Multi-wavelength aerosol absorption coefficient measurements: instrument inter-comparison and results of source and source-component modelling

In this work, some results obtained in the frame of a comprehensive measuring campaign carried out within the COST COLOSSAL activities at an urban background station at the University of Milano-Bicocca (Italy) will be presented. The work aims to gain insight into the performances of: · different instruments for the determination of the aerosol absorption coefficient (babs); · different models based on multi-wavelength (multi-λ) babs data to identify light absorbing aerosol sources (fossil fuel, FF; biomass burning, BB) and/or components (Black Carbon, BC; Brown Carbon, BrC). Twelve-hour resolved aerosol samples were collected on pre-fired (700°C, 1h) quartz fibre filters by a low volume sampler (flowrate 1m3/h). The samples were analysed by the multi-λ polar photometer PP_UniMI[1,2] at the University of Milan for off-line babs determination, by ThermalOptical Transmittance analysis (EUSAAR_2 protocol) for OC/EC and by HPAEC_PAD for levoglucosan at the University of Genoa. The site was also equipped with on-line instrumentation - Aethalometers AE31 and AE33 (Magee Scientific), and a Multi-Angle Absorption Photometer MAAP (Thermo-Fischer) - providing information on BC from babs determination. The babs from PP_UniMI and MAAP (data averaged to filter sampling time resolution) agreed very well. A high correlation (R=0.96) was found also between MAAP/PP_UniMI and Aethalometer data, but results from both AE31 and AE33 featured slopes significantly higher than 1 relative to the MAAP/PP_UniMI when using the Aethalometer tape specific multiple scattering parameter CAE31= 2.14 and CAE33= 1.57. Higher values of these parameters (e.g. CAE33>2.5) seem necessary to obtain the realistic babs from Aethalometers. Opposite, BC data from on-line instruments showed lower discrepancy, probably due to the different mass absorption crosssections (also considering the different λs of operation) set in the instruments. Multi-λ Aethalometer babs measurements were used as input to the Aethalometer model[3] and the MultiWavelength Absorption Analyzer (MWAA) model[4] for source and source-component apportionment, respectively. As for the Aethalometer model, a multi-λ fit was also attempted instead of choosing two fixed λs as performed in the literature. Differences depending on the chosen λs range were noticed with both approaches. The MWAA model provided αBrC estimate (3.8±0.4 considering the 470-950 nm range) during the campaign, in addition to BC and BrC apportionment. An episode characterised by unusual BrC contribution (identified by high Ångstrom absorption exponent in Aethalometer data) was identified and deeply investigated in terms of model performance.