Research forester Dr. Sarah McCaffrey explores some of the data, and counterarguments, surrounding that narrative. (Photo: Santa Rosa neighborhoods after the 2017 Tubbs Fire. Photo by David Loeffler, United States National Guard shared via Flickr Creative Commons.
How you tell a story influences what conclusions people draw from it (think Aesop’s Fables). Over the past decade, the overarching American wildfire narrative has become fairly focused on three dynamics: fuels buildup due to suppression, climate change, and the expanding wildland-urban interface (WUI). But what are these narratives based on?
There is a fair bit of research and debate as to when and where fuel overloading and climate change will have the most influence, in fact too much to be easily citable here. However, according to research for a paper that I’m developing with Matt Thompson of the Rocky Mountain Research Station and Courtney Schultz of Colorado State University, few data support the WUI story. Specifically, only limited data support the arguments for why and how the WUI is contributing to the wildfire problem.
For instance, a commonly cited concern is that 50–95 percent of wildfire suppression costs can be attributed to the protection of private property in the WUI. However, when our team followed the citations for this statement, we found that they all led back to data from a single 2006 Office of Inspector General (OIG) report (PDF, 1.57 MB). Further, the report reached that particular conclusion using data that most social scientists would find problematic at best. Stay tuned for my team’s paper to learn more about that report’s limitations.* We have found a few studies that examine the breakdown of suppression costs in a more rigorous manner, using meaningful metrics (e.g., the number of homes near the fire perimeter and the percent of private land). But, while they do find a positive association between homes being located in the WUI and suppression costs, none demonstrate causality or conclude that the costs of protecting private property come close to the low end of OIG’s estimates. Nor can we find any study that tracks changes in the WUI with changes in suppression costs over time, even though a key perceived “problem” with the WUI is that its expansion will automatically increase wildfire suppression costs.
Similarly, although it is often stated that wildfire-related housing loss is increasing, it is hard to find any data that support or deny this claim, as wildfire-related housing loss hasn’t been tracked consistently until fairly recently. Further, even the existing data are problematic. For instance, the National Interagency Coordination Center (NICC) claims that 2,638 homes were lost nationally in 2015 (see page 9 of this NICC report, PDF, 830 KB); while CAL FIRE reports that 3,217 homes were destroyed in California alone that same year. Further, while it is too short of a time span to draw any solid conclusions, it is worth noting that both datasets suggest that higher losses are periodic and due to specific wildfire events. For example, the graph below, which I developed using NICC annual summary reports, illustrates that when excluding the state that lost the most residences in a given year — say Tennessee in 2016, the year of the Gatlinburg fires — home loss (shown in green) is relatively flat.
Inaccurate Stories Are Unlikely to Lead to Effective Solutions
Accepting WUI narratives that lack sufficient supporting data can lead to developing solutions that are less likely to have the desired impact, because they may be targeting the wrong problem. Further, these proposed solutions may have unintended consequences because they don’t take existing data, or lack thereof, or the larger context into account. For instance, a common recommendation based on the WUI narrative is to limit development in fire-prone areas. However, while we do have fairly robust science concerning factors that make an individual home more fire resistant, we don’t have much information on what type of larger scale development is associated with improved wildfire outcomes. There are only a few relevant studies, and they offer conflicting “solutions.” For example, some studies suggest clustered housing would help with decreasing suppression costs, but others suggest clustered housing would lead to more losses.
Another common recommendation based on the WUI narrative is to form a national fire-insurance program similar to the National Flood Insurance Program, which appears to be seen as a means of ensuring that WUI residents bear the cost of living there. This recommendation ignores the significant evidence concluding that the National Flood Insurance Program has done little to shift the cost burden from federal taxpayers (and in fact has probably increased it) or to limit housing development in flood-prone areas. This argument also seems to assume that if individuals can’t obtain or afford insurance, then they will choose to not live in the WUI. However, there is no evidence that this is the case. It’s just as likely that individuals will continue to live there for a variety of reasons (e.g., amenities, affordability in other regards) and just take their chances, essentially making those with fewer resources even more vulnerable.
Further, insurance levers can raise an equity consideration: if insurance did have the desired effect, it is likely that only the wealthy would be able to live in the WUI since only they could afford to pay higher insurance premiums or to self-insure. It is worth asking whether that is an acceptable tradeoff.
Shifting Our Language
Through this process, I have begun to question whether the WUI is still a useful construct. While it may have been useful at one point to draw attention to the increasing intermingling of housing and natural vegetation, it now seems to create an artificial distinction that may be misleading. I doubt many of those who lost their homes in the recent California wildfires, particularly in Santa Rosa, thought of themselves as living in the WUI. Further, a preliminary analysis** by the University of Wisconsin shows that regarding the Tubbs Fire, many of the losses were not technically in the WUI. Instead, the largest portions of homes lost were in areas considered either too dense or not dense enough to be classified as the WUI. Forest Schafer’s recent blog post about how Lake Tahoe partners are rethinking their WUI is another example of the limitations of drawing artificial lines when assessing wildfire risk.
Forest Schafer’s recent blog post about how Lake Tahoe partners are rethinking their WUI is another example of the limitations of drawing artificial lines when assessing wildfire risk. Click on the icon above to read Forest’s post.
And then there is the question of how limiting development in the WUI would work. Beyond equity issues and the fact that we don’t have strong evidence regarding what “better” land-use planning for wildfire looks like, I always wonder, “So, where are people supposed to live then?” As one recent newspaper headline stated: “All Californians Live in Fire Country Now.” Perhaps just thinking about how we can best live in fire-prone landscapes may be a more useful way to frame the discussion, rather than creating artificial distinctions based on housing density and natural vegetation.
*This blog describes initial findings from a paper that Courtney, Matt and I are working on. The articles referenced in this blog post will be identified in that paper, which will be available on Treesearch once it is published.
**Author’s note, updated 2/2/2018: In clarifying the best contacts for the Tubbs Fire information described above, I also got an update on their findings. The proportion of the homes lost “outside of” the WUI (about 30 percent) is consistent with fires nationally, but distinct in that most of the losses from the Tubbs Fire were in areas of high density rather than very low density, which is more common nationally. More specifics and contacts are listed below as a comment from me, made on February 1.
Credit: Sarah McCaffrey
Sarah McCaffrey, Ph.D., is a research forester for the USDA Forest Service, Rocky Mountain Research Station. Her research focuses on the social aspects of fire management. This work has included projects examining wildfire risk perception, social acceptability of prescribed fire and thinning, characteristics of effective communication programs, and incentives for the creation and maintenance of defensible space. She has also initiated work examining social issues that occur during and after wildfires, including evacuation-decision making, agency-community interactions during fires, public perception of wildfire management overall, views on the Cohesive Strategy, and perceptions about what it means to be a fire adapted community. She received her Ph.D. in Wildland Resource Science in 2002 from the University of California at Berkeley.
Please note that comments are manually approved by a website administrator and may take some time to appear.
Thank you, Sarah, for raising these questions. I look forward to learning more from this research once it is published.
Thanks for the insightful post, Sarah. I really appreciate you helping us examine our terminology, beliefs, and narratives as we work to find effective solutions based on data. Do you happen to have a link to the Wisconsin analysis of the Tubbs Fire?
Thanks Sarah, there is a lot to think about here. It sounds like there is still important research needed to clarify the issues, but i can see important implications down the road. At a local level, how do communities prioritize and make designations? What are the implications for Community Wildfire Protection Planning?
Maybe it is time to shorten the title to Wildland Urban fires as the interface and intermix directs one to believe these “built fuel” fires only happen in or at the edge of the Wildland area. It is a set of conditions not necessarily a geographical location where many of these structures burn, more often than not by airborne embers/brands far removed from the Wildland fire. To better protect our homes from Wildfire, Residential building codes for low density as well as high density neighborhoods need to change to be ignition resistant. Landscaping and natural fuels within very short distances from the structure need to be mitigated and enforced with local ordinances if you are going to live or build within Wildland landscapes.
I liken our dilemma to nutrition…science says do this and then it says do that, and meanwhile you have to eat so you just cram something in your mouth and hope for the best. I hear the lesson and it’s that we really don’t know enough yet to say with authority what people should do, and there are conflicting studies, including one in the recent NFPA Journal that might contradict this new McCaffrey study. NFPA says that home loss IS increasing according to NIFC data. From 1985 to 2000 there were 400 homes burned per year in large wildfires (what about the small ones?), and from 2001 to 2011 the average jumped to 1,354, then jumped again between 2012 and 2016 to 3,456! 2017 will likely cause another uptick. Who do we believe? I’m going to crawl under the covers with a Big Mac and hope for the best.
Some great points here, Sarah. It reminds me of the book How to Lie with Statistics (1954). “The secret language of statistics, so appealing in a fact-minded culture, is employed to sensationalize, inflate, confuse, and oversimplify,” writes author Darrell Huff.
I agree that we need to be very cautious about the statistics we use to tell a specific fire narrative—structure loss statistics from wildfires have always been an elusive figure. And there are some clear limitations with WUI definitions that don’t account for contributing factors beyond a geographical area, as Gary points out.
We also have information we can leverage on how land use planning fits into the fire adapted community discussion. For example, as a wildfire profession, we know:
• How structures ignite.
• Standards that specify which construction materials, building techniques, and landscaping will reduce structure loss (when properly implemented and enforced).
• Standards for driveway and roadway requirements that allow apparatus to safely access neighborhoods and improve first responder and public safety during evacuations.
The legal mechanism to address structural vulnerabilities, access, and other similar topics is through land use development codes, WUI ordinances/codes, and/or building codes. Fostering relationships between fire response agencies and development decision-making authorities (e.g., planning departments, planning commissions, and development councils) is critical to helping both professions understand different perspectives and community planning objectives.
Not surprised there is confusion, for better or worse science rarely provides the unequivocal answer many desire, but we can at least be clear what the information does or does not say. A few thoughts on the NFPA paper. 1) I’ve run into the 1985-2000 numbers in my investigations but none provide specific year to year numbers to be able to assess the validity of the claim, generally just referencing NIFC – which sounds good but if you can’t verify……. 2) the time frames for all four periods of house loss are not the same so the numbers really aren’t comparable, especially as the time period chosen can really influence results. For instance, focusing on the 2011-2014 numbers of the graph above I would be able to show a downward trend in house loss , and if I chose the period from 1990-1994, which included four of the 20 most destructive California fires, including the Oakland Hills fire, I suspect the average would likely be much higher than 400/year. The only real point I wanted to raise is that there is a lot more uncertainty in WUI trends than is commonly thought. And in many ways it is indeed comparable to nutrition, (or Opioids- the whole argument that Opioids were not addictive that led to the current problem apparently can be traced to one letter to the editor in a journal )- developing too absolute a plan (no fat!) on limited or problematic data can lead to worse rather than improved outcomes.
Sarah,
It is worth noting that the numbers you cite from Redbook are “structures” and as such include outbuilding and “other structures” which, maybe not obviously, tend to increase loss numbers in rural areas where a single property often has a number of structures and only one “house”. Thus, conflating these metrics as houses is misleading.
It is true that accurate loss data has historically been a big problem, and even within our own agency the standards for counts have not been consistent, and further, the process for conducting damage inspections has varied by jurisdiction. Often, when we have urban conflagrations, its on municipal lands, and the local agency has been responsible for damage assessment, and they have done things their own way. We are making strides , but it actually is a a difficult data collection problem.
All that said, I think your ideas here are fertile to bringing the discussion forward. We in California have been mapping WUI for going on 20 years, and recognize that within that construct, there is a lot of different type of ground (and require different policies and actions). As you can imagine, the public seems conflicted in what constitutes the problem. At its base are some funadmentals of uncertainty and the inability to predict relatively rare occurences. However, if you look at North America, and trends, with 2016 and 2017, while the time series are problematic from an annualized timeframe, at the decade level they are unmistakable.
Hope you are well,
–dave
Good to hear from you Dave,
All good points. And excellent catch on the homes/structures distinction which is important- (its a point I’ve made elsewhere so I’m rather embarrassed I did not catch it). Going through my notes – looks like I was so thrilled I had finally found some multi-year numbers for California, which described them as homes lost, I missed identifying and correcting that problem when I finally tracked down the CalFire info source. My apologies for perpetuating that problem.
I’m not adverse to the notion losses are increasing – just hesitant to jump on that bandwagon without looking carefully at the floorboards and making sure they are solid. One question in this regard, which you are better placed than me to answer, is whether the house loss is rising proportional to the changing exposure (increase in number of housing units in the WUI overtime)? All things being equal one would expect higher house loss as more homes are exposed but are the losses increasing or decreasing proportional to changes in the number of homes exposed?
And I might argue it is more than the public that is conflicted about what constitutes the problem…… 🙂
Hi Steve,
In clarifying best contacts on the Tubbs fire information I also got an update on their findings which indicates that the proportion (~30%) of homes lost outside the WUI is consistent with fires nationally, but distinct in that most of those losses on the Tubbs fire were in areas of high density rather than very low density housing, which is more common nationally. Specifics and contacts are listed below.
“the majority (70%) of buildings destroyed in the Tubbs were in the WUI. This was lower than other CA fires (where 84% of destruction occurred in the WUI) and was unusual in that 27% of destroyed buildings were in high-density areas (compared 3% in CA fires overall). Note that these numbers are still preliminary, but we don’t expect them to change drastically with the finalized data. We’ve also been working on revisions for an accepted manuscript that shows 69% of national destruction (in fires between 2000 and 2013) to be in the WUI. In this paper we found that the majority of destruction outside the WUI occurred in areas of very low housing density (well below the threshold used to delineate the WUI), which lines up with what we found for CA fires overall “
In addition at the national level:, “93% of destroyed buildings outside the WUI were below the WUI housing density threshold of 6.17 homes/sq km. Also, the average distance of destroyed buildings outside the WUI to the WUI was only 1.6 km.”
Anu Kramer – hakramer[at]wisc[dot]edu
Miranda Mockrin – mhmockrin[at]fs[dot]fed[dot]us
Sarah, you are a great instigator of constructive discussion. There’s always more to learn about wildfire’s impact on communities. Funding for mitigation is NOT unlimited so I’m always supportive of focusing where we can get our best risk reduction results. If we truly got community risk reduction for the money we all spend on it, we’d be much further along the community fire adaptation trajectory. Knowing where to put out efforts is key. As we’ve discussed many times, linking community risk reduction to reducing the costs of response is futile …… while we may make communities safer, it is unlikely to greatly reduce the cost of response. What community wildfire risk reduction CAN do is save lives, homes/structures, a way of life, and a lot of heartache ….a worthy co-relation. Let’s keep talking.
I think it’s always important to consider whether we’re asking the right questions, and great that you’ve started a conversation about it. You’ve definitely convinced me that the research isn’t conclusive, and we often don’t draw WUI lines in a useful way. However, as Molly Mowery and Pam Leschak point out, there are things we do know about reducing risk, and there’s a lot to be said for focusing on the built environment in the WUI, even if it doesn’t reduce suppression costs.
I will add one point relative to your question, “‘So where are people supposed to live then?'”
Truly that question reaches far beyond the realm of wildland fire. With booming populations in many areas, there are numerous reasons to consider it carefully, and ideally to plan for it. It’s quite an oversimplification (and perhaps meant to be) to dismiss it.
I agree that we should acknowledge that many people outside the defined WUI have flammable vegetation around their homes, and beyond striving for defensible space, that isn’t likely to change. However, there are big differences in wildfire risk–and risk to responders–based on location. Gary is right–geography isn’t the whole picture. Yet it goes without saying that it is a factor–where structures sit relative to topography and varying vegetation should absolutely be considered. I’m back with you on the tricky question of how limiting development in the WUI would work!
I love the analogy of flood ins. as it has not paid for losses in flood plains, just a FEMA funds don’t cover wildland fire losses.
The actuarial data needs some serious examination and much better insurance consideration. Currently the occurrence rate is so low it is not considered as a major event threat by insurance companies.
As for the number of homes lost as change over time; the time frame currently being used, supposedly due to data access problems, is ridiculous. I can guarantee you, the insurance industry has data for fire losses going back to the 1930’s. {here is the gas} If they played nice and shared their actuarial data with an oversight agency, it would be very helpful.
The supposed lack of government agency data for such statistical treatments is blocked because of format and system changes. The data is still there going back to the forties, somebody just doesn’t want to do the system melds or more accurately doesn’t want to find the money to pay for doing the research work. Citing data from the last twenty years as the only data is just not accurate. imho
Did the findings described in this post ever get published?