Identification of Supercooled Water Clouds Using Combined Microwave and Terahertz Radiometer Observations

Supercooled water clouds (SWCs) exert significant influences on Earth’s radiative balance, weather modification, and aircraft icing hazards. This study proposes a new method for SWC identification by synergistically analyzing spectral information from the microwave (MW) and terahertz (THz) bands. Based on radiative transfer simulations, a random forest model was trained and evaluated using MW and THz brightness temperatures to assess the SWC identification capability of MW-only, THz-only, and combined MW–THz configurations. The results show that the MW–THz configuration achieves the best overall performance, with 90.3 ± 0.3% accuracy, 87.2 ± 0.4% SWC precision, 88.5 ± 0.4% SWC recall, and an F1-score of 87.9 ± 0.3%. THz information effectively reduces the confusion between SWCs and ice clouds, especially for cases with low-to-medium supercooled water path, whereas MW information helps maintain a stable separation between SWCs and warm water clouds. The channel feature analysis indicates that the 664.0 ± 4.2, 89.0, 165.5 ± 0.725, and 243.2 ± 2.5 GHz channels provide the strongest discriminative information for SWC identification. Further experiments using compact channel subsets, mixed-cloud scenarios, and noise sensitivity tests show that combined MW–THz observations can maintain good identification performance and exhibit stronger robustness under complex conditions. These results demonstrate the complementary advantages of MW and THz observations and provide a useful reference for future passive remote sensing identification of SWCs and channel selection.