Controls of seasonal high-level ozone statistics in the Northeastern US, part I: An examination of 19932012

- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner
Monday, 5 January 2015: 1:30 PM
228AB (Phoenix Convention Center - West and North Buildings)
E. M. Oswald, University Corporation for Atmospheric Research, Burlington, VT; and L. A. L. Dupigny-Giroux, E. M. Leibensperger, R. Poirot, and J. Merrell

The goal of this study is to support an assessment of the climate change impacts on ozone, and to produce new knowledge regarding the linkages between the climate system and surface-level ozone concentrations. This study focuses on the 1993-2012 time period for the Northeastern United States and the high side (right tail) of the distribution of ground-level ozone concentrations. The first portion of this study establishes relationships with meteorological-based predictors. The second portion of this study examines the links between ozone and large-scale teleconnection patterns within the climate system. Statistical models for each station are established via Principal Component Analysis and Multiple Regression Analysis.

In general, the strongest meteorological-based predictors of ozone in the study region are frequency of high temperatures and precipitation, as well as solar radiation flux. Statistical models of meteorological variables explained about 60-75% of the variability in the annual ozone time series, and had a typical error-to-variability ratio of 0.50-0.65. Teleconnection-based variables such as the Artic Oscillation, North American Oscillation, Quasi-Biennial Oscillation and the Pacific Decadal Oscillation are best linked to ozone in the region. Statistical models based on teleconnection-based variables explained 40-60% of the variability in the ozone annual time series, and had a typical error-to-variability ratio of 0.60-0.75.