Guest essay by Philip Lloyd
There are constant claims that extreme events are becoming more frequent, but when you really dig down, you cannot see any trends even in long-term data. Of course, the scaremongers claim that it hasn’t happened yet, but their models predict it is going to happen any day real soon now, just you wait! All agree it has been warming for at least the past 150 years. If there were any effect such as the models predict, surely we would have seen it by now? It surprises many, but there is no detectable trend in extreme events in the historical data sets.
However, it is not quite straightforward. For instance, how do you define an extreme event, particularly with phenomena that are not normally distributed? Do you only have to consider the high extremes, or must you also consider the low extremes? And how many extreme events does it take to determine a baseline, let alone a trend?
To illustrate the challenges, consider the longest rainfall record we possess, that of England and Wales, which has monthly data back to 1766. The annual totals are close to normally distributed, as shown in the figure. A multi-parameter distribution such as a Weibull would do a better job, but we can treat the distribution as normal for the purposes of this exercise. The average annual rainfall is 918mm with a standard deviation of 119mm.
The next figure plots the annual rainfall since 1766 with the upper and lower 95% (two standard deviations) confidence limits. There is a very slight increase of 0.19 ± 0.11 mm per year which is not statistically significant. However, it is of the order to be expected from the Clausius-Clapeyron relationship for the warming over this 250 years.
We would expect 12.5 extreme events in 250 years, if an extreme event is defined as one that exceeds the 95% confidence limits. The figure shows that there are seven such events above the upper limit and four below the lower, or 11 in total, where 12.5 had been expected. Given the slight skewness of the data and the approximation of normality, the difference is not significant.
What is significant, however, is that there is no detectable change in the frequency of the extreme events. Indeed, to detect such a change with any degree of confidence, you would need far more than eleven events or, in the present case, far longer than 250 years. So those who claim we are facing disaster from “climate change” need to reflect on the fact that even with a generous >95% measure of extremeness, it took 250 data points to approximate a baseline. How can we tell if an event is extreme if we have no baseline?
Is 95% generous? I think it is. Engineers typically design for the 1:100 year event, not 5:100. For really critical structures, they may use the 1:1000 year event. By and large, the engineers have been successful in protecting us against all manner of natural forces. The Great Kanto earthquake of 1923 devastated Tokyo; it had a magnitude of 7.9. The Great Tohoku earthquake of 2011, which caused the tsunami that destroyed the nuclear reactors at Fukushima, had a magnitude of 9.0 and the rebuilt, earthquake-proofed Tokyo was virtually unscathed.
When you hear that the effects of climate change will fall more strongly on poor nations, realize that it is probably true. It has, however, nothing to do with climate change, and everything to do with some poorly engineered infrastructure in those nations.
Energy Institute, Cape Peninsula University of Technology, Cape Town, South Africa