On the Sensitivity of Model-derived Estimates of Aerosol Indirect Effects and Forcings to Activation Schemes

Linking predicted aerosol physio-chemical properties and size distributions to cloud microphysics in modern global climate models (GCM) is accomplished by droplet activation parameterizations. Presumably, using simplified, physically-based models for this type of parameterization alleviates historical problems related to enforcing minimum cloud droplet number concentration (CDNC) in certain cloud regimes. However, tests of multiple state-of-the-art activation parameterizations in the framework of a GCM coupled with a detailed aerosol model using present day aerosol emissions differ by up to 400% in droplet number in heavily polluted parts of the world and by significant amounts elsewhere. Here, we show how the modeled aerosol indirect effect is critically a function of the baseline CDNC established by a particular activation scheme in much the same way that artificial limits on CDNC in earlier GCMs tended to suppress the indirect effect. The magnitude of changes in modeled shortwave cloud forcing between pre-industrial and present day emissions scenarios correlates directly to the background CDNC level. Additionally, the model's sensitivity to idealized scenarios in a simplified, aquaplanet-like configuration with imposed aerosol emissions is dependent on the embedded activation scheme's sensitivity in clean conditions - a regime where most of the schemes perform poorly.

We also show that similar relationships between modeled indirect effect and baseline CDNC levels exist in CMIP5 models which participated in experiments with historical forcings. These results help explain some of the spread between model-derived estimates of aerosol indirect effects, and strongly indicate the need for global climatologies of CDNC which can be used to validate and tune aerosol-cloud interactions in global models.