In recent years a large number of intensive field campaigns have been conducted to measure the size, chemical composition, and cloud condensation nuclei (CCN) activity of aerosol around the globe. Typically covering days to weeks, these campaigns provide a comprehensive snapshot of the atmospheric state in the given location and provide the basis for evaluating current theory and models. Köhler theory is the current state-of-the-art method for predicting CCN number concentrations from knowledge of the aerosol size distribution, chemical composition, and water vapor supersaturation. However, for computational efficiency, large-scale models must necessarily make approximations that introduce error into their CCN predictions (e.g., simplified chemistry and mixing state assumptions). To evaluate the impacts of these simplifying assumptions, many studies of ambient CCN attempt to close measured concentrations with those predicted from theory. Uncertainties usually tend toward over-prediction, although the magnitude of error can vary widely across locations and times. Ultimately, it is the effect of aerosols on clouds (i.e., cloud droplet number concentration, CDNC) that is most relevant for determining climate forcing. Yet, limited attention to date has focused on extrapolating these CCN closure studies to evaluate their impact on the overall CDNC uncertainty, and hence, climate change.

We present results from a combined experimental and modeling study incorporating field data from diverse environments in North America including the Alaskan Arctic, rural and agricultural areas, and urban centers in the southeastern and southwestern United States. The coupled adjoints of a CDNC parameterization and the GEOS-Chem chemical transport model were used to determine the sensitivity of the modeled CDNC to CCN over the North American domain. Using adjoint models is ideal for this study, as they provide the sensitivity of CDNC to CCN concentration without otherwise perturbing the model base state. The measured CCN uncertainties uncovered in each region and the modeled sensitivities are then combined to determine the overall uncertainty in CDNC. Uncertainty and sensitivity maps displaying the susceptibility of CDNC to changes in CCN concentrations will be presented.