In October 2015, devastating flooding across South Carolina over five days killed 19 people and caused an estimated $1.5 billion in damages. In addition, nearly 50 dams failed or were breached.
Eight years later in September 2023, unprecedented rainfall from Storm Daniel over the Wadi Derna watershed caused an extreme flooding event that hit the North African coastal city of Derna, Libya. And last July, the Guadalupe River in Texas surged nearly 30 feet as more than 100 people were killed, including 27 children and counselors at Camp Mystic.
A 2022 World Bank study estimated that 1.81 billion people are directly exposed to a 1-in-100-year flooding event. While extreme flooding cannot be prevented, Civil and Environmental Engineering Professor Jasim Imran’s team are studying causes and effects and developing prediction solutions for helping with improved early warning and evacuation strategies.
“Patterns have changed,” Imran says. “They’re not like before, so we need to be proactive with our infrastructure, our resources, and our understanding of the process. We need to be resilient to whatever comes our way.”
Last year, Imran’s team completed a study of the Texas floods, and graduate student Ayman Nemnem presented the findings at last month’s American Geophysical Union Conference. The findings included how two $300 sensors, built by the Adaptive Real-Time Systems Laboratory led by Mechanical Engineering Associate Professor Austin Downey, placed in the right location could have provided early warning signals and saved lives.
“By developing a model, we identified locations where the sensors could detect the rate at which the water was rising,” Imran says. “This would have provided four-to-five hours of lead time to evacuate the camp.”
Imran’s team also investigated catastrophic flooding in Libya, which claimed more than 10,000 lives. Their article, “How extreme rainfall and failing dams unleashed the Derna flood disaster,” which was published last May in Nature Communications, has been accessed more than 12,000 times with nine citations. The study has also been referenced by major international outlets, including The Washington Post and the United Nations Office for Disaster Risk Reduction.
The disaster was reconstructed using an integrated modeling approach that combined satellite imagery, simulations, machine learning, eyewitness accounts and digital elevation data to assess the impact of cascading dam failures. While the dams were in disrepair, the findings revealed that even if they were structurally sound, they would have only provided minimal protection against the extreme runoff.
The collapse of aging flood control infrastructures, compounded by inadequate risk assessment and emergency preparedness, escalated the disaster’s impact. Their findings underscored the urgent need for systematic dam safety evaluations, enhanced flood forecasting, and adaptive risk management strategies.
“The high number of casualties were only possible due to a dam failure,” Imran says. “We wanted to look more into what happened and if there was any data we could extract since there was not much to create a model.”
Imran’s team performed investigative work to obtain data, which included news reports and literature. Since acquiring rainfall data was challenging, they examined satellite data and compared it with a rain gauge that was near the affected area, allowing them to reproduce the event.
“Our main finding was that there would have been significant flooding and some casualties without any dams, but not at the scale of what happened,” Imran says. “The dams were not holding any water and since the rain fell in a short amount of time, it filled the wells over the top, causing a collapse.”
Imran’s interest in extreme flooding began with the devastating floods in South Carolina in October 2015. Imran’s team has published three articles on that event, including his most recent article in Water Resources Research this past November.
“We published the first article in the Journal of Hydraulic Engineering in 2017 after students working with Dr. [Hanif] Chaudhry and me conducted field sampling and analysis,” Imran says. “In a more recent study published in National Hazards Review, simulations led by Ayman showed that pre-storm pumping from some lakes could have reduced stress on downstream dams.”
According to Imran, dams that were constructed within the last 50 years were designed to withstand extreme storms. But a solid infrastructure is not effective without regular monitoring and maintenance. For example, if a small earthen dam is not maintained, animals can create a burrow and allow water to escape and cause a failure.
"I think South Carolina is working to make dams more regulated,” Imran says. “There are thousands of dams in the state, but the smaller ones are hard to manage and regulate because many of them are private and have different owners over time.”
Nemnem says that he is motivated in his work by knowing that he can help protect people from the devastation of extreme flooding. In addition, many of the processes that he studies as part of Imran’s team are not location specific and can be applied to dams and watersheds worldwide.
“We can improve preparedness through better forecasting, early warning and practical mitigation tools. Helping communities act sooner to reduce loss of life makes this work deeply satisfying,” Nemnem says. “Examining how failures at one dam can trigger downstream dam failures in South Carolina and Libya reinforced how transferable and urgently needed this research is for improving dam safety and emergency planning globally.”