Emerging computational and data analytics capabilities for Earth system science

The overall goal of CISL’s research activities is to sustain progress in Earth system science through both fundamental and applied computational and data science research. To this end, CISL’s research portfolio is guided by the NCAR strategic plan. Data science research supports NCAR’s strategic imperative of improving predictions of weather and climate and better estimating their impact. Our research also furthers strategic efforts in community model development. CISL research on the use of emerging architectures and technologies, such as GPUs and machine-learning techniques, plays a central role in yet another imperative: developing new computational resources.

A hallmark of CISL’s research portfolio is its impact on both researching new algorithms and developing new tools and capabilities based on that research. For example, thanks to our hard work in collaboration with NCAR’s Mesoscale and Microscale Meteorology Laboratory and The Weather Company, an IBM subsidiary, the Model for Prediction Across Scales is now slated to become the first global atmospheric model running in production on GPUs.

Other transformative topics in CISL’s research portfolio include: the lossy compression of climate and weather data; novel data assimilation capabilities to deal with non-Gaussian processes and localized processes; statistical techniques for the rapid evaluation of the consistency of a new simulation with a reference ensemble population; and applications of artificial intelligence and machine learning to a variety of problems, from downscaling to the emulation of physical processes typically represented by bespoke software. While this research is aimed at supporting NCAR’s science mission, it also has broad applicability across the geosciences and beyond.

Another hallmark of CISL research is collaboration. For instance, our Data Assimilation Research Section has collaborated with more than 70 UCAR-member universities in a variety of data assimilation projects involving literally dozens of models. The new Analytics and Integrative Machine Learning (AIML) Group has established collaborations with every NCAR laboratory. AIML also engages groups with similar focus at major government agencies such as the Department of Energy and the National Oceanic and Atmospheric Administration; Lawrence Berkeley National Laboratory and other national laboratories; the University of Colorado, Boulder, the University of Oklahoma, and other universities; internationally, with the U.K. Met Office, the Australian Bureau of Meteorology, and others; and with the private sector – for example, ClimaCell Weather Technology and IBM.

As noted earlier, CISL does applied computational science research in high-performance computing (HPC) to understand how to create highly scalable, performant, and portable applications for emerging exascale computing architectures. For example, CISL adapts and develops computational science methods and tools to accelerate the pace of code optimization and porting, enabling NCAR applications to exploit new technologies such as graphics coprocessors (GPUs). 

Regarding the rapidly advancing area of artificial intelligence (AI), CISL is interested in both understanding what HPC cyberinfrastructure (CI) is required for supporting AI research in the future (CI for the AI) and in applying machine learning to optimize the operation of HPC cyberinfrastructure itself (AI for the CI). 

Finally, CISL’s research is enhanced by a robust set of ongoing partnerships and community engagement activities. These partnerships take the form of joint appointments with other NCAR labs; university partnerships focused on HPC workforce development; research and development projects with public and private sectors to coproduce new technologies and capabilities; and recurring conferences, workshops, and training events that focus on fostering the understanding and effective use of emerging technologies and techniques within the community.

This work is supported by NSF Core funding and other sources as specified in the following sections.