Analysis of censored data in groundwater monitoring wells at the Savannah River Site

It is common in environmental analyses to deal with censored data. Censored data characteristically arise through laboratory analysis of samples with contaminant concentrations less than what the analytical method is able to reliably detect. These data are called ''less than detectable.'' Comparisons between downgradient or monitoring groundwater wells and upgradient or background wells are frequently done to determine if downgradient wells are more contaminated than background or some established maximum concentration limits (MCL's). In addition, parameter estimates are often desired. The presence of censored data complicates the statistics that can be used as estimators for individual populations or to estimate differences between two populations. This paper describes the current process at Savannah River Site (SRS) to determine constituents of concern (COC's) for complying with groundwater monitoring and clean-up regulations. COC's are analytes found in downgradient monitoring wells in concentrations significantly greater than in background wells or significantly greater than the MCL'S. Both parametric and non-parametric statistics are explored. Data plots are examined for outliers, trends, laboratory or sampling contamination, and unusually large detection limits for censored results. Wells are grouped by similar concentration levels to form a ''characteristic'' well, improving the estimation and decision process