Part of the difficulty in pinning down a value for climate sensitivity is the uncertainty in how clouds will respond to climate change. Now a team from the US National Center for Atmospheric Research has examined satellite observations of relative humidity in the subtropics – a physical factor related to, but more easily measurable than, cloud formation – to check the success of climate models in recreating today's climate. They found that models with a higher climate sensitivity – i.e. those that project a higher temperature rise in response to increasing greenhouse gases – performed better.

"The main goal of our work was to evaluate the representation of clouds in models in order to reduce uncertainty in future climate change arising from greenhouse gases," John Fasullo told environmentalresearchweb. "However clouds are tricky. Observations are fraught with errors and comparisons with models are not straightforward. Clouds are also tuned somewhat in models, thereby masking underlying problems. So as an alternative we evaluated the representation of relative humidity, as models use relative humidity as a key input in formulating their clouds."

Fasullo and colleague Kevin Trenberth used data from NASA's AIRS (Atmospheric Infrared Sounder) and CERES (Clouds and the Earth's Radiant Energy System) instruments to look at the relationship between seasonal extremes in relative humidity and albedo, which is related to the presence of clouds.

"Our findings suggest that many of the models are unable to represent the low-latitude dry zones – regions of the tropics and subtropics where subsident air driven by seasonally varying deep convection achieves relative humidity of about 20%," said Fasullo. "Moreover, as a systematic relationship exists between the aridity of these zones, clouds and future warming, our results suggest the lower-sensitivity models lack credibility in their future projections and that climate sensitivity is therefore very likely to be on the high side of the current estimated range."

According to Fasullo, the implications are substantial. Since the study suggests that future climate changes are likely to be on the high side of current projections, it has major implications for policy related to both mitigation and adaptation. "From a modelling standpoint, it also shows that models can likely be improved considerably using readily available observations and that correctly resolving the vertical structure of the troposphere should be a top priority for modelling efforts," he said.

Now the team is pursuing a similar approach aimed at better understanding projected changes in precipitation and the water cycle. Fasullo said the work is motivated by the fact that uncertainties in projected changes in drought and rainfall on a regional scale are considerably larger than for temperature, yet are central to the nature of future impacts.

The researchers reported their work in Science.