Abstract

Atmospheric aerosols are known to modify cloud properties, with a net cooling effect on the earth. However, the magnitude of this cooling effect is highly uncertain, largely due to the difficulty of separating the changes aerosols cause in cloud properties from those of other meteorological factors. For example, both aerosol optical depth (AOD – used as a proxy for aerosol number) and cloud fraction (CF) are strongly affected by relative humidity, such that the correlation between AOD and CF is not a good indicator of the causal effect of aerosols on CF. As CF is strongly correlated to many other cloud properties, this in turn obscures the causal effect of aerosols on these cloud properties.

The standard method for separating causal effects is direct experiment. Unfortunately, direct experiments on the atmosphere are often difficult, expensive and even unethical, requiring approaches based only on observation. In this talk, several possible methods for separating the effect of aerosols on clouds from those of the surrounding meteorology are demonstrated. These observational results are compared to results from global climate models to assess how well they can actually constrain causal aerosol-cloud relationships. Finally, techniques more common to economics and public health are used to evaluate the effect of aerosols on CF at a global scale, reducing the uncertainty in the anthropogenic forcing of the climate.