Air pollutant averaging times: Notes on a statistical model

Abstract The logarithmic standard deviation of time-averaged air pollutant concentrations has been shown by CAMP data to decrease as an inverse power B of the averaging time. If the correlation coefficient between the logarithm of the concentrations at distant times decreases as an inverse 2B power of the time difference, such behavior would be expected. This heavy-tailed correlation function suggests the existence of some possibly small, but highly persistent components of pollutant concentration. One possibility is that the observed process was not detrended. In order to test this hypothesis, it will be useful to calculate the average logarithm of the concentration instead of the logarithm of the average concentrations.