THE ROLE OF FIELD EXPERIMENTS IN ELUCIDATION OF DISPERSION IN SANDY AQUIFERS

Dispersion in groundwater first attracted attention in the 1960s, when the transport of radionuclides became a radioactive waste disposal issue. At that time, small dispersivity values from laboratory tests were used in analytic models of contaminant migration. This seemed satisfactory until the onset of numerical modeling. In the 1970s, 2-D numerical models fitted to apparent contaminant distributions from monitoring wells gave very large dispersivity values, both longitudinal and transverse, which were orders of magnitude above lab values. These apparent dispersivities resulted from sparse field data and numerical dispersion. With the benefit of hindsight, most of these modeling results were nonsensical. These model results, which were generally attributed to heterogeneity, went uncontested and more than a decade passed before the “very large dispersivity” view was assessed by appropriate field experiments involving natural groundwater flow. The experiments were simple in concept and design, although tedious to conduct. A solute tracer “slug” comprised of dissolved bromide or chloride was injected in the aquifer during several hours, followed by many months of detailed three-dimensional monitoring of transport by natural flow. The first few experiments were done at the Borden site, Ontario in the late 1970s and early 1980s and similar experiments were conducted elsewhere, most notably the Cape Cod site and another site in Denmark. These experiments, along with detailed three-dimensional monitoring of actual contaminant plumes in sand aquifers, showed that, although heterogeneities have an important influence, dispersion in the longitudinal direction and particularly in the transverse directions is much smaller, by orders of magnitude, than expected. The misleading modeling efforts could have been avoided if the appropriate field experiments had been done much earlier. The only road-block was our mind-set. Although dispersion in sand is now understood, little is known about dispersion at the field scale in fractured rock because appropriate experiments at the relevant field scales and detailed plume monitoring have not yet been conducted.