Science Advances · 13 November 2026 · From issue 67
Physics-informed and data-efficient machine learning for storm surge forecasting during typhoons
In simulations, a physics-based learning model beat standard forecasting past 16 hours out.
Huarong Xie, Changming Dong et al.
A new computer model predicts storm surge flooding from typhoons more accurately than a standard computer learning model. It mixes ocean physics rules with machine learning, so it needs less training data to work well. The model stays accurate for forecasts more than 16 hours ahead, when the older model starts to fail.
Integrity screen: passed (3 checks) Checked 17 November 2026. Read the source
The plain-language reading of this paper is in the full issue, for subscribers.
Read the full issue