Burn Probability Comes from Simulating Thousands of Fires Over Sampled Weather
A single spread simulation answers what happens under one weather sequence from one ignition point. Burn-probability modelling, of the FSim type, answers a different question: how often does this piece of ground burn at all. It draws thousands of synthetic fire seasons — sampling ignition locations from historical ignition density, weather sequences from a fitted weather generator, and containment from historical suppression outcomes — runs a deterministic spread model for each, and records which cells burned. The burn probability of a cell is the fraction of simulated fires that reached it, and the same run yields the distribution of fireline intensity there, which is what exposure and risk work actually needs. The output is a probability map, and reading it demands care: it is a long-run frequency under the simulated climate and fuel, not a forecast about this week, and it inherits every bias of the spread model run inside it, multiplied by thousands. After this Concept you can state what a burn-probability map is a probability of, and what it is not.
This Concept is waiting for its first lesson!
A single spread simulation answers what happens under one weather sequence from one ignition point. Burn-probability modelling, of the FSim type, answers a different question: how often does this piece of ground burn at all. It draws thousands of synthetic fire seasons — sampling ignition locations from historical ignition density, weather sequences from a fitted weather generator, and containment from historical suppression outcomes — runs a deterministic spread model for each, and records which cells burned. The burn probability of a cell is the fraction of simulated fires that reached it, and the same run yields the distribution of fireline intensity there, which is what exposure and risk work actually needs. The output is a probability map, and reading it demands care: it is a long-run frequency under the simulated climate and fuel, not a forecast about this week, and it inherits every bias of the spread model run inside it, multiplied by thousands. After this Concept you can state what a burn-probability map is a probability of, and what it is not.
Are you a teacher? Sign in to start contributing.
Sign In