All Climate Events

Webinar: Findings and future recommendations from a review of AI/ML activity across the NWS

Download as iCal file

Monday, 01 August 2022, 12:00

Monday, August 1, 2022. 12:00 PM. Webinar: Findings and future recommendations from a review of AI/ML activity across the NWS. Paul Roebber, University of Wisconsin Milwaukee. Sponsored by NOAA. More information here. Register here.

Abstract: A report was commissioned to summarize current AI/ML activity within the U.S. National Weather Service with a view towards identifying existing obstacles and recommending future directions. Although AI/ML activity is growing quickly, this activity is fragmented and lacks the needed infrastructure for improved coordination and efficiency of effort. Current obstacles to future progress include: lack of workforce training in artificial intelligence and machine learning, lack of curated datasets and software that can be used for development and evaluation of artificial intelligence/machine learning approaches, absence of a centralized clearing house available to weather service personnel for technical expertise/consultation, limited operational compute resources, and lack of a clear end-to-end project pathway that encompasses exploration, development, testbed/proving ground and operational implementation.

Each of these limitations is addressable. Partnering with the NOAA Center for Artificial Intelligence to develop National Weather Service specific training materials, using learning journey style materials, is of interest to that group and would help address the current knowledge gap within the weather service. The development of reference software and datasets and the establishment of a consulting team to work on specific projects with operational units will reduce siloed efforts and enhance productivity. By establishing funding vehicles for theme-based projects, and for which there is a sustainable pathway from initial exploration all the way through operational implementation, will help bridge the valley of death between research and operations. Agent-based modeling capability with the weather service is currently limited. Given NWS emphasis on Impact-based Decision Support Services (IDSS), agent-based modeling capability should be developed, since this approach can directly link natural and human systems, and can reveal non-intuitive, emergent properties of complex systems like decision support. Collaboration with academic experts in this area, through the above-mentioned sustainable funding pathways, can help to build this expertise.

Bio(s): Dr. Roebber received his BSc in Meteorology from McGill University in Montreal, Quebec, Canada in 1981. He earn a MS in Meteorology at the Massachusetts Institute of Technology in Cambridge, MA in 1983. In 1991, Dr. Roebber returned to McGill University where he completed is PhD in meteorology. Dr. Roebber's research interests include the following:

Synoptic and mesoscale meteorology

Climate dynamics

Systems modeling and data analysis

Numerical weather prediction

Location  Webinar