Evaluation of CMAQ sulfate simulations using surface and satellite measurements: Is cloud processing a source or sink of sulfate?

The Community Multiscale Air Quality (CMAQ v4.6) model system, with emission inputs prepared using the VISTAS inventory and the Sparse Matrix Operator Kernel Emissions (SMOKE2.2) Modeling System, is used to analyze air quality observations in 2002 over the United States. The ACM and RADM schemes are used in the model to test the sensitivities of model results to cloud processing modules. In this study, we focus on the evaluations of model simulations of sulfate, a major component of PM2.5. Observations from the SEARCH, IMPROVE, NADP, CASTNet and STN networks are applied in model evaluations. While the general seasonal cycles of SO2 and sulfate are simulated by the model, we find some systematic biases. Compared to an observed average of 4.08 ug/m3 of sulfate in the Southeast U.S., the model simulated averages of 2.62 ug/m3 (CMAQ/RADM) and 2.31 ug/m3 (CMAQ/ACM) are 35-45% lower Generally. Over the entire U.S., the model simulated averages of 1.65 ug/m3 (CMAQ/RADM) and 1.60 ug/m3 (CMAQ/ACM) are 33-35% lower than the observed average of 2.46 ug/m3. The underestimation is less in winter but more in summer. Application of the ACM scheme tends to worsen the low bias in summer compared to the RADM scheme. On the other hand, model SO2 simulations have a systematic high bias. Evaluations using deposition rates from the NADP and CASTNet networks show model biases consistent with the concentration evaluations. Four satellite products are used to compare to CMAQ diagnosed cloud water content; ACM simulated cloud volume ratios are higher than simulated by RADM and the satellite observations. Budget calculation reveals that cloud processing in the model is a larger sink (through wet scavenging) than source (through heterogeneous oxidation of SO2). Potential model biases in simulated cloud processing of sulfate are investigated.