Fire safety. This is an educational model, not an operational forecast. Never light fires outdoors in dry or windy weather. In a real wildfire, follow official warnings and evacuation orders straight away.
What you see
A 1.6 × 1.6 km landscape (160 × 160 cells of 10 m). Each cell holds a fuel type, height and slope. A fire front moves from cell to cell; how fast it moves depends on the fuel, how dry it is, the wind and the slope. Flame and smoke sprites are enlarged so you can see them from far away.
Rate of spread (Rothermel 1972)
The head of the fire advances when the heat it sends forward is enough to bring the next fuel to ignition:
R = I_R · ξ · (1 + φ_w + φ_s) / (ρ_b · ε · Q_ig)
I_R reaction intensity (heat released per area per minute)
ξ share of that heat that reaches the fuel ahead
ρ_b fuel bed bulk density, ε effective heating number
Q_ig = 250 + 1116·M heat to ignite fuel with moisture M (Btu/lb)
Each fuel model (grass, shrub, timber litter…) is a bed of particle sizes (1-h, 10-h, 100-h dead and live) with its own load, depth and moisture of extinction. The model combines them weighted by surface area (Albini 1976). Moisture damping: η_M = 1 − 2.59r + 5.11r² − 3.52r³ with r = M/M_x.
Wind and slope
φ_w = C · U^B · (β/β_op)^−E U = mid-flame wind = WAF × 20-ft wind
φ_s = 5.275 · β^−0.3 · tan²θ
Wind and slope act as vectors and are added (Rothermel 1983); the fire heads where the sum points. Wind is reduced to mid-flame height by a wind adjustment factor: about 0.36–0.55 in open grass and shrubs, about 0.15 under a forest canopy. Upslope, flames lean over the fuel ahead and preheat it, so fire climbs hills fast and creeps down slowly.
Fuel moisture from the weather
EMC (Simard 1968), T in °F, H = RH %:
H < 10: 0.03229 + 0.281073H − 0.000578HT
10–50: 2.22749 + 0.160107H − 0.01478T
≥ 50: 21.0606 + 0.005565H² − 0.00035HT − 0.483199H
1-h fuel ≈ 1.03 × EMC; 10-h and 100-h fuels +1 and +2 points (model value)
Fire shape
L/B = 0.936·e^(0.2566U) + 0.461·e^(−0.1548U) − 0.397 (U effective wind, mi/h)
R(θ) = R_head · (1 − e) / (1 − e·cos θ), e = √(L/B² − 1) / (L/B)
With no wind on flat ground the fire is a circle; wind and slope stretch it into an egg shape with the ignition point at the back focus.
How the grid spreads fire
Each burning cell sends the front to its 32 neighbours (8 adjacent, 8 knight moves, 16 longer moves). The time to reach a neighbour is distance ÷ rate in that direction, averaged over both cells; progress along each link is kept when the weather changes. A link may not cross a firebreak, road, stream or rock line. This is the minimum-travel-time idea used by operational tools such as FARSITE (Finney 2002).
Intensity and flame length
I_B = I_R · t_r · R (Byram fireline intensity, kW/m), t_r = 384/σ min
L_f = 0.0775 · I_B^0.46 (m)
| Flame length | What it means (general guidance) |
| < 1.2 m | Low intensity; ground crews can usually work at the edge |
| 1.2–2.4 m | Too hot for hand tools at the head; firebreaks may hold |
| 2.4–3.4 m | Hard to control; spotting and torching likely |
| > 3.4 m | Very intense; only indirect measures and evacuation planning |
Ember spotting (simplified)
Very intense cells (I_B > 2000 kW/m) occasionally loft embers downwind. Distance is drawn from a log-normal law whose median grows with flame length and wind (a model value, not the Albini lofting model). An ember starts a new fire with the probability of ignition of Schroeder (1969), which falls quickly as fine fuel gets wetter. Embers can jump firebreaks — the reason breaks alone do not stop every fire.
Model limits
- Surface fire only: no crown fire, so fires in timber litter look slow. Real timber fires can jump into the canopy.
- Wind is the same everywhere; real wind speeds up over ridges and swirls in valleys.
- Weather is constant unless you change it; no day–night moisture cycle.
- The grid under-estimates very elongated fires (L/B above about 3): about −6 % area at L/B 1.9, about −26 % at L/B 4 (measured against the exact ellipse).
- Spotting is phenomenological; no fire fighting is modelled.
Live checks (computed in your browser)
Running…
References
- Rothermel R.C. (1972) A mathematical model for predicting fire spread in wildland fuels. USDA FS INT-115.
- Albini F.A. (1976) Estimating wildfire behavior and effects. USDA FS INT-30.
- Anderson H.E. (1982) Aids to determining fuel models for estimating fire behavior. USDA FS INT-122.
- Anderson H.E. (1983) Predicting wind-driven wild land fire size and shape. USDA FS INT-305.
- Rothermel R.C. (1983) How to predict the spread and intensity of forest and range fires. USDA FS INT-143.
- Byram G.M. (1959) Combustion of forest fuels. In: Forest Fire: Control and Use.
- Finney M.A. (2002) Fire growth using minimum travel time methods. Can. J. For. Res. 32.
- Andrews P.L. (2018) The Rothermel surface fire spread model and associated developments. USDA FS RMRS-GTR-371.
Not affiliated with any fire agency. Equations are public USDA Forest Service research.