Can we correct climate models to better predict record weather?

Through Adam vaughan

Record-breaking climate events, such as Canada’s highest temperature on record, surpassed by nearly 5 ° C last month, will be increasingly likely over the next few decades, new research suggests. It comes as the ability of climate models to predict such extremes has been called into question following a series of intense weather events around the world.

Erich Fischer of ETH Zurich, Switzerland, and his colleagues ran computer models to simulate the average maximum temperature of the hottest week of the year in parts of North America and Europe to to see if they could produce temperatures that broke wide margins records. They might – in some emissions scenarios, records have been broken by more than one degree by 2030, not the 0.1 ° C or 0.2 ° C usually predicted.

The researchers conclude that the likelihood of such record-breaking events is largely due to the rate at which the Earth is warming, and not just how much it has warmed, which is 1.1 ° C up to present and continues to increase. “It’s really the rate of change,” says Fischer.

Geert Jan van Oldenborgh of the Royal Netherlands Meteorological Institute, who has linked the recent heat wave in North America to climate change, says it is worrying that some statistical models have indicated that the region’s records were impossible. Such models have a theoretical distribution of extreme values, which gives an upper limit for the temperatures in an area. This limit generally increases smoothly with climate change.

“Then this heat wave came and it was way above the upper limit [for the region]. It is rather surprising and unsettling that our theoretical picture of heat wave behavior has been shattered so brutally, ”says van Oldenborgh.

The heat wave is not the only event that has rocked climatologists in recent times. Germany has been hit by deadly floods while China’s Henan experienced its heaviest rainfall in a millennium, killing people in flooded subways. “It shocked me,” says Tim Palmer of the University of Oxford.

So what about future events? On a general level, climate models have done a good job of predicting large-scale changes due to climate change, says Peter Stott of the UK Met Office. “Not just the increase in global average temperature, but the increase in temperature extremes and precipitation. It has been made very clear, and it is indeed happening.

However, older models did not capture the intensity of some regional extremes like those seen in Canada, says Stott. The good news, he says, is that some new climate models have a higher level of spatial detail, closer to weather models, down to a grid of squares 2 kilometers in diameter, which will better predict extremes. local. Modelers also have a better understanding of the processes that cause short but intense precipitation, such as that seen in Germany and China.

However, the higher resolution required for some models usually requires more computing power – and resolution isn’t the only issue for projecting extremes: Another big issue is timescales. Much of climate modeling works on centuries-old timescales, but some scientists have now turned to decadal forecasts, which could be roughly described as weather forecasts designed to predict the next few years. These have already been shown to predict Atlantic hurricanes.

“There is definitely a movement towards these ten-year predictions. They are not meant to predict what climate change will do, but what climate change is doing now, ”says Ted Shepherd of the University of Reading, UK.

While many modelers say that more computing power alone isn’t a silver bullet for projecting extremes, it should help. One example is the computation required to produce numbers from the complex calculations of the Navier-Stokes equations, which can be used to model motion in the atmosphere.

More processing power would give more accurate numbers, Palmer says. “It basically comes down to IT.” He called for a ‘CERN for climate change’, a supercomputing project that he said could be executed for around 200 million euros per year. This has yet to happen, but initiatives are underway that could help climate models, such as an EU-backed project to build a ‘digital twin’ of Earth.

And it’s worth remembering that climate models are constantly improving, says Tim Osborn of the University of East Anglia, UK. He says models may not be able to simulate records like the heat in North America because they fail to capture a complex combination of processes, such as an interaction between a clear sky, a faint sky. soil moisture and wind direction, but the truth is we just don’t know yet.

Better climate models will be essential to adapt to climate change and inform early warning systems to prevent deaths. But it’s not like we need them to take action to mitigate the cause of climate change: humanity’s greenhouse gas emissions. “I don’t think it’s the models,” Shepherd said. “I think people don’t just take action on climate change for other reasons. They put their heads in the sand. It’s hard to imagine things that didn’t happen.

Sign up for Countdown to COP26, our free newsletter covering this crucial year for climate policy