When developing a site-specific ground-water sampling program, it is critical to have an accurate, three-dimensional understanding of the ground-water hydrology of the site under investigation. Investigators must understand how ground water will move across any given site and must identify what factors influence that movement, such as pumping centers, areas of artificial recharge, water-level fluctuations in an adjacent surface/water system, or tides. Samplers must also develop an understanding of how ground water behaves within the sampling point or monitoring well, both between sampling events and during purging and sample collection. Failure to develop this understanding typically results in selection and use of sampling protocols that are not well suited to the hydrogeology or geochemistry of the site. This, in turn, can result in generation of nonrepresentative data.
As illustrated in Figure 1, ground water moves through the subsurface in response to differences in hydraulic head, under laminar flow conditions in most hydrogeologic systems. Provided that the difference in hydraulic head is such that horizontal flow dominates in the formation, flow should continue in the same manner through well screens installed in the formation. Robin and Gillham (1987) demonstrated with tracer solutions that formation water moves through the well screen and that this water does not mix with the stagnant water that remains in the casing between sampling rounds.
Visual observations of movement of colloidal particles through well screens installed in granular aquifers made by Kearl et al. (1992) demonstrate that horizontal laminar flow in a formation does, in fact, continue through well screens and that water in the screen does not mix with that in the casing. Powell and Puls (1993), using dual tracer tests, also showed that ground water moves through the well screen with little interaction or mixing with the overlying stagnant water in storage in the well casing. Electrical conductivity data presented by Michalski (1989) clearly showed a fresh water zone in the well screen separate from the stagnant water in the casing, further supporting the observations made by Robin and Gillham. Robin and Gillham (1987) also theorized that the continual flow of water through the screen allows chemical reactions, such as desorption and adsorption, between well construction materials (well screen and filter pack) and constituents in ground water to approach equilibrium. Work done by Palmer et al. (1987) corroborates this and demonstrates the need to allow time for this equilibration to occur prior to the initial sampling event for a well, which is also suggested by Walker (1983).
These studies suggest that for wells in which horizontal flow dominates: (1) water in the screen at any point in time is indeed representative of water in the formation adjacent to the screen; (2) water samples taken directly from the screened interval are representative of ground water in the surrounding formation; and (3) provided that samplers can gain access to the water in the screened interval while minimally disturbing the water column in the well, purging multiple well volumes of water prior to sample collection is unnecessary.
However, where there is a difference in hydraulic head in the formation that results in vertical movement of ground water, and a well is screened across that zone, the well screen effectively acts to short-circuit ground-water flow and directly channels water from the zone with highest hydraulic head to the zone with lowest hydraulic head (Figure 1.2).
Several recent studies (i.e., McIlvride and Rector, 1988; Reilly et al., 1989; Church and Granato, 1996; Hutchins and Acree, 2000; Elci et al., 2001, 2003) have documented that, in areas with vertical hydraulic gradients, installation of a monitoring well with a long well screen (i.e.,10 ft long) may set up a localized vertical flow system that renders the well almost useless for sampling because the dilution that occurs in such a well would yield misleading and ambiguous data concerning contaminant concentrations and plume geometry. In some scenarios, installation of a well in this type of setting may result in the spread of contamination to parts of a formation that would not otherwise have become contaminated had the well not been installed.
Thus, samplers must be keenly aware of the hydraulic conditions that exist in a well so they can make appropriate decisions on whether a well should be sampled, and, if it is sampled, how to interpret the data generated by each sampling event. More detailed information on this topic is available in Nielsen and Schalla (2006) and Einarson (2006).
