How to extract a spectrum from hydrodynamic equations

Abstract

Practical results gained from statistical theories of turbulence usually appear in the form of an inertial range energy spectrum $\mathcal{E}(k)\sim k^{-q}$ and a cut-off wave-number $k_{c}$. For example, the values $q=5/3$ and $\ell k_{c}\sim \mathit{Re}^{3/4}$ are intimately associated with Kolmogorov's 1941 theory. To extract such spectral information from the Navier-Stokes equations, Doering and Gibbon (2002) introduced the idea of forming a set of dynamic wave-numbers $\kappa_n(t)$ from ratios of norms of solutions. The time averages of the $\kappa_n(t)$ can be interpreted as the 2$n$th-moments of the energy spectrum. They found that $1 < q \leqslant 8/3$, thereby confirming the earlier work of Sulem and Frisch (1975) who showed that when spatial intermittency is included, no inertial range can exist in the limit of vanishing viscosity unless $q \leqslant 8/3$. Since the $\kappa_n(t)$ are based on Navier-Stokes weak solutions, this approach connects empirical predictions of the energy spectrum with the mathematical analysis of the Navier-Stokes equations. This method is developed to show how it can be applied to many hydrodynamic models such as the two dimensional Navier--Stokes equations (in both the direct- and inverse-cascade regimes), the forced Burgers equation and shell models.