摘要：Abstract Marine boundary layer (MBL) clouds, ubiquitous to the world’s oceans, help govern the radiative balance of Earth’s climate system. Satellite remote sensing provides the most practical means to monitor cloud worldwide. Whereas visible‐based detection of MBL clouds from environmental satellites is relatively straightforward during the daytime, the night presents challenges. In certain conditions, the conventional infrared (IR) methods used for nocturnal cloud detection, such as the commonly used 11–3.9 μm brightness temperature difference, offer poor thermal and spectral contrast between clouds and the clear‐sky background—resulting in missed clouds. A less explored question is to what extent these IR techniques overstate the cloud field. The Day/Night Band (DNB), a novel low‐light sensor carried on the Joint Polar Satellite System constellation, is helping to address this question. By way of its daytime‐analog moonlight reflectance cloud detection, the DNB reveals situations where IR‐based techniques yield false‐alarm low clouds. These problems occur in conditions of cool surface and a warm/moist lower atmosphere, prevalent in areas of coastal upwelling, tidal mixing, river estuaries, and along oceanic fronts and cyclonic eddies. We show how differential sensitivity to MBL moisture can trigger algorithmic thresholds for cloud detection. Presented here are examples illustrating the problem and an evaluation of our hypothesis against idealized radiative transfer simulations. We consider the implications of such artifacts, including climate data records of sea surface temperature which rely upon IR‐based nocturnal cloud masks. The results provide a framework for designing algorithmic improvements to nighttime MBL cloud detection.
