Impact of chemistry scheme complexity on UK air quality modelling Met Office FitzRoy Road, Exeter, Devon, EX1 3PB United Kingdom Tel: Fax: M. Tilbee, C. Ordóñez, R. Thorpe, N.H. Savage, P. Agnew, L.S. Davis 1) Introduction AQUM (Air Quality in the Unified Model) is an on-line air quality modelling system which is used to provide the operational Met Office air quality forecast for the UK (Savage et al., 2013). The model uses the ‘Regional Air Quality’ (RAQ) chemistry scheme, which includes 40 advected tracers and 18 non-advected species. Lateral boundary fluxes for chemical species are taken from the MACC reanalysis. We have implemented and evaluated a new model configuration which uses the ‘Extended Tropospheric Chemistry Scheme’ (ExtTC), which has a greater level of sophistication and allows the interactive emission of biogenic ozone precursors dependent on meteorological conditions. We have conducted model simulations for both configurations and compared model ozone predictions against AURN surface observations. In an extension to this study we present results of an experimental configuration of AQUM which employs a simplified RAQ chemistry scheme emitting a reduced number of volatile organic compounds (VOCs). 2) Comparison of RAQ and ExtTC Chemistry Schemes in UKCA We have evaluated the performance of the RAQ and ExtTC chemistry schemes in AQUM. Results are more sensitive to the treatment of biogenic VOC emissions (interactive or fixed climatology) than to the complexity of the chemistry scheme (RAQ or ExtTC). There is a significant improvement in modelled ozone with the inclusion of interactive biogenic emissions. Further work will evaluate interactive biogenic VOC emissions for Northern Europe prior to implementation in the operational AQ forecast. We have demonstrated that a reduced VOC complexity RAQ chemistry scheme performs comparably to the standard AQUM configuration for modelling ozone. For short- term forecasting in small modelling domains, where advection from the boundaries plays a relatively large role, the use of a reduced complexity chemistry scheme may lower the computational expense of the model with no significant reduction in forecast skill. Species7758 Tracers5540 SOA Precursors Terpenes Aromatics C2+ alkenes C4+ alkanes Interactive Biogenics Isoprene Ethane Propane Acetone Basic NO x -HO x -CO-CH 4 -O 3 ExtTCRAQ ChemistrySchemes Fig.1 - Comparison of two chemistry schemes in UKCA Figure 3 illustrates the results of two sensitivity simulations run to separate the impact of the two main sources of difference between RAQ and ExtTC, namely the chemistry scheme and the biogenic emissions. F-ExtTC uses the same fixed isoprene emissions as RAQ, which introduces a slight positive bias. I-RAQ incorporates interactive emissions of isoprene within the RAQ scheme, with results broadly similar to ExtTC. There is no systematic positive bias, as is the case with RAQ. The essential characteristics of the ozone response in ExtTC can be captured by using interactive biogenic isoprene within the RAQ scheme; the additional complexities of the chemistry seem to make little difference. RAQ and ExtTC model configurations have been run for July Figure 2 shows bias and mean error statistics for the ozone hindcasts and demonstrates that ExtTC does not have the systematic positive bias in ozone shown by RAQ. 3) Results - RAQ vs ExtTC ExtTC RAQ Fig. 2 – Soccer plots of ozone concentrations at rural and urban sites for RAQ and ExtTC RAQR-RAQ Correlation Bias (µgm -3 ) RMSE (µgm -3 ) Fig. 4 – Time series of ozone at the urban site London Kensington (left) and frequency distribution of ozone across all sites (right) for RAQ (green) and R-RAQ (orange) 6) Conclusions Fig. 3 – Soccer plots of ozone concentrations at rural and urban sites for F-ExtTC and I-RAQ I-RAQ F-ExtTC AQUM has a relatively small domain, hence model predictions can be expected to exhibit sensitivity to the chemical LBCs. We have developed a preliminary LBC tracer of ozone (O3L) to quantify ozone contributions from boundary conditions. Loss of O3L is by dry deposition only. Figure 5 shows the ozone generated within the AQUM domain, excluding contributions from the LBCs. This varies significantly from day to day. On the 19 th July (right), with air coming from the continent, the production of in-domain ozone is considerably higher than on the 12 th July (left), when prevailing westerlies from the Atlantic lead to larger ozone contribution from the boundaries and less ozone generation within the domain. 4) Reduced-complexity RAQ scheme AQUM has been run with a simplified RAQ chemistry scheme (R-RAQ) emitting only three of the 12 VOCs currently emitted in the operational system. The total mass of anthropogenic VOC is apportioned 50:50 to ethene and ethane and biogenic isoprene is emitted. Fig. 5 – Daily mean modelled O3 - O3L concentrations (µgm -3 ) for the 12 th and 19 th July Table 1 - O 3 statistics for 4th June – 2nd August 2006 for RAQ and R-RAQ configuration 5) Ozone contribution from boundaries References 1. Savage et al., Air quality modelling using the Met Office Unified Model (AQUM OS24-26): model description and initial evaluation. GMD., 6, R-RAQ model ozone predictions are comparable to that of the standard AQUM configuration.