This video is no longer available.
To map the Chesapeake Bay watershed, researchers have long relied on aerial imagery, a time-consuming process to document every water source. But artificial intelligence just might help to change that.
Researchers at the University of Maryland, Baltimore County, along with the Environmental Protection Agency and Chesapeake Conservancy, used high resolution photos and AI to study the watershed so closely that they could detect changes in elevation along every square meter. In doing so, they were able to increase the number of documented stream miles in the watershed from about 100,000 miles to more than 200,000 miles.
It’s the most precise picture yet of how even the smallest drops of water eventually make their way into the bay.
“The way that most people have done this in the past is to use very detailed aerial photographs and to look for evidence of streams in those photographs, and even then, it would be a lot of work to manually set eyes on every single channel,” said Matthew Baker, professor of geography and environmental systems at the University of Maryland, Baltimore County. “So the idea is to let the computer do that work for you.”
This provided Baker and others a way to identify streams and stream beds that might often be dry, but under certain circumstances, such as when it rains, they become the starting point for where water, as well as any pollutants or sediment, might come from.
“In many headwater locations, the streams aren’t flowing all the time, and so the channels exist as small gullies or depressions, and sometimes they’re wetlands,” Baker said. “We don’t think of them as stream channels.
“There’s this understanding that stream networks have many small branches that are often not mapped, and that we felt as we increase the resolution and the precision of our mapping process, we would see a lot more channels. And we did,” he added. “That’s consistent with probably what we would have expected when we started this process.”
The computer algorithm visualizes the terrain in the same way as if someone was there in person — analyzing the terrain, underground drainage systems and stormwater outflows.
“Because almost everything the population in the bay watershed does influences how water gets to streams, and the way that water gets to streams and what it carries with it — whether it’s pollutants, sediment or just the energy — can influence the condition of downstream water bodies, streams and rivers, and ultimately, what happens in the bay watershed.”
“So there is a connection between what happens in the landscape and what happens in the bay, and that’s part of the reason that it’s taken a long time to restore the bay,” he added. “That population has continued to increase and our manipulation of the landscape has continued to increase.”
But having a more precise mapping of where the water is flowing as it eventually makes it way to the bay can then inform state and local governments on land use and how the land is developed. It might also make you rethink how you shape the land you own.
“There are many places in the bay where nitrogen and phosphorus, or even sediment forwarded to the bay in small streams,” Baker said. “Sometimes it’s very difficult to know exactly where those are coming from, and one of the key pieces of information to make that assessment correctly is knowing where the flow paths are, where the stream channels are that can carry those pollutants.”
It also might help detect future flooding issues.
“Because those streams aren’t always flowing, we sometimes forget that they are an important part of the drainage network. And so humans, on purpose or inadvertently, sometimes will develop over them, put bridges over them, roads that block those flow paths,” he said. “So having those stream channels mapped, having the valleys that convey those floodwaters mapped, can be a useful resource for better planning in the future.”
Get breaking news and daily headlines delivered to your email inbox by signing up here.
© 2025 WTOP. All Rights Reserved. This website is not intended for users located within the European Economic Area.
