Sensitivity and Identification of Cloud Phases in the Solar Spectrum Using Dual-View Satellite Observations

The cloud phase is expected to change in the warming world. Knowledge of the cloud phase is an initial and essential step in retrieving cloud optical parameters. Cloud optical parameters are generally retrieved by single-viewing remote sensing spectrometers. Identifying mixed-phase clouds (MPC), pure ice and liquid clouds simultaneously for cloud retrievals remains a research challenge. This study addresses this issue and uses the SCIATRAN radiative transfer model (RTM) to understand the sensitivities of the cloud phases and their varying optical and microphysical properties in the solar radiation spectrum in nadir and near-nadir viewing geometries. We use the dual-view measurements of the Surface Land and Sea Temperature Radiometer (SLSTR) to determine cloud phases. For the MPC, we introduce the parameter Ice Fraction (IF), defined as the fraction of the total extinction of solar radiation attributed to ice. To classify the cloud phases,  we use two indices: a) the  NIR  ratio of  1.64 µm to 2.25 µm backscattered intensities at the top of the atmosphere, which is sensitive to spectral absorption in the cloud,  and b) the dual-view ratio, using nadir and near-nadir intensities at 0.87 µm, which exploits the angular scattering variation of clouds. These indices form the NIR-dual view ratio (multiplication of the NIR and the dual-view ratio), enhance discrimination across the cloud phases, and especially allow MPC identification over ocean and snow surfaces. This ratio typically ranges from less than 2.75 for ice clouds and more than 3.50 for water clouds. The values in between attribute to the presence of MPC, except for MPC, having an ice contribution of > 80%. The NIR-dual view ratio is similar to water clouds for lower IF (e.g., IF < 20%). To test theoretical RTM results, we validated the NIR-dual view ratio calculated from SLSTR onboard Sentinel-3A and comparing with the CloudSat product  (2B-CLDCLASS-LIDAR).  Results demonstrate that our approach identifies all three cloud phases with more than 80% accuracy. This research highlights the potential of dual-view satellite observations to improve cloud phase classification, advancing the capabilities of cloud retrieval algorithms.