Noise, ripples, and other oscillations in numerical weather > prediction. A brief survey of how to treat them.

Abstract: The advent of inexpensive massively parallel computers in the past fifteen years has revolutionized the way numerical weather prediction, climate and ocean modeling are handled today. In this talk, I will describe how this revolution happened, the reasons that drove it, and what challenges are still to be fully addressed and resolved as exascale computing approaches.

I will concentrate on the use of high-order continuous and discontinuous spectral elements (SEM) as they are proving their mettle for solving the Navier-Stokes equations to model atmospheric motion. Because of the high susceptibility of SEM to Gibbs oscillations in the solution to non-linear problems, special attention will be given to understanding how their stabilization is still an active topic of research and how we are contributing towards its solution.

Computer scienceMathematics

Audience: researchers in the topic

Modelling of materials - theory, model reduction and efficient numerical methods (UNCE MathMAC)