Katabatic winds

Why does air on the Antarctic plateau reach gale force downhill with no storm system driving it?

Wind, we are taught, is air moving from high pressure to low, and the pressure pattern is made by weather systems — so a gale implies a storm somewhere. Yet on the slopes of the Antarctic ice sheet a gale blows through a cloudless sky, day after day, from very nearly the same bearing, with no low-pressure centre to point to.

Suppose the storm is simply absent rather than hidden. What could a bare, sloping, freezing surface do to the air resting on it?

Start with what the surface does through a polar winter that is one long night. Snow radiates infrared away efficiently and receives little back from a dry, cloudless sky, so it cools below the air above it and chills a shallow layer of that air by contact. That is the ordinary radiative-cooling story that whitens a garden lawn; the interesting part here is that the garden is tilted, and a thousand kilometres long.

Cold air is denser. At fixed pressure density goes roughly as the inverse of absolute temperature, so a layer 10 degrees colder than its surroundings, at around 250 kelvin, is about four per cent denser — 10 divided by 250. A denser fluid sitting on a slope beneath a lighter one is not in equilibrium: the component of gravity along the slope acts on the density excess. The acceleration is g times the fractional density excess times the sine of the slope angle. Take a slope of one in a hundred, which is a fair figure for the coastal flank of the ice sheet: 9.8 times 0.04 times 0.01 gives about 0.004 metres per second squared, a four-hundredth of gravity. Feeble.

Now ask the question that changes everything: feeble for how long? Let that acceleration act, without friction, over a hundred kilometres of continuous slope. The speed reached is the square root of twice 0.004 times 100,000 metres — about 28 metres per second. Over ten kilometres it would be only 9; over three hundred, nearly 50. So the answer to "gale without a storm" is not a strong force at all. It is a trivial force applied over an enormous fetch, which is exactly what a continental ice sheet uniquely provides. Friction and turbulent mixing hold the real wind well below that frictionless ceiling, but the ceiling is where the gale comes from.

One objection should be nagging. Air descending three kilometres from the plateau to the coast is compressed, and compression warms it by about 9.8 degrees per kilometre — some 29 degrees over the descent, far more than the 10-degree deficit we started with. Should the flow not warm itself out of existence?

It would, if the deficit were measured against a fixed thermometer. It is not: it is measured against the air being displaced at the same level, which is compressed equally. What survives descent unchanged is potential temperature — temperature corrected to a common pressure — and it is that deficit which supplies the buoyancy. So the wind arrives much warmer than it left and still colder than its neighbours. A foehn wind differs precisely here: foehn air was never chilled at the surface, so it arrives with no deficit and feels hot.

Two more things shape the flow rather than drive it. The Earth's rotation deflects it — to the left in the southern hemisphere — so the wind crosses the fall line at an angle of tens of degrees, with surface friction pulling it partway back. And terrain funnels it: where a broad catchment of slope drains into a concave coastal valley, the converging flow far exceeds what the local slope alone would give, which is why Cape Denison is a byword for wind.

Is there a feedback? Yes, and it is the brake rather than the engine. A faster wind mixes more vigorously, dragging warmer air down from above into the cold layer, which erodes the very deficit that drives it. Strong katabatic flow tends to be self-limiting. The forcing is the radiative deficit; the feedback is entrainment working against it.

How would we know any of this is right? If these winds were driven by passing pressure systems, their direction would swing as those systems came and went. Instead slope stations record extreme directional constancy — the wind returns to a bearing set by the local terrain, and that bearing differs at a station a hundred kilometres away with a different aspect. A slope station whose direction tracked the synoptic charts rather than the hillside, or one that blew hardest under thick cloud when radiative cooling was shut off, would kill the mechanism proposed here. What the records cannot settle is trend: the Antarctic automatic weather station network only becomes dense in the 1980s, and interannual swings tied to the large-scale circulation are large, so I would not offer a number for how katabatic strength is changing.

Three neighbouring ideas are easy to conflate with this one. Radiative frost explains why a surface layer gets cold and stops there; katabatic wind is the sequel, when that same layer sits on a gradient. A sea breeze needs a horizontal contrast between two different surfaces, whereas here one surface supplies the chill and gravity supplies the direction. And the geostrophic rule — wind following the isobars — describes the free atmosphere above the friction layer; a katabatic flow is often only tens to a couple of hundred metres deep, so buoyancy and friction dominate and it blows firmly across the isobars.

The misconception worth naming is that katabatic means "cold wind". It means a wind driven by negative buoyancy on a slope; its arrival temperature can be unremarkable, and what matters is that it is colder than the air it displaces at the same height.

• Why a katabatic surge sometimes ends in a hydraulic jump, with a visible wall of blowing snow.

Every term the collection defines is gathered in the glossary.