Consistency of Multiwavelength Infrared Retrievals for Marine Stratocumulus

Accurate prediction of climate change requires models for the behavior of cloud systems. Satellites promise to provide the observations of cloud properties required to gain this modeling skill. The properties retrieved from satellite data, particularly fractional cloud cover, cloud opacity and hydrometeor size, are, however, often difficult to verify using surface and aircraft based observing systems. Here the validity of retrieved products is examined through the analysis of independent information provided by rnultiwavelength satellite observations. A new retrieval method is used to extract regional scale (~60 km) estimates of emission for opaque clouds and an index of cloud particle size along with pixel-scale estimates of fractional cloud cover and liquid/ice water path for single-layered cloud systems. The retrieval scheme is applied to observations of emission at 3.7, 11 and 12 um obtained at night with the Advanced Very High Resolution Radiometer (AVHRR) for the Northeastern Atlantic during ASTEX. The results obtained using the coverage of the 11- 12-um radiance domain by pixel-scale radiances are compared with those based on the coverage of the 3.7-12-um domain. The weaker absorption and shorter wavelength coupled with a stronger sensitivity of the radiance to temperature at 3.7 um allows for retrievals that are more sensitive to shifts in cloud particle size and water/ice pathlengths. Allowing for uncertainties arising from noise in the observations and the approximations employed in the cloud-radiative transfer model used in the retrieval, regional-scale estimates of droplet size and emission for opaque clouds and pixel-scale estimates of fractional cloud cover and water/ice pathlengths retrieved using 11 and 12-um radiances agree with the properties obtained using 3.7 and 12-um radiances.