Robust Statistical Methods and Photochemical Air Pollution Data

Abstract

Robust statistical methods are those whose results are not dramatically affected by a small fraction of the observations. They are particularly useful when the goal is to determine the typical behavior of the data. Their importance in the field of air pollution is demonstrated by the analysis of data relevant to an understanding of the nature of atmospheric photochemistry in the New York City-New Jersey metropolitan region. Two types of analyses are presented. 1) For sunny workdays from May 1 to September 30, 1973, three robust, "average," diurnal curves, corresponding to three intervals of daily maximum O₃ in Bayonne, NJ, are presented for Bayonne O₃, NO_x, aldehydes, and total hydrocarbons. 2) By means of quantile-quantile plots it is shown that O₃ concentrations in the Camden-Philadelphia urban complex are generally lower than in the nonindustrial, low-traffic density area of Ancora, NJ, 37 km to the southeast.