Extreme events

I had thought of writing about Hansen’s recent paper, but I’m not really sure what to say about it. I did read some of the reviews – Thorne, Archer, Ruddiman – and, unsurprisingly, there appear to be some valid criticisms. On the other hand, the rate at which we’re releasing CO₂ is greater than it’s been released for at least 66 million years, so unexpected outcomes may not be all that surprising. If you’re interested in Hansen’s paper, Stoat has a post, I think Eli has one, but I can’t find it, and Chris Mooney’s article seems quite balanced.

Instead I thought I would write about another paper I encountered today. It’s by Michael Kelly, a Professor at the University of Cambridge. He was also part of a panel that assessed the Climatic Research Unit after the email controversy, typically referred to as Climategate.

The main conclusion of his paper can probably be gleamed from the first half of the abstract

It is widely promulgated and believed that human-caused global warming comes with increases in both the intensity and frequency of extreme weather events. A survey of official weather sites and the scientific literature provides strong evidence that the first half of the 20th century had more extreme weather than the second half, when anthropogenic global warming is claimed to have been mainly responsible for observed climate change.

What I think he misses pointing out here is that his sources are not simply official weather sites and the scientific literature, but also blogs, many of which could justifiably be regarded as promoting science denial.

One major problem with the paper is that the figures are so poor that I couldn’t even read the axes labels, so I won’t bother posting many here. Instead I’ll just comment on some of what he did, and then comment on what seems to be a glaring error. His first figure is the rate of change of surface temperature from 1850 to today. He points out that the rates were supposedly larger in the late 19th and early 20th century. Well, firstly, this appears to be mainly highlighting variability and completely masking the long-term anthropogenic trend. Also, extreme in this context refers to high temperatures, not high rates of change of temperature. Additionally, given the larger uncertainties during the earlier period, it’s maybe not surprising that it shows larger apparent variability.

He then plots precipitation and landfalling hurricanes in the US. There are two key points here. What is expected is an increase in intensity and frequency of extreme events. To consider this, you should really consider events that are regarded as extreme, not simply all such events (i.e., extreme hurricanes or extreme precipitation; not all hurricanes or all precipitation). Additionally, these are rare, so it’s hard to extract a signal in the first place. Considering only one location makes it harder, and – in the case of hurricanes – considering landfalling events, even more so.

He then illustrates that damage from flooding in the US, as a proportion of GDP, has decreased. So what? How has GDP changed over the same time interval? What other changes have helped to minimise damage from flooding? If we really want to understand extreme events, we should consider the actual events, not some proxy from which it is hard to extract a reliable signal. So, he hasn’t really shown anything like what he claims.

However, what I was going to focus on was his claim that NASA has shifted the 1980 temperature anomaly relative to 1940. He claims that in Hansen et al. (1981), the 1980 temperature anomaly was about 0.15^oC less than 1940. Now, it is supposedly 0.2^oC higher, and that this shift produces a significant fraction of the century variation. He illustrates it using the figure below.

Ignoring the fairly obvious issue that our understanding of how to analyse temperature data has changed substantially since 1980, I think there is a far more obvious problem. The left-hand figure is a 5-year running mean. If you compare that with the right-hand figure, there are some quite similar features; there’s a peak at around 1960, a dip just before 1970, and a few wiggles after that. It doesn’t really look – in the right-hand figure – as though these have moved up (relative to 1940) by anything like 0.35^oC. What appears to be the case is that the left-hand figure ends slightly before 1980 and, hence, does not include the fairly rapid increase (in the 5-year running mean) that starts just before 1980. He’s simply not doing the correct comparison, even if such a comparison were a reasonable one to make.

At the end of the day, detecting trends in extreme events is difficult. Given that nothing presented in this paper is a remotely reliable way of determining this, suggesting that there has been no increase is simply wrong. However, even if it is difficult, we do have some confidence that some extremes should show an increase in intensity and frequency; heatwaves and extreme precipitation events are two examples. Others are more complicated and uncertain, although there is some indication of an increase in the frequency and intensity of extreme Tropical Cyclones in some ocean basins. However, if we keep adding energy to the climate system, it would be extremely surprising if it didn’t have any effect on the intensity of extreme weather events, and – given that there’ll be more energy – it might be surprising if the most extreme events didn’t become more energetic.