Modeling of Ground-Level Ozone Exceedances Measured in Six National Parks Using the Jonckheere-Terpstra Test

The Clean Air Act (CAA) Amendments of 1990 established increased management of ground-level ozone concentrations by mandating implementation of ozone reduction proposals according to the severity of a region’s pollution. Analysis of the trend of ground-level ozone exceedances in national parks, which pick up pollutants from nearby cities, can be a good indicator of the effect of the CAA. We analyzed data from eight different sites in six national parks from 1987 to 2012 to observe ozone trends in various conditions. Based on the requirements of the act, we hypothesized that there would be an overall decreasing trend particularly in non-attainment areas. We utilized the approach of clustering exceedances over a predetermined threshold level at time intervals that yield independence between readings. Our chosen threshold was 0.068 parts per million (ppm), which maximized the number of clusters per year, and we clustered exceedance points within 72 hours of each other. Each cluster maximum served as the 'representative point' and, after being grouped by year, was evaluated using the Jonckheere-Terpstra Test, which stochastically ordered the years non-parametrically since the Generalized Pareto Distribution was ill-fitting. While at five of the sites chosen the p-values, at an alpha level of 0.05, precluded rejecting the null hypothesis, the other three sites indicated highly significant statistical differences in ozone levels that demonstrated a decreasing trend. This difference may be attributable to the ozone quality in surrounding regions. Rejecting the null hypothesis provides support for the CAA and is useful information for future air quality legislation.