|
|||||
|
|
Ground Water Monitoring Statistics Update: Part I: Progress Since 1988 |
|
| Authors: | Charles B. Davis Roger J. McNichols |
| Affiliation: | Charles B. Davis;is principal statistician with Environmet- rics &Statistics Ltd. (EnviroStat, 1853 Wellington Court, Henderson, NV 89014). After receiving his M.S. in mathematics and statistics and Ph.D. in statistics from the University of New Mexico, he joined the Mathematics Department of the University of Toledo to establish its graduate program in statistics with emphasis on consulting and applications. He and McNichols became involved in consulting and research related to statistical issues arising in RCRA ground water monitoring regulation in 1985 and formed EnviroStat in 1990. Davis left academia in 1992 to concentrate on environmental statistics. Roger J. McNichols;is professor and chairman of the Industrial Engineering Department at The University of Toledo, (Toledo, OH 43606). After receiving his Ph.D. in industrial engineering from The Ohio State University, he joined the faculty of Texas A and M University, where he directed the Maintainability Engineering Graduate Program at Red River Army Depot. At University of Toledo he is also chairman of the Systems Doctoral Program and has served as associate dean of engineering. His research and consulting interests include reliability, quality control, manufacturing, environmental monitoring, mathematical modeling, and applied statistics. |
| Abstract: | Current federal ground water monitoring statistical regulation dates from the revised RCRA Subtitle C Final Rule of 1988. That rule was a considerable advance over previous RCRA statistical rules. However, two major problem areas remained: facility-wide false positive rate (FWFPR) control and spatial variability. Progress has been made in the 1991 Subtitle D Final Rule and in guidance: the 1992 Addendum to Interim Final Guidance in particular includes a substantial conceptual advance toward resolving the FWFPR problem. Other areas of improvement include normality testing and distribution assumptions, dropping the four independent samples per monitoring period requirement, allowing a preliminary evaluation short of a 40 CFR Part 258 Appendix II assessment upon finding a statistically significant increase, and suggesting superior alternatives to analyses of variance (ANOVAs) and tests of proportions. The problem of dealing with natural spatial variability remains. Although certain techniques listed in the regulations can control for inherent spatial variability and the performance standards require doing so "when necessary," little attention has been paid to the ubiquity of such spatial variation. Moreover, regulatory traditions favoring upgradient-downgradient comparisons often make control of natural spatial variation difficult and ineffective. With new. lined facilities easily implemented statistical solutions are available; however, dealing with the several existing solid waste facilities which will now be regulated under Subtitle D will present major challenges. In short, the 1988 revision of the Subtitle C rules made it more possible to provide statistically sound monitoring programs, and there has been steady progress since then. Challenges remain, however. These vary from slate to slate, particularly with regard to controlling false positives and false negatives in the presence of natural spatial variability. |
| Keywords: | |