This paper utilizes a rotated principal components approach first used by Eder, Davis, and Bloomfield (Atmos. Environ. 27a (16) (1993) 2645) to characterize non-urban daily 8-h maximum ozone (O-3) concentrations within the eastern United States for the O-3 seasons of 1993-2002. The analysis proceeds by selecting a spatially representative O-3 database, imputing missing O-3 concentrations using a spatial interpolation scheme, applying a rotated principal components analysis to delineate spatial regions of homogeneous concentrations, and investigating the temporal patterns exhibited by concentrations in each of the regions. Spatially, the analysis divides the eastern United States into five regions: a Northeast region, a Great Lakes Region, a Mid-Atlantic region, a Southwest region, and a Florida region. Concentrations are near the domain average for the Northeast, Great Lakes, and Southwest regions, are highest in the Mid-Atlantic region, and are lowest in the Florida region. Within each region the temporal patterns (e.g., seasonal trends, persistency; and annual trends) Of O-3 concentrations were quite different. The Northeast, Great Lakes, and Mid-Atlantic regions display a moderate amount of seasonal variability with peak concentrations occurring in late July, late June, and late July time periods, respectively. Conversely, the Southwest region exhibits a small amount of seasonal variability with peak concentrations occurring in late August, and the Florida region exhibits a relatively large amount of seasonal variability with peak concentrations occurring during the months of April and May. Ozone concentrations are most persistent in the Florida and Southwest regions (3-4 days of persistence), less persistent in the Mid Atlantic region (2-3 days of persistence), and least persistent in the Northeast and Great Lakes regions (I or 2 days persistence). These results demonstrate the spatial and temporal variability Of O-3 concentrations in the eastern United States and indicate that regional characterization (e.g., through aggregation Of Clusters of monitors) Of O-3 air quality may be a powerful yet simplifying air quality metric for determining seasonal trends, annual trends, and a variety of other O-3 concentration characteristics. (C) 2004 Elsevier Ltd. All rights reserved.
