Angular dependencies of GOES-derived cloud properties over the continental United States

Satellite retrievals can provide valuable information about cloud and radiative properties over the entire Earth. Cloud properties retrieved from satellite imager data are used for validating climate models and are being assimilated into numerical weather prediction models. Thus, it is becoming more important to fully quantify the uncertainties in the retrieved parameters. A potentially large source of error is the effect of viewing and illumination geometry on the retrievals. The orbital positions, similarity in instrument suites, and nearly-simultaneous observations from the GOES-11 and GOES-12 geostationary satellites provide a unique opportunity to assess angle-dependent errors. This study will quantify biases in retrieved cloud properties due to viewing geometry as a function of solar zenith angle.

Clouds observed simultaneously from differing sun-target geometries can have differences in retrieved cloud properties for a variety of reasons. The vertical structure of cloud microphysics can cause viewing angle dependencies in cloud particle size retrievals because the partially absorbing channels are only sensitive to radiation from a finite path length downward from cloud top. Microphysical heterogeneity within in the cloud or 3D structure on the cloud tops can also affect the retrieval. These effects become more pronounced at high viewing or solar zenith angles. Sub-pixel cloud variability leads to biases in cloud amount when observed from large viewing angles. Because retrievals are performed using fixed ice crystal reflectance models and actual ice crystal habits can vary widely among clouds, errors can be large at particular scattering angles.

To evaluate these issues, data from GOES-11 and GOES-12 over the continental United States will be employed. This region was selected to obtain the maximum spatial and temporal coincidence of the GOES-11 and GOES-12 observations, and to include a wide range of viewing angles. Cloud properties will be obtained using a slightly revised version of the Visible Infrared Solar-Infrared Split-window Technique (VISST), which is an iterative model-matching plane-parallel technique that matches observations to theoretically calculated radiances to retrieve cloud properties. To minimize the effects of cloud heterogeneity, only single-layer overcast clouds within this domain will be considered. Additionally, ice cloud properties will be derived using both a smooth and rough ice crystal reflectance model to determine if the latter reduces the angle-dependent retrieval differences. If available, in situ data from the 2009-2010 ARM Small Particles in Cirrus Experiment will be used to further examine the ice cloud retrievals.