Statistical Models for the Analysis of Volatile Organic Compounds in Waste Disposal Sites

The occurrence of low‐level hits of volatile organic priority pollutant compounds is statistically modeled as a Poisson process. Methods are developed to estimate the mean of the Poisson distribution for a random sample of volatile organic measurements as well as 99% prediction limits and 99% tolerance limits. The prediction limits provide an interval estimate that will include values obtained for the next k future measurements based on a sample of n previous measurements with 99% confidence. The tolerance limits provide an interval estimate for the n previous measurements that will contain 99% of the population of background measurements with 95% confidence. These methods are illustrated with measurements obtained from 61 field blanks, 56 trip blanks, and 162 samples obtained from 29 upgradient wells. Both prediction and tolerance limits yielded extremely similar results in all three examples.