Identification of representative warm season periods for regional air quality (ozone) model simulations

There is a need to apply detailed air quality models to examine the fate of air pollutant emissions over periods of a year or more so that issues which involve annual or even multi-year standards can be addressed. This is a difficult task due to computational and model input demands. In this study we examine multiple years of ozone and meteorological data for eastern Canada to determine how many warm seasons (May–September) would need to be modelled to ensure that the range of typical conditions in the warm season is included. We also evaluate how well individual warm seasons can match these conditions, assuming that if a single, typical warm season can be identified then it may be a good candidate for future modelling studies. We found that representation of the long-term average warm season O3 distribution, cumulative O3 statistics, surface weather conditions and frequency of synoptic-scale weather patterns with a combined absolute error of less than 20% requires, on average, five consecutive warm seasons. The number can be reduced to three through selection of specific consecutive warm seasons. The single most representative warm season matches the long-term conditions to within 26% in Ontario, 29% in Quebec and 30% in Nova Scotia and New Brunswick, while inclusion of a second year reduces these errors by 4, 7 and 2%, respectively. Beyond two years the reduction in error from adding more warm seasons occurs at an increasingly slower rate.