ai-modeling.png

For decades, NSF NCAR has served a central role in the Earth system science community as both a developer and long-term steward of widely used community codes, including Community Earth System Model (CESM), Community Atmosphere Model (CAM), Weather and Research Forecasting Model (WRF), and Model for Prediction Across Scales (MPAS). Through partnerships with universities and sustained institutional investment, NSF NCAR has provided scientific leadership, software engineering expertise, computational infrastructure, user support, and governance frameworks that enable these traditional, physics-based models to evolve in response to advancing science and emerging societal needs.

By maintaining open repositories, coordinating community contributions, conducting rigorous testing and validation, and organizing workshops and training activities, NSF NCAR ensures that these modeling systems remain robust, extensible, and accessible to researchers worldwide. 

Advancing the Next Generation: AI + Physical Science

As NSF NCAR and the global Earth and Sun system science community further embrace AI/ML technologies as essential elements of their scientific workflows, NSF NCAR is leading the development and stewardship of a new breed of community-driven AI/ML-powered tools and products. These tools support hybrid modeling, large-scale emulation, model evaluation, and more while maintaining scientific rigor and reproducibility. NSF NCAR is committed to delivering community-driven, advanced, scalable, transformative, and extensible technologies that lower barriers to entry while upholding the highest standards of ethical and responsible use.

A Shared Modeling Ecosystem for AI-Enabled Science

We are building an open, shared ecosystem that supports development, testing, validation, and reuse of AI-enabled models across the global community.

Explore key components of this ecosystem:

Each platform is designed to promote collaboration, reproducibility, and long-term sustainability.

ai-modeling-2.png

Figure: Annual mean temperature at the bottom model level from CAMulator (Community Atmosphere Model emulator) built on the CREDIT framework. From Chapman et al. 2025.