How Science Works | Why Ocean Surface Water Can Move Sideways to the Wind

Wind acts on the ocean through stress, a transfer of horizontal momentum per unit surface area. Stress is measured in newtons per square metre. It is related to the relative motion of air and water and to conditions at their interface, but it is not the same quantity as wind speed.

The surface does not simply drag an entire water column as one rigid block. Turbulent motions transfer momentum downward, while current differences between depths create shear. The distribution of that transfer influences the vertical current profile.

Earth’s rotation changes the direction of moving water relative to the ground. In a horizontal momentum equation, the Coriolis term acts perpendicular to the velocity. It redirects motion; it is not an independent energy source that creates the current without forcing.

A steady wind-driven balance therefore combines surface forcing, rotational deflection and the vertical transfer of momentum. During adjustment, acceleration also matters. If the wind changes quickly, the ocean may never resemble the steady textbook solution over the interval being observed.

NOAA’s Ekman-spiral explanation introduces the layered response and its cumulative transport. It is a useful conceptual starting point, provided its idealized diagram is not mistaken for a universal profile measured everywhere in the ocean. NOAA explanation.

The first reading habit is to identify the forcing correctly. A quoted wind speed in metres per second cannot be inserted directly into a formula requiring stress in newtons per square metre. Converting between them requires an additional air–sea momentum-transfer model or measurement.

Contents · Next chapter

Velocity at a depth tells us how fast water there moves and in which direction. Depth-integrated volume transport adds the horizontal velocity contributions through the chosen layer. Its units are square metres per second, equivalently cubic metres per second per metre of horizontal width.

Consider an original arithmetic sketch with two ten-metre-thick layers. Let the upper layer move east at 0.10 metres per second and the lower layer west at 0.06 metres per second. Their transports are 1.0 square metre per second eastward and 0.6 westward, giving a net 0.4 eastward.

The surface speed is 0.10 metres per second, while the depth-average speed through the twenty-metre layer is 0.4/20 = 0.02 metres per second eastward. These two values are different because the lower contribution partly cancels the upper one.

This two-layer sketch is not an Ekman solution. It simply demonstrates why measuring one layer does not automatically determine the integral. In a rotating current, directions as well as speeds can vary through depth, so vector addition is required.

An Ekman calculation usually concerns the wind-driven component after an appropriate background current has been accounted for. Integrating the full measured current without distinguishing its pressure-driven, tidal or other contributions can give a transport that answers a different question.

Before interpreting an arrow labelled “current,” therefore, ask whether it represents a surface velocity, a depth-average velocity or a depth-integrated transport, and whether it includes the entire current or only an estimated component.

Contents · Previous chapter · Next chapter

Choose x eastward and y northward. Let Mx and My be the eastward and northward components of depth-integrated wind-driven volume transport. Let τx and τy be the corresponding wind-stress components, ρ the seawater density and f the Coriolis parameter.

For a steady, horizontally uniform deep-ocean Ekman component with negligible stress at its lower boundary, the integrated balances are −f My = τx/ρ and f Mx = τy/ρ. Thus Mx = τy/(ρf) and My = −τx/(ρf).

These equations rotate the stress direction through a right angle. In the Northern Hemisphere f is positive: northward stress gives eastward transport, and eastward stress gives southward transport. Both are ninety degrees to the right of the stress arrow. Negative f in the Southern Hemisphere reverses the orientation.

The derivation works because stresses exchanged internally between layers cancel when the whole layer is integrated. The remaining boundary forcing is the surface stress minus any stress at the lower boundary. Rotation balances that net forcing in the stated steady model.

This integrated relation does not require every depth to move at ninety degrees. Different vertical profiles can produce the same integral. It also does not describe a shallow water column with important bottom stress unless that additional boundary force is included.

Pressure-driven background motion must be treated consistently, and acceleration or horizontal momentum changes cannot simply be discarded when they are important. The perpendicular result is a consequence of a particular balance, not a geometrical command that every ocean current obeys.

Contents · Previous chapter · Next chapter

Take seawater density ρ = 1,025 kg per cubic metre, f = 1 × 10⁻⁴ per second and a northward wind stress of 0.10 newton per square metre. Set the eastward stress component to zero.

The eastward transport is 0.10/(1,025 × 10⁻⁴) ≈ 0.9756 square metres per second. The northward transport component is zero in this ideal example. The sign is positive eastward, consistent with a rightward deflection from a north-pointing stress arrow.

The units provide a useful check. Stress has units of kg per metre per second squared. Density times f has units of kg per cubic metre per second. Dividing leaves square metres per second, the correct units for volume transport per unit width.

If the same net transport were distributed through a forty-metre layer, its depth-average eastward velocity would be 0.9756/40 ≈ 0.02439 metres per second, or about 2.44 centimetres per second. This does not specify the surface velocity or imply a uniform profile. Forty metres is an explicitly chosen averaging depth, not a predicted Ekman-layer thickness.

Across a one-hundred-kilometre-long north–south section, uniform eastward transport of this magnitude would correspond to about 97,561 cubic metres per second crossing the section. The multiplication uses 100,000 metres of horizontal width and assumes the transport is uniform along it.

The numerical result is conditional on the stress, density, rotation and steady deep-layer assumptions. It is not a measured flow rate for a particular current, harbour or coastline.

Contents · Previous chapter · Next chapter

A classical solution assumes a constant vertical eddy viscosity, representing momentum mixing with one depth-independent coefficient. In that ideal deep-water model, the surface current is forty-five degrees from the wind stress and the velocity rotates and weakens with depth, producing an Ekman spiral.

The ninety-degree integrated transport and the forty-five-degree surface velocity are therefore different statements. The latter depends on more detailed assumptions about the profile. A reader who remembers only “water moves at an angle to the wind” can easily apply the wrong angle to the wrong quantity.

Eddy viscosity is a way to represent turbulent momentum transfer, not a claim that ocean turbulence is identical to molecular viscosity. Its depth variation can change the surface deflection and spiral shape while the integrated boundary-force balance remains informative under its own assumptions.

Dritschel, Paldor and Constantin’s 2020 theoretical study solved an Ekman problem with piecewise-uniform viscosity. It showed how the surface angle depends on the depths and mixing properties of the represented layers. The calculation provides a concrete reason not to treat forty-five degrees as a universal surface observation. Original research.

A sketch should therefore show separate arrows: one at the surface, several through depth and one for their integral. Label the assumed profile rather than drawing every arrow at ninety degrees. The strength of the classical solution is that it reveals a mechanism cleanly; its usefulness increases when the assumptions are visible rather than forgotten.

Contents · Previous chapter · Next chapter

Imagine an ideal Northern Hemisphere coast running north–south, with land to the east and ocean to the west. A southward wind stress gives westward Ekman transport under the stated sign convention. That is offshore transport for this particular coastline.

Water cannot flow out of the land to replace it. In an idealized coastal balance, offshore loss from the surface layer can be supplied by water rising from below and by other allowed flows. The coastal constraint turns a horizontal transport tendency into a vertical-exchange problem.

Reverse the stress to northward and the ideal transport becomes eastward, toward shore. Convergence can then favour downward motion and changes in sea level and pressure gradients. The actual response depends on geometry, time dependence, stratification and alongshore variations.

Diagram to sketch: put north at the top, land on the right and ocean on the left. Draw a south-pointing stress arrow along the coast and a west-pointing transport arrow into the ocean. In a side view, show a possible upward replacement flow beneath the surface-layer outflow. Label this as an idealized coastal mass balance, not the trajectory of every parcel.

A direction rule without a map is incomplete. The same wind direction at a differently oriented coast can have another onshore or offshore component. Moving to the Southern Hemisphere also changes the rotational sign.

Nor does wind-driven upwelling automatically determine biological productivity. Nutrient content, light, residence time and ecological responses add further conditions. The present explanation follows the physical transport step; it does not turn a current arrow into a guaranteed ecosystem outcome.

Contents · Previous chapter · Next chapter

Uniform horizontal transport across an unbounded region does not create upwelling everywhere merely because the water is moving sideways. Vertical exchange follows from convergence or divergence of the layer transport, together with the relevant mass-balance assumptions.

Use an original coastal-strip model. Let offshore transport increase from zero at the coastal boundary to 0.9756 square metres per second at an offshore boundary fifty kilometres away. Assume no net alongshore convergence, negligible surface-volume accumulation and no other significant volume sources or sinks.

Per metre of coastline, the strip loses 0.9756 cubic metres per second horizontally. Spread over a strip fifty thousand metres wide, the compensating area-average upward velocity at the layer base is 0.9756/50,000 ≈ 1.951 × 10⁻⁵ metres per second.

Multiplying by 86,400 seconds per day gives approximately 1.686 metres per day. This is an average volume-replacement rate for the model, not a claim that every deep-water parcel rises exactly that distance each day or that such a rate persists indefinitely.

The calculation uses the difference between boundary transports, not just the offshore transport at one point. If the same transport entered through one boundary and left through another, the net horizontal loss could be zero and no compensating upward flow would follow from this simplified balance.

Far from coasts, spatial changes in wind stress and in f can also produce transport convergence or divergence. This is one route to Ekman pumping or suction. The terminology concerns a vertically exchanging flow; there is no literal mechanical pump beneath the ocean surface.

Contents · Previous chapter · Next chapter

The Coriolis parameter is f = 2Ω sin φ, so its magnitude depends on latitude and it vanishes at the equator. The simple transport formula contains division by f. Letting f approach zero and claiming an infinite real transport is a misuse of the approximation, not a prediction of an infinitely fast equatorial current.

Near the equator, neglected spatial and time-dependent terms become important and a different dynamical treatment is needed. Likewise, a shallow shelf can have important bottom stress, while rapidly changing winds can produce acceleration and inertial motion rather than an immediate steady response.

For the example f = 10⁻⁴ per second, the inertial period 2π/|f| is about 62,832 seconds, or 17.45 hours. This is a rotational timescale. It is not a universal waiting time after which every ocean automatically reaches a steady Ekman state; mixing, forcing history and other processes affect adjustment.

Stratification can confine momentum transfer or change its vertical distribution. Surface waves can add Stokes drift, and pressure gradients can drive strong background currents. A floating object can also experience direct wind drag. None of those effects is determined by the depth-integrated Ekman formula alone.

A sound application begins by identifying which terms are plausibly small over the space and time of interest. If they are not small, retain them in the model rather than forcing observations to match a memorized right-angle rule. The theoretical result remains useful as a reference balance even when the complete ocean requires more physics.

Contents · Previous chapter · Next chapter

A current profile can contain a wind-driven contribution superimposed on a depth-varying pressure-driven flow. Removing an oversimplified background can distort the inferred Ekman spiral. A discrepancy between a textbook shape and a measured total current is therefore not automatically a failure of the wind-driven balance.

Roach and colleagues’ 2015 study used profiling-float observations north of Kerguelen in the Southern Ocean. Treating the background geostrophic current as vertically uniform suggested a compressed spiral. Accounting more accurately for geostrophic shear brought the observed spiral closer to classical predictions. In those data, the interpretation depended importantly on separating the background flow; it did not establish one universal explanation for all observed spiral shapes. Original study.

Sea ice introduces another momentum-transfer interface. Liu and Liang’s July 2026 analysis used a twenty-six-year Arctic ocean–ice state estimate to identify recurring combinations of wind, ice and geostrophic contributions to Ekman pumping. It was a model-constrained state-estimate analysis, not a direct measurement proving that the same balance applies everywhere. Original research.

Together, these examples show why oceanographers combine wind estimates, velocity profiles, density structure and dynamical models. Each helps constrain a different part of the momentum balance.

For a field claim, ask which current component was measured or inferred, how the background was removed and whether the averaging interval suits the steady approximation. A surface arrow alone cannot answer those questions.

Contents · Previous chapter · Next chapter

Start by drawing the wind-stress arrow and labelling the hemisphere. Under the ideal steady deep-ocean assumptions, rotate that arrow ninety degrees right in the Northern Hemisphere or left in the Southern Hemisphere to obtain the direction of net Ekman transport. Do not automatically assign that direction to a surface drifter.

Then check units and boundaries. Stress divided by density and f gives volume transport per unit width. A horizontal section width converts it to a volume flow rate. An assumed layer depth converts it to a depth-average velocity. These are different operations answering different questions.

The original northward-stress example gave about 0.9756 square metres per second eastward. Dividing by forty metres gave a depth-average speed near 2.44 centimetres per second. The separate coastal-strip balance gave about 1.686 metres per day of average upward replacement because transport diverged across the strip. Neither calculation specified the surface-current angle.

Finally, identify the limits: rapidly changing forcing, the equator, shallow bottoms, sea ice, stratification and background currents can require a richer treatment. A good explanation keeps those limits beside the result rather than treating them as exceptions discovered only after the formula fails.

Ekman transport reveals how a rotating fluid can respond sideways to surface forcing and link horizontal motion to vertical exchange. Continue through Oceanography for the wider current and water-mass system, or return to the How Science Works master guide.

Contents · Previous chapter · Continue to Oceanography