Eyes on Earth Episode 30 – Remapping LANDFIRE

HULT:

Hello everyone, and welcome to this episode of Eyes on Earth, a podcast focusing on our ever-changing Earth and on the people here at EROS and across the globe who use remote sensing to monitor the Earth. I'm your host, John Hult.

Today we are talking about Landscape Fire and Resource Management tools, a project more commonly known as LANDFIRE. The LANDFIRE group at EROS represent one of the partners in a geospatial mapping team that includes multiple federal agencies like the Department of Interior Office of Wildland Fire and the USDA Forest Service, and a non-profit partner called The Nature Conservancy. LANDFIRE products represent a dizzying array of data points on the nation's landscape that are useful for fire suppression and land management, but also for a host of research questions on ecosystems, hydrology, climate modeling and more. 

On today's episode we are joined by Randy Swaty, an Ecologist with the Nature Conservancy. Randy is going to walk us through what LANDFIRE is, tell us how it has been used around the country and tell us about a massive upgrade called the LANDFIRE Remap. Randy, welcome to Eyes on Earth.

SWATY:

Good morning! Thank you. Hi, everyone.

HULT: 

Ok, so let's get into this. First things first. What is LANDFIRE? Where did it come from?

SWATY:

Alright. I thought I would start with a quick story. Out of graduate school I got a job with Nature Conservancy out of Michigan where I worked with large landowners such as International Paper and Mead and the idea was to help them assess the landscape up here in Upper Peninsula Michigan. Naively I thought we had the data to do that. But we didn't. We had maybe good data for one area, no data for another area and maybe incompatible data for another area. So we couldn't answer basic questions such as how many acres of different ecosystems do we have across the landscape? Then along came LANDFIRE and I was super excited. So. LANDFIRE is a federal program aimed at mapping vegetation and fire characteristics across the United States including Alaska and Hawaii and the Insular Islands. It's a program that delivers over 2 dozen data sets, hundreds, literally hundreds of ecological models covering all the ecosystems of the United States. You know, if we would have had LANDFIRE data back then, it would have been easy to access the landscape and help managers prioritize what to do where. It maps over 9 billion pixels in the lower 48 and labels them with over 2 dozen attributes within each one of those data sets. So huge program available to all of us giving us all free data.

HULT:

And when you say pixel. Each pixel is...

SWATY:

30 meters by 30 meters.

HULT:

And again, you got into this from the perspective of someone who needed information and couldn't get it when you started. Was this the 90's or so when this happened?

SWATY:

This was early 2000s when I started working with the large industrial landowners and had to wait until about 2010 for some of the early data sets to come out.

HULT:

Are there studies and research projects out there that the public might be familiar with that are tied to the use of LANDFIRE data? You talked a little about what you are doing but what are some of the things that this is used for?

SWATY:

So now there's over a 1,000 research articles in Google Scholar that cite LANDFIRE. Those journal articles range from projects where they model effects from Atomic bombs and they use LANDFIRE vegetation data as part of that. It kind of blows me away, excuse the pun. There are papers mapping potential mycorrhizal fungi across the country. Papers where the researchers examine the potential water quality impacts of post fire erosion. So even though we are called LANDFIRE the data sets are used for all sorts of things. The public will probably never hear of LANDFIRE, like in media. But many of the land management decisions that they are aware of would be powered by LANDFIRE behind the scenes. So if you are out west and you see thinning projects to make it easier for firefighters to deal with wildfire in and around urban areas or cabins. Those decisions were probably powered by LANDFIRE data at some level.

HULT:

So in other words, people are going to see these things like forest thinning to help management and you talked about modeling hydrology, modeling stream flow, things of that nature. We are aware that kind of thing happens but this is maybe part of the mix of information sources that informs those decisions that we can see.

SWATY:

Right, right, right. Also, we need tons of data to keep firefighters safe when they're out on the line. Some of the models that power those decisions use LANDFIRE data as an input.

HULT:

It seems like maybe there are some places that have very good information about the landscape and the vegetation and then other areas that don't maybe have the high-resolution data. LANDFIRE is kind of the one national source for this kind of data. Is that right?

SWATY:

Right. That's a really good point. We are never out to replace a data set. If somebody has really great canopy cover data set for their landscape. Great. That's awesome. Use it, make the most of that. But if that data doesn't cover your whole landscape and you need something to fill in the gaps, we are great at that. As I mentioned before with my example, International Paper and other land managers like that. They have great standard inventory data but perhaps the neighbor doesn't. 

HULT:

Right.

SWATY:

Maybe it is a small family farm or something like that and they don't have data. But you want to know what would happen if a fire spread across that landscape so you need something to fill in the gap. That's where LANDFIRE really fits in well.

HULT:

Just for the record here. Do fires respect borders or....

SWATY:

Great question! No and as fire managers know well, spotting happens regardless. I am not myself in the fire management world but I have great respect for what they deal with.

HULT:

Right, so fires or like you say insects, disease things that span the landscape, they are going to jump from areas that have good data over to the area where there is not good data and LANDFIRE is the unifying factor that gives you the information.

SWATY:

Yeah, we try to be for sure.

HULT:

Let's jump into the Remap project here. That is the reason we are talking. Remap is completed for the conterminous United States. What is the remap project? Why did you guys take this up?

SWATY:

Well, first we need to have a couple terms in mind and get them understood. There is Refresh and there is Remap. So the Refresh process is where the mappers would take the original LANDFIRE data set called LANDFIRE National and where they had new knowledge, they would reclassify those pixels or that area. If in the National LANDFIRE data set they mapped an area as Ponderosa Pine Forest and they knew there was a replacement fire, they might reclassify that area as recently burned herbaceous or something like that. So they did that multiple times for 2008, 2010, 2012 and 2014. For the Remap process they have totally new base data from Landsat 8 and tons of new techniques including use of lidar data; so it was really important to stay relevant. You can only kind of refresh a data set so many times before it starts to degrade.

HULT:

I've heard the analogy that it's a bit like doing maintenance on a vehicle. The base map is 15 years old and you keep doing maintenance on the map just like you keep doing on a vehicle, but ultimately the newer model has features that you just cannot get. You just can't sort of build in. Is that fair?

SWATY:

Yeah, that's a really good one. You can still get a Toyota Corolla but maybe you want the 2020 model because it has more airbags, a better braking system and just new components compared to your 2005 Toyota Corolla. 

HULT:

And this is important, the base map itself is important. Cause, I think you got into a little bit of this ... all of these products, the hundreds of vegetation types, all the models....everything that you do with it, all the extra layers flows from the base map. It sort of is built up out of that base.

SWATY:

Right. For better or worse, yes. You know the whole junk in, junk out really applies. The other thing that is really exciting, I will mention briefly is the use of lidar data. You may know, lidar data to me is almost like echo location and bats. You have something that goes over the vegetation whether it is satellite, airplane or some handheld thing that is way up high sending laser packets down and then measuring the return rate to build this incredibly detailed vision of the canopy structure. LANDFIRE doesn't have lidar data everywhere but we can use it to train the processes so as a major improvement by the LANDFIRE mappers to incorporate lidar data super clever work, super hard work by the scientists with LANDFIRE and I think it is going to make a much better product for Remap.

HULT:

This gets into one of the basic things about LANDFIRE that's in the weeds: there is this idea that LANDFIRE is a landcover data set. There are landcover data sets out there but LANDFIRE is unique in its detail and one of those details is height. The height of the canopy is something you guys can provide for each of those pixels.

SWATY:

Right. And I think we can dig into this more if you want, but I'll explore this height thing a little bit. So, one of the improvements with Remap is the level of detail in the past if the life form is trees they would say the trees were: 0 to 5, 5 to 10, 10 to 25, 25 to 50 or 50 plus meters. The new data set has one meter increments. A pixel will say tree 10 to 11 meters so the level of detail is really incredible.

HULT:

How do you see the improvements in the Remap being useful to LANDFIRE users? I mean you are talking about this additional level of detail and this particular product. What are the improvements that are going to matter the most and how are they going to help people do their job?

SWATY:

As we know vegetation changes. Whether through succession or disturbance. So having the most modern up to date data set that is just a critical thing for managers. The Remap data set presents conditions on the ground as of 2016. So that just conceptually is super important. The other thing is those added details like I mentioned so the existing vegetation cover data set has a lot more detail as well. Those are 2 that have impacted me already for work I am doing on the Hiawatha National Forest. There's going to be a lot of accuracy gains in the new data sets. Again, I am not saying it is perfect. There are areas where there are a lot of room for improvement, I am sure. But I am feeling just by seeing the effort and the technologies used, there's going to be more accuracy which obviously helps users.

HULT:

Maybe I am wrong here but it seems like land managers and anyone doing this kind of work, land management work, fire management work.....the well is not endless, right? If you have a budget and constraints and you have a certain number of people who can do the work, I imagine you have to prioritize, right?

SWATY:

Yah and that is what my job is often about. You know, if we are working on a landscape in the central United States where land mangers tell me that if canopy cover of our Oak Hickories is over 70%, we need to do some thinning. Ok, I can map those with LANDFIRE data. If you give me some rules of what matters to you, I can usually combine LANDFIRE data sets in a creative way and apply rules to make those areas pop out. And what I tell people is I can tell you where you need to take the truck ride. I can't tell you exactly which acre to burn or thin or look at. And nobody should think of LANDFIRE data as being that detailed, but it helps point you to where to go to do things. It helps you understand the magnitude of issues.

HULT:

I want to get into a little bit of mapping stuff here some remote sensing stuff. There is another improvement that is noted in a lot of the information that we are looking at about Remap and that is the reduction of seamlines. What is a seamline and what are we talking about there? Did they cause problems before? How big of a deal is it that these seamlines in the map are disappearing in Remap?

SWATY:

Ok, when you say seamlines, the first thing I think of is pain.

HULT:

(Laughing)

SWATY:

It's a real drag. So what it is, is where, practically speaking it is where you will be cruising across visually and you will see some artificial line where maybe forest changes to Savannah and you will be like, "What? That doesn't happen!" Or you go from a Northern Horowitz Forest to Oak Hickory and it is just sudden. It is just like boom and you are like, "it is not a state boundary, it is not an ownership boundary, it is not some land management unit. What is that? It's a seamline and it's a problem. It's really hard to deal with. You know uhm, for me if I am showing a forester this data set and they see a seamline, they understandably question the data. What's going on there, I don't know if I believe this stuff and then analytically it is a problem. What do you do? Which side of that line is correct? And it is hard to avoid as well. Because almost no matter what we are having to break the landscape up into chunks. I don't understand exactly how they are eliminating those but I am really excited about it because, in the data sets I have seen so far the areas I have explored, they have done an amazing job reducing the impacts of those seamlines. It just looks a lot more realistic when you don't have these artificial lines on the landscape

HULT:

Why is it important for the research community to have access to these nation-wide data sets? Don't states, cities and counties have their own data?

SWATY:

Yah, well increasingly they do have their own data. It is getting easier and easier for even a graduate student to do remote sensing work. The challenge is that maybe that graduate student moves on, so that is a one-off data set. That is amazing for that time period. Maybe they did an amazing job at mapping fuels for an area. But they didn't map vegetation height or they didn't deliver that data set or they didn't deliver distance to the bottom of the forest canopy. If you are a researcher, it maybe important to have all those different data sets or even a sub-set and then it's important to have it across the landscape. For example, one of my favorite LANDFIRE powered studies was a Gauss hawk habitat map. And they took the input sort of the ingredients from LANDFIRE to build this map of Gauss hawk using succession classes, existing vegetation type and a couple other metrics and it's really hard to develop all that stuff. So those researchers benefited from LANDFIRE. I've been surprised how many researchers are stumbling into LANDFIRE and it's really cool to see how they have been putting the data sets together.

HULT:

How do you hope to see LANDFIRE data used in the future? Either avenues of study, research projects you see as potential benefactors especially with the Remap here.

SWATY:

One: Is just the idea of big data. I combined multiple LANDFIRE data sets this morning and generated a data table that has almost 700,000 rows in it. That's an area of research is just working with students coming out of school and other managers to deal with these larger data sets. Another thing is where more and more people are developing their own data sets. How can we work together to fill in the gaps? Another thing I wanted to mention is LANDFIRE does not get into climate change. We don't map climate change scenarios. But we do have state and transition models that users can use to jump start their climate change and other management scenario testing. You can take a LANDFIRE state and transition model, called a biophysical settings model, and then update it to represent current conditions with logging invasive species, fire suppression whatever the case maybe and then start to explore scenarios out into the future.

HULT: 

This has been a fascinating conversation Randy. I appreciate your coming on the program. Thank you so much for joining us.

SWATY:

My pleasure John.

HULT:

This podcast is a product of the U.S. Geological Survey, Department of Interior.