Mean of Autocorrelated Air Quality Measurements

The high degree of autocorrelation present among the observations in a continuous record of air quality has significant implications with respect to the estimation of the mean value of air pollutant concentration and to the determination of data collection requirements for estimating air quality levels. Based on the characterization of the form of the autocorrelation function, equations are derived for the equivalent number of independent observations in a sample of given size. An expression for the variance of the sample mean of an autocorrelated series is derived; and equations to estimate data collection requirements for a sequence of air quality measurements are developed. These equations yield both the number of equivalent independent measurements and the number of measurements in a continuous record needed to estimate the mean value at given levels of precision and significance. The derived results are applied to observed ambient carbon monoxide measurements.