Flash floods can strike without warning — this new technology could change that

On the morning of June 9, Laura Lin was working from home in Lanesville, Indiana, when heavy rain rapidly flooded her yard. Lin noticed debris floating near her barn, evacuated with her children, and sheltered with a neighbor; local reports indicate the town received more than eight inches of rain over just a few hours. The episode highlights how quickly flash flooding can arise, with floodwaters that may recede within hours but pose acute, immediate danger. Flash floods are among the most lethal weather events globally and the second-deadliest in the United States. National Weather Service offices — 122 forecast offices across the U.S. and territories — are responsible for issuing watches, advisories, and warnings; the agency defines a flash flood as flooding that develops in under six hours. Forecasters combine rain and stream gauge readings, radar and satellite observations, and local expertise to decide when to issue alerts, but observational gaps and limited lead time can constrain decisions. To improve early detection, researchers from the Scripps Institution of Oceanography at UC San Diego, the National Weather Service, and NASA developed TACLS, a tool that applies machine learning to satellite-derived and GNSS (Global Navigation Satellite System) data to flag areas at risk of transitioning from heavy rain to flash floods. Project lead Yehuda Bock and NWS collaborators say TACLS can provide additional indicators beyond conventional rain-tracking, addressing limitations such as sparser gauge networks in less populated regions and the fact that precipitation observations often arrive as the event is already unfolding. NWS staff involved in the project emphasize the potential life-saving benefits. Ivory Small, a science and operations officer at the NWS San Diego office, said the system could enable forecasters to issue warnings earlier in situations that might otherwise become deadly. Jayme Laber, a senior service hydrologist in Oxnard, explained how current alert categories are used operationally — watches issued 12–48 hours ahead to prompt preparedness, advisories for nuisance flooding, and warnings for imminent, life-threatening floods — and described how additional predictive signals could improve those decisions. TACLS was funded through NASA’s Earth Science Technology Office Advanced Information Systems Technology program and is being demonstrated initially in California with plans to extend to other forecast offices.