Clark Labs forecasts how land use will change across the globe by 2050

As leading scientists work to determine which areas of the Earth should be prioritized for conservation, they’re getting a helping hand from researchers at Clark University.

Clark Labs is in the process of using satellite imagery-derived landcover maps to forecast global changes to land use 30 years into the future. The project, which will help support sustainable development and conservation practices, focuses on identifying natural landscapes that are especially vulnerable to human-induced changes.

The project was recently highlighted by Jack Dangermond, D.Sc. ’11, president of mapping software maker Esri, in front of 80,000 virtual attendees of the 2020 Esri User Conference. The work being done by Clark Labs falls within a larger Esri initiative to support the E.O. Wilson Biodiversity Foundation’s Half-Earth Project, which aims to protect half the land and sea to ensure the long-term health of the planet and reverse the species extinction crisis.

“There are many different groups participating on that and we are one of the components. We are completing a vulnerability map on a global scale for loss of landscapes,” says Stefano Crema, chief applications research officer at Clark Labs. “The output of our work will highlight areas that are most vulnerable to change, given historical pressures.”

Clark and Esri already have created global and continental models that were presented at the Esri User Conference, and they are now working to model each country on Earth. Crema hired 12 students in Clark’s graduate and accelerated B.A./master’s degree programs to help with the work last semester and is working with seven students over the summer.

During last month’s conference, Dangermond discussed how researchers can help create a better future through geographic understanding. He shared the importance of partnerships like Clark and Esri’s in creating new experiences to address today’s challenges — both globally and locally.

“This gives a glimmer of the kind of scientific information we can expect more and more of in the future,” Dangermond told attendees. “This consortium envisions creating it, making it available, and most importantly, creating these new sorts of experiences that are easier to use than traditional GIS technology has been.”

Clark Labs has a long history of developing tools for assessing land cover change. For the current land-cover vulnerability project, the team is using data from the European Space Agency’s Climate Change Initiative. The Initiative maps global land cover on a yearly basis to predict changes to land cover in 2050, along with the vulnerability of land changing from a natural to human-modified class. Looking at maps from 2010 and 2018, Clark Labs is analyzing land-cover conversion over the eight-year period to forecast future changes.

“Given the two dates, we can extract for each 300-by-300-meter pixel on the globe what was there in 2010 and what was there in 2018. From there, we can quantify how much each one of those land covers lost to another land cover, and which land covers gained from another land cover,” Crema explains.

To predict how those areas will continue to evolve, the team must attempt to understand why a pixel changed over that eight-year period. Using variables such as elevation, distance to roads, and population density, the researchers test for an explanation for each type of change that occurred.

Then, the team uses the land change modeler developed by Clark Labs in the geospatial software system TerrSet to understand the dynamic of changes. From there, the researchers can determine how vulnerable each 300-by-300-meter pixel is to future changes.

“This map is a continuous map that highlights, based on historical information and on the variables included in the model, which particular pixels have all the characteristics to change. This is a vulnerability map,” Crema says. “When you think about predictions into the future — in this case, 2050 — one of the issues is that we could have an area that has a lot of potential to change, but it doesn’t mean it’s going to change. What dictates change is the demand of change — what is the proportion that we expect it to change in the future?”

To determine the demand of change, the researchers use the “business as usual” scenario, which assumes that the rate of alteration from 2010 to 2018 will continue into the future.

“The model is going to extrapolate the historical rate of change to find how many pixels are going to change for each one of the transitions modeled,” Crema explains. “The model is going to have to figure out if it is going to change — what is it going to change to? There’s a whole series of competition that takes place inside the model.”

At the end, the team is left with a land-cover map for the year 2050, and they can identify the most vulnerable areas across the globe.

“If you think about planning for conservation, which is part of the goal of the Half-Earth Project, and Esri’s goal as well, we want to find areas that are at high risk of conversion,” Crema says. “If we’re going to prioritize conservation areas, that’s part of the information we need to know.”

If an area is at high risk of human-induced changed and is also highly biodiverse, it will likely become a conservation priority, Crema explains. Areas that are at high risk of change but low in biodiversity may be a lower priority.

“When you think of conservation, there are many different goals you can try to extract out of this,” he says.

The project will run through August 2021. The new analytical layers created by Clark Labs are available to view through Esri’s ArcGIS Living Atlas of the World.

Student researchers include Anika Katherine Berger ’20, Olivia Rae Landry ’20, Rachel Elizabeth Corcoran-Adams ’21, Rowan Walter Moody ’21, Caroline Williams ’21, Juliette Gale ’21, Noah Jonathan Berkowitz ’20, Smruti Naik ’20, Isaack Okwaro Nyagweso ’20, Bryce Stouffer ’20, Josue Gabriel Yarleque Ipanaque ’20, Emily Evenden ’21, Miles Van Denburg ’21, and Lian Mingshuang ’21.