Development and application of a new air pollution modeling system — Part III. Aerosol-phase simulations

Result from a new air pollution model were tested against data from the Southern California Air Quality Study (SCAQS) period of 26–29 August 1987. Gross errors for sulfate, sodium, light absorption, temperatures, surface solar radiation, sulfur dioxide gas, formaldehyde gas, and ozone were lowest among parameters compared (1–40%). Gross errors for elemental carbon, organic carbon, total particulate mass, ammonium, ammonia gas, nitric acid gas, and light scattering, were larger (40–61%). Gross errors for particulate nit rate were largest (65–70%). Reducing the baseline land-based particulate emissions inventory to one-third its original value did not affect gross errors significantly, however, it did turn overpredictions into underpredictions for many species. Doubling emissions increased gross errors for nearly all parameters. Setting lateral boundary inflow concentrations of particles to zero caused slight (<1%) erosion of results for most species, large erosion (10%) for sodium and chloride, but slight improvement (<1%) for a few species. Setting both lateral inflow and initial concentrations for gases and particles to zero caused severe degradation of results for many species but relatively mild degradation or improvement for a few. Spinning up the meterological model 24 h in advance caused most gross errors to increase. Finally, the presence of aerosols reduced peak daytime surface solar radiation by approximately 6.4% (55 W m−2), increased nighttime temperatures by about 0.77 K, decreased daytime temperatures by about 0.08 K, and increased overall temperatures (day plus night) by 0.43 K compared to a no-aerosol case. The presence of aerosols also caused ozone mixing ratios to decrease by 2%.