Case Study: Caltech improves climate simulations with Google for Education and Google Cloud

CalTech scales climate simulations on Google for Education with up to 16 GPUs

The California Institute of Technology (Caltech) faced a challenge in climate modeling, specifically in simulating small-scale clouds that are crucial for accurate predictions but fall through the cracks of global models. To build a next-generation Earth System Model that could integrate vast datasets and reduce uncertainty, they partnered with Google for Education and utilized Google Cloud's Compute Engine and GPUs.

Google for Education provided the solution with its cloud infrastructure, enabling the CliMA team to develop the ClimateMachine model. This tool runs unified global and high-resolution local simulations efficiently. The result was a framework that demonstrated efficient scaling on Google Cloud, is expected to provide more accurate climate predictions, and is designed as an open-source tool to be accessible to all researchers.

CalTech