Hostname: page-component-7b7567c8c-2n9c9 Total loading time: 0 Render date: 2026-11-16T22:17:43.292Z Has data issue: false hasContentIssue false

An asymptotic model for the propagation of oceanic internal tides through quasi-geostrophic flow

Published online by Cambridge University Press:  12 September 2017

G. L. Wagner*
Affiliation:
Department of Earth, Atmospheric, and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, MA 02139-4307, USA
G. Ferrando
Affiliation:
Département de Physique, Ecole Normale Supérieure, 24, rue Lhomond, 75005 Paris, France
W. R. Young
Affiliation:
Scripps Institution of Oceanography, University of California San Diego, La Jolla, CA 92093-0213, USA
*
†Email address for correspondence: glwagner@mit.edu

Abstract

We derive a time-averaged ‘hydrostatic wave equation’ from the hydrostatic Boussinesq equations that describes the propagation of inertia–gravity internal waves through quasi-geostrophic flow. The derivation uses a multiple-scale asymptotic method to isolate wave field evolution over intervals much longer than a wave period, assumes the wave field has a well-defined non-inertial frequency such as that of the mid-latitude semi-diurnal lunar tide, assumes that the wave field and quasi-geostrophic flow have comparable spatial scales and neglects nonlinear wave–wave dynamics. As a result the hydrostatic wave equation is a reduced model applicable to the propagation of large-scale internal tides through the inhomogeneous and moving ocean. A numerical comparison with the linearized and hydrostatic Boussinesq equations demonstrates the validity of the hydrostatic wave equation model and illustrates how the model fails when the quasi-geostrophic flow is too strong and the wave frequency is too close to inertial. The hydrostatic wave equation provides a first step toward a coupled model for energy transfer between oceanic internal tides and quasi-geostrophic eddies and currents.

Information

Type
Papers
Copyright
© 2017 Cambridge University Press 
Figure 0

Figure 1.Comparison of the exact hydrostatic mode-$n$ Boussinesq dispersion relation $\unicode[STIX]{x1D6F4}/f$ with the dispersion relations $(\unicode[STIX]{x1D70E}+\unicode[STIX]{x1D70D})/f$ implied by (3.24) and (4.5). $\unicode[STIX]{x1D6F4}/f$ is given in (4.4) while $(\unicode[STIX]{x1D70E}+\unicode[STIX]{x1D70D})/f$ for (3.24) and (4.5) is given by (4.3) with $c=0$ and $c=-3/2$, respectively. All three are plotted in the main figure versus $k/\unicode[STIX]{x1D705}_{n}\sqrt{\unicode[STIX]{x1D6FC}}$ on a logarithmic $x$-axis. A grey dotted line shows the asymptote at $k/\unicode[STIX]{x1D705}_{n}\sqrt{\unicode[STIX]{x1D6FC}}=\sqrt{2(1+\unicode[STIX]{x1D6FC}^{-1})}$ where the dispersion relation implied by (3.24) is undefined. The inset shows the fractional error $(\unicode[STIX]{x1D70E}+\unicode[STIX]{x1D70D}-\unicode[STIX]{x1D6F4})/\unicode[STIX]{x1D6F4}$ versus $k/\unicode[STIX]{x1D705}_{n}\sqrt{\unicode[STIX]{x1D6FC}}$.

Figure 1

Figure 2.Scattering of a plane wave with frequency $\unicode[STIX]{x1D70E}=f\sqrt{2}$ and thus $\unicode[STIX]{x1D6FC}=1$ by two-dimensional turbulence with maximum vorticity $\text{max}(\unicode[STIX]{x0394}\unicode[STIX]{x1D713}/f)\approx 0.14$ in the linearized Boussinesq equations and the hydrostatic wave equation. Parameters and initial conditions are given in § 6.3. (a–c) Show the initial turbulent vorticity, speed and energy spectra. (d–o) Show wave field evolution in four snapshots: (d–g) shows speed ${\mathcal{V}}_{B}$ in the linearized hydrostatic Boussinesq system (6.13)–(6.15); (h–k) shows speed ${\mathcal{V}}_{A}$ in the hydrostatic wave equation (6.11); and (l–o) shows the logarithm of the spectral measure $\unicode[STIX]{x1D710}_{A}$ from the hydrostatic wave equation. ${\mathcal{V}}$ and $\unicode[STIX]{x1D710}$ are defined in (6.19) and (6.20).

Figure 2

Figure 3.A qualitative physical space comparison between snapshots of wave speed from the linearized Boussinesq equations and the hydrostatic wave equation for four initial value problems with wave Burger numbers $\unicode[STIX]{x1D6FC}=0.2$, 0.4, 0.8 and 1.6. The snapshots are taken at $t=10\unicode[STIX]{x1D6FC}$ wave periods. The initial value problems expose an initially planar wave field to a two-dimensional turbulent flow with $\unicode[STIX]{x1D716}\approx \max (\unicode[STIX]{x0394}\unicode[STIX]{x1D713}/f)\approx 0.064$ and are described in § 6.3. (a–d), (e–h) Show wave speed ${\mathcal{V}}_{B}$ from the linearized Boussinesq system and ${\mathcal{V}}_{A}$ from the hydrostatic wave equation, respectively, and (i–l) shows the absolute error $|{\mathcal{V}}_{B}-{\mathcal{V}}_{A}|$.

Figure 3

Figure 4.Comparison of normalized potential energy spectral amplitudes $\unicode[STIX]{x1D710}$ defined in (6.20) for the same simulations considered in figure 3. (a–d) Shows $\unicode[STIX]{x1D710}_{B}$ from the linearized Boussinesq equations, (e–h) shows $\unicode[STIX]{x1D710}_{A}$ from the hydrostatic wave equation and (i–l) shows the absolute difference $|\unicode[STIX]{x1D710}_{B}-\unicode[STIX]{x1D710}_{A}|$, all scaled logarithmically.

Figure 4

Figure 5.Comparison of the polar-integrated spectral measure $\unicode[STIX]{x1D6F6}(\unicode[STIX]{x1D718})$ defined in (6.23) in the linearized Boussinesq equations (solid lines) and hydrostatic wave equation (dashed lines), both normalized by $\int \unicode[STIX]{x1D6F6}\,\text{d}\unicode[STIX]{x1D718}$ from the linear Boussinesq result. (a) Compares four solutions with $\unicode[STIX]{x1D716}\approx 0.064$ with varying $\unicode[STIX]{x1D6FC}$ while (b) compares three solutions with $\unicode[STIX]{x1D6FC}=0.2$ and varying $\unicode[STIX]{x1D716}$. A dotted line indicates the wave field’s initial wavenumber $\unicode[STIX]{x1D718}=16$. (c,d) Show wave speed ${\mathcal{V}}$ from the two models for the case $\unicode[STIX]{x1D716}=0.064$ and $\unicode[STIX]{x1D6FC}=0.1$ to illustrate the spuriously intense small-scale features that develop in the hydrostatic wave equation solution when $\unicode[STIX]{x1D6FC}/\unicode[STIX]{x1D716}$ is small. All snapshots are taken at $t\approx 13\unicode[STIX]{x1D6FC}/\unicode[STIX]{x1D716}$ wave periods.

Figure 5

Figure 6.Integrated and maximum pointwise error in 60 solutions to the hydrostatic wave equation corresponding to three values of $\unicode[STIX]{x1D716}\approx \max (\unicode[STIX]{x0394}\unicode[STIX]{x1D713}/f)$ and 20 values of $\unicode[STIX]{x1D6FC}$. For every solution the error is computed at $t\approx 6.5\unicode[STIX]{x1D6FC}/\unicode[STIX]{x1D716}$ wave periods. The integrated error is defined in (6.24) and maximum error is the maximum pointwise error in speed defined in (6.25).

Figure 6

Figure 7.Comparison of $\unicode[STIX]{x1D70E}{\mathcal{A}}_{n}$ (dashed lines) and ${\mathcal{E}}_{n}^{B}$ (solid lines), both normalized by the initial wave energy ${\mathcal{E}}_{n}^{B}(t=0)$. (a) Shows three cases with $\unicode[STIX]{x1D6FC}=0.4$, 0.8 and 1.6 with $\unicode[STIX]{x1D716}\approx 0.064$ and (b) shows three cases with the same $\unicode[STIX]{x1D6FC}$ and $\unicode[STIX]{x1D716}\approx 0.14$. Both ${\mathcal{A}}$ and ${\mathcal{E}}^{B}$ are conserved to within a few per cent in all cases except $\unicode[STIX]{x1D6FC}=0.4$ and $\unicode[STIX]{x1D716}\approx 0.14$. Note that the panels have different $y$-axes.