Dim, JR, Takamura, T, Okada, I, Nakajima, TY, Takenaka, H (2007). Influence of inhomogeneous cloud fields on optical properties retrieved from satellite observations. JOURNAL OF GEOPHYSICAL RESEARCH-ATMOSPHERES, 112(D13), D13202.
 Analyses of solar radiation exchanges between the atmosphere and clouds are vital for the understanding of climate processes and cycles. Comparisons of satellite-to-satellite or satellite-to-ground-truth observations aiming at, elucidating the radiative behavior of atmospheric components (clouds, aerosols, gas, etc.), or validating data of a particular satellite are a common practice in global radiation investigations. In order to assess the quality of cloud optical properties derived from Geostationary Meteorological Satellite-5/Stretched Visible Infrared Spin Scan Radiometer (GMS-5/SVISSR), the former procedure (satellite-to-satellite comparison) was used. Data derived from GMS-5/SVISSR satellite were compared with those from the polar-orbiting Terra-Moderate Resolution Imaging Spectroradiometer (Terra-MODIS) satellite. This comparison showed serious discrepancies between cloud optical depth (COD) data retrieved from the two satellites' observations. GMS-5/SVISSR-retrieved COD appeared mostly lower than that of Terra-MODIS. To understand the origin of such differences, an identification procedure of the major factors likely to affect these data is conducted. Some of these factors were the satellite viewing and solar conditions, the cloud thermodynamic phase differentiation and particle effective radius, and the cloud inhomogeneity. Then emphasis was put on the examination of the latter effect (i.e., the cloud inhomogeneity). The analysis procedure was as follows: First, data having close-viewing geometries between both satellites were selected and used to understand the effects of the remaining factors. Among these, the cloud thermodynamic phase appeared to play the major role as analyses showed that most of the COD differences between both satellites were confined within ice clouds while warm clouds had the least discrepancies. This would suggest that the choice of a water cloud particle radiative transfer model to analyze a 2-phase cloud radiation data, as used here, may produce large uncertainties in ice COD retrievals from at least one of the satellites. To avoid the cloud phase problem, a more restrictive data set comprising only water clouds (besides close-viewing geometries between both satellites) was selected, and the impact of the degree of cloud inhomogeneity on the COD retrievals was evaluated. The study reveals that the 3-D radiative effects deriving from the external cloud inhomogeneity, i.e., cloud asymmetry and structured sides, were the most influencing properties here. The GMS-5/SVISSR interpretation of inhomogeneous cloud optical properties showed larger uncertainties than that of Terra-MODIS. Furthermore, COD values of GMS-5/SVISSR were systematically lower than those of Terra-MODIS for the pixels at shadow sides of the cloud, while at illuminated sides they often showed higher values. For gentle or near-plane-parallel cloud surfaces, fewer discrepancies were noticed ( the best agreement between both satellites' retrievals). At steep slopes of the shadow and illuminated cloud sides, GMS-5/SVISSR average COD data were respectively under- and overestimated compared to those of Terra-MODIS. COD differences between the two satellites could be sometimes higher than 30% for slopes steeper than 0.5 K/km.