There is only one chance in five that 2010’s record Moscow heatwave would have happened without a warming climate, researchers in Germany have found. “I was very surprised by how unlikely it was that this was just a random event,” Stefan Rahmstorf from the Potsdam Institute for Climate Impact Research told Simple Climate. “I don’t think many people appreciate how much the odds for such extremes have increased due to global warming – I certainly didn’t until we had performed this study.”
Not only do such weather extremes grab news headlines, they also often provide an important way to understand the climate. That’s what inspired Rahmstorf to investigate how frequently new temperature records are set. One of his motivations was that ‘climate skeptic’ researcher Richard Lindzen said in 2006. ‘Indeed, the absence of any record breakers during the past seven years is statistical evidence that temperatures are not increasing’. “But is it?” Rahmstorf asked. “Is seven years without a record evidence against an ongoing global warming trend? How often do we expect a new record with the ongoing warming trend? I did not know the answer, but I wanted to find out, so I started thinking about this problem.”
Research in this area dating from the late 1970s originally suggested that with a stable climate the further back in time we have temperature data from, the less chance there is of a record event today. As a result, most records occur near the first data collections and mark the highs and lows of “randomness”, or variability, in the climate. But, over the past few decades increasing temperatures and frequent records clearly show that the climate is behaving differently to this.
Consequently, Rahmstorf and his colleague Dim Coumou investigated how natural variability and underlying warming trends might combine to cause temperature records. Because they were studying chance rather than the physics of climate they did not need the general circulation models underlying so much research in this area. As in the case of a stable climate, for a steady underlying warming trend corresponding to that actually seen in the last century, they were able to calculate probabilities of global records.
In this approach, the number of record-breaking events depends on how quickly warming is happening and how variable temperature is normally. Writing in a Proceedings of the National Academy of Sciences of the USA paper published last week the scientists determined that rapid warming increases the number of new records, while variability reduces it. More rapid warming would therefore cause more records. Combining observed global numbers for these two phenomena they predicted 1.4 record years between 2000 and 2010, compared to the real-life situation of a record in 2005 that was matched again in 2010.
But Rahmstorf and Coumou also turned to comparatively “crude” Monte Carlo simulations – named after a casino where physicist Stanislaw Ulam‘s uncle would gamble away his money. These took the underlying warming trend and simply added or subtracted a random value in the variability range for each year in the century. Initially the scientists used the simulations to confirm their calculations for steady warming, but then they looked at more realistic scenarios where rates of warming change over time. “It is as crude as rolling dice 100,000 times to find out how often they roll a six,” Rahmstorf explained. “The crudeness does not consist in getting a bad answer, it is that it is not very elegant to have to try it out rather than finding a way to calculate it. For non-linear trends the question could only be answered by Monte Carlo simulations – and indeed we had 100,000 tries.”
Meanwhile, back to Russia
Using a constant rate of warming for the whole century, one-fifth of the Monte Carlo simulations had no temperature records, and around four-fifths had one, and the remainder had two or more. But simulations using the real warming trend – which levelled off in mid-century before warming especially quickly in the last 30 years – showed higher probability still of record temperatures. Nearly one third of the simulations reported three record years and a similar amount reported two. But chance means that periods with fewer records should still be expected. “We showed that Lindzen’s claim was wrong – there is nothing exceptional about seven years without a record whilst global warming continues at an unchanging pace,” Rahmstorf said.
Rahmstorf and Coumou then used the same method to specifically study the area around Moscow. They found that the recent rapid warming trend produces an almost 50:50 chance of a temperature record occurring there within the last decade. This compares to a less than one-in-ten chance during the same decade, if the climate were stable. Together these figures indicate just a one-in-five probability of a record in the Moscow region over the last decade not caused by warming.
The scientists underline that where temperature data is highly variable, like daily weather station outputs, warming has little influence on record occurrences. But in collected data where variation is much smaller, like monthly average weather station outputs, it plays a much greater role. “The expected number of extremes is now several times larger than that in a stationary climate, so the majority of monthly records like the Moscow heatwave must be considered due to the warming trend,” the scientists wrote.
- For more detail on this work, and in particular why it differs from a previous paper saying the Moscow heat wave was due to natural variability, read Stefan Rahmstorf’s own blog entry at RealClimate. The data and code he used in this experiment are also available on his personal blog.