Not Another Heatwave!

As the southern part of Great Britain enters another heatwave, it’s worth noting that the “1976” drought and heatwave actually began in 1975, so there’s a risk of problems continuing into 2023. It’s also troubling that the climate models used to predict the likelihood of heatwave conditions are reportedly poorly calibrated.

Brenda from Bristol talking about something else entirely!

Once again the Met Office is warning of a heatwave. With temperatures in, for example, London, forecast to reach near or over 30C every day for about the next week, and the threshold for a heatwave only a maximum of 28C for 3 days (less in some regions), an official heatwave is a nailed-on certainty.

Fig 1 Temperature forecast summary for 9-12/8/22 issued by Met Office on 8/8/22

So why is this happening?

A lot of people are saying that what we’re experiencing is “the new normal” with global warming, as if we’re going to experience drought and heatwave every year. That won’t happen, although when it does it’ll be more severe.

The good news, if you like, is that we’re experiencing a particular weather pattern, perhaps as a result of (or as part of) the extended “triple-dip” La Nina that is currently underway.

Fig 2 Timeline of La Nina episodes (from Wikipedia)

I can’t help noticing that extended La Ninas preceded the record-breaking UK summers of 1976 and 1911.

The bad news is that with the La Nina expected to continue into winter 2022-3, we may not be out of the woods yet. The Met Office notes that UK winter weather can be affected by La Nina, so that’s something to look out for. But one wonders whether the effects could continue into next summer.

All this is fairly speculative, of course, but what, for me, leapt out of the chart in Fig 3, again from the Met Office (they do produce some interesting stuff!), was the presence in the graphic of not only 2nd-5th July 1976, but also 3rd, 4th and 8th August 1975.

Fig 3, Highest UK average maximum temperatures from Met Office

What if, I’m sure you’re thinking, dear reader, what if 2022 is analogous to 1975 (and not 1976)? I noted back in June that we were not then experiencing a UK centred heatwave, as in 1976.

Fig 4 Temperature anomalies June 1976 vs June 2022, courtesy of Neale Upstone

To digress slightly, it turns out that these patterns actually have names. The 2022 pattern is a “wavenumber 5”, but to my untrained eye 1976 looks different. For starters an El Nino is developing in 1976 whereas in 2022 the sea off the west coast of South America is relatively cool. An El Nino might well happen in 2023, I could add.

Fig 5 Current weather map for UK region, from Met Office

Now that we have drought conditions in at least the south-eastern half of Great Britain, and high pressure centred at present over the UK, I suppose we could say we have our “own” heatwave, but it’s quite late in the summer, unlike in 1976 when the drought became severe earlier in the summer.

So, if, as I argued in my previous post, 2022 is not a worst-case scenario, why is it SO hot?

Ella Fitzgerald singing Cole Porter

The latest Met Office extreme heat warning notes that: “Temperatures are expected to peak at 35C on Friday [12th August] and Saturday, or even an isolated 36C on Saturday.”

It might be worth noting that until 1990, a temperature of 36C or higher had only been recorded in the UK on a few occasions (cursory research suggests these were 36.7C on 9th August 1911, 36.1C on 19th August 1932 and 36.0C on 22nd July 1911). As I noted last time, the highest temperature recorded in 1976 was “only” 35.9C, on 3rd July.

The record-books are rapidly being re-written, with Fig 6 superceding Fig 1 in my previous post.

Fig 6 Latest top 10 hottest UK days from the Met Office

Here’s a thing. According to World Weather Attribution, heatwaves as extreme as 18th-19th July 2022 in the UK are not really expected. They write:

“While Europe experiences heatwaves increasingly frequently over the last years, the recently observed heat in the UK has been so extreme that it is also a rare event in today’s climate. The observed temperatures averaged over 2 days were estimated to have a return period of approx. 100 years in the current climate. …

At three individual stations the 1-day maximum temperatures are as rare as 1 in 500 years in St James Park in London, about 1 in 1000 years in Durham and only expected on average once in 1500 years in today’s climate in Cranwell, Lincolnshire.

The likelihood of observing such an event in a 1.2°C cooler world is extremely low, and statistically impossible in two out of the three analysed stations.

The observational analysis shows that a UK heatwave as defined above would be about 4°C cooler in preindustrial times.

To estimate how much of these observed changes is attributable to human-caused climate change we combine climate models with the observations. It is important to highlight that all models systematically underestimate the observed trends. The combined results are thus almost certainly too conservative.

Combining the results based on observational and model analysis, we find that, for both event definitions, human-caused climate change made the event at least 10 times more likely. In the models, the same event would be about 2°C less hot in a 1.2°C cooler world, which is a much smaller change in intensity than observed.

This discrepancies [sic] between the modelled and observed trends and variability also hinders confidence in projections of the future trends.

Excerpt from World Weather Attribution summary of their report on the UK 40C heatwave of 18th-19th July 2022 (my stress)

Yeap, you read that correctly. The climate models don’t predict what’s actually been observed!

I’m a longstanding sceptic of the value of attribution studies in making quantitative (as opposed to qualitative or indicative) predictions – to put it in a nutshell (my three posts on the topic more than a decade ago being overly complex), the probability of an event in model runs is just that, a probability in the model, not the real world – but this sounds like another level of screw-up entirely.

Let’s just note in passing that at first blush it would seem difficult, to say the least, to conduct an attribution study when the models predict lower temperatures than actually occur (or higher temperatures without global warming than actually occurred) – maybe reading the full paper (pdf) would provide some clues as to how this problem was overcome.

What the hell are the models missing? I would have thought that the climate research community needs to get to the bottom of this ASAP, so that we have a better understanding of climate change risks.

I suspect the answer must have something to do with the huge uncertainties in warming caused by greenhouse gases and the more local cooling caused by aerosol emissions as shown in Fig 6 of my previous post, repeated here as Fig 7. Note the error bars:

Fig 7: IPCC AR6 Fig SPM.2

If the warming effect of greenhouse gases and the cooling effect of aerosols have both been underestimated, as is compatible with Fig 7, then a rapid increase in extreme temperatures is exactly what would be expected in a region, such as Europe, where aerosol levels have surely reduced significantly in recent decades.