Quantitative Evaluation of Three-Dimensional Cloud Radiative Transfer Effect and Morphology Using A-Train Data

This study evaluates 3D cloud radiative transfer effects quantitatively using a 3D radiative transfer model, MCstar, and A-train collocated cloud data. The 3D extinction coefficients are constructed by a newly devised Minimum cloud Information Deviation Profiling Method (MIDPM) (Okata et al., 2017) that extrapolates the scaled extinction coefficient profiles derived from CPR radar profiles and MODIS cloud optical thickness at nadir into off-nadir regions within MODIS swath. The method is applied to water clouds, for which the 3D radiative transfer (RT) simulations are performed.

The radiative fluxes thus simulated are compared to those obtained from CERES as a way to validate the MIDPM-constructed clouds and our 3D radiative transfer simulations for 89 cloud cases. The 3D cloud filed is 20km×20km×60km with 1km horizontal resolution and 240m vertical resolution based on CERES field of view size. The results show that the simulated shortwave fluxes agree with CERES values within 5~50 Wm^-2 One of the large biases arises from the 1D assumption for cloud property retrievals particularly for thin clouds, which tend to be affected by spatial heterogeneity leading to overestimate of the cloud optical thickness.

These 3D-RT simulations also serve to address another objective of this study, i.e. to characterize the “observed” specific 3D-RT effects by the cloud morphology. We extend the cloud field to 100km×100km×60km from instantaneous 1km horizontal and 240m vertical resolution for this purpose. The cloud morphology obtained from the observation is very complicated, and it is particularly sensitive to the solar insolation geometry. In this study, the 3D-RT effects are characterized by errors of 1D flux approximations to those of 3D cloud field. We then investigated the error dependence on the solar zenith angle (SZA) for several 1D idealized cloud fields. This method utilizes the fact that the SZA dependence is characteristically dependent on the 3D radiative transfer effect of the broken cloud system, and we define two indices from the error tendency. According to the indices, the 3D-RT effects are classified into three types which cannot be estimated by the 1D-RT corresponding to different simple three morphology types, i.e. isolated cloud type, upper cloud-roughened type and lower cloud-roughened type. The dominant 3D-RT effect for each cloud field is the light obscure effect in clear sky region, the light blocking effect by high clouds, and the light guide effect through clouds. We performed these simulations for the satellite-constructed real cloud cases, and quantitatively classified 32 cases into the characteristic 3D cloud RT effects and morphologies.

For visualization of the 3D-RT effects, we also made RGB composite maps from MODIS/Aqua three channels, which show cloud optical thickness and cloud height information. Such a color mapping offers a novel insight into the 3D-RT effects in a manner that relates to cloud morphology.