Following Up with the Fire Modelling Discussion

I received an answer from Brandon Collins, one of the authors of the paper we discussed here. At comment #14 here, I said that I would ask the author how they selected 5,000. It was a while back so thought I’d start a new post.

Here’s the answer from Brandon Collins, one of the authors of the paper.

I looked at your blog and the discussion around the issue of modeling fuel treatment tradeoffs is good. There are several valid points brought up. With regard to our choice to use 5000 randomly placed ignitions, the point of was to saturate the landscape with fire so that you essentially remove the influence of fire origin. By doing this you can get dominant fire spread patterns across the landscape, independent of assumptions about where fires may start. In our case we focused on how planned treatments (which now are being implemented) changed the occurrence of more problematic fire (flame lengths > 2m) across the landscape. By using the fire spread algorithm within RANDIG you can get at both on-site (within treated areas) and off-site (outside treatment areas) effects of fuel treatments. You cannot quantify the off-site effects of treatments if you don’t use a fire spread algorithm (e.g., stand-level modeling). This approach generates estimates of “conditional” burn probability, meaning the probability of a particular area (e.g., pixel) burning during a specified duration, given that a fire ignites in the analysis area. This is not an estimate of actual burn probability.

With regard to arguments that fuel treatment benefits (e.g., reduced fire behavior and effects) may not be realized when you take into account the actual probability of fire occurring in a particular area, we wrote this in a recent article published in Ecosphere 3(5):Article 38 (can be downloaded from http://www.cnr.berkeley.edu/stephens-lab/Articles.htm )

“The fact that we do not explicitly incorporate the probability of such an event occurring emphasizes that our analysis is not an actual risk assessment, i.e., expected loss multiplied by the probability of occurrence (Finney 2005). Rather, our assessment is more similar to a hazard analysis. While recent studies have included probability estimates of wildfire occurrence, and in particular occurrence of more severe fire in their assessments of fuel treatment impacts on C stocks (Ager et al. 2010, Campbell et al. 2011), there are several factors associated with the estimation of wildfire occurrence that lead to considerable uncertainty in results. First, the use of actual wildfire occurrence over the last two to three decades to derive annual burn probabilities (Campbell et al. 2011) reflects neither the historical (pre-Euro-American) occurrence of fire in frequently burned forest types (Stephens et al. 2007, Van de Water and Safford 2011), nor the projected changes in fire occurrence due to climate change in the future (McKenzie et al. 2004, Westerling et al. 2011). Second, changes in fire sizes resulting from fuel treatment implementation, particularly when considering the increased fire suppression efficacy associated with fuel treatments (e.g., Moghaddas and Craggs 2007), are not evaluated. This could lead to considerable over-estimation of fire occurrence in “treated” landscapes relative to “untreated” landscapes. A third source of uncertainty in the calculation of burn probabilities is related to the potential feedbacks associated with severe fires. In many dry forest types in the western US shrubs dominate for several decades following high-severity fire (McGinnis et al. 2010), and for that period of shrub-dominance the likelihood of reburning at high severity relative to intact forests is increased (van Wagtendonk et al. 2011). This feedback would affect calculations of high severity burn probabilities over time, particularly for “untreated” landscapes where more high severity fire would be expected.”

13 thoughts on “Following Up with the Fire Modelling Discussion”

  1. Key point “our analysis is not an actual risk assessment.” And the problems arise when people (authors or commentators) make policy recommendations based on these studies as if these were real risk assessments.

    If you saturate the landscape with fire then you remove the temporal variable related to actual fire frequency. Wildlife do not live in fire-saturated forests, they live in real forests that experience fire with some annual probability between 0 and 100, and the net benefits of fuel reduction are sensitive to the actual probability because it determines how likely it is that fuel treatments will subsequently experience fire (and yield benefits of modified fire behavior) and how many acres of habitat are treated and degraded unnecessarily. The study below showed a 10:1 ration of acres treated-and-did-not-burn vs treated-and-burned. De facto fire frequency (as modified by fire suppression) determines the actual ratio.

    Here’s a better explanation (again in the context of carbon) from John L Campbell, Mark E Harmon, and Stephen R Mitchell. 2011. Can fuel-reduction treatments really increase forest carbon storage in the western US by reducing future fire emissions? Front Ecol Environ 2011; doi:10.1890/110057

    “Abstract
    “It has been suggested that thinning trees and other fuel-reduction practices aimed at reducing the probability of high-severity forest fire are consistent with efforts to keep carbon (C) sequestered in terrestrial pools, and that such practices should therefore be rewarded rather than penalized in C-accounting schemes. By evaluating how fuel treatments, wildfire, and their interactions affect forest C stocks across a wide range of spatial and temporal scales, we conclude that this is extremely unlikely. Our review reveals high C losses associated with fuel treatment, only modest differences in the combustive losses associated with high-severity fire and the low-severity fire that fuel treatment is meant to encourage, and a low likelihood that treated forests will be exposed to fire. Although fuel-reduction treatments may be necessary to restore historical functionality to fire-suppressed ecosystems, we found little credible evidence that such efforts have the added benefit of increasing terrestrial C stocks.

    . . .
    “In a nutshell:
    “• Carbon (C) losses incurred with fuel removal generally exceed what is protected from combustion should the treated area burn
    “• Even among fire-prone forests, one must treat about ten locations to influence future fire behavior in a single location
    “• Over multiple fire cycles, forests that burn less often store more C than forests that burn more often
    “• Only when treatments change the equilibrium between growth and mortality can they alter long-term C storage

    “Conclusions
    “Across a range of treatment intensities, the amount of C removed in treatment was typically three times that saved by altering fire behavior.

    “the protection of one hectare of forest from wildfire required the treatment of 10 hectares, owing not to the low efficacy of treatment but rather to the rarity of severe wildfire event.

    “Long-term simulations of forest growth, decomposition, and combustion illustrate how, despite a negative feedback between fire frequency and fuel-driven severity, a regime of low-frequency, high-severity fire stores more C over time than a regime of high-frequency, low-severity fire.”

    Reply
    • But Tree, there is no such thing as a “real” risk assessment. All of the modeling you are talking about predicts the future. All are based on assumptions about the future.

      Yet, we often say “with climate change there are unknowable future events”. My view is that we would just try to deal with this qualitatively (what if there’s a lot more fire? what would happen with different times of drought? Because in reality we don’t even know that). Still we have many people using models to try to understand processes and interventions. Which is OK, but as you say how that is relevant to actual forest management is hard to say. That’s why sometimes the “best available science” gets us off the track.

      No one ever told me that they were doing a thinning project to store C, so I don’t know why you are citing that study…

      This is one of those situations in which people start out with a supposition that is not what people actually do (manage for C through thinning) in fire country, and then write a paper which is fine, but it’s not particularly relevant.

      Most fuels reduction is to provide defensible space for WUI’s. If we wanted to know why else people do it (restoration) say in Calfornia, we need to look at the purpose and need. I have never seen one about C storage.

      You say that doing fuel treatments where there is no fire, causes acres to be “treated and degraded unnecessarily”. But one could argue that in a predominantly unmanaged forest, as ours are, a thinning project provides important spatial-structural diversity.

      I do think you do have a point about affording to be able to do it. I think in the last couple of years there was a policy decision to focus fuel treatment bucks in the WUI. (does anyone have a cite to this?).

      We could possibly take a look at last years’ accomplishments and see where and how much mechanical treatment was out of the WUI, and check out the purpose and needs. I think this data might be relatively easy to find (the total acres by forest).

      Reply
      • Sharon, It is true that the future is uncertain. However that does not mean that we can’t try to make estimates of the probability of future events and that some estimates will be more accurate than others.

        You say “there is no such thing as a “real” risk assessment.” Maybe I can clarify by saying that a “real” risk assessment is one that makes an effort to use reasonable assumptions (like the anticipated annual probability of fire in that particular biophysical setting) instead of patently unreasonable assumptions (like 100% chance of fire after treatment).

        What I object to is policy recommendations that are based on completely erroneous psuedo risk assessments that assume 100% chance of fire during the period that treatments are effective, because this false assumption leads people to vastly over-estimate the risk of fire and vastly over-estimate the value of fuel treatments.

        If the results of modelling (and associated policy recommendations) are not sensitive to assumptions about fire frequency, then there is no problem, but when results are sensitive to estimates of fire frequency, we have to make as effort to make reasonable estimates. It’s as simple as that.

        Cases where it matters a lot include any proposal where short terms degradation is traded-off for long-term gains. Logging degrades habitat for species that prefer dense forests: spotted owl, fisher, goshawk, marten, pileated woodpecker, as well as reducing carbon storage. If we are going to claim that logging to reduce fire hazard provides net benefits that outweigh the unavoidable ecological costs, then we have to accurately model fire frequency (among other things).

        Reply
        • With a large part of the Sierra Nevada having 10 or more wildfires in the last 100 years, isn’t that close enough to “100% certainty” of wildfires?? I rest my case!

          Reply
  2. If you take away the hoped-for effect of thinning on fire behavior (slowing wildfire down, bringing it down from the crowns), the only reason to thin is for strictly silvicultural reasons of shifting the acre’s tree growth (cubic bd ft) from many small trees to fewer large trees. To propose any other “reason” for thinning is not accurate.

    Thinning (done properly) to reduce the severe aspects of wildfire in WUIs may be productive, but as stated by me before, we must critically examine the WUI mapping done previously. If these WUI depictions elsewhere are as silly and extensive as they are here in north Idaho, then Sharon’s rationale is weak. It is clear to me that the WUI maps for Kootenai and Shoshone Counties in Idaho were blatantly enlarged in an attempt to gain more intensive management (ie, logging or thinning) within. A direct effect of largely local decision-makers at the helm without “outside” review.

    Reply
    • Ed, you state that the only two “hoped-for” “reasons” for thinning forest stands is to slow wildfire and/or increase timber growth. That is not accurate. There are many (many) reasons to thin trees other than the two options you provide: e.g., to remove dangerous or diseased trees from a stand; to allow for more sunlight on the forest floor to encourage understory growth; to make money via log sales; aesthetics, etc.

      So far as “WUI’s” are concerned, they are just one more superfluous agency acronym that is used to help rationalize agency actions and budgets (in my opinion) and strategically divert public discussion and involvement. They are an imaginary construct that was only invented a few years ago. You can make them as big or as small as you like for any county anywhere and you still won’t be able to tell where the boundaries are without a map. Ask a local hunter or logger where they think their local WUI is. You might just get an honest answer.

      Reply
      • This is an interesting discussion. Maybe we need to develop a “landscape scale” fire suppression/management strategy that takes into account communities, developments of varying sizes, infrastructure (powerlines, dams, etc.), escape routes and highways with clear maps.

        It’s interesting also that Ed points out that local decisionmakers don’t have outside review. But they are elected officials and accountable in that sense. When we take things to court, I would argue that we lose accountability for the policy outcomes, as it becomes (is reduced to) a legal issue.

        It’s interesting that we did the experiment on local involvement plus a review by a national level group (Idaho Roadless and the RACNAC) for a contentious public policy issue, but that seems to have been highly unpopular with some national groups.

        We have large WUIs in Colorado too but it’s not because of our timber industry (on life support). I think it’s because people didn’t want to provide too detailed guidelines for their development and this is the predictable result. This result (what we might call “policy splatter”) is a natural consequence of “too loose” requirements.

        But I’m not judging here. Policy making is an art and not a science. Often in the decentralized Forest Service, when there is too much splatter, the FS changes the requirements and they do another round of whatever it is.

        Reply
      • Exactly, Bob! Some people seek to isolate aspects of thinning projects without looking at the whole of the many cumulative benefits that healthier, less dense forests bring, including wildfire resilience.

        Reply
    • Ed,

      Maybe another way to think about it was that the CWPP WUI designations were an attempt to gain, er, maybe even salvage opportunites for more intensive management in response to roadless area designations that were “blatantly enlarged”, thus effectively precluding any further active (and responsible) management of those landscapes? Sword cuts both ways…..

      I would agree though…using any sort of fuel reduction arguement is virtually pointless these days. Too much conflicting science and folks have assumed their stance. Lets just talk about the other benefits of responsible forest management already!?! Fuel reduction is a side of fries that comes with every meal….take ’em or leave ’em.

      Reply
  3. Good comment Ed, and basically the same thing here in Montana RE: WUI depictions silly and extensive. At one point back around 2005/06, when we were attempting to reach some common ground with some folks in Lincoln County, Montana (the extreme northwestern corner of our state) we were repeatedly told by county commissioners that the entire county, outside of a handful of designated Wilderness areas, was part of the “Wildland Urban Interface.” This, in a county with a population density of 5 people per square mile.

    Reply
    • I rather doubt that people living in and around the Bitterroot NF are “enjoying” the burned landscapes, the increased insect mortality on their properties, and their loss of property values. No, I am very sure I don’t want a sea of dead trees within a quarter mile of where I live, thank you very much!

      Reply
  4. I see no one addressing the human-caused fire issue. It isn’t sustainable to wait for a Fire Return Interval that doesn’t account for today’s human-caused wildfires burning in a tinderbox of drought-stricken and crowded forests. Just as we cannot go back to pre-European conditions, we also cannot go back to the pre-human conditions that “re-wilders” want. However, we CAN establish a new program of creating forests which will survive human-caused fires, as well as other sources of ignitions. And, for those preservationists who will surely claim otherwise, OF COURSE, we do not seek to fireproof entire forests. (Just heading some folks off at the pass that some of us know all-too-well.)

    Reply
    • Larry: The best way to stop human caused wildfires — assuming we want to retain a similar pattern of forests, woodlands, shrublands, and grasslands across the landscape — is to precede them with human-caused prescribed fires first, just like we’ve been doing for thousands of years.

      In the western US, our native plant and animal populations have adjusted to these regular acts of people over that period of time — ever since the glaciers receded and the megafauna went extinct. In the past 200 years we’ve added domestic livestock grazing, plowing, clearcutting, highway construction, exotic plants and animals, dams, parking lots, motors, plastic manufacturing and millions of more people to the landscape.

      We’ve also added the dehumanizing myth of “wildlands” to the mix by our attempts to “rewild” our native huckleberry fields, beargrass meadows, camas prairies, oak savannahs, and pine woodlands by renaming them Wilderness or roadless or WUI. And battling mightily for these changes in nomenclature and documented history with a legion wealthy lawyers, willing politicians, well intentioned dupes, and government grant writers.

      We cant’ go back to the past, and there really isn’t much reason for wanting to do so — but we can learn from the past: What grows where, and why; What happens when the weather changes, the sea rises, or the mountain explodes; Or a river is dammed or a city built.

      Or how to manage a forest. Indians didn’t have marking paint or chain saws, but they had fire and they had no reason to let it run wild on them. Probably the most important aspect of forest restoration, in my opinion, is restoring people (and fire) to the land. The native plants and animals would love it.

      Reply

Leave a Reply to TreeC123 Cancel reply