FLAM workflow #

FLAM (wildFire cLimate impacts and Adaptation Model) estimates daily burned area for every pixel. In the GUI you run it as a sequence of tools from the FLAM menu. Each tool writes rasters that the next one reads.

flowchart LR
  W[Weather] --> M[1 Daily Fuel Moisture]
  M --> P[2 Frequency Probability]
  D[Population, fuel, lightning] --> P
  S[Fire perimeters] --> L[3 Create Labeling Data]
  P --> B[4 Calculate Burned Area]
  L --> B
  B --> X[5 Spread Simulation]

Step	Tool	Main output
1	Daily Fuel Moisture	Daily fuel moisture `moist_YYYYMMDD.tif`
2	Frequency Probability	Daily ignition probability `P/P_YYYYMMDD.tif`
3	Create Labeling Data	Observed burned area rasters (needed only for calibration)
4	Calculate Burned Area	Suppression efficiency `q`, daily burned area `A_burn`
5	Spread Simulation	Burned area with spatial spread (optional)

You can skip step 1 if you already have fuel moisture data, and step 3 if you have gridded burned-area observations. To project future fire activity, see Future scenarios.

The model in brief #

P      = [P_h + (1 − P_h)·P_l] · (1 − F_supp) · P_b · P_m     daily probability of an escaped fire
a      = f(wind, fuel moisture)                               area one fire can burn in a day (km²)
A_burn = P · a · (1 − q)(2 − q) / q²                          expected daily burned area (km²)

P_h and P_l: ignition probability from people and from lightning.
F_supp: chance a fire is put out right away.
P_b and P_m: limits set by fuel amount and fuel moisture.
q: suppression efficiency (0–1). Higher values mean fires are put out sooner; a fire burns on average (1 − q)/q days. You calibrate q so that modelled burned area matches observed burned area.

The step pages give each equation in full.