The ACM awarded the Gordon Bell Prize for Climate Modelling to a team for developing a climate emulator that optimizes Earth System Models, enhancing accuracy while saving significant storage. Their contributions are pivotal in addressing climate change challenges using advanced computing methods. The award was presented during the SC24 conference in Atlanta, Georgia.
The Association for Computing Machinery (ACM) has awarded the prestigious ACM Gordon Bell Prize for Climate Modelling to a 12-member team for their groundbreaking project, “Boosting Earth System Model Outputs And Saving PetaBytes in Their Storage Using Exascale Climate Emulators.” This award celebrates innovative contributions in parallel computing that address the increasing concerns surrounding climate change. The team’s work enhances the accuracy of climate predictions while significantly reducing data storage needs.
The awardees hail from esteemed institutions including King Abdullah University of Science and Technology in Saudi Arabia, the US National Center for Atmospheric Research, NVIDIA, St. Louis University, and the University of Notre Dame. These scientists aim to tackle the growing crisis of global warming, exacerbated by human activities, by utilizing advanced computational strategies to refine climate models.
Exascale supercomputers, capable of executing quintillion calculations per second, present unprecedented opportunities for detailed climate analysis. However, the computational and energy-intensive nature of high-resolution Earth System Models (ESMs) raises substantial challenges concerning data storage. The team addressed this by implementing a climate emulator that optimizes storage capabilities while maintaining high model resolution, thereby significantly enhancing the modeling process.
Key to their methodology is the adaptation of mixed-precision computations along with the Spherical Harmonic Transform and Cholesky factorization techniques, which collectively provide a refined spatial resolution of 0.034 degrees, approximately 3.5 kilometers. The emulator efficiently handles vast data volumes, with the team asserting it could save several petabytes of storage, comparable to the capacity of around 170 high-end servers.
The team’s exceptional contributions were showcased at the International Conference for High Performance Computing, Networking, Storage and Analysis (SC24) held in Atlanta, Georgia, where they underscored the potential of their work not just for climate science, but also for advancing machine learning and AI forecasting techniques in the discipline.
The importance of climate modelling has escalated due to the alarming rise in global temperatures attributed to human activity, particularly fossil fuel consumption. Such changes have led to severe weather patterns, loss of biodiversity, and various environmental crises that threaten ecological and human health. Utilizing computational tools, scientists seek to grasp the nuances of climate dynamics better. The advent of exascale computing technology has allowed researchers to develop sophisticated Earth System Models that illustrate the climate with remarkable precision, despite challenges in computational load and data management that accompany these advancements.
In summary, the ACM Gordon Bell Prize for Climate Modelling recognizes the indispensable role of parallel computing in enhancing climate predictions. The prize-winning team’s innovative climate emulator promises significant advancements in the field, potentially leading to more effective climate strategies and interventions. As global warming poses an escalating threat to life on Earth, the application of such advanced computational methods is critical for future climate research and policymaking.
Original Source: www.eurekalert.org
