A comparison of the performance of four air quality models for the Southern Oxidants Study episode in July 1999

Modeling systems that are designed to investigate tropospheric air quality concerns must address several issues simultaneously: ozone, particulate matter, deposition and visibility. These modeling systems consist of three components: meteorology, emissions and air quality. A simulation was conducted for the July 1999 Southern Oxidants Study to evaluate one such modeling system, Models-3. Performance of the meteorological component, MM5, was evaluated against observations. Consistency of two emissions models, SMOKE and EPS 2.5, was evaluated by comparing their outputs. For comparison, the performance of CMAQ and three additional models (CMAQ-MADRID 1, CMAQ-MADRID 2, and REMSAD) was evaluated for the same time period. Nested simulations for a 32-km and an 8-km grid were conducted for CMAQ and CMAQ-MADRID 1. Results for CMAQ-MADRID 2 and REMSAD are available only for the 8-km grid. Performance was evaluated for PM and its components, ozone and wet deposition. Differences in model performance for PM2.5 and its components were greatest for OC and total PM2.5; performance was more consistent for the other components. Model performance was generally better on the 32-km grid than the 8-km grid for PM2.5 and its components. R-2 values ranged from 20 to 50% for NH4+ and EC and were lower for other PM2.5 components, indicating that the predictive capabilities of the models for PM2.5 are limited. Model performance for ozone met EPA guidance for MNB and MNE when a 60 ppb cutoff was used and was better on the 8-km grid than on the 32-km grid. Performance for wet deposition was good.