Weather and extreme events

For a long time, the only honest answer to “is this heatwave due to climate change?” was: nothing can be said about an isolated event.

That is no longer true. How it changed is a story worth telling.

What makes the weather

All weather comes from a single cause: the Sun does not heat the Earth evenly. Equatorial regions receive almost vertical light; the poles receive light at a low, glancing angle. Warm air is lighter and rises; cold air, being heavier, sinks and moves in to take its place. This permanent imbalance sets the atmosphere in motion, and that motion is the weather.

Wind is air moving from an area of high pressure towards an area of low pressure. The greater the difference between the two, the stronger it blows. That is why forecasts show pressure maps: they indicate where the air will go.

Clouds and rain come from the water that the air carries. Warm air can hold a great deal of water vapour, cold air much less. When a mass of air rises, it cools; at a certain point it can no longer hold all its vapour, which condenses into droplets: that is a cloud. If the droplets grow large enough, they fall.

This is enough to explain violent phenomena, which are not exceptions but extreme cases of the same mechanisms.

• A thunderstorm arises when warm, moist air rises very rapidly. The violent currents inside the cloud rub ice crystals and water droplets against each other, which separates the electrical charges — positive ones towards the top, negative ones towards the bottom; when the difference becomes too great, it discharges all at once — lightning. Thunder is the sound of the air suddenly expanding along its path.
• A cyclone — called a hurricane in the Atlantic, a typhoon in the Pacific — forms over a warm sea, above 26 °C — and this warmth must extend to a depth of around fifty metres, otherwise the churning of the waves cools the surface too quickly. The warm air rising from the sea feeds a spiral of ascending air that sustains itself as long as it stays over warm water, and that dies out over land, for lack of fuel.
• A heatwave is the opposite of movement: a mass of warm air held in place by a pressure pattern that prevents it from circulating. Nothing happens, and that is precisely the problem.¹¹

Weather and climate are not the same thing

Weather is what the atmosphere is doing at a given place on a given day. Climate is the statistics of that weather over decades. One is forecast a few days ahead, the other is projected over centuries — and the two exercises have neither the same methods nor the same limits.

The rest of this page deals with the boundary between these two notions. For a long time, it was considered impossible to link a specific event — this particular heatwave, that particular flood — to a change in the climate. What has changed is not the climate: it is what we know how to calculate.

Why it was a good answer

An event is a unique case. It happened once, in a specific place, on a given date.

It cannot be replayed. The observed reality cannot be compared with a parallel world in which the climate had not been altered by human activity — that world cannot be observed directly.

Saying “nothing can be concluded” was therefore not excessive caution. It was correct.

These are not our words. The authors of the paper that founded this discipline say so themselves, in their abstract:

It is an ill-posed question whether the 2003 heatwave was caused, in a simple deterministic sense, by a modification of the external influences on climate […] because almost any such weather event might have occurred by chance in an unmodified climate.²²

The nuance matters: they do not say that the answer is difficult. They say that the question is badly framed. It was by reframing it that they were able to move forward.

What changed: attribution science

Robust scientific methods have been developed to approach the question from another angle. This discipline is now called extreme event attribution science.

In 2021, the IPCC regarded its overall result as established: human-induced climate change is the main driver of the increase in the intensity and frequency of extreme heat.³³

The shift is this: the question of whether the event was caused is set aside, and it becomes possible to estimate by how much human activities have increased the risk of such an event occurring.⁴⁴

These methods therefore do not give a binary answer (yes or no). They measure a change in frequency and intensity: by how much an event of this type has become more likely, or more intense, because of anthropogenic emissions.

The principle: supercomputers are used to simulate the current climate a very large number of times (with the observed warming), then a “natural” climate a very large number of times (without human greenhouse gas emissions). One then compares how often an event of such magnitude appears in each.⁵⁵

The first case treated this way

The 2004 paper dealt with the European heatwave of the previous summer.⁶⁶

How can the event be defined without biasing the result? The authors chose a threshold for mean summer temperature over a large region of the continent — a threshold exceeded in 2003, and in no other year since records began, in 1851.⁷⁷

The care taken is notable: the threshold was not chosen after the fact to suit the result. It was defined by an event unique in a hundred and fifty years of records.

Their conclusion: it is very likely — with more than 90% confidence — that human influence has at least doubled the risk of a heatwave exceeding this threshold.⁸⁸

The phrase “at least” is essential. It is not an average estimate: it is a lower bound. The authors state that the factor is at least two, without claiming to know by how much it exceeds that.

A cautious science does not always give a central figure. Sometimes it gives a floor, and that is already a great deal.

“40% chance of rain”: what it means

Ask the people around you, and the question gets at least three different answers, all wrong:

• It is not: it will rain over 40% of the region.
• It is not: it will rain for 40% of the day.
• It is not: 40% of meteorologists think it will rain.

The real definition

In meteorology, the figure given is the Probability of Precipitation (PoP). It is a product of two factors:

PoP = C × A

• C = The forecaster’s certainty that it will rain somewhere in the area.
• A = The average fraction of the area that would receive rain if it rains.⁹⁹

If it is 100% certain (C = 1.0) that it will rain over 40% of the area (A = 0.40): 40%.
If it is 40% certain (C = 0.40) that it will rain over the whole area (A = 1.0): 40% as well.

The same figure covers two very different situations. That is why it is so poorly understood: it does not describe a simple spatial measurement, but the combination of an uncertainty and a geographical coverage.

The right way to say it

The rigorous rephrasing is as follows:

“If there were ten days with exactly these atmospheric conditions, it would rain at this place on four out of ten.”

It is longer, but accurate.

The real problem: testing a single case

A single probabilistic forecast can be neither confirmed nor refuted by an isolated fact.

If the forecast says 40% and it rains, it was not wrong. If it says 40% and it does not rain, it was not right. A single day tests nothing.

A forecaster or a weather model can only be evaluated by analysing many of its forecasts: if it has forecast 40% a thousand times, it should have rained on about four hundred of those days. If that is the case, its model is perfectly calibrated.¹⁰¹⁰

A probability never qualifies a single case. It qualifies a series.

That is why people feel that “the forecast is always wrong”: they are testing a probabilistic forecast with a single event — which is logically impossible.

A lesson that goes beyond the weather

We encounter probabilities all the time: in medicine, in insurance, in finance, in public decisions.

Each time, the same rule applies. A probability figure cannot be judged on a single event.

Each time, the same temptation: to conclude after the fact that “it was predictable” (hindsight bias) or that “the experts were wrong”. Neither can be deduced from a single case.

A forecast is not a promise

“Seventy per cent chance of rain” is not disproved by a dry day.

This number does not say that it will definitely rain. It says that in a large number of atmospheric situations equivalent to this one, measurable precipitation was recorded about seven times out of ten.

This definition rests on a specific precipitation threshold (generally 0.1 mm of rain at ground level over the period) and on a given geographical grid.¹¹¹¹

This is the central question of scientific rigour: what exactly does this figure count?

Two methodological reminders

Weather and climate differ in timescale. Weather describes the instantaneous state of the atmosphere; climate describes the statistics of that state over a long period (generally 30 years). A very cold day says nothing about a global trend — and neither does a very hot day. Our page on climate explains this.¹²¹²

Casualty figures are always given with their source (e.g. international databases such as EM-DAT or reports from official bodies), the date they were compiled, and exactly what they count (direct deaths vs overall excess mortality).¹³¹³

---

⚠ This page does not and will not contain any dramatised description of victims. The data are treated with strict analytical restraint.