Project

Discipline

Pollution source identification; Groundwater contamination modelling; Backward transport; Location probability; aquifer contamination; numerical simulation

Lead
Lay summary
An important issue in any remediation process of polluted groundwater is the assessment of groundwater contamination. This requires not only understanding of the dynamics of the aquifer system in which the contamination is propagating but also identification of unknown contamination sources and release histories that led to pollution. This is important both for the design of adequate remedial measures and to assign responsibilities.Many mathematical methods have been developed to deal with the identification of unknown pollution sources and contamination times, but all of them are subject to significant drawbacks and limitations, such as parameter homogeneity, simple flow geometries, high sensitivity to noise, previous knowledge of potential source locations and/or contamination time and history, or excessive computational effort.Recently, Milnes & Perrochet (2007) and Perrochet & Milnes (2008) developed an original simulation procedure based on the transfer function theory (e.g., Jury & Roth 1990) to identify both the unknown point-source location and contamination time with a single reversed flow field simulation. Although this new approach is seductive due to its simplicity, it is still bound to severe restrictions, such as perfectly known flow field conditions, conservative contaminants and known initial contaminant mass input. The present research proposal aims at extending and adapting the developed theoretical framework to alleviate these restrictive assumptions and to render the approach applicable to realistic contamination configurations. For that reason the research plan is subdivided into a theoretical and a practical work package, which partially overlap. The focus will first be put on theoretical developments to simulate cases where only few observation points are known and the contamination mass is unknown, where several point-sources pollution plumes are superimposed, where the contaminant is reactive and the flow field conditions are uncertain. Then, to evaluate and test the applicability of the developed simulation procedures, cross-validation will be done on real case sites: i) on historical data from the well-documented natural gradient tracer test in the Borden sand aquifer in Canada (Sudicky 1986), ii) on the Zürich-Affoltern gasoil accident in 1994 which has been monitored over the past decade (Munz et al. 2005), and iii) on specifically designed tracer tests that will be carried out in the Seeland test site in Switzerland.

Direct link to Lay Summary

Last update: 21.02.2013

Number	Title	Start	Funding scheme

An important issue in any remediation process of polluted groundwater is the assessment of groundwater contamination. This requires not only understanding of the dynamics of the aquifer system in which the contamination is propagating but also identification of unknown contamination sources and release histories that led to pollution. This is important both for the design of adequate remedial measures and to assign responsibilities.Many mathematical methods have been developed to deal with the identification of unknown pollution sources and contamination times, but all of them are subject to significant drawbacks and limitations, such as parameter homogeneity, simple flow geometries, high sensitivity to noise, previous knowledge of potential source locations and/or contamination time and history, or excessive computational effort. Recently, Milnes & Perrochet (2007) (see attached paper) developed an original simulation procedure based on the transfer function theory to identify both the unknown point-source location and contamination time with a single reversed flow field simulation. Although this new approach is seductive due to its simplicity, it is still bound to severe restrictions, such as perfectly known flow field conditions, conservative contaminants and known initial contaminant mass input.The present research proposal aims at extending and adapting the developed theoretical framework to alleviate these restrictive assumptions and to render the approach applicable to realistic contamination configurations. For that reason the research plan is subdivided into a theoretical and a practical work package, which partially overlap.The focus will first be put on theoretical developments to simulate cases where only few observation points are known and the contamination mass is unknown, where several point-source pollution plumes are superimposed, where the contaminant is reactive and the flow field conditions are uncertain. Then, to evaluate and test the applicability of the developed simulation procedures, cross-validation will be done on real case sites: i) on historical data from the well-documented natural gradient tracer test in the Borden sand aquifer in Ontario, ii) on the Zürich-Affoltern gasoil accident in 1994 which has been monitored over the past decade, and iii) on specifically designed tracer tests that will be carried out in the Seeland test site in Switzerland.