Satellite remote sensing of polar clouds: a review of cloud detection, phase identification, and microphysical property retrieval (2001–2025)

Polar clouds play important roles in regulating the surface energy budget, atmospheric moisture cycle, precipitation, and sea-ice variability, while their occurrence, vertical structure, thermodynamic phase, and microphysical properties remain difficult to characterize consistently across the Arctic and Antarctic. Satellite remote sensing enables broad-scale monitoring of polar clouds, yet high-albedo snow and ice surfaces, weak cloud–surface thermal contrast, frequent temperature inversions, and persistent mixed-phase conditions make cloud detection and property retrieval substantially more difficult than at lower latitudes. This review synthesizes progress during the quarter-century 2001–2025 in polar cloud satellite remote sensing with a focus on satellite-based polar cloud detection, cloud phase identification, and microphysical retrieval. It first analyzes the synergistic monitoring capabilities of multi-source satellite sensors, including passive multispectral imagers, active radars and lidars, and microwave radiometers. It then explores in depth the evolution of cloud detection methods from static physical thresholds to deep learning algorithms that integrate spatial information and multimodal features. For cloud phase identification and microphysical retrieval, this paper reviews the physical principles and technical limitations of representative methods, including threshold and Lookup Table (LUT) methods, Optimal Estimation Method (OEM), variational and Bayesian retrievals, analytical and empirical methods, and machine-learning approaches, with particular attention to the challenges associated with ice–liquid phase partitioning and vertical heterogeneity in polar mixed-phase clouds. Finally, it identifies the deficiencies in current research regarding the application of far-infrared and high-frequency submillimeter-wave bands, the construction of full-spatiotemporal coverage observation networks, and the physical consistency of data-driven models, and outlines priorities for developing a continuous, multi-dimensional polar cloud monitoring system based on coordinated high- and low-orbit observations, recent and forthcoming satellite instruments, and physically constrained artificial intelligence.