NOAA SCALABLE FRAMEWORK FOR COASTAL OCEAN MODELING EMULATION

NOAA’s Acquisition & Grants Office is preparing to award a sole‑source purchase order to Fathom Science, Inc. for the development of a scalable framework that emulates coastal‑ocean modeling on large‑scale unstructured meshes. The effort will refactor existing AI‑based NeuroCast code and integrate advanced neural‑operator technologies (GINO, PINTO, transformer attention) to support the National Ocean Service’s surge and tide operational forecast system (STOFS). The contract covers one base year with an option for a second year, beginning 15 Sep 2026. Although the solicitation is non‑competitive, interested firms may submit supporting documentation or request additional information within 15 days of the notice. No competitive bid package is available. • Develop and integrate Geometry‑Informed Neural Operator (GINO) and Physics‑Informed Transformer Neural Operator (PINTO) into NOAA’s existing modeling suite. • Refactor code to run on large‑scale unstructured meshes for coastal ocean applications. • Ensure compatibility with the AI‑based NeuroCast system and STOFS operational forecast workflow. • Deliver a functional prototype by 14 Sep 2027, with documentation and knowledge‑transfer support. • Provide optional second‑year support for further refinement and operational transition.