Eyes on Earth Episode 106 – EROS 50th: Land Cover, Part 1

Jane Lawson:

Hello, everyone, and welcome to another episode of Eyes on Earth, a podcast produced at the USGS EROS Center, which celebrates its 50th anniversary this year. Our podcast focuses on our ever-changing planet and on the people at EROS and around the globe who use remote sensing to monitor the health of Earth. My name is Jane Lawson, and I'll be hosting today's episode, where we're talking about the legacy of land cover mapping at EROS in the first of two episodes about land cover. EROS set up the Land Cover Characterization Program in 1995, after years of using Advanced Very High Resolution Radiometer, or AVHRR, sensor data with one kilometer resolution to produce greenness maps and eventually a database of land cover characteristics. The Land Cover Characterization Program led to the first global land cover data set at 1 kilometer resolution and also to the more detailed 30 meter resolution National Land Cover Database, or NLCD, which has become the definitive U.S. land cover product. Our guest today is here to talk about early land cover work at EROS, where she helped create that first global land cover data set. Research geographer Jess Brown has worked at EROS for 33 years and currently serves as the project manager for the integration of NLCD and another EROS land cover project, Land Change Monitoring, Assessment and Projection. Welcome, Jess, to Eyes on Earth.

Jess Brown:

Hi, Jane. Thanks for having me. 

Lawson:

Let's start off by defining the term land cover. What exactly does that include? 

Brown:

So land cover refers to a simplification of the land surface. It includes properties or characteristics of the Earth's surface. So it really is exactly what it says. It's what covers the land. Land cover includes describing the type and geographical patterns of features like vegetation, bare soil or rock, developed surfaces like what you get in urban situations, roads, that sort of thing, and agriculture. 

Lawson:

And what is the value of mapping land cover? 

Brown:

So land cover as a map or a geographical construct - and really, honestly, what we do here at EROS as a geospatial product - it's an ingredient or a necessary input into many different types of studies, studies like carbon modeling or mesoscale climate modeling, hydrologic modeling. Because it's a simplification of the land surface, it's been used in so many scientific studies that are necessary for us to understand how the world is constructed, what are the changes that human beings have made on the surface of the Earth, how our natural communities or natural landscapes are doing, the health of our natural landscapes and so on. 

Lawson:

So it's really a foundational product that helps others build on.

Brown:

Land cover is considered a higher order product as well that we make from inputs, remote sensing inputs, like Landsat. So Landsat itself, that surface reflectance that we collect and process here at EROS, is an ingredient from which we analyze and characterize these land cover types and turn into these higher order geospatial products of land cover. 

Lawson:

Can you describe the early land cover mapping efforts at EROS from your perspective at the time? And did any innovations come out of these efforts? 

Brown:

Oh, so that's a large question. As you said, I started at EROS over 30 years ago. When I first got here, I was hired to work on a project related to mapping land cover. It was the first ever attempt to map the land cover of the United States at a 1 kilometer resolution. So what set this project apart was that the resolution was 1 km, and at the time, which was 1990, 1991, somewhere in there, that was much higher resolution than what mapped land cover was available yet for our country. And I was fresh out of graduate school and quite excited to start working on this. I was the only woman on the team working with a number of really wonderful colleagues. But we started in 1991, and we published our data for the Conterminous US around 1993, and this launched a whole set of activities. USGS was collaborating at that time with a lot of different organizations and had, under the leadership of Tom Loveland, had made connections to something called the International Geosphere Biosphere Program. This program that was in existence from the late 1980s and actually wrapped up in 2015 was involved, and on the ground floor really, with a huge number of activities focused on global change. So the concern about our world changing drove the IGBP, or this International Geosphere Biosphere Program, to publish a report in 1990 that stated that existing global land cover data were inadequate for the task at hand, which was understanding global change. And it also stated that land cover data are perhaps the most important and universally used terrestrial data set - so needed for modeling agricultural production, mesoscale climate simulations, determining land surface roughness for modeling many different kinds of things like wind energy resources, or actually back to the climate modeling, they need to understand surface characteristics like roughness and albedo, and a lot of those are calculated based on the land cover type that's present. It was needed for modeling land cover change and degradation; understanding, you know, how deforestation and desertification are occurring across the world and the effects of that; carbon cycle modeling, hydrologic modeling - so that can include, you know, overland flow of water, what it's impacting, you know, flood risk and the like; and modeling trace gas emissions. And these were all seen as things - including many other topics - that were really critical to studying global change. The 1 kilometer conterminous US effort grew into a global 1 kilometer effort utilizing many of the same techniques that were used for the first mapping of the United States at 1 kilometer, but expanding to global. So as big promises were made, and it was a huge amount of work, although, you know, given what we're tackling today, I don't know if I'd say it was easier then. We were, you know, pathfinding - we were doing things that hadn't been done before, so there was a lot of experimentation. We didn't have to work with quite the data density that we do now. The data volumes and data densities that we have to work with in this day and age - much, much higher resolution data leads to very large data volumes. But back then we didn't have the compute power or the software, the infrastructure that we have today. So it was challenging to work on both those projects, the 1 kilometer conterminous US mapping and then the global 1 kilometer, but I think the bar was perhaps lower than it is today. We have - you know, the first time you do something, it's maybe, you know, it's a learning process. So you accept some of the problems, you learn as you go and then you try to improve on it. So I see those two efforts as being really foundational, certainly, to many, many mapping efforts today. You know, a lot of the publications that our team published in the 1990s and the 2000s are still cited today. And because there are many, many, you know, countless, activities mapping land cover today, too many to count. In that first project, the 1 kilometer land cover mapping of the United States, the conterminous US, we were funded to do that, at least in part, by the Environmental Protection Agency. And the EPA wanted to know more about sources for pollution. Specifically, it's referred to as non-point source pollution. So pollution that comes from water - generally speaking, water that drains off of agricultural lands and with it brings transportation of other elements like fertilizer and pesticides. So EPA wanted to know that, but in order to model that, they needed to know where agricultural lands were located and how they were distributed across the country. So they funded USGS to do that. And the proposal for that was written before I came to work, and that funding brought me to EROS originally. And in 2000, the first Terra platform was launched and established by NASA, and aboard that platform was the MODIS instrument. And around 2000, MODIS started producing land cover at multiple resolutions, I believe, to 50 meter or 500 meter and 1 kilometer, as a regular part of their higher order products. So the MODIS instrument now had many, many products created from the MODIS - the basic MODIS surface reflectance data that was collected, and land cover was seen as one of those essential data sets that needed to be provided for people. I believe the frequency at that point after MODIS was launched was probably once every quarter, perhaps once every month. So it was a very frequently produced data set. And then another thing that grew out of the, really at the same time, that grew out of the 1 kilometer conterminus US mapping that was done with AVHRR was the National Land Cover Dataset. That was another outgrowth of the 1 kilometer mapping. So you know, these early efforts were definitely pathfinders for activities that we're now still strongly involved in here at EROS. 

Lawson:

I'll also point out that EROS is the home of the MODIS data as well, which is archived and available to the public here. Would you like to give any examples of who might have benefited from these mapping efforts and how?

Brown:

When we surveyed our users in the mid 1990s, we found that modeling applications represented about half of the uses. So models such as mesoscale climate simulations, modeling land cover change and degradation, carbon cycle modeling, hydrological modeling, all those things that I mentioned before. About 15% of the users and uses reported employing the global land cover data for mapping. And some specific examples that we found were assembling an atlas on human demographics and the environment; ecoregion mapping, so that concerns more sort of landscape characteristics from the source of ecological uses and ecological benefits, and then mapping global biodiversity. And then another 15, or 13% use the land cover data for land management or conservation applications. And those included understanding forest reclamation after a disturbance and determining the effects of land use on ecosystems. So those were more related to probably human uses of the landscape. Education represented about 4% of the reported database usage, so people were making maps or utilizing the data to make three dimensional visualization of the land surface for public education. They used it for validating paleo-ecological vegetation maps and predicting the consequences of power plant accidents and determining the critical loads of acid rain for vegetation. Now, acid rain was a global change concern back in the 1990s, but it's something that over time was resolved when I believe coal power plants were required to clean their smokestacks better. So that's a success story. Looking back, I would certainly give Dr. Tom Loveland credit for the USGS's role in these activities. In the 1990s, there were much fewer players in this arena of mapping land cover, and because of the efforts of the conterminous US land cover mapping at 1 kilometer, we had demonstrated that we could do this. We had demonstrated our capacity and capability to map. So I believe we established a reputation from that initial mapping effort that then allowed us to expand. Tom Loveland was a good friend and colleague of a man named Jim Merchant. Jim and Tom either wrote this proposal together, or Jim wrote it, and they had this brainchild. I mean, I would have liked to have been a fly on the wall to hear them scheme on this because, you know, this just wasn't being done. And I'm assuming they wrote the proposal in '89 or '90 because the funding came in, and I started work in 1991. 

Lawson:

What kind of foundation did early mapping provide to efforts that continue today at EROS?

Brown: 

So a lot of these early mapping efforts led EROS to have an incredibly strong science department and to perform work that is strongly rooted in user requirements, where we focus on transparency, repeatability, robust accuracy assessment and continuous improvement of land cover mapping and monitoring. Certainly the LCMAP that you mentioned, the Land Change Monitoring Assessment and Projection project that took place at EROS started in 2016 and and continued until - and has just wrapped up this year, honestly - is rooted in that foundation. In LCMAP, we focused on change and annual change of land cover across the country. You know, Tom Loveland has said many times it's not just about cover and one-off mapping, one-time mapping of land cover, it's about determining what change is happening on the landscape, when it's happening and what are the causes and consequences of those changes. 

Lawson:

What are some of your favorite memories of land cover work at EROS?

Brown:

Prepping for this podcast has taken me on a trip down memory lane. Some of that's actually emotional for me because these were the jobs that I worked when I first came to EROS, and these were the people that I worked with when I was really starting my career. And the bittersweet part of it - the bitter part of it is that we've lost some of these people. Sweet part is how much I really enjoyed that work. It was quite cutting edge. It was very - it was full of creativity and experimentation, and a lot of camaraderie and teamwork was involved, which I, to this day, that's one of the things I enjoy the most about my work here at EROS is working with smart people. And as I've gotten older, you know, more and more you realize that you need to turn over the work to the people who are much smarter than you. I also felt very fortunate to work with such giants in the field, like Dr. Tom Loveland and Dr. Brad Reed and Dr. Jim Merchant; other colleagues, Zhiliang Zhu, Dr. Zhu, Don Ohlen and Dr. Limin Yang. I mean, I was surrounded by really smart and creative people. It was honestly a really fun time to work in land cover. In those days, the field was so much narrower than it is now, so we felt the pressure to do it right and and to give people the highest quality data that we could. That's not to say that we didn't make many, many mistakes along the way. So it was a learning experience.

Lawson:

An enjoyable one. 

Brown:

Looking back, my predominant memories are positive. Now, you know, sometimes we remember things better than they were, and that's probably a good characteristic. But I honestly feel so fortunate to have been part of these mapping efforts. And today I feel very fortunate to still be working here with the USGS at EROS on land cover.

Lawson:

Any closing thoughts you would like to share? 

Brown:

In closing, I'd like to say this is still an amazing time in remote sensing and in land cover and other higher order properties that we can derive from remote sensing data. The world today, if you go out into the literature on land cover, for example, you will see hundreds, if not thousands, of publications every month. There's many opportunities for people to create their own land cover now. That wasn't really possible or wasn't as easy to do back in the 1990s. People have access to different platforms, different cloud processing platforms like Amazon or Google, and a lot more maps are being made. But I guess I would say to people, if you're involved in this, definitely look at the earlier research just to keep you on track and and to remind you what's already been done and to learn from past history.

Lawson:

Thank you, Jess, for joining us for this episode of Eyes on Earth, where we looked back at early land cover mapping at EROS. And thank you to the listeners. Check out our EROS Facebook and Twitter pages to watch for our newest episodes. You can also subscribe to us on Apple and Google Podcasts. 

Various voices:

This podcast, this podcast, this podcast, this podcast, this podcast is a product of the US Geological Survey, Department of Interior.