A climatology of surface ozone and PM2.5 for North America

We present a surface climatology for ozone (2002-2009) and for particulate matter PM2.5 (2005-2009). The methodology is based on a sequential enhanced optimal interpolation with capabilities for adaptative error statistics for ozone and PM2.5 and a bias correction scheme for PM2.5. Optimal interpolation combining the chemical transport model CHRONOS (which was in use for operation forecast in Canada from 2001-2009) with AIRNOW US/EPA database which regroups most of representative air quality monitoring sites in North America are the basic of multiple year re-analysis. The error statistics for data ingestion have been computed using the Hollingsworth-Lonnberg's method (1986) followed by a quality control. The error statistics are then tuned “on the fly” using the Chi-squared diagnosis to achieve the "effective" minimization. The procedure is shown to verify significantly better than without tuning. Successful cross-validation experiments were performed with a set-up with an OA using 80%-90% of observations to build the objective analysis and with the remainder as independent set of data for verification. Moreover, comparisons with other source of independent information (MOZART, GEM-AQ models and other climatologies) show reasonable agreement. Applications of re-analysis and climatology are numerous: calculation of surface pollutant trend, correlation with epidemiological studies, model validation, etc.