Science in the Spotlight: Forecasting wildfire behavior

Welcome everyone uh to this NSF and CAR Explorer Series event. My name is Evan Portier and today we'll be talking about wildfires and how the US National Science Foundation National Center for Atmospheric Research or NSF ENCAR is working to provide solutions to this difficult environmental challenge. Just a couple of housekeeping notes. This is our second conversation on wildfires this month and you can find a link to the previous conversation in our slido chat. To access the slider chat, you will just scroll down and enter your name and click join. You'll also be greeted with several questions that if we mind answering, that really would help us out. And you can also find more information about the previous event and our future events on our website. Just search for NSF Encar Explorer series and you'll see our future events about all sorts of research. So today we'll be diving into wildfires, specifically forecasting wildfire behavior, how wildfires react and spread under different conditions. And to do that I am joined uh by NSF ENCAR's Jason Conneal. Jason, welcome. >> Thank you Evan. It's good to see you. I appreciate you arranging this conversation. Likewise. And so for the folks in the audience, Jason, I was wondering if we could just start off this conversation just telling a little bit more about yourself and broadly the work that you do here at NSFNCAR. >> I've been at the Foothills Lab at NSFNCAR since 2001 in a variety of different capacities. Most of my background, my education, and my experience is what we would call sort of traditional atmospheric science. studying weather in complex terrain. By that I mean mountains and valleys and deserts and coastlines, you know, weather that really feels what's on the earth's surface. Uh also over over cities. So that's that's more or less my background. But since 2018, I've been working on projects related to wildfire and uh it's been a really exciting time to be in this line of work. >> Great. And so for folks that don't know, the Foothills Laboratory is one of our several campuses here in the Boulder area and that the research applications laboratory is one of several labs um under NSF ENCAR. So like we mentioned earlier, Jason, we've been hearing a lot about wildfires um often in the news, but just conversationally. And so I'm curious to start off this conversation to hear your perspective on how these wildfires are becoming a growing problem and really a nationwide challenge. >> Yeah, it sure feels like something has changed. Uh I I moved to Colorado in the 1990s and I certainly do not recall wildfires being talked about so much um and and so much in the news. So I I I agree with you. it feels like something has changed. So, it's an interesting thing. Uh, one of the things that has not changed, at least in the nation as a whole, is the frequency of wildfires. If if we look back over the last half century or so, the last 50 years, yeartoyear, the number of fires nationwide goes up and down a little bit, as a lot of things do, but there's there's really no overall trend in the number of wildfires nationwide. There are some trends in regions of the United States, like in the western US where we are, we do see a little bit of an increase in in fires, but the overall number hasn't changed much. Now, that number is a shockingly high number. In in 2025, there were more than 77,000 wildfires in the United States. And when you hear a number that large, you might think, how is it then that we're not living in a constant state of like smoke and ash? That's an extraordinary number of fires. The answer is that most of those fires are quite small. They're really inconsequential. Inconsequential. It It's the way the distribution of fires work. It's the few large fires that make the news are the most troubling to us. And in fact, if you look over the western United States in most regions, um about 1% only 1% of the fires accounts for more than 80% of the total area burned. So that's that's why given the high number of fires, we it's not even worse than it is. So given that I just said that the number of fires nationwide isn't showing a strong trend, why does it seem like it is a more urgent problem now than it used to be? It's because the fires that we do have are burning hotter. They're moving faster and they're they're burning more area. And so the severity has gone up even though the number of fires nationwide has not particularly gone up. >> So, so just to recap, it's not that we're just seeing we're not seeing more fires just on a nationwide, but what we're seeing are these really larger fires that are burning hotter like you were mentioning. Are these kind of the mega fires that you were mentioning that kind of dominate the news like headlines? >> They are. They're they're the fires that um they dominate the headlines because these are some of the most destructive fires. They're they're the fires that are hardest to control. They're the fires that most threaten people and our property. And I I think an important context to everything we're talking about today is is the recognition that there are good fires. Wildfire is not fundamentally a problem to be solved and suppression of all fires is not the answer. A lot of landscapes um have fire as one of their components. There there are trees that are adapted to fire and actually require fire in order to reproduce. So fire itself isn't really the problem. However, there are bad fires and these are the fires generally that um can uh cause us all sorts of problems, life, property, the cost of fighting them. And so it's this is an important way to frame it. And a lot of I think what we're going to talk about today is the bad fires and and the need to forecast them better and mitigate them as a hazard. But I think it's important to keep in mind that that fire is just fundamental to nature. It's been around on the planet longer than humans have been around on the planet and we're not going to suppress them, nor should we try to suppress all of them. And so that's a really good point, Jason, and thanks for distinguishing that fire is kind of this natural process on these ecosystems, but there is a subsection of fires that we would deem problematic. And I think one of the reasons is this wildland urban interface. And I'm just wondering if you could touch upon why that is problematic in in terms of wildfires. >> Yes, I can. The the the WUI or the WOOI, the wildland urban interface. Uh I I can actually show you a picture of of what we mean by that and it might help other people uh get a sense of it. Here is just one example. Um, this is from Colorado, just up the hills from where we are right now. And it's it's the aftermath of um the Buffalo fire in 2018 in Silverthorn. So, if you look at this picture, what do you see on the left? You see what we sometimes call the built environment. This is our stuff. It's what we have built. It's where we live. And in the center, you see what we would call the natural environment. trees, grassland, sage brush, the mountains in the distance, and the Buffalo fire burned through this wildland urban interface where we and our stuff sort of meet the natural environment happens to be some of the the prettiest places to live. And in fact, the number of of people in the United States and in Colorado living in the wildland urban interface or the WOOI is uh is high and it's going up nationwide. about a third of the population either lives in or works in the WOOI. And in Colorado, it's even higher. It's about half. The reason that the WOOI is such an important topic if we're going to discuss wildfire is because not only is it in the WOOI that often um fires are the greatest hazard that we're most exposed to fire, but as we continue to build in the WOOI, there is the greater chance that we're going to start fires. In fact, although it varies quite a bit from state to state, overall most fires that happen are started by people. Not necessarily intentionally. Far more often than not, it's accidentally. But as we are in these areas that are fireprone more and more, not only are we exposed more, but there's more likelihood that we're going to cause fire, too. So, I think this is a really good context here. So, these fires are burning hotter. They're burning more area and but we're also um inventing ourselves in the this WOOI this wildland urban interface. And so, we know that this is a challenge nationwide. And so, I'm wondering if we can talk a little bit about the tools that are available to us to kind of help and predict wildfires. And so, you know, conversationally, we might hear about wildfire forecast, we might hear about wildfire models, and I'm wondering if you can kind of tease apart those two and expand upon what that really means. >> Yeah, I'd be happy to. Um, most generally, uh, predicting wildfire is somewhat like predicting the weather. You can do that with many different kinds of tools. One of the most powerful is um what we call a numerical model. This is just a set of computer code that contains equations that describe how the atmosphere or fires behave. You take that code and that'll give you a future prediction if you also tell the model what the current state is. Sort of a trivial but I think meaningful example of this is if I said to you you've got a number and I want you to add five to that number. What is the result? And you would come back to me and you'd say I I can't tell you the result until you tell me what was the original number. And if I say well one okay 1 plus 5 is six. So the point is that these equations that predict either the state of the atmosphere in the future or the state of a fire in the future, they have to have a starting point. You have to give the models the current conditions. So that's how a lot of these numerical models work. In the case of um the kind of modeling that we tend to do, not exclusively, but we tend to do at ENAR, we see the advantage of coupling fire models to atmospheric models. The reason for that is in nature, the atmosphere and fires communicate all the time. For example, fires behave in certain ways because of atmospheric conditions. Imagine drier than normal conditions, hotter than normal conditions, and then you've got wind on top of that. These are atmospheric variables, but they're going to define how a fire behaves. The fire then in turn communicates back to the atmosphere. Fires release heat, they release water vapor, they release gases and particles into the atmosphere. And this kind of exchange atmosphere to fire, fire to atmosphere is happening. nature all the time. And so we feel that there's a physical realism that you get from having what we call coupled modeling system. That coupling is those two models, the atmospheric model and the fire model behaving kind of in tandem. It's not the only approach. Models are tools. And if you imagine other kinds of tools, you know, for example, tools to clear snow off the driveway or the front porch in winter, there's no single tool that's best for all jobs. Sometimes you need a 15 horsepower three-stage snowblower. Sometimes a simple broom will do the job. The same is true of atmospheric models and fire models. The complexity that we gain that's an advantage when we're talking about coupled modeling also comes with disadvantages. So there's greater fe physical realism but that complexity means that even the most powerful computers take a long time to run these models. And so if you're in a position of needing a very very fast answer to what is this fire going to do in 3 hours or 6 hours and I can't wait 15 minutes to know or an hour to know. I need to know right now. There are alternative models that are simplified. They don't capture the full complexity of nature but they give you an answer a lot faster. And so an important thing to keep in mind anytime we're talking about these coupled modeling systems is it's one solution, but it's not the only solution to solving the problem of predicting what a fire is going to be doing, you know, in the next few hours, the next few days. So Jason, when you say a couple model, it sounds like you were saying that these two systems that you are trying to represent, so the atmosphere and the fire are kind of talking to each other. And is is that a fair thing to say when we say coupling, does that mean like there's feedback involved, right, between the two? >> There is. Feedback is the perfect word, Evan. Yeah. I think you've probably heard, I know I've heard uh in discussions of some of these mega fires that you referred to earlier that a fire is so big that it makes its own weather. And that's true. The big fires indeed do make their own weather. Uh but that's not exclusive to the big fires. Fires in general make their own weather. But the degree to which the weather's perturbed and how far away and how long lasting the perturbation is sort of scales with the size of a fire. Imagine a really small fire, something even as small as say a campfire. In my view, a campfire makes its own weather, too. You sit near it, you feel the heat. Uh water vapor is being released. You wouldn't necessarily notice that. That's invisible, but you see the smoke. And these are all very very local small changes, but nevertheless they're changes of the atmosphere right around the fire. And I would say, you know, in my view, they kind of qualify as weather changes. The bigger the fire, the the the bigger the change. And some of these mega fires, it's just extraordinary uh what they what they can do to the atmosphere. I've got a really dramatic example of one here that I can share with you. This is what's called a pyrou cumulo nimbus. It's a huge uh thundercloud. This one happens to be over the Sierra uh national forest in California. Remember that I mentioned that one of the the feedbacks, great word that you use. One of the feedbacks from the fire to the atmosphere is introducing heat and water vapor and particles. Well, it so happens that heat and water vapor and particles are three key ingredients to making clouds. So, what we'll often see above the most sustained fires, the largest fires, is clouds forming. They're not all necessarily this visually impressive. Some are just smaller uh cumuli. Nevertheless, this is one example of how fires feed back to the atmosphere. And it's one of the kind of things that we strive to capture when we're doing this coupled modeling. It so happens that pyroumulonus clouds are are difficult to simulate and there's still a lot of great research to be done in that area. Um but it is that kind of feedback that that we we like to capture with a couple modeling. >> Great. So it sounds like one of the tools that we can use to understand how wildfires spread and react with environment are these couple models, but it's not a like a one-size fit all. It really is dependent on I guess kind of those stakeholders or what folks are needing like you were mentioning if they need an answer quickly. Perhaps the couple model is not the the best tool available here. And so with that, we we've talked a little bit about the tools available here and we know that I think it's kind of just one of these things people say are like, "How can I trust the model or how can I trust this forecast?" And so I'm wondering if we can just dive into, you know, wildfires are complex and so there are challenges to to modeling this, especially like you mentioned, fires make their own weather and then the weather influences the fire. So can we dive into what are the current challenges of making or challenges of getting like accurate forecast? >> Yeah. Yeah. Absolutely. Um and and there's so many challenges we we could be here a long time but I'll I'll just hit on a few. In in coupled modeling the more accurate the atmospheric piece of that coupling is the better. So first and fundamentally you really want to get the weather forecast right. weather forecasts have been improving uh over the decades on a fairly regular basis and are much better now than they were a few decades ago. They're still imperfect. So um that is one challenge, you know, getting the atmosphere right. And in particular when you're talking about the state of the atmosphere right near fires when in fact they have been influenced by the fire uh getting those observations is really really important because recall that what I said is you can have an accurate model but if you don't give it a good starting point if you don't tell that model what the atmosphere's current state is through observations then it's going to have errors in it. It's very difficult to observe the atmosphere right in the vicinity of fires. Um it's it's a hostile environment and they at least wildfires tend to happen where we don't necessarily have a lot of instruments deployed. So that's one challenge. Another challenge on the fire side is that we need to know a lot of details about what is burning. What the we call this fuel. Fuel is really anything on the ground that either could be burning or or is burning. Trees, shrubs, grass in the natural landscape or unfortunately in the built landscape. This can be buildings, it could be cars, it could really be anything that the flames move through. We need to have an accurate characterization of that. and especially at at high detail because even large fires are relatively small compared to the size of a state or the size of the country. And so um understanding the the details of how a fire behaves really means we need to understand the details of the state of the fuel through which the fire moves. The state um really means several different things. What is it? How much of it is there? How wet or dry is it? And we could go into other details, but those are kind of the the the fundamental ones. What is it? How much of it is there? And how wet or dry is it? So that that's a challenge. Um >> wait, Jason, can I just interrupt you? >> Yeah, please do. >> Going back to our our first one where you were saying, you know, really one of the challenges with with accurate forecasting is the need for observations. And can you expand like what do you mean by that to to someone who's not familiar like what what exactly do we need? >> We need observations of the air temperature, the humidity in the air and the wind. And more would be even better. But I say at a minimum those are kind of the three main variables that we need to know because fire behavior will be fairly sensitive to those three atmospheric variables. In fact, it turns out turns out go going back to what we discussed before about the mega fires, the most dangerous fires, they tend to be fires where there is a high wind. These are really wind driven fires. We know this in Colorado. Um, but it's not exclusive to Colorado. California has similar problems. The the fires uh in Lahina uh in Hawaii th those were wind driven fires as well. And and this is true the world over really. If if you look at some of the most destructive fires, they tend to be the winddriven fires. And so getting accurate observations of the wind in the immediate vicinity of a fire is really important. Those are kind of the the minimum variables that we'd like to see on the atmospheric side. Does I hope that answers your question? >> It does. Yes. >> Okay. Yeah. And one one further thing I want to say about the fuels piece, it's characterizing fuels in the natural environment. So again, think trees, shrubs, grasses. While not easy, at least we've been doing that longer and it's a bit more straightforward than characterizing the fuels in the built environment. One of the huge opportunities for for greater basic science and applied science is understanding what happens when a fire makes its way into the wildland urban interface and starts burning those fuels, things we don't even want to think about as being fuels, houses for example, we have a lot of trouble representing that accurately in our models. And if we cannot accurately, if we don't understand that problem, then we can't write the equations. We can't write the computer code to put that into our models. And so our models, they kind of don't really know what to do once a fire makes its way into the built environment. We we we can make rough appro approximations. It's not like we are truly clueless about it, but if we're examining what are the areas that really would benefit from more research, understanding fires in the wildland urban interface and even more almost the purely built environment is, I would say, yet another challenge is if we're going down the the list of challenges. Uh, another one is early detection. Often when we first notice that there's a fire, it has been going for a while and that means it's a bit harder to respond to and certainly harder to suppress if it is one of those what we would call bad fires. This is a term of art in the business, bad and good fire. And uh our response to bad fires depends on really getting an early warning. There are ways of doing this. um satellite technology, cameras, but there's a long way to go. What we have found in some of our wildfire modeling with these coupled models is that if we introduce a little bit of an error in the where and when that a fire starts, you know, we take an ignition point and we say, let's actually pretend that it was 1 kilometer away to the north or to the south or let's pretend it was actually 6 hours earlier or 6 hours later than what we believe it to be. What does that do to our fire simulation or what would what would have been the fire behavior forecast at the time? Sometimes is sometimes it results in in a dramatic change in the fire. An extreme case imagine this. You move the ignition point away from dry brush to the middle of a small alpine lake. Suddenly you don't even get a fire anymore. So that's I mean that's an extreme example just to drive home the point that even pretty small errors in the what or sorry the where and the when of a fire starting can throw off our fire behavior predictions by quite a bit. So that's yet another challenge is that early detection and getting the location and the timing down really well. And then I'll close with one I would say is the is the la the last challenge I'll bring up. Certainly not the cha last challenge we faced. The atmosphere and fires are chaotic systems. That doesn't mean they're random. They still follow the laws of physics chemistry. By chaotic, what I mean is they're so complex and our ability to observe them is so limited that small errors in how we code them up in the models and how we observe them eventually translate to large errors in the prediction. We have ways of dealing with this. One way is is through probabilistic approaches. So instead of taking a single fire behavior simulation, giving you a single prediction of what that fire is going to be doing three, six hours from now, one day from now, you actually launch a whole series of simulations, each of which is slightly different from the others, but different in a plausible way. You don't want to have completely unrealistic simulations because they do you no good. But you can have a range. Sometimes we call these an ensemble, a collection of equally probable um outcomes and then you let all those model runs go and they give you a collection of results. It's almost like crowd sourcing in a way when you're sourcing a crowd of reasonable experts. Each of these model simulations is like an expert but they all differ slightly. If you take that collection of results, then you can begin to ask questions like, what is the probability that this fire is going to burn to the east or to the west or is going to be stopped by that stream? You can talk probabilities and you get a result that's like there's a 73% chance or there's a 12% chance. Then you can make really good objective decisions about how to employ resources, where to tell people to evacuate or maybe not evacuate. Probabilistic approaches to modeling is a really good way of dealing with these chaotic systems. And so doing that better because it's very computationally intensive is is the last challenge that I'll that I'll mention. >> Great Jason. So, we kind of we were walking through this where you've um talked about the tools that are available, these couple models that Encar has been developing. You've given us a lay of the land of some of the current challenges here. Not all of them, not an exhaustive list. Um so that leads us to a natural question. What are you and your colleagues doing in in this space? Um and specifically, let's jump into this um topic of embers because I find this really fascinating. And so can you expand upon your work um with embers and what are embers? >> Yeah, absolutely. I'd love to do that. Uh let's let's show you first an example of the kind of simulation that some of my colleagues are uh are working on right now. Let's get this going. There you go. This is the East Troublesome Fire from 2020. It it was notorious for several reasons. One of which is that in one 24-h hour period, it grew by something like, I don't know, 100,000 acres. Most fires never reach 100,000, let alone grow by 100,000 in a 24-h hour period. It was heavily wind- driven, which is something we've talked about already. And embers figured prominently in this, as did the importance of getting the fuels piece right. And let me let me show you what I mean about getting the fuels piece right. And that will lead into answering your question about some of our latest advancements here. What I've got on the screen right now is uh the observed perimeter. Say you're you're viewing this from up above. So you're looking straight down. That white poly line that you see in those two panels, and it's the same line on both panels, that represents the perimeter of the burned area at some point late in the fire's lifetime. So that outlines the area that was burned. What you see in yellow is what our fire behavior model predicted would be burned. The two yellows look different. Why are they different? Well, on the left, the smaller predicted burn area, ignored the fact that this fire burned through timber that had a large amount of uh a large number of trees that were killed by pine bark beetles. So, these are standing trees, in some cases down trees that are drier. They're not alive. Um they're drier than than the living trees. They tend to burn more readily. If our model doesn't know that, doesn't know that there was a lot of beetle kill, it it tends to underpredict how far the fire advances. On the right is what happens if we go into that data set of the fuels, what is being burned and we inform the model that a lot of uh these trees have been killed by beetles, we get a much more accurate prediction of the fire. So, one of the things that that my colleagues are doing is we're working on ways of improving the realism of the fuels data sets that go into the model. Now, I mentioned that um that this fire, these troublesome fire was notorious for several reasons. One is that it advanced through uh embers that were produced. And you ask what embers are. So, I've got a uh an illustration here from the Forest Service. Really nice diagram showing what we mean by embers. When you have uh fuels being burned, they will give off bits of material, glowing hot bits of material that are still burning, that are carried by the wind. They tend to rise up because they're warmer than their environment and then they're blown downwind. I'll bring up campfires. Again, you see this happen with with uh you know, they almost look like sparks, but they may last a little longer than sparks. Some of them are actually like little chips of wood that especially wet wood can spit out. You hear this pop and then you see a little ember flying and you hope it doesn't land on your tent. These these are embers. So in extreme cases really hot fires can produce a lot of embers and they can be lofted pretty far and carried downwind. Being carried down was important for several reasons. One is it can cross areas that otherwise would uh get in the way of the advance of a fire, a stream, a lake, a road, a parking lot, an expanse of um the ground that it itself will not burn. If you've got embers, it can hop right over those areas. Another reason embers are important is they can go way, way downwind from the main fire. There has been research in Australia, for example, showing that in some cases, embers can travel 10 to 20 miles away from the main fire, land on some burnal material, and start a new what we call a spot fire. That is such as an important it's such an important mechanism. And it turns out that in the built environment, the main way that fire moves from house to house to house is through embers that land on rooftops, land on uh landscaping that is flammable, fences. Embers can be sucked into the vents in the home and ignite the home from the inside. Embers are the main way that fire moves from house to house. So if our goal is to have realistic simulations of fire behavior in the built environment, we really need to account for this. I have a colleague Maria Frediani who is doing exactly that. She has figured out a way to model these embers by considering them to be more or less particles in the air. They're big particles. They are hot particles. Nevertheless, you can use an approach that is sort of tried and true and you can just model them um as particles and for a lot of fires in the build environment, that piece is key if you're going to get realistic results. I've got a simulation here I can show you of the infamous Marshall fire that affected so many of us here at NSFAR. This is the fire burning and that yellow line through sort of the upper middle part of the figure. That's highway 36. This simulation again accounts for ember transport and spot fires. That fire in the simulation is able to leap across the highway which itself is not burnable start additional fires. You'll also see another characteristic of this fire is that it expands and surges. It tends to move not methodically at the same pace but in these huge leaps forward. Some of that is ember transport. Some of that is just basically um this was a heavily wind- driven fire. You might remember at the time winds exceeded 100 miles an hour in places. So like a lot of the worst fires and this was the costliest in Colorado history. this fire was uh was heavily wind driven. So those are two areas that we're we're working on right now is better characterization of the fuels and and also um more accurate representation of the role that that embers play in producing spot fires. Uh and if you want I can go into a little bit more detail about some of the fuels as well. >> Yeah. Uh, and I think it's it's kind of important to mention it sounds like when you were showing us the map, um, the outline with the bark beetle kill, that one part of your work that you're doing is that you're looking kind of after the fact, right? And then you're kind of simulating these fires and you're kind of, it sounds like you're kind of testing or inputting different variables to see how you can get that realistic behavior that you actually observe in the fire. Is that is that correct to say? >> Yes, you you've nailed it exactly. It's it's fundamental to how we improve our models. You you have a model that does some things well. It does other things not very well and completely ignores yet more things. And you try that on data sets that you that you know and you see how well the model performs and you you look at what its strengths are, what its deficiencies are, and then you make modifications and then you try it again. And it sounds like I'm just sort of describing a trial and error thing, but it's a lot more directed. It's a lot more um based on fundamental and applied science than maybe what it sounds when I describe it that way. But there's this process of optimization that has to occur not just with numerical models really with with theoretical equations with with a lot of the basic tools by which science advances. Rarely do we come up with the perfect 100% correct answer right away. It's usually a step forward and you reassess. You take a few more steps forward, you reassess. And in the case of fires, especially the most severe damaging fires, even though it seems like during summer, especially hot, dry summers like this one, they're all around us. Each fire is so different from the others. And there are still relatively few of them that when new fires come along, they pose new challenges. We see things we've not necessarily seen before. Maybe we suspected it or maybe we knew knew they took place, but we didn't have the observations and lo and behold, there was a new fire this year or there will be a new fire next year that is observed better than any other fire and we use that then to further advance our our model. So this is an ongoing process and um it brings up one really more important almost philosophical point about weather modeling, fire behavior modeling. These models will always be imperfect just by nature of the the the chaotic systems we're dealing with and the fact that observations are never um comprehensive. They're never everywhere all the time to put it simply. However, just because these models are imperfect doesn't mean they're not very very useful. And in fact, they're getting better all the time and will give you very very good answers that you can base important decisions on, even if they're not quite perfect answers, and that's always going to be the case. But we we make progress by reassessing and and applying these models and these other techniques to known data sets. >> Great. That's a that's a really good point. It's just this iterative process that you know it sounds like sometimes a fire happens and you you need to really have it as like almost a case study to kind of refine your models here. And so I know you mentioned something about fuels and some of the work that you and your colleagues are doing and I think you mentioned this as a challenge. Uh so I think this would be a perfect time to dive in of yeah what uh work you're doing around these wildfire fuels. Yeah, I I mentioned that the um it's important to know what the fuels are, how much there is, usually that's measured in mass, for example, and then how how dry or wet the fuel is. Wetter fuels do not burn nearly as easily. So, fires that burn through wet fuels tend to not burn as quickly. They're not as hot. uh they produce very different kinds of smoke because the combustion when wet fuels burn is not as complete as the combustion when dry fuels burn. So understanding specifically the moisture level in these fuels is is really important and I have a figure that I can show you that's that's really exciting. It's it's um result of work being led by Pedro Himenez. And uh this is a figure, nice colorful figure. It shows what we call dead fuel moisture content. That's really how wet is all the dead stuff on the landscape. pine needles, twigs, small branches, logs, all the stuff that's down and on the ground that used to be living is no longer living. We refer to these as dead fuels. And they have a certain moisture content that is um based partly on how much it has rained, what the relative humidity is in the atmosphere, what the soil moisture is, you know. So these these fuels can get wetter and drier as the weather changes. We also talk about live fuel moisture and that is just the fuel moisture content of the vegetation that's still living, the trees that are still standing, the grass that's still green. Getting these observations right is fundamental to getting fire behavior modeling right. But it's a very very difficult thing to do because you have to take a lot of observations over a large area and update those observations frequently. So what we have done the work that Pedro has led is we use observations taken on the ground of the dry and wet and mo and sorry and live um the dead and live fuel moisture content. We combined those data with satellite data both both low earth orbiting and geostationary satellite data and then several other data sets that come from models. We use machine learning and what we can produce is a map of both the dead and live fuel moisture content. Again, this that you're seeing on the screen is is the dead fuel moisture content that's updated multiple times a day and covers all 50 states. I'm just showing the 48 contiguous states, but we have Alaska and Hawaii as well. This kind of data set fills in huge blanks in the areas where we really have no observations and we have to use what we call like interpolation. Think of it as a form of spatial averaging to guess at what exists between two stations that are widely separated where you do have obser obs observations and you don't necessarily in the middle. Not only does this data set then when fed into our models improve the results, but even without fire behavior modeling, just access to this kind of data set gives people a greater awareness of what the overall fire risk is. So that that's another area in which we're working. It's it's more on the land side than the atmosphere side, but is is vital for improving our modeling. >> Great J. Yeah, just really highlighting the utility of that. Even for someone who's not in that field, I think you can really get a sense of the different the fuel moisture that exists in your area. So, we've talked about some of the advancements that we're making um NSFAR is making here. Um I want to kind of wrap up some of this conversation with collaboration. We know that this is these are chaotic systems, wildfires, weather and so there are a lot of stakeholders, a lot of endusers involved here and I'm just wanting you to speak because of the laboratory. The research applications laboratory has a lot of collaborations. If you can talk about what you're finding most important, what collaborations you're finding most important in this field. I I'm glad you uh you brought up collaborations going back to how complex a problem wildfire is and and how many sectors and industries uh are are affected. It it's both a challenge and an opportunity. And the opportunity comes with the collaborations that you're referring to because it's it's if we focus on on how to lower the threat of wildfire in the future and we get people from different sectors talking to each other, working together who don't normally work together. This is hugely beneficial. So I'll go through a list like my other lists. This this one is not exhaustive. Uh first and foremost, chemists are so important. The single biggest cost from wildfires cost to society is because of smoke exposure. We we know from um fires that we've seen this year, last year, and years before. Smoke can go thousands and thousands of miles away from a fire and infect people who aren't the slightest bit worried about the flames themselves, but nevertheless are exposed to the smoke. Chemists allow us to understand the particles, the gases in the smoke, um how they change with time, what are the health effects. And so there's another opportunity for collaboration, people in the um the healthcare industry. Uh we work with um people in that sector to understand uh the kind of the outcomes um the threat to health that different kinds of exposure um has people then in other sectors that are affected by wildfires. think the transportation industry, you know, wildfires, especially in mountain communities where aren't that many roads in and out of a place, um, can have an effect. And then in the aftermath of fires, for years and years afterward, the landscape has changed. Glenwood Springs is a good example. Glenwood Canyon, where where Interstate 70 goes, you know, they've had their rounds of fires followed by rounds of debris slides, which interrupts transportation. The transportation industry is important. The utility industry especially uh electrical grid utilities, they're concerned about fires, not only, you know, unintentionally starting fires, but then also fires that that may be encroaching on their infrastructure. foresters, ecologists, soil scientists, really hot fires, they change the soils in profound ways such that long after the the fire is over, uh you you don't end up getting the same kind of nutrients, porocity um in in the soils that that have been burned. Water quality is an issue. So we have a group of hydraologists that we work with here at ENCAR and elsewhere as well to understand the close relationship between water and and fire and there are many close relationships there other sectors insurance um national defense you know you can see it's a long list and I'll stop there but all these areas they offer these opportunities for collaboration and doing science in in kind of new ways interdisciplinary or sometimes what we call convergence science science, which in some ways is the most exciting science that we're working on. >> Great. Uh just a reminder for any folks that want to ask Jason a question to put that into Slido. Uh Jason, last question for you here before we jump into our Q&A. Um if you can may wave your magic wand here and over like the next decade, like what advancements do you really want to see in this space and what role do you think NSF Encar has to play? I I would say uh first and foremost probably better observations of the atmosphere and the landscape in the vicinity of of fires. This has come up before so we're kind of repeating ourselves here but um that that's really really important. Where will those observations come from? Well, I think we're going to have better and better satellites up there. Um, more sophisticated sensors, higher resolution sensors, and those are a really good way of observing especially remote areas of the planet. Drones or uh more generally what we call uh uncrrewed um aerial vehicles, they they will offer opportunities. If you could imagine having teams deployed to fires where their sole function is to fly drones around or take observations of temperature, humidity, wind, even what's in the smoke, feed that back to uh wildfire prediction models. That would be hugely helpful. So I imagine we're going to be making a lot of of progress in that area. couple models have become more sophisticated because we'll know more about fire behavior especially in the built environment and so the equations in these models that make the predictions will be better. I think artificial intelligence really has a role to play here. AI models are showing just remarkable um utility in certain areas uh when they're relatively young. We've not been working on artificial intelligence nearly as long as we've been working on other models. So this is exciting, but I think it will be quite some time before um they can do everything that physical models do. So in the in the near future, I think what we're going to see is the kind of numerical models based on physics that we've been talking about combined with artificial intelligence and that duo that in a way it's a different sort of coupling that is probably going to yield some of the most accurate fire prediction um that we have workflows will improved. I I would love to see us get to a point where when you need a forecast of what a fire is going to do, you're just a few clicks away on your computer or your phone the way you are right now. If you need a forecast of what the weather is going to do, we've we've come to know that weather models are sort of they're operating in the background at the enterprise level nationwide fundamental improvements to to the accuracy of weather forecasting. I'd like to see us get to a place where we sort of consider fire behavior modeling kind of to be in the same vein and with a few clicks on my phone I I can see where a fire is likely to move in 3 6 12 hours. I I think we can get there. It's sort of a workflow problem. It's sort of a plumbing what we sometimes casually call a plumbing problem. We've got the pieces. Um and then I'd like to see other kinds of coupled modeling. Imagine that not only do you have really good predictions of fire behavior, but we feed that information into evacuation models and so emergency responders can tell people what roads to take outside of a threatened community. And um and that's done with input from the atmospheric modeling, the fire modeling, the modeling of of how people behave in these situations, the traffic flow. The more of these models that we can couple together and the more that we can capture the uncertainty that gets passed from one model to the next, the more we can make really objective decisions that that that help people um mitigate the the threat of fires. >> Great. Well, thank you Jason for that. Um on that note, we're going to jump into our Q&A. We definitely have some some questions from the audience. But before that, I want to make a quick plug just about our upcoming events here. Um on your screen, you'll see our four upcoming events from August 26. Uh that's going to be understanding the impacts of space weather on satellite orbits. That's local for folks that are to Boulder County. That's going to be at Frraasier. And then fall in the this Saturday, we'll have for models forecast tools for better water operations at the Leland Museum. And if you are a virtual audience and can't attend any in person or just like the comfort of your home, we do have a hybrid event at the Mesa Laboratory Cloud Makers: How Alaska Seasons Paint the Sky. And then lastly, upcoming on the 19th, we'll be talking about generally NSF Encar's science stories and breakthroughs over a 60-year period at the NOO library in Boulder. So if folks want to know more about these events, feel free to take a screenshot or again search for NSF ENCAR Explorer series. You can also sign up for our newsletter. And lastly, I want to make a plug. At the end of this conversation, you'll receive an email for an evaluation survey. Please, if you can fill that out, it really helps us understand the outcomes of events like this, but also you'll be participating in educational research project. So, you can also scan the QR code that is on the screen as well. And so, with that, Jason, we're going to dive into some of our Q&A questions here. Uh, Russ has a interesting question here. Along the Colorado front range, Excel has installed many utility pole mounted remote weather stations. Do the real-time weather readings from those stations make their way into some of these highresolution wildfire forecasts? Yeah, that's an excellent question. In some cases, yes, in some cases, no. We have had really good experiences with um talking with with uh electric grid operators, not just in Colorado, nationwide. In fact, we we've been regularly attending some of the meetings that that they hold to discuss for the entire industry um h how to deal with fires. they uh utilities have been some of the the most ambitious and aggressive about putting all kinds of weather stations out in remote places to take observations. It's it's hugely valuable and and so we would I I think you know to get more at this question in the future we would love to have the answer be universally yes that all those observations are are making it into uh models. At this point it is it depends. The answer is it depends. It's it's sort of hit or miss. But what I think what we can say is that even if some of the models that we're using here at NSFAR do not have access to those data, those data are used by the utilities to make um really well-informed decisions about how to manage their systems. And so going back to the point that wildfire forecasting takes many forms only one of which is using coupled models in a way those observations are informing the predictions that are made by the wildfire experts and the utilities whether or not they're making it into into our models. But in the future we'd like to see all of those data make it in. >> Speaking of which, we have a question from Luke about couple models here. How much more accurate are the coupled models compared to the up uncoupled models? >> It depends on the fire. In the case of uh heavily wind- driven fires where the atmospheric environment isn't changing much. So imagine the extreme case is you have a violent westerly wind that just blows steadily during the entire time of a fire. you really the the the feedback from the fire to the atmosphere is it's not zero but it's relatively small compared to the role of that wind. So in that example you would be able to get a pretty accurate prediction just by imposing an unchanging steady wind on the fire and letting it go and never allowing a fire feedback to the atmosphere. So that's one kind. At the other end of the extreme would be uh a case where almost all the fire behavior depends on the local perturbations to the atmosphere because fires can make their own wind. They make their updrafts horizontal wind and so if you don't have a largecale strong wind that that serves as the environment of a fire instead you have pretty much calm conditions imagine and then and then you have a fire erupt. In that case, the feedback to the atmosphere is going to be fairly important because those are the main changes that the atmosphere feels. So the the the summary answer to the question is what you lose or gain by not coupling or coupling depends a lot on the specific fire you're talking about. Sometimes it doesn't make much of a difference. Wonderful. We have time for one more question and I apologize. We had a lot of questions here. Uh if you don't get your questions answered, feel free to email the Explorer Series uh team and we can get your questions relayed to Jason. To kind of wrap us up here, Michelle has a question and a comment here. I really appreciate you mentioning the benefits of fire. Do you believe that the increase in large fast spreading fires has a connection with the lack of fire mitigation practices from the 80s to the early 2000s? Yes, that there's there's a lot of evidence um for that. As a matter of fact, I I mentioned the importance of of fuels in in understanding fire behavior and the and the specifically the aspect of how much fuel is there and that's exactly what your question is driving at. there is a lot of fuel to be burned because really dating back to the early part of the 20th century 1910 and on. Uh we've been pretty aggressive about suppressing fires and you know that was done based on reasonable understanding at the time uh among a lot of the decision makers and it seemed like a a wellfounded policy. Um, interestingly enough, Native Americans knew long before that the importance of having frequent fires in the landscape. They they did a lot of their own landscape management by having prescribed fires. And so they they often were able to to keep the amount of fuel load from building up to the point that when there is a fire, it is one of these blazing hot fastmoving fires. We've we've learned a lot since then. So if you go back to the early part of the 20th century, really up into probably the 1960s or 70s, this really aggressive suppression policy uh has this unintended consequ con consequence that the fuel buildup is massive. And now we're we're kind of grappling with that. It's the it's it's one of the reasons that prescribed fire is important and letting some fires burn, the good fires burn, because counterintuitively by letting those fires happen, we reduce the risk that future fires are going to be these hot, really threatening conflrations. >> Great. Thank you so much, Jason. And thank you everyone for joining us today. We really appreciate it. Again, if your questions didn't get answered, feel free to email the Explorer Series team and we'll get your questions answered. But with that, we'll see you at the next event and appreciate your time today. >> Thanks, Evan.