AAAR 32nd Annual Conference
September 30 - October 4, 2013
Oregon Convention Center
Portland, Oregon, USA
Abstract View
Evaluating Aerosols, Clouds, and Their Interactions in Three Global Climate Models Using COSP and Satellite Measurements
GEORGE BAN-WEISS, Susanne Bauer, Ralf Bennartz, Xiaohong Liu, Kai Zhang, Yi Ming, Ling Jin, Jonathan Jiang, University of Southern California
Abstract Number: 487 Working Group: Aerosols, Clouds, and Climate
Abstract Accurately representing the interaction of atmospheric aerosols and clouds is a great challenge in global climate modeling. This is in part because many aerosol and cloud physical processes operate on spatial scales that are much smaller than climate model grid cells. Thus, these processes are represented in climate models using parameterizations, making estimates of aerosol indirect forcing uncertain. To reduce uncertainty and improve parameterizations, signatures of aerosol indirect effects from climate models can be compared to those measured by satellites. These comparisons require special attention since aerosol and cloud measurements from satellites are inherently different than standard climatological output from climate models. In this work we compare signatures of how aerosols affect subtropical stratocumulus clouds in three climate models (NASA ModelE, NCAR CAM5, and GFDL AM3) to MODIS observations. We focus on near-coast marine areas near South Africa, South America, and Eastern Asia. These areas have persistent stratocumulus clouds and are subject to aerosol pollution from near-by land regions. Simulations are carried out for three years using prescribed SSTs. To maximize comparability to MODIS observations, cloud properties in climate models are diagnosed using the CFMIP Observations Simulator Package (COSP) and satellite overpass times are extracted. Aerosol-cloud interactions are analyzed using high frequency (3-hourly) climate model data. An algorithm is applied to determine cloud droplet number concentration (CDNC) from satellite observed and GCM simulated cloud optical depth and droplet effective radius. We find that the AM3 model matches well the observed CDNC in terms of spatial patterns, magnitude of the seasonal cycle, and sensitivity to aerosol loading. CAM5 overestimates the mean, seasonal cycle, and sensitivity to aerosol, while ModelE underestimates CDNC and cloud prevalence in the stratocumulus areas in general.