A Physical Basis for the Overstatement of Low Clouds at Night by Conventional Satellite Infrared-Based Imaging Radiometer Bi-Spectral Techniques

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 mu m 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. Plain Language Summary Marine boundary layer (MBL) clouds play a governing role in Earth's radiation balance. Satellite remote sensing provides the most practical means to collecting information about the frequency, distribution, and properties of MBL clouds, including their evolution across the day and night. Recent comparisons of operational cloud masks to novel nighttime moonlight reflectance observations from the Day/Night Band low-light sensor has revealed instances where too many MBL clouds are being analyzed (i.e., false alarms) in the operational algorithms. This study explores the underlying physical basis for these issues, and shows how a special combination of atmosphere/surface conditions can produce a clear-sky infrared (IR) spectral signature that is used in these algorithms as a test for the presence of low clouds. The false alarm mechanism is confirmed via radiative transfer simulations in both idealized and realistic environment scenarios. The implications of these findings are that localized to regional-scale biases may exist in nocturnal MBL cloud descriptions, translating to possible biases in downstream environmental parameters reliant on clear-sky scenes, such as sea surface temperature climate records. Increasing the frequency of nighttime visible observations and refining IR-based algorithms in areas prone to false alarm could help to mitigate these issues.

Automated summary

Marine boundary layer (MBL) clouds are important for Earth's radiation balance, and satellite remote sensing is the most practical method for collecting information about these clouds. Recent comparisons have shown that there are instances where too many MBL clouds are being analyzed as false alarms in operational algorithms. This study explores the physical basis for these issues and confirms the false alarm mechanism through radiative transfer simulations. The findings suggest that localized to regional-scale biases may exist in nocturnal MBL cloud descriptions, potentially impacting downstream environmental parameters such as sea surface temperature climate records. Increasing the frequency of nighttime visible observations and refining infrared-based algorithms in areas prone to false alarm could help mitigate these issues.