FRSM uses advanced statistical techniques such as logistic regression, generalized Pareto, and multinomial logit models to simulate the likelihood of ignition, fire growth, and severity fractions (low, moderate, and high). It incorporates climate forcing from both dynamically downscaled Weather Research and Forecasting (WRF) datasets and LOCA2 hybrid statistical downscaling, capturing the influence of temperature, precipitation, humidity, and wind patterns derived from CMIP6 global climate projections. The model also accounts for human factors—such as urban expansion and infrastructure density—that affect ignition rates and exposure in the wildland–urban interface.

By running thousands of Monte Carlo simulations, FRSM produces probabilistic fire forecasts that capture the inherent variability of fire behavior across decades. Outputs include maps and data layers representing ignition probability, area burned, and burn severity distributions, which are used to assess future wildfire risk under different climate, vegetation management, and urban growth scenarios. When coupled with LUCAS, FRSM forms part of a fully integrated system that links fire dynamics with vegetation growth and carbon feedbacks, allowing researchers and planners to explore how management actions, such as forest thinning or prescribed burning, can reduce long-term wildfire impacts.