Hybrid Physical Segmentation and Machine Learning Approach for Deep Convective Cloud Detection With Himawari-8 (opens in new tab)

Deep convective clouds (DCCs) are closely associated with extreme weather, including heavy precipitation and severe thunderstorms. Their rapid evolution requires continuous monitoring, underscoring the importance of geostationary satellites. Conventional identification methods based on infrared brightness temperature (BT) thresholds and visible albedo are often degraded by surrounding thick anvil (TA) clouds and perform poorly at night. Although data-driven machine learning (ML) approaches of...