Geophys. Res. Lett. 37, L01702 (2010)

Credit: MARAT KHAIROUTDINOV, STONY BROOK UNIVERSITY

Clouds are one of the largest sources of uncertainty in climate models. That's because global climate models cannot explicitly capture cloud formation. Instead they use a series of equations to describe the average conditions under which clouds form and decay, a technique called parameterization.

Now a team of researchers led by Cristiana Stan of the Center for Ocean-Land-Atmosphere Studies in Calverton, Maryland, has simulated the Earth's climate using a model that explicitly represents cloud processes. Stan and colleagues analysed simulations from a standard coupled climate model. They then replaced the cloud parameterization in their model with an embedded two-dimensional cloud process–resolving model and ran the simulation again. They found that when clouds were explicitly represented in the model, the climate simulation improved in several respects, including more accurate depiction of seasonal precipitation patterns and of several important climatic phenomena, such as the Madden–Julian Oscillation, the Asian monsoon and the El Niño/Southern Oscillation.

The study suggests that clouds should be explicitly represented in climate models for more accurate simulations of the climate.