The recovering of the contaminant release history in heterogeneous and partially known flow field

Abstract:

The recovering of the release history of a pollutant in an aquifer is a subject that obtains the researchers attention since 1990 because the knowledge of the pollutant discharge history can be a useful tool to share, among the responsible parties, the costs of the remediation actions. The methods developed are based on the perfect knowledge of the flow field where the pollutant event has occurred. But very often this perfect knowledge is only an unattainable hope: few boreholes and pumping tests are usually available and realizing new ones can lead to unacceptable costs.

An evaluation of the reliability of the results obtained with the procedures for the recovering of the release history in heterogeneous and partially known flow field can be a very useful subject under the point of view of researchers and the practitioners. In fact the different methods can be more or less sensitive to the problem or give information that includes or not an uncertainty evaluation; moreover it can be possible to quantify the adequate amount of the site characterization actions.

In this work we deal with a geostatistically based approach that in the past we have successfully applied on a perfect-known transmissivity field. The objective of the actual study is to investigate the importance of the knowledge of the hydraulic conductivity field and to identifying the minimum information required for recovering an acceptable release history. At this aim we built a numerical model of a 2-D confined aquifer with rectangular shape characterized by a heterogeneous hydraulic conductivity field. We estimate the conductivity field through a kriging interpolation starting from values sampled on a regular grids of different density and on that partially known field we estimate the pollutant release history. The results show that the reliability of the recovered release history, of course, increases with the density of the grid but meaningful indications can be obtained also with not very detailed discretization.