Multi-angular radiances of an isolated convective cloud: comparison between MISR measurement and Monte-Carlo simulation

Convective clouds are very important in the study of the climate system because they generally have large liquid water content. As for most cloud retrieval algorithm, their properties are obtained following the plane parallel assumption. It is, however, well known that this assumption induces large errors, which are due in part to cloud optical thickness variabilities but also to cloud geometry which tends to reduce the radiation reflected by the cloud top, the photons being lost by the cloud sides. This leads to an underestimation of the cloud optical thickness and total liquid water content. Moreover, this underestimation can be intensified by the fact that radiances tend to be saturated in case of deep convective cloud.

In order to study these 3D effects, we used the cloud geometry reconstruction obtained by Seiz and Davies (2005) from stereo retrieval using MISR data. The clear sky reflectances including ocean surface reflectivity and aerosol diffusion were set to match the MISR measurement in the nine directions. Once, we had the cloud envelop and the surface reflectance, we tried different hypothesis about the cloud composition and microphysics to simulate, with a new Monte-Carlo method, the radiances coming along the cloud contour. After a short description of the improvements added in this new Monte-Carlo method, the effects of the different assumptions made for the cloud properties will be analyzed and the simulated radiances will be compared with the MISR observations in the nine directions.