Â鶹ÒùÔº

Flooding is and costliest natural disasters in the U.S., causing in damage each year. In 2024 alone, floods destroyed homes in and .

Southeast Texas was hit by flash flooding repeatedly in the spring, and then hit . In one heartbreaking moment, a was swept away after his family's car was submerged during a thunderstorm near Fort Worth.

In the Upper Midwest, days of rainfall in May caused and its tributaries. A slow-moving storm in the Northeast in August caused .

The mountains of North Carolina and Tennessee saw some of the year's most devastating flooding as the remnants of Hurricane Helene hit in September. Heavy rain poured down mountains, turning creeks and rivers into torrents that washed away homes and vehicles. in North Carolina, and six workers drowned when their in Tennessee.

Storms like these are and producing that the land can't absorb fast enough. While many coastal areas are becoming more prepared for hurricane and tidal flooding, inland flood risk is less understood or easily anticipated.

These disasters underscore the importance of fast, accurate flood warnings. They're also a reminder that extensive gaps still exist in the systems that monitor U.S. stream levels.

Current coverage is less than 1% of waterways

The National Weather Service uses advanced models to issue flood warnings. These models rely on , land cover information and a network of —sensors that provide near-real-time data on precipitation, streamflow and water depth—to simulate water flow. Much of that data is .

However, the streamgage network covers less than .

Deploying a single sensor for permanent federal gauges, and nearly 70% of these . These high upfront costs, combined with , significantly limit sensor coverage—.The U.S. Geological Survey acknowledges that these sensors alone at fast enough intervals to fully address flood risk.

Without data, risk is often underestimated

Flood risk can be estimated in waterways without streamgages, but not as accurately.

In these areas, computers use . However, , along with and the evolving , .

The often and overlook . In particular, they can miss new risks in , where changes to the landscape and more pavement can quickly funnel water in risky ways.

These flood models are used for more than warnings. They also guide for and decisions on building protective infrastructure, so accuracy is important.

A case study in Philadelphia

A July 2023 flash flood in Lower Makefield, a suburb of Philadelphia along the Delaware River, highlights the challenges of insufficient data coverage in urbanizing watersheds.

, heavy rain transformed Houghs Creek, a small tributary of the Delaware River, into a deadly torrent, washing out Washington Crossing Road and trapping multiple vehicles. :

"All of a sudden, 3 inches of water, 4 inches of water, a foot of water just coming at us," Chloe Weissman said.

"This huge gush of water just came down from … down a hill," added Eli Weissman. "As it was coming down, cars were starting to float. [We] just tried to survive, laid on our back, feet heading down, grabbing trees, grabbing vines, grabbing whatever we could to stay afloat."

The National Weather Service issued a , but a phone alert wasn't triggered until 6:09 p.m.—after the flooding had begun. While the Delaware River has a nearby streamgage, flood models did not predict the rapid flooding along this small tributary.

Urbanization around Houghs Creek has made these events more dangerous and less predictable, as impervious surfaces quickly funnel the water into low-lying areas. The flash flood underscored the need for hyperlocal data to improve predictive models and allow earlier, more accurate warnings.

Expanding coverage of stream flood levels

Addressing data gaps .

One is expanding the streamgage network through public-private partnerships and encouraging and governments, , academic institutions and nonprofits to build and operate their own sensors. Greater coverage enables more accurate and timely flood forecasts, leading to improved warnings, more prepared communities and more effective emergency responses when disasters strike.

Engineers at the University of Michigan created one example of a low-cost, easy-to-deploy solution for flood monitoring. At its core is a controller connected to an ultrasonic sensor that measures water levels in a way similar to how bats navigate using sound. The data can be transmitted in real time for fast analysis.

The simplicity and affordability—around US$800 per sensor—of this system allows for widespread deployment, providing critical information to communities. Techniques such as validating readings against precipitation measurements, calibrating sensors with federal monitoring stations and using can build confidence in the value of this third-party and citizen-generated data.

Eventually, nonfederal sensors like these may be .

In the meantime, researchers have created that consolidate all known gauge data and allow the public to supply information. These combined datasets allow more advanced and robust flood models, such as , which covers large portions of the country.

Future of flood monitoring

Several universities are working together in a collaboration called to develop a system that integrates "floodcams," social media posts, smart city sensors and more to detect and warn residents of flash floods. Bringing these tools together could greatly expand the data available to meteorologists and emergency managers, improving flood risk assessments and warnings.

Combining diverse sources of data on a shared platform would establish . We believe that would with the information they need to , ultimately enhancing resilience in the face of climate change.

This article is republished from under a Creative Commons license. Read the .