Aerosol-cloud interaction from an observational and modeling perspective

Abstract: Clouds may respond strongly to changes in the atmospheric aerosol population, and the response of clouds to an increased global aerosol burden could to some extent mask the warming caused by enhanced greenhouse gas concentrations. However, estimates of the impact of aerosols on cloud properties are associated with large uncertainties, both because of difficulties representing the aerosol-cloud interaction within models, and because of problems of unequivocally isolating the effect of aerosols on cloud properties in observational data. This thesis focuses in part on underlying meteorological factors that significantly correlate with both aerosol and cloud properties, and on how sensitive clouds are to small variations in meteorological conditions. It was found that meteorological covariations must be taken into account when estimating the strength of the relationship between aerosols and cloud properties. By studying the response of shallow convective clouds to perturbations in meteorological conditions and aerosol concentration, it was further concluded that variations in meteorological conditions can enhance or mask the relationship between aerosols and cloud properties, making it difficult to isolate the aerosol signature from small meteorological differences. Additionally, the impact of deep convective clouds on the redistribution of aerosols within a cloud life cycle is examined. It was found that mid-tropospheric aerosols can have a substantial source in evaporating cloud droplets within deep convection. Lastly, this thesis focuses on the implications of meteorological analysis uncertainties, in part related to the difficulties of constraining meteorological variability in observational data of clouds and aerosols, but mainly the impact of analysis errors on atmospheric trajectory calculations. A method is presented to consistently estimate the uncertainty in trajectory calculations. It was concluded that the spatial and temporal trajectory error can be substantially underestimated if the analysis error is not taken into account.