A new software system called the Transient Artifact and Continuous Learning System (TACLS) uses satellites and machine learning to identify areas at risk of flash flooding and help the National Weather Service issue faster, more accurate warnings, according to a report published by The Verge. Scientists from the University of California San Diego, NASA, and the National Weather Service developed the technology, which is already operating in California and will expand to all 122 NWS forecast offices nationwide in the second half of October. The system addresses a critical gap in flood alert timing — one that left residents in Lanesville, Indiana stranded on June 9th when over 8 inches of rain fell within hours but warnings arrived only after flooding had already begun.

TACLS works by tracking moisture levels in the atmosphere using the Global Navigation Satellite System, a network of satellites and ground sensors typically used for earthquake prediction. The more water vapor present in the air, the longer the delay in communication between GNSS satellites and sensors on the ground, providing real-time data about atmospheric conditions before a storm develops. A machine learning model trained on years of atmospheric measurements, precipitation records, and past flash flood warnings analyzes this moisture data to determine whether conditions warrant issuing an alert. The system differentiates genuine weather patterns from false positives by checking multiple nearby GNSS stations — if several neighboring sensors detect the same signal, forecasters can trust it represents an actual atmospheric threat rather than a sensor malfunction.

According to Ivory Small, science and operations officer at the NWS San Diego Weather Forecast Office, TACLS "will help you save lives" by giving forecasters critical lead time. Without the system, a storm moving into an area "could kill some folks," Small notes, but with TACLS "you can put out the warning and save some folks." Jayme Laber, senior service hydrologist at the NWS Oxnard office, explains that the technology provides real-time atmospheric data to compare against forecasts, which aren't always accurate, helping forecasters "make better decisions in our warning decision process." The report notes that just 6 inches of fast-moving water can knock down an adult, 12 inches can lift a car, and 2 feet can sweep away trucks and SUVs — making early warnings essential as climate warming increases extreme rainfall events across the United States.

The system fills gaps left by existing flood monitoring tools, which face significant limitations. Current NWS satellites provide more detailed data over oceans than land, and tracking rainfall as it falls doesn't give enough warning time "because once there's precipitation, you're already in the event itself," according to TACLS project lead Yehuda Bock. Rain gauges don't cover every location, with deserts and rural areas especially sparse. TACLS addresses these issues by monitoring atmospheric moisture changes as storms develop, giving forecasters insight into what's actually happening in the skies rather than relying solely on predictions. The machine learning component uses long short-term memory architecture, which excels at analyzing weather patterns that evolve over time, though developer Bhavik Chandna emphasizes the system is designed to assist human forecasters, not replace their expertise.

The technology will see heaviest use in the Western United States, where the majority of GNSS sensors are concentrated in earthquake-prone zones, though Chandna believes the approach could work anywhere with sufficient sensors and local weather data. A newer version with improved graphics showing real-time rainfall and extreme weather events is nearly complete and will be available to all NWS offices by late October. For residents like Laura Lin, who watched floodwaters rise in her Lanesville yard before any alert arrived, faster warnings could make the difference between safe evacuation and being trapped: "It would have been helpful to know [water was coming] before I was locked in," she recalls. The expansion of warning systems built on atmospheric monitoring represents a shift in how meteorologists approach prediction, moving detection earlier into the formation cycle rather than reacting to precipitation already underway. Still, the effectiveness of any alert technology ultimately depends on whether communities have sufficient infrastructure and evacuation plans in place to act on warnings once they're issued.