Formulation and evaluation of IMS, an interactive three-dimensional tropospheric chemical transport model 3. Comparison of modelled C-2-C-5 hydrocarbons with surface measurements

In part 3 of this series of papers on a new 3-D global tropospheric chemica l transport model, using an Integrated Modelling System (IMS), an evaluatio n of the model performance in simulating global distributions and seasonal variations for volatile organic compounds (VOCs) in the atmosphere, is pres ented. Comparisons of model OH concentrations with previous model studies s how consistent modelled OH levels from the subtropics to midlatitudes, whil e more discrepancies occur over the tropical low latitudes, with IMS predic ting the highest levels of OH. The close agreement between modelled OH conc entrations over midlatitudes, where high surface NOx and VOC concentrations are also found, is indicative of the strong photochemical coupling between NOx, VOCs and O-3 over these latitudes. IMS OH concentrations in the North ern Hemisphere (NH) midlatitudes during summer are generally lower than ava ilable measurements, implying that models in general are underestimating OH levels at this location and time of year. Substantial differences between model OH concentrations over low latitudes clearly highlight areas of uncer tainty between models. IMS OH concentrations are the highest in general of the models compared, one possible reason is that biogenic emissions of spec ies such as isoprene and monoterpenes are highest in IMS, leading to higher O-3 levels and hence higher OH. Generally, the IMS VOC concentrations show a similar seasonality to the measurements at most locations. In general th ough, IMS tends to underestimate the NH wintertime VOC maximum and overesti mate the NH summertime VOC minimum. Such an overestimate in summer could be due to IMS underestimating OH levels, or an overestimation of VOC emission s or possibly a problem with model transport, all of these possibilities ar e explored. Except for n-pentane, the model underprediction of a VOC maximu m during the NH winter month strongly suggests a missing emission mechanism in the model or an underestimate of an existing one. It is very likely tha t there is a lack of time varying emission sources in the model to account for the seasonal change in emission behaviour such as increasing energy usa ge (e.g., electricity and gas), road transportation, engine performance, an d other anthropogenic factors which show strong seasonal characteristics. T he anomalous overprediction of wintertime n-pentane compared with its close summertime prediction with the measurements suggest that emissions in this case may be too high.