New satellite and machine learning tool aims to give earlier flash flood warnings across the US
TACLS uses GNSS signal delays to measure atmospheric water vapor and machine learning to flag storms shifting from heavy rain to flash flooding. The updated system is slated for all 122 NWS offices in the second half of October 2026.
A new software system called the Transient Artifact and Continuous Learning System (TACLS) is being deployed across National Weather Service (NWS) offices to help forecasters issue flash flood warnings sooner. Developed by scientists from UC San Diego, the NWS, and NASA, TACLS uses satellite data and machine learning to identify areas at risk of transitioning from heavy rain to dangerous flash flooding. The updated version is expected to be available to all 122 NWS forecast offices in the second half of October 2026.
The project was funded by NASA’s Earth Science Technology Office through its Advanced Information Systems Technology program. TACLS relies on the Global Navigation Satellite System (GNSS) — a network of satellites and ground sensors primarily used for earthquake prediction — to measure water vapor in the atmosphere. The more moisture present, the longer the communication delay between GNSS satellites and ground sensors, providing real-time data about atmospheric conditions ahead of a storm.
UCSD graduate student Bhavik Chandna spent roughly a year developing the machine learning component, training it on years of atmospheric measurements, precipitation data, and historical flash flood warnings. The system uses a long short-term memory architecture, which is suited to tracking conditions that change over time, such as developing storms. To reduce false positives, the system cross-references signals across multiple nearby GNSS stations rather than relying on a single reading.
TACLS is already in use at NWS offices in Los Angeles and San Diego, where flash flooding is common. Project lead Yehuda Bock, a research geodesist at the Scripps Institution of Oceanography, notes that current NWS satellite tools are more detailed over oceans than land, and that tracking rainfall as it happens leaves little lead time. TACLS is designed to address that gap by providing earlier atmospheric signals.
The stakes are significant. Floods are the second-deadliest weather event in the US and the deadliest globally. Just six inches of fast-moving water can knock an adult off their feet. Laura Lin, a resident of Lanesville, Indiana, experienced this firsthand on June 9th when over eight inches of rain fell in hours, flooding her property before any official warning reached her. “It came in so fast,” she said. Developers say TACLS is not intended to replace forecasters, but to give them better information when deciding whether to issue an alert.