Modelling of SOA markers : simulation through detailed mechanisms and validation by comparison with measurements. a new approach to understand SOA formation

Secondary organic aerosol (SOA) is formed via the oxidation of both anthropogenic and biogenic gas-phase organic compounds and is a large and often dominant fraction of total organic aerosol (OA). SOA markers from specific precursors have been identified and are currently used to assess Secondary Organic Carbon (SOC) sources. As a matter of fact, air quality models have still difficulties to reproduce observed particulate matter (PM) concentration levels due notably to a poor simulation of the SOA fraction. It is then necessary to develop atmospheric chemistry models that properly describe the formation of the SOA markers in the atmosphere to improve the understanding of SOA formation to improve air quality simulations (forecast, scenario analysis). In this context, the modelling approach developed must be compared with data obtained through field measurements. The aim of this work is to implement SOA marker mechanisms into the air quality model CHIMERE and to compare model results with field measurements data. This comparison gives an insight on the ability of the model to form SOA from specific precursors and on several processes (e.g. emissions, gas/particle partitioning). Measurements of SOA markers were performed at SIRTA station, representing the suburban background air quality conditions nearby Paris (25 km SW from city center). PM10 samples were collected every third day all over the year 2015. SOA markers have been quantified using native standard compounds by LC/MS-MS and/or GC/MS. The mechanisms of SOA markers formation implemented in CHIMERE were taken from the Master Chemical Mechanism (NCAS, Universities of Leeds and York) otherwise, data were issued from the scientific literature. The gas phase mechanism simulation was performed using MELCHIOR2, the gas-particulate partitioning was calculated using the thermodynamic model SOAP. Biogenic emissions were computed with the MEGAN 2.1 algorithm. Simulated markers included both biogenic, (e.g; pinonic acid, pinic acid and MBTCA:α-pinene oxidation), and anthropogenic (e.g. DHOPA and nitrophenols: toluene oxidation) precursors