1 / The question
Most fires get caught. That tells you less than you think.
Crews contain the great majority of new wildfires within hours. The success rate looks like a measure of what suppression achieves. It is not. To see what initial attack does, we have to pay attention to what would happen without it.
of new fires are contained by initial attack, counting a fire as contained once it is held under 5 hectares. Agencies report between 83% and 99%, largely due to different definitions.
100 fires, standing for the 111,087 in the study.
Three kinds of fire
Every new fire falls into one of three groups. A success rate adds the first two together. Only the second is the work of initial attack. Almost all the area burned comes from the third.
Stays small on its own
Damp fuels, a rain shower, a rocky ridge. These fires would have gone out or stayed small with no crew at all. They count as successes anyway.
Stopped by the crew
These fires would have grown large. Crews arrived while they were still small and cool enough to hold. This group is the effect of initial attack.
Escapes regardless
Too big and too intense by the time anyone arrives. The fewest fires, nearly all of the area burned and nearly all of the cost.
2 / The dice
You are the duty officer
A new fire is reported. Whether the crew holds it mostly depends on two things at the moment they arrive: how big it is and how intense it is. Size is set by how fast the fire spreads and how long the crew takes to get there.
- Fire size when the crew arrives
- Intensity class
Chance the crew contains it
Containment estimates derived from an expert judgement study assuming a medium crew with helicopter support (Hirsch, Corey and Martell 1998).
Most fires are easy to hold or nearly impossible
We can sort every fire in the study into a slot by its chance of containment. You might expect a hump in the middle, but the fires pile up at the two ends. Most are relatively easy to contain; a smaller group is extremely difficult.
Fires from the study by chance of containment
- Fires contained
A few fires fall in between. These are the contested ones. For them the outcome turns on details: the experience of the crew, the mix of vegetation, the stage of the growing season, how soon the fire was detected, and chance.
A faster crew response can make difficult fires winnable, and turn contested fires into easy ones. Some new fires will challenge crews however soon they arrive.
Each ball is 15 fires, from a sample of 6,000 in the study.
In this valley, location mattered
The fires come from a simulation of the Vermilion Valley in Kootenay National Park, in the Canadian Rocky Mountains. Each cell shows the average chance of containment for fires that start there.
The pattern of containment follows patterns in vegetation, lightning and crew access. On the orange ground, a new fire is more likely than not to escape.
3 / The landscape
Measure risk reduction, not fire numbers
The study simulates the same 100,000 fire seasons twice, once with every fire left to burn and once with initial attack. The gap between the two is the risk that suppression removed.
Drag to add initial attack
Left of the line, every fire is left to burn. Right of the line, initial attack removes the fires the crew would hold. Most of the valley cools to blue.
- Chance a point burns in a year, no suppression
- 1 in 450
- With initial attack
- 1 in 2,217
- Risk removed
- 78%
Colour shows burn probability, the chance a point burns in a given year. The figures are averages across the valley. The black line is the study area and the band outside it is the 5 km buffer. Pale grey is non-fuel.
The share of risk removed varies across the valley
86%recent burn
50%mature forest
Initial attack removed the largest share of risk where fire had recently burned, on the grey-tinged hillside. It removed the smallest share in the mature, continuous forest of the green valley bottom, where a new fire is most likely to outrun the crew.
The Vermilion Valley, looking south. Figures are the share of risk removed by initial attack in each kind of ground. Label positions are approximate.
4 / Why it matters
A debate that needs data
Wildfire policy is increasingly a matter of public debate. Some call for doubling down on full suppression: put every fire out, fast. Others argue that letting some fires burn where it is safe prevents worse ones later.
Arguments often run on vibes more than evidence. There is little reliable data on what a policy does to risk on the ground. This approach can help, in our decisions and in our conversations.
Know what suppression is doing
Agencies can separate the fires that needed a crew from the fires that did not, and report the difference initial attack made in terms of risk.
Test a response policy first
Full suppression is one policy among several. Alternatives can be simulated and their consequences compared before anyone commits to one.
Put mitigation where it counts
Prescribed burns and fuel breaks have the most to offer where initial attack removes the least risk. After the 2017 Verdant Creek fire, Parks Canada cut a fuel break through the zone this analysis had flagged and planned a prescribed fire for the same ground.
This study looked at one valley in the Canadian Rocky Mountains. Your fuels and suppression response may be quite different, but this method offers a way to find out.
The method in three steps
- Simulate the fire regime with no suppression, across the full range of weather and fire sizes.
- Simulate initial attack. Give each fire a chance of containment and remove the fires the crew would hold.
- Compare the two burn probability maps.