Last week I noted that Andrew Bolt had capitalised on a brief period in which Arctic sea ice extent was around the historical average. My primary point was that relying on short-term changes is a silly way to argue for or against global warming – as evidenced by the fact that the short-term trend in the graph Bolt relied on had already disappeared.
Now Jennifer Marohasy has made the same argument as Bolt – except that she did it only two days ago. She draws on a different data source than Bolt, but her argument is the same one Bolt advanced in May:
The latest satellite data on Arctic sea ice extent suggests that there is now a normal amount of sea ice in the Arctic – normal is defined as about average for the period 1979 – 2007.
Here is the latest graph from Marohasy’s data source:
Clearly this year’s Arctic sea ice extent (red line) was around the average for a while, but equally clearly it is trending below the average. In fact, it is already more than one standard deviation below the average again. So at this point in time, Marohasy’s fundamental assertion – that sea ice extent is close to normal – is wrong. But more importantly, Marohasy’s approach to reasoning about the Arctic sea ice data is also wrong.
The fact that the ice extent was around the historical average for a short time is not valid evidence that the Arctic isn’t melting. But at the same time, the fact that it is currently more than a standard deviation below the historical average is not valid evidence that the Arctic is melting. There will be natural variations in ice extent due to transient factors such as weather patterns. As the National Snow and Ice Data Center notes:
Two standard deviations provide an estimate of the expected range of natural variability.
But the crucial point is that, if we truly were seeing Arctic sea ice’s normal variations unaffected by climate change, we would expect the variation to be +/- 2 standard deviations – in other words, it would be equally likely that ice extent at any given time could be above or below the historical average. The fact that those who argue against global warming have been so worked up about ice extent merely approaching the average highlights how consistently the trend has been for ice extent to be lower than usual. As the NSIDC notes:
For the past few years, Arctic sea ice extent for most months has been more than two standard deviations below the 1979 to 2000 mean, particularly in summer … Because of cooler than average temperatures, Arctic sea ice extent at the end of April 2009 was within the expected range of natural variability.
It disappoints me when scientists, for whatever reason, demonstrate flawed reasoning and the inappropriate use of data. Aside from the direct effect that these scientists have when communicating with the public, it also provides fodder for columnists, commentators and bloggers to propagate the misinformation.
NB: The latest NSIDC sea extent data – the data source that Andrew Bolt relied on and which I examined in my previous post on this topic, is shown below:
4 Comments
I imagine it’s a danger with using one piece of evidence as the basis for supporting a position – the temptation is going to be draw out the bits that agree with your preconceptions. Something we all do, I suppose.
Those pictures could be used either way, as you say. A climate-change skeptic might point to the apparent ‘reversion towards the mean’ from the 2007 low – each year’s line looks on the face of it to be moving back towards the ‘norm’ based on the 1979-2007 period. While a ‘climate change believer’ might see just that all three recent year’s are below the norm. It’d be interesting to know what is typical for other unusually low years – say those that occurred during the 1979-2007. Chance are they might also show some reversion to the mean. I doubt that each year is independent of years immediately around it. A year where the Ice level is high/low is likely to be followed by other years that are high/low, but perhaps not so much so if it takes time for ice to melt or new ice to form. That’s a novice’s view, anyway… this novice, at least.
Besides, even the 1979-2007 base period for setting the norm is pretty short – only 29 years. The standard deviation estimate based on that is going to be pretty noisy, particularly if each year is not really an independant observation. So it would be hard to say anything with confidence based on those pictures. It’s when the evidence from a variety of sources is put together that the picture gets much stronger.
Nice post TZ – Interesting to note the slight variability even between the two graphs.
From the same NSIDC pages, this graph is slightly more convincing of the long-term trend. For those familiar with the “graphing” approach to the stockmarket, if the sea-ice extent were a listed company, the downward trend of the successive peaks spells a sick business model.
In my opinion, anyone professing a science background who knowingly ignores data that fails to reinforce their political viewpoint is a fraud.