3-D MODELING OF A FORCED-GRADIENT TRACER TEST AT A RIVER BANK FILTRATION SITE IN EL PASO, TEXAS: ANALYSIS AND SIMULATION OF DISPERSION PROCESSES

Bromide and microspheres were used as tracers during a multi-well tracer test conducted at an artificial bank filtration site in El Paso, Texas. The unconfined aquifer in the study site is part of the Rio Grande alluvial aquifer, which represents a heterogeneous alluvial environment and provides the opportunity to report scale dependent dispersivities at observation distances from 2 to 18 m.

The detailed stratigraphy and hydrogeology of the site was defined and helped developing a conceptual model for the site. The geology of the aquifer was found consisting entirely of fine to coarse-grained sand forming two major aquifer units. The upper (unconfined) and lower (semi-confined) aquifer units are separated by laminated fine and very fine-grained sand comprising a semi-confining (aquitard) layer.

Bromide was introduced from an instantaneous point source in an observation well at the channel margin. Differently colored fluorescent microspheres (1 micron, 6 micron, and 10 micron) to mimic Cryptosporidium oocysts and Giardia intestinalis were injected into the stream bottom and into two observation wells. The 427-hour tracer test using bromide and fluorescent microspheres provided initial results that are relevant to the transport of pathogens through the subsurface under riverbank filtration conditions.

The bromide recovery in the pumping well and in the deeper observation wells showed early and late peaks with a long tail (positively skewed curve) indicating that the geological medium at the field site may behaved like a double-porosity medium allowing the tracer to move relatively quickly through the higher conductivity units while being retarded in the low hydraulic conductivity units.

A complete analysis of the tracer test is available, which includes; analysis of breakthrough curves, tracking of plume spreading using two-dimensional geostatistics (interpolation) performed in GMS using the 2D Scatter Point module. Dispersion coefficients are estimated as part of a three-dimensional model calibration, i.e., as fitting parameters needed to match observed contaminant concentrations with concentrations predicted by the model. Transport of the sorbing solute in a three-dimensional steady and uniform flow field was modeled using GMS.