Issue No. 17 put a hurricane on the table as a heat engine, then let one clause do too much work: the gap between the warm ocean underneath and the cold air on top “sets a hard limit on how strong any single storm can get.” True, and a promise — and it moved on without the number.
There is a number. It is called the potential intensity, and it can be computed for any square of ocean on any day of the year. What earns it a whole issue is not that storms obey it. It is that they obey it and then stop well below it, apparently at random.
Two temperatures and a cube
Treat the storm as a turbine and the accounting goes plain. The intake is the sea surface, around 300 kelvin. The exhaust is the outflow near the top of the troposphere, where the storm dumps its leftover heat, around 200 kelvin. A Carnot engine between those reservoirs converts about a third of what it takes in — (300 − 200) ÷ 300 — into work.
The rest is a race between two quantities that scale differently. Energy comes in as wind strips warm, moist air off the sea, and that pickup rises in rough proportion to wind speed. Energy goes out as friction, which rises with the cube of it. A cube wins eventually, and where the curves cross is the ceiling — typically about 80 metres per second, 180 miles an hour, though it runs anywhere from zero to 100.
The best part is what becomes of the friction. It does not leave the budget. It turns the storm’s motion into heat a few tens of metres above the sea — precisely where the storm takes on fuel — so the waste feeds back into the intake. Bister and Emanuel worked out the consequence in 1998: in the efficiency term, the hot reservoir in the denominator is replaced by the cold one. (300 − 200) ÷ 300 becomes (300 − 200) ÷ 200; a third becomes a half; the ceiling rises about 20 per cent. The engine is stronger because it burns its own exhaust.
Nobody gets there
In 2000, Emanuel took the best-track records for the North Atlantic and the western North Pacific and divided each storm’s lifetime peak wind by the computed ceiling at that place and time, discarding every storm whose peak was cut short by land or cold water. Fifty-six Atlantic storms and seventy-three Pacific ones survived the filter.
The distribution is nearly a straight line. Not a peak, not a hump, not a long tail — a flat likelihood. A hurricane is about as likely to top out at four-tenths of its ceiling as at six-tenths, or at nine-tenths. And at the top the line simply stops: “there is virtually no possibility of exceeding potential intensity.” The ceiling is a wall. Everything under it is a coin toss.
Almost nothing else in geophysics looks like this. Earthquake magnitudes are lognormal, piled up at the small end and thinning away, and so are most natural intensity records. A flat distribution with a hard edge is close to unique, and Emanuel wrote plainly that it “begs a physical explanation” he did not have.
The flatness is itself the clue. If whatever holds storms down depended on how strong a storm had already become, the line would bend; a flat line says the brake arrives without regard to where the storm is in its climb. Idealised models feel no such brake — every one of them drives its storm right to the ceiling and parks there. That models always arrive and reality almost never does is the size of what the theory leaves out.
The wake it makes for itself
Three suspects are usually named and should not be merged. Vertical wind shear tears the column apart from outside. Eyewall replacement cycles, covered in No. 17, interrupt from within. And the storm chills the ocean beneath itself — its winds mix the warm surface layer down and haul colder water up, leaving a cold wake that can trail for hundreds of miles.
Emanuel’s own pick was the cold wake, and it is hard not to prefer it, because it is the only one of the three the storm does to itself. No. 17 was a machine with a single cancellable fuel line. This is that line, cut from the inside — and the cold water is only within reach because a summer ocean is stratified, which was No. 122’s subject exactly.
What the number will and will not do
It is rising, and the honest way to say so is dull. NOAA’s Geophysical Fluid Dynamics Laboratory assesses that tropical cyclone intensities globally are projected to increase, with medium-to-high confidence, by 1 to 10 per cent for two degrees of global warming. That is a claim about the ceiling, not about how many storms there will be; the frequency question is genuinely contested and nothing here settles it. Nor does the ceiling simply track how warm the sea is — what lifts it is water warm relative to the rest of the tropics, which is not the same thing at all.
And the theory knows its own error: if it were exact, that fitted line would meet the axis at 1.0. It meets it nearer 0.9 — Emanuel’s ceilings, by his own reading, about ten per cent high.
≈
One loop I’m watching
Next: there is a defence in your body that consists of nothing but being already occupied. No antibody, no cell, no recognition of any kind — a few hundred species of bacteria holding every seat in the room, so that something dangerous arriving finds nowhere to sit and nothing to eat. It works beautifully, it defends you every hour of your life, and a single course of the right drug can empty the room in three days.