Does prescribed burning reduce wildfire smoke?
Yes, it almost certainly does, but benefits in terms of net smoke reductions take a few years to be realized.
By: Marshall Burke
Wildfire smoke is a rapidly growing environmental health hazard. It’s the fastest-growing source of key air pollutants in the US and is unraveling decades of previous progress in improving air quality. Increasing wildfire activity is a consequence of a century of fire suppression, which led to increasingly abundant fuels, and a warming climate that has dried these fuels and made them more flammable. Climate-driven increases in wildfire smoke are likely to be the single largest source of climate damages in the US, by far.
A central proposal for limiting future increases in wildfire risk and resulting smoke is the expanded use of prescribed fire – or the purposeful re-introduction of low-severity fire into previously fire-adapted landscapes, with the goal of limiting the likelihood of extreme wildfire. ‘Fighting fire with fire’, if you like. Decades of great work by fire ecologists and others have demonstrated the benefits of expanded use of prescribed fire for limiting subsequent wildfire risk, often using case studies from individual fires or particular management areas (also here). Based on this understanding, California has proposed dramatically expanding the use of prescribed fire in the state, with a target of 1million acres/year burned in the state in 2025. This would be about 20x the annual average of area burned to prescribed fire in the entire Western US over the last few decades, and about 10x the recent area treated in CA in the last few years. So, a really dramatic scale-up of prescribed burning in the state.
How much will this expansion of prescribed burning actually help with the State’s wildfire smoke problem? Does prescribed burning in general actually reduce total smoke? The potential tradeoffs are pretty obvious, if poorly quantified: prescribed burning (henceforth, Rx fire) generates smoke with certainty, in the hopes that a given treatment will intercept a future wildfire and reduce burn severity and smoke output from that fire. So the net benefits of using Rx fire depend on how much smoke is generated by the Rx fire, the likelihood that a location treated with Rx fire will experience a subsequent wildfire, and the reduction in burn severity and emissions you get in the subsequent wildfire.
We have two new papers that try to quantify these tradeoffs, each using an alternate approach. The first, capably led by Makoto Kelp, uses the limited data we have on completed Rx treatments (186 in our dataset) and studies what happens to wildfire emissions when these treatments intersect with subsequent wildfires. Using recently-improved wildfire emissions data, Makoto compares wildfire emissions in an area of a fire that was recently treated with Rx fire to nearby areas in the same fire that were not treated. The approach is shown for the 2020 Creek Fire below, which had 59 different Rx burns that happened inside the eventual fire boundary but prior to the wildfire.
We find that areas treated with Rx fire burned at 16% lower severity in a subsequent wildfire, and that Rx fire led to a 14% reduction in overall smoke emissions, inclusive of the smoke emitted in the Rx fire itself. Treatments occurring outside the wildland-urban interface (WUI) appeared more beneficial for reducing subsequent emissions than treatments near urban areas, and Rx burning appeared more beneficial than other fuels treatments (including mechanical thinning). We estimate a smoke emissions benefit cost ratio of about 3.2 – i.e. each ton of emissions from Rx fire reduces subsequent wildfire emissions by 3.2 tons.
The second study, capably led by Iván Higuera-Mendieta, uses observed low-severity wildfire as a proxy for Rx fire and studies the impact of low-severity fire on subsequent wildfire activity and smoke. Unlike Rx fire, where historical treatments have been limited, there have been millions of acres burned by low-severity fire in CA alone, allowing large-scale assessment of their impact on the likelihood of subsequent extreme wildfire.
Iván first uses a synthetic control method to compare 82,000 1km-square pixels that were “treated” with low-severity fire to a very large set of matched control pixels that were similar across a range of characteristics (land type, previous fire history, elevation, slope, etc) but that had not recently burned. We then track the occurrence of future fire activity in subsequent years across treated pixels and matched control, allowing us to estimate the impact of low-severity fire on the risk and severity of future fires for more than a decade following the initial treatment. Below is the schematic of what this design looks like:
A key benefit of this approach is we can also measure “spillover” benefits to nearby untreated areas – i.e. measure the possibility that treating one pixel reduces the likelihood that a neighboring pixel burns. This possibility had been proposed and measured in case study settings but not measured at a broad scale across diverse landscapes.
Consistent with earlier work, we find that low-severity fires substantially limit the likelihood of subsequent severe wildfire: the immediate effect is a reduction in risk of severe wildfire by about 90%, with effects that last at least a decade. See figure below, which shows the “relative risk” of subsequent fire of various severity in the years following treatment with low-severity fire; a value of 0.6 means that the risk of fire in the treated pixels is only 0.6 as high as the risk in control pixels, or ~ 40% lower. Critically, we also find that these benefits spill over to nearby areas, with a 43% reduction in wildfire risk in unburned pixels within 2 km of burned pixels. These results hold in conifer forests but not in other land types, such as the chaparral systems that featured heavily in the recent LA wildfires.
To measure smoke impacts, we build on earlier work led by Jeff Wen in our lab that linked specific fires to observed surface smoke concentrations. This work allows us to estimate how changes in fire severity alter surface smoke concentrations.
Using these smoke data, and not accounting for the spillover benefits of low-severity fire, we estimate a very similar benefit/cost ratio to Makoto’s paper: the cumulative reduction in smoke from treating an acre of conifer forest with low-severity fire exceeds the initial smoke emitted by 2-4x. However, when you account for the spillover benefits, we estimate this number can be as high as 20x. Notably, in either case, it can take years for these benefits to arrive: benefits only appear if the treated acre burns in a subsequent wildfire, and the probability that any particular acre burns in a subsequent wildfire in a given year is low. But, this probability creeps up over time, and benefits (in terms of future smoke reduction) clearly emerge 3-5 years after initial treatment.
Finally, we simulate the impact on wildfire burned area and total smoke concentrations if CA had embarked on its proposed ambitious Rx burning policy a decade ago. We estimate that, accounting for spillovers, treating 500,000 acres/year annually starting in 2010 would have reduced cumulative smoke exposure by 23% after a decade. Treating “only” 125k acres/year would have reduced smoke exposure by about 8%. See figure below, which plots cumulative change in smoke in the years following (simulated) policy initiation in 2010. The colored lines are when we account for treatment spillovers, the grey line is when we don’t.
Importantly, however, we find that such a policy would almost certainly increase smoke in the first few years of the policy, and increase smoke in years in which wildfire activity was otherwise very low (e.g., 2019 in CA). This is the tradeoff of Rx burning: some initial increases in smoke for much larger reductions overall reductions after a number of years. We do not directly estimate the health impacts of these changes in smoke, but most evidence suggests that key health outcomes – including mortality and asthma-related emergency department visits – are basically linear in smoke: the more smoke, the worse the outcome. So a 20% reduction in smoke will correspond to a 20% reduction in health impacts.
We hope that our findings can help the state and others successfully navigate the possibly difficult politics of the tradeoffs around the use of prescribed fire. Our results basically say: each acre you burn brings eventual benefits much larger than costs, and the more treatments you do the more benefit (in terms of reduced overall smoke) you get. But our results also indicate that we’re in a pretty deep hole: you can burn 500k acres/year in CA for a decade and “only” reduce total smoke by about a quarter. We are nowhere close to burning that amount today. This almost surely means that to protect health from increasing wildfire smoke, we will need to combine a dramatically expanded prescribed burning effort with other tools – probably most specifically, better indoor air filtration. More on that to come.