Google, IBM & Meta Certificates — All 10,000+ Courses at 40% Off
One annual plan covers every course and certificate on Coursera. 40% off for a limited time.
Get Full Access
Explore a cutting-edge approach to glacier ice flow modeling using Universal Differential Equations in this 38-minute conference talk by Jordi Bolibar at the Alan Turing Institute. Delve into the world of inversion methods and their crucial role in glacier models for parameter calibration and estimation. Discover how machine learning, powered by differential programming, addresses the statistical and computational challenges of processing massive glacier datasets. Learn about a novel statistical framework for functional inversion of physical processes governing global-scale glacier changes, with a focus on inverting spatial variability of ice rheology. Understand the innovative technique of embedding neural networks within the Shallow Ice Approximation PDE to create Universal Differential Equations, aiming to minimize errors in simulated ice surface velocities. Gain insights into ODINN.jl, an open-source Julia package for global glacier evolution modeling, and its integration with Python libraries like the Open Global Glacier Model (OGGM) and xarray. This talk showcases the potential of combining machine learning with glaciology to derive general laws governing spatiotemporal variability of glacier parameters on a global scale.
