Retrieval algorithms are based on plane parallel radiative transfer, therefore the heterogeneity typically present in cloud fields introduces biases in these retrievals. These could be severe when Global Area Coverage (GAC) data are used to extract cloud properties because the horizontal resolution of the pixel is 1 x 4 km at nadir and the scene could be partly cloudy. We have examined histograms of optical depth and effective radius retrieved from low level clouds in the North Atlantic using two screening criteria - a simple threshold and the spatial coherence method. Not surprisingly, many partially filled pixels yield effectively low cloud optical depths. What is more intriguing is that there is no preferred bias in the retrieved effective radius.
In addition to the optical depth and effective radius, an attempt has also been made to estimate the liquid water path, geometrical thickness and droplet concentration of the clouds using ancillary data from a meteorological analysis. Since AVHRR GAC and the NCEP/NCAR reanalysis are available globally for several years, it is proposed that it may be possible to identify anthropogenic signatures in cloud microphysical properties on a global scale.