SOURCE-ORIENTED MODEL FOR AIR POLLUTANT EFFECTS ON VISIBILITY

A source-oriented model for air pollutant effects on visibility has be en developed that can compute light scattering, light extinction, and estimated visual range directly from data on gas phase and primary par ticle phase air pollutant emissions from sources. The importance of su ch a model is that it can be used to compute the effect of emission co ntrol proposals on visibility-related parameters in advance of the ado ption of such control programs. The model has been assembled by embedd ing several aerosol process modules within the photochemical trajector y model previously developed for aerosol nitrate concentration predict ions by Russell et al. [1983] and Russell and Cass [1986]. These modul es describe the size distribution and chemical composition of primary particle emissions, the speciation of organic vapor emissions, atmosph eric chemical reactions, transport of condensible material between the gas and the particle phases, fog chemistry, dry deposition, and-atmos pheric light scattering and light absorption. Model predictions have b een compared to observed values using 48-hour trajectories arriving at Claremont, California, at each hour of August 28, 1987, during the So uthern California Air Quality Study. The predicted fine particle conce ntration averages 62 mu g m(-3) compared to an observed value of 61 mu g m(-3), while predicted PM10 concentrations average 102 mu g m(-3) c ompared to an observed average of 97 mu g m(-3). The size distribution and chemical composition predictions for elemental carbon, sulfate, a nd sodium ion agree with observations to within plus or minus a few mi crograms per cubic meter, while ammonium and nitrate concentrations ar e underpredicted by the base case model by 3 to 7 mu g m(-3) on averag e. Light-scattering coefficient values are calculated from the predict ed aerosol size distribution and refractive index, and the model predi ctions agree with measured values on average to within 19%. The advant ages and limitations of the modeling procedure are discussed.