Oxidative potential modelling of PM10 : an indicator of aerosol health risk studied in France with the CHIMERE model

<p>Air pollution remains an outstanding issue due to its hazardous health and environmental impacts. For France, the number of premature deaths is estimated around 40,000 people per year mostly due to particulate matter (PM), while associated economical coast is estimated at 100 billion euros per year. In this context, action plans are implemented in order to reduce the PM mass concentrations in ambient air. However, considering mass concentration only can lead to an obvious bias: for instance, at equal mass concentrations, the exposure to fresh sea salts is assumed to have the same toxicity as soot particles containing toxic compounds. Therefore, accurate abatment measures need thorough knowledge on the PM chemical composition, which can then be used within receptor and/or chemical transport models to apportion their emission sources and secondary formation processes. The PM chemistry is also relevant in terms of adverse health effects. Based on the PM chemical species ability to generate oxidative stress through reactive oxygen species, the oxidative potential (OP) indicates the consumption of antioxidant per particles mass. This proxy estimates the imbalance between oxidants and antioxidants, responsible for inflammatory processes and chronic diseases. As a result, the aerosol&#8217;s oxidative potential has emerged as a promising indicator of PM adverse health impacts.</p> <p>To better evaluate PM health effects, we set-up a strategy to implement OP in the state-of-art air quality model CHIMERE and to simulate particles OP over the whole French territory for the year 2013 and 2014. To do so, a measurement derived and source specific OP determined by Positive Matrix Factorization (PMF) receptor modelling approach is combined with particle sources apportionment in CHIMERE using a tagging method called Particulate Source Apportionment Technology (PSAT). Alternatively, a source specific OP is obtained by linear regression of observed OP and simulated sources. Both methods are used to simulate OP over France for the years 2013 and 2014, and to determine the most affected areas and responsible sources.</p>