TWO YEARS OF OPERATIONAL AQ FORECASTING WITH GEM-MACH15: A LOOK BACK AND A LOOK AHEAD

GEM-MACH15 (Global Environmental Multiscale model-Modelling Air quality and CHemistry at 15 km) has been Environment Canada’s (EC) operational regional air quality (AQ) forecast model since it replaced the CHRONOS AQ model in November 2009 after a three-month test period. GEM-MACH15 (abbreviated G-M15) is a limited-area configuration of GEM-MACH, an on-line chemical transport model that is embedded within GEM, EC’s multi-scale operational weather forecast model. The operational version of GM15 is run twice daily at EC’s Canadian Meteorological Centre to produce 48-hour AQ forecasts on a North American grid (Anselmo et al., 2010). G-M15 forecasts provide important guidance to the Canadian AQ forecast program, whose key product is the national Air Quality Health Index (AQHI) for urban areas. The AQHI is a healthbased, additive, no-threshold, hourly AQ index with a 0 to10+ range that is based on a weighted sum of local O3, PM2.5, and NO2 concentrations. It was developed from a time-series analysis of air pollutant concentrations and mortality in Canadian cities (Stieb et al., 2008). Although G-M15 predicts hourly concentration fields of many gasphase and size-specific particle-phase species, the three most important fields are O3, PM2.5, and NO2, needed for the AQHI. The G-M15 forecast system has now been running operationally for more than two years but has undergone several modifications during that period (Sec. 2). This paper presents selected statistical analyses of model performance during this period and notes some impacts of model changes on model performance (Sections 3-5). G-M15 post-processing and future plans are then discussed in Sections 6 and 7.