Machine Learning in Extreme Weather Forecasting with William Drew Collins

Explore the groundbreaking application of machine learning in extreme weather forecasting through this kickoff webinar of the SIAM Geosciences Webinar Series. Led by William Drew Collins from Lawrence Berkeley National Laboratory, delve into a novel approach addressing the challenges of studying rare and impactful extreme weather events in a warming world. Learn how the FourCastNet ML algorithm, based on Fourier Neural Operators and Transformers, revolutionizes traditional numerical simulations by enabling the generation of massive ensembles for accurate analysis. Discover how this innovative method achieves high accuracy and fidelity at a computational cost five orders of magnitude lower than conventional techniques, allowing for unprecedented 1,000- or 10,000-member ensembles. Gain insights into how this breakthrough not only accelerates execution speed but also provides a data compression mechanism, tackling the computational and storage challenges associated with massive ensembles in extreme weather forecasting.