Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T21:42:35.966Z Has data issue: false hasContentIssue false

Simulating turbulent mixing caused by local instability of internal gravity waves

Published online by Cambridge University Press:  19 March 2021

Yohei Onuki*
Affiliation:
Research Institute for Applied Mechanics, Kyushu University, Kasuga, Fukuoka 816-8580, Japan
Sylvain Joubaud
Affiliation:
Univ Lyon, ENS de Lyon, CNRS, Laboratoire de Physique, F-69342 Lyon, FranceInstitut Universitaire de France (IUF), France
Thierry Dauxois
Affiliation:
Univ Lyon, ENS de Lyon, CNRS, Laboratoire de Physique, F-69342 Lyon, France
*
†Email address for correspondence: onuki@riam.kyushu-u.ac.jp

Abstract

With the aim of assessing internal wave-driven mixing in the ocean, we develop a new technique for direct numerical simulations of stratified turbulence. Since the spatial scale of oceanic internal gravity waves is typically much larger than that of turbulence, fully incorporating both in a model would require a high computational cost, and is therefore out of our scope. Alternatively, we cut out a small domain periodically distorted by an unresolved large-scale internal wave and locally simulate the energy cascade to the smallest scales. In this model, even though the Froude number of the outer wave, $Fr$, is small such that density overturn or shear instability does not occur, a striped pattern of disturbance is exponentially amplified through a parametric subharmonic instability. When the disturbance amplitude grows sufficiently large, secondary instabilities arise and produce much smaller-scale fluctuations. Passing through these two stages, wave energy is transferred into turbulence energy and will be eventually dissipated. Different from the conventional scenarios of vertical shear-induced instabilities, a large part of turbulent potential energy is supplied from the outer wave and directly used for mixing. The mixing coefficient $\varGamma =\epsilon _P/\epsilon$, where $\epsilon$ is the dissipation rate of kinetic energy and $\epsilon _P$ is that of available potential energy, is always greater than 0.5 and tends to increase with $Fr$. Although our results are mostly consistent with the recently proposed scaling relationship between $\varGamma$ and the turbulent Froude number, $Fr_t$, the values of $\varGamma$ obtained here are larger by a factor of about two than previously reported.

Information

Type
JFM Papers
Copyright
© The Author(s), 2021. Published by Cambridge University Press
Figure 0

Figure 1.(a) We consider an internal gravity wave obliquely propagating in a stratified fluid. Thick black curves represent the isopycnal surfaces and the arrows indicate the direction of the flow velocity. A Cartesian coordinate system $(x_1, x_2, x_3)$ is arranged such that $x_1$ points to the direction of the flow velocity, $x_3$ to the velocity gradient and $x_2$ perpendicular to them. (b) We cut out a small domain from the internal wave field, in which the velocity $U$ is linear in $x_3$ and oscillates in time. A time-dependent coordinate system $(\xi _1, \xi _2, \xi _3)$ that follows the background flow velocity is introduced. The shape of the model domain (thick parallelogram) is fixed in the $\xi$ frame and hence periodically distorted. (c) The wavenumber coordinates $(k_1, k_2, k_3)$, which are defined by taking the Fourier transform with respect to $(\xi _1, \xi _2, \xi _3)$, also vary with time.

Figure 1

Figure 2.A side view of the model domain. The periodic boundary condition connects points $a$ to $a'$ and $b$ to $b'$. Accordingly, the isopycnal surfaces $\rho = \rho _1$, $\rho = \rho _2$ and $\rho =\rho _3$ are connected with each other.

Figure 2

Table 1.Values of the control parameter $Fr \equiv S_0/N$, the total calculation time $t_{end}$ and the dimensionless quantities obtained in each run. Here, $Re_b \equiv \epsilon / (\nu N^2)$ is the buoyancy Reynolds number, $Fr_t \equiv \epsilon / (N \mathcal {E}_K)$ is the turbulent Froude number and $\eta _K \equiv (\nu ^3 / \epsilon )^{1/4}$ is the Kolmogorov scale. To obtain $\epsilon$, $\epsilon _P$, $\mathcal {E}_K$, $\mathcal {P}_K$ and $\mathcal {P}_P$, temporal averages are taken over $t_{end} - T < t \leqslant t_{end}$.

Figure 3

Figure 3.Results of the experiment with $Fr=0.4$. (a) Time series of the kinetic energy (red), the available potential energy (blue) and their sum (black). They are normalised such that the total energy becomes 1 at the end of the experiment. The vertical green lines indicate $t = 15T, 18T$ and $25T$, the times to which the following panels correspond. (b–d) Buoyancy perturbation $b'$ on the surface of the calculation domain at $t=15T$ (b), $18T$ (c) and $25T$ (d). Red and blue correspond to positive and negative values. (e–g) Energy spectra in the horizontal and vertical wavenumber space $E(k_H, k_V)$ at $t=15T$ (e), $18T$ (f) and $25T$ (g). Here, $k_H$ indicates the wavenumber against the horizontal directions $X_1$ and $X_2$, and $k_V$ indicates that of the vertical direction $X_3$. The spectra are integrated over the azimuthal angle and normalised such that $\int E(k_H, k_V) \,\textrm {d}k_H\,\textrm {d}k_V = 1$. The black dotted lines indicate $N \sin \phi = \omega$ and $N \sin \phi = \omega / 2$, where $\phi$ is the angle of the wave vector against the vertical axis.

Figure 4

Figure 4.The time series of the probability density function of the gradient Richardson number $Ri$ for the $Fr=0.4$ experiment. The probability density function $P(Ri, t)$ is normalised such that $\int P(Ri, t) \,\textrm {d} Ri = 1$. The white horizontal line indicates $Ri = 0.25$.

Figure 5

Figure 5.(a) Mixing coefficient $\varGamma \equiv \epsilon _P / \epsilon$ as a function of (external) Froude number $Fr \equiv S_0/N$. The colour represents buoyancy Reynolds number $Re_b \equiv \epsilon / (\nu N^2)$. (b) Mixing coefficient $\varGamma$ as a function of turbulent Froude number $Fr_t \equiv \epsilon / (N \mathcal {E}_K)$. The colour represents $Fr$. Here, the energy dissipation rates $\epsilon$ and $\epsilon _P$ and the kinetic energy density $\mathcal {E}_K$ are obtained by averaging each over $t_{end} - T < t \leqslant t_{end}$.

Figure 6

Figure 6.(a) Mixing coefficient $\varGamma \equiv \epsilon _P / \epsilon$ as a function of the ratio of the Ozmidov scale to the Thorpe scale $R_{OT} = \eta _O / \eta _T$. (b) Turbulent Froude number $Fr_t \equiv \epsilon / (N \mathcal {E}_K)$ as a function of $R_{OT}$ (the inset shows a log–log plot). Here, the energy dissipation rates $\epsilon$ and $\epsilon _P$ and the kinetic energy density $\mathcal {E}_K$ are obtained by averaging each over $t_{end} - T < t \leqslant t_{end}$. To estimate the Thorpe scale, we extract 800 samples of vertical density profile data within $t_{end} - T < t \leqslant t_{end}$.

Onuki et al. supplementary movie

Buoyancy perturbation on the model domain surface.

Download Onuki et al. supplementary movie(Video)
Video 9.7 MB