Urban flooding causes significant social, environmental and economic losses every year. In Baltimore, those losses have increased in frequency and intensity since flooding record began in 1748, and are expected to continue doing so as the climate changes. Traditional flood prevention approaches, which use concrete and other impervious surfaces to direct flow, often fail in urban settings because they impede rainwater infiltration. A team led by Penn State researchers worked to design mitigation techniques with those who know the city best - people who live in Baltimore. The team found that the collaboratively designed mitigation techniques, including planting rain gardens and cleaning storm inlets, could reduce flood depths by up to 55% in some locations.
The team published their findings in Sustainable Cities and Society.
"This study presents a co-designed urban flood modeling framework in which community and technical experts directly inform flood model development and adaptation design through long-term and iterative engagement," said Ava Spangler, a graduate student in the Department of Geosciences at Penn State and lead author of the paper.
This project is part of the larger Urban Integrated Field Program investigating climate hazards in American cities with the goal of bringing together interdisciplinary teams to work on science that is relevant to the communities. Each team brings together scientists from multiple disciplines like hydrology, ecology and landscape architecture with city planners and community members to best understand and address local climate hazards and impacts.
The urban flood modeling framework described in this paper focused on pluvial flooding - flooding caused by rainfall rather than river overflow or high tides from storms. Not only is it an understudied flood type, the researchers said, it was also a main concern expressed by the community members in Baltimore during workshops the team organized in the city. Over three years, the researchers learned about the community's concerns, as well as what type of mitigation techniques might be most feasible for residents. That information informed their model, which also needed to accurately account for the complexities of the terrain and of the flooding type.
"It's challenging to model this kind of flooding partly because we have to deal with complex urban micro-topography," Spangler said. "Remotely sensed elevation data captures big hills and valleys pretty easily, but struggles to capture very small topography, like the difference between a curb and the road. It's only about six inches, but that kind of topography makes a big difference in where the water flows."
Not only does urban topography make it hard to build a flood model, but pluvial flooding itself is challenging to capture because it is harder to predict and can occur very suddenly.
"Observations are one of the biggest limitations to building these types of models because floods come and go very quickly, and they can be hard to measure with tools like satellites," said Antonia Hadjimichael, assistant professor of geosciences and co-author on the paper. "The flood may only last a few hours, and that is not enough time to get a satellite to measure it, but it is plenty of time to get water damage."
The lack of observations also makes it hard to validate the model once it's built since there is sparse data to compare to, according to the researchers.
"We had a difficult time validating our model because there weren't any sensors placed in streets where this kind of flooding is happening," Spangler said. "Without sensors, we can't record how deep floodwaters got."
Despite the lack of direct sensor data, the team was still able to validate their model by using social media posts and news images with location tags, which let them compare the height of the water to known heights like curbs.
Spangler explained that this project helped fill a specific need to understand stormwater flooding and to help cities with adaptation strategies that community members are excited to have in their neighborhoods and city officials see as achievable.
"It's been really valuable and exciting to be able to talk to folks," Spangler said. "Over three years, we held several workshops where we talked to people about their concerns, priorities and objectives regarding flooding in the city, which helped us frame what questions to look into and areas to model."
Community engagement was especially important to Spangler, she noted.
"I love science, and I want to be able to help people, so I think that working with the community and being able to talk about what's actually feasible in the city was very powerful," Spangler said.
Hadjimichael explained that making sure projects and adaptations are relevant to the community can build real impact, and collaboration establishes relationships that build trust in science and scientists.
Through community workshops, the researchers discussed ideas for flood mitigation with local residents and city officials. After discussing the feasibility of different ideas, the researchers decided to model the impact of two adaptations: implementing green stormwater infrastructure and cleaning storm inlets.
Green stormwater infrastructure includes adding green spaces and permeable surfaces, like rain gardens, within existing city infrastructure to absorb water and slow the progress of a flood. The inlet cleaning adaptation is a maintenance solution that would require people to clear the leaves and debris that can clog the inlets built into a street or sidewalk curb designed to capture stormwater runoff. After modeling the impact of these adaptations, the researchers found that a combination of the two was the most effective at reducing flooding.
"Together these adaptations have the potential to reduce flood depths by as much as 55% in select locations, which would greatly reduce impacts on local stakeholders," Spangler said. "However, the combination was the best by only a small margin. Cleaning out inlets was by far the more impactful single mitigation strategy."
The researchers and their community collaborators continue to discuss the findings and work toward implementable solutions. In the future, they plan to evaluate how their findings compare to other climate adaptation studies in the city as part of the Baltimore Social Environmental Collaborative.
"Our next steps include going back to the bigger project and thinking about the interplay of solutions and co-benefits to see what can help multiple facets," said Hadjimichael, who is also a Wilson Faculty Fellow in the College of Earth and Mineral Sciences. "At the larger scale, we're investigating climate hazards like air quality and heat in addition to flooding. There may be co-benefits to some adaptations even if they aren't the best solution to a single problem. For example, rain gardens may not help flooding as much as inlet cleaning, but they may be able to help with heat."
The team also said they hope that more urban scientists use the strategy of community engagement for future projects across the country. Spangler said that the model is grounded in both hydrologic processes and community priorities, and the process could be carried out in more cities to create relevant and usable insights for communities preparing for climate impacts.
Additional Penn State co-authors include Carlin Blash, former undergraduate research assistant who earned her bachelor of science in earth sciences in 2025 and Hong Wu, associate professor of landscape architecture. Additionally, Masha Adib at Michigan State University, Mark Cameron at the Baltimore City Department of Public Works, Claire Welty at the University of Maryland, Doris Minor-Terrell at the Broadway East Community & Community Development Corporation and Benjamin Zaichik at Johns Hopkins University contributed to this work.
This research was funded by the Baltimore Social Environmental Collaborative Urban Integrated Field Laboratory Grant under award number DOE-SC0023217 and the U.S. National Science Foundation (NSF) Graduate Research Fellowship under award number DGE1255832. This content is the sole responsibility of the authors and does not necessarily reflect the views of the funders.
