Though not as feared as tornadoes, hail, or lightning, flash floods are one of the deadliest weather hazards in the United States. While the number of fatalities can vary dramatically from year to year, the national average for flood deaths is 88. Lightning accounts for an average of 41 deaths per year while tornadoes claim an additional 68 on average.
Part of the danger of flash floods comes in their incredible power. Waters can rise suddenly and violently, moving with thunderous power and sweeping away anything in their path. Many people fail to realize the power of water in a flash flood. A mere six inches of fast-moving flood water can knock you off your feet and 12 inches can carry away a car.
Flash floods have also been one of the hardest severe storm hazards to predict. Relatively minute differences in rainfall rates and location yield massive differences in outcome. In areas of hilly or mountainous topography, a shift in a storm’s location by just a few miles can be the difference between a manageable rain event and a catastrophe. In the 2025 Texas disaster, the rainfall was almost wholly concentrated in a single basin—the Guadalupe River basin. Had the storms moved just five to 10 miles south, the water would have been split across different basins, likely preventing the historic scale of the disaster.
“The bigger killers are the ones that are difficult to predict,” said JJ Gourley, a researcher at the NOAA National Severe Storms Laboratory (NSSL). “These are going to be driven by convective storms. They could be supercells or storms that train over the same area, over and over again. These tend to produce really intense rainfall rates and you can see several inches of precipitation just a matter of minutes.”
First things first: Get the rainfall right
For NSSL researchers, innovating new methods to predict exactly where that intensity will manifest is one of their primary areas of focus.
According to Senior Coordination Hydrologist Greg Waller of the NOAA West Gulf River Forecast Center (WGRFC) the difficulty begins with “getting the rainfall right.” Forecasting exactly where a thunderstorm will develop, its intensity, and its type is an exercise in managing extreme uncertainty.
“The location of a thunderstorm can be off by 25 to 50 miles, and that could be two to three river basins,” said Waller. “The precise location really matters in the precise response from each river system.”
One of the tools developed by NSSL to address this is the Multi-Radar Multi-Sensor (MRMS) system. Referred to by some in the weather community as “the gold standard” for precipitation estimates, MRMS is a ground-breaking product that combines raw data from weather radars, rain gauges, satellites, and numerical models into one live, high-resolution picture.
MRMS data feeds models to improve forecasts and in real-time it allows forecasters to see a complete, seamless image of what’s really happening in the atmosphere. That picture can reduce the mental workload for NWS forecasters and hydrologists, saving precious seconds when every moment counts for issuing precise warnings.
PREDICTIONS IN A FLASH
Understanding the rain as it falls is only half the battle. The next step is knowing how the water will move once it hits the ground.
FLASH (Flooded Locations and Analysis in Short-term Hydrology) is a suite of products developed at NSSL that translates MRMS rainfall data into runoff predictions using physics-based modeling. It predicts water flow for more than 10 million grid points across the United States every ten minutes.
“Many of the worst flash flooding impacts aren’t occurring on big rivers,” said Race Clark, NSSL research meteorologist and MRMS program lead. “They’re occurring on small streams and other areas that aren’t necessarily well-observed using traditional observations. MRMS and FLASH fill in those gaps.”
WARN-ON-FORECAST
With innovations like MRMS and FLASH in place, and our ability to forecast and measure rainfall is improving, NSSL researchers are now focused on developing the next generation of flash-flooding tools.
Using advances in computing, artificial intelligence, and machine learning, the Warn-on-Forecast System (WoFS) produces multiple storm scenarios, predicting where and how intensely rainfall and flooding could unfold. WoFS generates rainfall rates and other severe weather outputs several hours into the future and then those scenarios are run through hydrologic models to better anticipate where flooding could develop.
“We’re going to help forecasters improve their ability to forecast impacts and also at much finer resolution—even down to intersections—where we’re going to have flooded areas that can be anticipated hours in advance” said Gourley.
Finding new ways that science can prevent this kind of loss of life is at the core of the NOAA mission and what drives NSSL researchers. Ultimately, the innovations created at NSSL will become the tools that tomorrow’s NWS forecasters use to issue life-saving warnings and information.
Credit: NOAA Weather Partners and NOAA Research
