AI-augmented climate Simulators and Emulators

In recent years, AI has been disrupting conventional weather forecasting. We are beginning to witness the impact of AI on climate simulations. The fidelity and reliability of climate models has been limited by computing capabilities leading to inaccurate parametrizations of key processes such as convection, cloud, or mixing, and, consequently, to biases in large-scale phenomena such as temperature, rainfall, and sea level. These unresolved processes have posed a significant hurdle in enhancing climate simulations and their predictions. Here, we will discuss M2LInES’ new generation of climate models with AI representations of unresolved physics (ocean eddies, sea-ice, atmospheric convection), learned from observations and high-fidelity simulations, and their impact on reducing biases in climate simulations. The simulations are performed with the NOAA-Geophysical Fluid Dynamics Laboratory and Community Earth System Model (CESM) model components. We will further demonstrate the potential of AI to accelerate climate predictions and increase their reliability by generating emulators, which can reproduce decades of climate model output in seconds with high accuracy. This opens up the potential for large ensembles at scale, acceleration of model development, and more!