METEOROLOGICAL ADJUSTMENT OF YEARLY MEAN VALUES FOR AIR POLLUTANT CONCENTRATION COMPARISONS

摘要

Using multiple linear regression analysis we derive models which estimate mean concentrations of Total Suspended Particulate (TSP), sulfur dioxide (SO2), and nitrogen dioxide (NO2) as a function of several meteorologic variables, two rough economic indicators, and a simple trend in time. Considered are 24-hour averaged concentrations measured in Cleveland, Ohio, from 1967 to 1972 (approximately 450 observations for TSP and 400 observations at 13 sites for SO, and NO2) by the municipal Division of Air Pollution Control. Meteorologic data were obtained from the National Weather Service and do not include inversion heights. This is representative of data typically available to a local pollution-control agency. The goodness of fit of the esti¬mated models is partially reflected by the squared coefficient of multiple correlation which indicates that, at the various sampling stations, the models accounted for about 23 to 47 percent of the total variance of the observed TSP concentrations. If the resulting model equations are used in place of simple overall means of the observed concentrations, there is about a 20 per¬cent improvement in either (1) predicting mean concentrations for specified meteorological conditions or (2) adjusting successive yearly averages to allow for comparisons devoid of meteorological effects. This improvement can be obtained with no additional cost other than a moderate effort at statistical analysis. An application to source identification is presented using regression coefficients of wind velocity predictor variables.