Operational Spread Models Are Tuned With Adjustment Factors, Not Trusted Raw
Operational fire behaviour systems let an analyst multiply the predicted rate of spread by an adjustment factor, set separately for each fuel type. This is not a hack bolted onto the science; it is how the science is used. Surface spread equations were fitted to uniform laboratory fuel beds, while real fuel is patchy, partly cured and sitting under a canopy that shelters it from the wind the forecast reports. The result is a systematic bias, and the fire itself measures it for you: take the observed spread between two mapped perimeters, divide the observed rate by the predicted rate, and use that ratio as the factor for the rest of the run. The discipline is to record the factor, the observation it came from and the conditions under which it was measured, because a factor fitted in the morning's light wind may be badly wrong in the afternoon's, and a factor is not evidence that the model is right — it is a measured correction to a model you already know is biased. The confusion resolved is that tuning looks like cheating; an untuned run is a laboratory answer to a field question. After this Concept you can calibrate a spread run against an observation and state what that factor is and is not evidence of.
This Concept is waiting for its first lesson!
Operational fire behaviour systems let an analyst multiply the predicted rate of spread by an adjustment factor, set separately for each fuel type. This is not a hack bolted onto the science; it is how the science is used. Surface spread equations were fitted to uniform laboratory fuel beds, while real fuel is patchy, partly cured and sitting under a canopy that shelters it from the wind the forecast reports. The result is a systematic bias, and the fire itself measures it for you: take the observed spread between two mapped perimeters, divide the observed rate by the predicted rate, and use that ratio as the factor for the rest of the run. The discipline is to record the factor, the observation it came from and the conditions under which it was measured, because a factor fitted in the morning's light wind may be badly wrong in the afternoon's, and a factor is not evidence that the model is right — it is a measured correction to a model you already know is biased. The confusion resolved is that tuning looks like cheating; an untuned run is a laboratory answer to a field question. After this Concept you can calibrate a spread run against an observation and state what that factor is and is not evidence of.
Are you a teacher? Sign in to start contributing.
Sign In