Data-driven model reduction and climate prediction: nonlinear stochastic, energy-conserving models with memory effects
Seminar Room 1, Newton Institute
AbstractCo-authors: Mickael D. Chekroun (University of California, Los Angeles), Michael Ghil (University of California, Los Angeles)
This talk will focus on theoretical understanding and climate applications of a data-driven reduction strategy that leads to low-order stochastic-dynamical models with energy-conserving nonlinearities and conveying memory effects. New opportunities for climate prediction will be illustrated in the framework of "Past Noise Forecasting", by utilizing on the one hand estimated history of the driving noise by the low-order model, and on the other hand the phase of low-frequency variability estimated by advanced time series analysis.