Causality Detection and Multi-Scale Decomposition of the Climate System Using Machine Learning

Explore cutting-edge Machine Learning approaches for detecting causal relationships and identifying physically meaningful patterns in the complex climate system. Delve into Reservoir Computing's ability to systematically identify causal directions, coupling delays, and causal chain relations from time series data, highlighting its robustness to noise, computational efficiency, and effectiveness with high-dimensional data. Examine the application of Multi-Resolution Dynamic Mode Decomposition in extracting physically meaningful patterns from high-dimensional climate data, with a focus on its capability to identify the changing annual cycle. This 32-minute talk by Christian Franzke from PCS Institute for Basic Science offers valuable insights into advanced techniques for analyzing and understanding climate system dynamics.