Climate Change: It is the extremes that kill rather than the average

A lot of the information give about the affects of climate change are about what happens to the average temperature, the average rainfall etc. However, it is important to realise that changes in the average also affect the probability of extreme events much more significantly.

This post is not about any particular prediction but is about understanding some concepts. Therefore I shall be demonstrating this with idealised Gaussian distributions. In reality the distribution will be different but the concepts are the same.

Changes In The Mean

Consider the two (blue and red) distributions below. The peaks on the graph are the values that you are most likely to get i.e. the mean (note that the mean and the mode are the same since the distributions are symmetric).

The mean of the red graph is slightly shifted compared to that of the blue. The distribution could be for many variables – rainfall, winds or storms – but let us assume for the moment that we are talking about temperature.

Now let us say that an extreme high temperature is given by the green vertical line.

The probability of temperature higher than this is given by the area under the curve to the right of the green line. You will notice that in the red case the area (red shaded area) is very much larger than the blue case (blue shaded area). A small change in the mean temperature can dramatically increase the probability of extreme temperatures.

Changes In The Variation

Now let us consider the case where the mean changes the same amout but also the variation (i.e. the standard deviation) also increases.

As you can see this increases the probability of an extreme event even more.

If we increase the variation (standard deviation) even further then we get the following:

Not only has the probability of extreme heat increase the probability of extreme cold (pink area) has also increased.

Just a reminder that this does not have to be temperature – it could mean that there is an increased risk of flooding and increased risk of drought.

The Heat Wave Of 2003

In 2003 there was a heat wave across many countries in western Europe which claimed many lives. Although there have been many other heat waves (Pakistan in 2015 and 2017) which has claimed many lives this one has been more carefully studied.

About 15,000 people died due to the heat in France, which led to a shortage of space to store dead bodies in mortuaries. Temporary mortuaries were set up in refrigeration lorries.

The Heat Wave Of 2003, Met Office1

Some estimates give the death toll as 70,0002. (note that estimates vary partly because different workers use different methodology -e.g. excess deaths vs reported deaths).

It is not possible to attribute this heat wave to climate change. Such events always have a finite probability of happening (the blue area under the curve). What do know is that with increasing temperature the probability of such events increase very dramatically.

Michael Wehner of the Lawrence Berkeley National Laboratory3 in the USA has estimated using computer models that the risk of 2003 had doubled due to climate change at the time of the event. He went on to suggest that by 2023 the risk had increased 35 fold and by 2040 such extreme heat waves would be the normal summer temperatures (154 times the probability compared with the pre-industrial period).

So if the summer of 2003 becomes our new average after 2040 how many tens of thousands will die if we have an extreme heat wave?

As Professor Peter Stott from the Met Office told the recent UN climate summit in Katowice, Poland:

Humanity just won’t be able to cope with the world we are heading for.4

1 The Heat Wave Of 2003, Met Office (

2 Death toll exceeded 70,000 in Europe during the summer of 2003, Jean-Marie Robine et al, Comptes Rendus Biologies Volume 331, Issue 2, February 2008 (

3 High Performance computing of Extreme weather in a changing climate, Michael Wehner Lawrence Berkeley National Laboratory (

4 Climate change made UK heatwave 30 times more likely – Met Office , The Guardian, 6 December 2018 (