Application of inverse methods to contaminant source identification from aquitard diffusion profiles at Dover AFB, Delaware

Abstract

Abstract. This paper presents a refinement and expansion of our previously described efforts to estimate contaminant plume history from observed contaminant concentrations within a low-permeability aquitard at the site of a field-scale groundwater remediation experiment at Dover Air Force Base. At this site, a two-layer aquitard has been contaminated with tetrachloroethene and trichloroethene through diffusive mass transfer from an overlying contaminated aquifer. Measurements of contaminant concentrations in the aquitard are used, together with independently obtained information about the sorption and diffusion properties of the aquitard medium, to estimate the timewise variation of the boundary concentration at the interface between the aquitard and aquifer, thus providing evidence related to the overlying plume history. In our refined analysis, we assume the contaminant source history to be a function of time with unknown form, and we supplement our interpretations with a second coring result at a different location. The results demonstrate how "forensic" interpretations of this kind can provide useful and important information regarding the contaminant release history at sites of groundwater contamination and cleanup; however, the results also show that the forensic problem is a highly nonunique problem associated with potentially large uncertainty. Interpretation of the estimated results therefore requires careful consideration in the context of other available information.

Inverse modeling for locating dense nonaqueous pools in groundwater under steady flow conditions

“…Equipped with an increasing body of knowledge regarding contaminant fate and transport processes in groundwater, researchers are beginning to focus on realistic inverse modeling efforts aimed at reconstructing source information [e.g., Liu and Ball, 1999]. In this paper we solve the inverse problem to identify the location of a dissolving DNAPL pool on the bottom of a three-dimensional porous medium.…”

Source tracking of microbial intrusion in water systems using artificial neural networks

“…By contrast, the inverse problem is intended to find the unknown input variables that give rise to a partially known output state (Liu and William, 1999). In the source identification of a contaminant, the injection locations are unknown.…”

A hybrid regularized inversion methodology for highly parameterized environmental models

“…Though commonly employed for spatially varying parameters Tikhonov regularization has been employed in the reconstruction of a time-varying contaminant source [Liu and Ball, 1999]. Tikhonov relationships can be linear or nonlinear, and are often designed to yield smooth parameter fields.…”

“…This takes the form of a series of regularization equations, the weights for which are determined during the inverse process. Regularization equations are typically based upon differencing schemes, such as the difference of a parameter value from a preferred value, or the difference in value between parameters [Ory and Pratt, 1995;Liu and Ball, 1999]. Example regularization equations are described by Doherty [2003] and may be similar in form to prior information equations described by Cooley [1982].…”

“…Tikhonov relationships can be linear or nonlinear, and are often designed to yield smooth parameter fields. Smoothness is not always desirable [Vogel, 1997;Portniaguine, 1999] but has been demonstrated to produce numerical stability and conceptually agreeable results in groundwater models [Liu and Ball, 1999;Doherty, 2003].…”