BBC News 24 Jul 09;
Clouds over the North-East Pacific dissipate as the ocean warms, according to a study in the journal Science.
Researchers have described this as a "vicious cycle" of warming, as reduced cloud cover allows more of the Sun's rays to heat the Earth.
They say warming could gradually reduce the low-level cloud cover that is thought to help cool the globe.
But the team stressed that it was not yet possible to quantify how much this might impact on global temperatures.
They said that accurate simulations of these cloud effects would improve the models scientists use to predict future climate change patterns.
The accuracy of these models has been hampered by the uncertain influence of clouds on the global climate system.
The low-level clouds studied here are of particular interest, as they have been shown to have a net cooling effect on the Earth, by reflecting the Sun's rays.
Lead author Amy Clement, from the University of Miami, US, tried to resolve the uncertainty in the cloud data.
Each satellite that gathers cloud data is slightly differently calibrated, and observations made by people from the Earth's surface have to be subjective. "So if there were trends in either one of those (data sets), you would be very suspect," Dr Clement told the Science podcast.
So she and her colleagues looked at both satellite and surface-based observations together.
"Basically, our approach was to take imperfect but independent data sets and add them together," she told BBC News.
"Where they agreed we said it added confidence that the signals were real and not just some spurious trend."
Dr Clement described the findings as "almost shocking".
They noticed that, in the past 50 years, there had been a "positive feedback" cycle in the low-cloud cover, so when the surface of the ocean was warmer, there had been less cloud cover.
"These are subtle changes that take place over decades," Dr Clement commented. "But it's indicative of a vicious circle."
Better modelling
The authors then tested leading climate models and found that only one - a model designed by scientists at the UK Met Office's Hadley Centre - reproduced this cloud effect.
This particular model predicts one of the more pessimistic climate change scenarios, with temperature increases at the high end of the range of forecasts.
Dr Clement said that this range of different projected temperature increases, in response to greenhouse gases, had been "almost entirely" caused by the uncertainty about cloud feedback.
Daniel Lunt, a climate scientist from the University of Bristol, UK, said this was an "important finding", but that it would be a "quantum leap" to conclude that this single model's predictions about the effects of cloud cover on the future climate would be correct.
"Cloud feedbacks do not necessarily work in the same way under conditions of natural variability compared to (how they will work during) future carbon dioxide-induced warming," he explained.
But Dr Clement said: "This is coming from a model where this cloud feedback has been validated against observations. So the fact that (its predictions are) at the upper end of the range is something that merits more attention."
Dr Clement stated that the findings provided "a new way of looking at cloud changes".
"This can help improve the simulation of clouds in climate models, which will lead to more accurate projections of future climate changes," she said.
Matt Collins, a researcher from the Hadley Centre, said that the findings gave him confidence in the ability of models to predict climate change.
It was impossible to extrapolate this one test and say exactly what would happen in the future climate, he told BBC News. "But we have been studying cloud feedback mechanisms for years, and this gives us justification to pursue this project."
