Advective Thermocline Scaling

Advective Thermocline Scaling (anchored by the Welander M-equation and the Robinson–Stommel advective–diffusive scaling laws) is the theoretical framework in physical oceanography and Earth system science that explains how wind-driven Ekman pumping, surface buoyancy forcing, and turbulent diapycnal mixing determine the depth, vertical density gradient, and three-dimensional circulation of the permanent oceanic thermocline. Formulated simultaneously in 1959 by the Swedish oceanographer Pierre Welander and the American oceanographers Allan R. Robinson and Henry Stommel (Tellus), the theory resolved a central limitation of classical barotropic gyre models such as the Stommel gyre model: why the wind-driven subtropical circulation does not penetrate uniformly to the abyssal seafloor, but is instead concentrated within an upper warm-water lens of characteristic thickness D ≈ 400–1,000 m separated from the cold abyss by a sharp pycnocline. By combining the planetary geostrophic momentum balance, hydrostatic equilibrium, incompressibility, and steady density conservation into a single nonlinear partial differential equation for the vertically integrated Montgomery potential M, Welander (1959) showed that in the adiabatic limit (zero vertical eddy diffusivity, κ_v → 0), downward Ekman pumping w_E and lateral geostrophic advection establish an advective thermocline depth scale D_adv = (f w_E L² / g′)^1/2.

Between latitudes 15° and 45°, ocean profiles beneath the seasonal mixed layer (50–150 m) exhibit a main (permanent) thermocline across which temperature drops from 18–26 °C to 4–5 °C at 800–1,000 m depth (potential density increase Δρ ≈ 2–3 kg m⁻³), overlying a cold abyssal layer renewed by the thermohaline circulation. Two mechanisms were proposed in the 1950s to explain how this stratification is maintained:

1. Diffusive Regime (Robinson and Stommel, 1959; Munk, 1966): Slow abyssal upwelling (w > 0) balances downward turbulent heat diffusion (w ∂T/∂z ≈ κ_v ∂²T/∂z²), yielding scale depth D = κ_v/w in the deep abyss. However, in subtropical gyres, wind-stress curl drives downward Ekman pumping (w_E < 0), pushing warm surface water down rather than drawing cold water up.
2. Adiabatic Advective Regime (Welander, 1959): Warm water subducted out of the subtropical Ekman layer slides along sloping constant-density (isopycnal) surfaces while conserving density ρ and planetary potential vorticity q = −(f / ρ₀) ∂ρ/∂z, shaping the subtropical thermocline bowl even as κ_v → 0.

The Planetary Geostrophic Equations on a Sphere

At basin scales L ∼ 1,000–5,000 km exceeding the internal Rossby deformation radius L_d = (g′D)^1/2/f ≈ 30–50 km, the Rossby number Ro = U/(fL) ∼ 10⁻³ is small. Under the Boussinesq and hydrostatic approximations in spherical coordinates (longitude λ, latitude φ, height z), the steady planetary geostrophic equations and density conservation equation are:

− f v = − (1 / (ρ₀ R_E cos φ)) ∂p / ∂λ , + f u = − (1 / (ρ₀ R_E)) ∂p / ∂φ , 0 = − ∂p / ∂z − g ρ

(1 / (R_E cos φ)) [ ∂u / ∂λ + ∂(v cos φ) / ∂φ ] + ∂w / ∂z = 0

( u / (R_E cos φ) ) ∂ρ / ∂λ + ( v / R_E ) ∂ρ / ∂φ + w ∂ρ / ∂z = κ_v ∂²ρ / ∂z²

where f = 2Ω sin φ is the Coriolis parameter, R_E is Earth's radius, p is dynamic pressure perturbation, ρ is density anomaly relative to ρ₀, and κ_v is vertical turbulent diffusivity. Combining geostrophy, hydrostatics, and continuity yields the thermal wind relations and the planetary vorticity relation βv = f ∂w/∂z.

Reduction to Welander's Single Scalar Function M(λ, φ, z)

Welander (1959) showed that all five fields (u, v, w, p, ρ) derive from a single scalar potential M(λ, φ, z), defined by integrating dynamic pressure from a deep level of no motion (z → −∞) to z:

M(λ, φ, z) = ∫_−∞^z p(λ, φ, z′) dz′ + M₀(λ, φ)

Through geostrophy, hydrostatics, and f ∂w/∂z = βv, each physical variable is a derivative of M:

• Dynamic pressure and density: p = ∂M/∂z , ρ = −(1/g) ∂²M/∂z²
• Horizontal velocities: u = −(1/(2Ωρ₀R_E sin φ)) ∂²M/(∂φ∂z) , v = +(1/(2Ωρ₀R_E sin φ cos φ)) ∂²M/(∂λ∂z)
• Vertical velocity: w = +(1/(2Ωρ₀R_E² sin²φ)) ∂M/∂λ

Substituting these into the steady density equation yields the Welander–Robinson–Stommel M-equation:

( ∂M / ∂λ ) ( ∂³M / ∂z³ ) − ( ∂²M / ∂λ∂z ) ( ∂²M / ∂φ∂z ) + cot(φ) ( ∂M / ∂λ ) ( ∂³M / ∂z³ ) = ν_diff ∂⁴M / ∂z⁴

In the adiabatic Welander limit (κ_v = 0), the fourth-order diffusive term vanishes, leaving a third-order nonlinear partial differential equation expressing conservation of density ρ ∼ −M_zz and potential vorticity q ∼ fM_zzz along three-dimensional geostrophic streamlines.

Let L denote the horizontal basin scale (~5,000 km), f₀ ≈ 10⁻⁴ s⁻¹ the midlatitude Coriolis parameter, β ≈ f₀/R_E ≈ 2×10⁻¹¹ m⁻¹ s⁻¹, Δρ the surface-to-abyss density contrast, g′ = g Δρ/ρ₀ ≈ 0.02 m s⁻² the reduced gravity, and w_E ≈ 10⁻⁶ m s⁻¹ (~30 m yr⁻¹) the downward Ekman pumping velocity imposed by the wind stress curl at the base of the mixed layer (z = 0).

1. Purely Advective Scaling (Welander, 1959)

From the thermal wind balance, the horizontal geostrophic velocity scale U across a thermocline of thickness D is:

U ∼ ( g Δρ / (ρ₀ f₀) ) ( D / L ) = g′ D / (f₀ L)

From the planetary geostrophic continuity balance (βV ∼ f₀ w_E / D, where f₀ / (βL) ∼ O(1) on the planetary scale), the vertical velocity at the top of the thermocline must match the imposed Ekman pumping w_E:

w_E ∼ ( β L / f₀ ) U ( D / L ) ∼ U D / L

Eliminating U between the thermal-wind relation and the continuity relation gives the Welander advective thermocline depth scale:

D_adv = ( f₀² L² w_E / (β g′ L) )^1/2 ≈ ( f₀ w_E L² / g′ )^1/2

Using f₀ = 10⁻⁴ s⁻¹, w_E = 1.5×10⁻⁶ m s⁻¹, L = 5×10⁶ m, and g′ = 0.02 m s⁻² yields D_adv ≈ 430 m—in close quantitative agreement with the observed 400–700 m depth of the subtropical main thermocline. Notably, D_adv scales with the square root of the wind-stress curl (w_E^1/2) and inversely with the square root of the surface buoyancy contrast ((g′)^−1/2), independent of vertical eddy diffusivity κ_v.

2. Advective–Diffusive Scaling (Robinson and Stommel, 1959; Bryan and Cox, 1967)

If instead the thermocline thickness D is set by an internal balance between vertical advection w ∂ρ/∂z and vertical eddy diffusion κ_v ∂²ρ/∂z² (so w ∼ κ_v / D), substituting w ∼ κ_v / D into the thermal-wind and planetary continuity relations w ∼ g′ D² / (f₀ L²) yields the Robinson–Stommel diffusive depth scale:

D_diff = ( κ_v f₀ L² / g′ )^1/3

Welander's Self-Similar Solution and the q = F(ρ, B) Integral

Welander (1959) obtained an analytical similarity solution to the adiabatic M-equation of the form:

M(λ, φ, z) = P(λ, φ) e^{k(φ) z / sin φ}

which corresponds to a density profile decaying with depth as e^z/D(φ), where the vertical scale depth D(φ) = sin φ / k(φ) is proportional to the sine of latitude. In 1971, Welander generalized the adiabatic formulation by demonstrating that steady, inviscid, nondiffusive planetary geostrophic flow simultaneously conserves density ρ, the Bernoulli function B = p + ρgz, and planetary potential vorticity q = −(f / ρ₀) ∂ρ/∂z. Because any three scalar fields conserved along steady three-dimensional streamlines must be functionally related, every adiabatic thermocline satisfies:

q = − ( f / ρ₀ ) ∂ρ / ∂z = F(ρ, B)

Link to Luyten–Pedlosky–Stommel (1983) Ventilated Thermocline Theory

While Welander's similarity solutions required special mathematical forms of F(ρ, B), James R. Luyten, Joseph Pedlosky, and Henry Stommel (1983) solved the nonlinear adiabatic thermocline equations for arbitrary surface outcrop latitudes by partitioning the stratified ocean into discrete moving isopycnal layers. In their ventilated thermocline model (LPS) and the parallel Rhines–Young (1982) potential-vorticity homogenization theory:

• Surface water subducted downward by Ekman pumping across winter isopycnal outcrop lines carries its surface potential vorticity southward and westward into the ventilated zone, obeying Welander's advective depth scaling D_adv = (f w_E L² / g′)^1/2.
• Along the eastern boundary lies a stagnant shadow zone isolated from surface ventilation, while on the western side closed geostrophic contours experience mesoscale eddy mixing that homogenizes potential vorticity into a uniform value (Rhines and Young, 1982).
• Beneath the wind-driven ventilated lens (z ≈ −D_adv), Samelson and Vallis (1997) showed that a thin internal diffusive boundary layer of thickness δ_int = (κ_v D_adv / w_E)^1/2 bridges the adiabatic Welander upper thermocline to the diffusive Munk–Robinson–Stommel abyssal stratification below.

Why cannot a 1D upwelling–diffusion balance explain the upper subtropical thermocline?

In the subtropical gyres (roughly 15°–40° latitude), the curl of the trade winds and midlatitude westerlies forces convergent surface Ekman transport and downward Ekman pumping (w_E < 0, typically 20–50 m yr⁻¹ downward). A 1D advective–diffusive balance w ∂T/∂z = κ_v ∂²T/∂z² requires upward velocity (w > 0) to maintain a convex thermocline against downward heat diffusion; in the upper subtropical gyre where w < 0, three-dimensional geostrophic advection of density along sloping isopycnals (Welander's advective balance) is required to close the heat budget.

How does the thermocline depth respond to a doubling of the wind-stress curl?

According to the Welander advective scaling D_adv = (f₀ w_E L² / g′)^1/2, where the Ekman pumping velocity w_E = curl_z(τ) / (ρ₀ f₀) is directly proportional to the wind-stress curl, doubling the wind-stress curl increases the subtropical thermocline depth by a factor of √2 ≈ 1.41 (a 41 percent deepening of the warm-water bowl).

What is the physical meaning of the two-tiered thermocline structure discovered by Samelson and Vallis (1997)?

Microstructure tracer-release experiments in the upper pelagic thermocline measured diapycnal diffusivity around κ_v ≈ 10⁻⁵ m² s⁻¹—small enough that the upper 500–700 m is governed by the adiabatic Welander advective scale D_adv, while a sharp internal boundary layer of thickness δ_int ≈ 80 m at the base of the ventilated bowl matches the wind-driven lens to the deeper abyssal stratification governed by diapycnal upwelling and mixing.

1. Welander, P. (1959). "An advective model of the ocean thermocline." Tellus, 11(3), 309–318. https://doi.org/10.3402/tellusa.v11i3.9316
2. Robinson, A., & Stommel, H. (1959). "The oceanic thermocline and the associated thermohaline circulation." Tellus, 11(3), 295–308. https://doi.org/10.3402/tellusa.v11i3.9317
3. Welander, P. (1971). "Some exact solutions to the equations describing an ideal-fluid thermocline." Journal of Marine Research, 29(2), 60–68. https://elischolar.library.yale.edu/journal_of_marine_research/1202/
4. Luyten, J. R., Pedlosky, J., & Stommel, H. (1983). "The ventilated thermocline." Journal of Physical Oceanography, 13(2), 292–309. https://doi.org/10.1175/1520-0485(1983)013<0292:TVT>2.0.CO;2013<0292:TVT>2.0.CO;2)
5. Samelson, R. M., & Vallis, G. K. (1997). "Large-scale circulation with small diapycnal diffusion: The two-thermocline regime." Journal of Marine Research, 55(2), 223–275. https://doi.org/10.1357/0022240973224382