Core Model Evaluation for Physical Oceanography Applications

On behalf of the Department of Commerce s National Oceanic and Atmospheric Administration, the National Ocean Service (NOS) is conducting market research to understand capabilities related to open-source, community-supported oceanographic models and their current and planned future capabilities. This information may be used to support an evaluation to determine an optimal set of physical core model codes and approaches that can meet the NOAA mission; this will help position NOS to incorporate advances in computing infrastructure, data integration, and non-physics-based prediction methodologies over the coming decade. NOS currently maintains several physical oceanography applications based on a variety of community-supported core models developed and supported by the external modeling community. As examples, the Chesapeake Bay Operational Forecast System (CBOFS) was built using the core model called the Regional Ocean Modeling System (ROMS), the Salish Sea and Columbia River Operational Forecast System (SSCOFS) uses the core model called the Finite Volume Community Ocean Model (FVCOM), and the Surge and Tide Operational Forecast System (STOFS-2D-Global and STOFS-3D-Atlantic) application runs on the Advanced Circulation Model (ADCIRC) and Semi-implicit Cross-scale Hydroscience Integrated System Model (SCHISM) core models. While this diversity allows for supporting the entire NOS mission, it also dilutes resources as multiple software packages need to be supported. This resource dilution impacts human capital, appropriation of funds, High-Performance Computing (HPC) allocations, and software maintenance efficiencies. The scope of this effort centers on an evaluation of physical oceanography models. There is a critical need for NOS to evaluate and select an optimal set of core community model codes that can: Holistically meet NOS mission and/or modeling requirements (e.g., safe navigation, coastal resilience, etc.), Position NOS to better incorporate advances in computing infrastructure (e.g., HPC, cloud, graphics processing unit (GPU) architecture, etc.), Leverage non-physics-based prediction methodologies like AI/ML, and Integrate seamlessly with NOAA s UFS. This evaluation will be based on both quantitative and qualitative criteria and will, at a minimum, target the underlying core codes currently used across NOS (i.e., ADCIRC, SCHISM, FVCOM, ROMS), as well as novel AI/ML architectures, hybrid physics-AI systems and other approaches or emerging models supported by academia and private industry that have the potential to meet NOS modeling requirements (see Attachment A). Please see the attached Request for Information and Attachment A - NOS Modeling Requirements. If interested, please complete the linked Google Form (https://forms.gle/eZJSqg9zMdii4cxE7).