The best flood warning systems? Here's what works to save lives.

Washington Post

10-07-2025

The flash floods that killed at least 117 people in Central Texas over July Fourth weekend have amplified concerns that many communities — especially in rural areas — don't have good warning systems to keep people out of harm's way when sudden storms send walls of water down once-docile creeks and streams.
It's extremely difficult to anticipate exactly where a flood will strike, experts say. Weather models have been steadily improving over time, but if rain shifts a few miles in any direction — especially in hilly areas with many different streams, creeks and valleys — it can completely change which waterways will flood.
'Right now it is impossible to make site-specific flash flood warnings,' said Witold Krajewski, an environmental engineer at the University of Iowa. 'People say, 'How come we didn't predict this bridge would be compromised?' Well, I'm sorry, we just can't. It's just impossible right now.'
But with better river monitoring and more ways to get the word out, emergency managers can at least give people more time to react when flash floods strike.
Most small rivers and streams are blind spots for flood forecasters.
In Tennessee, for example, two-thirds of river basins have no gauges to measure how high the water is, according to Alfred Kalyanapu, an environmental engineering professor at Tennessee Tech University. 'One hundred eighty-two basins in Tennessee don't have any eyes on the ground,' he said.
That's because river gauges can be expensive. Top-of-the-line versions from the U.S. Geological Survey — which gather reams of useful data on water flow — can cost tens of thousands of dollars. Most are concentrated on big rivers near big cities, which helps protect urbanites but leaves rural residents on their own.
So, Kalyanapu developed his own DIY river gauges that cost about $500 a pop. The simple systems can be installed on bridges. They estimate water levels by bouncing soundwaves off the water surface every 15 minutes and measuring how long it takes for the echo to come back. The data is much rougher than the USGS sensors, which directly measure much more information about water speed and volume — but it's good enough to keep an eye on flooding, Kalyanapu said.
He and his students have built about 15 sensors for rural Tennessee rivers, including half a dozen for Humphries County after a flash flood killed 20 residents in 2021. Other rural areas could follow suit, he said.
If states fully commit to the idea, they could build a flood warning system like the Iowa Flood Information System, which is considered one of the best in the country. After a record-breaking 2008 flood, the state legislature set aside about $1.2 million a year for the University of Iowa to develop a statewide flood monitoring system.
University scientists deployed nearly 300 low-cost stream and river gauges on unwatched waterways. They combined those observations with USGS river gauges, radar data on rainfall and local hydrologic models that predict how water will move through Iowa's soil, streams and cement. Eventually, they created an online map that gives visitors a real-time look at water levels in rivers across the state and forecasts for future flooding that update every few minutes.
'Is this a foolproof system? No, of course not. We had people die in Iowa in flash floods. But nothing like Texas,' Krajewski said.
Experts say the Texas floods are also a reminder of the importance of investing in meteorologists at the National Weather Service, emergency managers at FEMA and scientists across federal agencies and universities who sharpen the weather models that forecast rain and floods. Funding for all this work has been cut or frozen since President Donald Trump took office — but without it, flood forecasts and disaster response could suffer.
'People really need the support,' said Brooke Fisher Liu, a University of Maryland professor who studies crisis communications. 'You can't just have people prepare on their own. It's not realistic in all situations.'
Once local emergency managers know a crisis may be coming, they need several ways to get the word out: cellphone alerts, sirens, weather radios, cop cars playing warnings over loudspeakers, neighbors knocking on doors and so on.
'You can't just have one alert or warning channel. When you only have one, things fail and communities suffer,' Liu said.
Sirens have gotten lots of attention after the Texas floods, because local officials in Kerr County had decided they were too expensive to install.
Sirens can be a useful piece of a warning system, some experts say.
'Sirens tell people, 'Hey, something is seriously wrong and you need to see what's happening,'' said Cruz Newberry, owner of Table Rock Alerting Systems. His company installed sirens in the unincorporated community of Comfort, Texas, which is just downriver from Kerr County — and which in 1987 lost 10 teenage campers in flash floods.
During last week's flooding, Comfort's sirens sounded as intended, as captured in a Facebook video posted by the Boerne Star, a local news outlet.
But experts say sirens aren't much help on their own, because they don't give any information about the threat.
'Let's say you're a traveler visiting an area on vacation and a siren goes off,' said Harry Evans, a senior research fellow in the Center for Research on Water Resources at the University of Texas in Austin. 'What is it telling you? What is your recommended action?'
That's why it's important to combine sirens with well-crafted alerts, according to Jeannette Sutton, an associate professor researching risk communication at the State University of New York at Albany. People should be able to understand from every emergency message who sent it, what the hazard is, where it's happening, when the danger is coming and how they should protect themselves.
'Otherwise, they're going to spend time searching the internet trying to figure out, 'What the hell are you talking about?'' Sutton said. 'If they need to act immediately … that's really dangerous.'
When cellphone companies first implemented the wireless emergency alerts that allow authorities to set off disaster alarms on phones in a certain areas, they limited messages to 90 characters — too short to be useful, Sutton said. They've since expanded the limit to 360 characters to give more space for detailed instructions, but Sutton said many local emergency managers still stick to the old 90 character limit because they lack training.
Sutton designed a free tool to help emergency managers quickly write good cellphone alerts for 48 kinds of hazards by plugging information into preset forms. 'It's just like Madlibs,' she said. She has trained 500 people on the tool, but she's still trying to get the word out to more emergency managers.