Cloud climatology of the SHEBA year derived from an automated Arctic cloud mask

Detecting clouds with satellite data over the Arctic is difficult due to the minimal contrast between clouds and the underlying snow surface in visible wavelengths. Polar clouds are frequently warmer or at the same brightness temperature as the background surface, complicating cloud detection. The frequent occurrence of some obscuring layer such as haze, diamond dust, or thin ground fog in the Arctic often makes it difficult to validate satellite cloud amounts. For this study, an automated Arctic cloud mask is used to discriminate clouds from the background snow surface over the Surface Heat Budget of the Arctic Ocean (SHEBA) ship site. The cloud climatology runs from November 1997 through September 1998. Input to the cloud mask includes NOAA-12 and 14 AVHRR 0.65um, 3.7um, 11um, and 12um data. The daytime cloud detection algorithm incorporates theoretical snow 3.7um reflectance models while using the framework of the Clouds and the Earth's Radiant Energy System (CERES) polar mask developed for MODIS. Cloud detection for nighttime scenes is determined strictly from a brightness temperature threshold approach. Cloud amounts from the automated cloud mask are validated by comparing them with surface observer, cloud radar, and lidar cloud amounts within a 25-km radius surrounding the SHEBA ship. Broadband longwave and shortwave fluxes are derived for clear sky and total sky scenes. Monthly-mean cloud cover, height, and cloud radiative forcing statistics are computed in the vicinity of the SHEBA ship and over a larger domain in the western Arctic Ocean. Additionally, preliminary TOA broadband fluxes are derived over the ARM North Slope of Alaska (NSA) site for March-July 2001 using matched CERES broadband and MODIS narrowband data.