The Ekman Spiral: How Wind Twists the Ocean Into a Corkscrew

The force balance: Coriolis against the wind's grip

Start with the horizontal momentum equations for a steady, homogeneous ocean far from the equator, keeping only the two forces that matter in the surface layer: the Coriolis force and the vertical turbulent friction that carries the wind's pull downward. Writing the horizontal velocity as (u, v), the balance is

• −f v = ν ∂²u/∂z²
• +f u = ν ∂²v/∂z²

Here f = 2Ω sin φ is the Coriolis parameter — Ω = 7.29×10⁻⁵ s⁻¹ is Earth's rotation rate and φ the latitude, giving f ≈ 1.0×10⁻⁴ s⁻¹ at 45° — and ν is the vertical eddy (turbulent) viscosity, typically 10⁻² to 10⁻¹ m²/s in the mixed layer, vastly larger than the molecular value 10⁻⁶ m²/s. The pressure-gradient and advective terms have been dropped: this is the pure Ekman problem, valid when the Ekman number Ek = ν/(f·D²) is order one over the layer thickness but the Rossby number Ro = U/(fL) ≪ 1 so nonlinear terms are negligible.

The elegance comes from combining the two into one complex equation. Define w̃ = u + iv. Then f u + i(−f v) rearranges to ν d²w̃/dz² = i f w̃, a second-order ODE whose decaying solution is a complex exponential. That single line contains the entire spiral.

Solving it: the corkscrew falls out of a complex exponential

The ODE ν d²w̃/dz² = i f w̃ has characteristic roots ±√(if/ν). Using √i = (1+i)/√2, the root that stays bounded with depth is +(1+i)/d where d = √(2ν/f) is the natural length scale. The physically bounded solution (velocity → 0 as depth z → −∞) is

• w̃(z) = V₀ · e^(z/d) · e^(i z/d), with z ≤ 0 measured downward from the surface.

Read this in two pieces. The real exponential e^(z/d) is an amplitude that shrinks with depth — speed falls by 1/e every d meters. The imaginary exponential e^(iz/d) is a rotation — the current vector turns steadily clockwise (Northern Hemisphere) as you descend. Splitting into components gives the classic Ekman result:

• u(z) = V₀ e^(z/d) cos(45° + z/d)
• v(z) = V₀ e^(z/d) sin(45° + z/d)

Applying the surface boundary condition — that the wind stress τ equals ρ ν (∂u/∂z) at z = 0 — fixes V₀ = √2 |τ|/(ρ f d) and, crucially, forces the surface current to sit at exactly 45° to the right of the wind stress. The Ekman depth D = π d = π√(2ν/f) is the depth at which the current has rotated a full 180° and points opposite to the surface flow; there its speed is only e^(−π) ≈ 4% of the surface value, so this is the practical bottom of the spiral.

The 45° and 90°: where the magic angles come from

Two angles define the whole phenomenon and they are not the same number. The 45° surface deflection is a local statement about the topmost water and it depends on the assumption of constant eddy viscosity — real oceans, with viscosity varying near the surface, typically show 20°–45°, matching Nansen's observed 20°–40° ice drift.

The 90° net transport is far more robust because it does not care about the messy vertical structure at all. Integrate the momentum equations over the entire layer. The friction terms telescope to just the surface stress τ and the (zero) stress at the base, leaving a clean algebraic balance:

• Eastward transport: M_x = τ_y / (ρ f)
• Northward transport: M_y = −τ_x / (ρ f)

The transport vector M = (τ × ẑ)/(ρf) is the wind stress rotated exactly 90° to the right. This is Ekman transport, and its independence from ν is what makes it a load-bearing result of oceanography. A 0.1 Pa wind stress at 45° latitude drives roughly M ≈ 0.1/(1025 × 10⁻⁴) ≈ 1 m²/s of depth-integrated transport — about one Sverdrup per 1000 km of coastline (1 Sv = 10⁶ m³/s).

The controlling variables: latitude, viscosity, and depth

The single parameter that sets the spiral's reach is the Ekman depth D = π√(2ν/f), and its dependence on the two knobs is instructive:

• Latitude (through f): D ∝ 1/√f, so D ∝ 1/√(sin φ). Near the poles f is largest and the layer is thinnest and tightest; approaching the equator f → 0, D → ∞, and the theory breaks down entirely (there is no Ekman spiral within a few degrees of the equator).
• Eddy viscosity (ν): D ∝ √ν. Because turbulent mixing is set by the wind itself, ν and hence D grow with wind speed. A useful empirical fit is D ≈ 4.3 W/√(sin φ) meters for wind speed W in m/s, giving D ≈ 45 m for a 10 m/s wind at 45° latitude and over 100 m in Southern Ocean storms.

Plugging numbers: with ν = 0.03 m²/s and f = 10⁻⁴ s⁻¹, d = √(2·0.03/10⁻⁴) = √600 ≈ 24 m, so D = π·24 ≈ 77 m. The surface current speed for τ = 0.1 Pa is V₀ = √2·0.1/(1025·10⁻⁴·24) ≈ 0.06 m/s — a few centimeters per second, well under 1% of the 10 m/s wind speed. The ocean is a sluggish, deep echo of the atmosphere above it.

Upwelling, gyres, and why the California coast is cold

The 90° transport rule turns wind patterns into biology. Along the California, Peru, Canary, and Benguela coasts, equatorward winds drive Ekman transport offshore (90° to the right in the north, left in the south). Surface water pulled away from the coast must be replaced from below, so cold, nutrient-rich water upwells from 100–200 m depth. This coastal upwelling feeds the planet's most productive fisheries — the Peruvian anchovy fishery alone once supplied ~10% of the world's fish catch — and it is why summer ocean temperatures off San Francisco hover near 12 °C while the same latitude in the Atlantic is far warmer.

On the basin scale, the curl of the wind stress sets Ekman pumping: convergent Ekman transport under the subtropical high-pressure systems pushes water downward at w_E = ∇×(τ/ρf), piling up a dome of warm water and driving, through the geostrophic balance, the great clockwise subtropical gyres. Ekman's little surface spiral is the engine that spins up the Gulf Stream and the Kuroshio. Divergence under the subpolar and equatorial wind fields does the opposite, lifting the thermocline and cooling the surface.

Measuring the spiral — and why it hid for a century

Ekman published the theory in 1905, but a clean, textbook spiral was not observed in the open ocean until the 1980s, and even then it looked compressed and flattened compared to the idealized corkscrew. The reasons are a catalogue of the model's assumptions failing gracefully:

• Constant viscosity is a fiction. ν varies strongly with depth; models with ν(z) increasing away from the surface reproduce the observed shallower, less-than-45° spirals. Price, Weller, and Schudlich's 1987 mid-latitude current-meter and drifter study finally resolved a spiral consistent with an eddy-viscosity profile.
• Stratification confines it. A shallow, strongly stratified summer mixed layer clips the spiral before it can fully rotate.
• The Stokes drift of surface waves adds a downwind, non-Ekman transport that must be subtracted before the spiral appears.

Modern verification uses moored acoustic Doppler current profilers (ADCPs), which measure velocity at meter resolution through the whole layer, plus satellite scatterometers (QuikSCAT, ASCAT) that map global wind stress and let oceanographers compute Ekman transport basin-wide. The 90° transport, being viscosity-independent, is confirmed to within observational error; the detailed angle of the spiral remains a sensitive probe of upper-ocean turbulence.

Subtleties and misconceptions

"The surface water goes downwind." No — that is the single most common error. The surface current is deflected 45° and the net water column moves 90° across the wind. Only the momentum, not the water, travels straight downwind.

"The Coriolis force does the work." The Coriolis force is always perpendicular to velocity, so it does zero work — it is a redirector, not an energy source. All the kinetic energy comes from the wind stress; Coriolis merely turns that momentum sideways, and viscosity dissipates it. This is the same reason a Foucault pendulum's plane rotates without gaining energy.

• Bottom Ekman layers exist too. Where a geostrophic current meets the seafloor, friction creates a mirror-image spiral that transports water 90° to the left of the interior flow (N. Hemisphere) — the mechanism behind the slow spin-down of ocean eddies and the drainage of a stirred teacup's leaves to the center.
• It fails at the equator. With f → 0 the Ekman depth diverges and the balance collapses; equatorial dynamics need a different theory.
• The eddy viscosity ν is not molecular. The molecular value (10⁻⁶ m²/s) would give a spiral only tens of centimeters deep. The observed tens-of-meters depth is proof that turbulent, not molecular, momentum transport rules the mixed layer.

The three dynamical regimes of a wind-driven, rotating ocean and how they differ

Property	Ekman layer	Geostrophic interior	Non-rotating (Stokes) layer
Force balance	Coriolis vs vertical friction	Coriolis vs pressure gradient	Inertia/friction vs pressure
Flow vs forcing	Spiral, 45°–90° right of wind	Along isobars (perpendicular to ∇p)	Parallel to the driving stress
Vertical extent	D ≈ 10–100 m (surface & bottom)	Whole water column below Ekman layer	δ ≈ √(2ν/ω), a few mm–cm
Governing number	Ekman No. Ek = ν/(fL²) ≪ 1	Rossby No. Ro = U/(fL) ≪ 1	Reynolds No. Re = UL/ν
Speed with depth	Decays as e^(−z/d), rotates	Roughly uniform in the interior	Decays and oscillates near wall