November 22, 2024 – A 12-member international team has been awarded the prestigious ACM Gordon Bell Prize for Climate Modelling for their groundbreaking project, “Boosting Earth System Model Outputs and Saving PetaBytes in Their Storage Using Exascale Climate Emulators.” Their innovation enhances the accuracy of climate predictions while significantly reducing the computational time and storage demands of high-resolution Earth System Models (ESMs).
Revolutionizing Climate Modeling
The winning team, comprising researchers from institutions across Saudi Arabia, the United States, and Europe, developed an exascale climate emulator designed to handle the enormous computational and storage requirements of ESMs. These models, made possible by advanced exascale supercomputers, offer unparalleled detail in understanding climate systems but are resource-intensive.
By leveraging Spherical Harmonic Transform (SHT) and Cholesky factorization, the team achieved ultra-high spatial resolutions of 0.034° (~3.5 km) and processed a staggering dataset with over 318 billion hourly observations across 54 million global locations. Their emulator also enables mixed-precision computations across thousands of nodes on supercomputers like Frontier, Alps, Leonardo, and Summit, saving several petabytes of storage space.
Why This Matters
As climate change accelerates, the need for precise and efficient modeling tools has never been greater. The emulator’s ability to reduce computational costs and storage demands will empower climate scientists to produce more detailed forecasts, aiding policymakers in developing targeted strategies to address the global climate crisis.
A Step Forward in Technology and Climate Science
The team’s emulator also highlights the potential of machine learning and AI-driven methods in climate research. Their work is expected to advance not only the science of climate modeling but also broader applications in data-driven prediction systems.
Key Contributors:
Researchers from King Abdullah University of Science and Technology (KAUST), Saudi Arabia, led the project.
Collaboration included experts from the NSF National Center for Atmospheric Research (USA), NVIDIA, St. Louis University, and the University of Notre Dame.
The award was presented at the International Conference for High Performance Computing, Networking, Storage and Analysis (SC24) in Atlanta, Georgia.
About the ACM Gordon Bell Prize for Climate Modelling
Established by ACM (Association for Computing Machinery), the award recognizes innovative use of parallel computing to address climate challenges. The prize, which includes $10,000, will be awarded annually for the next decade, spotlighting advancements in climate science through high-performance computing.
This breakthrough reinforces the vital role of computational innovations in addressing humanity’s most pressing challenges, from global warming to resource management.
