A Critical Review of the Risks to Water Resources from
Unconventional Shale Gas Development and Hydraulic Fracturing in
the United States
Avner Vengosh,*
,†
Robert B. Jackson,
†,‡
Nathaniel Warner,
§
Thomas H. Darrah,
∥
and Andrew Kondash
†
†
Division of Earth and Ocean Sciences, Nicholas School of the Environment, Duke University, Durham, North Carolina 27708,
United States
‡
School of Earth Sciences, Woods Institute for the Environment, and Precourt Institute for Energy, Stanford University, Stanford,
California 94305, United States
§
Department of Earth Sciences, Dartmouth College, Hanover, New Hampshire 03755, United States
∥
School of Earth Sciences, The Ohio State University, Columbus, Ohio 43210, United States
*
S
Supporting Information
ABSTRACT: The rapid rise of shale gas development through horizontal drilling
and high volume hydraulic fracturing has expanded the extraction of hydrocarbon
resources in the U.S. The rise of shale gas development has triggered an intense
public debate regarding the potential environmental and human health effects
from hydraulic fracturing. This paper provides a critical review of the potential
risks that shale gas operations pose to water resources, with an emphasis on case
studies mostly from the U.S. Four potential risks for water resources are
identified: (1) the contamination of shallow aquifers with fugitive hydrocarbon
gases (i.e., stray gas contamination), which can also potentially lead to the
salinization of shallow groundwater through leaking natural gas wells and
subsurface flow; (2) the contamination of surface water and shallow groundwater
from spills , leaks, and/or the disposal of inadequately treated shale gas
wastewater; (3) the accumulation of toxic and radioactive elements in soil or
stream sediments near disposal or spill sites; and (4) the overextraction of water
resources for high-volume hydraulic fracturing that could induce water shortages or confl icts with other water users, particularly
in water-scarce areas. Analysis of published data (through January 2014) reveals evidence for stray gas contamination, surface
water impacts in areas of intensive shale gas development, and the accumulation of radium isotopes in some disposal and spill
sites. The direct contamination of shallow groundwater from hydraulic fracturing fluids and deep formation waters by hydraulic
fracturing itself, however, remains controversial.
1. INTRODUCTION
Production from unconventiognal natural gas reservoirs has
substantially expanded through the advent of horizontal drilling
and high-volume hydr aulic fracturing (Figure 1). These
technological advances have opened vast new energy sources,
such as low-permeability organic-rich shale formations and
“tight-sand” reservoirs, altering the domestic energy landscape
in the United States.
1−3
The total production of natural gas has
increased by more than 30% during the past decade. In 2012,
unconventional shale gas and tight sand productions were
respectively accounting for 34% and 24% of the total natural
gas production in the U.S. (0.68 trillion m
3
).
4
The increase in energy production has been broadly
distributed across the United States (Figure 2) and densely
distributed within specifi c shale plays (Figure 3). Unconven-
tional hydrocarbon extraction from organic-rich shale for-
mations is now active in more than 15 plays in the U.S. In PA
alone, 7234 shale gas wells were drilled into the Marcellus
Formation
5
in addition to the 34 376 actively producing
conventional oil and gas wells in that state (2012 data; Figure
3).
6
At the end of 2012, the Marcellus Shale (29%), Haynesville
Shale (23%), and Barnett Shale (17%) dominated production
of natural gas (primarily methane, ethane, and propane) from
shales in the U.S., with the remaining 31% of total shale gas
production contributed by more than a dozen basins (Figure
1). Oil and hydrocarbon condensates are also targeted in
numerous basins, including the Barnett, Eagle Ford, Utica-Point
Pleasant, and Bakken.
4
Future energy forecasts suggest that U.S. unconventional
natural gas production from shale formations will double by
Special Issue: Understanding the Risks of Unconventional Shale Gas
Development
Received: November 18, 2013
Revised: February 9, 2014
Accepted: February 18, 2014
Critical Review
pubs.acs.org/est
© XXXX American Chemical Society A dx.doi.org/10.1021/es405118y | Environ. Sci. Technol. XXXX, XXX, XXX−XXX
2035 and generate ∼50% of the total domestic natural gas
production.
4
Similarly, U.S. domestic oil production from
unconventional shale formations is projected to increase by as
much as 15% over the next several decades.
7
Unconventional
extraction (hor izontal drilling and high volume hydraulic
fracturing) for shale gas has already expanded to Canada
8
and will soon be launched on a global scale, with signi ficant
reservoirs in South America, northern and southern Africa,
Europe, China,
9,10
and Au stralia.
11,12
The current global
estimate of natural gas reserves in unconventional shale is
approximately 716 trillion m
3
(2.53 × 10
13
Mcf).
11,12
Despite the large resource potentials and economic benefits,
the rapid expansion of shale gas development in the U.S. has
triggered an intense public debate over the possible environ-
mental and human health implications of the unconventional
energy development. Some primary concerns include air
pollution, greenhouse gas emissions, radiation, and ground-
water and surface water contamination.
1,3,13−36
These concerns
have been heightened because the 2005 Energy Policy Act
exempts hydraulic fracturing operations from the Safe Drinking
Water Act (SDWA). The only exception to the exemption is
the injection of diesel fuel. Additionally, because environmental
oversight for most oil and gas operations is conducted by state
rather federal agencies, the r egulation, monitoring, and
enforcement of various environmental contamination issues
Figure 1. Evolution of the volume of natural gas production from
different unconventional shale plays in the U.S. Data from the U.S.
Energy Information Administration.
4
Figure 2. Map of unconventional shale plays in the U.S. and Canada, based on data from the U.S. Energy Information Administration.
4
Environmental Science & Technology Critical Review
dx.doi.org/10.1021/es405118y | Environ. Sci. Technol. XXXX, XXX, XXX−XXXB
related to unconventional shale gas development are highly
variable throughout the U.S.
37−39
This paper provides an overview and synopsis of recent
investigations (updated to January 2014) into one set of
possible environmental impacts from unconventional shale gas
development: the potential risks to water resources. We identify
four potential modes of water resource degradation that are
illustrated schematically in Figure 4 and include (1) shallow
aquifers contaminated by fugitive natural gas (i.e., stray gas
contamination) from leaking shale gas and conventional oil and
gas wells, potentially followed by water contamination from
hydraulic fracturing fluids and/or formation waters from the
deep formations; (2) surface water contamination from spills,
leaks, and the disposal of inadequately treated wastewater or
hydraulic fracturing fluids; (3) accumulation of toxic and
radioactive elements in soil and the sediments of rivers and
lakes exposed to wastewater or fluids used in hydraulic
fracturing; and (4) the overuse of water resources, which can
compete with other water uses such as agriculture in water-
limited environments.
2. GROUNDWATER CONTAMINATION
2.1. Stray Gas Contamination. Elevated levels of methane
and other aliphatic hydrocarbons such as ethane and propane in
shallow drinking water wells pose a potential flammability or
explosion hazard to homes with private domestic wells. The
saturation level of methane in near-surface groundwater is
about ∼28 mg/L ( ∼40 cc/L) and thus the U.S. Department of
the Interior recommends monitoring if water contains more
than 10 mg/L (∼14 cc/L) of methane and immediate action if
concentrations rise above 28 mg/L. Several states have defined
a lower threshold (e.g., 7 mg/L in PA), from which household
utilization of methane-rich groundwater is not recommended.
Stray gas migration in shallow aquifers can potentially occur
by the release of gas-phase hydrocarbons through leaking
casings or along the well annulus, from abandoned oil and gas
wells, or potentially along existing or incipient faults or
fractures
40
with target or adjacent stratigraphic formations
following hydraulic fracturing and drilling (Figure 4).
27
The
latter mechanism poses a long-term risk to shallow ground-
water aquifers. Microseismic data suggest that the deformation
and fractures developed following hydraulic fracturing typically
extend less than 600 m above well perforations, suggesting that
fracture propagation is insuffi cient to reach drinking-water
aquifers in most situations.
41
This assertion is supported by
noble gas data from northeastern PA,
42
yet stray gas migration
through fractures and faults is considered a potential
mechanism for groundwater contamination.
40
Across the northeastern Appalachian Basin in PA, the
majority of shallow groundwater had detectable, naturally
occurring methane with thermogenic stable-isotope fingerprints
(e.g., δ
13
C−CH
4
and δ
2
H−CH
4
).
27−29,42,43
These findings
imply that the high methane in shallow aquifers from this
region is predominantly thermogenic in origin.
28,29,42,43
In
northeastern PA, however, a subset of shallow drinking water
wells consistently showed elevated meth ane, ethane, and
Figure 3. Map of active unconventional (yellow) and conventional (purple) oil and gas wells in Pennsylvania and West Virginia. Note areas of
coexisting conventional and unconventional development (e.g., southwestern PA and WV) relative to areas of exclusively unconventional
development (e.g., northeastern PA). Well locations were obtained from the West Virginia Geological Survey (http://www.wvgs.wvnet.edu/) and
the Pennsylvania Department of Environmental Protection’s oil and gas reporting Web site (https://www.paoilandgasreporting.state.pa.us/
publicreports/Modules/Welcome/Welcome.aspx). The background topographic map, Marcellus Formation outline, and state boundaries were
downloaded from http://www.pasda.psu.edu/ and the Carnegie Museum of Natural History.
148
Environmental Science & Technology Critical Review
dx.doi.org/10.1021/es405118y | Environ. Sci. Technol. XXXX, XXX, XXX−XXXC
propane concentrations (i.e., relatively low hydrocarbon ratios
(C
1
/C
2
)) and relatively enriched thermogenic carbon isotope
fingerprints in groundwater exclusively <1 km from shale gas
drilling sites. A subset of samples with evidence for stray gas
contamination display isotopic reversals (Δ
13
C=
δ
13
CH
4
−δ
13
C
2
H
6
> 0) and proportions of methane, ethane
and propane that were consistent with Marcellus production
gases from the region, while some other wells had natural gas
compositions consistent with production gases in conventional
wells from the overlying Upper Devonian formations.
27,29
New
evaluations of the helium content
29
and noble gas geo-
chemistry
42
in these samples further supports a distinction
between naturally occurring “background” hydrocarbon gases
and groundwater with stray gas contamination in wells located
near (<1 km) shale gas drilling sites. “Background” gases
typically had lower proportions of ethane and propane and
elevated helium concentrations that reflect the history of
natural gas migration from the Marcellus source rock to the
Upper Devonian reservoir rocks throughout geological
time.
29,42
Thus, the combination of gas geochemical finger-
printing suggests that stray gas groundwater contamination,
where it occurs, is sourced from the target shale formations
(i.e., the Marcellus Formation) in some cases, and from
intermediate layers (e.g., Upper Devonian Formations) in
others.
In cases where the composition of stray gas is consistent with
the target shale formation, it is likely that the occurrence of
fugitive gas in shallow aquifers is caused by leaky, failing, or
improperly installed casings in the natural gas wells. In other
cases, hydrocarbon and noble gas data also indicated that
fugitive gas from intermediate formations apparently flowed up
through the outside of the well annulus and then leaked into
the overlying shallow aquifers.
27,29,42
In these cases, the isotopic
signatures and hydrocarbon ratios matched the gases in
intermediate formations rather than Marcellus shale production
gases. In sum, the combined evidence of hydrocarbon stable
isotopes, molecular hyd rocarbon ratios, and helium geo-
chemistry indicate that stray gas contamination occurs in a
subset of wells <1km from drilling in northeastern PA.
In contrast to these reports, other investigators
22,43,44
have
suggested that higher methane concentrations in shallow
groundwater were natural and could be explained by topo-
graphic factors associated with groundwater discharge zones.
Geochemical data do suggest that some natural gas migrated to
shallow aquifers in northeastern PA through geologic time.
However, these characteristics occur in areas with higher
hydraulic connectivity between the deep and shallow
formations.
34,42
A recent analysis showed that topography
was indeed a statistically significant factor in some cases but did
not explain the variations in methane and ethane concen-
trations with respect to distance to gas wells.
29
Additional evidence for stray gas contamination because of
poor well construction is provided by the isotopic composition
of surface casing vent flow (SCV). Integrating the δ
13
C data of
methane (C1), ethane (C2), and propane (C3)
45−47
showed
that stray gas contamination associated with conventional oil
wells in Alberta, Canada reflected methane sourced from
intermediate formations leaking into shallow aquifers and not
from the production formations such as the Lower Cretaceous
Mannville Group.
48
Jackson et al. (2013)
49
listed several other
case studi es that demonstrate evidence for stray gas
contamination. While such studies have shown evidence for
methane, ethane, and propane contamination associated with
conventional oil production
48,50
and coal bed methane,
45
Muehlenbachs (2013)
51
also showed direct evidence for SCV
leakage from intermediate zones in newly completed and
hydraulically stimulated horizontal shale gas wells in the
Montney and Horn Ri ver areas of northeastern British
Columbia, Canada.
51
Methane leaking from the annulus of
conventional oil and gas wells was also demonstrated in PA.
52,53
Combined, these studies suggest that stray gas contamination
can result from either natural gas leaking up through the well
annulus, typically from shallower (intermediate) formations, or
through poorly constructed or failing well casings from the
shale target formations.
The migration of natural gas to the surface through the
production casing and/or well annulus is “ acommon
occurrence in the petroleum industry”
51
and can affect a large
fraction of conventional wells. Among the 15 000 production
oil wells tested from the Gulf of Mexico, 43% have reported
cement damage after setting that leads to sustained casing
pressure (SCP). These effects increased with time; whereas
30% reported damage during the first 5 years after drilling, the
percentage increased to 50% after 20 years.
54
Likewise, the BP
Deepwater Horizon oil spill was partly attributed to the fact
that “cement at the well bottom had failed to seal off
Figure 4. Schematic illustration (not to scale) of possible modes of
water impacts associated with shale gas development reviewed in this
paper: (1) overuse of water that could lead to depletion and water-
quality degradation particularly in water-scarce areas; (2) surface water
and shallow groundwater contamination from spills and leaks of
wastewater storage and open pits near drilling; (3) disposal of
inadequately treated wastewater to local streams and accumulation of
contaminant residues in disposal sites; (4) leaks of storage ponds that
are used for deep-well injection; (5) shallow aquifer contamination by
stray gas that originated from the target shale gas formation through
leaking well casing. The stray gas contamination can potentially be
followed by salt and chemical contamination from hydraulic fracturing
fluids and/or formational waters; (6) shallow aquifer contamination by
stray gas through leaking of conventional oil and gas wells casing; (7)
shallow aquifer contamination by stray gas that originated from
intermediate geological formations through annulus leaking of either
shale gas or conventional oil and gas wells; (8) shallow aquifer
contamination through abandoned oil and gas wells; (9) flow of gas
and saline water directly from deep formation waters to shallow
aquifers; and (10) shallow aquifer contamination through leaking of
injection wells.
Environmental Science & Technology Critical Review
dx.doi.org/10.1021/es405118y | Environ. Sci. Technol. XXXX, XXX, XXX−XXXD
hydrocarbons”.
55
In PA the overall reports of cementing, casing,
and well construction violations total 3% of all shale gas wells.
22
However a closer look at the distribution of violations shows
large variations in percentage with time (before and after 2009),
space, and type of wells.
5,56
In particular, the percentage of well
violations was much higher in northeastern and central counties
in PA (10−50%).
5
Consequently, reports of stray gas
contamination in areas of unconventional shale gas develop-
ment in the northeastern Appalachian Ba sin (U.S.) and
Montney and Horn River Basins (Canada) may be associated
with leaking of oil and gas wells.
In contrast to the results from the Marcellus, Montney, and
Horn River Basins, the Fayetteville Shale in north-central
Arkansas showed no evidence of methane contamination in
groundwater. Stud ies in this area showed low meth ane
concentrations with a mostly biogenic isotopic fingerprint.
36,57
The authors hypothesized the potential for stray gas
contamination likely depends on both well integrity and local
geology, including the extent of local fracture systems that
provide flow paths for potential gas migration.
36
In addition to groundwater, surface waters could serve as an
indicator of regional migration from unconventional shale gas
development. To date, streams in areas of shale gas drilling
have not shown systematic evidence of methane contamination.
A new methodology for stream-gas sampling as a reconnais-
sance tool for evaluating natural and anthropogenic methane
leakage from natural gas reservoirs into surface waters was
recently demonstrated using inorganic and gas geochemical
tracers and could be applied more widely in areas of oil and gas
development.
59
2.2. Groundwater Contamination with Salts or Other
Dissolved Constituents. The presence of fugitive gas in
shallow drinking water wells could potentially lead to
salinization and other changes of water quality in three possible
ways. First, the leaking of natural gas can be associated with the
flow of hydraulic fracturing fluids and saline formation waters to
overlying shallow aquifers. Given the buoyancy of gas, the flow
rate of denser saline water would be substantially slower than
the flow of natural gas, and would depend on both the pressure
gradients and hydraulic connectivity between the overpressur-
ized annulus or leaking sites on the wells and the overlying
aquifers.
53
An EPA study
60
near the town of Pavillion, Wyoming found
water contamination in two shallow monitoring wells. Although
this initia l stu dy was questioned for adequate s ampling
protocols,
22
a follow up study by the U.S. Geological Survey
confirmed elevated levels of specific conductance (1500 mS/
cm), pH (10−11), methane (25−27 mg/L), ethane, and
propane.
61
However, the mechanisms that caused the apparent
contamination of the shallow groundwater in this area are still
under investigation (i.e., contamination from surface ponds or
subsurface leaking cement from shale gas wells).
The ability to trace and identify contamination from shale gas
exploration is limited because of the relatively short time frame
since the beginning of large-scale shale gas exploration in early-
2000s c ompared to typical groundwater flow rates (i.e.,
decades). However, an evaluation of water contamination
associated with conventional oil and gas exploration provides a
much longer time frame for evaluating possible groundwater
contamination. Possible evidence of long-term (2000−2007)
increases in the sali nity of groundwater as sociated with
conventional oil and gas drilling was reported from Gar field
County, CO. There, a rise of chloride concentrations in
drinking water wells was associated with an increase of methane
with a thermogenic isotopic fingerprint, both of which were
associated with an increase in the number of conventional oil
and gas wells.
62
The fraction of drinking water wells that had
chloride concentrations >250 mg/L (EPA threshold for
drinking water) in groundwater from Garfield County doubled
between 2002 (4%) and 2005 (8%), with chloride up to 3000
mg/L in drinking water wells.
62
The parallel rise in salinity and
methane with a thermogenic isotope signature in Garfield
County could reflect either migration from leaking oil and gas
wells or contamination from unlined surface impoundments.
62
Overall, the geochemical composition of the salinized ground-
water in such scenarios would mimic the composition of either
the formation water in the production formations
34
or the
fluids in the shallower or intermediate units (that typically have
a different water chemistry). While there might be evidence for
water contamination in some areas of conventional oil and gas
exploration, groundwater sites in areas affected by stray gas
contamination near shale gas sites in northeastern PA have not
to our knowledge shown signs of salinization induced directly
by leaking natural gas wells.
27,29,34
Unlike other areas in PA,
northeastern PA was developed recently and almost exclusively
for shale gas (Figure 3), with few legacy wells reported in the
area. Thus, any water contamination in this area attributable to
natural gas extraction would be related to current shale gas
operations rather than to older legacy wells. Therefore
conclusions regarding contamination from saline water and
hydraulic fracturing fluids flow are restricted in both space and
time and further studies are needed to address this question.
A second mode of groundwater contamination that could
evolve from stray gas contamination is oxidation of fugitive
methane via bacteria l sulfate reduction.
50
Evidence for
dissimilatory bacterial sulfate reduction of fugitive methane
near conventional oil wells in Alberta, Canada, includes sulfide
generation and
13
C-depleted bicarbonate, with lower residual
sulfate concentrations relative to the regional groundwater.
50
Bacterial sulfate reduction reactions due to the presence of
fugitive methane could trigger other processes such as reductive
dissolution of oxides in the aquifer that would mobilize redox-
sensitive elements such as manganese, iron, and arsenic from
the aquifer matrix and further reduce groundwater quality. Low
levels of arsenic and other contaminants, recorded in some
drinking water aquifers in TX, were suggested to be linked to
contamination from the underlying Barnett Shale,
63
although
evidence for a direct link to the Barnett remains uncertain.
Athirdhypotheticalmodeofshallowgroundwater
contamination associated with the presence of stray gas
contamination is the formation of toxic t rihalomethanes
(THMs), typically co-occurring with high concentrations of
halogens in the saline waters. THMs are compounds with
halogen atoms (e.g., Cl, Br, or I) substituted for hydrogens in
the methane molecule. The formation of THMs were
previously recorded in untreated groundwater in the U.S.,
unrelated to shale gas activities, but associated with agricultural
contamination of shallow aquifers.
64,65
Numerous studies have
demonstrated that the presence of halogens together with
organic matter in source waters can trigger the formation of
THMs, specifically in chlorinated drinking water (see
references in Section 2.1). However, no data has to our
knowledge been reported for the presence of THMs in
groundwater associated with stray gas contamination from shale
gas wells.
Environmental Science & Technology Critical Review
dx.doi.org/10.1021/es405118y | Environ. Sci. Technol. XXXX, XXX, XXX−XXXE