The USGS definitive land cover database, National Land Cover Database (NLCD), published its earliest dataset for 1992. Its most recent publication based on a two- to three-year update cycle, NLCD 2019, characterizes land cover and land change from 2001 to 2019 produced by a 10-agency Multi Resolution Land Characteristics Consortium centered around NLCD. NLCD is currently working to produce an update to include data through 2021 with a planned release later in 2023.

Other, newer EROS-based land cover efforts include Land Change Monitoring, Assessment, and Projection (LCMAP), which offers land cover and land surface change data annually and with additional variables such as the magnitude and timing of changes. The most recent update, Collection 1.3, provides annual science products from 1985 to 2021 over the conterminous United States.

In addition, the Rangeland Condition Monitoring, Assessment, and Projection (RCMAP) focuses specifically on mapping western landscapes; the most recent update was released in December 2022 for the time series 1985-2021. The Landscape Fire and Resource Management Planning Tools (LANDFIRE) Program also provides a variety of land cover datasets, with the most recent LF 2020 Update released in 2022 for the years 2017-2020.  Both of these programs and product suites are part of the Multi-Resolution Land Characteristics (MRLC) Consortium and are integrated, in part or whole, into the NLCD suite of products. 

In addition to producing land cover and land surface change data, the LCMAP team has developed tools to help users more easily answer specific scientific questions. LCMAP data is complex, with multiple interconnected products, but the LCMAP tutorials can help automate the processing of products, and Pathfinder Workflows offer step-by-step instructions, with accompanying software code, for tackling examples of real-world research questions.

Giving Scientists What They Ask For

USGS Research Physical Scientist Chris Barber, who works on LCMAP, realized the importance of providing tools even before the release of the first LCMAP data in 2020. At the Pecora 21 conference in Baltimore, where Barber gave a workshop, he said, “one of the common themes I saw was users asking for tools.”

The goal of Pathfinder Workflows is to give scientists a path for using the data to find, say, the top four land conversions for the state of West Virginia from 1985 to 2021, said LCMAP data scientist Cole Krehbiel. With an automated process based in Python, scientists can easily tweak variables such as the variable, date range, or location—no need to redo point-click steps in a Desktop GIS environment to adjust to a different scenario. It’s especially useful for examining decades of data in an area that spans many pixels.

“College professors have asked me to send this to them so they can distribute it to their students. One of our target audiences is these new up-and-coming scientists and their big data,” Krehbiel said.

Sources/Usage: Public Domain. View Media Details

The EROS products are not the only remote sensing land change products available—“the science community is getting inundated with new products from different efforts around the world,” Barber said. Each is a little different than the others.

Scientists may find an advantage in using more than one of these products to answer a question more completely, and LCMAP recently released a pathfinder workflow to help with that, too, based on two EROS products.

The “Comparing NLCD and LCMAP Primary Land Cover” workflow helps scientists examine similarities and differences between the two products by using the Delaware River Basin as an example. Part of this is understanding crosswalks between NLCD’s 16-class Anderson 2 definitions and LCMAP’s 8-class Anderson 1 definitions, based on a classification system dating to the 1970s that was developed by USGS Chief Geographer James R. Anderson. 

“Sometimes, difference doesn’t mean wrong. Difference just means, well, it’s different. Sometimes there’s information in the differences,” Barber said. The differences may be explained by a difference in methodology or definition, he added.

For example, LCMAP may determine an area to be tree cover, while NLCD might call it a wetland. “You’re thinking, well, that’s kind of an egregious error,” Barber said—until you dive more deeply into NLCD land cover labels and discover the area is actually woody wetlands. “Flooded forest is still tree cover. It’s not necessarily one is wrong and one is right. It’s just that there’s different definitions.”

The threshold at which a pixel is considered or labeled as one type of land cover or another can vary from product to product, too.

Providing More Detail and Saving People Time

A tool like this can help compare not just the LCMAP and NLCD products but, with an automated process, it can be switched out for other products.

 “I developed a use case that was applicable to trying to find suitable habitat for a hypothetical bird species that needed large, intact, old hardwood forests,” said Krehbiel, who gave a presentation on it for the North American Congress for Conservation Biology. “I started by showing them how they can begin by using LCMAP to identify areas that have had no change in tree cover for 37 years. Then doing a little bit of GIS and finding the largest polygons of that intact forest for the entire time series. Then extracting NLCD data from there so we can get that higher thematic resolution. And then for users that need even more specific information, bringing in LANDFIRE to say that this is hardwood plantation forest from the Southeast.”

Krehbiel added, “It really shows, for all three of those products, how they can be combined to answer a question that would have been harder if they only went to the LANDFIRE data directly, or only NLCD, or only LCMAP.”

Research geographer Jennifer Rover, who leads the Applied Science component of LCMAP, said these tools are intended to “save the community a lot of time doing these processes themselves.”

“EROS has really tried to strive to make access to this data easier for the research community because it moves the research that much further along,” Rover added. “It’s like buying a cake mix instead of putting all the ingredients together yourself. This is the recipe that we use, and all they have to do is push the button. It’s a big time saver.”