The coronavirus pandemic has led researchers to shift gears or temporarily abandon projects due to health protocols or the inability to travel. But for Patrick Keys and Elizabeth Barnes, husband and wife scientists at Colorado State University, the past year has led to a productive research collaboration.
They teamed up with Neil Carter, an assistant professor at the University of Michigan, on an article published in Environmental research letters which presents a satellite map of human pressure on land around the world.
Keys, lead author and researcher at CSU’s School of Global Environmental Sustainability, said the team used machine learning to produce the map, which reveals where abrupt landscape changes have occurred around the world. The map shows a near-present overview of the effects of deforestation, mining, expanding road networks, urbanization, and increasing agriculture.
“The map we have developed can help people understand the important challenges of biodiversity conservation and sustainability in general,” Keys said.
This type of map could be used to track progress on United Nations Sustainable Development Goal 15 (SDG 15), “Life on Earth”, which aims to promote sustainable development while conserving biodiversity.
Eight algorithms to encompass data from around the world
Barnes, an associate professor in the department of atmospheric sciences at CSU, did the heavy lifting on the data side of the project.
While staggering parenting tasks with Keys, she wrote code like never before, working with billions of data points and training up to eight separate algorithms to cover different parts of the world. She then merged the algorithms to provide a transparent classification for the entire planet.
First, the two researchers had to learn to speak the other’s working language.
“Pat initially had an idea for this research, and I said, ‘Machine learning doesn’t work that way,’” Barnes said.
She then sketched out the components with it: The entrance is something we want to be able to see from space, like a satellite image; and the result is a measure of what humans do on Earth. The central part of the equation was machine learning.
Keys said what Barnes designed is a convolutional neural network, commonly used to interpret images. It’s similar to how Facebook works when the site suggests tagging friends on a photo.
“It’s like our eyes and our brains,” he says.
To develop the algorithm, they used existing data that categorized human impacts on the planet, factors such as roads and buildings, and grazing land for livestock and deforestation. Then, the convolutional neural network learned how to accurately interpret satellite imagery, based on this existing data.
From an analysis of a country to the world
The researchers started with Indonesia, a country that has undergone rapid change over the past 20 years. At the end of the summer, after being convinced of what they had identified in Indonesia using machine learning, Keys suggested looking at the whole world.
“I remember telling him it wasn’t possible,” Barnes said. “He knows that every time I say that, I’ll go back and try to make it work. A week later, we had understood the whole world.
Barnes said the use of machine learning is not foolproof and requires monitoring to ensure the data is accurate.
“Machine learning will always provide an answer, whether it’s garbage or not,” she explained. “Our job as scientists is to find out if this is useful.”
Keys spent many nights on Google Earth examining more than 2,000 places on the globe in 2000, then compared those sites to 2019. He noted the changes and confirmed the data with Barnes.
The research team also deepened their research in three countries – Guyana, Morocco and The Gambia – to better understand what they found.
Going forward, when new satellite data becomes available, Keys said the team can quickly generate a new map.
“We can plug that data into this now-trained neural network and generate a new map,” he said. “If we do this every year, we’ll have this sequential data that shows how human pressure on the landscape is changing.”
Keys said the research project had helped him lift his spirits over the past year.
“Honestly, I had a rough time during the pandemic,” he said. “Looking back, I was able to work on this exciting, fun, interesting and open project, and with great people. It shed a lot of light on the pandemic.”