Atmospheric mercury model evaluation

Model evaluation plays an essential role in atmospheric chemical transport models through comparison of observed and modeled data. Atmospheric mercury modelers use several statistics techniques to quantify model performance and uncertainty of mercury studies in Europe (MSC-East), North America, and Polar Regions. Standard statistical tests and procedures using with atmospheric pollutants may be applied to atmospheric mercury concentration because the concentration is generally normal distributed. However, the same tests are not suitable to evaluate wet deposition of mercury which has a non-normal distribution. In addition, some standard performance metrics (i.e. mean normalized bias/error, and mean fractional bias/error) can not apply to data including zero value as of mercury wet deposition (without compromising data completeness). In this study, techniques used by atmospheric mercury modelers will be reviewed and discussed. Continuous measurements of Total Gaseous Mercury (TGM) from the Canadian Atmospheric Mercury measurement Network (CAMNet) and wet deposition of mercury from the Mercury Deposition Network (MDN) will be used as a benchmark to evaluate the data generated by the Community Multiscale Air Quality model (CMAQ). The model domain covers East Asia and North America which is used in a trans-Pacific pollutants transport study of the intercontinental transport and Climatic effects of Air Pollutants project (ICAP).