Abstract:
Urban ozone arises as a consequence of the emissions of nitrous oxides and hydrocarbons into the atmosphere, but it is also very strongly affected by meteorological conditions. In analyzing ozone data, it is hard to separate genuine trends, that might be explained by changes in the levels of emissions, from apparent ones determined by meteorology. Previous statistical studies of this problem have used classification and regression techniques. In particular, Bloomfield et al. (1993) have developed a detailed nonlinear regression model fitted to weighted averages of measured ozone concentrations at 45 monitoring stations in the Chicago area. Here, based on the same data set, we develop an alternative analysis aimed specifically at characterizing the probability of exceeding a high threshold.
Keywords:
Ozone concentration, Meteorological adjustment, Regression, Extreme value analysis, Threshold exceedances.
tr6.pdf