An improved algorithm for cloud base detection by ceilometer over the ice sheets.

Optically thin ice clouds play an important role in polar regions due to their effect on cloud radiative impact and precipitation on the surface. Cloud bases can be detected by lidar-based ceilometers that run continuously and therefore have the potential to provide basic cloud statistics including cloud frequency, base height and vertical structure. Despite their importance, thin clouds are however not well detected by the standard cloud base detection algorithm of most ceilometers operational at Arctic and Antarctic stations. This paper presents the Polar Threshold (PT) algorithm that was developed to detect optically thin hydrometeor layers (optical depth τ ⩾ 0.01). The PT algorithm detects the first hydrometeor layer in a vertical attenuated backscatter profile exceeding a predefined threshold in combination with noise reduction and averaging procedures. The optimal backscatter threshold of 3×10^-4 km^-1 sr^-1 for cloud base detection was objectively derived based on a sensitivity analysis using data from Princess Elisabeth, Antarctica and Summit, Greenland. The algorithm defines cloudy conditions as any atmospheric profile containing a hydrometeor layer at least 50m thick. A comparison with relative humidity measurements from radiosondes at Summit illustrates the algorithm's ability to significantly differentiate between clear sky and cloudy conditions. Analysis of the cloud statistics derived from the PT algorithm indicates a year-round monthly mean cloud cover fraction of 72% at Summit without a seasonal cycle. The occurrence of optically thick layers, indicating the presence of supercooled liquid, shows a seasonal cycle at Summit with a monthly mean summer peak of 40 %. The monthly mean cloud occurrence frequency in summer at Princess Elisabeth is 47 %, which reduces to 14% for supercooled liquid cloud layers. Our analyses furthermore illustrate the importance of optically thin hydrometeor layers 25 located near the surface for both sites, with 87% of all detections below 500m for Summit and 80% below 2 km for Princess Elisabeth. These results have implications for using satellite-based remotely sensed cloud observations, like CloudSat, that may be insensitive for hydrometeors near the surface. The results of this study highlight the potential of the PT algorithm to extract information in polar regions about a wide range of hydrometeor types from measurements by the robust and relatively low-cost ceilometer instrument. [ABSTRACT FROM AUTHOR]