Evaluation of aerosol indirect effect including aerosols-precipitation interaction using a fast and accurate global microphysics model

Atmospheric aerosols perturb the earth's energy balance by modifying cloud microphysical properties through their ability to act as cloud condensation nuclei (CCN), known as the aerosol indirect effect. The large uncertainty in estimates of the aerosol indirect effect is due partly to uncertainties in CCN predictions, which can be improved with appropriate simulation of aerosol number and size. The TwO-Moment Aerosol Sectional (TOMAS) microphysics model incorporated in the Goddard Institute for Space Studies General Circulation Model II' (GISS GCM II') predicts accurately the evolution of aerosol number by aerosol microphysical processes such as condensation and coagulation. This thesis focuses on two broad topic areas: (a) the improvements of GISS-TOMAS model and (b) the study of aerosol indirect effects using the improved GISS-TOMAS model.;With the improved GISS-TOMAS model, the impact of mitigation of black carbon (BC) particulate matter on aerosol indirect forcing is studied. Mitigation of BC has been suggested as a strategy complementary to reduction of greenhouse gases due to its positive top-of-the-atmosphere (TOA) direct radiative forcing. However, BC-containing emissions can contribute to the CCN population when those particles become internally mixed with hydrophilic aerosol components, and a decrease in BC mass result in a reduction in TOA cloud radiative forcing. Our study shows that the reduction in aerosol indirect forcing by the BC mitigation partly offsets the reduction in TOA direct BC radiative forcing, which might indicate less climate benefits by the BC mitigation than what has been suggested.;Finally, we investigated the uncertainty of nucleation on cloud microphysical properties, such as cloud droplet number concentration (CDNC), cloud droplet effective radius (Reff), autoconversion rates, and aerosol indirect forcing using the GISS-TOMAS model. The global difference in nucleation rates (binary vs. ternary) is ∼106, but for CN10 is ∼2 and CDNC is ∼10%. The impact of nucleation on global-average autoconversion rates can be -10% to -60%, depending on the autoconversion parameterization chosen. In this study, the autoconversion rate predictions are only affected by the change in CDNC since other parameters (liquid water content and cloud droplet size distribution) are held constant. Thus, this large variation results from the fact that different autoconversion parameterizations have different power-dependence on CDNC. The choice of nucleation scheme influences the increase of global-average of CDNC from the preindustrial to the present-day (103% for the binary nucleation and 85% for the ternary nucleation) but has little effect on global average cloud albedo forcing (-0.74 W m-2 for binary nucleation and -0.72 W m-2 for ternary nucleation). All model results presented arc for January using GCM climatological meteorology, so this conclusion might be changed when using the annual average model outputs.;A global mineral dust model is developed and evaluated with dust observations. It completes the GISS-TOMAS aerosol model by adding mineral dust, the remaining climatologically important aerosol type. The GISS-TOMAS model is evaluated against aerosol optical depth measurements from AERONET, MODIS, and MISR, which provide well characterized spatial and temporal distributions. For this work, an aerosol optical depth calculation module is developed. A computationally efficient TOMAS global model is developed, permitting multi-year simulations. The computational speed increases 2 to 3 times with minimal increase in error. As a part of model evaluation, we investigated the impact of the representation of nucleation mode particles in global aerosol models on predicted aerosol number concentrations. This study shows that the accuracy of the nucleation mode parameterization depends greatly on the choice of the lower size boundary of the microphysical model; the lowest size boundary at 3nm performs well but a boundary at 10 nm does not, compared to explicit dynamics. However, the CCN mode is largely unchanged by the use of either lower size boundary.