On the morning of November 8, 2018, a blaze started north of the community of Paradise that would grow to be California’s deadliest and most destructive wildfire to date. In disbelief, Heather Green watched footage of the Camp Fire as it spread across Northern California. “It was horrific,” she says. “People lost their homes, livelihood, everything.”

Quickly, her attention turned to a different image: maps of the smoke drifting south and overwhelming San Francisco. That’s where Green works as the city’s capital planning director and deputy resilience officer.

“We realized that the smoke was going to be around for a long time, and it was going to get worse. It was already bad on the second day,” she says. “That was terrifying. The schools closed.” San Francisco’s Department of Emergency Management responded by publishing a list of park facilities, libraries, art museums, and other private buildings where residents could find respite from the smoke-filled air outside.

Smoke from the Butte County Camp Fire shrouds the City of San Francisco. The fire was one of three major fires burning across California in November 2018. Photo: Shutterstock/AlessandraRC

The need for adaptation

San Francisco was one of the only Bay Area cities to publish a list of buildings with filtered, clean air. Still, Mayor London Breed recognized that severe wildfires are a new normal here and the city could do more. After the Camp Fire, she ordered a study of how to improve the city’s response. She wrote: “In the last few years, California has experienced multiple wildfires throughout the state causing unprecedented and sustained poor air quality in the Bay Area. While we provide support and resources to those lives and communities directly impacted by wildfires, we must also improve our City’s ability to adapt and respond to growing climate threats, especially wildfires.”

Before 2017, San Francisco’s air quality rarely registered problems on the Air Quality Index, a tool that represents air pollution on a range from “good” to “hazardous.” But during the Camp Fire, San Francisco’s air pollution was in the “unhealthy” range for nearly two straight weeks.

Now, San Francisco is assessing its city facilities and identifying buildings with air filtration systems. City officials are poring over data to find public- and private-partner buildings that are large enough to function as community gathering spaces.

“The emergency plan for the city will be revised to acknowledge that this is a priority and a concern. We will have guidelines for what to do next time,” Green says. “Where should people go? Why should they go there? With the fires last year, there was a lot of information coming all at once from different sources and it was sometimes contradictory. We need to get ahead of that. We need to have clear recommendations.”

Nearly a year before the Camp Fire, in December of 2017, the Thomas Fire ripped across Ventura County in Southern California.

A few months earlier, the Ventura County Medical Center celebrated the opening of a $305 million new wing, which broke ground in 2013. The fire burned not far from the hospital, at one point consuming a nearby hillside.

Mara Baum thought the fire might overtake the hospital. Baum, a principal at HOK, a global architecture and planning firm, worked on the project. Thankfully, it did not. And because of good filters, the hospital operated throughout the fire. “In fact, they took in patients from other hospitals,” Baum says.

Filters are key, she explains. Filter efficiency is graded on the MERV metric scale of 1-16. The higher the number, the finer the particle filtration. Commonly, buildings are coded for MERV 8, but a MERV 13 filter is recommended for filtering wildfire soot or other combustion byproduct. Activated carbon filters are used to remove odor from the air, including volatile organic compounds, chlorine, and other gases.

Left: Mara Baum, a principal at HOK, a global architecture and planning firm, worked on the Ventura County Medical Center project. Right: Jeremy Dennis is the town manager for Portola Valley, California. Portola Valley was one of the first customers of One Concern, a software service that uses artificial intelligence and machine learning to help communities predict and prepare for disaster impacts. Photo: Nader Khouri.

Left: Mara Baum, a principal at HOK, a global architecture and planning firm, worked on the Ventura County Medical Center project. Right: Jeremy Dennis is the town manager for Portola Valley, California. Portola Valley was one of the first customers of One Concern, a software service that uses artificial intelligence and machine learning to help communities predict and prepare for disaster impacts. Photo: Nader Khouri.

Resilient public spaces

These are examples of how resilient buildings can be used to help communities during environmental disasters. At a moment of increasingly unpredictable and severe weather, San Francisco and Ventura County highlight the need for cities to incorporate sustainably constructed buildings into their emergency response plans.

For years, leveraging public buildings during disaster relief has been a point of discussion in the resilient design community. Increasingly, leaders see the U.S. Green Building Council’s (USGBC) Leadership in Energy and Environmental Design (LEED) rating system playing a crucial role in supporting communities.

“A LEED-certified building that has a good filtration system is capable of providing so much beyond just the day-to-day improved air quality,” says Brendan Owens, senior vice president of strategic partnership and growth with USGBC. “It is a resilience that is absent in many buildings that are out there.”

Wes Sullens, director of codes technical development at USGBC, says that LEED credits create the potential for better air quality, and communities can and should use these resources. “The buildings can be used in a disaster scenario like this,” says Sullens. “But it all comes down to maintenance and upkeep.”

Sullens added that recertification is important. “There’s a performance aspect,” he says.

Resilience through data

While wildfires are a major concern in California, they aren’t the only emergency for which sustainably designed buildings can be community resources.

California has about 10,000 earthquakes annually. While most are only small tremors, California has three major fault lines in the San Andreas, the Hayward, and the San Jacinto. A century has passed since California’s faults created a major, city-shaking quake, but that could mean that the next Big One is up this century. For this reason, seismic retrofitting is key for keeping people safe.

In the municipality of Portola Valley, California, an hour south of San Francisco, town manager Jeremy Dennis can see the line of the San Andreas fault from his office window. “Obviously, it’s had an impact on our thinking around emergency preparedness,” he says. A major earthquake in the Bay Area could affect millions, so this rural community of about 4,500 understands it could be on its own. “Getting the best information about what an earthquake could do to us, and how to prepare and respond, is critical,” says Dennis.

So when One Concern, an innovative tech company from across the highway in Silicon Valley, came out with a software service that uses artificial intelligence (AI) and machine learning to help communities predict and prepare for disaster impacts, Portola Valley partnered with the neighboring town of Woodside and the Woodside Fire Protection District (which includes the area’s unincorporated communities) to become one of the company’s first customers.

One Concern is an AI for disaster resilience company, says co-founder and CEO Ahmad Wani. Its AI-enabled platform models the behavior and impact of earthquakes and floods, with a wildfire program in the works, to help communities build for long-term resilience. One of a small number of such ventures (with others including Geospiza, The Field Innovation Team, forest fire researchers at the Universities of Alberta and Oklahoma, and some Microsoft-supported initiatives under development), the company responds to a growing need: Worldwide, the annual number of natural disasters has increased since the mid-twentieth century by an order of magnitude.

Wani’s personal experience of two of those disasters gave rise to his sense of mission. The devastation caused by an earthquake in his native Kashmir in 2005 inspired him to become a structural engineer; being flood-stranded for a week on the roof of his family’s house while visiting in 2014 highlighted for him the difficulty rescue services often have in knowing who needs help, and where.

Left to right: One Concern’s co-founder and CEO Ahmad Wani. Digambar Ganjre is One Concern’s director of customer success. Recently retired Woodside Fire Chief Dan Ghiorso.

Left to right: One Concern’s co-founder and CEO Ahmad Wani. Digambar Ganjre is One Concern’s director of customer success. Recently retired Woodside Fire Chief Dan Ghiorso.

After completing graduate studies at Stanford, Wani, in collaboration with computer scientist Nicole Hu and earthquake engineer Tim Frank, founded One Concern to bring the capabilities of AI and machine learning to emergency preparedness and response.

With a client base that now includes such disaster-conscious cities as San Francisco and Los Angeles, One Concern’s platform draws on data sets covering hundreds of attributes across three key vectors: the natural environment, with data from such sources as topographical maps, soil surveys, and the U.S. Geological Survey’s seismic sensors; the built environment, with building code and permit data, zoning maps, and satellite imagery; and demographics, with inputs from the U.S. Census Bureau. “The good thing is that One Concern already has this data for every jurisdiction in the United States,” says Digambar Ganjre, the company’s director of customer success. “So we don’t need this data from the cities, and that drastically reduces the time-to-value for our customers.”

By combining data on the three vectors, the software is able to predict the impact of a range of events on specific buildings or pieces of infrastructure. It then aggregates its predictions to the census-block level, with an accuracy estimated at about 85 percent, and presents them on a map showing where and to what degree damage is expected—either in the immediate aftermath of an event, or in advance, through simulations that foster enhanced preparedness.

“It’s phenomenal,” says Woodside Fire Prevention District Chief Dan Ghiorso. “I can get on this platform and within 15 minutes I can tell you where in my district we’re going to have some challenges.” The Woodside-Portola Valley team meets regularly, once a month for the first year and now bi-annually, to run tailored simulations of a range of earthquake magnitudes, locations, and depths, and to review and plan for the projected impacts. “Historically, there would be a fair degree of creativity in coming up with these scenarios, or a degree of rigorous analysis around one scenario that may or may not happen,” says Ganjre. “With our platform, you can look at the thousand most likely earthquakes, and then understand the aggregate overlapping risks—so you really home in on what’s most likely to affect your community.”

In some cases, the software’s predictions jive with what the team already knows. Some older school and college buildings show up as problematic, for example, and their owners are aware of the issue, with an upgrade plan already in place. In other cases, the simulations have offered some surprises.

For example, of only two routes connecting two of the three fire stations in the 32-square-mile district, the direct road, which crosses a marshy area, may in some scenarios be lost to liquefaction. “Not one person in the room had thought of that until we saw it in the simulation,” says Ghiorso. “Imagine finding out on the day of a disaster, ‘Oh my God, the road’s gone!’ Or having one of the fire engines driving when the road goes down.” Forewarned, the emergency management team has planned for an alternate route. “Not that we wouldn’t have figured it out at the time,” says Ghiorso, “but now we don’t have to.”

Another surprise was the degree of variation in projected impacts. “When you prepare for an emergency, you tend to think about it in a monolithic way,” says Dennis. “But in some scenarios, a particular area of town could have a unique geologic impact.” That impact might mean the difference between having a road open or not, or having gas lines rupture in one area when elsewhere they’re fine. “The realization is that your planning efforts have to be more nuanced than simply ‘earthquake,’” says Dennis. “They have to be much more thoughtful.”

The demographic modeling has also yielded some fresh perspective, helping to identify vulnerable populations. “Understanding that you have a senior population in a particular part of town may mean going to those residents and providing them with upfront information,” says Dennis, “not just assuming they’re getting the information they need from the internet.” The demographic information also helps emergency planners anticipate where support may be needed after an event, such as at child-care centers. “The data lets you think about emergencies in a way that wasn’t possible before,” Dennis says.

While preparation for future contingencies is the simulations’ main point, some benefits have manifested already. Regular meetings, shared problem solving, even a plan for a shared Emergency Operations Center after one scenario predicted the loss of Portola Valley’s town hall: “It’s created a great working relationship between the two towns,” says Ghiorso. Looking ahead, the townships plan to continue running their practice simulations, layering in One Concern’s wildfire program once it’s operational, preparing for what the AI predicts, and looking for contingencies they haven’t yet considered. “We hope we never have to use this thing in real life,” says Dennis, “but because of this system and the conversations that have come out of it, I think we’re as ready as we can be.”

Keeping California cool

Erik Ring is a principal and director of engineering at LPA, an architecture and design firm that works on private and public projects, including many schools. He notes two key areas for resilient building design in California beyond air quality: seismic retrofitting and air conditioning.

California’s weather is typically pleasant, especially in the Bay Area’s Mediterranean climate. Summers are temperate and dry, and winters are mild. But in recent years, California has experienced a series of early, record-setting heat waves. Sultry days in California tax an already stressed power grid, which can cause outages, exacerbate smog, and increase the danger of wildfire. Many homes aren’t equipped with air conditioning, which can cause heat stress for people, especially seniors.

A solution: using schools as community cooling centers. “The idea is that in the event of an emergency or a natural disaster the schools could serve the community by providing a place of refuge,” Ring says. “During a heat wave, if people in the community don’t have access to air conditioning, they go to their local public school to find cooling.”

He worked on the Montgomery Middle School in the Sweetwater Unified School District, a LEED Platinum building located in Southern San Diego County. Montgomery was designed as a school and as a community resource with tutoring centers and a public library.

“In a community emergency, it’s a place where people could gather,” Ring says. “That was an element of the design.”

California is a hotbed of emerging resilience strategies