Quantification of Monthly Mean Regional-Scale Albedo of Marine Stratiform Clouds in Satellite Observations and GCMs

Abstract

lanetary albedo?the reflectivity for solar radiation?is of singular importance in determining the amount of solar energy taken in by the Earth?atmosphere system. Modeling albedo, and specifically cloud albedo, correctly is crucial for realistic climate simulations. A method is presented herein by which regional cloud albedo can be quantified from the relation between total albedo and cloud fraction, which in observations is found to be approximately linear on a monthly mean scale. This analysis is based primarily on the combination of cloud fraction data from the Moderate Resolution Imaging Spectroradiometer (MODIS) and albedo data from the Clouds and the Earth?s Radiant Energy System (CERES), but the results presented are also supported by the combination of cloud fraction and proxy albedo data from satelliteborne lidar [Cloud?Aerosol Lidar and Infrared Pathfinder Satellite Observations (CALIPSO)]. These data are measured and derived completely independently from the CERES?MODIS data. Applied to low-level marine stratiform clouds in three regions (off the coasts of South America, Africa, and North America), the analysis reveals regionally uniform monthly mean cloud albedos, indicating that the variation in cloud shortwave radiative properties is small on this scale. A coherent picture of low ?effective? cloud albedo emerges, in the range from 0.35 to 0.42, on the basis of data from CERES and MODIS. In its simplicity, the method presented appears to be useful as a diagnostic tool and as a constraint on climate models. To demonstrate this, the same method is applied to cloud fraction and albedo output from several current-generation climate models [from the Coupled Model Intercomparison Project, phase 3 (CMIP3), archive]. Although the multimodel mean cloud albedo estimates agree to within 20% with the satellite-based estimates for the three focus regions, model-based estimates of cloud albedo are found to display much larger variability than do the observations, within individual models as well as between models.