New Machine Learning Software to Accelerate Flash Flood Warnings

Developed by scientists, the new software called Transient Artifact and Continuous Learning System aims to increase early warning times against flash flood hazards by combining satellite data and artificial intelligence.

Flash Flood Hazards and Global Climate Impact

Floods are the second deadliest weather event in the United States and rank first on a global scale. Warming climate conditions are continuously increasing the frequency of extreme precipitation.

Flash flood events in the town of Lanesville, southern Indiana, demonstrate how devastating sudden dangers can be, forcing families to abandon their homes and take refuge with neighbors.

How TACLS Software Works

The newly developed Transient Artifact and Continuous Learning System brings together satellite technologies and artificial intelligence to detect flood-risk regions much earlier.

Currently, forecast offices rely on rain and stream gauges, satellites, and weather radars. However, these traditional methods do not always provide sufficient time for real-time precipitation tracking.

Global Navigation Satellite System and Atmospheric Moisture Tracking

The system utilizes the Global Navigation Satellite System network, which is primarily used to predict earthquakes. This network measures atmospheric moisture to provide forecasters with a real-time data stream.

The artificial intelligence infrastructure developed by Bhavik Chandna utilizes a long short-term memory architecture to track moisture changes as storms develop.

Applications and Future Plans

TACLS is currently being actively used by the Los Angeles and San Diego weather forecast offices.

A new version featuring advanced graphics is planned to be made available to all National Weather Service offices in the second half of October.