Desert latitudes

Why do the world's great deserts cluster along the same two bands of latitude?

Lay a globe flat and the driest places arrange themselves suspiciously. The Sahara, Arabia, the Thar, the Sonoran and Chihuahuan deserts sit in one band north of the tropics; the Kalahari, the Namib, the Australian interior and the Atacama sit in a matching band to the south. Two stripes, roughly mirrored, at roughly the same distance from the equator.

Coincidence is not available at that scale. And the obvious guess — that deserts are where it is hottest, so the sun is nearest overhead — fails immediately, because the equator itself, which gets the most sun of all, is where the rainforests are. So what does latitude actually control?

Begin with what a desert is. Not hot: dry. The working definition is about the water budget — less water arriving as precipitation than the atmosphere could evaporate away. Antarctica qualifies. So the question is not where the sun is strongest, but where rain does not fall.

When does rain fall? Only when air rises. Air holds water vapour, and how much it can hold falls steeply as it cools; lift a parcel and it expands against lower pressure, cools, and past some point must shed its vapour as cloud and rain. Sinking air does the reverse — it is compressed, warms, and its capacity to hold water grows, so cloud evaporates rather than forming. That gives a clean rule to reason with: rising air rains, sinking air does not.

So the question becomes geometric. Where is the atmosphere rising, and where does it come back down? Near the equator the surface takes the most solar heating, warms the air above it, and that air rises — which is why the equator is wet. Follow that parcel: it rises, rains its water out, and now sits high, cold and dry, with more air rising beneath it. It has to go somewhere, so it spreads poleward.

Now, where does it come down? Not at the pole. As the poleward-moving air aloft conserves angular momentum it gains an ever stronger westerly component, and by roughly 30 degrees this flow can no longer be sustained — the circulation closes and the air descends. That descent is the Hadley cell's return arm, and it produces a belt of persistent high pressure at the surface near 20 to 35 degrees, the old sailors' horse latitudes. The air arriving there has two strikes against it: it lost its moisture over the equator, and it warms by compression on the way down, which drops its relative humidity further still. Clear skies, sinking motion, nothing to lift a parcel. That is the two stripes.

Does that finish it? It should not, and here the neat story needs policing. Test it: the Gobi is near 43 degrees, well outside the belt. The Patagonian desert is near 45. The Atacama and Namib are coastal, in latitudes where you might expect the sea to help. So subsidence cannot be the whole account.

Three other mechanisms do real work. A mountain range forces air upward on its windward side, wringing the water out there, so the leeward side receives air that has already rained — a rain shadow, which makes the Great Basin behind the Sierra Nevada and Patagonia behind the Andes. Distance from any ocean makes another kind: air travelling far inland rains as it goes, and by the deep interior of Asia there is little left, which is much of the Gobi and Taklamakan. And a cold upwelling current makes the strangest kind: cold water chills the air from below, so the lowest layer is colder than the air above it. That inversion is stable, the air cannot rise, and the result is a coast frequently foggy and almost never rained on — the Atacama and the Namib.

Notice that the Atacama is all three at once: subtropical subsidence, the cold Humboldt upwelling offshore, and the Andes blocking moisture from the Amazon side. The mechanisms stack.

The two refinements that keep this honest are the same one twice: latitude is a strong tendency, not a law.

First, the belt is not a fixed line. The whole circulation shifts with the sun through the year, which is why the desert margins — the Sahel, for instance — get a short wet season as the rain belt swings toward them, and why a small shift in that timing is a large event for anyone living there. Second, latitude only sets the background; local geography can override it in either direction. Coastal Ecuador sits on the equator and has an arid stretch, thanks to the same cold current; parts of the subtropics are wet because monsoon circulations reverse the flow seasonally.

I have also simplified the mechanics. Why the Hadley cell closes near 30 rather than 50 involves the balance between angular momentum and thermal driving, and the modern treatment is more subtle than "the air runs out of momentum". There is also active research on whether the belt is widening as the climate warms, and the size of that trend is contested.

• Why the rain belt's seasonal migration makes desert margins so vulnerable.
• How the same descending air makes the subtropical ocean gyres nutrient-poor.

Every term the collection defines is gathered in the glossary.