A General Statistical Procedure for Ground‐Water Detection Monitoring at Waste Disposal Facilities

Abstract

AbstractNonparametric upper prediction limits for ground‐water detection monitoring are developed. The limit is defined as the maximum of a previous background sample of size n (i.e., X(max, n)). The probability that at least one out of the next m measurements (i.e., Y(min, m)) will be less than X(max, n) at each of k monitoring wells is computed. In the context of ground‐water monitoring, m is equal to the initial new measurement in a particular monitoring well plus m –1 resamplings of that well. In this way, values of n and m can be selected to assure a desired confidence level (e.g., γ≥ .95). A series of typical examples are described, and a set of Tables that give these probabilities for values of n = 4 to 100, k = 1 to 100, and m = 1 to 3 are provided. Generalization to multiple variables (i.e., indicator parameters) is also discussed.