Observational and modeling studies of collision-coalescence in marine stratocumulus

Low clouds cover over a quarter of the planet's surface in the annual mean and exert a significant cooling effect on global climate. Despite their importance, the representation of such clouds in global circulation models remains a major source of uncertainty in projections of future climate. The lifetime of such clouds can be modified by precipitation, which is observed to occur 20--40% of the time in stratocumulus over the ocean. The processes by which precipitation forms are also highly uncertain and accurate quantitative precipitation forecasting remains a grand challenge in meteorology. The goal of this research is to explore and evaluate the representation of clouds in observations and models of the atmosphere, with a focus on improving our understanding of the primary process responsible for liquid precipitation formation, collision-coalescence.;The first goal of this research is to develop a metric for the dynamical "age" of small cumulus for use in contexting in situ observations. The lifetime of shallow cumulus is typically under an hour hence the metric must be accurate for a single measurement of the cloud. It is found that the deviation of cloud total water mixing ratio from the mean surface mixed layer total water mixing ratio is an effective measure because the mechanism by which total water mixing ratio is diluted, entrainment, is irreversible. Collision-coalescence is notoriously difficult to observe in situ and its implementation in models of the atmosphere is highly uncertain because it is an inherently local process, dependent on interactions of drops on length scales as small as micrometers. The second goal of this dissertation is to assess the ability of theoretically derived collision-coalescence rates to explain observations of cloud drops in marine stratocumulus off the coast of Monterey, California. Drop size spectra averaged over length scales of 1.5 and 30 kilometers are found to require enhancements of collision rates that vary as a function of the mode of the drop size distribution from 0.1 for the largest mode sizes to over 20 for the smallest sizes. Unresolvable small-scale variability in the DSD is likely a major factor in determining the spread in required enhancements.;The third goal of the research presented herein is to implement a high spectral resolution bin microphysics scheme in a large eddy simulation (LES) model. The high resolution scheme is validated by comparison with simulations using the default spectral resolution. Considerable differences in microphysical structure and planetary boundary layer turbulence are found when collision coalescence is active due to artificial acceleration of the process at the default resolution, which causes higher concentrations at large drop sizes than in the high resolution configuration.;Finally, the fourth goal of this research is to evaluate the ability of a turbulent collision kernel to improve the representation of precipitation formation. The net effect of the turbulent kernel is to deplete the cloud top of large drops, but the effect of this varies depending on total drop concentration such that precipitating and non-precipitating cases have opposite responses. Precipitating cases experience an enhanced sedimentation flux throughout the boundary layer at default resolution, while nonprecipitating cases and high spectral resolution simulations experience further suppression of precipitation. We conclude that the effects of changing spectral resolution dominate the effects of the turbulent kernel. This finding is contingent on the assumption that microphysical boundary conditions (i.e. cloud condensation nuclei concentrations) should be constant across spectral resolution configurations.