Presets
Playback
Parameters
Camera
Layers
Quality
Model check
The science
Spread of a fire. Without wind a fire grows as a circle at the no-wind spread rate b (metres per minute). With wind it grows as an ellipse with the ignition point at its rear focus (Huygens principle, Richards 1990): ROS(θ) = b / (LB (1 - e cosθ)), where θ is the angle from the wind direction, e = √(1 - 1/LB²) and the length-to-breadth ratio LB comes from Anderson (1983): LB = 0.936 e0.2566 U + 0.461 e-0.1548 U - 0.397 with U the mid-flame wind in miles per hour, capped at 8. Head, flank and back rates therefore differ by factors of about LB.
Fuel and moisture. The no-wind rate is R0 (grass 4, shrub 2.2, timber 0.9 m/min, illustrative round numbers) times Rothermel's moisture damping coefficient ηM = 1 - 2.59 r + 5.11 r² - 3.52 r³, with r = moisture / moisture of extinction. Water and bare ground cannot burn.
Solving it on a map. Each 30 m cell gets a spread rate. Arrival times are found with Dijkstra's algorithm on a 32-direction graph; the time to cross an edge is the integral of 1/ROS along it. When you drop retardant or change the wind the solver keeps everything already burning and recomputes the rest.
Retardant. A tanker load V laid at coverage C (litres per square metre) covers V / (C × 48 m) metres of a 48 m wide line. Cells whose coverage is at least the requirement for their fuel (0.5, 1.0, 1.5 L/m² for grass, shrub, timber; illustrative values after the US coverage-level idea) cannot burn; lower coverage only slows the fire by 1 - C/Creq. The ends of a line are tapered.
Checks. The solver is compared with a circle (no wind), with the ellipse length-to-breadth ratio and area, with the head-fire distance rate on the wind axis, with a closed retardant ring (nothing outside may burn), and with the exact delay a partly effective strip should cause.
Simplifications. Flat ground, uniform fuel inside a cell, one fuel layer, steady wind at mid-flame height, no slope, no ember spotting, no crown fire, no fire-induced wind, no diurnal weather, retardant that never washes out, and a spread rate that does not depend on fuel load. The aircraft and trees are drawn larger than their scale. This is an educational model, not a tool for planning real fire suppression.