Wildfire Risk By Market: What I Would Actually Do | Part 2

Five wildfire-exposed markets. Three states. The FEMA data from Part 1 showed where the risk concentrates and why county averages hide everything that matters. In Part Two, Principal Architect Patrick Chopson, AIA, goes market by market and does what most wildfire design content does not: rank the design responses in order, from the ones that determine whether a building survives to the ones that are expensive reassurance.

Part 1 established the thesis: wildfire risk hides inside county averages, the tracts where it concentrates carry loss rates that exceed the worst hurricane markets in America, the insurance market has already repriced them, and the code has not moved. It also made a claim I want to defend properly here, which is that fire-resilient design is the most effective cultural preservation tool an architect has in a wildfire market, because a building that survives is a neighbor who stays.

As I write this, the National Preparedness Level is at 5 for the first time since October 2024. Ninety-two large fires are burning. More than 25,000 personnel are committed. Forty-two thousand fires have burned four and a half million acres this year, and the Northwest alone has 48 uncontained large fires. This is not a forecast. It is Wednesday.

So let me be useful instead of ominous. Below is the order I actually specify things in, followed by what changes market to market.

First principle: materials do not work alone

The most important finding from the January 2025 fires is not about any product. IBHS surveyed 252 structures in the Eaton and Palisades burn areas within a week of the event, and the pattern they quantified is this: homes avoided damage when there was greater spacing between structures and lower connective fuel density, which allowed noncombustible building materials to perform as intended.

Read that clause twice, because it is the whole game. Noncombustible materials have a performance envelope. Exceed the radiant and ember load they were tested for and they stop working.

The Washington Post documented two homes in La Viña, twenty-five feet apart, both with stucco siding and cement tile roofs. One burned to the ground. The other did not.

That is why product-first wildfire design fails. You cannot buy your way out of a site plan. Fire-resilient design is a system with a sequence, and the sequence has an order of operations that almost nobody follows, because the expensive visible items get specified first and the cheap invisible ones get value-engineered out.

Here is the correct order.

Tier 1: nothing else matters if these are wrong

1. Structure separation and connective fuel. This is a site plan decision and it is free.

Radiant exposure from a burning neighbor at twenty feet will defeat assemblies that would have survived the wildland fire front. The most consequential wildfire decision on most projects is made at entitlement, in plan, by someone who thinks they are doing yield optimization. Separation, and the elimination of continuous fuel between structures (fences, hedges, wood decks, trellises, side-yard storage), buys more survivability per dollar than every product below combined. In a conflagration, your neighbor is your primary hazard.

2. The roof assembly, including the parts nobody draws.

“Class A” is the beginning of the sentence, not the end. A Class A tile roof with open ends at the eave and rake is the failure mode I saw across the Palisades. Specify it completely: fire-rated underlayment, bird stops or mortar closures at eaves and rakes, metal valleys and drip edge, no exposed sheathing, no gaps larger than an ember. On a tile roof this is a few thousand dollars of detailing that determines whether the roof is a shield or a sieve.

3. Listed ember-resistant vents. Not hardware cloth.

Code-minimum 1/8-inch mesh slows an ember. It does not stop an ember storm. Specify vents tested and listed to ASTM E2886, which resist both ember intrusion and flame penetration. This is the cheapest high-consequence line item in the entire scope, and it directly addresses the interior-ignition failure mode that took most of the houses I walked through.

4. Box the eaves.

Exposed rafter tails are an architectural choice that costs you the attic. Enclosed soffits in noncombustible material, with the vents from item 3. There are honest ways to get shadow and depth in an elevation without building a heat trap.

5. Zone 0: the first five feet, noncombustible, no exceptions.

Noncombustible ground surface. No bark mulch. No plastic. No firewood, no patio cushions, no stored anything against the wall. No combustible fence or gate touching or within five feet of the structure. California has failed for six years to make this a rule, San Diego made it one itself, and it remains the best dollar-for-dollar mitigation available. Specify it, draw it, and put it in the CC&Rs so it survives the first landscape contractor.

6. Glazing.

Tempered on both panes, smaller lites on high-exposure elevations, and noncombustible or fire-rated frames. No vinyl frames in the WUI: they deform and drop glass at radiant loads well below flame contact, and once the glass is out, the interior is the fuel. Where there are operable openings, exterior noncombustible screens and, on the highest-exposure elevations, deployable shutters.

7. The holes people forget: garage doors, gutters, decks.

The garage door is usually the largest unsealed opening in a house. Gasket it. Noncombustible gutters, and either guards or no gutters at all. Under-deck enclosure in noncombustible material with metal screening at the base, because an open deck is a firebox with a house attached.

Tier 2: real, worth the money, buy in this order

8. Interior residential sprinklers.

Since the documented failure mode is interior ignition from ember entry, the one active system with a real case is the one that fights the fire that already got in. An ember in the attic at hour two, while pressure still exists, is exactly the fire a sprinkler head is designed for. Be honest with clients about the limit: municipal supply frequently degrades during a conflagration, so this protects you against the early ignition, not against hour six.

9. Backup power and independent communications.

Grid restoration in WUI corridors after a major fire is measured in days to weeks, and Public Safety Power Shutoffs mean you may lose power before the fire. This is what makes a shelter-in-place strategy real rather than aspirational, and it is non-negotiable for senior housing and any mid-rise residential in an interface tract.

10. On-site water storage for suppression, where hydrants are absent or thin.

Real, often required, and frequently misunderstood. Size it with the fire authority per NFPA 1142. But understand what you are buying: this is water for firefighters defending your building. It is not a self-defense system.

11. Project-wide noncombustible fencing.

This is the Marshall Fire lesson and it only works collectively. Wood fencing carried that fire structure to structure through a suburb. Metal, masonry, or fire-rated composite at the project level, in the specs and in the governing documents, costs marginally more than cedar and eliminates a documented failure mode. One lot doing it accomplishes nothing.

Tier 3: theater, unless

I get asked about these constantly, usually by well-meaning clients who have seen a compelling video. Here is my honest ranking.

12. Exterior sprinkler systems, pool-fed or roof-mounted: mostly theater.

Trace the failure chain. The pump needs power, and power fails, often preemptively. Municipal pressure collapses when hundreds of hydrants draw simultaneously and pump stations lose electricity. A wet film on a roof evaporates in minutes at sixty miles an hour and eight percent relative humidity. And most of these installations have no listed standard behind them at all.

Give me an unhardened envelope with a roof sprinkler and I will show you a burned house. Give me a hardened envelope with no sprinkler and I will show you a standing one. If a client insists, the only version worth building has dedicated stored water, generator or gravity feed, and a designed nozzle layout, and it comes after everything in Tier 1. Note also that California’s framework is beginning to recognize active defense systems for insurance credit. Earning a credit and surviving a fire are not the same achievement.

13. Automated vent closure: solve chemistry with chemistry, not motors.

A motorized damper is a mechanical device with a control board that requires power on the worst power day of the decade. Passive intumescent vents do the same job with no actuator, no wiring, and no failure mode. Do not buy a motor to solve a materials problem.

14. Radiant heat barrier products: the real version is geometry.

The radiant threat in a dense WUI is your neighbor’s house at twenty feet, and the answer is separation and glazing performance, not a wrap. Setback is a radiant barrier. Twenty feet of noncombustible ground is a radiant barrier. Buy those first, and usually you will find there is no budget left for the product, which is the correct outcome.

15. Green roofs as fire barriers: no.

An irrigated succulent roof in July is a wet mat. The same roof in October, during a PSPS with no pump running, is a mat of dry biomass sitting on your sheathing. Add the dead load fighting your seismic budget in California and this is a bad trade three different ways. If you want a noncombustible roof, specify a noncombustible roof. I am a climate advocate and I will still tell you this is the wrong place for that gesture.

16. And the big one: concrete and masonry versus wood frame.

The honest answer will not please anyone selling premium structural systems. Framing type is second-order. The assembly is first-order.

Chapter 7A does not require noncombustible structural framing for most residential rebuilds, and it does not need to, because the documented survivals are overwhelmingly wood-framed houses with noncombustible exteriors and no openings. The much-photographed Palisades home whose owners had already lost one house to fire in 1993 survived with steel-reinforced walls and a metal roof, but the operative detail is the one everybody skips: it had no eaves and no roof vents. What saved that house was the absence of openings, not the presence of steel.

Where mass construction genuinely changes the calculus is at multifamily scale. Type I or II construction buys fire separation, acoustic performance, insurance treatment, and a seventy-five-year asset in a single decision, and the cost delta lands inside a pro forma instead of on a family. That is a real conversation and we have it often.

But do not sell a homeowner concrete as fire protection when the same money buys far more survivability in the roof closures, the vents, the glazing, and the first five feet. That is not engineering. That is upselling with a safety story attached.

Now, market by market. The data is the same federal dataset. The design problem is different in every county.

San Diego: the broadest exposure in the dataset

Source: FEMA National Risk Index v1.20 (December 2025). Analysis by cove.

Twenty-eight percent of San Diego County’s total expected annual loss comes from wildfire, the highest share of any county here. That is $430 million a year in wildfire EAL across $639 billion in building value. Sixty-three percent of tracts carry some wildfire exposure, 11 percent carry rates above $500 per million, and the worst tract hits $16,667 per million, nearly five times New Orleans’ county-wide hurricane rate.

San Diego’s problem is not a handful of canyon-edge tracts. Decades of building into chaparral hillsides and canyon systems produced a wildland-urban interface that runs through established residential and commercial neighborhoods, not just along an exurban fringe. The county also carries significant earthquake (28 percent) and inland flood (42 percent) exposure, so a wildfire-only strategy addresses roughly a quarter of the hazard.

What changes here: San Diego did the thing the state would not. Zone 0 is now in the municipal WUI code, effective for new structures as of February 28, 2026, and for existing structures on February 28, 2027. If you are designing in the City of San Diego, the first five feet is not best practice anymore. It is plan check.

What I specify:

Ignition-resistant construction as the baseline for anything within a mile of vegetated fuel, not as a high-risk upgrade. Complete Tier 1, and treat the ember pathway audit as a formal review deliverable at DD, signed off like an egress review, because it is the single most consequential half-day an architect spends on a WUI project.

Then the thing most teams get wrong: multi-hazard detailing. San Diego’s near-even split across wildfire, earthquake, and flood means noncombustible assemblies have to be seismically detailed and the seismic system has to maintain the continuity of the fire envelope. Fire resistance and lateral performance are not separate conversations; they are the same set of connections. The 1994 Northridge event is the reminder that these hazards arrive in sequence.

By building type: For multifamily, the broad WUI means a project in an established neighborhood can carry the same exposure as one on raw land. The spec scales with the data, not with how rural the site feels. For senior living on a canyon edge, evacuation timeline is a life-safety design input, not an operations plan. More than 800 residents of nursing homes, group homes, and assisted living facilities were evacuated during the January 2025 LA fires, and many were separated from essential services and from each other for months. Design for that.

Los Angeles: invisible until it is not

Source: FEMA National Risk Index v1.20 (December 2025). Analysis by cove.

LA County’s county-wide wildfire rate of $91 per million is the lowest of the five, because 72 percent of tracts have zero wildfire EAL and mathematically swamp the ones that do. The roughly 3 percent where it concentrates run up to $7,373 per million, 81 times the county average.

LA’s wildfire problem is a set of specific geographic corridors: the Santa Monica Mountains, the San Gabriel foothills, the canyons through Malibu, Pacific Palisades, and Bel Air, the hillside communities above Pasadena and Glendale. Vegetation, topography, wind, and building density combine into rapid spread.

The rest of the profile is earthquake (60 percent) and inland flood (37 percent). Wildfire is 2 percent of the county total and 100 percent of the story in the places where it exists.

What I specify:

The first design decision is a data question: is this site in the interface? A project two miles from the Santa Monica Mountains ridgeline is a different building than one in the canyon. The tract data draws the line. Our job is to design to it rather than to the neighborhood’s self-image.

Ember exposure is the dominant failure mode and I will keep saying it until it is boring. Santa Ana winds carried firebrands well ahead of the front in January 2025. Houses ignited from the inside. Full Tier 1, with particular attention to glazing and tile roof closures, because coastal LA is full of beautiful barrel tile roofs that are ember funnels.

Earthquake-plus-fire is LA’s signature combination and it constrains material choice more than most teams expect. Every building in the county needs to perform seismically; in the WUI tracts it also needs to perform against ember and radiant load, with integrated detailing that satisfies both.

And backup systems. Utility disruption in LA’s WUI corridors after a major fire runs days to weeks. Backup power, water reserve, and communications independence belong in the basis of design for any mid-rise residential or senior housing in an interface tract.

By building type: In the hillside communities the wildfire spec and the seismic spec are both non-negotiable and they interact. For senior living, evacuation on narrow canyon roads makes shelter-in-place capability a design requirement, not an operational plan. Eighteen of the nineteen deaths in the Eaton Fire occurred west of Lake Avenue, and the majority in the area that did not receive its first evacuation order until 3:25 in the morning. Evacuation is a system that fails. Design as though it will.

For anyone rebuilding here right now: understand that the 110 percent footprint rule governs your schedule, your code cycle, and your all-electric obligations simultaneously. That is a design decision made in week one, and getting it wrong costs a year.

El Paso County (Colorado Springs): building into the fire path

Source: FEMA National Risk Index v1.20 (December 2025). Analysis by cove.

Ninety-seven percent of El Paso County’s census tracts carry some wildfire EAL, the highest penetration rate of the five counties. Nineteen percent carry rates above $500 per million. The county rate of $348 per million is third-highest here.

Colorado Springs is a growth market expanding west and north into the foothills, straight into the interface. The Black Forest Fire in 2013 destroyed more than 500 homes and was the most destructive in state history at the time. Waldo Canyon in 2012 burned into established neighborhoods.

The hazard mix is the most diverse of the five: wildfire 19 percent, inland flood 49 percent, tornado 12 percent, hail 8 percent, plus winter weather, avalanche, cold wave, and lightning. This is a genuinely multi-hazard market in a way the California counties are not.

What I specify:

Design to the actual fire environment, not the mapped boundary. Colorado adopted the 2021 International WUI Code with amendments, but the development frontier is moving faster than the maps. If there are vegetated fuels within 300 feet of the site, the WUI spec applies regardless of what the boundary layer says. Nobody will make you do this. Do it anyway.

Roof and wall assemblies carry the survivability here, and dry air, low humidity, and frequent high wind mean ember transport distances are long. The envelope is the firebreak.

The detail that gets missed: hail and fire interact on the roof. At 8 percent of total EAL, hail is not a footnote in El Paso County. A Class A roof compromised by hail impact in year three is a fire vulnerability in year five. Specify impact-rated roofing that maintains its fire rating across its service life, and defend that line item, because it is a classic VE casualty and the failure shows up a decade later on somebody else’s watch.

Site access is a design-phase decision, not a civil afterthought: emergency vehicle access, turnaround capability, and alternative egress in communities with limited road networks.

By building type: With 97 percent tract penetration there is no low-risk submarket to retreat to. The WUI spec is the county baseline. For commercial and light industrial, the hail-plus-fire roof interaction belongs in the capex model, not in a surprise assessment after the first storm.

Boulder County: the Marshall Fire proved the thesis, and the state ignored it

Source: FEMA National Risk Index v1.20 (December 2025). Analysis by cove.

Boulder County’s wildfire rate of $225 per million is the second-lowest of the five, but 96 percent of tracts carry some exposure and the worst tracts reach $3,286 per million. Wildfire is 11 percent of total EAL, with inland flood at 50 percent, avalanche 11 percent, tornado 9 percent.

On December 30, 2021, a wind-driven grass fire destroyed more than a thousand homes in Louisville and Superior. It remains the most destructive wildfire in Colorado history, and it happened in a suburb that did not consider itself a wildfire market. It moved through developed neighborhoods, not remote forest. Structures ignited from embers landing on combustible roofing, wood fences, and landscape materials.

That is the proof point for this entire series. Wildfire risk is invisible until it concentrates, and the design of the building determines whether it survives.

What actually happened next is the part worth arguing about.

Boulder County ran a program called Rebuilding Better, an incentive-based, voluntary effort to get homeowners to rebuild above code on energy and resilience. Roughly 70 percent of eligible households participated. Seventy percent, voluntarily, from people who had just lost everything and had every excuse to build back cheap.

Colorado eventually adopted the Colorado Wildfire Resiliency Code in July 2025, three years after Marshall and after FEMA denied the state $101 million in resilient infrastructure grants partly because it lacked a statewide standard. Full enforcement began July 1, 2026. But the code applies only to new construction in WUI areas, the existing stock that burned is unaffected, and stakeholders involved in drafting it have called the requirements modest. The homeowners chose resilience at a 70 percent rate. The state wrote a floor. I do not know how to describe that gap except as a failure of nerve.

What I specify:

Grass fire in a suburb is a different design problem than forest fire on a ridge. Forest resilience emphasizes defensible space measured in hundreds of feet. Grass fire resilience in a built-out subdivision is almost entirely about the immediate envelope and the first five feet: noncombustible roofing, noncombustible siding at the first floor, boxed eaves, listed vents, and the removal of every combustible attachment. Wood fences connected to structures. Combustible deck materials. Bark mulch in beds against the wall.

Fencing and landscape are architectural decisions in Boulder County, not landscape-consultant decisions. The Marshall Fire spread structure to structure through wood fencing. A project-wide noncombustible fence policy breaks that chain.

Benchmark to the post-fire standard, not the pre-fire minimum. The code that permitted the houses that burned is not the target.

And design for wind, because wind was the accelerant: gusts of 60 to 100 miles an hour for more than ten hours. Conveniently, the details that resist ember intrusion (sealed penetrations, listed vents, continuous sheathing) also improve wind performance. That is the argument to use with a value engineer.

By building type: Any multifamily project in Louisville or Superior will face scrutiny on its fire hardening from residents, lenders, and insurers who lived through this. For mixed-use, the ground-floor commercial envelope is usually the weak link: storefront glazing, ventilation intakes, and recessed entries that trap embers.

Jackson County, Oregon: the small market with outsized risk

Source: FEMA National Risk Index v1.20 (December 2025). Analysis by cove.

Jackson County carries the second-highest wildfire loss rate in this analysis at $457 per million. Ninety-eight percent of tracts have some exposure. The worst tract reaches $3,605 per million. Wildfire is 21 percent of total EAL, with earthquake at 29 percent and inland flood at 44 percent.

The Almeda Fire on September 8, 2020, destroyed more than 2,400 structures and killed three people, burning through the Bear Creek corridor between Ashland and Medford: mobile home parks, subdivisions, commercial areas. Like the Marshall Fire, it was wind-driven and it moved through developed areas. The structures it took were largely unprotected against ember exposure, with combustible roofing and siding.

This is live right now. On July 10, 2026, the East Evans Creek Road Fire ignited about 26 miles north of Medford, grew past 1,500 acres in hours, and triggered Level 3 “Go Now” evacuations. The Governor invoked Oregon’s Emergency Conflagration Act for the first time this season and the State Fire Marshal mobilized four task forces. As of late July, Oregon and Washington together account for 48 uncontained large fires.

Jackson County is a market of 223,000 people. It does not have the capital depth, institutional developer base, or insurance capacity of LA or San Diego. The Almeda rebuild has been slower and shaped by what carriers will cover and what construction capacity exists locally.

What I specify:

Code-minimum construction is what burned. Composition shingle, wood or vinyl siding, open eaves, unscreened vents. The design response is the exact inverse, and it is not a luxury package, it is the documented delta between what burned and what did not.

Manufactured and modular housing gets the same spec. Almeda disproportionately destroyed manufactured homes and mobile home parks. Ignition-resistant standards apply regardless of construction type, and the fact that HUD-code housing is cheaper is not a reason to build the next generation of it to the standard that failed.

Earthquake is the second conversation. At 29 percent of EAL, in the Cascadia Subduction Zone’s influence area, the structure has to address seismic performance and fire resistance together.

And here is the one unique to small markets: recovery capacity is a design input. Post-fire recovery in a 223,000-person county is constrained by local workforce, material supply, and claim timelines. Designing for speed of repair, with standardized assemblies, locally available materials, and modular strategies where appropriate, is a resilience decision. In a small market, a two-year rebuild and a five-year rebuild are two different communities.

By building type: The Bear Creek corridor was residential, retail, office, and industrial burning together, so this applies to every type in the valley, not just the hillsides. For affordable housing, which was a large share of what burned, the wildfire spec adds cost that has to be in the development budget from day one. The alternative is building the next generation of affordable housing to the standard that failed in 2020. That is not thrift. It is a deferred loss with a lower interest rate.

What connects these five markets

Five counties. Three states. Different climates and development patterns. One consistent finding: wildfire-resilient design is a set of envelope, material, and site decisions, most of them made in the first weeks of design, and almost none of them required by code.

The insurance market has moved ahead of the code in every one of these markets. Carriers are pricing at the building. Buildings that can document ignition-resistant construction, ember protection, and defensible space stay in the insurable pool. The ones that cannot migrate to the residual market or go bare.

The same three futures from our hurricane series apply. In the quiet decade, a fire-hardened building carries modestly better insurance terms. In the event, it is the difference between a standing structure and a slab. In the slow reprice, it is the qualification for institutional capital in markets carriers are exiting.

A fire-hardened building performs in all three. A code-minimum building performs in one.

But I want to end somewhere other than the pro forma, because the pro forma is not why I do this.

The thing worth protecting is the place

Go back to what I saw in the Palisades. The basketball hoop over the empty driveway. The oak that lived. The rose bush and the fence.

Eighteen months later, twenty-seven homes have certificates of occupancy out of over 6,800 destroyed. Investors have bought something like 40 to 45 percent of the lots. The 1950s rent-controlled walkups are not coming back as 1950s rent-controlled walkups. The debris and contaminated soil took the better part of a year to haul out before anyone could pour a foundation, and the soil testing program only wound down this month.

What arrives next, if nobody intervenes, is competent, code-compliant, insurable, and generic. Suburban plan sets dropped onto one of the most specific landscapes in North America, a place with canyon walls and ocean light and a hundred years of eclectic accumulated architecture, rebuilt as though it were anywhere.

That is the second loss, and it is permanent in a way the first one is not. A burned house can be rebuilt. A neighborhood’s character, once replaced with a plan set, does not come back.

Here is why it happens, and it is not because developers hate beauty. It is because designing a house that genuinely belongs on a specific canyon lot, in the material language of a specific neighborhood, that also satisfies Chapter 7A, stacks the available insurance credits, clears the 110 percent footprint rule, and lands inside an insurance settlement that is a million dollars short, is slow and expensive to produce by hand. A pre-approved plan set is fast and cheap. After a fire, fast and cheap wins, because families are out of runway.

Nobody is going to be talked into a slower, costlier process while their ALE clock runs out. The intervention is to remove the penalty on specificity.

Beauty without rigor is scenery. Rigor without beauty is machinery. A rebuild needs both, and it needs them fast, because the clock on the community is twenty-four months and the clock on the permit is longer than that.

The federal data told us which tracts would burn. The insurance market told us what it would cost. The code told us the minimum. None of that tells us what to build.

That part is still ours.

For methodology, data sources, and caveats, see the appendix in Part 1.

About cove’s Principal Architect Patrick Chopson, AIA

Patrick Chopson, AIA, is Co-Founder and Principal of cove, an AI-powered architecture firm transforming how buildings are designed and delivered. A licensed architect with 20+ years of experience, technologist and building scientist, Patrick focuses on the intersection of AI and Architecture.

He previously co-founded the building performance consultancy Pattern r+d and co-authored Build Like It’s the End of the World (Wiley, 2025), a guide to decarbonizing AEC. His work has been featured in Architect Magazine, TechCrunch, and ArchDaily, and he regularly collaborates with developers and industry leaders on next-generation solutions.