Taylor Column

Taylor Column is a vertical, two-dimensional column of trapped or closed-streamline fluid that forms directly above a submerged three-dimensional obstacle—such as a laboratory cylinder, an oceanic seamount, or a planetary topographic ridge—in a rapidly rotating, homogeneous (barotropic) fluid. First predicted mathematically by the British mathematician Joseph Proudman in 1916 and demonstrated experimentally in rotating tank studies by Geoffrey Ingram Taylor in 1917 and 1923, the phenomenon is the classic physical manifestation of the Taylor–Proudman theorem, which dictates that slow, steady, inviscid flows in a rapidly rotating frame have zero velocity gradient parallel to the rotation axis: (Ω · ∇)u = 0. When the Rossby number (Ro = U / 2ΩL) and the Ekman number (Ek = ν / 2ΩH²) are both much smaller than unity, fluid columns cannot compress or stretch vertically over a submerged bump of height h₀ without violating geostrophic and hydrostatic balance. Instead, the approaching horizontal flow splits and deflects laterally around an imaginary vertical cylinder extending from the obstacle up to the upper boundary of the fluid, exactly as if the short bottom obstacle were a solid pillar spanning the entire fluid depth H. In stratified geophysical fluids (the ocean and planetary atmospheres), baroclinic stratification modifies the classical depth-independent column into a vertically decaying Taylor cone (or Taylor cap) trapped over seamounts and submarine banks.

Derivation in a Rapidly Rotating Frame

Consider an incompressible, homogeneous fluid of uniform density ρ₀ and kinematic viscosity ν rotating at constant angular velocity Ω = Ωk about the vertical z-axis. In the rotating reference frame, the momentum and mass continuity equations are:

∂u / ∂t + (u · ∇)u + 2Ω × u = − (1 / ρ₀) ∇P + ν ∇²u , ∇ · u = 0

where P = p + ρ₀gz − (1/2)ρ₀|Ω × r|² is the reduced pressure absorbing both gravitational hydrostatic pressure and the centrifugal potential. Non-dimensionalizing horizontal lengths by the obstacle radius L, vertical lengths by fluid depth H, velocities by the upstream speed U, time by the advective scale L/U, and reduced pressure by the geostrophic scale 2ρ₀Ω*UL* yields the dimensionless momentum equation:

Ro [ ∂u^* / ∂t + (u · ∇)u ] + k × u = − ∇P^* + Ek (∇)²u

where the two governing dimensionless parameters are the Rossby number (Ro, measuring inertial advection relative to Coriolis acceleration) and the Ekman number (Ek, measuring viscous friction relative to Coriolis acceleration):

Ro = U / (2ΩL) = U / (fL) , Ek = ν / (2ΩH²) = ν / (fH²)

In the asymptotic limit of rapid rotation, steady slow motion, and negligible friction (Ro ≪ 1 and Ek ≪ 1), the inertial and viscous terms vanish to leading order, leaving three-dimensional geostrophic and hydrostatic balance:

2Ω × u = − (1 / ρ₀) ∇P

Taking the curl of both sides eliminate the curl-free pressure gradient (∇ × ∇P = 0):

∇ × (2Ω × u) = 2Ω(∇ · u) − 2(Ω · ∇)u = 0

Because the fluid is incompressible (∇ · u = 0) and Ω = Ωk is constant and aligned with the z-axis, this vector identity simplifies to the Taylor–Proudman theorem:

2Ω ( ∂u / ∂z ) = 0 ⇒ ∂u / ∂z = 0 , ∂v / ∂z = 0 , ∂w / ∂z = 0

Vertical Rigidity and the Taylor–Proudman Paradox

The three component equations ∂u/∂z = ∂v/∂z = ∂w/∂z = 0 carry profound kinematic consequences for flow over bottom topography z = h(x, y) in a container with a flat rigid upper lid at z = H:

1. Two-Dimensional Horizontal Motion: Because ∂u/∂z = 0 and ∂v/∂z = 0, the horizontal velocity vector u_h(x, y) is identical at every depth z; vertically aligned parcels move together as rigid Taylor ink walls or vertical vortex filaments.
2. Vanishing Vertical Velocity (w = 0): Because ∂w/∂z = 0 throughout the interior and the impermeable upper lid enforces w(x, y, H) = 0, the vertical velocity must vanish everywhere in the interior: w(x, y, z) = 0, which simultaneously forces horizontal non-divergence ∂u/∂x + ∂v/∂y = −∂w/∂z = 0.
3. Topographic Steering Along f/H Contours: At the sloping lower boundary z = h(x, y), the kinematic boundary condition requires w(x, y, h) = u_h · ∇h. Since w = 0, the leading-order interior flow must satisfy u_h · ∇h = 0: fluid parcels cannot climb up or down a topographic slope at O(1), and streamlines must follow contours of constant total depth H − h(x, y).

Around an isolated circular bump or seamount, the contours of constant depth H − h(r) are closed circles. Therefore, an approaching uniform upstream flow cannot cross the closed topographic contours at leading order, giving rise to a stagnant, isolated cylinder of fluid above the obstacle—the Taylor column.

Potential Vorticity Conservation and Threshold Topographic Height

In a realistic flow where the Rossby number Ro = U/(fL) is small but finite (Ro ≈ 0.01 to 0.2), fluid columns can undergo small O(Ro) vertical compression as they pass over a gentle topographic elevation of maximum height h₀ ≪ H, provided the fractional depth change h₀ / H does not exceed a threshold value relative to Ro. Under quasigeostrophic theory on an f-plane, every fluid column conserves its barotropic potential vorticity q along streamlines ψ(x, y):

q = ( ζ + f ) / ( H − h(x, y) ) ≈ ( f / H ) [ 1 + ( ζ / f ) + ( h(x, y) / H ) ] = constant along ψ

where ζ = ∂v/∂x − ∂u/∂y = ∇²ψ is the relative vorticity (scaled by U / L, so |ζ / f| ~ Ro). Far upstream where h = 0 and the flow is uniform (u = U, v = 0, ζ = 0), every column has background potential vorticity q_∞ = f / H. When a column is advected over an obstacle of height h(x, y) > 0, vortex-tube compression demands that it acquire anticyclonic (negative) relative vorticity ζ ≈ −f (h / H) < 0 (clockwise rotation in the Northern Hemisphere).

Integrating the induced anticyclonic circulation around a right circular cylinder of radius L and height h₀ in an inviscid flow yields the Ingersoll (1969) and Huppert (1975) streamfunction combining the uniform background flow −Uy and the topographically induced anticyclonic vortex:

ψ(r, θ) = − U r sin θ + ψ_topo(r)

Closed streamlines (representing a trapped Taylor column with a stagnation point where the anticyclonic vortex velocity cancels the upstream flow U) appear above the obstacle if and only if the dimensionless topographic Hide parameter B_T = (h₀ / H) / Ro exceeds a geometry-dependent threshold B_crit:

B_T = ( h₀ / H ) / Ro = ( f L h₀ ) / ( U H ) ≥ B_crit

• For a right circular cylinder with vertical side walls (the geometry of Taylor's 1923 tank experiment), Andrew P. Ingersoll (1969) proved that closed streamlines form for any B_T > B_crit = 2.
• For a smooth axisymmetrical conical or Gaussian seamount, Herbert E. Huppert (1975) showed that B_crit ranges from 1.3 to 2.0 depending on the radial slope profile. When B_T < B_crit, streamlines deflect anticyclonically over the bump without closing; when B_T > B_crit, a closed, recirculating Taylor column forms on the anticyclonic flank of the obstacle and grows to encompass the entire obstacle summit at large B_T.

Quantitative Comparison of Classical, Stratified, and Planetary Taylor Columns

In the real ocean, vertical density stratification characterized by the buoyancy (Brunt–Väisälä) frequency N = [−(g/ρ₀)(∂ρ/∂z)]^1/2 decouples deep flow from surface layers through thermal wind shear:

f ( ∂u_g / ∂z ) = + ( g / ρ₀ ) ( ∂ρ / ∂y ) , f ( ∂v_g / ∂z ) = − ( g / ρ₀ ) ( ∂ρ / ∂x )

As demonstrated by Nelson G. Hogg (1973) and Huppert (1975), the three-dimensional quasigeostrophic potential vorticity equation in a fluid of constant N and f is Laplace's equation in stretched vertical coordinates:

∂²ψ / ∂x² + ∂²ψ / ∂y² + ( f² / N² ) ∂²ψ / ∂z² = 0

Consequently, a topographic disturbance of horizontal radius L at the seafloor (z = 0) does not extend unchanged to the sea surface when the Burger number Bu = (NH / fL)² is greater than unity. Instead, the anticyclonic vorticity and closed streamlines decay vertically with z as e^−z/H_R over the Prandtl–Rossby deformation height:

H_R = ( f L ) / N = H / Bu^1/2

For a midlatitude oceanic seamount of radius L = 25 km, f = 10⁻⁴ s⁻¹, and N = 2.5 × 10⁻³ s⁻¹, the vertical trapping scale is H_R ≈ 1,000 m. In a 4,000-meter-deep ocean, the closed streamlines form a submerged Taylor cone (or Taylor cap) extending ~1 km above the seamount summit without reaching the sea surface. By thermal wind balance, the anticyclonic circulation above the seamount requires dense, cold water to dome upward over the summit (isopycnal uplifting), creating a localized cold-core dome that traps nutrient-rich water, phytoplankton, and pelagic fish assemblages over seamounts such as Fieberling Guyot in the eastern North Pacific and Rockall Bank in the North Atlantic.

How did G. I. Taylor demonstrate the Taylor column in the laboratory?

In his landmark 1923 paper presented to the Royal Society of London, Geoffrey Ingram Taylor filled a rotating cylindrical glass tank with water, spun it at uniform angular velocity until the fluid reached solid-body rotation, and slowly towed a short solid wax cylinder (occupying only the bottom quarter of the tank depth) across the flat floor. By releasing streaks of colored dye ahead of the cylinder, Taylor showed that water in the upper three-quarters of the tank—well above the top of the physical obstacle—parted and flowed around an invisible vertical cylinder situated directly over the moving wax bump, while fluid inside the vertical cylinder was carried intact across the tank with the obstacle.

Why does a Taylor column have anticyclonic rotation?

By conservation of potential vorticity q = (ζ + f) / h_col, any fluid parcel pushed onto the summit of a submerged obstacle experiences a decrease in total water-column height h_col = H − h(x, y). To keep the ratio (ζ + f) / h_col constant as h_col shrinks, the absolute vorticity (ζ + f) must decrease, which requires the relative vorticity ζ to become negative when f > 0 (or positive when f < 0)—corresponding in both hemispheres to anticyclonic rotation (clockwise in the Northern Hemisphere, counterclockwise in the Southern Hemisphere) over the topographic high, while the displaced vortex tube originally sitting atop the bump is shed downstream as a cyclonic eddy during spin-up.

Is Jupiter's Great Red Spot a Taylor column?

In 1961, the British geophysicist Raymond Hide proposed that Jupiter's Great Red Spot might be a planetary-scale Taylor column anchored above a permanent topographic mountain or magnetic anomaly in Jupiter's deep interior. However, subsequent spacecraft gravity measurements (Pioneer, Voyager, Galileo, and Juno) and numerical fluid dynamics showed that Jupiter is a fluid hydrogen–helium planet lacking a solid surface at the depth of its weather layer, and that the Great Red Spot drifts zonally in longitude relative to Jupiter's internal magnetic rotation period (System III); modern atmospheric dynamics classifies the Great Red Spot not as a topographically locked Taylor column, but as a long-lived, self-sustaining baroclinic/quasigeostrophic anticyclonic vortex embedded between opposing zonal jets.

1. Proudman, J. (1916). "On the motion of solids in a liquid possessing vorticity." Proceedings of the Royal Society of London. Series A, 92(642), 408–424. https://doi.org/10.1098/rspa.1916.0026
2. Taylor, G. I. (1923). "Experiments on the motion of solid bodies in rotating fluids." Proceedings of the Royal Society of London. Series A, 104(725), 213–218. https://doi.org/10.1098/rspa.1923.0103
3. Ingersoll, A. P. (1969). "Inviscid fluid flow over topography in a rotating frame." Journal of the Atmospheric Sciences, 26(6), 1191–1198. https://doi.org/10.1175/1520-0469(1969)026<1191:IFFOTI>2.0.CO;2026<1191:IFFOTI>2.0.CO;2)
4. Hogg, N. G. (1973). "On the stratified Taylor column." Journal of Fluid Mechanics, 58(3), 517–537. https://doi.org/10.1017/S0022112073002302
5. Huppert, H. E. (1975). "Some remarks on the initiation of inertial Taylor columns." Journal of Fluid Mechanics, 67(2), 397–412. https://doi.org/10.1017/S0022112075000377