Clouds aren't as easy to track deep into the past as carbon dioxide. But like CO2, clouds can play an important role in climate change: Either they can trap heat in the atmosphere, magnifying to the greenhouse effect, or they can reflect sunlight back into space, cooling the planet.

So will clouds contribute to climate change or help mitigate it? "Right now, we don't know what that relationship is," says NASA atmospheric scientist Anthony Del Genio, who works at the NASA Goddard Institute for Space Studies in New York City.

But a new study, published today in Science, takes another step forward in trying to piece together this complicated dynamic, which will be crucial to understand for accurate pictures of what Earth will be like decades and centuries from now.

When it comes to predicting climate change, not all clouds are created equal, notes lead study author Amy Clement of the Rosentiel School of Marine & Atmospheric Science at the University of Miami. Higher-level clouds, such as thunderheads, have a greenhouse effect (producing moisture and reemitting radiation back onto the surface), she explains, whereas low-level clouds act more like an umbrella, sheltering Earth from the sun's warming rays.

Clement and her team examined low-level stratiform clouds over the northeastern Pacific Ocean. By comparing independent sets of observational data gathered by sailors and satellites during the past 50 years, they hoped to see how the cloud cover reports matched up with temperature and wind circulation—and with one another. To their surprise, the space and sea observations were uncannily similar, which helped add credence to these data sources, which many had criticized as unreliable.

They found one climate model (from the U.K.'s Hadley Center for Climate Change) that complemented their data especially well. It showed that warming surface temperatures and decreasing air circulation—trends which are both predicted to continue in a changing climate—meant fewer low-level clouds. And that meant even warmer surface temperatures.

But a lot more work remains to be done. "I think it's a very impressive piece of observational analysis," says Del Genio, who wasn't involved in the study. "It's the first time anyone has demonstrated these decadal changes." But, he adds, more modeling will be the only way to back up their tentative conclusions.

Low-level clouds are a difficult target for modeling, Clement admits. "They're forming on a microphysical scale." The Hadley model was probably the most successful, Clement notes, because it had the most information about the complex processes that happen in the lower atmosphere, where it comes in contact with Earth's surface—a zone, Clement explains, that is much more difficult to model than large-scale circulation in the upper atmosphere.

And, as always, there is the question of how one reconciles day-to-day weather events with the long-term climatic trends. As Clement points out, "If you look at the hour-to-hour cloud processes, you get a very complex story." But, "the data, on the decadal time scale, seems to have this very simple story: When the ocean surface is warm and the circulation is weak, the cloud cover is reduced."