Stochastic capture zone analysis of an arsenic-contaminated well using the generalized likelihood uncertainty estimator (GLUE) methodology

• Published in: Water resources research 2003-06, Vol.39 (6), p.n/a
• Main Authors: Morse, Brad S., Pohll, Greg, Huntington, Justin, Rodriguez Castillo, Ramiro
• Format: Article
• Language: English
• Subjects: Bayesian; capture zone; GLUE; Mexico; Zimapan
• ISSN: 0043-1397; 1944-7973
• DOI: 10.1029/2002WR001470

In 1992, Mexican researchers discovered concentrations of arsenic in excess of World Heath Organization (WHO) standards in several municipal wells in the Zimapan Valley of Mexico. This study describes a method to delineate a capture zone for one of the most highly contaminated wells to aid in future well siting. A stochastic approach was used to model the capture zone because of the high level of uncertainty in several input parameters. Two stochastic techniques were performed and compared: “standard” Monte Carlo analysis and the generalized likelihood uncertainty estimator (GLUE) methodology. The GLUE procedure differs from standard Monte Carlo analysis in that it incorporates a goodness of fit (termed a likelihood measure) in evaluating the model. This allows for more information (in this case, head data) to be used in the uncertainty analysis, resulting in smaller prediction uncertainty. Two likelihood measures are tested in this study to determine which are in better agreement with the observed heads. While the standard Monte Carlo approach does not aid in parameter estimation, the GLUE methodology indicates best fit models when hydraulic conductivity is approximately 10−6.5 m/s, with vertically isotropic conditions and large quantities of interbasin flow entering the basin. Probabilistic isochrones (capture zone boundaries) are then presented, and as predicted, the GLUE‐derived capture zones are significantly smaller in area than those from the standard Monte Carlo approach.