A peer-reviewed paper showing that variation in global mean temperatures over the past decades is due to variations in the Southern Oscillation index? Authors claiming that it disproves anthopogenic influence on global warming?
“Figure seven from our original paper showed there’s really not much room in there for man to do anything about it.
“As the temperature’s going up and down pretty much in synchronisation with the southern oscillation seven months earlier, I can’t see that man plays much role at all.” – John McLean
Whoa, whoa! Back up there, Johnny. In fact, the paper wasn’t about climate change or global warming at all. It applied a ’12-month averaging differential filter’ to long-term temperature data to remove long term trends, and with those trends removed, much of the variability was explained by the SOI. It’s somewhat interesting, but already a well-known phenomenon.
If you don’t know what that type of filter is (I didn’t at first), here’s my (own layperson) analogy:
We’ll use daily temperature as an example, and apply an 8-hour rolling differential filter. To do that, for any 8-hour period, we take the average value. We then subtract that value from the mean of the next 8-hour period, to find a ‘differential’. So, at 4pm one day, the average temperature in the preceding 8 hours might be 30C, and from 4pm to midnight, it’s 25C. So we have a differential of -5C. Here’s such a trend for Amberley Airport over the past day:
OK, so that looks a fair bit like what you’d expect of a mean temperature graph for a day, right? However, it peaks around 18:00 (6pm), which is later than you might think for the hottest part of a day. That’s OK – it’s what you expect with a rolling average setup, because it lags behind the actual temperature. Also, you’ll notice it varies above and below zero – the data is relative, not absolute. That means you can get a differential from +5 to -5 C in winter or summer; they’d look identical if plotted on the same graph.
So what is controlling this pattern? I think it’s the sun! Let’s superimpose a plot of ‘time from the sun’s zenith’ (how high the sun is) onto our plot above:
So, as the red line goes up, the sun comes up, but it looks like the temperature differential lags behind a bit. So let’s correct the sun’s trajectory by bumping it forward a few hours:
Wow, what a correlation! If we push the height of the sun forward 5 hours, it matches almost perfectly with the 8-hour average temperature differential. Clearly, this explains almost all of the variation in mean temperature near Amberley. If we extended this calculation out for 365 days, it would look much the same. There’s “not much room” for the effect of anything else. Like, say, seasons.
Huh? Seasons don’t have an effect on the temperature at Amberley?
That doesn’t follow from the graphs above! All I showed is that, if you compare the difference between one 8-hour period of temperature with the next 8-hour period of temperature, it will peak about 5 hours after the sun reaches its highest point. In other words, day will normally be warmer than night, because of the sun rising and setting. Mhm. And this holds in both winter and summer! You heard it here first.
However, correlating daily temperatures with the sun rising and setting says nothing about longer-term trends – the Earth’s 24-hour rotation does not cause seasonal variation, for example. Instead, it’s the fact that the Earth is tilted on its axis of orbit, so the angle of the Sun in the sky varies. If we look at a monthly, rather than a daily, trend, the angle of the sun in the sky will be a more important factor influencing temperature – the variation from the sun going ‘up’ and ‘down’ cancels itself out. There’s a long-term trend (season) superimposed on a short-term trend (day/night).
McLean used this differential average technique with air temperatures and the Southern Oscillation Index, which looks at the differences in air pressure (related to temperature) on either side of the South Pacific Ocean. He found that there was a good correlation with the SOI and global temperatures over the past decades, when global temps were smoothed using the differential method.
That’s certainly more interesting than me proving that sunlight warms up the Amberley area, but it’s been known about for decades. However, like good climate skeptics, McLean and co. used their paper as evidence against AGW in the media. It’s a nasty ploy, because their analysis (seems, to my eye), scientifically sound, but the conclusions they’re drawing from it and touting in the media are, simply, unsupported by their paper. By making it look like the SOI controls global temperature, and without explicitly stating that they’d taken out the long-term warming trend, they can pretend they’ve disproved the climate change consensus. It’s like me saying that my day/night data disproves the “Season alarmists” who tell the world that it’s going to get cold in winter, because the sun rising and setting is the main factor in the temperature variation we measure.
Fortunately, a group of other climate scientists published a reply in the same journal, criticisng the McLean paper and clearly laying out the difference between the findings and the anti-scientific interpretation being perpetuated in the media.
More detail on the story can be found over at The Drum. I’d recommend the article simply because it outlines why peer review is a useful tool: it’s not infallible, but it doesn’t discriminate based on name, or institution, or ideology: it functions based on the data and logic presented, and the conclusions drawn from that. You don’t need to be a ‘scientist’ to publish in the peer-reviewed literature (though you could probably call yourself one if you do), but if you want to wave your science around in the media in support of your position, you’d better make sure it actually backs up what you’re saying!
NB: I’m no climate scientist, so there could be errors in this post! I am fairly certain the thrust is correct, though. More links and commentary about this issue here.