A comparison between four different retrieval methods for ice cloud properties using data from the CloudSat, CALIPSO, and MODIS satellites

Ice clouds are of great importance to the global water cycle and the earth's radiation budget. A better representation of these clouds in current weather prediction and climate models is only possible with greater observational resources and an improved understanding of ice cloud microphysics. The A-Train constellation of satellites provides an array of different ice cloud measurements together with the new capability to obtain vertical profiles of clouds from CloudSat and CALIPSO, leading to more detailed measurements of ice cloud microphysics.

A new ice cloud product, VarCloud, combines the CloudSat radar and CALIPSO lidar data to retrieve vertical profiles of ice cloud properties. Using optimal estimation theory, VarCloud takes advantage of the complementary nature of the two instruments, providing a smooth retrieval of ice water content and extinction coefficient for a more complete ice cloud profile. For instance, the radar and lidar combined estimate of global ice cloud occurrence in the troposphere is 15.1%, whilst for the radar and lidar individually it is only 10.0% and 9.9% respectively.

In this presentation, the VarCloud retrieval is compared with the CloudSat ice-only retrieval of ice water content, an empirical formula which derives ice water content from radar reflectivity and temperature, and the MODIS retrievals of ice cloud optical properties. The results are interpreted in terms of their dependence on factors such as the radar scattering model and the ice particle assumptions used in the retrieval.

The comparison of VarCloud retrieved ice water content with the CloudSat ice-only retrieval indicates a sensitivity to ice particle shape assumptions in the non-Rayleigh scattering regime. The highest ice water content values are retrieved using an empirical formula derived from aircraft measurements of reflectivity and temperature, followed by an adjusted VarCloud-Mie product, which both assume spherical particle shapes for non-Rayleigh scattering. A clear reduction in large ice water content values occurs for the standard VarCloud product, which models particles as oblate spheroids, whilst the CloudSat ice-only product achieves a reduction in retrieved ice water content through the application of a Mie correction factor to large reflectivity values.

The comparison of VarCloud retrievals of ice cloud optical depth and mean effective radius with MODIS retrievals indicates consistent factors of 0.5 and 2.0 difference between the two properties respectively, which leads to a cancellation of these factors in the calculation of ice water path. An adjusted VarCloud version using an ice particle model similar to that used by MODIS appears to provide a better match between the two products, which indicates the importance of ice particle assumptions to the retrieval of cloud optical properties.

The new combined radar-and-lidar product VarCloud compares well for the bulk of ice water content retrievals from the CloudSat ice-only product. Differences at high ice water content retrievals, as well as various differences with the MODIS retrievals highlight the importance of ice particle models and their impact on retrieved variables. A better understanding and further comparison of ice cloud retrieval methods and their assumptions is necessary before these observations are used to evaluate numerical weather prediction models.