Weather/fire modeling for fire risk assessment

To quantify the uncertainties of weather/climate model predictions, particularly at extended ranges, numerical weather centers produce multiple model simulations, each with a different initial condition. From the ensemble of these, various statistics can be calculated that represent the forecast uncertainties. Within the fire management community, a similar practice is emerging in fire behavior prediction, which is also beset by modeling uncertainties, including those due to errors in the weather inputs that drive simulated fire behavior. This paper describes how multiple simulations of fires in a Southern California national forest are being used to generate probability maps of fire spread in the area. When combined with co-registered maps of resources at risk, a powerful tool for fire risk assessment is available for various fire planning applications, including incident and vegetation management. This study considered fire risk to structures, water and recreational resources.