Mean flow from phase averages in the 2D Boussinesq equations

Abstract

The atmosphere and ocean are described by highly oscillatory PDEs that challenge both our understanding of the dynamics and their numerical approximation. This paper presents a preliminary numerical study of one type of phase averaging applied to mean flows in the 2d Boussinesq equations that also has application to numerical methods. The phase averaging technique, well-known in dynamical systems theory, relies on a mapping using the exponential operator, and then averaging over the phase. The exponential operator has connections to the Craya-Herring basis pioneered by Jack Herring to study the fluid dynamics of oscillatory, nonlinear fluid dynamics. In this paper we perform numerical experiments to study the effect of this averaging technique on the time evolution of the solution. We explore its potential as a definition for mean-flows. We also show that, as expected from theory, the phase averaging method can reduce the magnitude of the time rate of change of the PDEs making them potentially suitable for time stepping methods.