How Extreme Weather Increases With Global Warming - The Basic Version

A more technical version of this article, written by Rob Painting, is available at Skeptical Science.

It seems common sense that as the Earth warms we will see more record-breaking warm extremes, and less cold ones. What the authors of a recent study, Rahmstorf & Coumou (2011), sought to find out is how much of these extreme events can we put down to the slowly evolving change in climate, and how much is due to random variations in weather.

The authors developed a statistical approach to evaluate record-breaking events. They found that long-term warming increased the odds of record warm events in global temperature, and when applied to the 2010 monster summer heatwave in Moscow, Russia, they calculated an 80% probability the record-breaking heatwave would not have happened without climate warming.

Analogy time - If it floats your boat
Before going any further, a useful analogy here is to consider a boat moored in a marina. The incoming tide is our slowly changing component (warming climate), and waves are our rapidly fluctuating short-term component (weather). Over a period of time we measure the height of the boats mast, from the top of the jetty to which it is tied. Was the incoming tide responsible for the greatest recorded height?, or was it because of random wave action alone?
Rolling the dice
In order to find out how much each process (weather/climate warming) contributed to record-breaking extremes, the authors turned to Monte Carlo simulations. These are calculations in a computer program, where random numbers are run over and over again. The best way to think of this is rolling a dice. Rolling the dice once tells us nothing about the probability of rolling a six, but roll it 100,000 times (as in the experiment) and you can calculate the probability of a six turning up.

Using the actual data from both the NASA GISS global and July Moscow temperature observations (see figure 1) the authors carefully applied statistical analysis to strip out the weather component from the long-term nonlinear climate trend (see figure 2). The remaining long-term climate trend then gave them a template upon which they could run Monte Carlo simulations of the weather (this is comparable to stripping out the waves, in our analogy, and using the rising tide as a template, or backbone for wave simulations).

Figure 1: GISS global, and Moscow July temperature time series 1911-2010. Both data sets have been normalized (i.e. scaled into a common refererence frame so comparison can be made). NB: Normalization (scaling) has the effect of making the change at Moscow look smaller than it actually is, due to very large annual fluctuations in temp. The long-term increase over the 100 year period is 1.8°C, more than twice the global average.  Adapted from Rahmstorf & Coumou (2011).