The ACM Gordon Bell Prize for Climate Modelling awarded a 12-member team for their project on exascale climate emulators, enhancing climate predictions and reducing storage needs. Their work allows for ultra-high-resolution climate modelling, addressing the urgent challenges posed by global warming. The award highlights the role of advanced computing in climate science and policy development.
The ACM Gordon Bell Prize for Climate Modelling has been awarded to a team of twelve researchers for their innovative project titled “Boosting Earth System Model Outputs And Saving PetaBytes in Their Storage Using Exascale Climate Emulators.” This prestigious award, bestowed by the Association for Computing Machinery (ACM), recognizes advancements in parallel computing that contribute to addressing the global climate crisis. The team consists of distinguished scientists from various esteemed institutions, including King Abdullah University of Science and Technology and NVIDIA.
Recent warnings from scientists about global warming underscore the urgent need for advanced predictive models. The increased frequency and severity of extreme weather events highlight the gravity of the situation, with climate change posing a substantial threat to biodiversity and life on Earth. Recent developments in computational tools, especially the introduction of exascale supercomputers, enhance our grasp of climate dynamics and forecast capabilities. These supercomputers can execute quintillion calculations per second, offering an unprecedented level of insight into climatic changes.
Recognizing that high-resolution Earth System Models (ESMs) are often computationally intensive, the prize-winning team has developed a sophisticated exascale climate emulator to alleviate these challenges. Their emulator allows substantial reductions in storage requirements for massive datasets generated during climate simulations while ensuring efficient computational processes. The team’s results indicate potential storage savings of several petabytes, which is equivalent to the capacity of around 170 high-end servers.
The team’s pioneering methods include advanced high-performance computing techniques, such as Spherical Harmonic Transform (SHT) and Cholesky factorization, allowing them to model the Earth’s climate at an ultra-high resolution. Their findings indicate a spatial resolution of approximately 3.5 kilometers across more than 54 million locations globally, integrating billions of observations.
Furthermore, the implementation of their model on diverse computing platforms demonstrates the emulator’s capabilities in streamlining climate simulations. The developments hold profound significance for advancing climate research and policy formulation. Additionally, the research also positions itself for further developments in machine learning and AI methodologies pertinent to climate forecasting.
This award was presented at the International Conference for High Performance Computing, Networking, Storage and Analysis (SC24) held in Atlanta, Georgia, showcasing the importance of computational innovation in tackling pressing global issues like climate change.
Climate modelling has been a key scientific endeavor since the 1950s, serving as a critical tool in understanding and predicting changes in the Earth’s climate. In light of escalating global warming due to human activity, there is an increasing demand for enhanced predictive models to inform policy and response strategies to mitigate climate impacts. The advent of exascale computing, capable of processing an immense volume of calculations, offers scientists unprecedented opportunities to create high-resolution models that are essential for detailed climate analyses. However, these models are often limited by their immense computational and storage requirements, necessitating the development of more efficient methodologies, such as the use of climate emulators to streamline processes and save resources.
In conclusion, the ACM Gordon Bell Prize for Climate Modelling recognizes innovative advancements in climate prediction and modelling technologies. The award-winning team’s development of an exascale climate emulator represents a significant leap forward in understanding the complexities of climate dynamics while addressing the crucial challenges of computational efficiency and data storage. As global warming continues to threaten ecological balance, such advancements are vital for enhancing climate research and formulating effective strategies for mitigation efforts.
Original Source: www.eurekalert.org
