Assimilating an Observed Perimeter Restarts the Forecast Where the Fire Actually Is
A running spread simulation drifts. Wind forecasts are wrong, the fuel map is old, and after a few hours the simulated fire and the real one are in different places. Data assimilation corrects it: when a new observed perimeter arrives, from an infrared mapping flight or a satellite hot-spot product, the simulation state is nudged toward the observation instead of being either ignored or thrown away and restarted from scratch. In the ensemble form, many simulations are run with perturbed winds, fuel moistures and ignition times; the spread of that ensemble at the observation time says how uncertain the model is, the observation carries its own error, and the update weights the two accordingly, exactly as a filter weights a prediction against a measurement. What often matters more than the corrected perimeter is what the update implies about the parameters: a fire consistently ahead of the model suggests the wind or fuel input is wrong, and correcting that improves the next forecast too. After this Concept you can explain why a fire forecast improves when a new perimeter is folded in rather than replacing the run.
This Concept is waiting for its first lesson!
A running spread simulation drifts. Wind forecasts are wrong, the fuel map is old, and after a few hours the simulated fire and the real one are in different places. Data assimilation corrects it: when a new observed perimeter arrives, from an infrared mapping flight or a satellite hot-spot product, the simulation state is nudged toward the observation instead of being either ignored or thrown away and restarted from scratch. In the ensemble form, many simulations are run with perturbed winds, fuel moistures and ignition times; the spread of that ensemble at the observation time says how uncertain the model is, the observation carries its own error, and the update weights the two accordingly, exactly as a filter weights a prediction against a measurement. What often matters more than the corrected perimeter is what the update implies about the parameters: a fire consistently ahead of the model suggests the wind or fuel input is wrong, and correcting that improves the next forecast too. After this Concept you can explain why a fire forecast improves when a new perimeter is folded in rather than replacing the run.
Are you a teacher? Sign in to start contributing.
Sign In