Flooding is one of the deadliest and most costly natural disasters in the U.S., causing billions of dollars in damage each year. In 2024, floods devastated over a dozen states, destroying homes and claiming more than 165 lives.
Southeast Texas experienced multiple flash floods during the spring, followed by Hurricane Beryl. In one heartbreaking incident, a 4-year-old boy was swept away after his family’s car was submerged during a thunderstorm near Fort Worth.
In the Upper Midwest, prolonged rainfall in May caused significant flooding along the Mississippi River and its tributaries. A slow-moving storm in August also caused catastrophic flooding in Connecticut.
In September, the remnants of Hurricane Helene brought severe flooding to the mountains of North Carolina and Tennessee. Torrential rains turned rivers and creeks into raging torrents, sweeping away homes and vehicles. More than 100 people died in North Carolina, and six workers drowned in Tennessee when their plastics factory was inundated.
Such storms are intensifying, with faster development, slower weakening, and more extreme rainfall that overwhelms the land’s ability to absorb water. While coastal areas are improving their preparedness for hurricane and tidal flooding, inland flood risks remain less understood and harder to anticipate.
These events emphasize the need for timely and accurate flood warnings. They also reveal gaps in the systems used to monitor U.S. stream levels.
Currently, the National Weather Service issues flood warnings using advanced models that rely on historical trends, land cover data, and a network of over 11,800 streamgages—sensors that provide near-real-time data on precipitation, streamflow, and water depth.
While much of this data is available online, the streamgage network covers less than 1% of the nation’s rivers and streams.
The high cost of deploying these sensors—over $25,000 for federal gauges—limits their coverage, especially in small and urban watersheds that are vulnerable to flash floods. Furthermore, the U.S. Geological Survey acknowledges that these sensors alone do not provide enough data at the necessary intervals to fully address flood risks.
Flood risk can be estimated in areas without streamgages, but such estimates are less accurate. In these areas, computer models use data from similar waterways to estimate stream flow, but these assumptions introduce uncertainty, particularly in fast-developing urban areas where changes to the landscape can quickly alter water flow patterns.
The limitations of flood models are highlighted by a flash flood that occurred in Lower Makefield, a suburb of Philadelphia, in July 2023. The heavy rain caused Houghs Creek, a small tributary of the Delaware River, to flood, washing out roads and trapping vehicles.
While the National Weather Service issued a flash flood warning, the alert was not triggered until after the flooding had already begun, as the models did not predict the rapid flooding of this small creek.
The urbanization around Houghs Creek made the flood more dangerous and unpredictable, as impervious surfaces funneled the water into low-lying areas. This incident underscored the need for more localized data to improve flood forecasting and provide more accurate warnings.
To address these data gaps, experts suggest expanding the streamgage network through public-private partnerships and encouraging local governments, businesses, and academic institutions to deploy their own sensors.
One promising solution comes from the University of Michigan Digital Water Lab, which has developed a low-cost, easy-to-deploy flood monitoring system. At around $800 per sensor, this system can be widely deployed, providing real-time data to help communities prepare for floods.
In addition to low-cost solutions, open-source databases have been created to consolidate known gauge data and allow the public to contribute information. This collaborative effort is helping to build more robust flood models, such as Google’s flood forecasting model, which covers large parts of the country.
Looking ahead, several universities are collaborating on a project called FloodAware, which aims to integrate floodcams, social media posts, smart city sensors, and other tools to detect and warn residents of flash floods.
Combining these diverse data sources into a shared platform could greatly enhance flood monitoring, risk assessments, and warnings, providing communities with the information needed to advocate for protective measures and build resilience in the face of climate change.
