Extreme weather – it doesn’t take much warming to make the extremes much more likely

I’m in New Zealand right now, in the middle of pouring rain. But I’m enough of an Australian, and an insurance professional, that I can’t help reading everything my very slow internet connection will let me read about the bushfires.

I’m so relieved that despite needing that “Catastrophic” rating for bushfire danger, NSW at least seems to have escaped rather more lightly than Victoria did nearly four years ago.

I wrote a post after that extreme event that bears repeating now.

One of the lessons I learned early on as an actuary is that the extremes of any distribution do not behave the same way as the middle. So if the middle of a distribution of temperature rises by 1 degree, the effects on the extremes are not intuitive.

Something that was a 1 in 100 chance of occurring in the old distribution is not just slightly more likely. Moving the distribution up the curve can significantly increase the chance of that 1 in 100 event. For an extreme event that is three deviations away from the mean of a distribution, moving the mean up 20% of a standard deviation doubles the chance of that extreme event. So put it in temperature terms – say the summer mean maximum temperature is 25 degrees, with a 1 in 400 chance of a 40 degree day. Increase the mean maximum temperature to 26 degrees, and there is a 1 in 200 chance of a 40 degree day. And that is assuming that the climate follows a nice stable normal distribution model.

It seems the Bureau of Meteorology has had to recognise this increase in the extremes; as the SMH reported today, they have added two more colours to their temperature charts to recognise the possibility of temperatures above 50 degrees celsius.

Given that the flash point of eucalyptus oil is 49 degrees, bushfire risk does not just increase in a linear way with temperature. Brace yourself for many more days of frayed nerves, extreme heat, and guarding against stray sparks.