I still remember the first time I was asked about how climate change affects El Niño. It was given as a group exercise during a winter school in Les Houghes (in France) back in February 1996. Since then, I have kept thinking about this question, and I have not been the only one wondering about this. Now I had my hopes up as a new study was just published on the evolution and forcing mechanisms of El Niño over the past 21,000 years (Liu et al., 2014).
The main results published by Liu et al., (2014) rely on simulations with one climate model (CCSM3), but we know that different climate models indicate that a global warming will have different effects on El Niño. So why would I trust new computer simulations with a model that has a coarser representation of the atmosphere and ocean?
The spatial resolution of the model they used was described by the cryptic description “T31x’3″. The “T31” part of this refers to the atmosphere and is a technical term describing how the calculations are made.
Since our planet is spherical, it is possible to represent the differential equation describing the motion of air as a set of spherical harmonics rather than calculating the results for each grid box seperately. Basically, this is a clever mathematical “trick” also used in weather forecasting. “T” is short for triangular truncation of spherical harmonics, and “T31” means that the model represents all total wave numbers from 0 to 31.
In other words, “T31” means that the description of the atmosphere is rather coarse (resolution of 3.75×3.75 degrees) in terms of its spatial details. This is understandable because it would require too much computer resources to compute the climate with more details if you want to account for 21,000 years.
You can use a similar model with very high degree of detail for a weather forecast because you only want to compute the atmospheric state for a few days.
A climate model also includes an ocean component, and the description of the ocean usually requires finer details (higher spatial resolution) than the atmospheric model (due to different Rossby radius of deformation).
Although the atmosphere model is low resolution in this case, the ocean is relatively high. But some crucial aspects are not clear from the paper: if the ocean resolution is 0.3 degrees or 0.9 degrees near the equator.
Nevertheless, the ability to represent the El Niño phenomenon is limited in the model experiment, and indeed, its time scale tends to be closer to 2 years than the observed 2-7-year span.
I think the most interesting aspect of the model experiment was running the model several times with different inputs: CO2 forcing, changes in solar forcing due to orbital changes, the effect of changes in the ice sheets, the effect of meltwater from shrinking ice sheets, and one with all combined.
Also, I liked the way the paper tried to draw the lines to different causes, both through the experimental set up and the choice of diagnostics.
The results indicate that El Niño is fairly sensitive to the different aspects, but the picture is quite messy. Nevertheless, (Liu et al., 2014) reported a dominant role of precessional (orbital changes) forcing, but they also found that the glacial melt had an effect on El Niño.
The figure reproduced below also suggests that the ice (green curve) had a strong effect at around 14,000 years ago (a big jump). Yet, the orbital (light blue) and meltwater (dark blue) gave an evolution that was closest to the effect of the combined forcing (black).
However, all this is very qualitative. I don’t see a clear picture on the connection between the forcers and El Niño, and I’m no wiser. The quote from the press release “El Niño is driven by an intricate tango between the ocean and the Earth’s atmosphere.” [Kelly April Tyrrell] is still quite descriptive in my mind.
Fig.3 from Liu et al., (2014): a, Amplitude of ENSO. b, Amplitude of the annual cycle.
- Z. Liu, Z. Lu, X. Wen, B.L. Otto-Bliesner, A. Timmermann, and K.M. Cobb, "Evolution and forcing mechanisms of El Niño over the past 21,000 years", Nature, vol. 515, pp. 550-553, 2014. http://dx.doi.org/10.1038/nature13963