Highlights• Emission scenarios were implemented in eight ozone air quality models• NOx-and VOC sensitivities for peak ozone levels were highly variable between days• Filtering by model performance minimised apparent conflicts between models
KeywordsAir quality modelling, model intercomparison, episodic peak ozone levels, emission sensitivities, NO x and VOC emissions 2
ABSTRACTSimple emission scenarios have been implemented in eight United Kingdom air quality models with the aim of assessing how these models compared when addressing whether photochemical ozone formation in southern England was NO x -or VOC-sensitive and whether ozone precursor sources in the UK or in the Rest of Europe (RoE) were the most important during July 2006. The suite of models included three Eulerian-grid models (three implementations of one of these models), a Lagrangian atmospheric dispersion model and two moving box air parcel models. The assignments as to NO x -or VOC-sensitive and to UK-versus RoE-dominant, turned out to be highly variable and often contradictory between the individual models. However, when the assignments were filtered by model performance on each day, many of the contradictions could be eliminated. Nevertheless, no one model was found to be the 'best' model on all days, indicating that no single air quality model could currently be relied upon to inform policymakers robustly in terms of NO x -versus VOCsensitivity and UK-versus RoE-dominance on each day. It is important to maintain a diversity in model approaches.