A model. Run it many times over, sampling the inputs differently each run. Count how many of those runs burned each cell. Divide by the number of runs. Map it.
“Probability was estimated simply as the fraction of the total simulations where fire arrives at a particular cell.”
The right-hand column is Finney et al. (2011), the peer-reviewed paper describing the ensemble method behind FSPro — the model a Long Term Fire Analyst runs on a real fire. Same method, same arithmetic. One of them runs on a laptop in a classroom; the other helps decide whether to move a fire camp.
Wildfire ▸
A big fire runs on people with laptops. Someone maps it, someone predicts where it goes tomorrow, someone reads the infrared imagery at 3 a.m., someone orbits above it directing aircraft, someone tells the public the truth, and someone walks the burn afterwards to work out what the soil will do when it rains. Every one of those is a job, and most of them are computational.