Citation: Davini, P. & Cagnazzo, C. (2013), On the misinterpretation of the North Atlantic Oscillation in CMIP5 models, Clim. Dyn. doi:10.1007/s00382-013-1970-y.
The North Atlantic Oscillation (NAO) is a big hitter in the world of atmospheric dynamics. It’s everywhere, and it (or its close relations the Arctic Oscillation and Northern Annular Mode) invoked in pretty much every explanation of why midlatitude countries get the weather they do. But what is it?
In the simplest terms, the North Atlantic Oscillation is an expression of the location of the jet stream, a band of strong winds which circles the globe at a latitude of about 45 degrees. Weather systems form on the jet stream and are steered along with it. The jet stream wobbles north and south (like most fluids, it forms eddies), and depending on its latitude it might send a storm into the Mediterranean or to the United Kingdom.
The NAO is often used as a neat way to summarise the behaviour of the jet stream. In the real atmosphere it describes a specific pattern of variation in its position, but as this paper (and others) shows, this pattern isn’t exactly the same in climate models.
An aside before we begin. It’s not really correct to say the NAO explains why the jet stream and weather systems are where they are. The NAO must be a certain way if the jet stream is where it is. It’s just a description, like saying, ‘we got a storm because we are at the latitude where the storm was’. The reasons for the way the jet shifts north and south are complex and really not obvious, and we’re learning more about it all the time.
So that’s what the NAO is in general terms. When we analyse atmospheric data we give it a mathematical form. This involves a tricky but really rather clever analysis technical called Empirical Orthogonal Function analysis (or Principal Components analysis). I won’t go into it here but it’s basically a way to pull out a pattern according to which something varies. If you think of the pattern of variation of a swinging pendulum, it’s main variation pattern is a back and forth swing. For the jet stream, the main variation is a north and south wobble – the NAO.
The convention is, when the NAO is in a negative phase the jet is shifted southward, and when it is in a positive phase it is shifted northward. Generally, it’s also wavier in its negative phase than its positive phase, and more waviness means more ‘blocking’, which is when we get big meanders, producing high pressure systems which hang around for a week or so and give stable, calm weather.
But some climate models don’t have an NAO that behaves like this. For some of them, the main variability includes too much northward wobbling and not enough southward wobbling. Another group has variability which is more like a pulsing of the speed, going from fast to slow and back again, without any wobbling. For some the north-south wobbling is simply too weak.
This paper shows the weakness of the mathematical definition of the NAO – it’s not the same between different models! The examples above represent different physical processes so it’s meaningless to lump them together like this. But, importantly, all the models do have something like the real-world NAO – that is, the north-south wobbling. It’s just that the standard mathematical definition doesn’t always pick that out. This is the peril of using a mathematical construction with no physical basis to define variations in the jet latitude which are the result of specific physical processes.
Obviously it’s better to think of things, if we can, in terms of the physical processes involved rather than this rather obscure mathematical idea. Davini and Cagnazzo end their paper with this recommendation:
…since climate models represent a slightly different world with respect to the real one, special caution must be applied…when the NAO is studied. We conclude suggesting that instead of using the NAO to study the North Atlantic variability it would be better to adopt diagnostics based on jet stream position and strength