Atmospheric measurements of ratios between CO2 and co-emitted species from traffic: a tunnel study in the Paris megacity

Measurements of CO2, CO, NO and selected Volatile Organic Compounds (VOCs) mole fractions were performed continuously during a 10-day period in the Guy Moquet tunnel in Thiais, a pen-urban area about 15 km south of the centre of Paris, between 28 September and 8 October 2012. This data set is used here to identify the characteristics of traffic-emitted CO2 by evaluating its ratios to co-emitted species for the first time in the Paris region. High coefficients of determination (r(2) > 0.7) are observed between CO2 and certain compounds that are characteristic of the traffic source (CO, NOx, benzene, xylenes and acetylene). Weak correlations (r(2) < 0.2) are found with species such as propane, nbutane and i-butane that are associated with fuel evaporation, an insignificant source for CO2. To better characterise the traffic signal we focus only on species that are well-correlated with CO2 and on rush-hour periods characterised by the highest traffic-related mole fractions. From those mole fractions we remove the nighttime-average weekday mole fraction obtained for each species that we infer to be the most appropriate background signal for our study. Then we calculate observed Delta species / Delta CO2 ratios, which we compare with the ones provided by the 2010 bottom up high-resolved regional emission inventory from Airparif (the association in charge of monitoring the air quality in Ile-de-France), focusing on local emission data for the specific road of the tunnel. We find an excellent agreement (2 %) between the local inventory emission CO / CO2 ratio and our observed Delta CO / Delta CO2 ratio. Former tunnel experiments carried out elsewhere in the world provided observed Delta CO / ACO(2) ratios that differ from 49 to 592 % to ours. This variability can be related to technological improvement of vehicles, differences in driving conditions, and fleet composition. We also find a satisfactory agreement with the Airparif inventory for n-propylbenzene, n-pentane and xylenes to CO2 ratios. For most of the other species, the ratios obtained from the local emission inventory overestimate the observed ratios to CO2 by 34 to more than 300%. However, the emission ratios of NON, o-xylene and i-pentane are underestimated by 30 to 79 %. One main cause of such high differences between the inventory and our observations is likely the obsolete feature of the VOCs speciation matrix of the inventory that has not been updated since 1998, although law regulations on some VOCs have occurred since that time. Our study bears important consequences, discussed in the conclusion, for the characterisation of the urban CO2 plume and for atmospheric inverse modelling of urban CO2 emissions.