Federal Grant Award Summary

Award: NSF EPSCOR Research Fellows Program | Integrative Activities (CFDA 47.083) Funding Amount: $295,993 | Award Period: May 1, 2026 – April 30, 2028

Boise State University, in collaboration with Massachusetts Institute of Technology, will develop artificial intelligence tools and machine learning methodologies to enhance ocean modeling efficiency and accuracy. The primary deliverable is a deep reinforcement learning framework that automatically determines optimal mesh resolution in large-scale ocean circulation simulations, replacing traditional heuristic refinement methods. The project will integrate adaptive-grid ocean modeling with machine learning to improve computational resource allocation, enabling more efficient weather forecasting, coastal resilience planning, and natural disaster preparedness capabilities. The framework will employ reward functions designed for nonlinear shallow water systems to dynamically adjust computational mesh based on learned policies rather than user-defined thresholds.

The fellowship supports an Associate Professor and graduate student training at Boise State University, with the dual objective of advancing ocean modeling technology and building scientific machine learning research capacity in Idaho. Through this National Science Foundation (NSF) Office of Integrative Activities award, the project will strengthen institutional research capabilities and deepen inter-institutional collaboration between Boise State and MIT while delivering technical innovations applicable to large-scale computational ocean science.