As wildfires grow more frequent and devastating, they expose vulnerabilities in infrastructure, governance, and community preparedness. Tackling this escalating threat demands interdisciplinary solutions that address not just the immediate risks but also the broader systemic changes driving extreme weather events.
Recorded on February 18, 2025, this Matrix on Point discussion featured Christopher Ansell, Professor of Political Science and Executive Director of the UC Berkeley Center for Catastrophic Risk Management (CCRM); Kenichi Soga, Distinguished Professor of Civil and Environmental Engineering and Director of the Berkeley Center for Smart Infrastructure; and Marta Gonzalez, Associate Professor of Civil and Environmental Engineering and City and Regional Planning. Louise Comfort, Professor Emerita and Project Scientist, Department of Civil and Environmental Engineering, moderated.
This panel was co-sponsored by the UC Berkeley Department of City and Regional Planning, the Charles and Louise Travers Department of Political Science, the Department of Civil and Environmental Engineering, the Center for Catastrophic Risk Management, and the Center for Information Technology Research in the Interest of Society (CITRIS).
The panel was presented as part of Matrix On Point, a discussion series promoting focused, cross-disciplinary conversations on today’s most pressing issues. Offering opportunities for scholarly exchange and interaction, each Matrix On Point features the perspectives of leading scholars and specialists from different disciplines, followed by an open conversation. These thought-provoking events are free and open to the public.
Podcast and Transcript
Listen to a recording of this panel below or on Apple Podcasts.
Transcript
[UPBEAT MUSIC]
WOMAN’S VOICE: The Matrix Podcast is a production of Social Science Matrix, an interdisciplinary research center at the University of California, Berkeley.
CORI HAYDEN: Welcome to the Matrix. My name is Cori Hayden. I am the interim Director of Social Science Matrix this semester from the Anthropology Department, otherwise.
It’s a real delight to welcome you all here for this panel, which was organized on relatively short notice. And thank you so much to Louise Comfort and colleagues here for helping put this together, not just helping put it together, for taking the initiative and making this happen, obviously, in response to the devastating wildfires in Los Angeles. We’re going to explore some of the many dimensions of the issues that come up with urban and wild wildfires threats, bringing together a range of absolutely critical perspectives from political science, civil and environmental engineering, and, of course, from the Center for Catastrophic Risk Management.
We’d like to thank our co-sponsors as well. City and Regional Planning, Center for Information Technology Research in the Interest of Society, CITRIS. Now, I know that Louise Comfort, as moderator, will have some substantive introductory comments, so I will keep it short here.
I do want to– let’s see if I can do this. Oh, yes. –do my due diligence as center director and remind you or let you know of some upcoming events at Matrix.
We have a very packed semester of extraordinarily timely conversations. Upcoming next Monday– Virtual Realities and Digital Spaces. Thursday, March 6, really interesting panel, lunchtime panel, on mainstreaming psychedelics.
Monday, March 17th– An Author Meets Critics Panel With Areej Sabbagh-Khoury from the Department of Sociology to discuss her recent book, Colonizing Palestine. Additional events are up here. Please do keep us in mind as you are looking for interesting things to do with all of your copious spare time. But today, as our topic is the Los Angeles wildfires. And we’re going to be talking about risk, resilience, and collective action. I will leave the floor to our panelists, but let me introduce our moderator, Professor Dr. Louise Comfort.
Louise Comfort is project scientist with the Department of Civil and Environmental Engineering and the co-principal investigator for the NSF grant, Designing Smart, Sustainable Risk Reduction and Hazard Prone Communities, which runs from 2022 to 2025 here at Berkeley. She’s Professor Emerita of the Graduate School of Public and International Affairs and was the director of the Center for Disaster Management at the University of Pittsburgh from 2009 to 2017. As a fellow– she’s a fellow of the National Academy of Public Administration and received the 2020 Fred Riggs Award for Lifetime Achievement on International Comparative Administration.
She studies the dynamics of decision-making in response to urgent events, earthquakes, tsunamis, hurricanes, wildfire, COVID-19, among them. That list was probably going to just keep getting longer. Without further ado, let me turn it over with enormous gratitude to Louise Comfort. Thank you.
LOUIS COMFORT: Yeah.
[APPLAUSE]
Thank you. Thank you very much. And I am delighted to introduce this panel, a sad panel but an important panel and one that we really need to pay attention to.
Let me just give you just a very brief context of these, were catastrophic– and we’re looking at catastrophic risk, a set of conditions that’s the extreme of the extremes. And this is probably the most consequential, the most costly, the most difficult wildfire we have seen.
Just some brief statistics. It was not just one wildfire. It was 10 wildfires burning successively. In the two big wildfires, the Eaton wildfire and the Palisades fire, were burning simultaneously.
So this was a situation where we had– it burned over 16,000 acres. The numbers keep changing. And there were at least 25 people who were dead, found dead, but likely the consequences from of smoke, pollution, et cetera, will lead to more and thousands and thousands of people harmed.
The evacuation was really rather remarkable. About 200,000 people were evacuated in a matter of hours. It was chaotic. It was difficult.
Cars were being pushed off the road, but they got people out. And the total cost was estimated at $250 billion, largest, most expensive wildfire we’ve had. The critical thing about this event is that was the initial event.
And already what we see is the rains came, and the mudslides that followed, and the floods. And then the whole discussion of recovery. So we’re looking at these catastrophic events of very complex systems that are interacting and interrelated. And I’m very pleased to say that we have, from our own Berkeley faculty, some expert analysts to address exactly this.
And we have Kenichi Soga from the Department of Civil and Environmental Engineering. And Kenichi Soga, if I read all of this, I think I’m going to shorten it a little bit. He is not only professor, and he’s a distinguished professor of all kinds of things, but he’s also the director of the Center for Smart Infrastructure and also the principal investigator of a project that is focused on designing resilience for communities at risk and the interaction between the built infrastructure, the organizations that manage it, and the people in the communities and interacting with them.
And second, we have Marta Gonzalez, professor of city and regional planning, who has looked at modeling in complex systems. And then we have Chris Ansell, who is professor of political science, is focused on governance issues, and especially governance across jurisdictions. So this is what you’ve seen, but I do have to point out this was a photograph.
And anyone who lived through the ’91 fires here will see that ominous standing chimney. And then this was the critical issue. It’s where we are now in– excuse me, cascading and compound risks.
And this is the burn scar from both the Eaton and the Palisades fire that is now subject to mudslides, landslides, flooding. And so our challenge is really to understand these complex conditions, the dynamics that are driving these extremes, and to anticipate the risk for the next time, which will surely come. Now, I’ll turn it over to Kenichi.
[APPLAUSE]
KENICHI SOGA: So I’m Kenichi Soga from the College of Engineering, Civil and Environmental Engineering Department. I see my colleagues here. So it’s great to have our College of Engineering and Civil Engineering colleagues here.
We’re coming from a very technical side. And I think gradually it goes to Marta and then Chris into social science side. So please bear with me.
We have a project really promoting what we call socio-technical digital twin. But before I do that, two weeks ago, Louise and I were in LA area looking at the aftermath of the wildfire. And these are the photos that we took. And we took a lot of photos in Palisades and also Altadena.
And one thing that you see– the burned the car on the right. And it’s the first time I saw no EV.
So we start to see this new technology creating an issue on the other– when you have a wildfire, what do we deal with this EV-related cars and that sort of things? So that’s another interesting area of research, I think, that we need to think about.
During the weekend, I went through a variety of news articles and figured out what are the issues on the evacuation, which I’m focusing a little bit more today. But you can see there are a lot of articles, and you’ve probably read some of the articles related to that. The one on the right– I’m not sure you can see it. It’s the purple one. It’s showing the evacuation order that happened. It’s from The Wall Street Journal.
I have to go through, check with the– going to LA to find out more about it, whether this is true or not. But you can see that the evacuation order was done on the one side of the street that you see on the purple side. And then, the after eight hours, evacuation was on the left side of the street. And then you can see on the bottom figure showing that who were killed in that particular incident you see more on the left side.
So there’s an interesting– think about evacuation and that thing. And this is where we come familiar earlier with Paradise and Camp Fire event that I started working in this area with Louise. And I’m going to show you more on that.
At College of Engineering and Center for Smart Infrastructure, many of our colleagues know that we run a lot of simulations. And the simulation is coming from the street level to the regional scale. So in the left, we see the area where we are. There are about 7 million people. There are about 15 million trips every day.
We model each individual trip and see what’s going on in each of the roads. And then we say, OK, Bay Bridge goes down, what’s going to happen? On the right is our water network, is pipelines that you see. And then, when earthquake happens, what’s going to happen to the pipelines, and then how we recover? So these are the things that we do in our simulations.
And really we’re looking at systems of systems or how one system affects the other through the simulations. And then, we hope that will inspire some of– or work together with our social scientist colleagues.
The SimCenter is an NSF-funded Center. So it’s a collection of our research colleagues working in not only wildfire, but also earthquakes and also on tsunamis and other things. So if you want to know more about it, please let us know.
This is an example of when earthquake happens. There’s a Hayward earthquake that happens here. And you can see the ground shaking, which is shown on the left top. And then, you can start to simulate which part of the pipeline will get damaged and then how much water will go, so you do a simulation of the water. And then you start to see earthquake may not happen in one particular location.
It’s probabilistic. So you have to do a lot of simulation to figure out. And then, we can also simulate what happens to the red-type buildings, and then traffic disruption. So these are the interactions that we do simulate.
And then looking at the functional recovery, how long it will take, and how do we recover. So these are working with East Bay MUD. This is a project with Caltrans.
For example, Caltrans have thousands of bridges. Of course, Bay Bridge, Golden Gate Bridge are good, but then there are thousands of bridges. When earthquake happens, which one is going to be? So we have a project to see which bridge is very important so that everybody has access to hospitals, everybody has access to fire station, everybody has access to police stations. So these are the things that we’re working with Caltrans, and looking at emergency recovery and functional recovery.
The project that I want to highlight today is what we call a smart and connected community. It’s a National Science Foundation project that we’ve been working together with the counties of Alameda and Marin in particular. And Louise is one of the colleagues. But Steven Collier, who you may know, and then Michael Goldner from the mechanical engineering, we all work together with the UC Santa Cruz and UC Davis colleagues.
For my part, it’s really using socio-technical digital twin, the simulation that you see, a wildfire simulation that you see on the top, a traffic simulation. So what you see in the dot is like a usual traffic. It’s ordinary traffic that happens.
And then, you start to see some people in the blue evacuating. And then, what are the interaction? What are the bottlenecks?
Obviously, it’s a simulation. So it’s not going to be true. Every event will be different, but you start to see what-if scenario. What are the things happening? And then it allows the community to discuss with the local government by looking at each other and trying to understand each other.
So I’ll go through this a little bit more carefully, and that’s what we do. So what we do in this particular project is that Louise will look at through the interviews and looking at the network. So Louise may talk about this later on, but then there is a formal network, how people talk to each other during the wildfire event, but also there’s an informal network like fire councils and all these organizations. And then, the link may be related to PG&E, for example, not through the local government, Cal Fire and Cal OES.
So on the formal network, you see Cal Fire and Cal OES linking closely together. But then, in the informal one, it’s really fire councils that I see some of the colleagues here linking to PG&E. We do simulations on wildfire modeling and agent base. And then important part in the most important part is that how to model the communication, how the organization talk to each other, how the organization talk to the public, and how do we model that is a tricky part. But I rely on Louise to work on that.
And so we can combine everything together to see what’s going on. And I’ll show you some later on. And then my colleagues from UC Davis and UC Santa Cruz creates games out of that. So because these are complex system, outcomes are quite complex sometimes. So how do we bring that into gaming environment is what we’re working on.
In terms of social science questions, these are the three questions that my colleagues highlighted. You see on the left? One is called social dimension, which is complex time. How do people look up time, how they think about time?
Your time is different from my time. Sometimes, my time go quickly. Your time goes very slowly.
What are the times? And I guess that’s something about complex time scales of risk and capacity for action. So again, whether it’s a local, or government, or regional, how do that– risk and capacity can be looked at and make action?
And the action is related to the third one, collective cognition and action, is that how community– it’s not only a person making action. It’s really how the community collectively make a– realize what is going on at that time and make the right action.
So again, it’s about cognition and action as a community. And that’s what the– another third question that my colleagues came up.
So in the project, these are the important questions to answer. And then, myself, I will work on the right side on the technological dimension, is to figure out how it works together. And obviously, by showcasing those together in community is what we want to do.
So Camp Fire is probably, you may remember, in 2018, that happened in November 8. And, of course, the Camp Fire wildfire happened, as you can see, on the simulation. And we see how fast it went to the Paradise. And within one hour or two hours, it consumed all the town. And obviously, everybody has to evacuate, as you can see on the right bottom.
So we do simulation of this in terms of wildfire and then communication model and how traffic goes. Oops, sorry. And then, how the traffic simulation happens? And then, compare that with the reality and try to see.
Obviously, there are a variety of issues. For example, if you evacuate, you may have two cars in your household, so you want to use two cars. You may have three cars, so you may have three cars. But it may be better to just have one car because it don’t create traffic jams. So the question is that if you only have one car, what would have happened?
At that particular day, cell tower went down, so people didn’t have a communication. And what would have happened? So again, if you don’t have that communication part, what would have happened? And that’s something that we can look at as well.
The most important part is really looking at the timeline of what happened. And then, this is the video of the fire chief going to the site in the morning, where actually the fire has consumed the city, and then town, and then people are evacuating. So Louise and I and ourselves got together to really looking at the timeline because that’s so important in terms of understanding what really happened and how organization behaves. So we really want to do this for the LA Fire as well. And we just submitted a supplementary request to NSF, but we’ll see how NSF goes.
The incident is that fire arrived at Paradise at 8:00 AM. And then 10:30, the communication went down. Actually, the Paradise had 14 zone divided, and they had an award-winning evacuation plan. But then that assumed a staged evacuation. But in this case, it went so fast the fire chief had to decide everybody evacuate at the same time.
And the question really is that the right thing to do? And this is something that we can simulate and understand. You can see that there’s a four exits, different roads.
And then, at every point, because the fire was coming from the right side, they had to stop the access road stop. And then you can see when that happens. So again, that allows us to look at the simulation, how the people evacuate.
They did use a contraflow, meaning that there are four lanes to two, but then they 4 lanes. But then the issue was that sometimes if you’re on the right side, you’re not very used to it. Your police chief and the other cars are coming against you. And so you have to figure out is that the right thing to do. So there was confusion on that particular part as well.
So that is a finding that we made. And then we’re trying to simulate the most important part of the communication challenge is really how people decide to evacuate. And that, we see that on the wildfire in LA as well, is that you– people, the fire chief will come and say there’s a siren saying that evacuate, but people don’t usually evacuate.
It’s only your neighbor knocking the door saying you’ve got to evacuate that you evacuate. So these are the things that we do see often, but we have to create a timeline. So that’s what you see on the right, is that we inform population and how they evacuated. So these are an important part of our simulations.
And at the end, we do simulation like this for this particular car. Each individual cars are modeled. But then you can see time it took for 90% of evacuees to leave Paradise. So that’s what you see on the right top, which is called baseline.
So if you see on the right top, I’m not sure I can show you. This is the baseline, which took about five hours, six hours to evacuate. But let’s say if you had one vehicle to go, then it takes about three hours. And then if you do contraflow, what would the effect?
We have a phased evacuation. It did have an effect, but then actually doing an immediate effect did have a good results as well. Real-time information is about if your smartphone is working, if you knew about the traffic.
But in this case, you only have one way to go anyway, so it doesn’t really help. But in Berkeley Hills, there are different ways to evacuate because there are small roads, so real time– we know that real-time traffic information does help to evacuate faster.
Obviously, it’s not only evacuation, but then if you’re evacuating under the fire, it becomes your lifetime trauma as well, being exposed to fire. So you can be safe, but you’re under the fire, creates a psychological, mental issue in the future. And therefore, we also look at how long you are in the fire as well.
So the simulation does like this. Actually, what you see on the right is simulation. We’re working with City of Berkeley. Maybe you’re familiar with Marine Avenue, Marine Circle. And there we do simulation of that and then trying to work together what are the issues on that particular Marine Avenue, and then how do we evacuate effectively?
We do have a game. So if you go to a QR code, you can look at some of these games that my colleagues from UC Santa Cruz are developing. There are entertainment game developers, but then they also call them serious game developers. So we’re working with the community to showcase these ones.
We also have a startup called WUI-Go!, trying to use this particular simulations and then trying to make more evacuation personalized. So you can see some of the apps that we start to create. And these embedded simulations are part of that. But at the same time, this particular software is– trying to make it small so that it works without cell coverage.
And then we’re trying to do that as well. So we tried to put the information as possible inside your small computer on your smartphone. So if the internet goes down, at least you can see how you evacuate. So it gives you different routes by putting, if it is blocked here, what’s going to happen?
So this is my last slide just to see two sites that we looked at. Altadena, which is on the left– and then, Palisades on the right. And just looking at the geographic but also street, they’re all quite different.
In other words, we also see in the other cases, Berkeley Hills, we work on Sunol, we work with Marin County, Inverness area, but also with the Novato. They’re all different street condition, which means that they’re all different scenarios that we need to think about. One is more important than the other, and that’s where it’s very important that what you learn from one does not really reflect the other.
What we need to do is to find out some other things, and hopefully that what we want to do with these simulations is not just understand what your belief is correct, but then also find out what your belief may not be correct, and find out more scenarios in your community.
And therefore, when something happens, you can see which one is the right one for your community to take and make the right decision. So that’s what we’re doing at the moment. Thank you very much.
[APPLAUSE]
MARTA GONZALEZ: All right. We were in the civil engineering domain. I belong to civil engineering and city planning. So I would like to bring some data-driven planning and modeling in wildfire research.
I had to put things in context. And what is surprising to me is, if you see, from 1985 until a few years ago, the number of fires, contrary to what you may be perceiving, is not increasing. However, the area burned is increasing a lot, which is this black line here.
And then, the suppression cost increases. And California is right there with the US suppression costs. But what is most surprising to me is the number of infrastructures that have been destroyed.
So if you see here, from the last 16 years, it is only this amount of infrastructure. But in the orange line, until 2019, we have this. If we include the wildfire of LA here, the red zone is already going to be all this size because this is the campfire with 18,000 buildings infrastructures, and we have 18,000 infrastructure.
So the amount of infrastructures destroyed is growing an exponential rate. And that brings us to the urban planning domain. The problem is that we know how costly housing is in our state.
And we have this model of growth that is low density growth, also called sprawling. And that brings us to go from the urban side to what we call the WUI, the wild urban interface. And that is a problem we need to address.
And, well, it’s a political problem, not only a modeling problem. So that is what brought me, the urban planning had brought me into this problem. And here is just– from 2020 until today, how many more blocks increase the WUI.
So here we have LA, is in a dramatic context and also the Bay Area, so all over California. This is the PNAS 2024 by Greenberg is showing us where the WUI blocks are increasing. And here, in the right-hand side, we can say the climate.
So this is the infrastructure planning growth, and this is the climate, which means that we have more bigger areas fire. So the fire is getting worse, and we are building in the WUI. That is our problem that we need to tackle.
And it’s a sobering case. I will not solve it. But let me go into three examples in which I think data and modeling has helped us.
And this is the first one, what brought me into this topic. So it’s basically how we predict wildfire behavior and intensity of spread. It was through the realization of this type of law. We want to spread when– this is in the wildfire, not in the WUI.
Just in the wildland, when we are spreading the fire, we need fuel type, fuel moisture, wind speed and slope. And the physics of that is the so-called Rothermel equation, which are not microscopic physics like the one Michael Goldner is doing. This is semi-empirical laws. And these laws are in different simulators, Prometheus in Canada, Farsite here in the US, like the one Kenichi was using.
And the problem is that this physics is old. The simulators were built in the ’50s probably. And I thought as a physicist, and engineer, and data scientist, we should be doing better, something that everybody can use, that is friendly. Let’s bring the IT here. Then I had this student from Iowa, a brilliant student, that developed– we shared this thing. It has to be better. Let’s develop it ourselves.
He wrote down the equations in a cellular automata model, and it is open source. And here what we have is what’s the angle of the ellipse, and what is the rate of spread in the front, in the back, and in the flank. So we are able to– because it’s relatively simple physics, we are able to write down the equation, Cristobal was, and develop the model.
And here is where Minho comes, our PhD student, to follow up on the thesis of Cristobal. And we have here the fire spread of our model itself to fire in Santa Barbara versus the standard fire simulator that is Farsite. And we can do better because the physics of these problems are simple. Then we have the typical problems, that here is the real fire scar. Here is what the simulator saw. So we are falling short in how to model the fires.
We have the hope that applying machine learning would give us better models. Sobering story. Just simple black box optimization was better. And we set the fire, change the shape of the ellipse, parameters and get better results.
So right now we have an open-source fire spread simulator that can produce fire scars. It’s not very elegant because, again, it’s not the physics of the problem, is semi-empirical models. But we have it and we expect to use our models with CRM to have something open source. And as we know, Berkeley is very keen in open source.
Then, the second story is facility allocation, is again, where is the communities? And there is one aspect that I see, the environmental science modeling is not together with the human models. And this is what brings us into the second story.
We have the ability to download and process the whole 2 million nodes of the road network in California. We are able to download all the fire stations, and we want to see something as simple as the accessibility to fire station of the different census blocks in the state. However, we need to do a facility allocation, not only take into account where the people is, that is more the planning had. That is not doing the environmental science part.
We also have in every block what is the fire behavior, that is, the fire intensity in the block in the state. And then, we have this optimization model that is for every block of 0.1km, what is the shortest travel time from the fire station, and what is the fire behavior in that block? Then, you have a risk index that is the combination of these two. That is the shortest time to the fire station.
If you have these too long, your risk is very high. You have the fire behavior. And here comes the number of infrastructures and amount of people.
So we have all the different variables. And these type of thing, when we are doing optimization, we come still– I think we fall short. The best ideas for facility allocation may come from the social science.
And here, we have the fire behavior only is worse in the North. If we only look at the amount of people and infrastructure is here in the South. This is a LA County. And with this, together having fire behavior and sociodemographic utility, we have this weighted utility to then minimize the risk to every block and allocate the facilities.
In this paper, it was before the fires in LA, what we show is that all this region in the South are lacking access to fire stations. And if we relocate the facilities, we have all this shortened in the shortest travel time to the fire station. So here is where we come from, infrastructure data and optimization.
The last story brings us more into the WUI and land use. And it’s assessing the mobility in LA in context. And here we have 21 cities around the world. Notice that we have from the more sprawl to the less sprawl.
And there, if you look at the name, we have Latin America, Europe, China, and California. And what we see interestingly is that when we compare, let’s say Boston area with LA, those are the probability density function of wind in a radius of travel for every mobile phone user in these cities. And what we see, in short, is that in cities like Bogota, the poorest population live in the skirts and need to travel more. In cities like LA, the poorest population live in downtown. So who is the sprawling is the surprise, surprise, the richer neighborhoods in California?
And then we started seeing something very interesting, is depending– what distance you are from the CBD, how many people you find? And here you see in LA all this amount of people that travel everywhere. In Boston, you travel more if you are far from the CBD. This is a monocentric city. But LA, mobility wise, is a polycentric city.
So we are sprawling and traveling everywhere. And when we compare polycentric, monocentric, which is the travel behavior, with the development of the land where people is, which is the Gini of the population distribution, we see that LA is in this a box, that is, polycentric and dispersed. And when we see this, a plot of the 21 cities mentioned, California cities are in the worst parts, dispersed and polycentric.
So we got to put this into our model. So with slide, I would like to wrap up. We need to develop physics-based models informed, hopefully by AI, to improve prediction of fire behavior.
Data science and decision science allow us to target vulnerable locations at the state level. Our next step includes market design, Mansi, I am looking at you, for land use planning and more compact urban development. And that’s it.
LOUISE COMFORT: Thank you.
[APPLAUSE]
And now, I’d like to introduce Chris Ansell, who will take us into the governance nexus.
CHRISTOPHER ANSELL: Good.
LOUISE COMFORT: Thank you.
CHRISTOPHER ANSELL: OK, hi, everybody. Thanks, Louise. Thanks to the Matrix for organizing this.
My co-panelists, thank you very much. Strangely, our topics complement one another, even though I don’t have any models here at all. So Louise suggested that I focus some of my comments on governance, which is what I’ve done.
And before I get into the details, and maybe I’m dating myself, but I want to make an analogy between the Los Angeles fires and Hurricane Katrina. Some of you may remember that. That was very formative for some of us to think about. I know Louise worked there as well.
And just like in the LA fires, there was a blame game that went on. And I’m sure you spotted it in the newspaper. It started very early in Katrina and continued. And you may recall that a lot of the blame fell on FEMA.
And FEMA was the villain of the story. And it was argued to be too slow and too unresponsive to the needs of the local community. OK, and so FEMA certainly didn’t perform amazingly well, and there were definitely some mistakes made. But I recall a conversation that was important for me with a colleague who Louise knows named Arjen Boin, a Dutch scholar, also a well-known crisis management scholar.
And he said– he ended up writing a book about Katrina. And one of the things he said is everybody was focused on how to analyze the errors of FEMA.
He said that’s kind of misleading. He said if you really look at it, the best emergency management agency in the world would have failed in Katrina. And focusing on their errors, not that they’re unimportant, that’s not the point.
Not that they’re unimportant, but focusing on the errors of FEMA is distracting you from the big picture. And that stuck with me as an important thing. And I think a similar thing as I read the newspaper articles.
A similar thing is true about LA. We got focused on a lot of blaming across– between the mayor, and between the governor, and between the LA Fire Department, and other people. And if you focus on that, you can find errors, things they did wrong, but it sort of distracts you from the big picture. So I want to try to put it a little bit in a big picture for you.
And this quote up here by John Keeley, who is a US Geological Service Survey fire ecologist based in California. And just to paraphrase, he said something like, and this caught my attention as I was thinking about this, how this puts things in perspective. He basically said when the winds are blowing like this, all bets are off.
So the severity of the winds were just overwhelming almost anything you could do in response. Now, again, that doesn’t mean that there weren’t ways you could improve, that there were things you could do better. I’m not saying that, but just kind of trying to get you to focus on some of the bigger issues.
OK, and what are the bigger issues? Sticking with John Keeley, I’m impressed by a particular distinction he makes in his work with some colleagues. And he distinguishes between fuel-dominated fires and wind-dominated fires. And fuel-dominated fires are driven by the buildup of excessive fuel.
You’ve probably heard a lot about this, about how in the West we suppress fire so long and it built up fuel. And that led to our President saying that the problem in California was we were not raking the forests enough. But it turns out that the fires in LA were wind-driven, not fuel-dominated fires.
And that actually has implications for how you think about public policy, as I’m going to try to bring out a little bit. And the first big thing is, and you can see down here, is that a lot of the wind-driven fires, which, by the way, are mostly along the Coast of California, except for the camp Camp Fire– it’s a little further in. But they’re all being impacted or lit, or many of them are being lit by power line failures. So there, that’s an important public policy issue there to think about governance.
Now, if you think about power lines, they’re kind of a tricky issue. It’s an important issue to think about, but it’s a tricky issue. There’s speculation, by the way, that the Eton fire, but not the Palisades fire, was caused– speculation.
There’s still investigation. Speculation that it was caused by Southern Edison power lines. Is that right? Southern Edison? Yeah.
Now, one of the ways that you can address the power line issue, this is a governance question. You can put the power lines underground. Really expensive, effective, but really expensive.
Now, what Keeley says is that that’s a good public policy solution. Whether people will pay for it or not is another issue. He says we can be a little bit more selective.
We know, we’ve done modeling of extreme winds. We have a lot of data on that. We can tell, show you where the extreme winds are likely to come in the future, and we can selectively put things underground. But if you see pictures of where the Eaton fire was, up in the hills, it’s a pretty hard place to put things underground. So that’s another point to keep in mind.
OK, so now let me talk about a second issue. And Marta already brought this up. And this is the issue of what are called the wildland-urban interface.
And the wildland-urban interface or WUI– that interface is where you get basically high density of vegetation coming into contact with high density of settlement or buildings. Where those two things come together, it’s not a very good place for big wildfires. A lot of the big devastating wildfires we’ve had are in these or near these WUI zones.
And so what I wanted you to– first, I should say that people who do a lot of research on this have found that there’s different ways to measure WUI. And this one is using remote sensing. It says that it’s an improvement on several other measures.
But anyway, what I really wanted you to see was this, a map of California in the middle, B. And what I want you to see is that these are measuring the amount of area of WUI in a particular county. And you see, the worst place from this perspective in California is in San Diego. It’s not so good up in Sonoma, either.
OK, that dark brown area is San Diego down there. You go two counties up. That’s LA. That’s like the second worst county to be in.
OK, now think about this in governance terms. We have fire-dominated– fire-driven fires in areas that are dominated by wildland, by WUI. So this is a real challenge in LA.
OK, now this is leading me to my next point, which is going back to the blame game. But before the fires, the LA– there’s two fire departments in LA. One, the County Fire Department, and one, the City Fire Department.
Before the fire, about a month before, the Fire Chief had– the City Fire Department sent a message to the City Council and the Mayor saying, you cut our budget by this amount of money, and we’re not really prepared to deal with a lot. You’ve undercut our capacity. So this is part of the issue.
And in fact, one of the other issues was that the LA Fire Department didn’t really respond very quickly and very adequately to the fire. There were other issues that came up in the blame game. And you may have heard of them, the water and fire hydrants. The water basically ran out for the fire personnel, firefighters. And that made national news.
It turns out, just to say something about this, again, Connie, you need to put this in context of a larger perspective of the magnitude of the fire. There’s a guy, a water resources expert at UCLA, named Greg Pierce, and he basically said there is no fire, there is no water system in the world that would have been able to function perfectly under these conditions. And then he said, and this is also– I’m getting you to think about costs and what the trade-offs are.
You can put– you can make the water system really reliable for giant wildfires, but it’s going to be really costly. And the question is whether that’s the best use of the money. Is it better that money goes into maybe burying power lines, for instance?
So now I want to come back. There’s a bigger structural issue, and it was brought up in Marta’s talk about the polycentric and dispersed nature of Los Angeles. And that is, if you look at LA County and LA City, you see that the fire departments cover this incredible size geographical region between the two of them. It’s much bigger than New York City’s metropolitan region, by the way.
And one of the things they found, I wasn’t able to find reliable data on this, but one of the things they found is that the number of firefighters that you have in LA– they’re very thin on the ground. And it’s partly because they got to cover these big geographical distances. So this goes back to the finding that Marta found.
Now, when the fire chief said, “Mayor, you’ve cut our budget. You’re undercutting our capacity.” That was, I’m sure, true to some extent. But the Mayor said, well, we had some tough choices to make. And you can understand that.
They had to make some tough choices. We put this into education, or do we put it into the firefighting, et cetera, et cetera? Do we put it into affordable housing, which is a big issue in LA, or do we put it into firefighting? You can see that there’s tough choices.
Now, what I expect is after this, the fire department will get a budget increase, and the budget increase will improve their capacity at the margins. But I don’t think that LA is ever going to have fire capacity the same as New York or Chicago because it has to deal with this basic structural problem of a big dispersed polycentric area.
OK, now let me come to my next point. And this brings me, in terms of thinking about governance, to what some of the planning and regulatory approaches are for dealing with issues of wildfire. And let me just say, take one step back and say people like John Kelly who’ve been studying these wind-dominated fires– he says you’re really better than thinking about response because that’s how we think of LA.
The fire department is a response function. You should think about preparedness, prevention, preparedness, and resilience of community like Kenichi is working on. And that’s probably a better use of money. But what’s happening there when you actually look at that?
Well, one of the things that’s happening with planning is that there’s a couple of different ways that local– well, first of all, I need to tell you that a lot of planning is very localized for wildfire. It’s basically 88 cities in LA County. They’re all doing their own little planning for the issue.
And one of the things you learn from the research on this is that there’s a lot of variation in the quality of those plans. Some communities are really doing a good job, and some aren’t doing such a good job. And then, there are different mechanisms for planning. And there’s three big ones that I found. One is called community wildfire protection plans.
These were prompted by federal legislation, but they take place locally. And it turns out they’ve done research on these. And they find out that some do good work, but a lot of them are superficial. So they don’t do great stuff, unfortunately.
And other research has found that these hazard mitigation plans, which are sponsored by FEMA– that they actually do a better job than the wildfire plans. And the reason why they found is interesting to me is because they found that local planners have to check more boxes in order to actually get the plan approved, and that leads them to be more systematic. And then there’s also something called the general plan, which all counties do. They put together a general plan, but those– the best approach is through these hazard mitigation plans.
OK, another thing I want to tell you, which is interesting, I think, is that California and Los Angeles are pioneers. They’re really out ahead of a lot of other places in terms of wildfire planning. In some ways, they’re doing a really good job, although it’s limited in ways that I’m going to tell you in a second. But they’re out ahead of other states and many other cities.
So one of the ways that California has been ahead of things is that they have what are called– they’ve developed maps of high-risk areas, and they’ve connected local planning regulation to those maps. So basically, in these high-hazard zones, you don’t actually– you do actually have to build to a higher standard. And that’s something that California has been ahead in.
OK, now that brings me to regulation. And I found that there’s two big types of regulation at the local level that are designed to make communities more resilient. One is basically to get rid of the vegetation around your house.
That’s called producing a defensible– what is it called? A defensible space. Thank you. And the other is called home hardening, basically making your home more resistant to lighting on fire in the first place through the materials you would use on your roof.
Now, vegetation. For instance, I told my wife, OK, there’s a new rule that says we’re going to– there’s going to be a new state rule that says you have to clear stuff within 5 feet of your house. I told my wife that. She says– her reaction is, “Oh my god, that’s like half of our garden.”
[LAUGHTER]
And I think– and I just use my wife as an example of– I think a lot of people feel that way. And one of the things that they found is that there’s a lot of pushback on these local regulations. So there’s a political angle to this.
OK, in terms of home hardening, again, California has been a leader on that, has very strict building codes. But one of the limits of this is that this only works for building, new buildings. And old, wooden houses like mine in Berkeley– they’re like tinderboxes.
And we’re not really not really improving those. So there’s a long-term transition to really move towards greater resilience in this way. OK, I’m almost done here.
And so my conclusion about these regulatory measures is that they’re good, they’re important, but they don’t really meet the need for responding to fire on a big level. And I had some more slides, but I’m going to jump to my conclusion, which is the governance of large fires in Los Angeles and California is, to be very understated, a big challenge.
I think the good news from what I– from my reading is that cities like Los Angeles and the state of California are really leaders in addressing wildfire risks, contrary to what our President has been saying.
And also, I saw a lot of evidence in reading things that there are– that fires do lead to incremental improvements in wildfire safety. So you do see learning going on, although I would describe it as mostly incremental.
The bad news is that addressing these more fundamental challenges is really confronts some significant political and financial barriers, addressing the problem of utility lines, or building up the capacity for fire response, or accelerating home hardening. All these things are expensive, and they pose trade-offs across these different programs.
So if you look at local planning and regulation, it’s kind of mixed. It’s very variable by community. Some communities don’t have the capacity to do it very well. There’s limits in the willingness of citizens to go along with local regulations.
So I hope in the end, I’ve provided a little bit of perspective on the governance. I hope I didn’t go too far past. Yeah, thanks.
LOUISE COMFORT: Thank you.
[APPLAUSE]
Well, thank you, Chris, very much. We’ve had three different perspectives from different disciplinary of views. And now, I’d like to open it up for questions if anyone would like to ask a question. Anna.
AUDIENCE MEMBER: Thanks for all the presentations. I have a question regarding to the first presentation. How do you couple the information from the network’s analysis with the simulations that Kenichi does?
KENICHI SOGA: So when you say network analysis, our simulation is a network analysis.
AUDIENCE MEMBER: The network analysis is a network analysis.
KENICHI SOGA: Oh, I got it. So I asked Louise to create– we are agent-based model. So each agent decides when to evacuate. So that means that the agent’s decision is made from the model. Does that make sense?
AUDIENCE MEMBER: I see.
KENICHI SOGA: Yeah.
AUDIENCE MEMBER: OK.
KENICHI SOGA: So wildfire propagates seeing the– so there is an agent who will see a fire and say I got to evacuate. So there are certain proportion of that saying that some proportion will say, well, somebody said so I’m evacuating. So Louise is creating that model for that.
AUDIENCE MEMBER: To inform form the agent-based model?
KENICHI SOGA: Yeah, for the agent-based models. Yes, yes.
LOUISE COMFORT: We’re actually looking at the network of people and managers and the communication between the people and how that is communicated actually to the managers of, say, the traffic system, who sets the traffic lights, and which direction? When the roads are closed, how is that information communicated to the people?
So this is the sociotechnical aspect of a digital twin, recognizing that the roads are fixed, but people can change their minds and they can redirect and go in a different direction if the communication is there. So it’s communications and traffic. And it becomes a dialogue between the two.
AUDIENCE MEMBER: Thank you.
AUDIENCE MEMBER: Thank you all. Wonderful presentations. I have a question for Kenichi.
So your SNCC proposal was about Alameda and Marin. And so you’ve transferred that knowledge down to Southern California. So I’m just curious.
I have maybe two questions. In terms of modeling what’s going on in Southern California, were there variables that you had to consider that you weren’t expecting? So what came up as different– as the most prevalently different from your Bay Area modeling? And what is the spin-up time for transferring those models to other places?
KENICHI SOGA: So the first question is currently working on Alameda and Marin. And actually colleagues from El Cerrito, which is a Contra Costa, recently contacted, please include us. So that is a really because fire doesn’t know the boundaries. So that is a very big issue.
And, of course, we have to confine ourselves. So we started with city of Berkeley. We now have LBNL. We now have our campus colleagues, Office of Emergency Services colleagues, coming together and thinking about it. And, of course, we’re trying to expand that to Kensington and the area.
So I think that’s a challenge. But then I’m hoping we go or some of the startups or these colleagues will create a little bit more how to scale up. Yeah, and then our SimCenter also helps to scale that up.
Number 2 is what we see in LA. And here, I think that’s something that we really want to find out more because we find that every locations are different. We see in Palisades that they did do education pretty effectively, but they did say that issues were there. It’s one way out.
But then, typically, if you’re in one way out, people are a little bit more aware of the issues. So it works better. It’s really the Eaton fire was a little bit more sporadic.
Maybe they were not prepared for the wildfire because they’re quite spread. And so we do see that differences. But then I really want to find out why.
And that’s why it’s very important to find the details to get that because I think every fire is going to be different. But it’s a good question that– and it’s really about community understanding that, is that one person you may have a past experience of wildfire.
Sometimes, the issue is that they think that’s the case. It’s going to happen again, but that may or may not be the case. So realizing that is very important. Sorry, I’m maybe talking too much.
AUDIENCE MEMBER: It’s interesting.
AUDIENCE MEMBER: So I appreciate all of your presentations. And what I wanted to ask about is– so Berkeley has extreme fire weather and suggests people to pre-evacuate. Trying to get people to pre-evacuate– ooh, that’s hard.
OK, we did, but a lot of people didn’t. But what I’m wondering is, do you simulate in your simulations– because you were talking about a lot of wind-driven fires. And if we get the spot weather forecast, which Berkeley’s been doing, they say very low humidity predicted, high winds, Diablo winds around here, so please evacuate. Do you ever put that in your simulation in a way to maybe encourages more of that?
KENICHI SOGA: Yeah, that’s a good question, and we would like to. So we are hearing different scenarios that you can think of and trying to see what– would that have an effect?
Maybe your community may have a good effect, but then the other community may not have an effect. Just understanding that a little bit more is an important one to start. And what we want to do is that– maybe City of Berkeley has that particular evacuation notice that we received.
I live in North Berkeley as well, but then maybe the public don’t understand that. But having perhaps, and I’m not sure this is the case, but we start to see that people see the simulations, and they start to say, OK, this may be the case. But having that particular dialogue is important rather than telling that this simulation is true because it is not going to be true because every simulations are different. Yeah, but I think we want to try that. And working with the community is very important.
AUDIENCE MEMBER: Hi. I have a question for Professor Gonzalez especially, but if others want to share, too. I’m curious about the fire growth simulations, the optimized simulations that you showed us. I’m curious how feasible it is to be producing those and using those in real time for localities and local fires.
I assume that these might take a while to complete. And also then there’s the issue of– I do work in the rural counties in California, like Plumas, Butte, more rural, that are impacted by fire and don’t have the resources, also local computing resources and would rely on, I’m guessing, like Cal Fire having those resources to then distribute that information. So I’m curious just whether those can be used in real time and what the feasibility of that is.
MARTA GONZALEZ: Yes, that is what brought me to the topic. I said, if there is so much computation, we got to be doing better than this. And then I discovered that the accuracy, meaning the physics of the models, to make– just the physics of fire is a whole complex work in itself.
Then, I refer to Michael Goldner, Kenichi’s collaborator. That type of problem is the frontier in terms of the research. To make it more for operational purposes, real a decision-making, it would be these semi-empirical models.
And right now, with my students, we just make the search of what are the existing models. And the existing models are Farsite, Prometheus, you name it. Every country has its own. And it was built long ago.
So the idea was, OK, let’s do our own. It’s open-source, and let’s improve from that. Then, it brings us to one sobering realization that is high-resolution wind injection. Wind data. What we call data ingestion is limited. So I believe one of the main components is the good high resolution of wind, is one big limitation to make real-time meaningful because you can do a video game, a movie, but it is not accurate. Then ingesting wind in high resolution is one frontier that we would like to. And that brings CCRM, our center, with a new initiative in campus that is the Environmental Data Science Center.
LOUISE COMFORT: Exactly.
MARTA GONZALEZ: That would be the idea. So it’s a social good problem. And what I realize is not going to be done, let’s say, by Google.
The technology is there– we need to be aiming into that. And I believe in a public university like Berkeley it could be done. It’s not there yet.
AUDIENCE MEMBER: Great. Thank you. One quick follow-up. How long do those optimize– how long does it take for you to get those simulations? What is the actual computing time?
MARTA GONZALEZ: –just behind you?
AUDIENCE MEMBER: Are you referring to those simulations? Running those simulations don’t take that much time. And so they’re definitely able to be run in real time. I will say in a matter of minutes.
AUDIENCE MEMBER: OK, great.
AUDIENCE MEMBER: Thank you for great presentations. One topic that hasn’t come up here that’s quite in the public mind is insurance. And from, I guess, beyond the governance level.
So perhaps could one of you comment about the direction that discussion is going in, and who might know where the next catastrophic wildfire is going to occur in California in a probabilistic sense? Is it somebody like First Street or the risk– the analysts with their proprietary models, quite in contrast to the wonderful open-source models that you’re presenting here, which are very, shall we say, propagation models rather than general risk models that a real estate insurer might use?
So maybe start with the provocative part of the question. Who knows where the– in a probabilistic sense, where the next catastrophic wildfire will occur in California? And what do we do about it?
LOUISE COMFORT: I’ll respond to that because I’ve interviewed an awful lot of fire chiefs in California. The fire Chiefs know that there are certain areas that burn repeatedly. Malibu that burned in the Palisades fire burned in 1993 in three-four different times.
The fire chiefs know that it’s the combination of the geographic terrain, the fuel, and the winds that come. So they are making investments in mapping those areas. And this is where the select– Kenichi mentioned this, and also Marta, and Chris.
The selective, for instance, undergrounding of power lines might be a good strategic decision. But I will say that CAL FIRE has invested in the last five to seven years enormous amounts of money in modeling equipment and training their own personnel to do this. The difficulty is that there is a direct, almost one-to-one correlation by the increase in emissions and the increase in the size and ferocity of the fire.
The critical issue that we can change, and this is social action, is literally reducing the number of emissions that go into the air. And framing that as an issue, Chris, is a public policy issue. And so this is why addressing this problem of increasingly exponential wildfire is a interdisciplinary inter-jurisdictional issue.
And we have to do it smart. We have to do it recognizing that we’re dealing with a very complex set of interconnected systems. And this is why I think the University of California, with campuses across the state, is in an excellent position to do this. But it’s not going to be easy, and it’s not going to be fast, and it’s not going to be soon.
KENICHI SOGA: Chris, do you want to talk about insurance issue? That’s a big issue.
CHRIS ANSELL: Well, one thing I’ll say about insurance because I have been following a little bit is that the FAIR Plan– the FAIR Plans are the last-resort plan. I lost my insurance this year, by the way, in my house.
The FAIR Plans basically were driven bankrupt by this. Or they can’t fulfill the claims because of the LA fires. And what the state did was it allowed the FAIR Plans to basically– I guess it’s– what do they call it? Not meta insurance but reinsurance.
Basically, the FAIR Plans could take $1 billion from other insurers in the state, who are set up in the state. And I think you can maybe see some– one of the implications of that is we’re all going to be paying for the LA fires. Yeah.
AUDIENCE MEMBER: [INAUDIBLE] produce this?
CHRIS ANSELL: Yeah. I don’t know that, but it’s a good question. Yeah.
LOUISE COMFORT: I will say one of the members of our team, Steven Collier in Urban and Regional Planning, has focused on insurance. And his basic quick assessment is private insurance is almost going to be gone in California. And that’s a really difficult thing. So looking at alternative plans and the FAIR Plan, publicly supported is one of those.
KENICHI SOGA: Yeah, that’s the next workshop, I think, which is a very big one.
AUDIENCE MEMBER: –much for bringing this to us. I have a question. So Louise, you mentioned almost in passing, but as a take, something very obvious, we need to reduce emissions. And this is the elephant in the room as there’s something that the models are maybe taking for granted, which is– I’m curious what you all do with the fact of the driving causes of those winds that mean that all bets are off, or of climate change, these changing conditions.
And so I’m not– I’m assuming that this is very much on all of our minds. There’s one question one could ask of like, well, the governance issue is how do we put pressure on Exxon and fossil fuel companies. And that was part of the blame game.
Actually, some community– some affected folks are trying to sue some of the fossil fuel companies and name them as the responsible parties. So there’s a governance question there with the blame game. That’s probably the one blame game to be played.
But I’m just curious, and from where you all sit and the expertise you bring to the table, where does that conversation– what input is that into the simulations? Are you taking for granted that things are just going to get worse and the winds are going to get worse? Or how do you factor this very messy political world into this story? And I ask that, having just read that Trump just agreed with–
MARTA GONZALEZ: Yes, actually, I like to quote Dan Kammen from ERG in this topic. And it is that a housing policy is a climate policy. And what the data is telling us, why the areas burn is getting worse is the drought. So we have drought, and then is– fuel is worse and that’s why it’s getting worse.
However, interestingly, we cannot continue building at risk. That’s why I like to put our cities in the international context. The model of land use development that we have in the US and particularly in California, that this low dense that we all love is not sustainable.
And then, right now, even CARB, California Air Resources Board, is funding a call for proposals that brings housing policy with vehicles miles traveled. So we need to reduce vehicle miles travel, and that means sustainable transportation, but even how far we need to travel. And California is aware of that marriage.
And then, housing policy is a climate policy. And we are now being affected by the fires. So it’s all intertwined.
KENICHI SOGA: I’m not trying to promote our work, but then I think we can really go to the details to model what you see in a climate change model right now. So I’m hoping– Stephen Collier is a great example. He doesn’t believe on models that we do, but then he creates different ideas and said, wow, that’s interesting. Maybe we can model that and see how that goes.
[LAUGHTER]
So it allows us to do that right now, I think. And so I think Stephen starts to see maybe what we do maybe link to what he’s thinking. And that’s what we see in this project. But probably before the project, maybe he did believe at all what we’re doing.
So I think that’s where we can have an interesting discussion on what can be integrated if we can because there are lots of interesting ideas that come from you that we may want to think about. And Louise has been very promoting, yes, I can do it in the communication style. So we need a model to put it in that. Of course, the model may not be correct, but then at least we try.
LOUISE COMFORT: Thank you. I’m looking at the clock. It is 1:30. Others may have other appointments.
If there’s any last question, you might ask any one of us. But I really want to thank all of you for coming. And I’ll ask one big favor.
Keep this in your mind. Start thinking about it. We need an innovative approach to deal with these increasingly catastrophic risks. And the one thing that we can change is how we think and act about risk. So let’s– please join me in thanking–
[APPLAUSE]
[MUSIC PLAYING]
[WOMAN’S VOICE] Thank you for listening. To learn more about Social Science Matrix, please visit matrix.berkeley.edu.
