Migration behavior of storm waters within water-saturated porous media - Laboratory experiments and reactive transport simulations

Tracer tests are routinely employed to determine the transport characteristics of aquifers. However, the value of any tracer test for identifying transport parameters strongly relies on the selected tracer and the employed type of tracer test. Many of the commonly applied tracer substances can be subject to various detrimental impacts that can eventually cause a non-conservative transport behavior and therefore add bias to the tracer test interpretation.

This study assesses the applicability of a-priori sampled liquid precipitation, i.e., rain and snowmelt, as a reliable and cost-effective groundwater tracer. For this, we investigate the migration behavior of the two main tracer components, i.e., the stable isotope signatures (δ²H and δ¹⁸O) as well as the electrical conductivity (EC). While environmental tracer studies often use the same tracer components, their usefulness is often impacted by the lack of precise knowledge of the tracer input function. The active and well-controlled injection of the tracer, however, provides this crucial information directly and therefore reduces bias and uncertainty.

The validation of this 'rain tracer' approach was performed exemplarily for a fluviatile aquifer in Pirna, Germany. The investigation included three key parts: Firstly, a series of laboratory-scale experiments was performed comparing the migration behavior of natural and synthetic storm waters to NaCl, KBr and uranine. The setup included column experiments as well as a 24 h batch study, each performed for sediments from sand to silty sand representing the lithology at the study site. Secondly, reactive 1D transport modeling (PhreeqC) was used to analyze possible modifications of the EC value as a result of, among others, ion exchange and mineral reactions. Here, the character of storm waters was mimicked by employing waters with very low ionic strength. Finally, field-scale model simulations were performed to underpin possible tracer test designs. For this, injection volumes and initial isotopic shifts required for the conduction of three selected test scenarios (push-drift-pull tracer test, natural gradient tracer test and two-well tracer test) were estimated using a MODFLOW/MT3D-MS model simulation. Here, observable breakthrough signatures and the associated accuracy were used as objective function.