Satellite remote sensing of cloud optical thickness involves matching observed reflectivities at a single angle, usually at nadir, to a model. This single-view, radiance-based approach has several deficiencies, most of which are a direct consequence of the underlying assumption of plane-parallel homogeneous layers. The non-linear relationship between reflectivity and optical thickness results in a plane-parallel albedo bias when area averages are calculated. Saturation of reflectivity with optical thickness renders the retrieval of large optical thicknesses impossible. The marked difference between observations and model computations of reflectance as a function of solar zenith angle yields a bias in retrievals which gets worse with solar zenith angle. With a single view, it is impossible to check the validity of the plane-parallel model.

Multiangle measurements obtained by the MISR instrument onboard the Terra satellite may help improve this situation. This study investigates the information content of multiangle observations, using a set of marine boundary layer cloud scenes as an example. Comparing the angular distribution of measured reflectivities to that of model predictions reveals how well and how frequently the plane-parallel model applies to these clouds. The degree of consistency among the angular optical thicknesses enables the accuracy of the retrievals to be estimated, which is impossible with the single view approach. First results indicate that, for the marine stratocumulus example, the observed angular distribution of radiances matches the plane-parallel model better at coarse than at fine spatial resolution, despite the presence of significant spatial heterogeneity. A comparison of the angular consistency test with a spatial homogeneity test suggests that the latter is a strickter criterion for the applicability of the plane-parallel model.