Monitoring parameters and statistics: A land treatment unit monitoring case study

Abstract

Statistical evaluation procedures for monitoring data at facilities permitted under the Resource Conservation and Recovery Act (RCRA) are frequently established before monitoring begins. Selecting the statistical method before background data have been collected often leads to the use of statistical procedures that are inappropriate for the actual monitoring data. Such was the case for unsaturated zone monitoring at a permitted land treatment unit in the Gulf Coast area of Texas. Due to the large number of “not detected” results in the background database for lysimeters, statistical evaluation procedures specified in the original RCRA permit yielded an artificially low standard deviation, resulting in background values that were strongly biased on the low side. An alternate statistical procedure based on probability plots was developed and was accepted by the state environmental regulatory agency. This technique, which has wide applicability for many types of environmental monitoring data, significantly reduced the chasing of false positives, thus saving potentially expensive investigations and remediations.