Validation of AERONET and SKYNET columnar aerosol size distributions by comparison with aircraft in situ measurements at Cape Verde

The Sunphotometer Airborne Validation Experiment (SAVEX-D) was carried out in the Cape Verde archipelago, along with the ICE-D (Ice in Clouds Experiment - Dust) campaign, led by the Met Office and the Universities of Leeds and Manchester (United Kingdom) in summer 2015. The aim of the SAVEX-D experiment was to verify and validate the retrievals of both AERONET’s and SKYNET’s ground based sun-photometers for columnar aerosol optical properties, particularly the aerosol size distributions, with the use of airborne in situ measurements. To this end, a Cimel CE318 (AERONET) and a Prede POM01 (SKYNET) were installed at Praia (Cape Verde), and their retrievals were compared with those obtained with several in-situ instruments equipped in the FAAM BAe-146 research aircraft, such as PCASP, CDP and 2DS. In addition, data from AERONET’s long-term site on Sal was employed. The motivation behind this experiment was related to the non-negligible discrepancies between the retrieved aerosol properties from AERONET and SKYNET networks reported in previous studies. In fact, it is essential to better characterise aerosol retrievals from sun-photometers, since these are employed as inputs in climatological studies, aerosol model verifications and validation of satellite products. The location of the site, downwind of the Sahara Desert (a major source of dust aerosols) is a key aspect of the experiment. In the two flights that were performed (16th and 25th August 2015), dust dominated the aerosol mixture, with a measured AOD about 0.4 - 0.6. Cimel and Prede measurements were inverted through different available versions of the AERONET and SKYNET algorithms. Among the retrieved aerosol properties, aerosol size distribution is of particular interest, as it can be compared to the vertically integrated size distributions obtained from the in-situ instrumentation in the aircraft. From the results, it is observed that both AERONET version 2 and different versions of Skyrad algorithm from SKYNET are consistent in the interval 0.2 - 2.0 µm. However, discrepancies between networks become more noticeable at the extremes of the distributions. In addition, it is shown that AERONET underestimates the size distribution of coarse particles, and that differences in the coarse mode are significant when non-sphericity is considered in the MRI version of the Skyrad inversion algorithm. Further plans include the comparison with new versions of both AERONET and SKYNET algorithms, and GRASP retrievals obtained from inversion of both Cimel and Prede measurements.