Sensitivity of Near-Infrared Bands to Cloud Phase: An Assessment Using Dual-View Satellite Measurements

Abstract

Accurate cloud phase classification in the near-infrared is challenging due to the overlapping radiative properties of water, ice, and mixed-phase clouds. This study presents a new composite Phase Classification Index (PCINIR,DV) for near-infrared satellite measurements in a dual-viewing geometry. The index is defined as the product of two physically derived components: (1) a spectral ratio of top-of-atmosphere radiances at 1.61 μm and 2.25 μm, which exploits the differences in absorption between water and ice, and (2) a directional ratio of 0.87 μm radiances from oblique and nadir views, which are influenced by scattering. Theoretical simulations using the SCIATRAN radiative transfer model demonstrate that the PCINIR,DV effectively distinguishes between pure water and ice clouds, enabling mixed-phase clouds to be identified. Sensitivities are analyzed for ranges of particle sizes, ice fractions, and surface types. Theoretical results show that water clouds, excluding thin clouds over snow surfaces, exhibit high PCINIR,DV values (above 3.5), ice clouds yield low values (below 2.75), and intermediate values correspond to mixed-phase clouds. Validation of PCINIR,DV derived from the Sea and Land Surface Temperature Radiometer (SLSTR) dual-view observations (onboard Sentinel-3A) against CloudSat-CALIPSO phase classifications confirms its applicability, yielding 86% classification accuracy, including over 63% for mixed-phase clouds. The results demonstrate that PCINIR,DV provides a robust physical framework for dual-view satellite missions, which aim to measure the cloud phase.