Sensitivity of TOA irradiances derived from CERES observations over the Arctic to the error in the cloud cover

Identifying clouds and estimating cloud cover over snow and sea ice from radiances measured by passive sensor such as MODIS is challenging. Top-of-atmosphere (TOA) shortwave irradiances derived from CERES radiance measurements depend on cloud and snow/sea ice fractions over a CERES footprint because the anisotropic factor to convert the radiance to irradiance depends on these fractions. In this study, cloud and snow/sea ice fractions are perturbed. Anisotropic factors (angular distribution models) are computed as a function of cloud and snow/sea ice fractions, as well as viewing and solar zenith angles. These anisotropic factors are then applied to CERES radiances to estimate TOA shortwave irradiances to understand the TOA irradiance sensitivity to scene ID error. In estimating the TOA irradiances, cloud and snow/sea ice fractions are also perturbed as the same amount as those perturbed in building angular distribution models. This process simulates the entire process of angular distribution model, from building them to deriving TOA irradiances, used for polar regions in the CERES project. A preliminary result indicates that perturbing the cloud fraction by 0.05 changes the TOA shortwave irradiance as large as 10 Wm-2 over some 1 degree by 1 degree regions averaged over a month (June). The effect on the estimating a TOA albedo trend over the Arctic due to the error caused by the cloud fraction error will also be discussed.