The guys and gals at World Weather Attribution have pulled a white rabbit out of the hat yet again to confirm to a breathlessly waiting queue of climate activists, alarmists and media hacks their worst suspicions – yes, humans did indeed make the recent June heatwave in France much more likely and more intense. Peter Stott is one of the authors of the retrospective rapid extreme weather attribution study. Quite why, I’m not sure because he already knew that it was climate change wot dunnit before the heatwave actually happened! Was it tea leaves, Tardis or entrails I wonder which bestowed this remarkable foresight upon Peter? Actually, Peter would make quite a good Doctor, don’t you think? Perhaps he should think about auditioning for the BBC.

Joking aside, let’s get to the serious stuff. Let’s actually examine the study which headlines across the world declared was proof positive that the European heatwave was man-made. Did the writers of those headlines actually read the study? I doubt it somehow. Most of them were probably just briefed by the scientists involved or they got their info second hand from other media sources.

 

What the papers say

So before getting into the nitty gritty, let’s look at a few of those official tweets, headlines and media stories which have promoted the idea that your SUV, washing machine and toaster were responsible for France baking in extreme high temperatures for three days from June 26-28.

CNN headline says:

Climate crisis made European heat wave ‘at least’ five times more likely

In the text, Friederike Otto, another of the above study’s authors, is quoted as saying:

 

Scam (Scientific American) reports:

For one group of climate scientists, the event presented a rare opportunity: to rapidly analyse whether the cause of the heatwave — which made headlines around the world — could be attributed to global warming. After a seven-day analysis, their results are in: climate change made the temperatures reached in France last week at least five times more likely to occur than in a world without global warming.

The scientists with the World Weather Attribution Project decided to take action when they saw the heatwave coming and ended up performing a real-time analysis while at a climate conference in Toulouse, France.

“Some say the uncertainties are too big,” says Otto. “There are indeed caveats, mostly to do with imperfect climate models. But even with large uncertainty bars we think it is useful to provide quantitative evidence for how climate change is affecting extreme weather,” she says.

Using their models, the researchers calculated that the average temperatures reached over the three hottest days in France — around 28°C — were at least five times more likely to happen because of climate change.

But in a second analysis that looked at historical temperature records over the past century rather than models, the team calculated that the likelihood of such a heatwave in June has in fact increased 100 times since around 1900, owing to the combined influence of climate change and other factors such as air pollution.

Here’s what the British version of Scam, New Scientist, tweeted:

The Graun’s headline and sub-heading says:

The record-breaking heatwave that struck France and other European nations in June was made at least five – and possibly 100 – times more likely by climate change, scientists have calculated.

Such heatwaves are also about 4C hotter than a century ago, the researchers say. Furthermore, the heatwaves hitting Europe are more frequent and more severe than climate models have predicted.

Dr Friederike Otto, of the University of Oxford, one of the scientists behind the new analysis, said: “This is a strong reminder again that climate change is happening here and now. It is not a problem for our kids only.”

The researchers, many of whom happened to be at a conference on extreme events and climate change in Toulouse, then used temperature records stretching back to 1901 to assess the probability of a heatwave last month and in the past. They also examined climate change models to assess the impact of global heating.

Global heating caused by human activities made the French heatwave at least five times more likely, said Otto, based on combining the observations and climate models. Analysis of the observations alone indicated the heatwave was at least 10 times more likely than a century ago, and potentially 100 times.

However, these bigger increases in probability may result in part from changes in land use, soil moisture and irrigation, the growth of towns and cities, and air pollution, all of which can affect temperature.

You get the idea. We’re all going to fry, not boil slowly, like frogs in a gently heated saucepan, but like prawns, flash-fried in a wok, via heatwaves so intense that not even the climate models predicted just how intense they would be. The climate crisis is happening here and now.

 

The World Weather Attribution rapid attribution study

So what does the attribution study actually say? Well, in summary, at the beginning, it states:

The observations show a very large increase in the temperature of these heat waves. Currently such an event is estimated to occur with a return period of 30 years, but similarly frequent heat waves would have likely been about 4ºC cooler a century ago. Climate models have systematic biases in representing heat waves at these scales and show smaller trends, more year-on-year variation and fewer really severe heat waves than the observations. Combining models and observations we conclude that the heat wave was made at least 5 times more likely.

Why did the heatwave occur?

The authors tell us:

The heat wave occurred rapidly after a rather cool period in the early phase of June. This sudden change was mostly due to the specific dynamical conditions that were present, with a mid-tropospheric “cut-off low” system that formed off the coast of the Iberian peninsula (see the 500 hPa map on 27/6/2019 in Figure 2a). This system created a transport of heat from low-level Saharan air, and induced extreme temperatures at 850 hPa (Figure 2c) and in mountainous areas, such as 29°C at 1600-1800m in the Alps. These dynamical conditions, transporting hot Saharan air or air arising from the Iberian Peninsula are unusual (see analysis below). In particular the fact that the resulting transport of air masses remained at a very low level across the Mediterranean sea over five days was very unusual.

So it’s clear. The reason the heatwave occurred was because of a rare meteorological set of events, precipitated as we have already noted by what is known as a phase-locked wave 7 pattern of the northern hemisphere jet stream which is currently in an exaggerated wavy (meridional) configuration.

Event definition

We defined the event as the highest 3-day averaged daily mean temperature in June (TG3x). The daily temperature is taken as the average over France and at the city of Toulouse in southern France.

As the intense heatwave only really lasted three days, this would seem to be an obvious choice, though it doesn’t qualify as a heatwave according to the WMO’s definition:

The World Meteorological Organization defines it [a heatwave] as five or more consecutive days during which the daily maximum temperature surpasses the average maximum temperature by 5 °C (9 °F) or more.

The short duration reflects the fact that there were apparently no deaths reported attributable directly to heat exhaustion or hyperthermia. For comparison, 15,000 people died during the prolonged heatwave of 2003.

What observational temperature dataset was used?

For the average over France we use the official Météo France definition, which is the average over 30 long-term homogenised stations. Unfortunately, this series is not public but the area average of the E-OBS daily mean temperature over metropolitan France gives virtually identical results. This time series shows previous hot events around 1950, one in 1976, and further ones in 2005 and 2017, in addition to the present event, which set a record for June of 27.5 °C (the 2003 heat wave was more intense but occurred in August, not June). The 10-yr running mean (green line) shows a combination of the increasing trend due to greenhouse gases and a cooling phase around 1980 mainly due to air pollution (aerosols).

The dataset which can’t be examined because it is not public goes back only as far as 1947 and is virtually identical to another set of observations which cover metropolitan France. Hmm. UHI bias anyone? Here is the graph of the time series which the authors used:

Note how the authors attribute the ‘warming trend’ to anthropogenic GHGs and explain away the mid 20th century cooling to anthropogenic aerosols. By doing so, they have virtually done the attribution study for observations there and then – it’s all anthropogenic according to them. No confirmation bias there then! Rational people with a fair knowledge of climate change and its causes would look at that graph and note that it could actually be the truncated part of a cycle – a natural cycle – with mid 20th century cooling, pre 1950 warming and post mid 70s rapid warming at least partly explained by natural internal variability. rational people would also note that there’s not much of a warming trend across the entire (somewhat short) series and that it is only June 2019 which stands out. Similar heat events occurred at the beginning of the series and indeed the June 1976 heat spike occurred right at the end of the mid 20th century cooling, when the authors contend that aerosols were masking the GHG warming. So according to their analysis, 1976 3 day June temperature would have been even hotter still, maybe hotter than 2019, when atmospheric CO2 concentration was around 330ppm vs. 410ppm in 2019!

Note also that the hottest 3 day max June temperature was about 25.5C in 1947 and showed no increase before 2017, which year it edged up to about 26.3C, then leapt to 27.5C this year apparently. This, we’re told, is climate change in action. But where did the 4C figure come from?

How the attribution study estimated 4C anthropogenic warming of 3 day June heatwaves

Simple really. They used the 3-day max series from 1947-2018 and plotted it against the Hadcrut 4.6 anomaly for western Europe during June July August and then drew a linear trend line through the highly variable data. Then they extrapolated the linear trend back to 1900 when anthropogenic warming was thought to be negligible. Hence they ended up with 4C of 3-day heatwave warming (since 1900) which they attributed in its entirety to GHGs released since 1900!

We fit a Generalised Extreme Value distribution (GEV) to the French series (figure 5). The trend is included by allowing for a shift in proportion to the European summer temperature. This gives a good description of 3-day heat in France. We extrapolated the linear relationship found to 1900 to obtain an estimate of the total warming since a time when anthropogenic influences were small

The mean observed summer land temperature warming in western Europe according to Hadcrut 4.6 is about 2C apparently, so how to explain the 4C 3 day heatwave warming? The authors write:

This implies a much higher warming trend in France in June compared to that of the average European land summer temperature, which has warmed by about two degrees. More research is needed to understand this feature, but a major factor in the difference is likely the amplification that soil moisture drying contributes to temperature means and extremes in regions with transitional climate between dry and wet conditions. This effect is known to be strong in southern France and other regions with Mediterranean climate and is getting stronger in mid-latitude regions with global warming (because of decreased evaporative cooling if soil moisture levels become limiting for plants’ transpiration; e.g. Seneviratne et al, (2010), Mueller and Seneviratne (2012). Nonetheless, the derived value is particularly high compared to the warming in the European land and could also be affected by additional contributing factors or reflect some shortcomings of the analysis. For instance, the fit was derived from 60 years of data (Figure 5, left) and it is possible that there could be non-linearities in the relationship, e.g., because of a threshold behaviour which could imply a different relationship to European temperatures earlier in the century compared to the derived relationship. In addition, the role of soil moisture feedbacks could be different for means vs extremes, both in France as well as for the European temperatures (since it may also play an amplifying role during extreme conditions in more northern regions). It should be noted that the strength of the feedback may also depend on the month considered (e.g. likely stronger drying at the end of the summer season). Other processes that could also contribute are landscape changes or specific dynamical conditions occurring during hottest extremes in France, which could include hot advections from Spain or Sahara, making the underpinning physical processes different for mean vs extreme cases.

A lot of uncertainties and unknowns and caveats in there, very few of which found their way into the media reports and which the authors themselves – more especially Friederike Otto – appeared not to be too concerned about expressing, preferring instead to concentrate on the climate crisis narrative.

What did the study say about dynamics (i.e. meteorological effects/the jet stream)?

Basically, it concluded that the dynamical set up which is the direct cause of the heatwave cannot be attributed to climate change.

We do not identify a trend. This analysis thus does not suggest a strong role of climate change in this dynamical aspect of the extreme June heat.

They conclude that the meteorological situation was highly unusual in June 2019 but that no trend can be observed in the data going back to 1949.

What do the climate models tell us about the attribution of the European heatwave?

The fundamental basis of an extreme weather attribution study is a comparison between a world without added GHGs and just natural climate variability and a world with added GHGs as simulated by running climate models. Comparing natural to anthro model runs in the context of the observed event should enable an estimate of the influence which GHGs have played. That’s the whole point of the exercise. So what did the climate models used in this study reveal? Nothing, basically, they were useless, not fit for purpose!

Before undertaking the attribution analysis using climate models we have to evaluate whether the models readily available to us are fit for purpose and represent the statistics of the TG3x June event well. We considered the following models to investigate the changes in heat waves in France (Table 1)

 

Taking the two tests together we find that only one model ensemble or single model (barely) passes the test that the fit parameters of the tail of the distribution have to be compatible with the parameters describing the observations; a situation similar to the one encountered for area-averaged heat waves in the eastern Mediterranean (Kew at al, 2019). A more careful study would also investigate whether the observational analysis did not introduce biases in the fit values for the observations.

For a single station, Toulouse, we find similar results: the two models that do not have an incompatible scale parameter (Figure 11) have incompatible shape parameters (Figure 12), so here no model passes the formal tests. From the analyses for the shape parameter, we see that the CMIP5 multi-model ensemble actually has some overlap with the observations (Fig. 11), however it is outside the range of the observed scale parameters (Fig. 10). Hence, we are formally left for the present analysis with no suitable climate models or model ensembles to do the attribution (though we did not check the suitability of each single CMIP5 models and cannot exclude that some might be suitable for both parameters).

Got that? No models were found to pass the suitability tests so the attribution analysis could not be done. So what the hell did they do?

Having thus no well suited models from the analysed sample for the investigated event and locations at this stage, we decide not to give a synthesised overall result drawn from observations and models as in previous studies (refs) but still proceed with analysing all models, noting that the results are only indicative at best when drawing conclusions.

Can you believe that? Presumably so desperate to a get a non peer reviewed attribution study out in the public domain as quickly as possible they went ahead and did it anyway, using models which they knew were not suitable for the purpose! It worked though didn’t it. They got the global headlines they wanted – but the analysis is total BS, based on using models which are, by the authors own admission, not fit for purpose.

Then comes the pièce de résistance of this French heatwave study, the glacé cherry on top of the BS cake:

For the average over France we find that the probability has increased by at least a factor five (excluding the model with very strong bias in variability). However, the observations show it could be much higher still, a factor 100 or more. Similarly, the observed trend in temperature of the heat during an event with a similar frequency is around 4 ºC, whereas the climate models show a much lower trend.

Let’s deconstruct this short, innocent looking paragraph. Using models which are total crap, they find the probability of such a heatwave occurring has increased by a factor of five due to climate change. Looking at the observations (which prove nothing about attribution), they conclude that it could be as much as a factor of 100! Using the ‘observed trend’ of 4C (which is not an observed trend, it’s a linear extrapolation) the authors go public and declare that heatwaves in France are 4C hotter, implying this is due to climate change and furthermore state that an increase in probability of 5 times is a “conservative estimate” and the “best estimate” is 100 times. This “best estimate” appears to have been plucked out of the blue with no reference to the climate model runs (which cannot be relied upon anyway), simply by looking at 60 years of observations with limited amounts of data on similar heatwaves and extrapolating this back to a point 118 years ago when it is assumed that anthropogenic forcing of climate started, assuming also that anthropogenic aerosols were responsible for mid 20th century cooling.

The estimated 4C warming since 1900, which the climate models comprehensively failed to simulate, giving much lower warming generally, was used as the benchmark for the “best estimate” of the increased probability of the heatwave occurring, i.e. 100 times more likely. It’s hard to stress just how wrong this is. It’s attributing observed changes to man-made climate change using no evidence whatsoever other than the mere fact of the observations themselves. Astonishingly bad science and deliberate deception of the public in my opinion.