Ocean circulation

Ocean circulation comprises wind-driven, buoyancy-driven, and tidal motions of the ocean over a wide range of spatial and temporal scales.

Meridional overturning circulation connects upper and deep ocean transports with vertical exchange. Surface density changes help form deep water, while winds, mixing and interbasin exchange contribute to the global return pathways. Its heat, carbon and nutrient transport affects climate and ocean ventilation. The Atlantic meridional overturning circulation is its Atlantic-basin expression, not a synonym for every wind-driven current in that basin.

A large-scale circulation with upper and deep return branches and exchanges between them. Different branch temperatures give meridional heat transport even if their net volume transport cancels. The Atlantic meridional overturning circulation is an important regional component. Wind forcing, surface buoyancy exchange and interior mixing all matter, so a purely thermally driven picture is incomplete.

Upward water motion bringing subsurface water toward the surface. For example, wind-driven surface divergence near the equator can be supplied by upwelling of colder water from near the thermocline. A deeper thermocline can reduce the cooling produced by a given rate of upwelling.

A change in the amount of ocean water belonging to a specified density class, driven by surface heat flux, freshwater exchange or interior mixing. Advection along an isopycnal transports a water mass without necessarily transforming it; diapycnal mixing can transfer water between density classes.

A dense current that leaves a marginal sea over a sill and descends into a deeper basin. Entrainment increases its volume flux while reducing its density anomaly; the resulting mixed water can contribute to a deep overturning branch.

The Antarctic Convergence is the circumpolar oceanic transition where cold Antarctic waters meet and mix with warmer waters farther north. It is associated with the Antarctic Polar Front, a boundary between contrasting water masses rather than a fixed line on a map. The Australian Antarctic Program identifies this mixing region and its enhanced nutrient concentrations.^[ref]

A ship crossing the region can record a marked change in near-surface temperature together with changes in the vertical water-column structure. The physical distinction is between water masses with different density and thermal properties, so a temperature section is more informative than latitude alone. The Antarctic Circumpolar Current follows an eastward path around the continent; the Antarctic Convergence identifies a frontal region within this broader circulation.

The Antarctic Coastal Current is a mainly westward flow near the Antarctic coast. It can transport icebergs along the continental margin; coastal geometry, winds and regional circulation make trajectories variable.

The Antarctic Circumpolar Current is the eastward ocean circulation encircling Antarctica and connecting the Atlantic, Indian and Pacific oceans. Its direction differs from the predominantly westward Antarctic Coastal Current closer to the continent. Wind stress, oceanic density gradients and seabed geometry shape the current; it is not a single uniform-speed ring.^[ref][ref]

The current illustrates why ocean dynamics require both forcing and boundaries. Atmospheric winds supply momentum, geostrophic balance relates much of the large-scale flow to horizontal pressure gradients, and underwater ridges redirect its path. Measuring the current therefore involves velocity and water-column structure across a section, rather than tracking one surface parcel and assigning that speed to the entire ocean.

Antarctic bottom water is a dense southern-origin water mass that ventilates much of the abyssal ocean. Cooling and brine rejection form dense shelf water; its descent and mixing with adjacent waters contribute to Antarctic bottom water production. Shelf production is geographically concentrated and does not mean that every polynya exports water to the abyss.

Wind stress is the tangential traction exerted by the atmosphere on the ocean surface. Its curl injects vorticity into the ocean.

The vertical curl of surface wind stress is $({\nabla }_h\times \mathbf{\tau }{)}_z={\tau }_{y,x}-{\tau }_{x,y}$. In a layer of density ${\rho }_0$ and depth $h$, its normalized potential vorticity source is $W=({\nabla }_h\times \mathbf{\tau }{)}_z/({\rho }_{0}h)$. It drives Sverdrup balance in the weakly nonlinear steady ocean interior.

An Ekman layer is a viscous boundary layer in a rotating fluid. For constant kinematic viscosity $\nu$ and Coriolis parameter $f$, its characteristic thickness is ${\delta }_E=\sqrt{2\nu /|f|}$, and its velocity turns with distance from the boundary.

For $f>0$, a stationary lower no-slip boundary condition below uniform exterior geostrophic flow $W_g$ produces $W_E=-W_g{e}^{-(1+i)z/{\delta }_E}$. The upward shear component at the bottom is $T_b=\rho \nu W_E^{\prime }(0)=\rho \sqrt{f\nu /2}(1+i)W_g$; the bottom traction on the fluid is $-T_b$. This law cannot be assigned to an arbitrary prescribed surface wind stress without an additional boundary condition.

A surface Ekman layer is the stress-driven boundary correction below a rotating fluid’s upper boundary. For upward $z\le 0$, $f>0$, constant kinematic viscosity and imposed complex wind stress $T$, its steady velocity is $W_E=T{e}^{\lambda z}/(\rho \nu \lambda )$, $\lambda =(1+i)/{\delta }_E$. The underlying geostrophic flow is independent of $T$ unless an additional boundary velocity condition couples them.

A viscous no-slip boundary below a geostrophic current creates a Bottom Ekman layer of thickness $\delta =(2\nu /|f|{)}^{1/2}$. Its depth-integrated ageostrophic transport has a component opposite the geostrophic current and a transverse component set by the sign of $f$.

Ekman transport is the depth-integrated ageostrophic flow in a frictional rotating boundary layer. Beneath a surface stress $\mathbf{\tau }$,with the stated orientation convention.

Spatial convergence or divergence of Ekman transport drives vertical motion at the base of the Ekman layer. With upward vertical coordinate,

In a thin Bottom Ekman layer, slowly evolving geostrophic flow drives upward Ekman pumping into a uniform-depth exterior layer. With $f>0$, $\alpha =\sqrt{f\nu /2}$ and $R_D=\sqrt{gH}/f$, its balanced height satisfies $({\nabla }_h^2-R_D^{-2}){\eta }_t=-(\alpha /H){\nabla }_h^2\eta$. Small-scale Fourier modes decay on $H/\alpha$; large-scale modes obey a diffusion equation with diffusivity $g\alpha /f^2$. The balanced approximation requires $|f|\tau \gg 1$.

Bottom Ekman pumping damps a horizontal quasi-geostrophic Fourier mode while its influence penetrates vertically only over $|f|/(Nk)$. For a half-space with a vertically uniform initial mode, the boundary amplitude decays exponentially while the remote interior remains unchanged.

An ocean gyre is a basin-scale rotating circulation driven principally by wind stress and shaped by planetary rotation and boundaries.

Sverdrup balance equates meridional advection of planetary vorticity to wind-stress curl in the weakly frictional ocean interior.

With depth transport $(-{\psi }_y,{\psi }_x)$ and a depth-mean bottom-drag closure, the wind-driven equation is $\beta {\psi }_x=S-r{\nabla }^2\psi +\nu {\nabla }^4\psi$, where $S$ is wind-stress curl divided by density and $r=\gamma /({\rho }_{0}H)$ is the drag rate. Its zonal boundary modes solve $\nu {\lambda }^3-r\lambda -\beta =0$. The Munk boundary layer and Stommel boundary layer are different limits of this combined equation.

If $r^3\gg {\beta }^2\nu$, the broad western Stommel boundary layer has thickness $r/\beta$, but small viscosity supplies no-slip skins of thickness $(\nu /r{)}^{1/2}$ at both east and west walls. The western solution therefore contains two decaying exponentials, while the east has the rapidly decaying skin. Setting viscosity identically to zero removes the modes needed to satisfy all four lateral no-slip conditions.

For a uniform normalized upper-layer wind stress curl $W$ and no lower-layer forcing, a uniform local Sverdrup balance isThe upper layer flows meridionally and the lower layer rests. Although the relative vorticity vanishes, the potential vorticity has the zonal gradients ${\bar{q}}_{1,x}=-FV$, ${\bar{q}}_{2,x}=FV$. These are essential for the perturbation stability and do not vanish merely because the background velocity is uniform.

Let $K^2=k^2+l^2$, $A=K^2+F$, $\mathcal{B}=K^2(K^2+2F)$ and use $e^{i(kx+ly-\sigma t)}$. The uniform upper-layer flow $V$ over a resting lower layer has the dispersion relationIts discriminant is $4F^2{\beta }^2{k}^2+\mathcal{B}{K}^2(K^2-2F)l^2{V}^2$. A negative value gives exponential baroclinic instability. This model includes advection across the interface-induced potential-vorticity gradients in a meridional two-layer current.

For purely meridional wavevectors, $k=0$, $l\ne 0$, the growth rate of a uniform meridional upper-layer flow $V$ over a resting lower layer isGrowth requires $V\ne 0$ and $0<l^2<2F$. The perturbation has no northward velocity, so planetary vorticity advection does not explicitly stabilize it. At fixed normalized wind stress curl, $V=W/\beta$ means stronger $|\beta |$ nevertheless reduces the shear and growth rate.

For ${\bar{\psi }}_1=Vx$ and ${\bar{\psi }}_2=0$, the two-layer quasi-geostrophic potential vorticity is ${\bar{q}}_1=f_0+\beta y-FVx$, ${\bar{q}}_2=f_0+\beta y+FVx$. A perturbation therefore feels both the northward gradient $\beta$ and opposite zonal gradients $\mp FV$. Dropping the zonal gradients removes the instability mechanism from the linearization.

The unbounded constant-depth steady transport response to a unit point wind-curl source solves $r{\nabla }_h^2\psi +\beta {\psi }_x=H\delta (x)\delta (y)$ for $r,\beta >0$. An exponential substitution reduces it to a modified Helmholtz equation, yielding the expression shown in terms of the Modified Bessel function of the second kind. Its finite-level streamfunction contours are nearly circular near the source and stretched into a western wake far away. The wake width is of order $\sqrt{(r/\beta )|x|}$.

An ocean transport streamfunction defines depth-integrated horizontal transport. For constant depth $H$, it is $H$ times the velocity streamfunction. Its difference between two boundaries gives signed transport across a connecting line.

The ocean transport streamfunction convention $Hu=-{\psi }_y$, $Hv={\psi }_x$ gives relative vorticity $\zeta =H^{-1}{\nabla }^2\psi -H^{-2}\nabla H\cdot \nabla \psi$. This identity is exact for variable positive depth and does not assume that depth variations are small.

For an island in a constant-depth ocean with Sverdrup balance, choose a counterclockwise contour running from the island's southern tip east to the basin wall, north along that wall, west to the northern tip, and south along the island's western coast. If that contour avoids dissipative boundary layers, the ocean transport streamfunction on the island isHere $\mathbf{W}$ is the wind acceleration appearing in the depth-averaged momentum equation. The basin-boundary streamfunction is zero. This integral fixes the otherwise undetermined constant on the island without requiring a friction law, as in Godfrey's original island-circulation model.

When depth varies but the Coriolis parameter is constant, forcing can balance advection of $f/H$ rather than advection of planetary vorticity. This topographic potential-vorticity gradient plays the role normally played by the beta effect.

With $f=f_0+\beta y$ and parabolic depth, the topographic potential-vorticity pseudovelocity is $(-\beta /H,-f{H}_x/H^2)$. For positive $f,\beta$, it points west and turns south on the western slope, north on the eastern slope. Its characteristics are the parabolas $y=Qax(L_x-x)/\beta -f_0/\beta$, where $Q=f/H$. Wind-driven ocean gyres are steered by these depth gradients; drag and boundary data determine the actual ocean transport streamfunction. The zero-depth shoreline is a singular limit that requires a separate physical regularization.

In the steady small-Rossby number transport equation, this coefficient field advects the ocean transport streamfunction along contours of background shallow-water potential vorticity $f/H$. It is not a physical velocity: when $\psi$ has units of volume transport, its units are inverse length squared per time. For uniform depth on a beta plane it points westward, $\tilde{\mathbf{u}}=(-\beta /H,0)$. With no forcing or drag, $\psi$ is constant on each connected characteristic; locally $\psi =\Phi (f/H)$ at regular points.

In a slowly varying inviscid layer, parcels tend to follow contours of $f/H$. If depth increases in the direction of motion, conservation of potential vorticity requires a poleward displacement or a compensating change of relative vorticity. Ekman pumping permits streamlines to cross those contours.

A western boundary current is a narrow, intense return flow that closes a broad wind-driven interior circulation. The beta effect selects the western side for this frictional boundary layer.

An inertial closure balances advection of relative vorticity with planetary vorticity advection. If a basin of width $L$ and meridional size $B$ has interior speed $V_I$, transport continuity gives $V_b\delta \sim V_{I}L$. Equating $V_b^2/(B\delta )$ with $\beta V_b$ yields ${\delta }_I^2\sim V_{I}L/(\beta B)$, reducing to $V_I/\beta$ when $B$ and $L$ are comparable. This is a scale estimate rather than a full boundary-current profile.

The Stommel boundary layer is a frictional western boundary current that returns the transport carried by a broad wind-driven Sverdrup balance interior. In a flat-bottomed, linear beta plane model, use streamfunction $\psi$ with $(u,v)=(-{\psi }_y,{\psi }_x)$ and dimensional linear drag acceleration $-r\mathbf{u}$, where $r>0$ has units of inverse time. The steady vorticity equation iswhere $\beta =df/dy>0$ and $F$ is the wind-stress-curl forcing per unit mass and depth.

For a narrow boundary correction near the western wall $x=0$, the leading balance isIts decaying solution is ${\psi }_{\mathrm{b}}=C(y)e^{-x/{\delta }_S}$, withThe dimensions are length because $[r]=T^{-1}$ and $[\beta ]=L^{-1}{T}^{-1}$. If a dimensionless drag parameter $\gamma$ is defined by $r=\gamma f_0$ for positive reference Coriolis parameter $f_0$, the same width is $\gamma f_0/\beta$. For a dimensional drag term $-\gamma {\nabla }^2\psi$ in the vorticity equation, the width instead reads $\gamma /\beta$.

The exponentially decaying correction can enforce the western no-normal-flow boundary condition while leaving the interior Sverdrup balance approximately unchanged. The underlying linear wind-driven model and the role of latitude-dependent rotation were established by Henry Stommel. ^[ref]

For streamfunction $\psi =\Psi (x)\cos (my)$ with impermeable walls at $x=0,L$, let $A=-F_0/(r{m}^2)$ and positive exponents satisfy $\lambda -\gamma =\beta /r$, $\lambda \gamma =m^2$. The exact profile is $\Psi =A[1-C_W{e}^{-\lambda x}-C_E{e}^{\gamma (x-L)}]$, where $C_W=(1-e^{-\gamma L})/(1-e^{-(\lambda +\gamma )L})$ and $C_E=(1-e^{-\lambda L})/(1-e^{-(\lambda +\gamma )L})$. Substitution proves the ordinary differential equation and both boundary conditions. The usual narrow western boundary current limit requires $\lambda L\gg 1$ and $\gamma L\ll 1$, with $\lambda \simeq \beta /r$ and $\gamma \simeq r{m}^2/\beta$; the opposite first inequality does not describe weak drag.

If the dimensional linear drag acceleration is $-r\mathbf{u}$, then $r$ has units of inverse time and Stommel boundary layer balance $\beta v=-r{v}_x$ gives ${\delta }_S=r/\beta$. For returned volume flux $Q$ in depth $H$, the boundary speed is $Q/(H{\delta }_S)$ and relative vorticity is of order $Q/(H{\delta }_S^2)$. A gyre return varying along a distance $B$ has a generic nonlinear-to-planetary vorticity advection ratio $V_b/(\beta B{\delta }_S)$. Large relative vorticity alone does not force nonlinear advection: a perfectly parallel $v=v(x)$, $u=0$ has zero material relative-vorticity advection.

A Munk boundary layer closes a Sverdrup balance interior using lateral viscosity. At a zonal wall, balancing $\beta {\psi }_x$ with $\nu {\nabla }^4\psi$ gives the width $(\nu /\beta {)}^{1/3}$.