The structure and function of watersheds throughout the Western US are impacted by the interactions of fire and other disturbances with vegetation dynamics and hydrology. The watershed is the intermediate scale at which these dynamic interactions occur directly, and thus it is important to understand and predict the impacts of a changing climate at that scale. This requires the synthesis of knowledge across multiple disciplines (including hydrology, ecology, plant physiology, fire science, and biogeochemical cycling), which is challenging because these disciplines have different metrics for the state variables that must interact in an integrated model. In this study, we have integrated WMFire, a stochastic raster-based model of fire spread, with the Regional Hydro-Ecological Simulation System (RHESSys), a spatially distributed model of climate-water-carbon interactions, in order to simulate expected fire regimes for watersheds throughout the Western US. In the bidirectional coupling of the two models, RHESSys passes to WMFire values for litter load, relative moisture deficit, wind direction, and topographic slope; WMFire combines these to calculate the probability that a fire spreads from one pixel to a neighboring pixel. The pixels affected by fire are returned to RHESSys, which will then calculate relevant fire effects. In a preliminary analysis that compared simulated to expected regimes we were able to match expected patterns in the spread of individual fires, but in some locations fire occurrence was over-predicted. The integrated model was improved by separating the process of fire occurrence from that of fire spread, thereby reducing prediction uncertainty with a modest increase in model complexity. The rate of possible fire starts is scaled to the size of the watershed under study, and for each possible fire start a successful ignition is related to the conditions in the understory rather than across the entire canopy. In addition to providing substantial progress to solve the challenge of integrating a fire-spread model with an eco-hydrological model, this work demonstrates a path forward for understanding uncertainty in the integration of models from different disciplines in the study of the Earth system.
Kennedy, M.C.; Mckenzie, D.; Tague, C.; Bart, R.R. (2016): Ecohydrological Projections of Fire Regimes: Balancing Uncertainty and Complexity to Integrate Cross-disciplinary Simulation Systems. Fall Meeting, American Geophysical Union, December 2016. Abstract GC51E-1236..