Wind-Driven Circulation: Ekman, Sverdrup, Stommel and Munk
CSIR NET Ocean Sciences: Ekman transport and pumping, Sverdrup balance, Stommel and Munk western intensification, vorticity reasoning and upwelling types.
- 19 min
On this page · 14 sections
Prerequisites
- Geostrophy, Dynamic Topography and Eddies: geostrophic balance, \( f \), \( \beta \) and the Rossby number.
- Ocean Circulation, Ekman Spiral, ENSO and IOD: the full Ekman spiral solution, written out there.
- Vorticity from fluid mechanics: \( \zeta = \partial v/\partial x - \partial u/\partial y \); curl of a vector field.
The Hook: Why Is the Gulf Stream on the West?
The Hierarchy at a Glance
| Theory | Year | Balance | Key result | What it cannot do |
|---|---|---|---|---|
| Ekman | 1905 | Coriolis vs vertical friction in a surface layer | Ekman spiral; net transport 90° to the right of the wind (NH) | no pressure gradients, no basin circulation |
| Sverdrup | 1947 | \( \beta v \) vs wind-stress curl, vertically integrated | interior meridional transport from the curl alone | no boundary currents; cannot close the circulation |
| Stommel | 1948 | wind curl + \( \beta \) + linear bottom friction | westward intensification is due to \( \beta \); western boundary layer width \( r/\beta \) | idealised friction |
| Munk | 1950 | wind curl + \( \beta \) + lateral eddy viscosity | realistic gyre pattern from observed winds; boundary layer width \( (A_H/\beta)^{1/3} \) | linear; transports too small without nonlinear recirculation |
graph TD
W[Wind stress on sea surface] --> E[Ekman layer: transport 90 deg to wind]
E --> P[Convergence or divergence: Ekman pumping w_E]
P --> S[Interior: vortex stretching balanced by beta v - Sverdrup]
S --> B{Return flow must close}
B --> ST[Stommel: bottom friction on west]
B --> MU[Munk: lateral friction on west]
ST --> WBC[Western boundary current]
MU --> WBCEkman Dynamics
The Ekman layer
In the surface layer the wind stress \( \boldsymbol{\tau} \) is carried down by turbulent friction and balanced by the Coriolis force. With constant vertical eddy viscosity \( A_z \), Ekman (1905) found:
- Surface current 45° to the right of the wind in the NH (left in the SH).
- The current rotates clockwise (NH) and decays as \( e^{\pi z/D_E} \) with depth: the Ekman spiral.
- Ekman depth \( D_E = \pi\sqrt{2A_z/|f|} \), where the current is opposite to the surface current and reduced to \( e^{-\pi} \approx 4\% \).
- The net transport is 90° to the right of the wind (NH), independent of \( A_z \).
Ekman worked from the observation, made during Nansen’s Fram drift, that sea ice drifted to the right of the wind. The full solution is on the marine geology Ekman page; here we use its integral result.
Ekman transport
Integrating the Ekman-layer equations over depth gives
$$\boxed{M_x^E = \frac{\tau_y}{f}, \qquad M_y^E = -\frac{\tau_x}{f}}$$as mass transport per unit width (kg m\(^{-1}\) s\(^{-1}\)). Dividing by \( \rho \) gives the volume transport per unit width, \( U_E = \tau/(\rho f) \) (m\(^2\) s\(^{-1}\)).
Wind stress comes from the bulk formula \( \tau = \rho_a C_D U_{10}^2 \) with \( \rho_a \approx 1.2 \) kg m\(^{-3}\) and \( C_D \approx 1.2 \)–\( 1.5\times 10^{-3} \). An 8 m s\(^{-1}\) wind gives roughly 0.1 N m\(^{-2}\).
Ekman pumping and suction
Where the Ekman transport converges, water is pushed down out of the Ekman layer (Ekman pumping, downwelling); where it diverges, water is drawn up (Ekman suction, upwelling). The vertical velocity at the base of the Ekman layer is
$$\boxed{w_E = \frac{1}{\rho}\,\hat{\mathbf{k}}\cdot\nabla\times\left(\frac{\boldsymbol{\tau}}{f}\right) \approx \frac{1}{\rho f}\left(\frac{\partial\tau_y}{\partial x} - \frac{\partial\tau_x}{\partial y}\right)}$$(the approximation ignores the variation of \( f \)). With \( z \) upward, \( w_E > 0 \) is upwelling.
| Wind-stress curl (NH) | \( w_E \) | Where | Result |
|---|---|---|---|
| Negative (anticyclonic): westerlies to the north, trades to the south | downward (pumping) | subtropical gyres | thick warm layer, SSH high, convergence |
| Positive (cyclonic) | upward (suction) | subpolar gyres, north of the Findlater Jet in the Arabian Sea | domed thermocline, nutrients |
The bottom Ekman layer
Near the sea floor, friction slows a geostrophic flow. In the bottom boundary layer the flow turns to the left of the interior flow in the NH (to the right in the SH), giving a net cross-isobar transport toward low pressure. Under a cyclonic eddy this transport converges and forces upward motion out of the bottom layer. This Ekman spin-down is the bottom friction that Stommel’s model represents with a linear drag.
Vorticity and the Sverdrup Balance
Planetary vorticity and vortex stretching
Water carries planetary vorticity \( f \) (the vorticity of the rotating Earth at its latitude) and relative vorticity \( \zeta \). For a layer of thickness \( H \), the potential vorticity
$$q = \frac{f + \zeta}{H}$$is conserved in the absence of forcing and friction. For large-scale interior flow \( \zeta \ll f \), so \( f/H \approx \) constant: if the column is squashed (\( H \) decreases) it must move to smaller \( f \), toward the equator; if stretched, toward the pole.
Sverdrup’s result
Under the Ekman layer, linear vorticity balance in a geostrophic interior gives \( \beta v = f\,\partial w/\partial z \). Integrated over the whole depth (with \( w = w_E \) at the top and \( w = 0 \) at the bottom), it becomes the Sverdrup balance:
$$\boxed{\beta V = \frac{1}{\rho}\,\hat{\mathbf{k}}\cdot\nabla\times\boldsymbol{\tau}}$$where \( V \) is the total meridional volume transport per unit width (Ekman plus geostrophic), and \( \beta = 2\Omega\cos\phi/R_E \approx 2\times 10^{-11} \) m\(^{-1}\) s\(^{-1}\) at mid-latitudes.
Physical reading for a subtropical gyre: negative curl → Ekman pumping squashes the water columns → they must move equatorward. The whole interior of a subtropical gyre therefore flows slowly toward the equator. The poleward return must happen somewhere else.
With a purely zonal wind, \( V = -\frac{1}{\rho\beta}\frac{\partial\tau_x}{\partial y} \); the zonal transport follows from continuity, integrating westward from the eastern boundary (where the normal flow must vanish).
Sverdrup (1947) applied this balance to the eastern tropical Pacific, where it accounts for the North Equatorial Countercurrent from the curl of the trade winds near the ITCZ. Sverdrup transport computed from observed winds agrees with measured interior transports to within errors over much of the subtropics, which is good evidence that the theory captures the gyre interior.
Western Intensification
Stommel (1948): bottom friction and \( \beta \)
Stommel modelled a rectangular, homogeneous, flat-bottomed ocean with a sinusoidal zonal wind and a linear bottom drag (\( -r\mathbf{u} \)). His three cases:
| Case | Result |
|---|---|
| No rotation | symmetric gyre |
| Uniform rotation (\( f \)-plane, \( \beta = 0 \)) | still symmetric |
| \( f \) varying with latitude (\( \beta \neq 0 \)) | streamlines crowd against the west: a narrow, fast western boundary current and a broad, slow interior |
So the cause of westward intensification is \( \beta \), the variation of the Coriolis parameter with latitude. The Stommel boundary-layer width is
$$\boxed{\delta_S = \frac{r}{\beta}}$$With \( r = 10^{-6} \) s\(^{-1}\) and \( \beta = 2\times 10^{-11} \) m\(^{-1}\) s\(^{-1}\), \( \delta_S \approx 50 \) km.
Munk (1950): lateral friction and real winds
Munk replaced bottom drag with lateral eddy viscosity \( A_H \) and drove the model with observed zonal-mean winds. The interior is still Sverdrup; the western boundary layer (the Munk layer) has width scale
$$\boxed{\delta_M = \left(\frac{A_H}{\beta}\right)^{1/3}}$$\( A_H = 10^3 \) m\(^2\) s\(^{-1}\) gives \( \delta_M \approx 37 \) km at 30°. With the observed wind pattern, Munk’s model produced the main features of the North Pacific and North Atlantic: a subtropical gyre with a strong western boundary current, a subpolar gyre to the north, and the countercurrent structure near the equator. Its transports were of the right order but smaller than observed, because the real boundary currents recirculate through nonlinear (inertial) effects that the linear model omits.
The vorticity argument, step by step
Why must the return flow be on the west? Follow the vorticity budget of a subtropical gyre (NH):
- The wind puts in negative (anticyclonic) vorticity everywhere.
- In the interior, water moves equatorward to smaller \( f \): consistent with the input (Sverdrup).
- In a western boundary current, water moves poleward fast, gaining planetary vorticity. The current’s shear against the coast (fast offshore, zero at the wall) is strong positive relative vorticity, and friction at the wall is what supplies it. That positive vorticity balances both the wind input and the planetary-vorticity gain. Balance is possible.
- In a hypothetical eastern boundary current flowing poleward, the shear against an eastern wall has the opposite sign: friction would add negative vorticity, the same sign as the wind. No balance.
Hence the return flow is concentrated on the west, and the eastern boundary currents are broad and weak. The same argument works for subpolar gyres (positive wind curl, equatorward western boundary currents such as the Labrador Current and Oyashio) and in the SH.
Upwelling
Upwelling is the rise of subsurface water (colder, richer in nutrients, often poorer in oxygen) into the surface layer. Typical vertical speeds are 1–10 m per day, tens of times faster than gyre-scale Ekman pumping (about 0.1 m per day), but over narrow regions.
Types of upwelling
| Type | Mechanism | Examples |
|---|---|---|
| Coastal (Ekman) upwelling | alongshore wind with the coast on its left (NH) or right (SH): offshore Ekman transport, replaced from below | California, Canary (NW Africa), Benguela, Humboldt (Peru–Chile); Somalia and Oman in the southwest monsoon; parts of the southwest coast of India |
| Equatorial upwelling | easterly trades: Ekman transport poleward on both sides because \( f \) changes sign, so the surface flow diverges at the equator | equatorial Pacific and Atlantic cold tongues |
| Open-ocean (curl-driven) upwelling | positive wind-stress curl (NH): Ekman suction | subpolar gyres; north (left) of the Findlater Jet in the Arabian Sea; Costa Rica Dome |
| Eddy-driven | cyclonic eddies dome isopycnals; eddy–wind interaction | Bay of Bengal cyclonic eddies |
| Topographic and current-driven | flow over banks, shelf-break fronts, current meanders | shelf breaks, headlands |
Eastern-boundary coastal upwelling also involves curl: near the coast the wind weakens, and the resulting positive curl adds Ekman suction offshore of the coastal band.
Coastal upwelling: the numbers
The offshore Ekman transport per unit length of coast is \( U_E = \tau_{\text{alongshore}}/(\rho|f|) \). It is supplied from below within a band roughly one Rossby radius wide (10–50 km), so
$$w \approx \frac{U_E}{R_d}$$Signatures and consequences
- Low SST in a band along the coast; sharp upwelling fronts; filaments carrying cold water offshore.
- Isotherms and isopycnals tilt upward toward the coast; a coastal undercurrent often flows poleward beneath the equatorward surface jet (e.g. under the California Current).
- High nutrients and chlorophyll: the four eastern-boundary upwelling systems cover a tiny fraction of the ocean’s area but support a large share of the world fish catch.
- Decay of the resulting organic matter at depth produces oxygen minimum zones (Peru–Chile, the Arabian Sea).
- Upwelling indices: alongshore wind stress converted to offshore Ekman transport per 100 m of coast is used as an index of upwelling strength.
Downwelling
The reverse wind direction (coast on the right in the NH) piles water against the coast: sea level rises, isopycnals tilt down, surface water sinks. Downwelling coasts are nutrient-poor; downwelling-favourable winds also raise coastal sea level and add to storm surges.
Memory Tricks
Don’t Confuse
| This | vs | That |
|---|---|---|
| Ekman transport \( \tau/(\rho f) \): per unit width, horizontal | Ekman pumping \( w_E \): vertical velocity from the curl | |
| Ekman spiral depends on \( A_z \) | Ekman transport is independent of \( A_z \) | |
| Sverdrup transport: total (Ekman + geostrophic), interior only | Boundary current transport: closes the circulation | |
| Stommel: bottom friction, \( \delta = r/\beta \) | Munk: lateral friction, \( \delta = (A_H/\beta)^{1/3} \) | |
| Cause of westward intensification: \( \beta \) | Friction only sets the boundary-current width | |
| Coastal upwelling: alongshore wind | Curl-driven upwelling: spatial variation of the wind | |
| Surface Ekman layer: transport right of wind (NH) | Bottom Ekman layer: flow turns left of interior flow (NH) |
Common Mistakes
Practice
All questions below are original practice questions written for this page. They are not from any CSIR NET paper.
Level 1: Foundation
- In which direction relative to the wind is the net Ekman transport in the southern hemisphere?
- Which of the four classical theories first introduced the \( \beta \)-effect into the wind-driven circulation?
- What is the unit 1 Sverdrup?
Answers
- 90° to the left of the wind.
- Sverdrup (1947); Stommel (1948) then showed that \( \beta \) causes westward intensification.
- \( 10^6 \) m\(^3\) s\(^{-1}\).
Level 2: Part B Style
Practice Q1. On a north–south coastline with the ocean to the west, at 35°N, upwelling is favoured by a wind blowing toward the:
(a) north (b) south (c) east (d) west
Answer: (b) south
A southward (equatorward) wind has the coast (east) on its left; Ekman transport is to the right of the wind, i.e. west, offshore. This is the California situation.
Practice Q2. At 15°S, an alongshore wind stress of 0.08 N m\(^{-2}\) drives offshore Ekman transport. The transport per unit length of coast is closest to (\( \rho = 1025 \) kg m\(^{-3}\)):
(a) 0.5 m\(^2\) s\(^{-1}\) (b) 1.0 m\(^2\) s\(^{-1}\) (c) 2.1 m\(^2\) s\(^{-1}\) (d) 4.2 m\(^2\) s\(^{-1}\)
Answer: (c) 2.1 m² s⁻¹
\( |f| = 2\times 7.292\times 10^{-5}\times\sin 15° = 3.77\times 10^{-5} \) s\(^{-1}\). \( U_E = 0.08/(1025\times 3.77\times 10^{-5}) = 2.07 \) m\(^2\) s\(^{-1}\).
Practice Q3. In Stommel’s 1948 model, the boundary current width depends on:
(a) \( A_H/\beta \) (b) \( (A_H/\beta)^{1/3} \) (c) \( r/\beta \) (d) \( r/f \)
Answer: (c)
Stommel used linear bottom friction with coefficient \( r \) (s\(^{-1}\)); \( r/\beta \) has units of length. Option (b) is Munk’s.
Practice Q4. In a NH subtropical gyre interior, Sverdrup dynamics predicts that the depth-integrated meridional flow is:
(a) poleward, because the wind curl is positive (b) equatorward, because the wind curl is negative (c) zero, because the flow is geostrophic (d) eastward
Answer: (b)
Westerlies poleward and trades equatorward give a negative (anticyclonic) curl; Ekman pumping squashes the columns, \( \beta V < 0 \), equatorward flow.
Level 3: Part C Style
Practice Q5. A model basin at 30°N is 6000 km wide. The zonal wind stress is \( \tau_x = -\tau_0\cos(\pi y/L) \) with \( \tau_0 = 0.1 \) N m\(^{-2}\) and \( L = 3000 \) km (\( y = 0 \) at the southern edge of the gyre). Take \( \beta = 1.98\times 10^{-11} \) m\(^{-1}\) s\(^{-1}\) and \( \rho = 1025 \) kg m\(^{-3}\). Which statements are correct?
- The wind-stress curl is negative throughout \( 0 < y < L \).
- The maximum Sverdrup transport per unit width is about 5.2 m\(^2\) s\(^{-1}\), at \( y = L/2 \).
- The western boundary current must carry about 31 Sv northward at mid-gyre.
- Doubling \( \tau_0 \) and halving \( \beta \) quadruples the boundary-current transport.
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 1, 2, 3 and 4 (d) 2 and 4 only
Answer: (c) all four
- \( \partial\tau_x/\partial y = \tau_0(\pi/L)\sin(\pi y/L) > 0 \) for \( 0 < y < L \), so curl \( = -\partial\tau_x/\partial y < 0 \). Correct.
- \( |V|_{\max} = \frac{\tau_0\pi/L}{\rho\beta} = \frac{0.1\times\pi/(3\times 10^6)}{1025\times 1.98\times 10^{-11}} = \frac{1.047\times 10^{-7}}{2.03\times 10^{-8}} = 5.16 \) m\(^2\) s\(^{-1}\), where \( \sin = 1 \), at \( y = L/2 \). Correct.
- \( 5.16\times 6\times 10^6 = 3.10\times 10^7 \) m\(^3\) s\(^{-1}\) ≈ 31 Sv southward in the interior, so 31 Sv northward in the WBC. Correct.
- \( V \propto \tau_0/\beta \): \( 2/(1/2) = 4 \). Correct.
Practice Q6. Consider a NH subtropical gyre and these statements about its vorticity balance:
- In the interior, the negative vorticity input by the wind is balanced by southward movement to lower planetary vorticity.
- In a western boundary current, frictional stress at the coast supplies positive relative vorticity.
- An eastern boundary current carrying the same northward return flow could also balance the vorticity budget, but it would be wider.
- With \( f \) constant (\( \beta = 0 \)), Stommel’s solution has no boundary current at all.
Which are correct?
(a) 1 and 2 only (b) 1, 2 and 3 only (c) 1, 2 and 4 only (d) 2 and 3 only
Answer: (a) 1 and 2 only
1 and 2 are the standard argument. 3 is wrong: friction in a poleward eastern boundary current would add negative vorticity, the same sign as the wind, so no balance is possible however wide it is. 4 is wrong: on an \( f \)-plane the gyre is symmetric, with frictional boundary flow on both sides, not absent.
Practice Q7. Off a coast at 20°N, satellite data show SST 4 °C lower within 30 km of the coast than offshore during a period of steady equatorward alongshore wind of 9 m s\(^{-1}\). Take \( \rho_a = 1.2 \) kg m\(^{-3}\), \( C_D = 1.3\times 10^{-3} \), \( \rho = 1025 \) kg m\(^{-3}\). The upwelling velocity in the 30 km band is closest to:
(a) 0.5 m per day (b) 2.1 m per day (c) 7.1 m per day (d) 21 m per day
Answer: (c) about 7 m per day
\( \tau = 1.2\times 1.3\times 10^{-3}\times 81 = 0.126 \) N m\(^{-2}\). \( f = 4.99\times 10^{-5} \) s\(^{-1}\).
\( U_E = 0.126/(1025\times 4.99\times 10^{-5}) = 2.47 \) m\(^2\) s\(^{-1}\).
\( w = 2.47/(3\times 10^4) = 8.2\times 10^{-5} \) m s\(^{-1}\) \( = 7.1 \) m per day.
Quick Revision Table
| Idea | Formula or fact |
|---|---|
| Wind stress | \( \tau = \rho_a C_D U_{10}^2 \) |
| Ekman surface current | 45° right of wind (NH) |
| Ekman depth | ( D_E = \pi\sqrt{2A_z/ |
| Ekman transport | \( \tau/(\rho f) \), 90° right (NH), left (SH) |
| Ekman pumping | \( w_E = \rho^{-1}\,\mathrm{curl}(\tau/f) \); negative curl → downward |
| Potential vorticity | \( (f + \zeta)/H \) conserved |
| Sverdrup | \( \beta V = \rho^{-1}\,\mathrm{curl}\,\tau \) |
| Stommel | \( \beta \) causes westward intensification; \( \delta_S = r/\beta \) |
| Munk | lateral friction, real winds; \( \delta_M = (A_H/\beta)^{1/3} \) |
| Coastal upwelling | coast on left of wind (NH), right (SH); \( w \approx U_E/R_d \) |
| Equatorial upwelling | easterlies, divergence because \( f \) changes sign |
| Curl-driven upwelling | positive curl (NH) |
Key Takeaways
What’s Next
- Gyres, Boundary and Equatorial Currents, Somali Current: the real currents these theories explain, and Arabian Sea upwelling.
- Geostrophy, Dynamic Topography and Eddies for the balance assumed in the interior.
- Monsoon, Cyclones and ENSO for the winds that force the Indian Ocean.
- Back to the Physical Oceanography overview.