Layer detection algorithm for CALIPSO observation based on automatic segmentation with a minimum cost function

Abstract CALIPSO (cloud-aerosol lidar and infrared pathfinder satellite observation) provides unique opportunities for profiling global cloud and aerosol. It is crucial to accurately detect the boundaries of cloud and aerosol layers from CALIPSO observation because the detecting error will be passed to further retrieval. Considered superior to other layer detection methods, the threshold method is the core of the selective iterated boundary location (SIBYL) algorithm developed for producing the CALIPSO official products. However, the threshold method can miss many tenuous layers, and the use of the slope method to refine the layer base in SIBYL leads to considerable uncertainty due to its high sensitivity to noise. This study proposed a new layer detection algorithm based on an automatic segmentation method with a minimum cost function. Results show that the new algorithm determines 21% and 13% more layers than SIBYL at 1 km and 1-5 km resolution, respectively, which indicates that the new algorithm has higher detection efficiency. Moreover, the layers detected by the new algorithm are 170 m thicker than that detected by SIBYL on average, which indicates that the SIBYL misses layer edges where the signal to noise ratio is low. The new algorithm can improve the accuracy and resolution of the layer products of CALIPSO as well as other space-based lidars.