Atmospheric Predictability and a Factory for Gray Swans from Differentiable Weather Models

Name: Atmospheric Predictability and a Factory for Gray Swans from Differentiable Weather Models
Start: 2026-04-13T10:00:00-04:00
End: 2026-04-13T11:00:00-04:00

Prof. Greg Hakim

University of Washington

Monday April 13, 2026, 2 PM ET

Abstract:

Recently developed weather models use computing frameworks that exploit the efficiency of GPU hardware and have built-in methods for computing gradients. I will argue that this combination enables a revolution in the application of nonlinear adjoint methods to a wide range of problems, including novel exploration of dynamical process in weather and climate variability. Adjoints, which derive from gradient operations on a model, are useful for measuring the sensitivity of model outputs to inputs and parameters. They have historically been used with tangent-linear models primarily for optimization applications such as data assimilation. New computationally efficient and differentiable models allow for very deep, fully nonlinear, optimization. I will show how my group has used these methods to discover shadowing trajectories for testing the limit of predictability, and a factory for manufacturing gray swan extreme events.

Biosketch:

Greg Hakim is a leading atmospheric scientist currently using deep-learning models for novel research on predictability, data assimilation, sensitivity analysis and extreme events in current and historical time periods. He is the author of over 100 scientific papers and two textbooks on weather and dynamic meteorology. Greg has previously chaired the advisory panel for the Mesoscale and Microscale Meteorology Division at the National Center for Atmospheric Research (NCAR) and served on the Advisory Committee to the Geosciences Division at NSF, and the President’s Advisory Committee on University Relations (NCAR). He has undergraduate degrees in Math and Atmospheric Sciences, and a PhD in Atmospheric Science from the University at Albany, State University of NY. After a postdoctoral fellowship in the Advanced Study Program at the NSF National Center for Atmospheric Research, Greg joined the Department of Atmospheric Sciences at the University of Washington in 1999 where he is Professor and served as Department Chair from 2012 to 2017.