Cloud patterns have been shifting over the past 30 years in ways that a new study says are possibly due to global warming—-and may even lead to more warming in the future.

Climate scientists believe cloud changes are one of the biggest sources of uncertainty in climate models and understanding how cloud patterns respond to rising greenhouse gas levels is critical to determining how much and how quickly global temperatures will rise.

This new study, published Monday in the journal Nature, provides for the first time a reliable record of past cloud changes spanning nearly three decades and a comparison of those changes with climate models. This brings researchers much closer to solving the mysterious cloud-climate relationship.

Most climate models have projected that global warming would cause the tops of certain clouds to move higher in the atmosphere and also trigger a decrease of cloudiness in the subtropics, expanding the dry zone there. The models also predict that these patterns will trigger more warming, creating what’s called a positive feedback loop.

The new findings offer “more evidence that clouds are going to be … an exacerbating factor” on climate change and not a mitigating one, lead author Joel Norris told InsideClimate News.

Norris is a climate professor at the University of California San Diego’s Scripps Institution of Oceanography. He conducted this study with five scientists from the University of California Riverside, Lawrence Livermore National Laboratory and Colorado State University.

Because the only satellites available to monitor clouds for decades weren’t designed for this task, scientists haven’t been able to conclusively track cloud changes and compare them with the model results—until now. The study shows that these two cloud pattern changes predicted by the models have already been occurring in parts of the globe since at least the 1980s.

“It think it’s a very good study but it’s not really the smoking gun in terms of proving we know” how exactly clouds are impacting global warming, said Dennis Hartmann, a professor of atmospheric sciences at the University of Washington. Hartmann was not involved in the study.

Measuring cloudiness has been difficult, if not impossible, for researchers because they’ve had to rely on satellites designed to measure weather patterns, not clouds. Scientists seeking to use this data grapple with a daunting array of challenges: satellites shift orbits over time; sensors degrade; instruments have to be replaced. So the data are inconsistent, and previous attempts to correct for these issues have left flaws (called “artifacts”) in the data that have led to incorrect or ambiguous interpretation.

To fix that, Norris and his colleagues developed a way to systematically find and remove these data artifacts. This allowed them to clean up the data from two satellite projects—International Satellite Cloud Climatology Project (ISCCP) and Extended Pathfinder Atmospheres (PATMOS-x)—to clearly observe cloud patterns between 1983 and 2009.

“It’s a huge accomplishment,” said Joyce Penner, an atmospheric science professor at the University of Michigan who was not involved in the study.

Their findings showed an increase in cloudiness in some regions and a decrease in others. They found that the subtropical dry zones—longitudinal bands containing many of the world’s deserts—are expanding. These areas could expect to see warmer surface temperatures and more evaporation, and possibly exacerbated droughts too.

The corrected data also showed the tops of the highest clouds are moving even higher in the atmosphere across the globe. This pattern could be even worse for the climate because clouds both absorb thermal radiation emitted from the Earth’s surface and emit some of that radiation to space. How well a cloud emits radiation to space depends on its temperature. Low clouds are more effective than high clouds at emitting radiation.

The researchers confirmed their findings with three other sources of satellite data used to indirectly measure cloudiness during the same time period.

After confirming that the cloud patterns they’d observed were real, Norris and others ran thousands of climate model simulations to see how they compared. Models set up to track changes in cloudiness due to natural variability did not match the observed patterns. Nor did models measuring the impact of ozone or other factors.

But models tracking cloud changes linked to greenhouse gas emissions or the impact of volcanoes spewing particles in the atmosphere did match. This means it is likely climate change along with the recovery of the atmosphere from volcanic eruptions has influenced the cloud changes observed between the early 1980s and late 2000s.

Penner said, however, the study doesn’t attempt to prove these cloud changes will then exacerbate global warming.

Hartmann praised the study, but said it raised a lot of questions, including why the observations revealed much more dramatic changes to cloudiness than the models predicted. Hartmann also said the study’s short time period was a limitation and questioned the findings that natural variability did not appear to account for the changes to cloudiness in the last few decades.

The study identified general changes in cloud patterns and did not attempt to measure and quantify such changes, either globally or in specific regions. Without these details, it’s difficult to predict how changing cloud patterns might impact Atlantic hurricane frequency or paths, for example, or other regional weather phenomena.

“It’s very enticing,” Penner said. “It leaves all these issues hanging out there that need further work.”