Cross-wave Stokes drift Created 2026-10-14 Updated 2026-11-05
Two plane internal gravity waves can each have zero Stokes drift but give a nonzero drift through interference. The displacement of one wave need not be perpendicular to the other wave's wave vector. For upward waves with wave vectors and and vertical amplitudes , the cross-wave drift is with . It has zero spatially averaged vertical flux. It cannot on its own satisfy an impermeable oscillating boundary: the second-order Eulerian mean velocity supplies the necessary normal correction there.
Initial and mean parcel labels in a surface wave Created 2026-10-07 Updated 2026-11-05
A fluid displacement measured from an initial position must vanish at the initial time. Integrating a harmonic velocity therefore introduces phase-dependent constants. Mean-orbit labels absorb those constants and leave purely oscillatory displacements. The instantaneous displacement correction to velocity can differ under the two reference conventions, but its period mean gives the same leading Stokes drift.
A fluid particle follows a Lagrangian trajectory, so its position satisfies rather than the velocity evaluated permanently at its initial position. Assume , and write
Set and . A Taylor expansion of the velocity about gives
Thus , and
At the next order,
so the required second-order trajectory in a sinusoidal velocity field is
Both corrections vanish at . All retained terms have units of length. The secular contribution gives the Stokes drift
although the Eulerian time average of the given velocity is zero. The particle samples the wave at its displaced position, producing this second-order drift. The expansion is in the dimensionless parameter , for fixed times; it ceases to be uniform when the accumulated drift changes the wave phase by order unity.
Put . Taking the gradient of the velocity potential gives
The motion is not generally incompressible; a velocity potential ensures irrotationality, not zero divergence. Both components have zero time average at each fixed point because their sine and cosine averages vanish.
The fluid acceleration is the material derivative . Direct differentiation gives
Averaging over one period at fixed therefore yields
This need not be the derivative of the mean velocity: the nonlinear advective term remains after averaging.
The dyed particle obeys the Lagrangian trajectory equations
Expand and , for bounded times. The first-order equations give , , so . At second order,
Integrating with zero initial values verifies
The velocity averaged along this particle over is its displacement divided by . Hence
This Stokes drift is a Lagrangian mean; it does not contradict the zero Eulerian mean velocity at a fixed point. The secular term is a small-parameter expansion, not an assertion of uniform accuracy as .
Let and . For a deep-water gravity wave, and the first-order velocity potential is . Therefore
Relative to mean parcel labels, integrate at fixed zeroth-order position to obtain the oscillatory fluid displacement
The Stokes drift is the period mean of the displacement correction to the first-order velocity field:
Thus , in the direction of propagation. A parcel samples faster forward motion near the top of its orbit and weaker backward motion lower down. This is the leading second-order drift of small-steepness waves, , relative to the Eulerian mean velocity.
Choose the positive-frequency branch with , let , and write . The internal gravity wave dispersion relation is . The phase velocity and group velocity are
They are perpendicular: the phase advances normal to the crests while energy travels along them. For a general signed , the positive frequency is and the horizontal component of group velocity changes sign with .
If , there is no real wave vector satisfying this relation. For prescribed real horizontal wavenumber, , so a bounded boundary response is an evanescent wave, rather than a freely propagating plane wave.
For a single incompressible plane wave, internal-wave polarization gives . Its oscillatory displacement is parallel to and therefore also perpendicular to . Every spatial derivative of the plane-wave field is proportional to a component of , so
There is no leading Stokes drift for that single wave. The fluid parcels oscillate along straight lines within phase planes and never sample a spatial change of the wave along their displacement. This conclusion does not prohibit cross terms between different wave vectors.
Assume and impose upward group velocity in . The two positive-frequency components then have wave vectors and with . Equal lengths are automatic, and orthogonality gives . Hence
The given orthogonal-wave condition is therefore possible only at this forcing frequency. The upward radiation condition excludes the branch.
Use superposition with vertical amplitude in each component and internal-wave polarization:
The two sets of crests have slopes and ; their energy propagates upward and outward at angles of , while their phases travel downward. Their interference makes fixed horizontal bands of vertical-velocity nodes, at , and horizontal-velocity nodes, at .
Put , . The first-order displacement relative to mean parcel labels is
The horizontal displacement correction vanishes identically, while
Consequently the formal cross-wave Stokes drift is
It has alternating downward and upward columns, no depth decay in this ideal infinite plane-wave field, and zero horizontal-period-averaged vertical transport. Each wave alone has zero drift; the interference terms create this result.
This calculation is a kinematic second-order consequence of the linear field, not a complete prediction of mean transport. In particular, an impermeable oscillating lower boundary cannot have that nonzero mean normal parcel velocity. Its first-order displacement is . Expanding the exact kinematic boundary condition on gives
This Eulerian mean velocity cancels the computed normal Stokes drift at the material boundary. Away from it, the second-order mean-flow and buoyancy equations must also be satisfied. Thus the alternating drift is a valid formal cross-wave effect, but treating it alone as persistent transport with zero Eulerian mean would be physically incomplete. A prescribed permeable boundary would require its own second-order flux specification instead.
The leading Stokes drift comes from evaluating the oscillatory velocity at the displaced particle position. Taylor expansion gives
Here and . Inserting the initial-position displacements from the preceding solution yields
In particular the instantaneous difference is zero at , as it must be. The unaveraged constant equality in the PDF is incompatible with its initial labels. Averaging over a period removes the oscillatory term:
Equivalently, using mean parcel labels and the purely oscillatory displacements gives at this order. This recovers the intended constant result. The period-mean drift is in the wave-propagation direction and decreases as . The corresponding second-order vertical difference is for initial labels and has zero mean. The resulting horizontal displacement per cycle is ; closed first-order circles do not imply zero second-order transport.
The dyed particle satisfies the Lagrangian trajectory equations
For the proposed approximation, and
Also
The initial conditions hold, verifying
through order .
Over one period, the periodic terms return to their initial values while the secular term changes by . Hence the dyed particle has Stokes drift
to this order, despite the zero Eulerian mean velocity.
For prescribed one-dimensional velocity , let , and expand the Lagrangian trajectory as . The Taylor expansion of gives and . Initial position requires and
The secular velocity is Stokes drift, even though the Eulerian time average vanishes. This fixed-time expansion requires and is not uniform once the secular displacement changes the phase by order unity.
Stokes drift Created 2026-09-24 Updated 2026-11-05
Stokes drift is the difference between the mean velocity of moving fluid particles and the mean velocity measured at fixed spatial positions. An oscillating particle samples a slightly different part of the velocity field during each part of its cycle. Correlation between its displacement and the spatial variation of the field can therefore produce a net drift even when the first-order Eulerian coordinate velocity average is zero.
For a small-amplitude deep-water gravity wave with surface elevation , no imposed mean current and mean particle height , the leading drift is
It follows the propagation direction and decreases with depth. Its being quadratic in wave amplitude explains why the linear particle orbits nearly close but do not close exactly.[ref]
A marked fluid particle and a fixed velocity probe measure different averages, so a wave-tank comparison must also account for any compensating return current. Surface float motion can include additional effects from wind or the float's own response. Stokes drift is consequently a precise difference of averages, rather than the name for every observed motion of floating debris.