Ozone ensemble forecasts: 1. A new ensemble design

[1] A new Ozone Ensemble Forecast System (OEFS) is tested as a technique to improve the accuracy of real-time photochemical air quality modeling. The performance of 12 different forecasts along with their ensemble mean is tested against the observations during 11 - 15 August 2004, over five monitoring stations in the Lower Fraser Valley, British Columbia, Canada, a population center in a complex coastal mountain setting. The 12 ensemble members are obtained by driving the U. S. Environmental Protection Agency (EPA) Models-3/Community Multiscale Air Quality Model (CMAQ) with two mesoscale meteorological models, each run at two resolutions (12- and 4-km): the Mesoscale Compressible Community (MC2) model and the Penn State/NCAR mesoscale (MM5) model. Moreover, CMAQ is run for three emission scenarios: a control run, a run with 50% more NOx emissions, and a run with 50% fewer. For the locations and days used to test this new OEFS, the ensemble mean is the best forecast if ranked using correlation, gross error, and root mean square error and has average performance when evaluated with the unpaired peak prediction accuracy. Ensemble averaging removes part of the unpredictable components of the physical and chemical processes involved in the ozone fate, resulting in a more skilful forecast when compared to any deterministic ensemble member. There is not one of the 12 individual forecasts that clearly outperforms the others on the basis of the four statistical parameters considered here. A lagged-averaged OEFS is also tested as follows. The 12-member OEFS is expanded to an 18- member OEFS by adding the second day from the six 12-km "yesterday'' forecasts to the "today'' ensemble forecast. The 18-member ensemble does not improve the ensemble mean forecast skill. Neither correlation nor a relationship between ensemble spread and forecast error is evident.