The UK has provisionally breached 40C for the first time ever, reaching 40.2C at London Heathrow airport. Before this heatwave, temperatures of 40C had never even been forecast for the UK – let alone recorded.
Yet earlier this month, scientists began sharing the first-ever weather forecasts predicting 40C temperatures in the UK for mid-July. That a model could project such temperatures in the UK was unusual, scientists said, but they stressed that such heat was unlikely to materialise.
As the days passed, however, more and more models began to converge on higher temperatures – and the likelihood of seeing 40C heat in the UK rose steadily.
The UK Met Office implemented an “amber” weather warning a full six days before the heatwave struck – the earliest heat warning that it has ever released. And, by 15 July, the Met Office released its first ever “red” warning.
This is “the most insane weather story I have seen in this country”, a meteorologist tells Carbon Brief, adding that “the silent revolution of weather technology and understanding allows meteorologists to share life-saving information days in advance”.
How do scientists forecast heatwaves?
Weather is “chaotic”, meaning that small changes in temperature, wind or humidity can have a significant impact on conditions hours or days later. This feature of the Earth’s atmosphere and oceans is often called the “butterfly effect”.
However, chaos can make it tricky to produce accurate weather forecasts – as even small errors in the “initial conditions” of a forecast can result in inaccuracies that grow as the forecast runs into the future. This means that as the “lead time” of the forecast increases – the time between the forecast being made and the event happening – it can become less accurate.
To account for inaccuracies, forecasters often use a large group of simulations called an “ensemble”, which are each run with slightly different starting conditions. By looking at the range of outcomes, forecasters can determine which outcome is the most likely.
This is shown in the graphic below, with the blue lines indicating the simulations in an ensemble and the blue blob showing the spread of potential outcomes. The red line shows a single “deterministic” forecast for comparison.
“Meteorologists increasingly use ensemble modelling in their forecasts, as opposed to relying on a single computer model run every six hours…When many different computer model runs are forecasting roughly the same outcome over a consistent period of time, we can have more confidence in that outcome. It’s more likely to happen – although nothing is ever guaranteed.”
Weather forecasts have improved significantly over the past 50 years. Dr Simon Lee – a research scientist at Columbia University in New York – tells Carbon Brief that “for large-scale patterns across the northern hemisphere, a seven-day forecast today is about as accurate as a five-day forecast in 2000 or a three-day forecast in the 1980s”.
Meanwhile, the Met Office’s four-day forecast is now as accurate as a one-day forecast was 30 years ago. McGivern tells Carbon Brief that this is in part due to improvements in individual models – for example, higher spatial resolution – and partly because more model runs can now be produced simultaneously, thanks to greater computing power.
The UK 2022 heatwave
On 23 June, the European Centre for Medium-Range Weather Forecasts released the results of a 32-day “subseasonal” forecast model run, which predicted unusually high temperatures for 18-25 July over the UK and US.
This model run was the first indication of what would eventually become an unprecedented heatwave, attracting global media attention. However, at the time, it was not given much attention, as it was only one in a large range of available model runs. Furthermore, Lee tells Carbon Brief that this type of forecast is “generally low in skill”, meaning not particularly accurate.
On 1 July, one of the 30 model runs produced by the Global Ensemble Forecast System predicted temperatures of 40C in the UK for 14 July. Scientists stressed that these temperatures were unlikely to materialise, given the limited number of model runs suggesting extreme temperatures. Nevertheless, some highlighted this result as “historic”.
As the days went on, model runs from a range of institutions continued to suggest high temperatures around the middle of July. McGivern tells Carbon Brief how the likelihood of high temperatures increased over July:
“At the start of July, a minority of computer model runs were indicating exceptionally high temperatures for the middle of the month. From around 8-9 July, however, most of the operational model runs and their ensemble counterparts started converging on the same idea: heat building from the south during the 16-17 July weekend and peaking in the mid- to high-30Cs.”
The Met Office released an amber warning on 11 July, stating that “exceptionally high temperatures are possible from Sunday [17 July], lasting into early next week”.
“It’s still not often that so many computer models agree on the weather and temperatures more than five days ahead,” McGivern tells Carbon Brief. He adds that this six-day lead time “is further ahead than most weather warnings – and certainly most amber warnings – that the Met Office tend to issue”.
The day after the warning was released, the Met Office press office told Carbon Brief why the warning was released so early:
“At this stage we have enough confidence that extreme heat is likely [to] develop across the areas warned and given the potential impacts, we have issued the warning with maximum lead time to allow services and the public to respond effectively.”
On 13 July, the Met Office global model showed maximum temperatures of 40C in parts of central England on 19 July. Met Office senior forecaster Alex Deakin explained in a video posted to Twitter that, “at this stage, 40C is possible early next week, but unlikely, with [the] chance being less than 10%”. However, he added:
“It is still pretty remarkable that temperatures over 40C are being generated at all by these computer simulations.”
On 15 July, the Met Office issued its first-ever red warning for “exceptional” heat across parts of England. In a press release, the organisation said “this is the first time we have forecast 40C in the UK”.
Met Office chief meteorologist Paul Gundersen added:
“Currently, there is a 50% chance we could see temperatures top 40C and 80% we will see a new maximum temperature reached.”
The Met Office raised its heatwave forecast on 17 July, giving a 90% chance of the UK seeing a temperature record in the upcoming week and a 60% chance of reaching 40C.
This story of increasing model certainty is similar with different forecasting centres. On 18 July, ECMWF released a press briefing, stating that the heatwave “was forecast at least three weeks in advance, but, until last week, there was significant uncertainty about the extent and severity of the heatwave”.
This is “the most insane weather story I have seen in this country”, meteorologist Scott Duncan tells Carbon Brief – likening the UK heatwave to the “record-smashing” temperatures seen in the Pacific northwest last year.
“It’s been remarkable that we’ve seen extreme scenarios picked up by ensembles at the longest lead-times,” Lee says.
“The heatwave was very well forecast, some models were flagging the extreme potential even 10 days out. Naturally, you have to be cautious to place much faith in details that far out but the pattern for delivering the heat was captured.
“Even the magnitude of the heat has been captured fairly well. A meteorologist always raises an eyebrow when a record is in the models. [But] this event was so clear about the exceptional nature of the air mass, records were inevitable.”
Forecasting unprecedented events
Temperatures of 40C are already 10 times more likely in the UK than they would have been in a world without climate change – but this milestone temperature is still only expected once every 100-300 years in the current climate.
Nevertheless, as the climate warms, temperature records will repeatedly be broken, as increasingly hotter temperatures are reached around the world.
Prof Hannah Cloke, professor of hydrology at the University of Reading, tells Carbon Brief that climate change can make it “tricky” to forecast extremes, because they are often unprecedented in the historical record. Cloke explains:
“We haven’t had [these extremes] in the record before so that the models are not kind of set up with all of that information. So we have to rely on physics, and putting together lots and lots of different models to understand whether something’s more likely or not. So that’s a real kind of technological challenge.”
Dr Rebecca Emerton – a research scientist at the ECMWF – adds that “based on the science and our understanding of the Earth system, [models] can simulate how events might unfold, including events that we’ve never observed before but are physically possible”.
Nevertheless, Cloke tells Carbon Brief that the ability of models to “predict the unprecedented” is one of the more “exciting” advances in modelling.