Stewartson Layer Dynamics in Rotating Fluids

• The Stewartson layer is a thin viscous boundary layer in rapidly rotating fluids that bridges the geostrophic interior flow with rigid no-slip boundaries.
• It exhibits a hierarchy of asymptotic sublayers, including classical E^(1/4) and E^(1/3) scalings, to accommodate velocity adjustments in various geometries.
• Its dynamics critically impact turbulent convection, secondary circulations, and heat transport in both laboratory experiments and geophysical applications.

Searching arXiv for recent and foundational papers on Stewartson layers. Found relevant results; using them to ground the article in published work. A Stewartson layer is a thin viscous boundary layer that forms where a rapidly rotating, nearly geostrophic flow meets a rigid boundary that cannot satisfy the interior balance. In rotating fluids, the Coriolis force and pressure-gradient force enforce an approximately two-dimensional interior through the Taylor–Proudman theorem; at a vertical sidewall, or on the tangent cylinder of spherical Couette flow, that interior state must be matched to no-slip boundary conditions by a shear layer now identified as the Stewartson layer. In cylindrical rotating Rayleigh–Bénard convection it sits on the vertical sidewall and closes Ekman-layer-driven secondary circulation; in spherical shells it forms on the tangent cylinder separating fluid regions constrained by different rotations (Kunnen et al., 2011, Wei et al., 2010).

1. Geostrophic origin and boundary-layer closure

In the rapidly rotating limit, the interior flow is geostrophic to leading order. The Stewartson layer arises because this geostrophic interior cannot satisfy the no-slip condition at a rigid vertical boundary. In rotating Rayleigh–Bénard convection, the same mechanism operates at the cylindrical sidewall: the sidewall layer adjusts the interior azimuthal motion to zero at the wall and carries a weak secondary circulation that closes the flows driven by Ekman layers at the top and bottom plates. At those horizontal plates, the Ekman-layer thickness scales as $\delta_E \sim H\,Ek^{1/2}$, and the associated suction or spin-up/spin-down produces radial outflow in the plate layers; mass continuity then forces return flow through the Stewartson layer (Kunnen et al., 2011).

In spherical Couette flow the corresponding object is attached not to a material sidewall but to the tangent cylinder, the cylindrical surface $s=r_i$ just touching the inner sphere and parallel to the rotation axis. In the limit of strong overall rotation, the Taylor–Proudman constraint again enforces nearly columnar motion. If $\Omega_i \neq \Omega_o$, fluid inside the tangent cylinder tends to rotate with the inner sphere while fluid outside is locked to the outer sphere, and the mismatch produces a thin Stewartson shear layer on that cylinder (Wei et al., 2010).

This shared geostrophic origin is the central unifying feature across geometries. A common misconception is to regard the Stewartson layer as only an azimuthal shear adjustment. The cited studies show that it also carries weak but dynamically consequential meridional or axial flow, either as the return branch of Ekman pumping or, in centrifugal convection, as a buoyancy-driven circulation constrained by rotation (Kunnen et al., 2011, Lai et al., 9 Sep 2025).

2. Classical thickness scalings and nested sublayers

For a cylindrical container of height $H$ rotating at rate $\Omega$ with kinematic viscosity $\nu$, the Ekman number may be written either as $E=\nu/(2\Omega H^2)$ or, in the direct numerical simulations discussed for rotating Rayleigh–Bénard convection, as $Ek=\nu/(\Omega H^2)$. Matched-asymptotic analyses cited in the rotating-convection study identify two characteristic Stewartson-layer thicknesses at a vertical wall: an outer or “classical” layer

$\delta_{s,1/4}\sim H\,Ek^{1/4},$

and an inner Stewartson sublayer

$\delta_{s,1-3}\sim H\,Ek^{1/3}.$

Using $s=r_i$0, these become

$s=r_i$1

The outer layer carries most of the azimuthal-velocity adjustment, whereas the thinner inner sublayer adjusts vertical velocity to satisfy impermeability and supports an internal recirculation (Kunnen et al., 2011).

The local asymptotic structure is likewise explicit. In the $s=r_i$2 region, with $s=r_i$3, the azimuthal velocity obeys

$s=r_i$4

where $s=r_i$5 is the interior azimuthal-velocity deficit. In the thinner $s=r_i$6 sublayer, with $s=r_i$7, matched-asymptotic solutions yield a peaked vertical velocity $s=r_i$8 with zero net flux and a small radial recirculation $s=r_i$9 (Kunnen et al., 2011).

Spherical Couette flow introduces a more elaborate nested structure on the tangent cylinder. One asymptotic description identifies a quasi-geostrophic $\Omega_i \neq \Omega_o$0 layer inside the cylinder, an $\Omega_i \neq \Omega_o$1 layer outside it, and an ageostrophic $\Omega_i \neq \Omega_o$2 shear layer centered on the cylinder; near the equator this structure merges with an $\Omega_i \neq \Omega_o$3 equatorial Ekman layer of axial length $\Omega_i \neq \Omega_o$4 (Marcotte et al., 2016). Another description, focused on the instability problem, uses a primary $\Omega_i \neq \Omega_o$5 layer together with secondary $\Omega_i \neq \Omega_o$6 and $\Omega_i \neq \Omega_o$7 components (Wei et al., 2010). Taken together, these analyses show that the Stewartson layer is not a single uniform slab but a geometry-dependent hierarchy of asymptotic sublayers.

3. Rotating Rayleigh–Bénard convection and sidewall transport

When the classical Rayleigh–Bénard system is rotated about its vertical axis, three regimes are identified in a cylindrical cell of aspect ratio one at $\Omega_i \neq \Omega_o$8–$\Omega_i \neq \Omega_o$9, $H$0–$H$1, and $H$2. In regime I, corresponding to weak rotation, the large-scale circulation is the dominant flow feature. In regime II, under moderate rotation, the large-scale circulation is replaced by vertically aligned vortices. Regime III, under strong rotation, is characterized by suppression of vertical velocity fluctuations. The azimuthal sidewall temperature signal changes accordingly: regime I shows a cosine-shaped azimuthal wall-temperature profile together with a vertical temperature gradient due to plumes travelling close to the sidewall with the large-scale circulation, whereas in regimes II and III the cosine profile disappears even though a vertical wall-temperature gradient remains (Kunnen et al., 2011).

The persistence of that vertical wall-temperature gradient in regimes II and III is a central result. Its origin is different from the large-scale-circulation mechanism of regime I. Instead, it is caused by rotating-flow boundary-layer dynamics that drive a secondary circulation carrying hot fluid upward along the sidewall in the lower part of the container and cold fluid downward along the sidewall in the upper part. The Ekman-to-Stewartson circulation therefore imprints a sidewall temperature signature even when the large-scale circulation is absent (Kunnen et al., 2011).

Direct numerical simulations at $H$3, $H$4, and $H$5 with $H$6 quantify the mean profiles after azimuthal and time averaging. The azimuthal velocity $H$7 is negative in the bulk, corresponding to anticyclonic motion, and reverses to a thin positive cyclonic band within $H$8, consistent with a Stewartson layer of $H$9 thickness. The vertical velocity $\Omega$0 at $\Omega$1 exhibits a sharp upward spike peaking at $\Omega$2, associated with the singular eruption of Ekman-layer suction into the inner Stewartson layer, and this spike is flanked by weaker downward flow in the outer part of the layer. The temperature is elevated above bulk levels within the Stewartson layer because warm fluid is transported upward along the sidewall in the lower half of the cell (Kunnen et al., 2011).

The significance of these results lies in their connection between turbulent convection and laminar asymptotics. The study concludes that even in strongly turbulent, plume-dominated rotating Rayleigh–Bénard convection, the mean secondary circulation is well captured by classical linear Ekman–Stewartson theory, and that the observed sidewall temperature signals provide evidence for the $\Omega$3 and $\Omega$4 sublayers. It also states that these findings provide a basis for refined models, including Grossmann–Lohse extensions that include sidewall-driven secondary flows (Kunnen et al., 2011).

4. Stewartson layers on the tangent cylinder in spherical Couette flow

In spherical Couette flow, the governing equations are conveniently written in the frame rotating with the outer sphere at angular speed $\Omega$5, using the gap width $\Omega$6 as length scale, $\Omega$7 as time scale, and $\Omega$8 as velocity scale. The dimensionless Navier–Stokes equations are

$\Omega$9

with Ekman number $\nu$0 and Rossby number $\nu$1. In the basic state one takes $\nu$2, so the leading balance is $\nu$3. The inner-sphere boundary condition is $\nu$4 in dimensionless form, while the outer sphere satisfies $\nu$5 (Wei et al., 2010).

The thickness scaling of the tangent-cylinder Stewartson layer follows from matched asymptotics. A naive Coriolis–viscous estimate gives $\nu$6, which is the Ekman-layer scaling, but the cylindrical Stewartson layer instead satisfies

$\nu$7

where the numerical constant $\nu$8 depends on geometry and on the precise matched-asymptotic solution. In a spherical shell, that constant must be computed numerically or by more elaborate asymptotics (Wei et al., 2010).

The instability problem is formulated by adding small non-axisymmetric perturbations $\nu$9 and linearizing about the basic Stewartson-layer profile. Writing the Reynolds number as

$E=\nu/(2\Omega H^2)$0

one determines critical values $E=\nu/(2\Omega H^2)$1 for neutral stability. For typical rapid-rotation regimes $E=\nu/(2\Omega H^2)$2, onset occurs at large $E=\nu/(2\Omega H^2)$3. The direction of differential rotation matters. For $E=\nu/(2\Omega H^2)$4, when the inner sphere spins faster in the same sense as $E=\nu/(2\Omega H^2)$5, the instability is a Kelvin–Helmholtz-type shear-layer mode and the most unstable azimuthal wavenumber $E=\nu/(2\Omega H^2)$6 increases as $E=\nu/(2\Omega H^2)$7 decreases. For $E=\nu/(2\Omega H^2)$8, when the inner sphere spins opposite to $E=\nu/(2\Omega H^2)$9, the $Ek=\nu/(\Omega H^2)$0 symmetry is broken by the Coriolis term, and one obtains an anomalous $Ek=\nu/(\Omega H^2)$1 mode concentrated near the intersection of the inner-sphere Ekman layer and the Stewartson layer; the corresponding critical $Ek=\nu/(\Omega H^2)$2 is substantially larger, indicating greater stability (Wei et al., 2010).

These features explain why Stewartson layers are relevant to both laboratory and geophysical systems. The spherical-Couette study explicitly identifies planetary cores, gaseous-giant interiors, and rotating spherical-shell experiments as settings in which a Stewartson-type shear layer on the tangent cylinder can form and where its non-axisymmetric instabilities can influence the dynamics (Wei et al., 2010).

5. Equatorial Ekman layer and the full asymptotic hierarchy

A more complete asymptotic account of the spherical Stewartson problem resolves how the tangent-cylinder layer interacts with the sphere Ekman layers near the equator. In axisymmetry one introduces a streamfunction $Ek=\nu/(\Omega H^2)$3 and angular velocity $Ek=\nu/(\Omega H^2)$4 through

$Ek=\nu/(\Omega H^2)$5

and the linearized steady equations in the rotating frame become

$Ek=\nu/(\Omega H^2)$6

Outside thin layers the mainstream is geostrophic, but near the tangent cylinder $Ek=\nu/(\Omega H^2)$7 the Proudman solution fails and a nested hierarchy is required (Marcotte et al., 2016).

That hierarchy comprises several distinct layers. First are the $Ek=\nu/(\Omega H^2)$8 Ekman layers on the inner and outer spheres. Second are the quasi-geostrophic shear layers that smooth the jump in $Ek=\nu/(\Omega H^2)$9 across the tangent cylinder: an inner layer of radial width $\delta_{s,1/4}\sim H\,Ek^{1/4},$0 and an outer layer of width $\delta_{s,1/4}\sim H\,Ek^{1/4},$1. Third is an ageostrophic $\delta_{s,1/4}\sim H\,Ek^{1/4},$2 shear layer centered on $\delta_{s,1/4}\sim H\,Ek^{1/4},$3, with radial coordinate $\delta_{s,1/4}\sim H\,Ek^{1/4},$4 and $\delta_{s,1/4}\sim H\,Ek^{1/4},$5. Its leading equations are

$\delta_{s,1/4}\sim H\,Ek^{1/4},$6

Finally, very near the point $\delta_{s,1/4}\sim H\,Ek^{1/4},$7, the ageostrophic core merges with the inner-sphere Ekman layer to form an equatorial Ekman layer with radial width $\delta_{s,1/4}\sim H\,Ek^{1/4},$8 and axial length $\delta_{s,1/4}\sim H\,Ek^{1/4},$9 (Marcotte et al., 2016).

Under the inner scaling

$\delta_{s,1-3}\sim H\,Ek^{1/3}.$0

the local equatorial problem is independent of $\delta_{s,1-3}\sim H\,Ek^{1/3}.$1 to leading order. The reduced equations are

$\delta_{s,1-3}\sim H\,Ek^{1/3}.$2

with mixed boundary conditions on the equatorial plane distinguishing the inner-sphere side from the outer side. To represent far-field decay in a finite computational domain, the numerical solution employs a non-local integral boundary condition at truncated height $\delta_{s,1-3}\sim H\,Ek^{1/3}.$3,

$\delta_{s,1-3}\sim H\,Ek^{1/3}.$4

obtained from a Fourier representation of the decay mode (Marcotte et al., 2016).

The asymptotic and numerical analysis extends Stewartson’s similarity solution for the $\delta_{s,1-3}\sim H\,Ek^{1/3}.$5 shear layer and shows that the combined hierarchy resolves the Proudman rigid-rotation paradox, satisfies the global angular-momentum budget, and explains how the tangent-cylinder shear layer is connected to the equatorial singular structure (Marcotte et al., 2016).

6. Gravity-driven Stewartson flow in centrifugal convection

A recent extension of the subject considers centrifugal convection in a closed annular container with inner radius $\delta_{s,1-3}\sim H\,Ek^{1/3}.$6, outer radius $\delta_{s,1-3}\sim H\,Ek^{1/3}.$7, and height $\delta_{s,1-3}\sim H\,Ek^{1/3}.$8, rotating at angular speed $\delta_{s,1-3}\sim H\,Ek^{1/3}.$9. The inner cylinder is held at $s=r_i$00, the outer cylinder at $s=r_i$01, gravity acts downward, centrifugal acceleration acts radially outward, all boundaries are no-slip, the top and bottom are adiabatic, and the sidewalls are isothermal. Using the gap width $s=r_i$02 and free-fall speed $s=r_i$03 with $s=r_i$04 and $s=r_i$05, the study defines

$s=r_i$06

with $s=r_i$07 and aspect ratio $s=r_i$08 (Lai et al., 9 Sep 2025).

The Stewartson layer in this system is identified from the mean vertical velocity $s=r_i$09 on an $s=r_i$10–$s=r_i$11 section. For weak gravity, $s=r_i$12, the mean flow consists of four quadrupolar vortices driven by boundary friction and Ekman pumping, with no Stewartson layer. Beyond a critical gravitational forcing, $s=r_i$13, corresponding to inverse Froude number $s=r_i$14 of order $s=r_i$15–$s=r_i$16, two elongated vortices appear near the inner and outer walls. At still larger $s=r_i$17, these sharpen and merge into classical Stewartson layers with weak downward jets near the inner wall and weak upward jets near the outer wall, of magnitude $s=r_i$18–$s=r_i$19 in nondimensional units (Lai et al., 9 Sep 2025).

The thickness scaling remains rotational: $s=r_i$20 The study derives this from balancing Coriolis and viscous stresses in the sidewall layer. Its distinctive result concerns the internal circulation strength. In centrifugal convection, the vertical velocity inside the Stewartson layer is argued to be driven by gravitational buoyancy rather than Ekman pumping alone. Assuming a steady viscous–buoyancy balance in the vertical momentum equation and using the radial variation across the layer, the analysis yields

$s=r_i$21

for the $s=r_i$22 case. Numerical tests varying $s=r_i$23 from $s=r_i$24 down to $s=r_i$25 and $s=r_i$26 from $s=r_i$27 to $s=r_i$28 show that $s=r_i$29 collapses onto $s=r_i$30 and that $s=r_i$31 follows an $s=r_i$32 power law in the centrifugal-dominated regime. When gravity dominates, $s=r_i$33, deviations appear and the flow approaches the vertical-convection scaling $s=r_i$34 instead of the viscous–buoyancy balance (Lai et al., 9 Sep 2025).

This suggests that the Stewartson layer should be understood not only as a rotational matching layer but also, in some settings, as a site where weak buoyancy-driven axial transport is organized and constrained by geostrophic balance. The same study proposes implications for heat transport, mixing, rotating heat exchangers, centrifuges, and rotating annular channels in geophysical contexts (Lai et al., 9 Sep 2025).