A wind-driven or uphill fire does not grow as a circle: the front runs fastest downwind or upslope and slowest backwards, so the burned area is close to an ellipse with the ignition near its rear focus. The page computes the head spread rate from a Rothermel-form formula, the shape of the ellipse from Anderson's length-to-breadth ratio, and then advances the fire across a height map cell by cell (a continuous-time cellular automaton: every cell gets an arrival time).
R_head = R0 eta_M (1 + phi_W + phi_S) R0 = no-wind, no-slope rate of the fuel
phi_W = C (waf U)^B phi_S = 5.275 beta^-0.3 tan^2(slope) (vector sum: wind + upslope)
eta_M = 1 - 2.59 r + 5.11 r^2 - 3.52 r^3, r = M / Mx (moisture damping)
L/B = 0.936 e^(0.2566 Ue) + 0.461 e^(-0.1548 Ue) - 0.397 (Ue in mph, capped at 8)
e = sqrt(1 - 1/(L/B)^2) R(theta) = R_head (1 - e) / (1 - e cos theta)
a = R_head t / (1 + e), b = a / (L/B), area = pi a b (uniform conditions)
cell-to-cell: shortest travel time with direction-dependent speed (Dijkstra order + triangle updates)
- Wind and slope add - a steep slope acts like a strong wind (phi_S grows with tan squared); wind across the slope tilts the head direction, and wind against the slope can cancel it.
- Firebreak versus prescribed burn - a firebreak is bare soil: no fuel, so no cell in it can burn and a strip at least one cell wide stops the model fire. A prescribed-burn strip is timber litter kept at 20 % of its spread rate (burned earlier, so there is little fuel): the fire slows but crosses. The sim shows the burned area with and without the strip at the same time.
- Why the ellipse area check is not exact - the cell grid favours 48 directions, so the numerical area is a few percent below the analytic one for a stretched ellipse and converges as cells shrink (checks 2-3).
- Honest simplifications - coefficients (R0, C, B, waf, beta, Mx) are illustrative values with the structure of Rothermel's model, not a calibrated fuel-model table. No spotting or embers jumping a break (so a modelled break is more effective than a real one), no crown fire, no fire-generated wind, no diurnal change of weather or moisture, fuel moisture and wind uniform, 30 m cells. Educational use only; it is not a prediction tool and contains no suppression or incident-command content.