Predicting Wildfire Spread
Predicting wildfire spread is a difficult task, and current approaches often fall short in reliably describing fire behavior during severe conditions.
We covered our experimental use of the Pyrecast fire spread forecasting models in depth with this article on 8/27.
Since we ran the model 2 days ago, the large-scale weather forecast models have changed their predictions for incoming winds, and accordingly, the fire spread models for the next two days have changed markedly – specifically, they are now predicting more southerly spread on active fires than the previous forecasts which pushed fires to the east.
Current Model Predictions
Here are 2-day runs starting today, 8/29, and running thru midnight on the 31st. These are fairly crude visualizations based upon large-scale weather models and much more detailed fuel and terrain models, and we are sharing them mainly to keep readers abreast of big-picture potential with forecast north winds. We’ll revisit these maps after the forecast period ends, and use the results to talk about how well the models did or didn’t perform, and what their performance portends for larger discussions around technology’s role in dealing with megafires.
Significantly, these spread models assume no fire suppression will take place on active areas of the current fires.
Shades of gray show predicted fire arrival time.
Smith River Complex – 48 Hour Fire Spread Potential – 70th percentile weather 
Discussion: Models are running more conservative than they were 2 days ago, and limit spread on the Smith River Complex to about 1-2 miles toward the south.
Mosquito and Bluff Fires – 48 Hour Fire Spread Potential – 70th percentile weather
Discussion: Run under the same parameters as the Smith River Complex, not far to the north, these fires are more exposed to ridgetop winds, and show potential significant growth of 4 or more miles to the south. Interesting, the model doesn’t push the two fires together down Bluff Creek to the south, which is aligned perfectly with the north winds, as shown in the graphic below (IR from last night).
Pearch Fire – 48 Hour Fire Spread Potential – 70th percentile weather 
Discussion: The model assumes no suppression on the heat from the firing operations on the south end of the fire. The wide swath of active fire from the fresh firing operations acts as a linear ignition source which sweeps south in a wide front. A more likely scenario here would be scattered spot fires from the firing which grow independently. The terrain immediately south of the Pearch fire is deeply dissected east-west, and much of it is sheltered from the north winds, as shown in image below, with last night’s IR. Fires spotting into the South Fork of Pearch Creek could easily make slope/wind-aligned runs to the next ridge, near the Flat Fire, and hopscotch their way south.
Elliot, Swillup and U-Fish – 48 Hour Fire Spread Potential – 70th percentile weather 
Discussion: I don’t have much to say about this run. The SW flank of Elliot has been fairly cold for several days and the bottom of the canyon in Dillon Creek may be fairly sheltered from the larger-scale winds. We could see individual runs of this length on these fires in places with alignment, but it seems unlikely to me we’ll see spread on such a large front. The image below looks south over Elliot Fire at aligned topography on the other side of Dillon Creek. Most of the flank of Elliot is cold in this section as shown by lack of yellow/scattered heat on these IR maps.
Blue 2 Fire – 48 Hour Fire Spread Potential – 70th percentile weather 
Discussion: The Blue 2 Fire is one of the most alpine fires burning in the current complexes. This may render it more exposed to large-scale winds. The adjacent areas to the south burned in the major fires of 2008, and have heavy fuel loading., but this run seems overpredicted to me.
Looking south over Blue 2 Fire, lots of active fire in the IR, but lack of aligned topography to the south.
So that’s all for now. The fires look to be waking up on the heat satellites, should be a dynamic couple of days – it’s heartening to see the winds shift away from Orleans, for now, but these models are just basically Minecraft for fire nerds, so we’ll have to wait and see how well they end up matching the real world!