fire occurrence, and area burned are predicted to drastically increase with climate change. Dynamic Vegetation Models (DVMs) are process-based models that are used to predict climate change impacts on terrestrial vegetation. Realistic representation of fire dynamics within DVMs is an area of current development. Here we present results from integrating a mechanistic fire module, LMfire, into the DVM LPJ-GUESS. LMfire simulates fire occurrence, behavior, and impact from dynamic processes. Many landscape forest models prescribe fire size based on an empirical relationship with fire season length and historical fire size. Instead, LMfire represents fire extent by calculating a rate of spread based on weather and topography and allows multi-day burning and coalescence of fires to more realistically represent fire behavior. Another level of mechanization is that fire mortality is a function of crown scorch and cambial damage based on the current crown height and bark thickness for each plant type, instead of imposing mortality based on a generic probability rate.
Results/Conclusions
Our development of LPJ-GUESS-LMfire resulted in improved representation of area burned, fire size, and fire severity compared to observed fire activity from 1984-2015 for the Greater Yellowstone Ecosystem. Further model improvements included coding the fire routines to operate on age-based plant cohorts (as opposed to uniformly aged populations), enabling representation of mixed-aged stand dynamics. Other landscape forest models either do not represent these fine scale processes or are computationally limited, only simulating areas of 100s to 1,000s km2. Our model advancement with LPJ-GUESS-LMFire enables the simulation of future forest demographics under a changing climate at the ecosystem and regional scale (10,000s to 100,000 km2). We also improved representation of crown fires and cell-to-cell fire spread. These new approaches to mechanistic fire modeling significantly improved representation of stand-replacing fire regimes in the western U.S.