STATISTICAL PREDICTION INTERVALS FOR THE EVALUATION OF GROUNDWATER QUALITY

Factors for a normal distribution are given such that one may be 99% confident that the two-sided prediction interval x OVER BAR plus or minus rs or the one-sided prediction interval x OVER BAR plus rs will contain all of the k future values, where x OVER BAR and s are the sample mean and standard deviation obtained from n previous values. In the context of ground-water monitoring, the future samples may represent new monitoring values at each of k downgradient wells, and the n previous values might be the historical monitoring results for one or more upgradient wells. The Tables provided in this paper allow the computation of one-sided and two-sided 99% prediction intervals for previous sample sizes of n equals 4 to 100 and future samples of k equals 1 to 100. Modification of these intervals for log-normally distributed data is also presented.