Climate change has brought forth extreme fire events, like the Palisade and Eaton fires in Los Angeles, which devastated communities in Altadena and the Pacific Palisades in the beginning of 2025. And it’s becoming harder to not wonder: Is this just the world we live in now? Under the constant threat of catastrophic fires? Fortunately, we have the perfect guests to answer that question. We traveled to NASA’s Goddard Space Flight Center and spoke with two of their scientists who study fires from space.
Transcript of this Episode
Doug Morton: My journey that led me to study wildfires started at the end of a dirt road in the middle of the Brazilian Amazon.
I remember standing in the middle of a clearing that was tens of thousands of acres and seeing smoke columns rising in every direction, sleeping at night with a T-shirt wrapped around my face so I could prevent some of the impacts of the smoke that would settle over the town where we were staying in the middle of the night and waking up with my eyes burning and that acrid taste in your mouth because you just couldn't get away from the smoke.
Sam Jones: That’s Doug Morton, a fire scientist at NASA. It’s been more than 20 years since he got his start studying how deforestation affects fires in the Amazon. Today, more and more people can probably relate to Doug’s vivid experience with fires. Climate change has brought forth extreme fire events, like the Palisade and Eaton fires in Los Angeles, which devastated communities in Altadena and the Pacific Palisades in the beginning of 2025.
Deboki Chakravarti: I moved to Claremont, California, when I was twelve, and there were wildfires really close to where I lived my first two years there. In 2003, there was this horrible wildfire called the Grand Prix fire that got right to the edge of my high school. And one of my main memories was at nighttime, it looked like someone was drawing the outline of mountains with the fire, and it was so striking but also so scary. And for the rest of my time in high school, all of those hills around the campus stayed almost completely barren. So even though I don’t live in California anymore, every time there’s a wildfire, I can’t help but wonder: is this just the world we live in now? Under the constant threat of catastrophic fires?
Sam: It’s a fair question and, fortunately, I found the perfect guests to answer it. I traveled to NASA’s Goddard Space Flight Center in Maryland and spoke with two of their scientists who study fires from space.
Welcome to Tiny Matters, a science podcast about the little things that have a big impact on our society, past and present. I’m Sam Jones and I’m joined by my co-host Deboki Chakravarti. Today on the show, you’ll hear from NASA scientists who are using satellites and machine learning to monitor fires around the world in real time and better predict fires in the future.
When I sat down with Doug and his colleague Mark Moussa, who you’ll hear from a bit later in the episode, it was only about a month after the LA fires, so they were still very fresh in our minds.
Doug Morton: It's devastating you know more than 200 of our colleagues at NASA's Jet Propulsion Laboratory lost their homes in the Eaton fire in particular. And the impacts on the communities, both those directly affected, those communities will never be the same. And then the folks who were supporting them. Also, the folks who were breathing that smoke are forced to think about how to handle the toxic cleanup that will result from those fires. It's heartbreaking.
Deboki: The LA fires was the second deadliest wildfire in California’s history. It claimed 30 lives, destroyed more than 15,000 structures, and displaced thousands of families. Doug says the conditions in LA leading up to the fire, like prolonged drought and dry, strong winds called Santa Ana winds, made the likelihood of a catastrophic fire, sadly, predictable.
Doug Morton: We're starting to see sets of fire weather conditions that present an insurmountable challenge. When you have conditions that are hot, dry, and windy with winds at times approaching a hundred miles an hour, there's very little we can do to stop those firestorms.
Sam: In Los Angeles, it took thousands of firefighters 24 days to fully contain both fires.
Deboki: And while the fires were burning, NASA scientists used satellites to monitor the situation and relay critical updates to officials on the ground.
Doug Morton: Today we have a system which is called the ultra real time system. The satellite is broadcasting data in real time as soon as it's collected, received on the ground, processed into data products and distributed through one of NASA's most well-known systems, the Fire Information Resource Management System or FIRMS.
Sam: That ultra real time data can be transmitted to the ground in less than 60 seconds. This is critical information, especially in fires where high winds make it almost impossible to do any monitoring by planes.
Doug Morton: If you get into a situation room within a fire response, whether that's US Forest Service or FEMA, they're taking data from that FIRMS page. The applications that have been built to provide local alerts to communities, many of which were featured in the response to the LA wildfires, are taking data from the FIRMS site. It really is probably one of the best examples of using satellite data for a public good that I can think of. We've been delivering active fire information globally for free for 25 years.
Deboki: About 25 years ago, NASA launched the first two satellites that would eventually provide this public service. The first one, called Terra, launched in 1999 and measures multiple things to study the climate, for instance the aerosol particles and carbon monoxide in Earth’s atmosphere as well as vegetation, snow, and ice cover. A few years later, the satellite Aqua launched to collect information on the water cycle. Both satellites were equipped with instruments that track fires.
Sam: Back in the day, we used to rely on rangers sitting in watch towers to look for whiffs of smoke. So these satellites were a huge step forward when it came to monitoring fires.
Deboki: Totally. So now that scientists had this bird’s eye view of fires around the world, Doug says they could begin to measure how fires have changed over time.
Doug Morton: And it might surprise you and maybe your listeners to know that actually on an area basis, the amount of burned area has actually been decreasing over the last 25 years by more than 1% per year.
Sam with Doug and Mark: Interesting.
Doug Morton: So looking down from space today, we see 25% less fire on the planet than we did in 2000, primarily in grasslands and savannas. And so the crazy part about that, if you think about how ecosystems like grasslands have evolved to be adapted to fire, is that those fire adapted ecosystems are burning less.
Sam: That may sound really counterintuitive, at least it did to me at first. But Doug says there’s a few reasons for this decline in fires in grasslands and savannas, in places like the Serengeti in Tanzania or the Eurasian Steppe. These landscapes are being fragmented for agriculture, and people are using the land to raise livestock that graze the grass. So that means less big open space full of grass to burn. Finally, as more people build homes and structures, they’re less likely to turn to fire to manage the land.
Doug Morton: At the same time, forests and shrublands are burning in record ways that they haven't in the past. And so even if they don't burn as much area, they're burning hotter, faster and longer in places that are become drier and that have fuel to burn.
Deboki: Research shows that climate change is behind these dry conditions, leading to twice as many large fires between 1984 and 2015. And when it comes to starting fires, it turns out humans are playing an even more direct role, especially when you remember that fire is a common tool for agricultural management and land clearing, especially in the tropics.
Doug Morton: People start 99% of the fires we see from space.
Sam: That is absolutely wild to me. But we also can’t discount natural causes like lightning strikes which may become more frequent due to climate change.
Doug Morton: Lightning is an important source of ignitions, especially in remote regions. And lightning ignitions can be particularly troublesome if your goal is to suppress those fires. Because when we have a storm come through, you could have tens of thousands of lightning flashes in the matter of an hour. And so you can start lots of fires at the same time.
Deboki: Understanding how fires start is key to predicting where they may go. Scientists at both public and private institutes rely on fire data from the first few hours, along with other measurements, to model how that fire will spread.
Sam: And those are really helpful predictions. But ideally you’d want to take it a step further and be able to predict these fires before they ever get started. Doug likens it to being able to look up next week’s weather.
Doug Morton: And so can we apply that same kind of framework to thinking about fire weather and can we anticipate with the same degree of confidence those fire weather conditions three, four days in advance that would allow for a different set of responses, preparedness, decision making? And of course, we see this on the ground in California. You see the utilities trying to make decisions about rolling blackouts. You see parks and other protected areas having to make tough decisions about when to exclude people from backcountry areas or other kinds of permitted activities.
Deboki: Some areas of the world have fairly predictable patterns, like in the tropics, where Doug has done a lot of work. These predictable patterns could allow people to prepare for fires even two to three months ahead.
Doug Morton: It doesn't tell you exactly where the next fire is going to start, but it would allow you to pre-position resources and imagine allocating more effort towards fighting fires in those areas on a year to year basis. That’s different from that three to four day forecast I'm talking about. And that's different from the real time monitoring that also is where Mark and I are engaging our efforts right now. But those are all pieces of, I think, that same puzzle.
Sam: He’s referring to Mark Moussa, an AI and machine learning engineer at NASA. Doug and Mark’s goal is to launch more satellite instruments to observe fires, and to make those instruments smaller and less expensive.
Doug Morton: So Mark and I are funded to build a couple of new instruments called the Compact Fire Imager, one of which we’ll test on an airplane starting this spring that'll participate in the NASA Fire Sense campaigns that we'll be using in partnership with the Forest Service, other state agencies to better demonstrate the capability of the new technology, and also to learn from our partners about what exact type of information would be most helpful for them.
Do they need to know more about the position of the fire line, the temperature, the intensity of the burning, the duration of the burning in any one location? Those are the kinds of questions that we want to be able to answer in an airplane because we can take many overpasses over the same wildfire, just a few minutes apart, and we're building a space version, which is about the same size, about the size of a Kleenex box. And that puts us in a position to be ready to take the next ride that's available, whether that's a small CubeSat type satellite bus, or whether that gives us a spot on the International Space Station. We'll be ready in terms of trying to make some new observations from space using an updated set of technology that's smaller, more powerful, and more consistent so we can keep taking science quality data to identify fires around the world.
Deboki: A key feature of this new technology will be using machine learning to process the data. And that’s where Mark comes in.
Mark Moussa: So before, a long time ago, I think scientists and engineers, we largely had the problem that we didn't have enough data about the Earth or we didn't have enough high resolution and different types of data. So the scientists and engineers here at NASA and around the world, we did a really great job of collecting and building tools and satellites and towers to collect so much data, but now we have too much.
And so it takes scientists a long time sometimes on the order of years to sift through this data, make sense of it, not to mention all of these beautiful tools, some of which Doug mentioned, they give you different types of data. And so you have a lot in the kind of search space that scientists need to go through. So machine learning I think really helps find patterns on such a large search space that it would take either humans a long time to, or maybe even come up with novel stuff that humans wouldn't have seen in the first place.
Sam: It seems like we hear the terms machine learning and AI thrown around all the time. But what exactly are they?
Mark Moussa: I think even people in my space are confused about exactly what to call it. That's why I say AI and machine learning, because there's even people attribute different definitions to those same terms.
The best way I can describe it is being able to, through what is essentially statistics and calculus, train a computer to learn to predict patterns based on the data. So a good analogy would be: think about when you are in class and you want to study for a test, you have a textbook filled with data that you want to learn, and you have a study guide. And in that study guide you'll have answers, you'll have questions, and you'll have the answers to those questions. So when you want to study for an exam, what do you, you go over the textbook, you comb over it a number of times just to make sure that you can memorize the information, and then you use the study guide and you try and answer the questions. If you get the answer wrong, you look at the answer, you figure out why you got it wrong, and then you go over it again. So largely that's what machine learning actually does.
Sam: You can apply that to something like detecting wildfires. So machine learning models are fed tons of data and then asked, “is this a fire or is this not a fire?” Then it checks the results and goes through the process a bunch more times, getting better and better at identifying fires correctly.
Deboki: That sounds great. But the technology also poses major challenges.
Mark Moussa: One of the problems I think that Doug and I are working on is machine learning models are generally very computationally heavy, so they require a lot of computational resources, electricity, power, that kind of thing. Getting that and running it on a satellite to get as near time as possible or on an airborne mission, like Doug mentioned, the size of a Kleenex box, is tough. So to get the model to be able to run with a huge constraint on your resources is an obstacle all on its own. One of the number one obstacles in wildfire specifically I think for machine learning is the ability to reduce false positives and false negatives especially. So what you don't want is this machine learning model to pass over a fire, process the data and then say there's no fire, because that's just not good.
Deboki: Another challenge is that satellites can’t be everywhere you want them to be at all times of the day.
Doug Morton: The reality is today we have satellites that are traveling at 17,500 miles an hour and get a look at the whole earth every day, but they only overpass the same location every 12 hours. So I only get to look at fires twice a day from each of our satellites that are giving us global information.
And those happen to be at times a day that aren't particularly helpful for studying fires because we didn't launch those satellites specifically to study fires. It's one of the many things they can do. So we have satellites that overpass in the middle of the morning because that's when we tend to have lower cloud cover, and we have satellites that overpass in the early part of the afternoon because that's when you have higher cloud cover. The higher cloud cover is good for weather forecasting. The lower cloud cover is good for looking at the land, and fires tend to be most common and most intense in the late afternoon when we have no lower earth orbiting satellites coverage.
Mark Moussa: So I think Doug touched on something really important here and that I think machine learning can help with, and that's clouds. So when you are looking through a satellite down at Earth, sometimes clouds can get in the way, and if you're looking at clouds, that's great, but if you're trying to look at the actual ground for whatever reason, wildfire or otherwise…
Sam with Mark and Doug: It's frustrating.
Mark Moussa: It's frustrating because they kind of occlude what you want to see. And I think machine learning is really well poised to be able to take clouds and sort of mask them out in a lot of ways. I don't necessarily think the technology's there yet, but I know that researchers are currently working on it. And I've seen some preliminary good results in the research space, so it's something exciting.
Doug Morton: One scientist’s signal is another scientist's noise.
Mark Moussa: Agreed.
Deboki: Plenty of challenges remain to make these new technologies a reality. But Doug and Mark say it’s an exciting time to be working on these problems.
Mark Moussa: I think the space of machine learning is progressing very rapidly.
I can't speak to the operational and the logistics side of things. I think as far as the actual technology and the capability of doing some of this stuff goes, I think it's quite possible soon. And I think humans are generally bad at judging exponential curves, and I think, this is my personal opinion, but I think technology-wise, we're kind of on an exponential curve. It seems that often when I think of something that's very sci-fi related in the world, and I'm like, oh, that'll happen so long from now it, especially these days, happens sooner than I think.
Sam: One of the big things that Mark is excited about is called data prioritization.
Mark Moussa: So when you're downlinking from satellites, whether it's on earth or somewhere out in space, the amount of data that you collect sometimes is really a lot to kind of send down when the actual downlinking is kind of constrained. If you have machine learning models on board, then they can do that for you. And so instead of sending gigabytes of data down all at once, you're sending the most important stuff first right away and prioritizing that so that scientists can then see it.
Doug Morton: The collaboration with Mark is really exciting because speed and accuracy are two critical elements of working in fire. The ability to deliver trusted data has always been at the kind of heart of the work that we do here. And having the scientific rigor to have both sort of our physics understanding and our machine learning capabilities coming together, I think is really powerful.
Deboki: Doug says the reality is that fires already impact billions of people around the world and that number is only growing.
Doug Morton: Every year from space, we detect and track more than a million large wildfires.
Sam with Doug and Mark: Oh wow.
Doug Morton: We've talked about the LA wildfires. They were not remarkable in many respects in terms of the size of the fire, the duration of the fire. They were incredibly devastating for the communities and they will be for many years to come.
I think that we're continuing to try to figure out how to continue to identify and anticipate risk, detect, and respond to new wildfires and to do this all in a more flammable context. So keeping good fires in the landscape and preventing fires in landscapes that we would like to protect is a challenge that's going to get harder, but one where we're committed to do our part.
Mark Moussa: You don't have to be a scientist to start doing science. So NASA, like Doug mentioned, does a lot of open sourcing of our data. So if you're interested in this kind of thing, data is everywhere online. If you're super young, if you're super old, you can get your hands on it and kind of start just seeing it and looking at it and seeing, how can I help? I think citizen science is a beautiful thing, and so if listeners are interested, I just kind of want to encourage them in general that you can do a lot with the data.
I think my main goal in life is to just help my fellow human as much as I possibly can. And so I think wildfires are a great way to do that.
Deboki: That’s a lovely note to end on. We’ve left a link to the website for FIRMS — the Fire Information for Resource Management System — in the episode description and the references list at the end of this week’s transcript. On the FIRMS site you can access fire maps, download data, set up email alerts and look at satellite images with NASA Worldview.
Sam: Shall we Tiny show and tell?
Deboki: Let's do it.
Sam: Well, mine's fun. Is yours fun?
Deboki: Yeah, mine's fun too. No bummers today.
Sam: Okay. Well, I was going to say we should end with a fun one, but I guess since we both have a fun one, I'll just go. So Deboki, I'm going to tell you about a sea lion with rhythm.
Deboki: Oh.
Sam: This sea lion's name is Ronan. And Ronan was born in the wild in 2008, but kept getting stranded due to malnutrition.
Deboki: Aw.
Sam: I know. And so after three strandings, she was actually seen walking down Highway 1 near Santa Cruz.
Deboki: Oh my …
Sam: And it was pretty clear she was not going to be releasable. And so UC Santa Cruz adopted her in 2010 and she became a permanent resident of what they call the Pinniped Lab. So I don't know if you remember this, this was a long time ago. It was 2013. Researchers found that Ronan could bob her head to the beat of music and she could also adjust to tempo she hadn't heard before, which I vaguely remember this, but this is now 12 years ago.
Deboki: Yeah, that sounds familiar.
Sam: Yes. So obviously a lot of time has passed. Ronan's been part of a bunch of different projects, looking at memory and diving and all this other stuff, but really not a ton of stuff related to rhythm. And the thing that's cool that I was reading about with Ronan is they're like, yeah, she kind of does stuff when she wants to and when she doesn't, she doesn't have to.
Deboki: Amazing.
Sam: She gets fish if she does these activities, but sometimes she's like, I'm over it. And they're like, all right, see you tomorrow, Ronan. So it sounds like Ronan lives a pretty sweet life. Anyhow, she hadn't been doing a lot of stuff related to rhythm, but after this 2013 study, apparently there were some bio-musicality theorists who were talking a little trash. This is me. This is not the language that was used when I read that.
Deboki: You're editorializing.
Sam: I am definitely editorializing right now. And they were saying that Ronan's performance wasn't as precise as a human's.
Deboki: Oh no.
Sam: Yeah, I know, like calm down. So, I'm not saying calm down to you. I'm saying it to these theorists. These researchers were like, well, let's test it. And so what they did was they got 10 UC Santa Cruz undergrads matched up against Ronan, and instead of bobbing their heads, because that's not really like a big head bob... Ronan is not a typical human movement when listening to music. What they did was they would move their preferred, or dominant arm up and down in a fluid up and down motion to the beat of a percussive metronome.
And so there were three different tempos played 112, 120, 128 beats per minute. Ronan had never been exposed to 112, or 128 beats per minute. And what they found was that Ronan was really on the beat, which was not shocking. At 120, which she had some experience in, she would hit within 15 milliseconds of the beat, which is very, very good. Apparently, blinking your eye is 150 milliseconds. So very close to the exact beat. Yeah. And so this was just as good, if not better than the humans. Take that bio-musicality theorist. That's really it. I mean, Ronan just has incredible rhythm and is so precise, and even with beats per minute that she's never heard before. So I thought that was just so cool. And Ronan is so cute. I'm actually going to share with you the press release that has a little video. I mean, you can just see a picture of Ronan. She's adorable.
Deboki: Let's see this… I feel like they could have had the humans doing this head bobbing.
Sam: I know. Hey, you know, maybe that's the next study.
Deboki: Yeah. Well, I also have some animal videos to share with you today.
Sam: Oo.
Deboki: Yeah. I will queue up this link for you. So this is all from a New York Times article that I was reading about how bats are able to drink water while they fly, which scientists were really interested in because it's multitasking. And multitasking is pretty important for animals to do. The more you can do it once, especially essential things for a bat like flying and drinking water, the more likely you can do the things you need to do to survive. And I didn't know this, but there are two main ways that bats can drink water. So some species will fly over the water and then also lap at it with their tongue, which I feel like is kind of what I would've expected. But also that sounds really hard.
Sam: I'm looking at this video, also this loop of a bat doing that and it's really cute.
Deboki: Exactly. Yeah. It's amazing. And you're kind of like, oh, that is kind of hard. So the other way that I really love that they drink water, which is not the focus of this article, but I just really love it. It's a technique called belly dipping, and basically they just get their fur wet and then they lick it and they're like, nom nom nom, there's my water.
Sam: That's sweet. I like that.
Deboki: It's a little bit funny. And also maybe a little bit gross in my head, I'm like, on the one hand, this is cute, but on the other hand, how do you know that your fur is not holding other things?
Sam: You don't. But I don't know. I feel like how animals lick their fur and-
Deboki: True. Good point. So in this research, they just wanted to get a better sense of how bats are able to do that method where they're lapping the water as they're flying. So they used all these cameras, they caught all these different angles of it, and so there were a few things that they found. I will say that on its own, I don't think any of these results are super, super surprising, but I do think it's really cool that they caught it. So for example, they saw that the bats will reduce their flight speed as they're getting close to the water.
The one that I found really interesting is that they do also really control their tongue super precisely to get the water into their mouths. And I just thought that was interesting because I assume we do that too, and also just don't think about it, right? It is shocking to me to think of bats really precisely controlling their tongues to get water. But I was also like, why am I shocked? We also have to do things like that, right?
Sam: Right.
Deboki: But yeah, anyways, I just really, really love this video. That's mostly why I'm sharing this because it’s just really cute.
Sam: It's worth sharing for just the video alone for sure. And of course we'll link to it.
Deboki: Yeah. Today is our animal video edition of Tiny Matters.
Sam: I know, it's so funny, same wavelength.
Deboki: I like how we were like, we're not doing any bummers, and instead we went full like...
Sam: I know. It's a good switch. We’ve got to switch it up sometimes, for sure.
Deboki: Thanks for tuning in to this week’s episode of Tiny Matters, a podcast brought to you by the American Chemical Society and produced by Multitude. This week’s script was written by Tien Nguyen and edited by me and Michael David, and by Sam, who is also our executive producer. It was fact-checked by Michelle Boucher. Audio editing was done by Jeremy Barr. The Tiny Matters theme and episode sound design was by Michael Simonelli and the Charts & Leisure team.
Sam: Thanks so much to Doug Morton and Mark Moussa for joining us. We’ll see ya next time.
References:
- L.A. Fires Death Toll Rises to 30 After Remains Are Found
- The dangerous combination fueling the L.A. fires: Exceptional dryness and strong winds
- LA fires aftermath: How people are rebuilding after losing almost everything
- FIRMS
- Terra
- Aqua Earth-Observing Satellite Mission
- Wildfire climate connection
- Lightning fires threaten planet-cooling forests
- NASA Worldview
- Rhythmically trained sea lion returns for an encore—and performs as well as humans
- How Bats Enjoy an In-Flight Beverage Service
;void(0);){kind=link}