Coriolis & Global Winds · Guide · glidepathscience.org
GuideCoriolis & Global Winds
Contents
This Guide explains the ideas and equations behind every Walkthrough step, Concept Challenge question and Predict round. Part 1 prepares the Intro track and Part 2 the Advanced track; each takes about ten minutes. The worked examples use places and numbers that are not the sim’s own problems.
Part 1 · The essentials
Math you’ll need: algebra, the sine of an angle, and unit conversions.
One motion, two frames
Picture a puck sliding across a spinning turntable with no friction. Someone standing in the room sees it go in a straight line: nothing pushes it sideways, so by Newton’s first law it keeps its velocity. A rider on the turntable sees something else. While the puck slides, the disk turns under it, so the straight line, traced on the turning floor, comes out curved. Both observers are right; they describe one motion from two frames.
The rider can still use Newton’s laws, but only by adding forces that have no agent: a sideways Coriolis force on anything that moves across the disk, and an outward centrifugal force. These are called fictitious forces, and the sim draws them dashed. On a disk turning counterclockwise seen from above, the Coriolis force bends motion to the right; reverse the spin and it bends motion to the left.
Earth is a turntable too. Seen from above the North Pole it turns counterclockwise, so in the Northern Hemisphere moving air and water are deflected to the right of their motion. Seen from above the South Pole it turns clockwise, so in the Southern Hemisphere they are deflected to the left. The deflection is not a push from the ground; it is how motion looks to observers who turn with Earth, which is all of us.
How fast Earth turns, and the Coriolis parameter
Earth turns once relative to the stars in a sidereal day, 86,164 s, a few minutes shorter than the 24 h solar day because Earth also moves along its orbit. Its rotation rate is one turn, 2π radians, in that time:
$Ω = 2π/T_{sid}$
A turntable at 10 turns a minute spins about 14,000 times faster. Earth’s turning is slow, but it never stops, and winds travel for days.
Not all of that spin turns a horizontal wind. Only the part about the local vertical counts, and the local vertical lines up with Earth’s axis only at the poles. At latitude $φ$ the part about the vertical is $Ω\sin φ$, and the Coriolis parameter is defined as twice that:
$f = 2Ω\sin φ$
It is zero at the equator, where the vertical is at right angles to the axis, largest at the poles ($2Ω$), and negative in the Southern Hemisphere, which is the sign of the leftward bend there.
How big is the Coriolis acceleration?
Something moving horizontally at speed $U$ has a Coriolis acceleration of size
$a_C = fU$
at right angles to its motion. Because it is always sideways, it changes the direction of motion but never the speed: it does no work.
Accelerations this small are never felt in a car: the road’s friction cancels them at every instant. What makes them matter for the air is time. Nothing but pressure differences opposes them, and a wind keeps blowing for hours or days over hundreds of kilometers. A useful clock is the time Coriolis takes to turn a flow by one radian (57°), $t = 1/f$: at 60° N, 1 / 1.263 × 10⁻⁴ s⁻¹ = 7,918 s, or 2.2 h.
That clock also settles the bathtub-drain myth. A sink drains in under a minute, far too short for Coriolis to turn the water noticeably. The swirl left over from filling it, and the basin’s shape, decide which way it drains, in either hemisphere. Only in a large, symmetric tank left to settle for a day has Earth’s effect been seen (Shapiro, 1962).
The pressure push and geostrophic balance
Air is pushed from high pressure toward low pressure. On a weather map, lines of equal pressure are isobars; if neighboring isobars differ by $Δp$ and are a distance $Δn$ apart, the push per kilogram of air of density $ρ$ is the pressure-gradient acceleration
$a_P = (1/ρ)Δp/Δn$
pointing across the isobars toward low pressure. Air released from rest starts moving that way. As it speeds up, Coriolis turns it (to the right in the north), and the turning goes on until the wind blows along the isobars. There the Coriolis force, at right angles to the wind, points straight away from the low and exactly balances the pressure push. This is the geostrophic wind: $fv_g = a_P$, so
$v_g = Δp/(ρfΔn)$
Starting from rest, the parcel overshoots and swings about the balance, as the figure shows; in real air these swings spread away and the wind settles. Closer isobars mean a stronger wind. Nearer the equator $f$ is smaller, so the same isobars give a stronger wind. Only at the equator itself, where $f$ = 0, does the balance have no solution at all. Close to it, though, the balanced wind would be far faster than real winds, and a wind’s own curving and friction matter as much as Coriolis. The Rossby number measures this: $Ro = U/(fL)$ for a wind of speed $U$ curving around a system of size $L$, and geostrophic balance is a good guide only when $Ro$ is well below 1. That is why the Geostrophic Map draws no arrows within 5° of the equator: a cutoff the map chooses, not a law of nature.
Since the low is on the side the pressure push points to, and the push balances a Coriolis force to the wind’s right, the low is on the wind’s left in the Northern Hemisphere. That is Buys Ballot’s law (1857): stand with your back to the wind and low pressure is on your left in the north, on your right in the south.
Friction near the ground, and why storms spin
In the lowest kilometer or so, friction with the ground slows the wind. A slower wind feels less Coriolis force but the same pressure push, so the push wins a little and turns the wind across the isobars toward the low. Near the ground, then, the wind is weaker than geostrophic and crosses the isobars toward low pressure, typically by a few tens of degrees over land and less over the sea.
Now look at a low from above. Air near the ground flows in toward it from every side. In the Northern Hemisphere each inflowing stream is bent to its own right, and from every side that adds up to a counterclockwise circulation around the low; in the Southern Hemisphere, clockwise. Hurricanes and midlatitude storms turn this way. Coriolis does not create the low: the pressure difference draws the air in, and Coriolis sets the sense of the spin. Around a high, air flows outward near the ground and turns the other way.
The three cells and the wind belts
Sunlight heats the tropics most. Warm, moist air rises near the equator in a band of clouds and rain, the ITCZ (intertropical convergence zone), flows poleward high in the troposphere, and sinks near 30° latitude, where the dry, sinking air builds the subtropical highs and many of the world’s deserts. It returns toward the equator along the surface. This loop is the Hadley cell.
The returning surface air is turned by Coriolis: to the right in the north, so air heading south comes from the northeast, and to the left in the south, so air heading north comes from the southeast. These are the steady trade winds. Between about 30° and 60° the surface air flows poleward and is turned east: the westerlies, the belt in which most midlatitude weather travels. Beyond 60°, near the subpolar lows, cold polar air flows outward as the polar easterlies. The middle loop, the Ferrel cell, is the average effect of the storms that pass through it, and the polar cell closes the picture.
Part 2 · Going further
Math you’ll need: Part 1, vectors and the cross product, derivatives, a quadratic equation; the inertial-oscillation section solves a pair of first-order differential equations.
Newton’s law in a turning frame, and the f-plane
In a frame turning at a steady rate $\vec Ω$, the acceleration measured there is
$\vec a′ = \vec F/m − 2\vec Ω × \vec v′ − \vec Ω × (\vec Ω × \vec r′)$
The last term, the centrifugal acceleration, depends only on position, so on Earth it is folded into gravity: effective gravity is $\vec g = \vec g^* − \vec Ω × (\vec Ω × \vec r)$, and it defines the local vertical (Earth’s slightly flattened shape keeps the ground at right angles to it). The sim’s dish does the same for the turntable: its slope cancels the centrifugal term, and the rider sees only Coriolis.
With $x$ east, $y$ north and $z$ up, the horizontal Coriolis terms are $du/dt = fv$ and $dv/dt = −fu$. The terms with $2Ω\cos φ$ couple horizontal and vertical motion; for a thin atmosphere moving almost horizontally they are dropped, the traditional approximation. Holding $f$ fixed at one latitude gives the f-plane, the model of every map in this sim.
Inertial oscillations and the Rossby number
With no pressure gradient and no friction, $du/dt = fv$ and $dv/dt = −fu$ give $d^2u/dt^2 = −f^2u$: the velocity keeps its size $U$ and turns at the rate $f$. The parcel moves on an inertial circle,
$R = U/|f|$, $T = 2π/|f|$
clockwise in the north. The period is half a pendulum day: a Foucault pendulum at the same latitude turns once in $2T$. After a storm passes, ocean currents often trace these loops for days.
Whether rotation matters is decided by comparing the acceleration a flow needs to curve over its own size $L$, about $U^2/L$, with the Coriolis acceleration $fU$. Their ratio is the Rossby number, $Ro = U/(fL)$. A midlatitude cyclone has $Ro$ well below 1 and is nearly geostrophic; a tornado or a sink has $Ro$ in the thousands, and Coriolis plays no part. Toward the equator $f$ shrinks, so $Ro$ grows for any flow: for the Geostrophic Map’s centers (20 hPa deep, $L$ = 450 km, $U$ their fastest geostrophic wind) it is 0.46 at 45°, 2.0 at 20° and 30 at 5°.
Geostrophic, gradient and cyclostrophic balance
Geostrophic balance in vector form is $f\hat k × \vec v_g = −(1/ρ)∇p$: the wind runs along the isobars at $|\vec v_g| = Δp/(ρ|f|Δn)$. Upper-air maps show the height $z$ of a pressure surface instead; there $−(1/ρ)∇p$ becomes $−g∇z$ and
$|\vec v_g| = (g/|f|)Δz/Δn$
with no density at all, so one contour spacing means one wind at any level. For example, 500 hPa heights 40 m apart every 250 km at 40° N give 9.81 × 40 / (9.375 × 10⁻⁵ × 250,000) = 16.7 m/s.
When the isobars curve, the flow on a circle of radius $R$ also needs a centripetal acceleration $V^2/R$. Around a low, $V^2/R + |f|V = (1/ρ)∂p/∂n$, and the positive root is the gradient wind, $V = −fR/2 + \sqrt{f^2R^2/4 + Rfv_g}$, weaker than $v_g$. Around a high the curvature adds to the pressure push, $|f|V − V^2/R = (1/ρ)∂p/∂n$, and the wind is stronger than $v_g$; but a solution exists only while $∂p/∂n ≤ ρf^2R/4$, which caps the pressure gradient near a high’s center and keeps its winds gentle.
In a tornado or dust devil Coriolis drops out and the pressure push supplies only the centripetal acceleration, the cyclostrophic balance $V^2/R = (1/ρ)∂p/∂n$. It allows either sense of spin; most tornadoes turn cyclonically because the storms that spawn them do. A dust devil with a gradient of 2 Pa/m at 5 m from its center turns at √(5 × 2 / 1.225) = 2.86 m/s.
The friction layer and the thermal wind
Near the ground, turbulence carries momentum down to the surface. The sim models it as a linear drag, $−k\vec v$; the steady wind is then slower than geostrophic by the factor $|f|/\sqrt{k^2 + f^2}$ and crosses the isobars toward low pressure at $α = \arctan(k/|f|)$. A fuller treatment, the Ekman layer, gives a spiral of wind with height, but the result is the same: around a low, the friction layer’s inflow converges and must rise, so lows bring clouds and rain and slowly fill (Ekman pumping); around a high the layer diverges and air sinks, bringing clear skies.
Higher up, the geostrophic wind changes with height wherever the temperature changes horizontally. Pressure falls faster with height in cold air than in warm, so where it is colder toward the pole the pressure surfaces slope down poleward more steeply with height, and the westerly wind grows:
$Δu = −(g/(fT))(∂T/∂y)Δz$
This is the thermal wind. The sharpest temperature contrasts concentrate it, and the jet stream sits above them.
Angular momentum and the Hadley cell’s edge
A ring of air around the globe that leaves the equator at rest, high in the Hadley cell, and moves poleward without friction keeps its angular momentum about Earth’s axis, $M = (Ωa\cos φ + u)a\cos φ$. Closer to the axis it must spin faster, and its eastward wind becomes
$u = Ωa\sin^2φ/\cos φ$
That is 32.2 m/s at 15°, but well over 100 m/s at 30°, far faster than real subtropical jets. Eddies carry angular momentum away, and a ring that fast would be unstable: the cell cannot carry it much farther. Held and Hou (1980) balanced this wind against the thermal wind of the heated tropics and found the edge, for small angles, at $φ_H = [5gHΔ_H/(3Ω^2a^2)]^{1/2}$, with $H$ the tropopause height and $Δ_H$ the radiative equator-to-pole temperature difference as a fraction. Typical values put the estimated edge close to the observed one. The estimate scales as $1/Ω$: for a planet turning half as fast it doubles, past where the small-angle form can be trusted, but the trend is real. Slowly turning planets have wide Hadley cells.
Vorticity: how storms spin up
A column of air carries the planet’s spin, $f$, plus its own, the relative vorticity $ζ$ (positive counterclockwise seen from above). Without friction, the absolute vorticity times the column’s area is kept, $(ζ + f)A$ = constant, just as a spinning skater who pulls in their arms speeds up. Air converging into a low shrinks its columns, so $ζ$ grows with the sign of $f$: cyclonic, counterclockwise in the north and clockwise in the south.
This is how Earth’s slow turning becomes a storm’s fast one: convergence concentrates the planet’s vorticity, and its sign is fixed by the hemisphere.
Reference
Common mistakes
Part 1
- Leaving out the 2 in $f = 2Ω\sin φ$, or the sine of the latitude.
- Thinking Coriolis speeds the wind up or slows it down: it only turns it.
- Expecting winds to blow straight from high to low: aloft they blow along the isobars.
- Saying Coriolis causes storms: pressure differences do; Coriolis sets their spin.
- Believing sinks drain one way in each hemisphere: they are far too small and fast.
- Forgetting to convert hPa to Pa and km to m.
Part 2
- Taking the negative root of the gradient-wind quadratic.
- Using density in the geostrophic wind on a height map.
- Treating the Held–Hou edge as exact: it is an axisymmetric, small-angle estimate.
The equation sheet
| Part 1 | |
|---|---|
| Coriolis parameter | $f = 2Ω\sin φ$, $Ω$ = 7.292 × 10⁻⁵ rad/s |
| Coriolis acceleration | $a_C = fU$ |
| pressure push | $a_P = (1/ρ)Δp/Δn$ |
| geostrophic wind | $v_g = Δp/(ρfΔn)$ |
| turning time | $t = 1/f$ |
| Part 2 | |
|---|---|
| inertial circle | $R = U/|f|$, $T = 2π/|f|$ |
| Rossby number | $Ro = U/(fL)$ |
| on a pressure surface | $|\vec v_g| = (g/|f|)Δz/Δn$ |
| gradient wind (low) | $V = −fR/2 + \sqrt{f^2R^2/4 + Rfv_g}$ |
| cyclostrophic | $V^2/R = (1/ρ)∂p/∂n$ |
| thermal wind | $Δu = −(g/(fT))(∂T/∂y)Δz$ |
| angular-momentum ring | $u = Ωa\sin^2φ/\cos φ$ |
$Ω$ Earth’s rotation rate, $φ$ latitude, $f$ Coriolis parameter (s⁻¹), $U$ and $V$ speeds, $ρ$ = 1.225 kg/m³ air density, $Δp$ pressure difference across a spacing $Δn$, $v_g$ geostrophic wind, $R$ radius, $T$ period, $L$ size of a flow, $g$ = 9.81 m/s², $Δz$ height difference, $T$ in the thermal wind the mean temperature (K), $a$ = 6371 km Earth’s radius, $u$ eastward wind, $ζ$ relative vorticity.
Further reading
- J. R. Holton and G. J. Hakim, An Introduction to Dynamic Meteorology, 5th ed. (Academic Press, 2013).
- I. M. Held and A. Y. Hou, “Nonlinear axially symmetric circulations in a nearly inviscid atmosphere,” Journal of the Atmospheric Sciences 37, 515–533 (1980).
- A. Persson, “How do we understand the Coriolis force?”, Bulletin of the American Meteorological Society 79, 1373–1385 (1998).
- OpenStax, University Physics Volume 1, Chapter 6 (Applications of Newton’s Laws): inertial forces and the Coriolis force.
- This sim’s Field Manual (H): the model and what it leaves out.
Coriolis & Global Winds v1.0 · © 2026 Dr. David Meier · CC BY-NC-SA 4.0 · glidepathscience.org