Assessing background air pollutant concentrations for modelling studies: Evaluation of addition equations under Irish conditions

Values adopted for background pollutant concentrations can have a significant effect on the accuracy of the final results in air quality modelling studies. In the absence of a reliable method of combining modelled and background concentrations it has been common practice to sum the percentiles or annual means of each contribution to obtain a value for comparison with limit values. This is often not appropriate as in many cases the meteorological conditions producing high concentrations from the source do not correspond to those resulting in high background concentrations. The validity of a number of equations derived in the UK to add background NO₂ and PM₁₀ concentrations to modelled stack contributions has been examined for Irish conditions. The equations allow a total percentile concentration to be predicted at a given receptor based on an annual mean background concentration and hourly modelled concentrations. A theoretical point source was modelled using the point source Gaussian plume equation and corresponding meteorological data, and the addition equations applied using monitored background NO₂ and PM₁₀ data. The equations were also tested for a line source, modelled using the General Finite Line Source Model (GFLSM). Baseline values were calculated by addition of the relevant hourly or daily background concentration to the modelled concentrations to produce a full year of total hourly or daily concentrations. Percentiles and annual mean values, and corresponding 95% confidence limits were calculated directly from this data set. Percentile concentrations predicted by each of the equations were compared to the baseline values. It was found that all equations produced values outside the confidence limits, indicative of systematic variation. Certain methods, however, provide an improvement to commonly applied addition methods and may be useful for screening purposes. Results for the line source were poorer than those for the point source. Further research in this area is required to develop more accurate methods for the addition of background concentrations to modelled contributions, particularly in the case of line sources.