Diagnostic Model Evaluation for Carbonaceous PM2.5 Using Organic Markers Measured in the Southeastern U.S.

Abstract

Summertime concentrations of fine particulate carbon in the southeastern United States are consistently underestimated by air quality models. In an effort to understand the cause of this error, the Community Multiscale Air Quality model is instrumented to track primary organic and elemental carbon contributions from fifteen different source categories. The model results are speciated using published source profiles and compared with ambient measurements of 100 organic markers collected at eight sites in the Southeast during the 1999 summer. Results indicate that modeled contributions from vehicle exhaust and biomass combustion, the two largest sources of carbon in the emission inventory, are unbiased across the region. In Atlanta, good model performance for total carbon (TC) is attributed to compensating errors: overestimation of vehicle emissions with underestimations of other sources. In Birmingham, 35% of the TC underestimation can be explained by deficiencies in primary sources. Cigarette smoke and vegetative detritus are not in the inventory, but contribute less than 3% of the TC at each site. After the model results are adjusted for source-specific errors using the organic-marker measurements, an average of 1.6 microgC m(-3) remain unexplained. This corresponds to 26-38% of ambient TC concentrations at urban sites and up to 56% at rural sites. The most likely sources of unexplained carbon are discussed.