Weather — Why Tomorrow Can Be Told and Next Week Cannot

Weather is air on the move

Air is stacked above our heads. Invisible, but heavy enough to press down on us. That pressing is air pressure.

It is not the same everywhere. Somewhere the air piles thick, somewhere thin, and air runs from the thick side to the thin. That running is wind. Where the closed curves on a weather map crowd together, the wind is strong.

Why thick in one place and thin in another? Because the sun does not warm the ground evenly — tarmac, paddy, hill and sea all take it differently. Air over the warmer patch expands, lightens and lifts. That is convection.

Rising air cools, and cooled air cannot keep all its water vapour, so the vapour beads into very small drops. Those drops together are cloud. Let them collide and grow and they fall. That is rain, and the beading is condensation.

So why does it blur after three days?

If it is arithmetic, the answer ought to be single. The sky does obey very definite laws. And still next week cannot be told. Here is the surprise of this piece.

A man was computing weather on a machine. To resume an earlier run he typed the numbers back in from a printout, trimmed a few decimal places — and the weather months later came out entirely different.

A tiny difference at the start growing as time passes: that is chaos. Not an absence of law. The law is exact, and it happens anyway.

It is hard to believe on the page. In the board below, shrink the difference between the two starting values yourself: the parting moves later and later, and never goes away.

Tool The Forking Sky — Why Forecasts Blur After Three Days Run today's sky twice under almost identical conditions. Two lines that start on top of each other become different weather a few days on. See where a speck of difference forks, and why a forecast blurs past three days. (Korean only)

Modern forecasting is what came after accepting that. The sums are not run once: they are run dozens of times from slightly different starting values, and the share of runs that produced rain is the probability of precipitation.

Sixty percent does not mean rain over sixty percent of the area, nor for sixty percent of the hours. It means that in ten situations like this one, six bring rain. The forecaster speaks in probabilities not from timidity but because the probability is the honest answer.

A little further in

Measuring — the sky is thinner observed than you would think

To start computing you must know the sky as it stands. On the ground the stations are dense; above it, matters differ.

1. Surface stations — temperature, wind, rain, pressure. Several hundred across the country, most of them unstaffed.
2. Radiosondes — a small instrument flown up under a balloon, reporting temperature, wind and moisture by height until the balloon bursts. Twice a day, holidays included.
3. Weather radar — radio pulses returned by raindrops give their position and motion. Usually sited on a summit.
4. Weather satellites — a satellite holding station above the region watches the cloud continuously, filling in the sea and everywhere else people cannot go.
5. Ocean buoys and ships — pressure and waves at sea, which is where typhoons travel.
6. Airliners — measuring wind and temperature at altitude along their routes and sending it down.

Even all together it is sparse against the whole sky. This is where the wall comes from.

Computing — cutting the sky into cells

The sky is divided like a game board. Each cell takes the present temperature, wind and moisture, and the laws air obeys give the values a short while later; then again, and again. This is numerical forecasting.

Smaller cells catch a single valley or ridge and give better answers — but halving the cell size multiplies the work more than tenfold. Hence the large supercomputer, the machine of the fourteenth piece.

Not every country builds this method; most borrow one. Korea, centred on the meteorological research institute on Jeju, built its own model and forecasts with it. Being able to fit it to our own terrain and our own seas is what borrowing does not allow.

It is also why a typhoon track is drawn as a widening cone rather than a line. The cone is how far the many runs disagree. It broadens with time not from carelessness but from honesty.

If three days cannot be told, how can a century?

A common question. The answer is that weather and climate are not the same thing.

Think of dice. Nobody knows the next throw. Everybody knows roughly the average of a thousand throws. Weather is the next throw; climate is the average. Different questions take different answers.

And climate calculations include forces with a direction, such as the greenhouse effect. The die has been tilted: the single throw stays unknown while the drift of the average can be stated.

What you would study to do this

Weather work is more than the forecaster. Some measure, some compute, some keep the instruments alive.

1. Forecaster — atmospheric science. The one who judges last, choosing which of the machine's several answers to say aloud.
2. Model builder — mathematics and computation, turning the laws of the sky into something a computer can solve.
3. Instrument technician — keeping summit radars and unstaffed stations running, climbing snowed-in roads to mend a generator or a link. Without these people there are no numbers to compute with at all.
4. Radiosonde launcher — flying the balloon at the appointed hour without fail, and more often when a typhoon comes.
5. Satellite data specialist — turning what the satellite sends into numbers a forecast can use.
6. Private meteorology — airlines, shipping, farming and power companies all buy the weather they need; there are professional certificates and expert-assessment work.
7. Military and airport meteorology — for flight and operations, weather is safety.

The third line is the unexpected seat here. It is a path open to anyone at home with machinery or electricity, and a good share of forecast accuracy rests in those hands. The finest supercomputer computes nothing without numbers to put in.

There are places to visit, too: weather science centres at Daegu, Miryang, Yeosu, Chungju and Jeongeup, and in Seoul a weather museum in the old observatory building. The Yeosu one takes the weather of the sea as its own subject.

The question that remainsHow should we hear a forecast that speaks in probabilities — knowing that six in ten also means four in ten with no rain at all?