All this fuss about the “F-word” should not distract from the fact that the mainstream media has problems reporting a science story even if we do pitch it right. Take, for example, Bryden et al.‘s 2005 Nature paper, which reported a 30% slowdown in the strength of the thermohaline circulation (THC), also known as the meridional overturning circulation (MOC), measured across the Atlantic basin at 25ºN. In the wake of this current’s starring role in the ever-so-slightly unrealistic The Day After Tomorrow, this result got widely and breathlessly covered in the press. Even at the time, though, it was clear that this result was preliminary at best. As I commented at the time (in one of my earliest blog posts):
Because the current strengths are being estimated by an indirect method, the potential errors are quite large; in fact, the authors estimate that the error is about the same size as the apparent decrease in [THC] flow. They also argue that because the observed reduction is associated with a particular component of the flow, rather than distributed at all levels, the effect is real, which is a fair point but not a conclusive one. The other problem is that there is virtually no data on the natural variability of the thermohaline circulation, so it is also possible that the most recent survey is just detecting this rather than a sustained reduction.
Some of these caveats did make it into the better write-ups, such as the BBC one linked to above, but most of them treated the reported slowdown as a pretty solid result. End of story. Except that, of course, it wasn’t.
First off, there was the rather inconvenient fact that such a large drop in THC strength was not consistent with other data and models:
[climatologists’ doubts] stem from two main sources. The first is observational: a reduced thermohaline circulation should reduce heat transport into the North Atlantic, but there has been no drop in high latitude sea surface temperatures, and Europe has been warming, not cooling, in the last decade. Secondly, the estimated increase in fresh water added to the surface North Atlantic is lower, by an order of magnitude, than that required by modelling to stop NADW formation. Much more warming is required before the thermohaline circulation should be seriously affected.
To be fair to Nature, they gave climate scientists the space to point this out; but I don’t recall any follow-ups in the Independent.
Then there was the fact that the cruise which produced the final snapshot profile used in Bryden et al. had installed a series of buoys designed to continuously monitor the circulation. And what did preliminary results presented at the end of last year show?
It seems that over timescales of a few days there are fairly sizeable shifts in the amount of North Atlantic Deep Water (NADW) crossing the 25ºN transect. These results pretty much destroy any hope of extracting any meaningful long-term trend from the existing data set of decadally spaced, one-time snapshots of the circulation; the ‘weakening’ from 23 to 15 Sv since 1957 proposed in Bryden et al. …is very close in magnitude to the observed natural variability (and their 2004 hydrographic section underestimated the average strength of the present circulation by 2 Sv).
This did get some play in the media, but rather than focusing on the real implications, more attention was focused on an apparent 10-day hiatus in the flow across 25ºN; as I pointed out at the time, this was unlikely to be a result of suddenly weakened downwelling in the Norwegian Sea, but once again prompted talk of “dramatic weakening” (the BBC did a better job, illustrating that framing is indeed often in the eye of the reporter, rather than the scientist).
And now, we have the latest episode in this saga. The high short-term THC variability revealed by the RAPID moorings clearly creates difficulties for extracting any long-term decline or increase, especially if you’re relying on infrequent isolated measurements. Someone has done the math (presumably assuming a net slowdown over the sorts of timescales suggested by the models – maybe a third or so by 2100):
Johanna Baehr, a physical oceanographer formerly at the Max Planck Institute of Meteorology in Hamburg, Germany, has now shown that with this type of information — snapshots from shipboard measurements carried out every 10 to 20 years — it would take well over a century to reliably detect a 30% reduction of the flow.
Which isn’t surprising, but even with continuous measurements we’re going to have to wait a while:
Baehr, who is now at the Massachusetts Institute of Technology in Cambridge, calculates that even with these continuous measurements it will take 20-30 years to detect a statistically significant trend.
So here’s the thing. What we’ve learnt in the last 15 months or so demonstrates that the Bryden et al. paper couldn’t have detected what it claimed to have detected. Another and possibly more far-reaching implication of all this short-term variability is that any measurements of the state of the oceans based on snapshot measurements on research cruises may not be quite as representative as we once assumed. But is that knowledge going to reach many of the people who read or heard the original story? I can understand why the media have not made so much of a fuss about the latest stuff (“we won’t know for some time if we’re doomed or not – more at 11!”), but what’s worse is that someone searching for information on this topic on the web is far more likely to come across stories relating to the original paper than any of the follow up. And in science, the follow-up is almost as important as the original result, if not more so. Does it stand up to the test of time? How is it affected by other new discoveries? What new research does it guide or inspire? In science, the story is always breaking – there’s always something new to be added. The story of a particular study can continue for years after the headlines have died down; and that’s a hard thing for scientists, who can track citations through the literature, to follow, let alone journalists and the public. But if we’re ever going to get people to really understand how science works, we need to illuminate this long tail of small refinements and occasional reverses.
I don’t have any quick or smart solutions to this problem: encouraging the media to get out of their “dead trees’ mentality and judiciously add new links to old stories might be a start, although implementing it might be quite fiddly and time consuming. Science bloggers should perhaps think about doing this too, and to practice what I preach, I’m going to forward-link all of my THC posts. It’s a start, if only a small one. Beyond that, I’m still thinking…
Comments (1)