Editor’s note: Joseph Hardin, a computational scientist at Pacific Northwest National Laboratory in Washington state, provided the following summary of the 2020 ASR/ARM Topical Workshop on Machine Learning and Statistical Methods for Observations, Modeling, and Observational Constraints on Modeling. This virtual workshop took place October 19 and 20.
A group of scientists from DOE national labs and academia recently organized a workshop on machine learning (ML) and statistical constraint on atmospheric models with over 100 attendees from across the research landscape.
Participants discussed how ML and statistical methods could be used to develop improved observational products and for combining observations with modeling to constrain atmospheric models and their development.
During the workshop, more than 20 talks, including three keynote talks, were presented. Speakers shared their experiences and expertise on a wide variety of applications.
Keynote presentations explored the future of artificial intelligence (AI) and ML in DOE science (Rick Stevens, Argonne National Laboratory); a new National Science Foundation institute for trustworthy AI in weather (Amy McGovern, University of Oklahoma); and an exploration of interpretable ML techniques for climate sciences (Elizabeth Barnes, Colorado State University).
Additional presentations covered topics ranging from ML-based parameterizations and retrievals to Bayesian statistics for modeling constraints.
The workshop promoted thought-provoking discussions and a need for developing novel ML tools and algorithms for scientific applications aimed at improving the understanding of aerosol-cloud-radiation processes in earth system models of DOE interest.
Watch recordings from the workshop on this ARM YouTube playlist. You can view the agenda to see the full list of presentations and speakers.
Check back soon on ARM.gov for an in-depth article about the workshop and related ARM projects.