Quantitative assessment of uncertainties for a model of tropospheric ethene oxidation using the European Photoreactor (EUPHORE)

Methods of uncertainty analysis were used for comparison of the Master Chemical Mechanism version 3 (MCMv3) with measurements made in the European Photoreactor (EUPHORE) at Valencia (Spain) to investigate model–measurement discrepancies and to obtain information on the importance of wall effects. Two EUPHORE smog chamber measurements of ethene oxidation, under high and low NOx conditions were analysed by the following methods: (i) local uncertainty analysis, (ii) the global screening method of Morris and (iii) Monte Carlo (MC) analysis with Latin hypercube sampling. For both experiments, ozone (by 25% and 30%, respectively) and formaldehyde (by 34% and 40%, respectively) are significantly over-predicted by the model calculations, while the disagreement for other species is less substantial. According to the local uncertainty analysis and the Morris method, the most important contributor to ozone uncertainty under low NOx conditions is HOCH2CH2O2+NO→HOCH2CH2O+NO2, while under high NOx conditions OH+NO2→HNO3 is the main contributor. The MC simulations give an estimate of the 2σ uncertainty for ozone as ∼20% in both scenarios at the end of the experiment. The results suggest systematic disagreement between measurements and model calculations, although the origin of this is not clear. It seems that chamber effects alone are not responsible for the observed discrepancies.