AI/ML Emulators and Dynamic Time Interpolators for km-scale Earth System Models

This NASA Postdoctoral Program fellowship offers a unique research opportunity to develop AI/ML models for emulating high-resolution Earth system models. The project focuses on creating dynamic time interpolators that can reconstruct atmospheric states between checkpoint intervals, providing a fast and accurate data compression mechanism for km-scale, non-hydrostatic simulations. The research addresses grand challenges in Observing System Simulation Experiments (OSSEs), particularly the lack of global cloud-resolving simulations representing convection and cloud-aerosol interaction processes. Unlike existing AI/ML foundation models for weather and climate which typically have lower spatial resolution, this project requires emulating the full km-scale, non-hydrostatic state of the atmosphere, including the full vertical column. The successful applicant will work on a prototype model based on CNN-Transformers with Laplacian eigenmap embeddings and have opportunities to contribute to AI/ML methods for data assimilation. This fellowship is part of NASA's competitive postdoctoral program designed to advance NASA's missions in space science, Earth science, aeronautics, space operations, exploration systems, and astrobiology.