Russian Heat Version 2

This is the second post about the Russian heat wave. The subject for this post is another study about the Russian heat wave but this study argues that contrary to the study done by Dole et al that humans did had influence on 2010 Russian heat wave. But there is a big contrast between the two studies. Were Dole et al relied on climate models and observations takes this study a very different approach. The study done by Stefan Rahmstorf and Dim Coumou uses a quantitative approach, they have a mathematical approach. They do all sorts of calculations with temperature data, more precisely they use analytical solutions and Monte Carlo simulations. If you don’t know what Monte Carlo simulations are than is Here a short explanation, given by the University of Nebraska, Lincoln. Before I start summarising and discussing about this article I just want to emphasize how much the approaches used by both studies differ from each other. If you can remember all graphs and maps with airflows from the last the study about the Russian heat wave and contrast those graphs and maps with this:

I won’t give all the details about the meaning of all the letters and numbers, just contrast this with the last study of the Russian heat wave. This was one of their main tools to find about how much chance there was that warming caused a heat wave. But this formula  was not of course the only formula they used. See the methods in the article for an explanation of all the math used in this article. The calculations they used are based on observed temperatures. The basis for the formula above was a Gaussian probability density function. If you want don’t know what  means Here is an introduction from Columbia University, but  Gaussian processes  are often used in modelling things such as temperature over time. The authors of this article see weather with a changing number of extreme weather events as non-linear. 

All this they applied on the Russian July mean temperatures at Moscow weather stations. The data results gave a graph:

The linear trend over the last 100 years is 1.8 degree Celsius warmer with an inter annual variability of 1.7 degree Celsius. The ratio is 0,011 heat waves due to warming per year  which brings the total per decade on 0,29. This is much higher than the 0,105 in a stationary, non-warming climate. This makes the chance that a heat wave is caused by warming 64%(0.29 − 0.105 ∕0.29)  in the last decade in a linear climate. When using the more realistic non-linear warming trend, weather is never linear  the expected number of records is 0,85 which means that in the last decade it was an 88% probability (0.85 − 0.105∕0.85) that a heat wave was caused by warming in the last decade in Moscow. They say explicitly warming instead of a phrase like ‘’Human caused warming’’ because maths can’t say anything about the physics behind an extreme event.  As you can see there is large rise in the last decade, especially in the year 2010. This influences the results significant, so the author decide to use instead the data between 1910-2009. The result is much lower when 2010 is excluded. The expected number of heat waves is reduced to 0,47 but this still implies that 78% 0.47 − 0.105∕ 0.47 of the heat waves from the last decade are caused by warming climate. 

Prior to the 80’s there were much less heat waves in Moscow but from the 80’s onwards the numbers of heat waves started to rise again. The authors of this article say contrary to the last study about the Russian heat wave that the chances of  warming having an influence are almost 80% based on tests with Monte Carlo simulations and statistical analyses about mean July temperatures in Moscow. But this only started to develop after the 80’s. The authors do some further tests to examine the effect of the urban heat island but this is only very limited. Because the authors use math to determine the chances they can’t say anything about the physics behind the warming but based on other studies and publications by the IPCC they argue that the chance of humans causing this warming is very high. 

The major downside to this approach is that you don’t what is causing the trend that you are investigating. All the calculations say that the number of heat waves in the Moscow area are growing but they can’t explain why they are growing. Math does not say anything about the physics that create a trend of a growing number of heat waves. The authors say that the recent rise in heat wave to large is to be explained by only natural causes but it remains a miss that the authors can’t say anything about the forces that produce the recent rise. Secondly the author leave the year 2010 but that exactly the event were the whole study from Dole et al was about. The authors of this study, Rahmstorf and Coumou argue that there Is still almost 80% chance that the 2010 heat wave happened due to warming and more specific man made warming but that means that there is still more than 20% chance that something else caused the heat wave such as atmospheric blocking were Dole et al talks about. Thirdly Rahmstorf and Coumou only examine the Moscow area were Dole et al examines whole Western Russia. These differences in scale could also result in different outcomes. 

Both the studies use very different approaches and were Rahmstorf and Coumou focus on almost a  100 years of heat waves around Moscow Dole et al focus only on the year 2010. One studies uses a more qualitative approach and the other a quantitative approach. I think that Rahmstorf and Coumou are right on the longer run but I’m not so sure about the specific case of the 2010 heat wave although 80% chance is very high. But they don’t counter the argument of  atmospheric blocking at all. And that is such an event that could occur due to the remaining 20%.  

Rahmstorf, Coumou ( 2011) Increase of extreme events in a warming world. Proceedings of the National Academy of Sciences, vol. 108, issue 44, pp. 17905-17909 Here

Dole et al (2011) Was there a basis for anticipating the 2010 Russian heat wave.  Geophysical Research Letters VOL. 38, Here