On the role of surface emissivity in polar night-time cloud detection

Clouds play a key role on the radiation budget of the Earth. However, several sources of uncertainties affect the ability to detect cloud presence from passive satellite measurements, especially in the polar regions. In fact, cloud detection in polar regions is complicated by the following aspects:  little thermal and visible contrast between clouds and snow/ice surface  persistence of atmospheric temperature inversion (ambiguous temperature-height relationship)  low optical thickness  presence of liquid, ice, and mixed-phase clouds  cold background temperatures (low signal-to-noise ratios) During polar nights, cloud detection is even more difficult, because the scarcity of sunlight makes information on texture unavailable. Moreover, techniques based on differential reflectance in the visible and near infrared (e.g. using 1.6 m channel) are not usable during winter months. Therefore, an approach relying on infrared (IR) observations only must be used. A variety of threshold methods based on multiand hyper-spectral observations have been proposed recently with encouraging results, although problems still exists with thin clouds and weak inversions. As such, uncertainties related to surface emissivity may play an important role, since spectral emissivity for ice/snow surfaces may differ significantly depending upon microphysical properties. Furthermore, in particular cases the spectral emissivity of ice/snow surfaces may resemble spectral signatures of clouds when observed by satellite, and could potentially confuse detection techniques relying on thresholds. Thus, we first show the effect of realistic sea/ice/snow emissivity spectra on simulated radiances in the IR range often used for cloud detection (700-1200 cm). Then, we discuss the similarities of surface emissivity effects with cloud signatures obtained with a variety of cloud microphysics parameters. Finally, we analyse the impact of surface emissivity uncertainties on the performances of currently available polar night-time cloud detection techniques based on multispectral (MODIS) and hyperspectral (AIRS, IASI) observations.