Probing the reasons behind the changing pace of warming

Some bits of the puzzle have been known for a while. For instance, during El Nino years, warm water spreads out across the equatorial Pacific and the ocean releases heat into the air, warming the air measurably. That is what happened in 1998.

But while El Ninos come and go within a year or so, there are other cycles in heat distribution and circulation of the oceans that operate over decades. They include the Pacific Decadal Oscillation and the Atlantic Multidecadal Oscillation (AMO), both of which have been implicated in climate fluctuations in the 20th century. So have these or other ocean cycles been accelerating the uptake of heat by the oceans?

Virginie Guemas of the Catalan Institute of Climate Sciences in Barcelona, believes so. In a paper published in Nature Climate Change this week, she claims to provide the first "robust" evidence linking ocean uptake of heat directly to what she calls the recent "temperature plateau" in the atmosphere.

By plugging detailed measurements of recent atmospheric and sea temperatures into EC-Earth, a European model of interactions between atmosphere, oceans, ice and land surfaces, Guemas found that about 40 percent of the take-up was in the tropical Pacific, and another 40 percent in the tropical and North Atlantic.

She told me that it seems likely the changing thermohaline ocean circulation, which starts in the North Atlantic, plus the cycles of El Nino and perhaps the AMO, may play a prominent role. She thinks her model could have predicted the recent slowdown of atmospheric warming ahead of time.

That would be a breakthrough, but nobody has done it yet. Meanwhile, the climate modellers are skating on thin ice when they make predictions that play out over the timescales of a decade or so on which ocean cycles seem to operate. These forecasters can claim that, all things considered, they have done pretty well. But the forecasts remain hostages to fortune.

If anything, the recent pause shows the model forecasts in a good light. Myles Allen, a climate modeller at Oxford University in England, reported in Nature Geoscience last month on an audit of one of his own forecasts, which he made in 1999. He had predicted a warming of a quarter-degree Celsius between the decade that ended in 1996 and the decade that ended in 2012. He found that, in the real world, temperatures got too warm too soon during the 1990s; but the slackening pace since had brought the forecast right back on track.

That shows the forecast is performing well so far, but Allen admitted it might not stay that way. If temperatures flat-line out to 2016, his model's prediction for that year will look no better than a forecast based on a series of random fluctuations.

Some in the mainstream climate community privately admit that they were caught out by the slackening pace of warming in the past decade or so. Back in the 1990s, some suggested - or at least went along with - the idea that all the warming then was a result of greenhouse gases. They were slow to admit that other factors might also be at work, and later failed to acknowledge the slowdown in warming. As Allen pointed out earlier this year: "A lot of people were claiming in the run-up to the Copenhagen 2009 [climate] conference that warming was accelerating. What has happened since then has demonstrated that it is foolish to extrapolate short-term climate trends."

Not surprisingly they have been taken to task for this hubris. Roger Pielke Jr., an environmental studies professor at the University of Colorado at Boulder, who enjoys baiting the mainstream, told me last month: "It is good to see climate scientists catching up with the bloggers. They should ask why it took so long to acknowledge what has been apparent to most observers for some time."

But modellers are now responding more actively to the new real-world temperature data. For instance, the Met Office's Stott reported last monththat global temperatures were following the "lower ranges" of most model forecasts, and that higher projections were now "inconsistent" with the temperature record.

And last December, the Met Office downgraded its best guess for temperatures in the five years to 2017 from 0.54 degrees C higher than the average for the late-20th century average to 0.43 degrees higher. It said the new forecast was the first output of its latest climate model, HadGEM3, which incorporates new assessments of natural cycles.