A Coupled Fire-Atmosphere Model Lets the Fire Rewrite Its Own Wind Field
Ordinary spread models take wind as an input from a weather forecast and never talk back. A coupled fire-atmosphere model, of the WRF-SFIRE type, runs a numerical weather model and a fire front on the same grid and passes information both ways each time step: the fire model reports its heat and moisture release into the atmospheric model's lowest layers, and the atmospheric model returns the updated near-surface wind that drives the front. That two-way link is the only way to reproduce behaviour a fire causes in its own environment — indraft toward the fire, the wind reversal on the flanks, a plume that leans and then straightens, and the plume-dominated behaviour and pyroconvection the fire-weather area describes. It is also the only way to model a fire's response to its own downdrafts. The cost is that you are now running a weather model, so a forecast that took seconds takes hours on many processors, and errors in the weather model become errors in the fire. After this Concept you can say what feedback a coupled model captures that an uncoupled spread model structurally cannot.
This Concept is waiting for its first lesson!
Ordinary spread models take wind as an input from a weather forecast and never talk back. A coupled fire-atmosphere model, of the WRF-SFIRE type, runs a numerical weather model and a fire front on the same grid and passes information both ways each time step: the fire model reports its heat and moisture release into the atmospheric model's lowest layers, and the atmospheric model returns the updated near-surface wind that drives the front. That two-way link is the only way to reproduce behaviour a fire causes in its own environment — indraft toward the fire, the wind reversal on the flanks, a plume that leans and then straightens, and the plume-dominated behaviour and pyroconvection the fire-weather area describes. It is also the only way to model a fire's response to its own downdrafts. The cost is that you are now running a weather model, so a forecast that took seconds takes hours on many processors, and errors in the weather model become errors in the fire. After this Concept you can say what feedback a coupled model captures that an uncoupled spread model structurally cannot.
Are you a teacher? Sign in to start contributing.
Sign In