Date of Award

4-3-2012

Document Type

Dissertation

Degree Name

Doctor of Philosophy (PhD)

Department

Atmospheric Sciences

First Advisor

Xiquan Dong

Abstract

Despite recent advancements in global climate modeling, models produce a large range of climate sensitivities for the Earth. This range of sensitivities results in part from uncertainties in modeling clouds. To understand and to improve cloud parameterizations in Global Climate Models (GCMs), simulations should be evaluated using observations of clouds. Detailed studies can be conducted at Atmospheric Radiation Measurements (ARM) sites which provide adequate observations and forcing for Single Column Model (SCM) studies. Unfortunately, forcing for SCMs is sparse and not available for many locations or times. This study had two main goals: (1) evaluate clouds from the GISS Model E AR5 SCM at the ARM Southern Great Plains site and (2) determine whether reanalysis-based forcing was feasible at this location. To accomplish these goals, multiple model runs were conducted from 1999–2008 using forcing provided by ARM and forcing developed from the North American Regional Reanalysis (NARR). To better understand cloud biases and differences in the forcings, atmospheric states were classified using Self Organizing Maps (SOMs). Although model simulations had many similarities with the observations, there were several noticeable biases. Deep clouds had a negative bias year-round and this was attributed to clouds being too thin during frontal systems and a lack of convection during the spring and summer. These results were consistent regardless of the forcing used. During August, SCM simulations had a positive bias for low clouds. This bias varied with the forcing suggesting that part of the problem was tied to errors in the forcing. NARR forcing had many favorable characteristics when compared to ARM observations and forcing. In particular, temperature and wind information were more accurate than ARM when compared to balloon soundings. During the cool season, NARR forcing produced results similar to ARM with reasonable precipitation and a similar cloud field. Although NARR vertical velocities were weaker than ARM during the convective season, these simulations were able to capture the majority of convective events. The limiting factor for NARR was humidity biases in the upper troposphere during the summer months. Prior to releasing this forcing to the modeling community, this issue must be investigated further.

Download

Share

COinS