A multispectral method for retrieving overlapping cloud top heights from passive radiometers

Cloud top height (CTH) is an important factor influencing its radiative effects, weather and climate change, and spaceborne passive radiometers widely used to retrieve CTH mostly use an oversimplified single-layer cloud assumption. Nevertheless, natural clouds exhibit complex vertical inhomogeneity, such as overlapping clouds. This study investigates the potential of multispectral observations for characterizing overlapping cloud properties and introduces a new CTH retrieval algorithm for overlapping CTHs. Synthetic radiative transfer simulations and an information content analysis are applied to demonstrate the theoretical basis of using passive radiometers for overlapping cloud property retrievals. Specifically, this method leverages the merits of four shortwave infrared (0.87, 1.6, 2.13 and 2.25 μm) channels in distinguishing cloud optical and microphysical properties in different phases and the capabilities of longwave infrared (8.6, 11 and 12 μm) channels for the corresponding CTHs. The method performs effectively for overlapping clouds with an optically thin but detectable ice layer (optical thickness less than ∼7) above a liquid water layer. We consider collocated Advanced Himawari Imager (AHI)/Visible Infrared Imaging Radiometer Suite (VIIRS) and Moderate Resolution Imaging Spectroradiometer (MODIS) measurements to provide sufficient spectral information. Compared with conventional single-layer-assumed CTH retrievals, our CTHs for both upper-layer ice and lower-layer water clouds show closer agreement with those indicated by active instruments. Statistically, the average bias of the upper ice CTH results is constrained from −2.89 km to −1.36 km and we obtain lower water CTHs with an average bias of −1.21 km. The cloud properties based on our retrievals also make the simulated multichannel radiances more consistent with the observed ones. The new method will greatly improve our understanding of the vertical structures of overlapping clouds and their parameterization in general circulation models; thus, we suggest that including increased numbers of cloud-sensitive channels is necessary to better determine cloud properties.