Cirrus cloud detection by micro pulse lidar: Algorithm development and testing

Micro pulse lidar (MPL) recently has been developed for profiling cloud and aerosol structure over long time periods. MPL offers advantages over previous lidars by providing more horizontal data due to a high pulse repetition rate and long data collection times. The lidar operates at low energy levels (~1 μJ), requiring a more statistical approach for obtaining relevant cloud properties such as cloud base height. Due to the high volume of time versus height backscatter data, an automated algorithm is required. This paper presents an automated algorithm for cirrus cloud detection and develops a simulated cloud model used to test the algorithm. The increased amount of information along the time axis allows one to take advantage of horizontal correlations in the data. Local running standard deviations are taken both vertically and horizontally to determine threshold criteria for cloud boundaries. Image processing techniques are incorporated in the algorithm developed to improve confidence levels in detected cloud boundaries