2. Frequency Probability

Frequency Probability #

FLAM › Frequency Probability calculates, for each pixel and day, the probability that a fire starts and isn’t put out right away (P). It combines population, lightning, fuel load and fuel moisture.

Steps #

  1. Parameter file. Click New to create an .ini file with default values, or Load to open an existing one. Keep the file: you need the same parameters for scenario runs.
  2. FLAM Parameters. Adjust the values, then click Save. This writes the .ini file and opens the next window.
  3. FLAM Preprocess Probability. Set the inputs and click Run.

Parameter file dialog

FLAM Parameters window (frequency)

Parameters #

P_h    = min(1, (p / p_up)^p_e)                         ignition by people
F_supp = clip(Supp_max − exp(−C_supp · p), 0, 1)        instant suppression by people
P_l    = B_L / (B_L + exp(1.5 − 6·B_L))                 ignition by lightning
         B_L = clip((L_f − L_f,low) / (L_f,up − L_f,low), 0, 1)
P_b    = clip((B − B_l) / (B_u − B_l), 0, 1)            fuel availability
P_m    = 1 − tanh(1.75 · m / m_e)^m_e_pw                fuel moisture

p is population density (people/km²), L_f lightning density (flashes/km²/month), B fuel load (gC/m²) and m fuel moisture (fraction).

ParameterDefaultMeaning
p_up300Population density at which P_h reaches 1
p_e0.43Shape of the P_h curve; smaller values make P_h rise faster at low density
Supp_max0.9Maximum probability of instant suppression
C_supp0.025How quickly suppression increases with population density
L_f,low0.02Lightning density below which P_l = 0
L_f,up0.85Lightning density above which P_l is at its maximum
B_l200Fuel load below which fire can’t ignite (P_b = 0)
B_u1000Fuel load above which fuel doesn’t limit ignition (P_b = 1)
m_e0.35Moisture of extinction
m_e_pw2.0Shape of the P_m response to moisture

The defaults are a generic starting point. Adapt them to your region.

The formulas shown in the GUI window contain typos for B_L. The equations on this page match the code.

Inputs #

FLAM Preprocess Probability window

FieldTime stepRequired fills from
Start Date, End DateyesConfig dates
PopulationConstant / YearlyyesPopulation
Lightning (Opt.)Constant / Monthly / YearlynoLightning
FuelConstant / YearlyyesFuel
MoistureDaily (folder)yesDaily Mois
Output FolderyesResults

How the Population, Lightning and Fuel fields work:

  • Each field takes a file.
  • For Constant, the tool uses that file.
  • For Yearly or Monthly, select any one file of the series. The tool uses all rasters in that file’s folder.
  • guesses the time step from the number of files in each folder: one file means Constant. Check the guess before you click Run.

The configuration stores Lightning as a folder. After clicking , use Open to select a file inside that folder. Otherwise, Monthly and Yearly runs read the wrong folder.

Output #

The tool creates these subfolders in the Output Folder:

Output subfolders

FolderFilesContent
P_h, F_suppP_h.tif or P_h_YYYY.tifHuman ignition and suppression
P_lP_l.tif, P_l_YYYYMM.tif or P_l_YYYY.tifLightning ignition (empty without lightning data)
P_bP_b.tif or P_b_YYYY.tifFuel availability
P_mP_m_YYYYMMDD.tifFuel moisture effect
PP_YYYYMMDD.tifDaily fire probability per pixel. NoData −999.

The equations give the probability for a reference area of 1000 km². Before saving, AFEIM converts it to each pixel’s area A (km²) with P_pixel = 1 − (1 − P)^(A/1000).

The subfolder names are fixed. Use a separate Output Folder for each scenario so that runs don’t overwrite each other.