Eyes on Earth Episode 15 – Burn Severity Mapping

STEVE YOUNG

Hello everyone, welcome to another episode of Eyes on Earth. Our podcast discusses our ever-changing planet, and the people here at EROS and across the globe who use remote sensing to monitor and study the health of planet Earth. I'm your host Steve Young. Today's guests are Birgit Peterson, a Geographer and Senior Scientist at EROS, and her colleague at the center, Software Engineer Gail Schmidt. We're going to talk about their work mapping burn severity after fires in the Black Hills, their use of unmanned aerial systems to assist in that work, and how that work may benefit both western South Dakota and the forested areas around the country. Welcome Birgit!

BIRGIT PETERSON

Hi Steve.

STEVE

And on the telephone with us, welcome Gail!

GAIL SCHMIDT

Hi Steve.

STEVE

Let's start by talking about mapping burn severity. Why do we do that?

BIRGIT

One of the reasons we do is because how we react to that burn severity on the ground matters. Landowners want to know if their area has been burned severely, so if they need to do mitigation work, such as reseeding in an area to kind of keep erosion under control, that's something that they want to know from the forefront. If it's a moderate or low severity, that can actually work in our favor, because that can work as essentially a fuel treatment. But they need where those different types of fires occurred, because when you have a big fire perimeter, and that's kind of what we often see on maps that are out there, not everything burns the same within that fire perimeter. In fact you might have islands that didn't burn at all, so how you manage lands after that fire occurred, that type of burn severity information is really important.

STEVE

Does everyone in areas where fires have occurred understand that fires can be good, that it burns fuel loads or do you have to fight that perception that, you know, we need to put every fire out that comes along?

GAIL

Ultimately, fire is still perceived as a bad sort of thing in the forested areas, however we do try and preach "good fire." Fire truly is, if it's on the ground and handled correctly, it's good fire and that's what we do with our prescribed burning is we try and burn out the low surface fuels, and maintain those low surface fuels so that when the hot months come, when your hot July, August, September, dry and hot months show up, the surface fuels are low and the fire does not burn near as hot, and ultimately, Mother Nature has a tendency especially out here in the Black Hills we usually see fire, it's healthy to see fire in the area every 10 to 20 years to maintain those low surface fuels. So, if we don't have fire or if we have firefighters go and extinguish that fire quickly, then we don't allow Mother Nature's job of fire keeping those surface fuels low to do its job correctly.

STEVE

Well then how do we use remote sensing to assist us in mapping burn areas? How does that work?

BIRGIT

Traditionally we've used what's called spectral remote sensing, so if people are familiar with the Landsat suite of data, we've used that heavily to look at the spectral response of a landscape after a fire. So, healthy vegetation looks a certain way on these images, and then once it's been burned, it definitely has a different reflection. It looks different. It's basically a picture taken from above that allows us to do before and after comparisons. And that's been a key component and a key way of making these assessments over the spatial areas that we need to cover. I mean people certainly go in and they do ground measurements to validate a lot of this, but that takes a lot of effort and a lot of money to do. If you could do it through remote sensing, that is certainly a cost savings. And the other advantage that it gets you is you can be consistent in your measurement or your observation over time and over space, so when you want to start doing comparisons--okay, how are things tracking over time, how is one area comparing to another?--having that consistent, somewhat unbiased look can be very beneficial when you're looking at these trends.

GAIL

And Steve, ultimately if we have a decent map of the previous fires throughout a particular area from the firefighting perspective, if we have another fire and it starts to grow into that area, we can use that previous fire area to help us control ultimately what's happening with this fire and we know that if there was a previous fire there, we're going to have lower fuel loads and ultimately that could be a fuel break for us as we're fighting that fire.

STEVE

I know we want to talk here about the work you're doing in the Black Hills, you're potentially moving toward a three-dimensional burn severity mapping technology. Before we discuss that, tell me why the Black Hills? Why are you doing this work in the Black Hills?

BIRGIT

Part of it is serendipity. We just happen to have a fire that occurred relatively close to EROS. We're located in the southeast portion of the state of South Dakota, so we're surrounded pretty much by plains and farmland, so for those of us who are interested in wildfire and forested systems, we have to travel quite a distance to do so, so to have one in the Black Hills that's of significant size--it was over 50,000 acres, which is a big fire for that area--was fortunate for us in terms of access, being able to get there with our equipment and being able to drive.

GAIL

And from my perspective, having that opportunity to work with the private b and federal on the U.S. Forest Service side, the state on the state of South Dakota side, and the National Park entities and their different management practices and being able to look at areas within each of those management practices, that's important but also then the follow-up and being able to see how the fire area recovers and being also able to talk to those local land managers and better understand from their perspective what they're seeing, and allow them to utilize the information that Birgit and the team area collecting, is exciting to see what they pick up from it from their perspective as to how that might help them in the long run with their management practices.

STEVE

Would it be fair to say, Gail, that because you work fighting fires out there that your relationship with all those different entities maybe makes this project go off a whole lot better than it might? 

GAIL

You know, initially I was able to obtain some contacts to allow Birgit and the rest of the team to get in and do those studies, and Birgit and the team have also maintained those contacts, and we'll just continue to grow from there. But yeah, being a local volunteer Fire Chief and a firefighter out here has allowed us to at least know those areas and those contact individuals so we can get on their land.

STEVE

So again you're working towards a potential 3D burn severity mapping methodology. Tell us about that. What would that entail?

BIRGIT

Well as we talked about before, the data that we've typically used for remote sensing and looking at burn severity is the spectral data. So we're looking at really a two-dimensional image, much like a photograph from above. But there's changes that occur on the landscape that can be added to that information that we think might help us define burn severity a little bit better or refine the legend as it currently exists. So what we're really looking at is that third dimension. Where is material removed in the canopy and what portions of the canopy is it removed from? Is it only in the lower portion of the canopy? Ground fires, surface fires, fires that don't go into the main part of the canopy, those are typically good. If you want to category fires in terms of good and evil, that's kind of what clears out that fuel layer that then can keep another fire from occuring in that area that would be more devastating, and you can't get that from two-dimensional imagery so what we use is LiDAR (Light Detection and Ranging) that in different scales, at different scopes, that will sort of get at that picture of how structure has changed from before and after the fire. We've done it by going in and establishing plots in the burned area and taking what's called terrestrial LiDAR so these are measurements that are made from the ground. We have looked at spaceborne data, there's two systems up right now, one called ICESat-2, the other called GEDI, and we're looking at coincident data from those systems, so we've got the really, really local footprint level and then we've got the really broad-scale spaceborne and what we're trying to fill out with the UAS data is that middle ground. How do we scale up from really detailed point locations to the spaceborne picture, and then how do we link that with the other data, the spectral data that we already have on hand, because we're not saying that those data are worthless, what we want to do is supplement that with the structure data we can get from the LiDAR.

STEVE

So you have started using UAS systems--the general public might call them drones, but we refer to them as UAS--how many opportunities have you had to use that out in the Black Hills now and how's it been going with it?

BIRGIT

We had a couple of flights over our study sites within the fire perimeter back in August, and then in September we actually expanded, so there's some flights that occurred that were beyond the Legion Lake burn severity, and on National Forest land in areas where they're planning to do some prescribed burning. So what we were able to do is fly a couple of the UAS with different instruments on them, and do some data collections in the areas that they were hoping to do prescribed burns on this fall. What we were able to do through that was collect data pre-fire. That's one thing that we haven't really been able to do at the Legion Lake fire, because that was again just by chance, most of the data collection that we have done has been post-fire, and what we're really interested in is doing this pre- post-fire comparison. So the idea there is to collect data pre-fire, and then after the prescribed burning has occurred, go back and recollect data over those same areas, the same transects and start doing some of these comparisons. Now in prescribed fire again you tend to look a little bit more at that lower severity or that lower intensity of a fire and not the devastating canopy fire, but still, it's going to give us, what can we actually detect with these instruments and start help us figure out what metrics we can pull from the LiDAR point cloud or the other instruments that we have available to us.

STEVE

To a John Q Public, when you talk about point clouds, is there an easy definition of what that is? 

BIRGIT

So, when you're looking at a traditional image, you've got these individual pixels. Anybody who's done any type of digital photography now, and I think most people have at this point, are kind of familiar with that. If you look at it in an inappropriate resolution, you get this pixelation. Point clouds are something very different from that. So what the instrument is doing is actually sending out a laser, and that laser is interacting with what interferes its pathway as it's going towards the ground surface. So if it hits a leaf, or hits a branch, or it might go all the way down to the ground, and that triggers a return back to the instrument. So what's recorded is that location where it interacts with something, some object that's near the ground, whether it's natural, whether it's from a built environment, or what have you. So these points they have X, Y, Z coordinates, we can look at those and the individual point might not tell you much, but when you start getting collections of these points, you can start detecting objects within them. 

STEVE

Is it too early to say what we're learning from this 3D technology, or we've already ascertained some important information?

BIRGIT

We know we can get basic canopy metrics from LiDAR; that's something that's been done nigh on to decades now: things like height or canopy cover. What we're now teasing out is what can we infer from that interior vegetation structure. So one metric that comes up quite a bit when you're talking fire is "canopy base height" so, where does the bottom of the canopy really start, and that's harder to define. You know, where does a shrub layer end and then an actual canopy layer start? That's the type of thing that we want to use this 3D point cloud perspective to kind of pull apart and then also see what did it look like before the fire, and then what does the picture look like after, and how much volume was removed from that location.

GAIL

And Birgit, one of the things that we've talked about several times with these Rx burns, the prescribed burns, the moderate, the low severity and moderate severity burns again as you had talked about don't necessarily affect that canopy, and so being able to study pre- and post- in those low and moderate severity using these other technologies is important as compared to our traditional satellite imagery that usually is blocked by that--as you alluded to earlier--that dense canopy, so those prescribe burns are going to be important for us in the study area.

STEVE

Well, who might be the biggest beneficiaries of this kind of fire mapping, burn mapping, and the information you come up with? And beyond, who might benefit from it? How might they potentially use it?

BIRGIT

Really what we're looking at is refining the burn severity maps that are already there. So certainly the audience of people who is using what's already produced on an operational basis, I think would be the same. But where they really start using the data is to make again these management decisions. And if I have a better idea of how truly severe was the fire in my location, I can probably target what I want to do after the fire a little bit more precisely than I could before. I think it's really just going to offer better decision making capacity for those people who already using this suite of products. Another group that uses the burn severity maps is our USGS Landslide Hazards program. They want to be able to say if a fire has occurred, especially in an area with a high slope, and there's a follow-on significant rain event, what is the chance of a debris flow after that. So if we can start targeting, again, areas that experience especially high severity fire, and make good decisions about managing those lands after the fact, we might be able to mitigate some of those post-fire effects.

GAIL

I also feel that our discussions with some of these local land managers, that ultimately when they start to see the data that is made available from this project, they will likely have some additional ideas on how to utilize these data sets that maybe we as a team haven't thought about.

STEVE

I understand you're doing this in the Black Hills. I would surmise that maybe the hope is that based on what comes out of this work in the Black Hills that it could be used in other areas around the country?

BIRGIT

I think that would be our hope, is that if we can demonstrate a prototype out in the Black Hills, which is what we're doing right now and what we're working on, we would like to look at that in other areas as well. Where a challenge lies for us, being with the USGS, is that we're not a land management agency, so we always have to work in partnership with other land owners to get access to their lands. For example if we have an opportunity to work, say, in the Grand Canyon on the National Park Service in collaboration with them, and look at, okay, do the same metric, does the same process apply in a different type of forested system and just try it in different locales, so we can see well how does this scale, how does this work in different vegetation types, different forest types, different management types? Then we can start making assumptions about okay can this be something that folds into something that's operational? Often a lot of what we target is we move from experimental projects, which this is right now, to something that ends up being an operational system.

STEVE

Where do you go from here? I mean you've started doing this work, what future work, do you have plans, and what kind of timeline are you working on, if any? 

BIRGIT

Right now, we've done a lot of data collection, so we're in a processing phase right now. So we've done our T-LiDAR data collection, all those data need to be processed before we can really use them to do any type of analysis. So the airborne data or the UAS data that we collected, same deal. Also with the spaceborne data, we're looking at new systems such as ICESat-2 or GEDI. So we've kind of done the data collection stage and are pretty far along on the processing stage, and next comes the analysis, and that's kind of the fun part, where we can look at what other people, learn from that, but also the main thing is really pulling apart the metrics we can use to quantify what it is we're looking at, and that's going to be the challenge for us.

GAIL

And Steve, obviously we have to have fire on the ground yet right, so in order for us to go in and do post-fire in these areas, like Birgit had said earlier, there's Rx burns that need to be completed, and so ultimately as the conditions allow, and the conditions are part of that prescribed burn management plan, then the local agencies will be able to put that fire on the ground and we can come back in and do the post-fire analysis.

STEVE

Last question: A 3D technology to do burn mapping... is what we're doing unique at all, or how does it fit into the work that's being done nation-wide?

BIRGIT

Good question. The 3D technology that we're really looking at is the LiDAR data. And LiDAR has been around for quite a while now--a couple of decades at least--and been used in the vegetation community for a bulk of that time. And pretty much since its inception I think there's been this recognition that this would be a really cool tool for fire. It hasn't been as integrated as you would think. There's a fair amount of experimental work that has gone on, but that bridge to doing something operational really hasn't occurred, and I don't think it's for lack of trying but I think part of it is just the investment that needs to go into figuring out what works, what works well, and what can work in different locations relatively consistently. So I think the advent of the two spaceborne systems that we have now is really going to maybe cause a new awakening in this interest because we have data that area freely available, they're there for the entire country if not most of the world. We could start doing these exercises and it becomes more relevant to look beyond your own management unit, your own park, your own stake in the ground, and see what can happen beyond that and I think that's where EROS fits in really well, because a lot of what we do is regional to national in scope, and so I think we have a good role in bridging that local to national data integration. 

STEVE

We've been talking to EROS Geographer and Senior Scientist Birgit Peterson and her EROS colleague Software Engineer Gail Schmidt about their efforts to move toward a three-dimensional burn severity mapping methodology and the significance of that work. Thanks for joining us Birgit and Gail.

BIRGIT

Thank you Steve.

GAIL

Thanks Steve.

STEVE

We hope you come back for the next episode of Eyes on Earth. This podcast is a product of the U.S. Geological Survey, Department of the Interior. Thanks for joining us.