On Preserving Scientific Integrity for Climate Model Data in the HPC Era

Abstract

Over the last 30 years, the Computational Science Graduate Fellowship (CSGF) program has played an integral role in preparing a large and diverse community of computational scientists to push the limits of high-performance computing (HPC). To celebrate the CSGF program’s enduring influence, I share my perspective from the climate modeling community, which has used HPC to better understand the earth’s climate system. While the benefits of HPC in climate science have been enormous, rapid computing advances have brought new challenges. One difficulty is quality assurance—ensuring that large and complex codes running on multiple platforms are correct. A second is mitigating the increasingly large data volumes. Here, I describe our software quality assurance framework and strategies to reduce climate data volumes. In these complementary efforts, our focus has been preserving the scientific integrity of the simulation output data, a priority shared across HPC model simulation disciplines.