Characteristics of drift effects in the quasi-geostrophic equation arising from nonlinear symmetry
Masakazu Yamamoto
TL;DR
This work analyzes the large-time behavior of two 2D diffusion-type models in meteorology: the quasi-geostrophic equation $\partial_t \theta + \mathcal{R}^\bot \theta \cdot \nabla \theta = \Delta \theta$ and the convection-diffusion equation $\partial_t \rho - \Delta \rho = a \cdot \nabla (|\rho|\rho)$. Using the Escobedo–Zuazua (EZ) expansion with renormalization, the authors decompose solutions into a Gaussian core and gradient corrections to isolate nonlinear contributions. They find a stark contrast: the convection-diffusion model exhibits a logarithmic drift term and a symmetry-distortion in the asymptotics, while the quasi-geostrophic model shows no nonlinear drift due to the odd-type symmetry and skew-adjointness of the Riesz transform, effectively mitigating the logarithmic growth. This reveals that nonlinear symmetry strongly governs the large-time profiles, aligning meteorological intuition that Coriolis effects preserve the radially symmetric component without altering the mean potential temperature. $\,$
Abstract
This paper compares two similar diffusion equations that appear in meteorology. One is the quasi-geostrophic equation, and the other is the convection-diffusion equation. Both are two-dimensional bilinear equations, and the order of differentiation is the same. Naturally, their scales also coincide. However, the direction in which the nonlinear effects act differs: one acts along the isothermal surface, while the other acts along the temperature gradient in a specified direction. The main assertion quantifies this difference through the large-time behavior of their solutions. In particular, the nonlinear distortions in the asymptotic profiles of both equations are compared. In this context, the spatial symmetry of the first approximation plays a crucial role, but the solutions require no symmetry.
