PhD Research

Impacts of Anthropogenic Aerosol and Greenhouse Gas Emissions on Clouds, Convection, and Precipitation as Simulated by a Super-parameterized Global Climate Model

Scripps Institution of Oceanography, University of California, San Diego

Dissertation Abstract

    Clouds strongly influence Earth's climate by reflecting and absorbing radiation, transporting latent heat, and generating precipitation. Changes in cloud properties in response to anthropogenic greenhouse gas and aerosol particle emissions can both dampen and amplify climate trends. Conventional global climate models (GCMs) poorly represent the multi-scale nature of these processes, which range from micrometer-scale droplet nucleation to large-scale convective systems, and thus contribute significant uncertainty to future projections. A new approach called super-parameterization replaces conventional statistical parameterizations with embedded cloud-resolving models that explicitly simulate sub-grid convection. A super-parameterized version (SPCAM) of the standard Community Atmosphere Model (CAM) is shown to improve the variability and intensity of simulated convection, and representation of aerosol-cloud interactions.

    Natural modes of variability influence aerosol and cloud distributions such that isolating statistically significant aerosol indirect effects requires long simulations. SPCAM improves aerosol-cloud relationships compared to conventional GCMs, but the added computational cost of resolving convection makes long integrations prohibitively expensive. An alternative Newtonian relaxation approach applied here uses nudging to constrain simulations with pre-industrial and present-day aerosol emissions toward identical meteorology. This reduces differences in natural variability and dampens feedbacks to isolate aerosol indirect forcing estimates from short simulations. Nudging facilitates a meaningful evaluation of one-year SPCAM simulations, which produce a substantially weaker indirect effect (-0.81 Wm-2) than CAM (-1.19 Wm-2).

    Most GCMs do not realistically represent the physical mechanisms that generate convection in the Central US during summer, and models disagree on the sign of future precipitation trends. A realistic convection signal in a climate change-capable GCM has recently been documented in SPCAM. A new empirical orthogonal function-based index developed here efficiently demonstrates that nocturnal, eastward propagating mesoscale convective systems are a robust effect of super-parameterization. The signal is sensitive to aspects of model implementation and is most realistic in the latest version. Employing a time-slice climate change experiment design, Central US convection is further shown to be sensitive to higher CO2 concentrations, which increase the intensity of precipitation generated by propagating storms. Changes in these storms are one manifestation of the general shift toward more extreme rainfall captured in SPCAM, but not in CAM.