Comparison of two automated aerosol typing methods and their application to an EARLINET station

In this study we apply and compare two algorithms for the automated aerosol-type characterization of the aerosol layers derived from Raman lidar measurements over the EARLINET station of Thessaloniki, Greece. Both automated aerosol-type characterization methods base their typing on lidar-derived aerosol-intensive properties. The methodologies are briefly described and their application to three distinct cases is demonstrated and evaluated. Then the two classification schemes were applied in the automatic mode to a more extensive dataset. The dataset analyzed corresponds to ACTRIS/EARLINET (European Aerosol Research Lidar NETwork) Thessaloniki data acquired during the period 2012-2015. Seventy-one layers out of 110 (percentage of 65 %) were typed by both techniques, and 56 of these 71 layers (percentage of 79 %) were attributed to the same aerosol type. However, as shown, the identification rate of both typing algorithms can be changed regarding the selection of appropriate threshold criteria. Four major types of aerosols are considered in this study: Dust, Maritime, PollutedSmoke and CleanContinental. The analysis showed that the two algorithms, when applied to real atmospheric conditions, provide typing results that are in good agreement regarding the automatic characterization of PollutedSmoke, while there are some differences between the two methods regarding the characterization of Dust and CleanContinental. These disagreements are mainly attributed to differences in the definitions of the aerosol types between the two methods, regarding the intensive properties used and their range.