The real story behind Walker Circulation headline

The recent article, Slowdown of the Walker circulation driven by tropical Indo-Pacific warming, by Tokinaga et al. (Nature, 15 November 2012, 491, pp 439–443) could prove very important for Australian climate.

Tokinaga et al. claim that the Walker circulation (an East-west overturning circulation in the tropical atmosphere) has slowed down since 1950, and show using climate model simulations that this can be explained by the changing pattern of tropical sea surface temperature (SST), in particular the zonal gradients in the Indo-Pacific region.

A global-warming induced slowdown of tropical atmospheric overturning is expected due to energetic arguments, and a downward trend has already been reported in observations. Tokinaga et al. point out that they can explain the Walker slowdown from SST changes, which they claim is at odds with the energetic explanation.

But I believe this contradiction results from mistakenly conflating the overturning and Walker circulations: the former must be energetically regulated within the atmosphere, while the latter---which is only part of the total overturning---should be largely controlled by sea surface temperatures. This distinction is analogous to that between the thermohaline and meridional overturning circulations in the Atlantic Ocean, which has long vexed oceanographers.

However, lurking beneath this debatable headline is the real significance of the Tokinaga et al. study: when they look at subsets of SST data that are relatively immune from inhomogeneities due to changing observing practices (in particular, bucket temperatures or nighttime near-surface air temperatures), they find a very different SST trend pattern than the one in standard datasets such as HadlSST.

In particular, the Maritime Continent region warms less than does the East Pacific, while the opposite occurs in HadlSST. The newly derived SST trends, applied to an atmospheric model, predict sea level pressure trends in the Australian and Maritime-Continent region that agree with those in ship observations, but are opposite to those predicted with standard SST forcing datasets. This is very important since, regionally, rainfall is closely related to sea level pressure.

The tropical atmosphere is extraordinarily sensitive to small changes in SST patterns. This study shows that regionally varying errors in the standard datasets are enough to change the sign of predicted climate trends in Australia since 1950.

This may explain why SST-forced runs of atmosphere models have typically failed to reproduce Australian rainfall and other climate trends, and points to a possible resolution of this problem -- which may not be due entirely to bad models, after all.

Author: Prof Steven Sherwood

UNSW logo ANU logo Monash logo UMelb logo UTAS logo