Duke Study: Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing

Methane contamination of drinking water
accompanying gas-well drilling and
hydraulic fracturing
Stephen G. Osborna, Avner Vengoshb, Nathaniel R. Warnerb, and Robert B. Jacksona,b,c,1
a
Center on Global Change, Nicholas School of the Environment, bDivision of Earth and Ocean Sciences, Nicholas School of the Environment, and
c
Biology Department, Duke University, Durham, NC 27708

Edited* by William H. Schlesinger, Cary Institute of Ecosystem Studies, Millbrook, NY, and approved April 14, 2011 (received for review January 13, 2011)

Directional drilling and hydraulic-fracturing technologies are dra-
matically increasing natural-gas extraction. In aquifers overlying
the Marcellus and Utica shale formations of northeastern Pennsyl-
vania and upstate New York, we document systematic evidence for
methane contamination of drinking water associated with shale-
gas extraction. In active gas-extraction areas (one or more gas
wells within 1 km), average and maximum methane concentrations
in drinking-water wells increased with proximity to the nearest
gas well and were 19.2 and 64 mg CH4 L−1 (n ¼ 26), a potential
explosion hazard; in contrast, dissolved methane samples in neigh-

ENVIRONMENTAL
boring nonextraction sites (no gas wells within 1 km) within similar

SCIENCES
geologic formations and hydrogeologic regimes averaged only
1.1 mg L−1 (P < 0.05; n ¼ 34). Average δ13 C-CH4 values of dissolved
methane in shallow groundwater were significantly less negative
for active than for nonactive sites (−37 Æ 7‰ and −54 Æ 11‰,
respectively; P < 0.0001). These δ13 C-CH4 data, coupled with the ra-
tios of methane-to-higher-chain hydrocarbons, and δ2 H-CH4 values,
are consistent with deeper thermogenic methane sources such as
the Marcellus and Utica shales at the active sites and matched gas
geochemistry from gas wells nearby. In contrast, lower-concentra-
tion samples from shallow groundwater at nonactive sites had
isotopic signatures reflecting a more biogenic or mixed biogenic/
thermogenic methane source. We found no evidence for contam- Fig. 1. Map of drilling operations and well-water sampling locations in
ination of drinking-water samples with deep saline brines or frac- Pennsylvania and New York. The star represents the location of Binghamton,
turing fluids. We conclude that greater stewardship, data, and— New York. (Inset) A close-up in Susquehanna County, Pennsylvania, showing
possibly—regulation are needed to ensure the sustainable future areas of active (closed circles) or nonactive (open triangles) extraction. A
of shale-gas extraction and to improve public confidence in its use. drinking-water well is classified as being in an active extraction area if a
gas well is within 1 km (see Methods). Note that drilling has already spread
groundwater ∣ organic-rich shale ∣ isotopes ∣ formation waters ∣ to the area around Brooklyn, Pennsylvania, primarily a nonactive location at
water chemistry the time of our sampling (see inset). The stars in the inset represent the towns
of Dimock, Brooklyn, and Montrose, Pennsylvania.

I ncreases in natural-gas extraction are being driven by rising
energy demands, mandates for cleaner burning fuels, and the
economics of energy use (1–5). Directional drilling and hydrau-
Concerns for impacts to groundwater resources are based on
(i) fluid (water and gas) flow and discharge to shallow aquifers
due to the high pressure of the injected fracturing fluids in the
lic-fracturing technologies are allowing expanded natural-gas gas wells (10); (ii) the toxicity and radioactivity of produced water
extraction from organic-rich shales in the United States and else- from a mixture of fracturing fluids and deep saline formation
where (2, 3). Accompanying the benefits of such extraction (6, 7)
waters that may discharge to the environment (11); (iii) the
are public concerns about drinking-water contamination from
potential explosion and asphyxiation hazard of natural gas; and
drilling and hydraulic fracturing that are ubiquitous but lack a
(iv) the large number of private wells in rural areas that rely on
strong scientific foundation. In this paper, we evaluate the poten-
shallow groundwater for household and agricultural use—up to
tial impacts associated with gas-well drilling and fracturing on
one million wells in Pennsylvania alone—that are typically unre-
shallow groundwater systems of the Catskill and Lockhaven
formations that overlie the Marcellus Shale in Pennsylvania and gulated and untested (8, 9, 12). In this study, we analyzed ground-
the Genesee Group that overlies the Utica Shale in New York water from 68 private water wells from 36- to 190-m deep in
(Figs. 1 and 2 and Fig. S1). Our results show evidence for
methane contamination of shallow drinking-water systems in at Author contributions: S.G.O., A.V., and R.B.J. designed research; S.G.O. and N.R.W.
least three areas of the region and suggest important environmen- performed research; A.V. contributed new reagents/analytic tools; S.G.O., A.V., N.R.W.,
tal risks accompanying shale-gas exploration worldwide. and R.B.J. analyzed data; and S.G.O., A.V., N.R.W., and R.B.J. wrote the paper.

The drilling of organic-rich shales, typically of Upper Devo- The authors declare no conflict of interest.
nian to Ordovician age, in Pennsylvania, New York, and else- *This Direct Submission article had a prearranged editor.
where in the Appalachian Basin is spreading rapidly, raising Freely available online through the PNAS open access option.
concerns for impacts on water resources (8, 9). In Susquehanna 1
To whom correspondence should be addressed. E-mail: jackson@duke.edu.
County, Pennsylvania alone, approved gas-well permits in the This article contains supporting information online at www.pnas.org/lookup/suppl/
Marcellus formation increased 27-fold from 2007 to 2009 (10). doi:10.1073/pnas.1100682108/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1100682108 PNAS Early Edition ∣ 1 of 5

Fig. 2. Geologic cross-section of Bradford and western Susquehanna Coun- Fig. 3. Methane concentrations (milligrams of CH4 L−1 ) as a function of dis-
ties created from gas-well log data provided by the Pennsylvania Department tance to the nearest gas well from active (closed circles) and nonactive (open
of Conservation and Natural Resources. The approximate location of the Law- triangles) drilling areas. Note that the distance estimate is an upper limit and
renceville-Attica Lineament is taken from Alexander et al. (34). The Ordovician does not take into account the direction or extent of horizontal drilling un-
Utica organic-rich shale (not depicted in the figure) underlies the Middle derground, which would decrease the estimated distances to some extraction
Devonian Marcellus at approximately 3,500 m below the ground surface. activities. The precise locations of natural-gas wells were obtained from the
Pennsylvania Department of Environmental Protection and Pennsylvania
northeast Pennsylvania (Catskill and Lockhaven formations) and Spatial Data Access databases (ref. 35; accessed Sept. 24, 2010).
upstate New York (Genesee formation) (see Figs. 1 and 2 and SI
Text), including measurements of dissolved salts, water isotopes The average δ13 C-CH4 value in shallow groundwater in active
(18 O and 2 H), and isotopes of dissolved constituents (carbon, drilling areas was −37 Æ 7‰, consistent with a deeper thermo-
boron, and radium). Of the 68 wells, 60 were also analyzed for genic methane source. In contrast, groundwater from nonactive
dissolved-gas concentrations of methane and higher-chain hydro- areas in the same aquifers had much lower methane concentra-
carbons and for carbon and hydrogen isotope ratios of methane. tions and significantly lower δ13 C-CH4 values (average of −54Æ
Although dissolved methane in drinking water is not currently 11‰; P < 0.0001; Fig. 4 and Table 1). Both our δ13 C-CH4 data
classified as a health hazard for ingestion, it is an asphyxiant in and δ2 H-CH4 data (see Fig. S2) are consistent with a deeper ther-
enclosed spaces and an explosion and fire hazard (8). This study mogenic methane source at the active sites and a more biogenic
seeks to evaluate the potential impact of gas drilling and hydrau- or mixed methane source for the lower-concentration samples
lic fracturing on shallow groundwater quality by comparing areas
from nonactive sites (based on the definition of Schoell, ref. 14).
that are currently exploited for gas (defined as active—one or Because ethane and propane are generally not coproduced
more gas wells within 1 km) to those that are not currently asso-
during microbial methanogenesis, the presence of higher-chain
ciated with gas drilling (nonactive; no gas wells within 1 km),
hydrocarbons at relatively low methane-to-ethane ratios (less
many of which are slated for drilling in the near future.
than approximately 100) is often used as another indicator of
Results and Discussion deeper thermogenic gas (14, 15). Ethane and other higher-chain
Methane concentrations were detected generally in 51 of 60 hydrocarbons were detected in only 3 of 34 drinking-water wells
drinking-water wells (85%) across the region, regardless of gas from nonactive drilling sites. In contrast, ethane was detected in
industry operations, but concentrations were substantially higher 21 of 26 drinking-water wells in active drilling sites. Additionally,
closer to natural-gas wells (Fig. 3). Methane concentrations propane and butane were detected (>0.001 mol %) in eight and
were 17-times higher on average (19.2 mg CH4 L−1 ) in shallow two well samples, respectively, from active drilling areas but in no
wells from active drilling and extraction areas than in wells from wells from nonactive areas.
nonactive areas (1.1 mg L−1 on average; P < 0.05; Fig. 3 and Further evidence for the difference between methane from
Table 1). The average methane concentration in shallow groundwater wells near active drilling sites and neighboring nonactive
water in active drilling areas fell within the defined action level sites is the relationship of methane concentration to δ13 C-CH4
(10–28 mg L−1 ) for hazard mitigation recommended by the US values (Fig. 4A) and the ratios of methane to higher-chain hydro-
Office of the Interior (13), and our maximum observed value of
64 mg L−1 is well above this hazard level (Fig. 3). Understanding
Table 1. Mean values Æ standard deviation of methane
the origin of this methane, whether it is shallower biogenic or
concentrations (as milligrams of CH4 L−1 ) and carbon isotope
deeper thermogenic gas, is therefore important for identifying
composition in methane in shallow groundwater δ13 C-CH4 sorted
the source of contamination in shallow groundwater systems.
by aquifers and proximity to gas wells (active vs. nonactive)
The δ13 C-CH4 and δ2 H-CH4 values and the ratio of methane to
higher-chain hydrocarbons (ethane, propane, and butane) can ty- Water source, n milligrams CH4 L−1 δ13 C-CH4 , ‰
pically be used to differentiate shallower, biologically derived Nonactive Catskill, 5 1.9 ± 6.3 −52.5 ± 7.5
methane from deeper physically derived thermogenic methane Active Catskill, 13 26.8 ± 30.3 −33.5 ± 3.5
(14). Values of δ13 C-CH4 less negative than approximately −50‰ Nonactive Genesee, 8 1.5 ± 3.0 −57.5 ± 9.5
are indicative of deeper thermogenic methane, whereas values Active Genesee, 1 0.3 −34.1
more negative than −64‰ are strongly indicative of microbial Active Lockhaven, 7 50.4 ± 36.1 −40.7 ± 6.7
methane (14). Likewise, δ2 H-CH4 values more negative than Total active wells, 21 19.2 −37 ± 7
about −175‰, particularly when combined with low δ13 C-CH4 Total nonactive wells, 13 1.1 −54 ± 11
values, often represent a purer biogenic methane origin (14). The variable n refers to the number of samples.

2 of 5 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1100682108 Osborn et al.

Using a Bernard plot (15) for analysis (Fig. 4B), the enriched
δ13 C-CH4 (approximately > − 50‰) values accompanied by
low ratios of methane to higher-chain hydrocarbons (less than
approximately 100) in drinking-water wells also suggest that dis-
solved gas is more thermogenic at active than at nonactive sites
(Fig. 4B). For instance, 12 dissolved-gas samples at active drilling
sites fell along a regional gas trajectory that increases with reser-
voir age and thermal maturity of organic matter, with samples
from Susquehanna County, Pennsylvania specifically matching
natural-gas geochemistry from local gas wells (Fig. 4B, orange
oval). These 12 samples and local natural-gas samples are con-
sistent with gas sourced from thermally mature organic matter
of Middle Devonian and older depositional ages often found
in Marcellus Shale from approximately 2,000 m below the surface
in the northern Appalachian Basin (14–19) (Fig. 4B). In contrast,
none of the methane samples from nonactive drilling areas fell
upon this trajectory (Fig. 4B); eight dissolved-gas samples in
Fig. 4B from active drilling areas and all of the values from non-
active areas may instead be interpreted as mixed biogenic/
thermogenic gas (18) or, as Laughrey and Baldassare (17) pro-
posed for their Pennsylvanian gas data (Fig. 4B), the early migra-
tion of wet thermogenic gases with low-δ13 C-CH4 values and
high methane-to-higher-chain hydrocarbon ratios. One data
point from a nonactive area in New York fell squarely in the para-

ENVIRONMENTAL
meters of a strictly biogenic source as defined by Schoell (14)

SCIENCES
(Fig. 4B, upper-left corner).
Carbon isotopes of dissolved inorganic carbon (δ13 C-DIC >
þ10‰) and the positive correlation of δ2 H of water and δ2 H
of methane have been used as strong indicators of microbial
methane, further constraining the source of methane in shallow
groundwater (depth less than 550 m) (18, 20). Our δ13 C-DIC
values were fairly negative and show no association with the
δ13 C-CH4 values (Fig. S3), which is not what would be expected
if methanogenesis were occurring locally in the shallow aquifers.
Instead, the δ13 C-DIC values from the shallow aquifers plot
within a narrow range typical for shallow recharge waters, with
the dissolution of CO2 produced by respiration as water passes
downward through the soil critical zone. Importantly, these
values do not indicate extensive microbial methanogenesis or
sulfate reduction. The data do suggest gas-phase transport of
methane upward to the shallow groundwater zones sampled for
this study (<190 m) and dissolution into shallow recharge waters
locally. Additionally, there was no positive correlation between
the δ2 H values of methane and δ2 H of water (Fig. S4), indicating
that microbial methane derived in this shallow zone is negligible.
Overall, the combined gas and formation-water results indicate
that thermogenic gas from thermally mature organic matter of
Middle Devonian and older depositional ages is the most likely
Fig. 4. (A) Methane concentrations in groundwater versus the carbon source of the high methane concentrations observed in the shal-
isotope values of methane. The nonactive and active data depicted in Fig. 3 low water wells from active extraction sites.
are subdivided based on the host aquifer to illustrate that the methane A different potential source of shallow groundwater contam-
concentrations and δ13 C values increase with proximity to natural-gas well
drilling regardless of aquifer formation. Gray areas represent the typical
ination associated with gas drilling and hydraulic fracturing is
range of thermogenic and biogenic methane taken from Osborn and Mcin- the introduction of hypersaline formation brines and/or fractur-
tosh (18). VPDB, Vienna Pee Dee belemnite. (B) Bernard plot (15) of the ratio ing fluids. The average depth range of drinking-water wells in
of methane to higher-chain hydrocarbons versus the δ13 C of methane. The northeastern Pennsylvania is from 60 to 90 m (12), making the
smaller symbols in grayscale are from published gas-well samples from gas average vertical separation between drinking-water wells and
production across the region (16–18). These data generally plot along a tra- the Marcellus Shale in our study area between approximately
jectory related to reservoir age and thermal maturity (Upper Devonian 900 and 1,800 m (Fig. 2). The research area, however, is located
through Ordovician; see text for additional details). The gas-well data in in tectonically active areas with mapped faults, earthquakes, and
the orange ovals are from gas wells in our study area in Susquehanna County,
lineament features (Fig. 2 and Fig. S1). The Marcellus formation
Pennsylvania (data from Pennsylvania Department of Environmental Protec-
tion). Gray areas represent typical ranges of thermogenic and biogenic
also contains two major sets of joints (21) that could be conduits
methane (data from Osborn and McIntosh, ref. 18). for directed pressurized fluid flow. Typical fracturing activities in
the Marcellus involve the injection of approximately 13–19 mil-
lion liters of water per well (22) at pressures of up to 69,000 kPa.
carbons versus δ13 C-CH4 (Fig. 4B). Methane concentrations not The majority of this fracturing water typically stays underground
only increased in proximity to gas wells (Fig. 3), the accompany- and could in principle displace deep formation water upward into
ing δ13 C-CH4 values also reflected an increasingly thermogenic shallow aquifers. Such deep formation waters often have high
methane source (Fig. 4A). concentrations of total dissolved solids >250;000 mg L−1 , trace

Osborn et al. PNAS Early Edition ∣ 3 of 5

toxic elements, (18), and naturally occurring radioactive materi- tivity of the fracture system. The reduced pressure following the
als, with activities as high as 16;000 picocuries per liter fracturing activities could release methane in solution, leading to
(1 pCi L−1 ¼ 0.037 becquerels per liter) for 226 Ra compared to methane exsolving rapidly from solution (29), allowing methane
a drinking-water standard of 5 pCi L−1 for combined 226 Ra and gas to potentially migrate upward through the fracture system.
226 Methane migration through the 1- to 2-km-thick geological
Ra (23).
We evaluated the hydrochemistry of our 68 drinking-water formations that overlie the Marcellus and Utica shales is less
wells and compared these data to historical data of 124 wells likely as a mechanism for methane contamination than leaky well
in the Catskill and Lockhaven aquifers (24, 25). We used three casings, but might be possible due to both the extensive fracture
types of indicators for potential mixing with brines and/or saline systems reported for these formations and the many older, un-
fracturing fluids: (i) major inorganic chemicals; (ii) stable isotope cased wells drilled and abandoned over the last century and a half
signatures of water (δ18 O, δ2 H); and (iii) isotopes of dissolved in Pennsylvania and New York. The hydraulic conductivity in the
constituents (δ13 C-DIC, δ11 B, and 226 Ra). Based on our data overlying Catskill and Lockhaven aquifers is controlled by a sec-
(Table 2), we found no evidence for contamination of the shallow ondary fracture system (30), with several major faults and linea-
wells near active drilling sites from deep brines and/or fracturing ments in the research area (Fig. 2 and Fig. S1). Consequently, the
fluids. All of the Naþ , Cl− , Ca2þ , and DIC concentrations in high methane concentrations with distinct positive δ13 C-CH4 and
wells from active drilling areas were consistent with the baseline δ2 H-CH4 values in the shallow groundwater from active areas
historical data, and none of the shallow wells from active drilling could in principle reflect the transport of a deep methane source
areas had either chloride concentrations >60 mg L−1 or Na-Ca- associated with gas drilling and hydraulic-fracturing activities. In
Cl compositions that mirrored deeper formation waters (Table 2). contrast, the low-level methane migration to the surface ground-
Furthermore, the mean isotopic values of δ18 O, δ2 H, δ13 C-DIC, water aquifers, as observed in the nonactive areas, is likely a nat-
δ11 B, and 226 Ra in active and nonactive areas were indistinguish- ural phenomenon (e.g., ref. 31). Previous studies have shown
able. The 226 Ra values were consistent with available historical that naturally occurring methane in shallow aquifers is typically
data (25), and the composition of δ18 O and δ2 H in the well-water associated with a relatively strong biogenic signature indicated
appeared to be of modern meteoric origin for Pennsylvania by depleted δ13 C-CH4 and δ2 H-CH4 compositions (32) coupled
(26) (Table 2 and Fig. S5). In sum, the geochemical and isotopic with high ratios of methane to higher-chain hydrocarbons (33), as
features for water we measured in the shallow wells from both we observed in Fig. 4B. Several models have been developed to
active and nonactive areas are consistent with historical data explain the relatively common phenomenon of rapid vertical
and inconsistent with contamination from mixing Marcellus Shale transport of gases (Rn, CH4 , and CO2 ) from depth to the surface
formation water or saline fracturing fluids (Table 2). (e.g., ref. 31), including pressure-driven continuous gas-phase
There are at least three possible mechanisms for fluid migra- flow through dry or water-saturated fractures and density-driven
tion into the shallow drinking-water aquifers that could help buoyancy of gas microbubbles in aquifers and water-filled frac-
explain the increased methane concentrations we observed near tures (31). More research is needed across this and other regions
gas wells (Fig. 3). The first is physical displacement of gas-rich to determine the mechanism(s) controlling the higher methane
deep solutions from the target formation. Given the lithostatic concentrations we observed.
and hydrostatic pressures for 1–2 km of overlying geological stra- Based on our groundwater results and the litigious nature of
ta, and our results that appear to rule out the rapid movement of shale-gas extraction, we believe that long-term, coordinated sam-
deep brines to near the surface, we believe that this mechanism pling and monitoring of industry and private homeowners is
is unlikely. A second mechanism is leaky gas-well casings (e.g., needed. Compared to other forms of fossil-fuel extraction, hy-
refs. 27 and 28). Such leaks could occur at hundreds of meters draulic fracturing is relatively poorly regulated at the federal level.
underground, with methane passing laterally and vertically Fracturing wastes are not regulated as a hazardous waste under
through fracture systems. The third mechanism is that the process the Resource Conservation and Recovery Act, fracturing wells
of hydraulic fracturing generates new fractures or enlarges exist- are not covered under the Safe Drinking Water Act, and only re-
ing ones above the target shale formation, increasing the connec- cently has the Environmental Protection Agency asked fracturing

Table 2. Comparisons of selected major ions and isotopic results in drinking-water wells from this study to data available on the same
formations (Catskill and Lockhaven) in previous studies (24, 25) and to underlying brines throughout the Appalachian Basin (18)
Active Nonactive Previous studies (background)
Lockhaven Catskill Catskill Genesee Lockhaven Catskill formation Appalachian brines
formation formation formation group formation (25) (24) (18, 23)
N¼8 N ¼ 25 N ¼ 22 N ¼ 12 N ¼ 45 N ¼ 79 N ¼ 21
Alkalinity as HCO− ,
3
mg L−1 285 ± 36 157 ± 56 127 ± 53 158 ± 56 209 ± 77 133 ± 61 150 ± 171
mM [4.7 ± 0.6] [2.6 ± 0.9] [2.1 ± 0.9] [2.6 ± 0.9] [3.4 ± 1.3] [2.2 ± 1.0] [2.5 ± 2.8]
Sodium, mg L−1 87 ± 22 23 ± 30 17 ± 25 29 ± 23 100 ± 312 21 ± 37 33,000 ± 11,000
Chloride, mg L−1 25 ± 17 11 ± 12 17 ± 40 9 ± 19 132 ± 550 13 ± 42 92,000 ± 32,000
Calcium, mg L−1 22 ± 12 31 ± 13 27 ± 9 26 ± 5 49 ± 39 29 ± 11 16,000 ± 7,000
Boron, μg L−1 412 ± 156 93 ± 167 42 ± 93 200 ± 130 NA NA 3,700 ± 3,500
δ11 B ‰ 27 ± 4 22 ± 6 23 ± 6 26 ± 6 NA NA 39 ± 6
226
Ra, pCi L−1 0.24 ± 0.2 0.16 ± 0.15 0.17 ± 0.14 0.2 ± 0.15 0.56 ± 0.74 NA 6,600 ± 5,600
δ2 H, ‰, VSMOW −66 ± 5 −64 ± 3 −68 ± 6 −76 ± 5 NA NA −41 ± 6
δ18 O, ‰, VSMOW −10 ± 1 −10 ± 0.5 −11 ± 1 −12 ± 1 NA NA −5 ± 1
Some data for the active Genesee Group and nonactive Lockhaven Formation are not included because of insufficient sample sizes (NA). Values represent
means Æ1 standard deviation. NA, not available.
N values for δ11 B ‰ analysis are 8, 10, 3, 6, and 5 for active Lockhaven, active Catskill, nonactive Genesee, nonactive Catskill, and brine, respectively. N
values for 226 Ra are 6, 7, 3, 10, 5, and 13 for active Lockhaven, active Catskill, nonactive Genesee, nonactive Catskill, background Lockhaven, and brine,
respectively. δ11 B ‰ normalized to National Institute of Standards and Technology Standard Reference Material 951. δ2 H and δ18 O normalized to Vienna
Standard Mean Ocean Water (VSMOW).

4 of 5 ∣ www.pnas.org/cgi/doi/10.1073/pnas.1100682108 Osborn et al.

firms to voluntarily report a list of the constituents in the fractur- Methods
ing fluids based on the Emergency Planning and Community A total of 68 drinking-water samples were collected in Pennsylvania and New
York from bedrock aquifers (Lockhaven, 8; Catskill, 47; and Genesee, 13) that
Right-to-Know Act. More research is also needed on the mechan-
overlie the Marcellus or Utica shale formations (Fig. S1). Wells were purged
ism of methane contamination, the potential health consequences to remove stagnant water, then monitored for pH, electrical conductance,
of methane, and establishment of baseline methane data in other and temperature until stable values were recorded. Samples were collected
locations. We believe that systematic and independent data on “upstream” of any treatment systems, as close to the water well as possible,
groundwater quality, including dissolved-gas concentrations and and preserved in accordance with procedures detailed in SI Methods.
Dissolved-gas samples were analyzed at Isotech Laboratories and water
isotopic compositions, should be collected before drilling opera- chemical and isotope (O, H, B, C, Ra) compositions were measured at Duke
tions begin in a region, as is already done in some states. Ideally, University (see SI Methods for analytical details).
these data should be made available for public analysis, recogniz-
ing the privacy concerns that accompany this issue. Such baseline ACKNOWLEDGMENTS. We thank Rebecca Roter, Peggy Maloof, and many
data would improve environmental safety, scientific knowledge, others who allowed us to sample their water wells; Laura Ruhl and Tewodros
Rango for coordination and field assistance; Nicolas Cassar for thoughtful
and public confidence. Similarly, long-term monitoring of ground- suggestions on the research; and Kaiguang Zhao and Rose Merola for help
water and surface methane emissions during and after extraction with figures. Jon Karr and the Duke Environmental Isotope Laboratory
would clarify the extent of problems and help identify the mechan- performed analyses of δ18 O, δ2 H, and δ13 C of groundwater samples. William
Chameides, Lincoln Pratson, William Schlesinger, the Jackson Lab, and two
isms behind them. Greater stewardship, knowledge, and—possi- anonymous reviewers provided helpful suggestions on the manuscript and
bly—regulation are needed to ensure the sustainable future of research. We gratefully acknowledge financial support from Fred and
Alice Stanback to the Nicholas School of the Environment and from the Duke
shale-gas extraction. Center on Global Change.

1. Pacala S, Socolow R (2004) Stabilization wedges: Solving the climate problem for the 20. Martini AM, et al. (1998) Genetic and temporal relations between formation waters
next 50 years with current technologies. Science 305:968–972. and biogenic methane: Upper Devonian Antrim Shale, Michigan Basin, USA. Geochim
2. Tour JM, Kittrell C, Colvin VL (2010) Green carbon as a bridge to renewable energy. Cosmochim Acta 62:1699–1720.

ENVIRONMENTAL
Nature Mater 9:871–874. 21. Engelder T, Lash GG, Uzcategui RS (2009) Joint sets that enhance production from

SCIENCES
3. Kerr RA (2010) Natural gas from shale bursts onto the scene. Science 328:1624–1626. Middle and Upper Devonian gas shales of the Appalachian Basin. Am Assoc Pet Geol
4. Raupach MR, et al. (2007) Global and regional drivers of accelerating CO2 emissions. Bull 93:857–889.
Proc Natl Acad Sci USA 104:10288–10293. 22. Pennsylvania Department of Environmental Protection (2011) (Pennsylvania Dept of
5. US Energy Information Administration (2010) Annual Energy Outlook 2010 with Pro- Environmental Protection, Harrisburg, PA), Marcellus Shale, http://www.dep.state.
jections to 2035 (US Energy Information Administration, Washington, DC), DOE/EIA- pa.us/dep/deputate/minres/oilgas/new_forms/marcellus/marcellus.htm.
0383; http://www.eia.doe.gov/oiaf/aeo/pdf/0383(2010).pdf. 23. New York State Department of Health, Bureau of Environmental Radiation Protection
(2009) (New York State Dept of Health, Troy, NY), Comments, July 21, 2009, Supple-
6. US Environmental Protection Agency (2011) Hydraulic Fracturing. (US Environmental
mental Generic Environmental Statement on the Oil and Gas Regulatory Program
Protection Agency, Washington, DC), http://water.epa.gov/type/groundwater/uic/
Well Permit Issuance for Horizontal Drilling and Hydraulic-Fracturing to Develop
class2/hydraulicfracturing/.
the Marcellus Shale and other Low Permeability Gas Reservoirs; http://www.
7. Kargbo DM, Wilhelm RG, Campbell DJ (2010) Natural gas plays in the Marcellus shale:
riverkeeper.org/wp-content/uploads/2010/01/Riverkeeper-DSGEIS-Comments-Appendix-
Challenges and potential opportunities. Environ Sci Technol 44:5679–5684.
3-NYSDOH-Environmental-Radiation-Memo.pdf.
8. Revesz KM, Breen KJ, Baldassare AJ, Burruss RC (2010) Carbon and hydrogen isotopic
24. Taylor LE (1984) Groundwater Resources of the Upper Susquehanna River Basin, Penn-
evidence for the origin of combustible gases in water supply wells in north-central
sylvania: Water Resources Report 58. (Pennsylvania Dept of Environmental Resources-
Pennsylvania. Appl Geochem 25:1845–1859. Office of Parks and Forestry—Bureau of Topographic and Geologic Survey, Harrisburg,
9. Zoback M, Kitasei S, Copithorne B Addressing the environmental risks from shale gas PA) 139.
development. Worldwatch Institute Briefing Paper 1 (Worldwatch Inst, Washington, 25. Williams JH, Taylor L, Low D (1998) Hydrogeology and Groundwater Quality of the
DC), http://blogs.worldwatch.org/revolt/wp-content/uploads/2010/07/Environmental- Glaciated Valleys of Bradford, Tioga, and Potter Counties, Pennsylvania: Water Re-
Risks-Paper-July-2010-FOR-PRINT.pdf. sources Report 68. (Commonwealth of Pennsylvania Dept of Conservation and Natural
10. Pennsylvania Department of Environmental Protection, Bureau of Oil and Gas Resources, Harrisburg, PA) p 89.
Management (2010) 2009 Year End Workload Report. (Pennsylvania Dept of Environ- 26. Kendall C, Coplan TB (2001) Distribution of oxygen-18 and deuterium in river waters
mental Protection, Bureau of Oil and Gas Management, Harrisburg, PA), http://www. across the United States. Hydrol Processes 15:1363–1393.
dep.state.pa.us/dep/deputate/minres/oilgas/2009%20Year%20End%20Report-WEBSITE. 27. Van Stempvoort D, Maathuis H, Jaworski E, Mayer B, Rich K (2005) Oxidation of
pdf. fugitive methane in groundwater linked to bacterial sulfate reduction. Ground Water
11. Colborn T, Kwiatkowski C, Schultz K, Bachran M (2010) Natural gas operations from 43:187–199.
a public health perspective. Hum Ecol Risk Assess, in press. 28. Taylor SW, Sherwood Lollar B, Wassenaar LI (2000) Bacteriogenic ethane in near-
12. Pennsylvania Department of Environmental Protection (2011) Private Water Wells surface aquifers: Implications for leaking hydrocarbon well bores. Environ Sci Technol
in Pennsylvania. (Pennsylvania Dept of Environmental Protection, Harrisburg, PA), 34:4727–4732.
http://www.dep.state.pa.us/dep/deputate/watermgt/wc/Subjects/SrceProt/well/. 29. Cramer B, Schlomer S, Poelchau HS (2002) Uplift-related hydrocarbon accumulations:
13. Eltschlager KK, Hawkins JW, Ehler WC, Baldassare F (2001) Technical Measures for the the release of natural gas from groundwater. 196 (Geological Society Special Publica-
Investigation and Mitigation of Fugitive Methane Hazards in Areas of Coal Mining (US tions, London), 447–455.
Dept of the Interior, Office of Surface Mining Reclamation and Enforcement, 30. Geyer AR, Wilshusen JP (1982) Engineering characteristics of the rocks of Pennsylvania;
Pittsburgh). environmental geology supplement to the state geologic map, 1982 Pennsylvania
Geological Survey. (Dept of Environmental Resources, Office of Resources Manage-
14. Schoell M (1980) The hydrogen and carbon isotopic composition of methane from
ment, Harrisburg, PA).
natural gases of various origins. Geochim Cosmochim Acta 44:649–661.
31. Etiope G, Martinelli G (2002) Migration of carrier and trace gases in the geosphere:
15. Bernard BB (1978) Light hydrocarbons in marine sediments. PhD Dissertation (Texas
An overview. Phys Earth Planet Inter 129:185–204.
A&M Univ, College Station, TX).
32. Aravena R, Wassenaar LI (1993) Dissolved organic carbon and methane in a regional
16. Jenden PD, Drazan DJ, Kaplan IR (1993) Mixing of thermogenic natural gases in north-
confined aquifer, southern Ontario, Canada: Carbon isotope evidence for associated
ern Appalachian Basin. Am Assoc Pet Geol Bull 77:980–998.
subsurface sources. Appl Geochem 8:483–493.
17. Laughrey CD, Baldassare FJ (1998) Geochemistry and origin of some natural gases in 33. Coleman DD, Liu C, Riley KM (1988) Microbial methane in the shallow Paleozoic
the Plateau Province Central Appalachian Basin, Pennsylvania and Ohio. Am Assoc Pet sediments and glacial deposits of the Illinois, USA. Chem Geol 71:23–40.
Geol Bull 82:317–335. 34. Alexander SS, Cakir R, Doden AG, Gold DP, Root SI (2005) Basement depth and related
18. Osborn SG, McIntosh JC (2010) Chemical and isotopic tracers of the contribution of geospatial database for Pennsylvania: Pennsylvania Geological Survey, 4th ser., Open-
microbial gas in Devonian organic-rich shales and reservoir sandstones, northern File General Geology Report 05-01.0. (Pennsylvania Dept of Conservation and Natural
Appalachian Basin. Appl Geochem 25:456–471. Resources, Middletown, PA), http://www.dcnr.state.pa.us/topogeo/openfile.
19. Repetski JE, Ryder RT, Harper JA, Trippi MH (2006) Thermal maturity patterns in the 35. Pennsylvania Spatial Data Access (PASDA) Online mapping, data access wizard, oil
Ordovician and Devonian of Pennsylvania using conodont color alteration index (CAI) and gas locations. (Pennsylvania Dept of Environmental Protection, Harrisburg, PA),
and vitrinite reflectance (%Ro). Northeastern Geology Environmental Sciences http://www.pasda.psu.edu/uci/SearchResults.aspx?searchType=mapservice&condition=
28:266–294. OR&entry=PASDA.

Osborn et al. PNAS Early Edition ∣ 5 of 5

Duke Study: Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing

More Related Content

What's hot (19)

Viewers also liked (7)

Similar to Duke Study: Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing (20)

More from Marcellus Drilling News (20)

Recently uploaded (20)

Duke Study: Methane contamination of drinking water accompanying gas-well drilling and hydraulic fracturing