Project Grant 2425658

The National Science Foundation (NSF) awarded a $443,674 Project Grant under the Geosciences Program (CFDA 47.050) to The Pennsylvania State University (Penn State) to develop advanced artificial intelligence (AI) models for predicting the evolution and initiation of thunderstorms. The project aims to enable breakthroughs in understanding the physical and dynamical mechanisms controlling convection initiation through explainable AI techniques and uncertainty quantification methods. Key objectives include: (1) designing a prognostic, diagnostic, and generative deep learning convection nowcasting system using satellite, radar, and surface weather data; (2) assessing the predictability and quantifying the uncertainty of the AI system's predictions compared to traditional ensemble numerical weather prediction; and (3) discovering the physical processes governing convection initiation predictability through physically-informed explainable AI. The project is expected to generate a convection initiation benchmark dataset and train students in cross-disciplinary collaborative research, thereby advancing both geosciences and AI capabilities.