Paris emission inventory diagnostics from ESQUIF airborne measurements and a chemistry transport model

During the Atmospheric Pollution Over the Paris Area (ESQUIF) experiment a series of airborne measurements were collected in the vicinity of the city of Paris during smog episodes. They are used in combination with an air quality photochemical model in order to diagnose uncertainties in the current emission inventory. Diagnostics are made by comparing simulated with observed concentrations for nitrogen oxides, carbon monoxide, and primary hydrocarbons, taking into account the chemistry and transport processes of these compounds. An emphasis is put on the uncertainty of the results, taking into account the finiteness of the measurement samples, possible errors in the model transport, and chemistry and measurement errors. We examine, in particular, possible sources of bias in the model. For instance, we show that boundary layer depth is underestimated by at most 30% on average. However, sensitivity experiments showed that these model biases, taken individually, cannot alter the qualitative aspects of our results. Only a conspiracy of these biases could possibly shift all our diagnostics toward significantly different results. There is reasonable consistency between simulated and measured concentrations. NOy simulations agree with measured concentrations to within 35%; CO concentrations agree to within a factor of 2. There are significant underestimations and overestimations in some individual primary hydrocarbons. However, the total mass and reactivity of the measured hydrocarbon mixture, which accounts for only about half of the total emitted mass, agree with modeled values to within an estimated uncertainty of 40%. The analysis of results provides clues for improving emission inventories. It is found, for instance, that temperature dependence, which is not considered here, can be a key factor and that hydrocarbon emissions from solvent use may suffer from inaccurate totals or speciation. Another source of uncertainties may be the temporal or spatial distributions of solvent activities.