AI-Powered Maps Double Documented Stream Miles in Chesapeake Bay Watershed, Enhancing Restoration Efforts

A new hydrography dataset for the Chesapeake Bay Watershed has more than doubled the documented stream miles, increasing the total from approximately 150,000 to nearly 350,000 miles. This Hyper-Resolution Hydrography Data, a collaborative effort between the University of Maryland, Baltimore County (UMBC), the Environmental Protection Agency's Chesapeake Bay Program (CBP), and the Chesapeake Conservancy (CC), provides unprecedented detail on the watershed's water flow patterns. The project leverages high-resolution LiDAR data and artificial intelligence to create these detailed maps. LiDAR, a laser-based system deployed via aircraft, captures elevation data with centimeter-level accuracy, creating a three-dimensional terrain model. AI algorithms, using resources from UMBC's High-Performance Computing Facility (HPCF), process this data to identify and trace stream channels, estimate their width and depth, and refine the maps by excluding non-natural features like detention ponds and crop furrows. This innovative approach significantly reduces the cost, time, and labor required for stream mapping, completing the entire watershed in just two weeks—a task that could take years using traditional methods. Matthew Baker, a professor of geography and environmental systems at UMBC and a lead researcher on the project, highlights the importance of understanding the landscape's water flow. "The landscape is shaped by running water. Stream networks are the primary conduit between the watershed and the Bay, and now we can characterize that connection in ways that we've never been able to before," he says. The enhanced accuracy, validated by David Saavedra, senior geospatial technical lead at the Chesapeake Conservancy, achieves 94% for existing streams and 67% to 82% for previously unmapped streams. These detailed stream maps offer a tenfold boost in resolution, moving from a 1:24,000 map scale to a 1:2,400 map scale, where each pixel represents one square meter. This improvement in resolution allows for better alignment with newly developed land cover maps of the same resolution, enabling a more comprehensive view of the watershed. The new dataset will be instrumental in various environmental and conservation efforts. Environmental groups and government agencies, including the CC and CBP, can prioritize restoration projects, such as streamside tree plantings to mitigate erosion and improve water quality. Farmers and urban planners can use the data to manage agricultural runoff and development more effectively, avoiding flooding and minimizing impacts on wildlife habitats. David Saavedra emphasizes the potential impact: "When people begin using our hyper-resolution hydrography in conjunction with the one-meter land use data, it will be eye-opening to see just how connected the landscape is to our waterways. There are so many opportunities to improve our region's water quality that were not readily apparent with previous data." Labeeb Ahmed, a geographer at the EPA's Chesapeake Bay Program, adds that the lack of consistent high-resolution hydrography data has been a significant barrier to achieving conservation goals, such as mapping forest buffers, non-tidal wetlands, and species habitats. "This data release will enable novel and interesting research and scientific inquiries. I'm excited to see how other researchers and stakeholders will use this data in their conservation and restoration efforts," he says. The dataset, released on June 26, 2025, represents over six years of hard work and innovation. It sets a new standard for stream mapping and offers a powerful tool for sustainable management of one of North America's most critical ecosystems, the Chesapeake Bay Watershed, which spans six states and supports millions of residents and iconic wildlife like blue crabs and migrating shorebirds. Industry experts agree that this advancement is crucial for advancing hydrography and environmental management. The success of this project demonstrates the potential of AI in accelerating and enhancing geographic data collection, making it easier to update and apply in other watersheds. This could lead to more effective and targeted conservation efforts globally, setting a precedent for similar projects in other regions. The University of Maryland, Baltimore County (UMBC) is known for its strong programs in geography and environmental science, and the High-Performance Computing Facility (HPCF) played a pivotal role in processing the vast amount of LiDAR data efficiently. The Chesapeake Conservancy and the Environmental Protection Agency's Chesapeake Bay Program are leading organizations in the region, dedicated to protecting and restoring the bay and its surrounding areas. Together, they have created a dataset that promises to revolutionize environmental management practices.