Project Grant 2616256

The National Science Foundation Office of Integrative Activities awarded The Research Foundation For The State University of New York $677,967 on August 1, 2026, to develop physics-informed artificial intelligence emulators for stratosphere-troposphere coupling under the Geosciences program (CFDA 47.050).

The project addresses a critical gap in current AI weather and climate models: they forecast near-surface conditions accurately but fail to represent the stratosphere and its connection to surface weather, limiting subseasonal-to-seasonal (2 weeks to 2 months) predictability. The work will build physics-informed machine learning tools that incorporate fundamental dynamical constraints into model loss functions to achieve robust representation of middle and lower atmosphere interactions. By clarifying when stratospheric information provides additional predictive value for surface weather, the project aims to identify forecasts of opportunity for high-impact events such as severe winter cold snaps and heavy snowfall across North America and Eurasia. The effort will build on existing weather emulators (Graphcast) and climate emulators (ACE2) and will deliver open-source software pipelines to advance weather forecasting and climate modeling across the scientific community. The project also includes cross-training of students at the intersection of geoscience and computer science and curriculum integration.

Performance occurs in Albany, New York, with a period of performance through July 31, 2029. The Research Foundation For The State University of New York serves as the fiscal and administrative agent for the State University of New York system.