3.7 Identifying Groundwater Flow through Fractures

Environmental tracers also provide excellent tracers of groundwater flow in aquifers dominated by fractures – an area where traditional hydraulic approaches for estimating rates and patterns of groundwater flow have limitations. Groundwater age tracers are particularly useful for identifying the depths of active groundwater circulation. A good example of this application comes from a field site in southern Ontario (Figure 36), where clay-rich glacial till deposits extend from within a meter of the ground surface to depths of several tens of meters. Fractures have been observed to depths of more than 9 m in these sediments, but whether these fractures contribute to significant groundwater flow is difficult to determine. An investigation into flow through these fractured tills involved installing piezometers to 16-m depth at 14 sites, and sampling groundwater from the piezometers for analysis of 3H. Results from four of these sites are presented in Figure 37. All sites show high concentrations of 3H in the upper few meters and concentrations decreasing rapidly with depth. Above-background concentrations of tritium are observed to a depth of 7.5 to 12 m. The authors calculated that 3H could move 1 to 2 m by diffusion, and so concluded that active groundwater flow occurred to a depth of 5 to 10 m. Similar depths were suggested by mapping of fractures, and the maximum depths at which seasonal variations in hydraulic head were observed (Ruland et al., 1991).

Map showing location of field site in southern Ontario, Canada
Figure 36 – Location of field site in southern Ontario, Canada, where 3H profiles were measured to determine depth of groundwater flow (white circle). Figure 37 for 3H depth profiles.
Profiles of <sup>3</sup>H obtained from four sites in southern Ontario
Figure 37 – Profiles of 3H obtained from four sites in southern Ontario. High 3H concentrations indicate active groundwater flow, which extends to depths of approximately 5 to 10 m (After Ruland et al., 1991).

36Cl has more sensitivity than 3H as a tracer of modern recharge, in part because its signal has not been significantly attenuated by radioactive decay since the period of nuclear fallout, and this makes it an extremely useful tracer for detecting recent rainfall in groundwater. Investigations into the potential for fracture flow at the proposed United States high-level nuclear waste repository at Yucca Mountain, Nevada (Figure 38), relied heavily on results from 36Cl testing (Fabryka-Martin et al., 1997; Wolfsberg et al., 1999). An 8km long, nearly horizontal tunnel was drilled in a loop beneath Yucca Mountain, at the same depth and in the same formation as the proposed repository. The tunnel formed the basis of several tests on fluid flow at the site, including collection of rock samples from the tunnel walls, and leaching of salts dissolved in pore water from the rocks for 36Cl analysis. Samples were collected at regular intervals of approximately 200 m, and collected from selected features where preferential flow was considered to be more likely, such as faults and fractures.

Map showing location of Yucca Mountain, Nevada, USA
Figure 38 – Location of Yucca Mountain, Nevada, USA. As a proposed storage facility for high-level radioactive waste, Yucca Mountain has been the site of several isotope studies to define rates of infiltration leading to aquifer recharge (Figure 39), and rates of groundwater flow.

The natural 36Cl/Cl ratio of modern precipitation at the site is approximately 500 × 1015, but this value has varied historically, and was as high as 1,200 × 1015 approximately 12,000 years ago. However, during the 1950s and 1960s, high concentrations of 36Cl were added to rainfall from atomic testing of nuclear devices (Figure 6). The peak 36Cl/Cl value may have been as high as 200,000 × 1015, although this signal is diluted by mixing with pre-fallout chloride in the soil. Thus, samples with 36Cl/Cl ratios between approximately 500 and 1200 × 1015 probably reflect pre-fallout precipitation, indicative of slow percolation of water through the rock mass, with travel times of several thousand years. Values in excess of 1200 × 1015 are interpreted as reflecting the input of 36Cl from nuclear weapons testing. Whereas samples collected at regular intervals along the tunnel did not yield values greatly in excess of pre-1950s values, many samples collected close to potential conduits (fractures, faults, breccia and unit contacts) yielded high 36Cl/Cl values (Figure 39). This suggests higher water velocities (of at least several meters per year) through such features (Fabryka-Martin et al., 1997; Campbell et al., 2003). Subsequent validation studies yielded mixed results, with some 3H and 36Cl/Cl ratios indicative of post-1950s water, but 36Cl/Cl values significantly lower than those reported by Fabryka-Martin et al. in 1997. The validation studies also failed to replicate the earlier elevated values measured in the vicinity of the Sundance Fault (Figure 39). However, this does not necessarily invalidate the earlier results. Water flowing rapidly through fractures within the unsaturated zone is unlikely to be detected large distances from individual fractures. Thus, repeat sampling is unlikely to yield identical results, particularly if it occurs after the initial tunneling when the unsaturated zone moisture may have redistributed itself in response to the new conditions created by the presence of the tunnel (Marshall et al., 2012).

Chlorine-36 ratios of salts leached from rock samples from a section of the Exploratory Studies Facility at Yucca Mountain, Nevada.
Figure 39 – Chlorine-36 ratios of salts leached from rock samples from a section of the Exploratory Studies Facility at Yucca Mountain, Nevada. The shaded area shows the approximate range of 36Cl/Cl values in rainfall prior to the 1950s, and the vertical lines denote the locations of faults that have been mapped at the surface. Feature-based samples were taken from fractures, faults, breccia and unit contacts. From Fabryka-Martin et al. (1997), as reproduced in Philips (2000), and Wolfsberg et al. (1999).