Article pubs.acs.org/est
Groundwater Ages and Mixing in the Piceance Basin Natural Gas Province, Colorado Peter B. McMahon,*,† Judith C. Thomas,‡ and Andrew G. Hunt§ †
Colorado Water Science Center, U.S. Geological Survey, Denver Federal Center, Bldg. 53, Mail Stop 415, Lakewood, Colorado 80225, United States ‡ Colorado Water Science Center, U.S. Geological Survey, 764 Horizon Drive, Grand Junction, Colorado 81506, United States § Crustal Geophysics and Geochemistry Science Center, U.S. Geological Survey, Denver Federal Center, Box 25046, Mail Stop 964, Denver, Colorado 80225, United States S Supporting Information *
ABSTRACT: Reliably identifying the effects of energy development on groundwater quality can be difficult because baseline assessments of water quality completed before the onset of energy development are rare and because interactions between hydrocarbon reservoirs and aquifers can be complex, involving both natural and human processes. Groundwater age and mixing data can strengthen interpretations of monitoring data from those areas by providing better understanding of the groundwater flow systems. Chemical, isotopic, and age tracers were used to characterize groundwater ages and mixing with deeper saline water in three areas of the Piceance Basin natural gas province. The data revealed a complex array of groundwater ages (50,000 years) and mixing patterns in the basin that helped explain concentrations and sources of methane in groundwater. Age and mixing data also can strengthen the design of monitoring programs by providing information on time scales at which water quality changes in aquifers might be expected to occur. This information could be used to establish maximum allowable distances of monitoring wells from energy development activity and the appropriate duration of monitoring.
1. INTRODUCTION Advances in horizontal drilling and hydraulic fracturing technologies have enabled oil and gas development in the United States to rapidly expand into new areas. This trend has increased concerns at local, state, and federal levels about potential degradation of groundwater quality related to this industry. Assessments of groundwater quality in aquifers overlying hydrocarbon reservoirs have been done in many locations.1−5 Nevertheless, reliably identifying the effects of energy development on groundwater quality remains difficult because baseline assessments of water quality completed before the onset of energy development are rare and because interactions between hydrocarbon reservoirs and aquifers can be complex, involving both natural and human processes.5−10 Moreover, groundwater monitoring in energy basins often relies on monitoring points of opportunity (preexisting wells and springs) that may not be located close to energy development activity. Changes in water quality related to energy development at these distant locations may not be discernible at the time scale of monitoring programs except in the most dynamic flow systems.11 Designing effective monitoring programs and correctly interpreting monitoring data requires a good understanding of groundwater flow systems. Measurements of groundwater age and mixing can greatly improve that understanding by © 2013 American Chemical Society
providing essential information on recharge, water and solute sources, flow velocity, time scales for changes in water quality, contaminant flushing times, and other flow-system characteristics.12−16 Groundwater age refers to the time since water recharged an aquifer. A few studies have used groundwater ages to understand aquifer−reservoir interactions.17−20 In this paper, chemical, isotopic, and age (3H, CFCs, SF6, 14C, and 4He) tracers are used to characterize groundwater ages and mixing in three areas of the Piceance Basin natural gas province in northwestern Colorado. Concentrations and sources of CH4 in groundwater are interpreted in the context of groundwater age and mixing.
2. METHODS 2.1. Sampling Locations. Natural gas in the basin is largely produced from tight sandstones of the Upper Cretaceous Mesaverde Group by hydraulically fracturing the sandstones.21 The Mesaverde Group is overlain, from oldest to youngest, by the Tertiary Wasatch, Green River, and Uinta Formations. Received: Revised: Accepted: Published: 13250
June 3, 2013 September 18, 2013 November 4, 2013 November 4, 2013 dx.doi.org/10.1021/es402473c | Environ. Sci. Technol. 2013, 47, 13250−13257
Environmental Science & Technology
Article
Forty wells in three areas of the basin were sampled from 2009 to 2012 for a broad suite of chemical, isotopic, and age tracers. Area 1 is located in Rio Blanco County (Figure S-1, Supporting Information). Samples were collected from 14 monitoring wells completed in the Green River and Uinta Formations (Figure S-2 and Table S-1, Supporting Information). Area 2 is located in Garfield County north of the Colorado River. Samples were collected from 10 domestic wells completed in the Wasatch Formation. Area 3 also is located in Garfield County but south of the Colorado River. Samples were collected from 16 domestic wells completed in the Wasatch Formation. Monitoring wells in Area 1 are 10% saline water. For s4 and s9, the isotopic and Cl−/ Br− data provide generally consistent information about mixing. That is not the case for s8. The δ2H and δ18O values for s8 were not particularly enriched even though the Cl−/Br− data indicate that it contained a relatively large fraction of saline water. Of the three aquifer areas studied, Area 3 appears to be the most connected to underlying geologic units on the basis of the δ2H, δ18O, and Cl−/Br− data.
The presence of 3H in groundwater from Areas 2 and 3 does not preclude the possibility that some groundwater also contained a fraction of pre-1950s recharge. The δ2H, δ18O, and Cl−/Br− data indicate that some groundwater from Area 3 mixed with deep water, which could have been relatively old. 4 He data were used to evaluate if any samples from Areas 2 and 3 contained a fraction of pre-1950s recharge. 4He concentrations in groundwater increase with residence time because of 4 He production from the α-decay of U−Th series nuclides in aquifers and sometimes also because of 4He fluxes from deeper crustal or mantle sources. 4He diffusion from mineral grains in the aquifers is expected to be small because of the age of the rocks.30 There is no evidence for a mantle component of 4He in the samples on the basis of 3He/4He ratios and noble gas concentrations (Supporting Information). Thus, 4He production in the aquifers, and possibly a 4He flux from deeper crustal sources, are considered the main sources of excess 4He in the aquifers. All but one sample from Area 2 had 4He concentrations that generally could be explained by air−water equilibration and incorporation of some excess air at the time of recharge (Figure 3; Table S-5, Supporting Information). Only three samples from Area 3 had 4He concentrations that could be explained solely by atmospheric sources. Samples from Areas 2 and 3 that contained 3H and had predominantly atmospheric sources of 4 He are interpreted as entirely post-1950s in age. This interpretation is consistent with CFC data that also indicate minimal mixing with pre-1950s water in those samples (Table S-6, Supporting Information). The remaining samples from Areas 2 and 3 had 4He concentrations that were at least 7 to 1,000 times higher than air−water equilibrium (Figure 3, Table S-5, Supporting 13252
dx.doi.org/10.1021/es402473c | Environ. Sci. Technol. 2013, 47, 13250−13257
Environmental Science & Technology
Article
Information), indicating a substantial amount of excess 4He was present in many of the samples. For the samples that contained 3 H, in situ 4He production could not account for the excess 4He if that water was entirely post-1950s in age on the basis of an 4 He production rate of 5 × 10−12 cm3 STP/g yr (Figure 3). This rate is based on aerial gamma-ray estimates of the U and Th content of rocks in the Wasatch outcrop area.31,32 Instead, these samples plot along an array of mixing lines between samples that contained 4He from atmospheric sources (post-1950s recharge) and sample s9 (Figure 3). Sample s9 had very high 4He concentrations (>4000 × 10−8 cm3 STP/g) and 10% old water, which also is generally consistent with the Cl−/Br− data. The presence of a small amount of old water in n4 was not apparent from the Cl−/Br− data and indicates the utility of tracers such as 4He in identifying mixing processes. Multiple age tracers were used to refine the groundwater age estimates. For Area 1, 14C and 4He data were used to estimate apparent ages, assuming piston flow because all but one sample were recharged prior to the 1950s. Adjusted 14C ages ranging from 6000 to 22,000 years were determined for five samples using NETPATH (Table S-3, Supporting Information).35,36 4 He ages calculated for the same five samples on the basis of an in situ 4He production rate of (4.3 ± 1.6) × 10−11 cm3 STP/g yr were in good agreement with the 14C ages (Figure S-3, Supporting Information), indicating that external 4He fluxes generally were negligible in Area 1. The 4He production rate is based on direct measurements of the U and Th contents of rock cores from Area 1 (Supporting Information). 4He ages ranging from 50,000 years were calculated for the remaining samples that could not be 14C dated (Figure S-4 and Table S-5, Supporting Information). In Areas 2 and 3, SF6, CFC, and (or) 3H/3He data were analyzed with lumped-parameter models to estimate the age of the young fraction of groundwater in the samples (Table S-6, Supporting Information).37−39 In Area 2, three datable samples had ages ranging from 8 to 22 years (Figure S-4, Supporting Information). For the remaining samples in Area 2, more precise ages could not be determined with the available data, but they are considered to be entirely post-1950s in age on the basis of the 3H and 4He data. In Area 3, the ages ranged from 22 to 64 years (Figure S-4, Supporting Information). Results from this study reveal a complex array of groundwater ages and mixing fractions in Piceance Basin aquifers that was previously unknown (Figure 4). Area 1 was dominated by pre-1950s recharge (some ages >50,000 years) that generally was not mixed with saline water from zones below the aquifers. Area 2 was dominated by unmixed post-1950s recharge. Area 3 had the most complex array of ages and mixing fractions, from
Figure 4. Characterization of groundwater samples collected from Areas 1 (circles), 2 (triangles), and 3 (squares) with respect to age and mixing with saline water. Each symbol represents one sample.
unmixed post-1950s recharge to highly mixed pre-1950s recharge that was likely to be at least several thousand years old. 3.4. Controls on Groundwater Age and Mixing. The broad range of modeled groundwater ages and mixing fractions observed in the Piceance Basin reflects important differences in well locations in the flow systems, land use, aquifer confinement, and geologic structure between the three study areas. The datable fractions of post-1950s recharge in Area 3 were significantly older than the datable fractions of post-1950s recharge in Area 2 (p = 0.01, Wilcoxon rank-sum test) (Table S-6 and Figure S-4, Supporting Information). Groundwater from Area 1 was older than groundwater from both Areas 2 and 3 (Figure S-4, Supporting Information). Mean sample depth below the potentiometric surface for the datable samples increased according to Area 2 (3.4 m) < Area 3 (9.5 m) < Area 1 (156 m), indicating that the shallowest groundwater was the youngest groundwater. Recharge in Areas 2 and 3 also is more widespread than in Area 1 because of return flow from irrigation and septic systems in those areas. Greater recharge in Areas 2 and 3 would reduce groundwater ages in those areas compared to Area 1. Aquifer confinement also is likely to have affected groundwater ages. Whereas the Wasatch Formation in Areas 2 and 3 largely consists of interbedded fluvial sandstones and mudstones and contains no laterally extensive confining layers,40 the aquifers in Area 1 are confined by regionally extensive lacustrine oil-shale layers that impede downward water movement (Figure S-2, Supporting Information),41 which can result in low recharge and old groundwater ages in those aquifers. Overall, the combination of shallower sample depths, more widespread recharge, and less aquifer confinement in Areas 2 and 3 are consistent with the presence of younger groundwater in those areas than in Area 1. The confining layers and an evaporite mineral zone at the base of the aquifer system in Area 1 (Figure S-2, Supporting Information) are likely to be effective seals against the upward movement of saline water from deeper formations and may also explain the apparent absence of a deep 4He flux in Area 1. Widespread mixing with old saline water was more evident in Area 3 than in Areas 1 and 2 (Figures 1−3). It is unlikely that the absence of an extensive confining layer in Area 3 could be the sole reason for this mixing pattern because Area 2 also lacked such a confining layer. Almost all the wells in Area 3 that contained fractions of old saline water were located in the 13253
dx.doi.org/10.1021/es402473c | Environ. Sci. Technol. 2013, 47, 13250−13257
Environmental Science & Technology
Article
vicinity of the Divide Creek anticline in the eastern part of Area 3 (Figure S-1, Supporting Information). The Wasatch Formation has been partially to completely eroded near that structure so the Mesaverde Group is closer to land surface there than in the western part of Area 3 and in most of Area 2.40 Moreover, fracturing associated with the anticline could provide fluid migration pathways.21,42 The combination of erosion and fracturing near the anticline probably accounts for much of the widespread mixing in the eastern part of Area 3. Improperly sealed boreholes in gas wells could be locally important conduits for upward fluid migration in each of the areas. 3.5. Methane Concentrations and Sources. CH4 is one of the most commonly measured constituents in groundwater monitoring programs that focus on energy development, so understanding the factors affecting CH4 concentrations and sources in groundwater is important. CH4 concentrations were highly variable in Areas 1 and 3, ranging from
