Hostname: page-component-7b7567c8c-tnkh4 Total loading time: 0 Render date: 2026-11-16T02:26:17.049Z Has data issue: false hasContentIssue false

Mean zonal flows induced by weak mechanical forcings in rotating spheroids

Published online by Cambridge University Press:  14 April 2021

David Cébron*
Affiliation:
ISTerre, Université Grenoble Alpes, CNRS, 38000 Grenoble, France
Jérémie Vidal
Affiliation:
ISTerre, Université Grenoble Alpes, CNRS, 38000 Grenoble, France
Nathanaël Schaeffer
Affiliation:
ISTerre, Université Grenoble Alpes, CNRS, 38000 Grenoble, France
Antonin Borderies
Affiliation:
ISTerre, Université Grenoble Alpes, CNRS, 38000 Grenoble, France
Alban Sauret
Affiliation:
Department of Mechanical Engineering, University of California, Santa Barbara, CA 93106, USA
*
†Email address for correspondence: david.cebron@univ-grenoble-alpes.fr

Abstract

The generation of mean flows is a long-standing issue in rotating fluids. Motivated by planetary objects, we consider here a rapidly rotating fluid-filled spheroid, which is subject to weak perturbations of either the boundary (e.g. tides) or the rotation vector (e.g. in direction by precession, or in magnitude by longitudinal librations). Using boundary-layer theory, we determine the mean zonal flows generated by nonlinear interactions within the viscous Ekman layer. These flows are of interest because they survive in the relevant planetary regime of both vanishing forcings and viscous effects. We extend the theory to take into account (i) the combination of spatial and temporal perturbations, providing new mechanically driven zonal flows (e.g. driven by latitudinal librations), and (ii) the spheroidal geometry relevant for planetary bodies. Wherever possible, our analytical predictions are validated with direct numerical simulations. The theoretical solutions are in good quantitative agreement with the simulations, with expected discrepancies (zonal jets) in the presence of inertial waves generated at the critical latitudes (as for precession). Moreover, we find that the mean zonal flows can be strongly affected in spheroids. Guided by planetary applications, we also revisit the scaling laws for the geostrophic shear layers at the critical latitudes, and the influence of a solid inner core.

Information

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

Figure 1.Comparison of the azimuthal component of the time-averaged zonal flow $\langle \bar {V}_\phi \rangle$, as a function of cylindrical radius $s$ in the equatorial plane $z_0=0$, between theory (dashed curves) and experiments/numerics (solid curves). (a) Libration-driven zonal flows with dimensionless angular frequency $\omega =0.1$ in a full sphere, extracted from figure 3(b) and figure 4 in Sauret et al. (2010). Experiment with $E = 1.15 \times 10^{-5}$, and $\epsilon =0.08$. DNS with $E = 5 \times 10^{-5}$, and $\epsilon =0.2$. The vertical axis has been normalised by $\epsilon ^2$, where $\epsilon$ is the dimensionless forcing amplitude. (b) Precession-driven zonal flows extracted from figure 8 in Noir et al. (2001b).

Figure 1

Figure 2.(a) Spheroidal geometry of the forced problem. (b) Equatorial (circular) section.

Figure 2

Figure 3.Azimuthal component of the time-averaged zonal flow $\langle \bar {V}_\phi \rangle /\epsilon ^2$ as a function of cylindrical radius $s$ in the equatorial plane $z_0=0$. DNS in a sphere with $E=10^{-4}$, $\epsilon =10^{-2}$ and forcing frequency $\omega ={\rm \pi}$, using Nek5000 (red solid curves) and xshells (blue dashed curves). (a) Longitudinal librations. (b) Latitudinal librations.

Figure 3

Figure 4.Rotation rate of the mean zonal flow, as a function of the cylindrical radius $s$, in rotating spheres subject to weak longitudinal librations (with $\varOmega _0=0$, $m=0$). (a) Our analytical results (solid coloured curves) are compared with the theoretical results (black dashed curves) of Sauret & Le Dizès (2013). (b). Comparison between the theory (black dashed curves) and DNS (solid coloured curves) at $E=10^{-7}$ and $\epsilon =10^{-4}$.

Figure 4

Figure 5.Rotation rate of the mean zonal flow, as a function of the cylindrical radius $s$, in rotating spheroids subject to longitudinal librations with $\omega = {\rm \pi}$. (a) Comparison between DNS with $E=10^{-5}$ and $\epsilon =5 \times 10^{-4}$ (solid coloured curves) and theory (dashed black curves). Numerical profiles have been truncated for $s\leq 0.05$, because there are not enough grid points near the centre to get reliable numerical averages with formula (2.8) for the rotation rate. (b) Comparison of the theoretical rotation rate between the sphere, oblate spheroids ($r_{pol}/r_{eq}=0.2$) and prolate spheroids ($r_{pol}/r_{eq}=10$). Horizontal dashed blue curve indicates the theoretical profile for the cylinder (Wang 1970; Sauret et al.2012).

Figure 5

Figure 6.Rotation rate of the mean zonal flow, as a function of the cylindrical radius $s$, for a rotating sphere subject to weak latitudinal librations. (a) Comparison between theory (black dashed curves) and DNS (coloured solid curves) at $E=10^{-7}$ and $\epsilon =10^{-4}$ in the regime $\omega \geq 2$. (b) Theoretical zonal flows when $\omega <2$ with divergent flows at the critical latitudes.

Figure 6

Figure 7.Rotation rate of the mean zonal flow, as a function of the cylindrical radius $s$, for a precessing sphere. (a) Comparison of theoretical profiles between our theory and Busse's predictions (black squares, extracted from figure 1 in Busse 1968b). (b) Comparison between our theory (black dashed curve) and DNS (solid curves) for different values of $P_o$ with $\gamma =P_o/E^{1/2}$.

Figure 7

Figure 8.Mean zonal flows driven by precession in spheres. (a) Value at $s=0$ of $f$, as a function of $\gamma = P_o/E^{1/2}$ in DNS (symbols). Dashed blue line shows a slope $-0.15$. (b) Rotation rate of the mean zonal flow, as a function of the cylindrical radius $s$. Comparison between theory (black dashed curve) and DNS (solid coloured curves) choosing values of $\gamma$ such that $f(0)=0$.

Figure 8

Figure 9.Rotation rate of the mean zonal flows driven by the tidal-like forcing of Suess (1971), obtained with $\varOmega _c=\omega =0$ and $m=2$ in expression (2.4). (a) Thick solid red curve: present theory. Solid orange curve: experimental results of Suess (1971). Dashed blue curve: erroneous theory of Suess (1971). Inset shows the azimuthal velocity $sf$. (b) Various contributions to $f$ (solid red curve), made of the sum of two contributions: the one due to $\boldsymbol {u}_B$ (dotted curve), which was mistakenly considered to be $f$ by Suess (1971), and the one due to $\boldsymbol {u}_C$ (dashed curve).

Figure 9

Figure 10.Geostrophic velocity $sf$ (or angular velocity $f$ in the insets) of the mean zonal flow, as a function of the cylindrical radius $s$, for a sphere subject to (a) latitudinal librations at $\omega =0.1$ and (b) longitudinal librations for two $\omega$. Comparison between theory (black dashed curves) and DNS (solid coloured curves) at $\epsilon =10^{-4}$. The curve at $E=10^{-7}$ is indistinguishable from the one obtained with $\epsilon =10^{-5}$ (not shown). In (b), DNS have been performed at $E=10^{-7}$, and we have checked that the numerical profile $f$ at $\omega =1$ is unchanged for $\epsilon =10^{-3}$ and $\epsilon =10^{-6}$ (not shown). Insets show that $f(0)$ remains finite in DNS.

Figure 10

Figure 11.Mean zonal flows in rotating spheroids subject to longitudinal librations in the presence of critical latitudes ($\omega ={\rm \pi} /2 \leq 2$). (a) Comparison of the mean zonal velocity $sf(s)$ given by the theory (black dotted curve for the sphere, black dashed curve for $r_{pol}/r_{eq}=0.8$) and DNS (coloured solid curves) at $E=10^{-5}$ and $\epsilon =10^{-4}$. (b) Comparison of the theoretical rotation rate between the sphere, an oblate ($r_{pol}/r_{eq}=0.2$) and prolate ($r_{pol}/r_{eq}=4$) spheroid.

Figure 11

Figure 12.Geostrophic shear associated with the critical latitudes of longitudinal librations with $\omega \leq 2$. (a) Comparison between theory and DNS (Lin & Noir 2020) in a spherical shell for libration-driven zonal flows computed with $\epsilon =10^{-2}$. The grey area indicates the tangential cylinder $s\leq 0.1$ associated with the inner core in Lin & Noir (2020). (b) Schematic regime diagram for the evolution of $\delta u_g$ and $\delta s$, as a function of $\omega$ and $E$. Pink area indicates the regime dominated by the scalings of the Ekman boundary layer. In the blue area the dominant scalings are given by (5.1a,b). Inset illustrates the geostrophic shear centred on $s_c$ in a meridional section, where the blue area represents the Ekman boundary layer of thickness $E^{1/2}$.

Figure 12

Figure 13.Geostrophic shear associated with the critical latitudes of longitudinal librations (empty points) and latitudinal librations (filled points) in a sphere. The symbols $\omega =1$ performed for longitudinal librations have been extracted from figure 10 in Lin & Noir (2020). The black cross indicates our DNS of a full sphere at $\omega =1$ and $\epsilon =10^{-4}$, which agrees with Lin & Noir (2020). Distance $\delta s$ between peaks in (a), and peak-to-peak amplitude $\delta u_g/\epsilon ^2$ in (b), as a function of $E$.

Figure 13

Figure 14.Normalised mean zonal velocity $\langle \bar {V}_\phi \rangle /\epsilon ^2$ at various dimensionless heights $z=\sqrt {1-s^2}$ in homoeoidal shell geometries (with aspect ratios $\eta _{pol}=\eta _{eq}=0.5$) for libration forcings. The DNS profiles are computed at $z=z_0$ for $s \leq \eta _{eq}$, and at $z_0 = 0$ for $s > \eta _{eq}$, giving a single profile for each DNS. Grey area shows the tangent cylinder $s \leq \eta _{eq}$. Forcings with $\omega _{in}=\omega$ and $\epsilon _{in}=\epsilon$ on the no-slip inner boundary, which is subject to the same forcing as the outer boundary (except for the stress-free case, labelled SF). Black dashed curves indicate the theoretical profiles in the shell geometry. (a) Spherical shell $(r_{pol}/r_{eq}=1)$ with a no-slip (cyan and orange solid curves) or a stress-free (red and pink solid curves) inner boundary. Blued dotted dashed curves illustrate the full-sphere analytical profiles. DNS performed at $E=10^{-6}$, $\omega =3$ and $\epsilon =10^{-6}$ for the two forcings. (b) Homoeoidal (i.e. $\eta _{pol}=\eta _{eq}$) spheroidal shells $(r_{pol}/r_{eq}\neq 1)$ subject to longitudinal librations (solid coloured curves). DNS performed at $E=2.5 \times 10^{-6}$, $\omega ={\rm \pi}$ and $\epsilon =5 \times 10^{-4}$.

Figure 14

Figure 15.Normalised geostrophic mean velocity $\langle \bar {V}_\phi \rangle /\epsilon ^2$ at various heights $z=\sqrt {1-s^2}$ in shells with $\eta _{eq}=0.5$. Black dashed curves indicate the theory in the shell. The DNS profiles computed at $z=z_0$ for $s \leq \eta _{eq}$ and at $z_0 = 0$ for $s > \eta _{eq}$, giving a single profile for each DNS. Grey area shows the tangent cylinder $s \leq \eta _{eq}$. (a) Non-homoeoidal shells subject to longitudinal librations (solid coloured curves), with a spherical inner boundary (i.e. $r_{pol}^{in}=r_{eq}^{in}$) and a spheroidal outer boundary (i.e. $r_{pol} \neq r_{eq}$). DNS performed at $E=2.5 \times 10^{-6}$, $\omega =\omega _{in}={\rm \pi}$, $\epsilon =\epsilon _{in}=5 \times 10^{-4}$. (b) Homoeoidal shells with $\eta _{pol}=\eta _{eq}=0.5$ and different inner and outer forcings. Case 1: DNS at $E=10^{-6}$ in a spherical shell subject to latitudinal librations at outer boundary (with $\omega =3$ and $\epsilon =10^{-6}$), and to longitudinal librations at inner boundary (with $\omega _{in}=2$, $\epsilon _{in}/\epsilon =2$). Case 2: Spheroidal shell with $r_{pol}/r_{eq}=0.8$ subject to longitudinal librations. DNS at $E=5 \times 10^{-6}$ with $\epsilon =5 \times 10^{-4}$ and $\omega ={\rm \pi}$ at outer boundary, and with $\epsilon _{in}/\epsilon =\omega _{in}/\omega =2$ at inner boundary.

Figure 15

Figure 16.Normalised geostrophic mean velocity in a spherical shell ($\eta _{pol}=\eta _{eq}=0.35$) subject to longitudinal librations at $\omega =\sqrt {2}$. In panel (a), $\epsilon _{in}=\sqrt {2}/100$ and $\epsilon =0$, whereas, in panel (b), $\epsilon _{in}=0$ and $\epsilon =\sqrt {2}/100$. Legend in panel (b) is identical to the one in panel (a). Coloured curves have been reproduced from figures 13(a) and 17(a) in Lin & Noir (2020), where $\langle \bar {V}_\phi \rangle$ is computed as in formula (2.8), with $z_{max}=1$, but only considering only the $m=0$ component. Theory is given by formula (5.5). Grey areas indicate the tangent cylinder $s \leq 0.35$.

Figure 16

Figure 17.Schematic regime diagrams for the existence of mean zonal flows driven by mechanical forcings (planetary values estimated from Noir et al.2009; Cébron et al.2012; Lin et al.2015; Vantieghem et al.2015). Red circles: precession. Blue squares: longitudinal librations. Black triangles: latitudinal librations. Magenta stars: tides. (a) Transition between laminar and turbulent boundary layers (BL). Transition occurs when $\epsilon \sim K E^{1/2}$ (with the typical values $K = 20-150$ shown by the grey area, $K = 55$ by the dashed dotted line). (b) Competition between bulk- and boundary-driven generation of mean zonal flows. Input Rossby number $Ro = {O}(\beta )$ for tidal forcing or $Ro = {O}(\epsilon \beta )$ for precession and libration forcings, where $\beta$ is the typical boundary (equatorial or polar) ellipticity. Instabilities and bulk turbulence (hatched area) onsets when $Ro \sim K E ^{1/2}$ (dashed line with the value $K =10$).

Figure 17

Figure 18.Normalised (instantaneous) rotation rate $V_{\phi }/(\epsilon s)$ as a function of time for longitudinal librations of a sphere ($E=10^{-6}$ and $\epsilon =10^{-4}$), at position $(s=0.9,q_2={\rm \pi} /2)$ in the equatorial plane. DNS are given by the solid red lines and the theory $V_{\phi }/(\epsilon s)=\Im _m ( f_0 \, \mathrm {e}^{\mathrm {i} \omega t} )$, where $f_0$ is given by (B15a), is shown as dotted blue and dashed green lines, when calculating $\mathcal {S}$ with (B15b) or with its approximation for $\omega =0$, respectively. (a) $\omega = 10^{-2}$ and (b) $\omega = 3$.

Figure 18

Figure 19.(a) Same as in figure 18(b) but at $s=0.1$ and showing the solution obtained from (B18) with the black dashed line. (b) Snapshot $\hat {v}_{\phi } = - 2 \partial _z \hat {\varPsi }_0^1/(\mathrm {i} \omega )$ in a meridional section of the theoretical azimuthal flow obtained from (B18). We obtain an excellent agreement with the analogous snapshot from DNS (not shown).

Figure 19

Figure 20.Mean zonal flow driven by longitudinal librations of a sphere, from theory and DNS ($E=10^{-7}$, $\epsilon =10^{-4}$). (a) Geostrophic velocity $sf$ as a function of the cylindrical radius $s$, at the inertial mode eigenfrequency $\omega =\sqrt {12/7}$ (Zhang & Liao 2017). DNS (solid line) does not agree well with the theory, whether ${}_H\boldsymbol {U}_0^1$ is taken into account (dashed line) using (B20) or not (dotted line). (b) Rotation rate $f$ at $s=0.6$, as a function of the libration frequency $\omega$, given by the theory (solid line) assuming ${}_H\boldsymbol {U}_0^1=\boldsymbol {0}$. Dashed lines show the four least damped inertial modes in figure 3 of Aldridge & Toomre (1969), including $\omega =\sqrt {12/7}$.