A research paper published in the Proceedings of the National Academy of Sciences. The paper evaluated the level of methane in groundwater in Colorado going back 25 years. It finds the rate of groundwater methane did not change after the introduction of horizontal drilling combined with high-volume hydraulic fracturing in 2010. That is, fracking does not increase methane migration.

1. Groundwater methane in relation to oil and gas
development and shallow coal seams in the
Denver-Julesburg Basin of Colorado
Owen A. Sherwooda,1
, Jessica D. Rogersb
, Greg Lackeyb
, Troy L. Burkeb
, Stephen G. Osbornc
, and Joseph N. Ryanb
a
Institute of Arctic and Alpine Research, University of Colorado, Boulder, CO 80309; b
Department of Civil, Environmental and Architectural Engineering,
University of Colorado, Boulder, CO 80309; and c
Department of Geological Sciences, California State Polytechnical University, Pomona, CA 91768
Edited by Peter H. Gleick, Pacific Institute for Studies in Development, Environment, and Security, Oakland, CA, and approved June 7, 2016 (received for
review November 24, 2015)
Unconventional oil and gas development has generated intense
public concerns about potential impacts to groundwater quality.
Specific pathways of contamination have been identified; how-
ever, overall rates of contamination remain ambiguous. We used
an archive of geochemical data collected from 1988 to 2014 to
determine the sources and occurrence of groundwater methane
in the Denver-Julesburg Basin of northeastern Colorado. This
60,000-km2
region has a 60-y-long history of hydraulic fracturing,
with horizontal drilling and high-volume hydraulic fracturing
beginning in 2010. Of 924 sampled water wells in the basin, dis-
solved methane was detected in 593 wells at depths of 20–190 m.
Based on carbon and hydrogen stable isotopes and gas molecular
ratios, most of this methane was microbially generated, likely within
shallow coal seams. A total of 42 water wells contained thermogenic
stray gas originating from underlying oil and gas producing forma-
tions. Inadequate surface casing and leaks in production casing and
wellhead seals in older, vertical oil and gas wells were identified as
stray gas migration pathways. The rate of oil and gas wellbore failure
was estimated as 0.06% of the 54,000 oil and gas wells in the basin
(lower estimate) to 0.15% of the 20,700 wells in the area where stray
gas contamination occurred (upper estimate) and has remained
steady at about two cases per year since 2001. These results show
that wellbore barrier failure, not high-volume hydraulic fracturing in
horizontal wells, is the main cause of thermogenic stray gas migration
in this oil- and gas-producing basin.
unconventional oil and gas | hydraulic fracturing | groundwater |
methane | stray gas
Horizontal drilling combined with hydraulic fracturing has
revolutionized the petroleum industry. It has also generated
persistent concern about environmental impacts to groundwater
quality (1, 2). Rates and pathways of groundwater contamination
resulting from drilling and production operations remain con-
troversial (3–6). Complicating factors include spatial overlap be-
tween legacy and newer development and the presence of naturally
occurring hydrocarbons in the shallow subsurface. Geochemical
data can help resolve these factors and clarify potential impacts to
groundwater (1, 7–11); however, there is a general lack of time
series data of sufficient regional extent in most petroleum-
producing basins. This problem underlies the rationale behind
moratoria on unconventional petroleum development within
various municipal and state/provincial jurisdictions (8).
We focus here on the Denver-Julesburg (DJ) Basin of north-
eastern Colorado (Fig. 1), notable for having a 60-y-long history
of hydraulic fracturing. The technique was introduced in 1950 and
expanded during the 1970s–1990s to stimulate production from
low-permeability reservoirs (SI Appendix) (12–14). Approximately
half of the 49,800 vertical wells in the basin have been hydrauli-
cally fractured, typically with two to three stages of fracturing
(15). In 2010, horizontal drilling was introduced in the Watten-
berg Field (Fig. 1) to exploit the Niobrara shale/chalk forma-
tion (16). Virtually all of these 4,180 horizontal wells have
been hydraulically fractured, typically with 20+ stages of frac-
turing (15). The DJ Basin produces both gas (446 billon cubic feet
in 2014) and oil (88 million barrels in 2014) (15), comprising the
full range of thermogenic maturity from black oil to condensate
and dry gas (12, 13). Previous studies on impacts to groundwater
focused mainly on shale gas plays with a relatively short history of
hydraulic fracturing (3–10); this study focuses on unconventional
oil and gas with a long history of hydraulic fracturing.
The presence of gas in DJ Basin groundwater was noted at
least as early as the 1880s, when artesian wells drilled to depths
of 365–425 m produced flammable gas (17). In 1982, residents near
the towns of Hudson and LaSalle complained of well water that
was oily, flammable, and undrinkable (18). In 1984, gas was found
leaking from five abandoned water wells (240–360 m) in LaSalle,
one of which caused a destructive explosion and fire. Geochemical
analysis determined that the gas originated in the Codell formation
(SI Appendix), which was being developed at the time, but the
specific migration pathways were never identified (19).
Following the events of the early 1980s, the Colorado Oil and
Gas Conservation Commission (COGCC) took over responsi-
bility for investigating water quality complaints. Beginning in
2005, regulations requiring geochemical testing of water wells
were introduced (SI Appendix). As a result, the COGCC has
accumulated groundwater and natural gas geochemical mea-
surements going back to 1988, constituting one of the most
comprehensive groundwater geochemical datasets related to
petroleum development anywhere. The data span a time range
Significance
The impact of unconventional oil and gas development on
groundwater quality remains controversial. We use an archive
of public domain data to examine factors influencing the dis-
tribution and sources of groundwater methane in the oil- and
gas-producing Denver-Julesburg Basin of Colorado. Thermo-
genic stray gas sourced from deep oil and gas reservoirs im-
pacted 42 water wells in 32 separate cases at a rate of about
two cases per year from 2001 to 2014. The rate did not change
after the introduction of horizontal drilling combined with
high-volume hydraulic fracturing in 2010. The risk of stray gas
contamination ranged from 0.12% of 35,000 water wells in the
basin (lower estimate) to 4.5% of the 924 water wells that
were tested (upper estimate).
Author contributions: O.A.S. and J.N.R. designed research; O.A.S., J.D.R., G.L., T.L.B., and
S.G.O. performed research; O.A.S., J.D.R., G.L., T.L.B., and S.G.O. analyzed data; and O.A.S.
and J.N.R. wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Freely available online through the PNAS open access option.
1
To whom correspondence should be addressed. Email: owen.sherwood@colorado.edu.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1523267113/-/DCSupplemental.
www.pnas.org/cgi/doi/10.1073/pnas.1523267113 PNAS Early Edition | 1 of 6
ENVIRONMENTAL
SCIENCES

2. over which oil and gas wellbore construction evolved from
vertical wells with uncemented sections to horizontal wells with
fully cased and cemented vertical sections (20). We obtained
data from the COGCC in October 2014, and after rigorous
data screening and quality control protocols (SI Appendix), we
examined the distribution and origins of groundwater methane in
relation to oil and gas development in the DJ Basin.
Results and Discussion
Dissolved methane was detected in 593 water wells, representing
a majority (64%) of the 924 water wells for which analysis for
methane was conducted. The COGCC reporting threshold of
1 mg/L methane was exceeded in 261 wells. The hazard threshold
of 10 mg/L was exceeded in 122 wells, and the hazard mitigation
level of 28 mg/L was exceeded in 5 wells (SI Appendix). Long-
term repeatability in dissolved methane measurements was
assessed from a subset of 92 water wells that underwent follow-
up sampling over a period of up to 20 y. Sample pairwise dif-
ferences varied by 67% (1 SD, n = 136; SI Appendix), which
highlights well-known problems in repeatability of dissolved
methane measurements due to hydrological variability and
sampling and analytical methods (11, 21). Nevertheless, overall
spatial and temporal patterns of dissolved methane provide
valuable insights to origins and associated hazards of ground-
water methane in the DJ Basin (Fig. 1).
Genetic origins of groundwater methane were assessed from
stable carbon isotopes of methane (δ13
CC1) and gas molecular
ratios [C1/(C2 + C3)] in 211 of the water wells (22) (Fig. 2 and SI
Appendix). Dissolved methane concentrations were >1 mg/L in
all but four of the water wells. Of these 211 wells, 169 had
methane with δ13
CC1 < −60‰ and C1/(C2 + C3) > 100, char-
acteristic of microbial methane. Stable hydrogen isotope ratios
(δ2
HC1) further showed that the microbial methane is in-
termediate between the CO2 reduction and acetate fermentation
methanogenic pathways (SI Appendix). Another 29 water wells
contained methane with δ13
CC1 > −55‰ and C1/(C2 + C3) < 50,
characteristic of thermogenic methane. A final 13 wells had
isotopic and molecular values representing mixed microbial-
thermogenic methane [δ13
CC1 < −55‰ and C1/(C2 + C3) < 100]
Denver
Wyoming Nebraska
Kansas
Colorado
Denver
Dissolved CH4 (mg/l):
not detected
>0 - 1
>1 - 10
>10 - 28
>28
0 25 50 75 10012.5
Kilometers
State
County
DJ Basin
Wattenberg Field
Coal Fields
Coal-Bearing Areas
Wrench Faults
Dawson
Denver
Arapahoe
Laramie
Laramie-Fox Hills
Dakota-Cheyenne
High Plains
Aquifer Formations:
Thermogenic CH4
!
Oil/Gas Wells
Boulder-Weld
Coal Field
Scranton
Coal Field
Coal Samples
B
A
Fig. 1. Study area in the DJ Basin of Northeastern Colorado. (A) Overview
map showing distribution of dissolved methane and thermogenic gas oc-
currences in 924 groundwater wells in relation to aquifers and oil and gas
wells. (B) Closeup of Wattenberg field, also showing the distribution of coal-
bearing deposits and coal mining fields (23), and northeast-trending wrench
faults (12). Location of coal samples discussed in text is shown in both panels.
Aquifers of the Denver Basin aquifer system (Dawson through Laramie-Fox
Hills) are listed from youngest (Top) to oldest; Dakota-Cheyenne and High
Plains represent separate aquifer systems (44). Fig. 2. Genetic characterization plot (22) of C1/(C2 + C3) vs. δ13
CC1 for
groundwater aquifers compared with natural gases from producing for-
mations in Wattenberg field. Repeat samples from water wells are included.
Groundwaters fall within microbial and thermogenic domains and along a
mixing line (bold curve) calculated for thermogenic [δ13
CC1 = −46‰; C1/(C2 +
C3) = 5] and apparent Laramie-Fox Hill (LFH) microbial end-member values
[δ13
CC1 = −72‰; C1/(C2 + C3) = 1,000]. Arrows show oxidation trends using
indicated fractionation factors (αCH4-CO2), following calculations in ref. 22.
There were no data for Dawson or High Plains aquifers.
2 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1523267113 Sherwood et al.

3. and fell along a mixing line between the microbial and thermogenic
end-members. Six samples had δ13
CC1 and C1/(C2 + C3) values that
might be considered impacted by oxidation (Fig. 2). Repeatability of
δ13
CC1 measurements (1.2‰, n = 63; SI Appendix) was excellent.
Repeatability of C1/(C2 + C3) was poorer (42%, n = 76; SI Appendix),
but this would affect determination of genetic origin in less than 5%
of samples (Fig. 2). Median (±bootstrapped 95% confidence limits)
concentrations of dissolved methane categorized by genetic origin
increased in the following order: unknown (0.03 +0.00=−0.01 mg/L),
thermogenic (7.1 +2.8=−1.6 mg/L), microbial (9.2 +0.8=−0.6 mg/L),
mixed (12.5 +2.0=−2.5 mg/L) (SI Appendix).
Spatial and stratigraphic relationships suggest that the micro-
bial methane is sourced from shallow coal seams (Figs. 1 and 3
and SI Appendix). Microbial methane occurred primarily in wells
screened in the Denver aquifer (181 ± 10 m depth), which contains
lignite, and the confined part of the Laramie-Fox Hills aquifer
(189 +10=−8 m), which is interbedded with subbituminous coal
(23, 24). Within the confined Laramie-Fox Hills aquifer, ele-
vated methane concentrations occur in a region overlying and
extending down-dip of the Boulder-Weld coal field; within the
Denver aquifer, elevated methane occurs near the Scranton
coal mining district (Fig. 1). Reports of gas venting, suffoca-
tions, fires, and explosions from coal mines, and gas-in-place
measurements of up to 24 ft3
/ton indicate coalbed methane
potential in this area, with a total resource estimated at 2 trillion cubic
feet (23, 24). Samples of Laramie formation coals from 185 to 234 m
depth contained methane with δ13
CC1 = −69 ± 0.2‰ (n = 5; SI
Appendix). This value is consistent with dissolved methane in the
confined Laramie-Fox Hills aquifer at similar depths and supports a
microbial, not thermogenic, origin of the coalbed methane. Note also
the subbituminous B/C rank of Laramie formation coal throughout
the basin is inconsistent with thermogenic methane production (23–
25). Methane in Denver formation lignites has not been analyzed;
however, the more depleted values of δ13
CC1and δ2
HC1 in Denver
aquifer dissolved methane (Fig. 2 and SI Appendix) are consistent
with the isotopic composition of lignites in general (26).
Although coal is a likely source of microbial methane in the
Laramie-Fox Hills aquifer, we cannot rule out the possibility that
the microbial methane may also originate from slightly deeper
gas-bearing reservoirs. Gas from two water wells drilled to the
so-called “1100 foot sandstone” at depths of ∼381 m had
δ13
CC1 = −67.09 and −67.26‰ and C1/(C2 + C3) = 310 and 711
(SI Appendix). These values are slightly isotopically heavier and
chemically wetter than gas in the Laramie-Fox Hills aquifer.
Methane occurrence is also controlled by groundwater redox
conditions. Both the Denver aquifer and the confined Laramie-
Fox Hills aquifer contain Na-HCO3 waters with low sulfate (SO4)
concentrations (≤30 mg/L; SI Appendix), which is compatible with
microbial methanogenesis (22, 26). In contrast, the unconfined
Laramie-Fox Hills (the part of the Fox Hills formation not over-
laid by the Laramie formation) contains Ca-Na-HCO3-Cl water
with high sulfate (>300 mg/L), similar to groundwater in the over-
lying Quaternary alluvium and in the Dakota-Cheyenne aquifer in
the northern half of Wattenberg field (SI Appendix). Higher sulfate
in these aquifers is considered to be incompatible with microbial
methanogenesis (22, 26).
Based on δ13
CC1 and C1/(C2 + C3) criteria described above, a
total of 42 water wells, all located in or near the Wattenberg
field, had thermogenic or mixed microbial-thermogenic methane
detected at least once in their sampling history (SI Appendix).
The thermogenic gas comes from underlying (>1,000 m deep)
production reservoirs, as indicated by δ13
CC1 and C1/(C2 + C3)
compositions that overlap with or fall along a mixing line extending
to gases of the Sussex, Codell, Niobrara, or J-Sand formations (Fig.
2). Measurements of the δ13
C of ethane (δ13
CC2) and propane
(δ13
CC3) from water wells, where available, also overlap with that
of production formations (Fig. 4). Because microbes produce very
little ethane and propane, thermogenic ethane and propane are
not significantly diluted by the presence of microbial gas (22) (SI
Appendix). Because of this, δ13
CC2 and δ13
CC3 can be used to trace
the source of stray gas to a specific oil and gas reservoir, provided
that the isotopic signature of the source reservoir is known and no
isotopic fractionation or mixing has occurred during migration and
accumulation (21, 27, 28) (Fig. 5). Measurements of δ13
CC2 and
δ13
CC3 also rule out the possibility that heavier δ13
CC1 in the water
samples are caused by oxidation (22). Finally, C2–C6 molecular
Fig. 3. Plot of dissolved methane concentrations vs. groundwater aquifer,
colored by methane genetic origin. Underlying vertical lines and boxes repre-
sent median ± 95% confidence limits (SI Appendix). Numbers to right of each
data series represent number of samples. Repeat water well samples included.
−80−70−60−50−40−30−20
1/Carbon Number
δ13
C(‰)
0.33 0.5 1
C3 C2 C1
Groundwater
Sussex
Niobrara
Codell/Niobrara
J Sand/Codell/Niobrara
Codell
J Sand/Codell
J Sand
Thermogenic
Microbial
Mixing
Fig. 4. Natural gas isotope plot (45) of δ13
C vs. reciprocal carbon number for
groundwater aquifers compared with production gas formation averages.
δ13
CC2 and δ13
CC3 more clearly distinguish between microbial and thermogenic
gases compared with δ13
CC1. Water samples with thermogenic isotope signa-
tures overlap with production gases. Repeat water well samples included.
Sherwood et al. PNAS Early Edition | 3 of 6
ENVIRONMENTAL
SCIENCES

4. compositions of thermogenic gases in water wells also overlap with
those of production gases (SI Appendix).
Due to clustering, the 42 water wells with thermogenic gas
represent 32 separate cases of stray gas migration (SI Appendix).
The 32 cases are based on complaint reports filed with the COGCC
due to water quality concerns, bubbles in water, explosions, or
thermogenic gas detected during baseline sampling. In six of the
cases, two to four water wells in close proximity and screened in the
same aquifer contained thermogenic gas, thus comprising a cluster
impact. Complaint files include information about sequence of
events, investigations of nearby oil and gas wellbore construction
and integrity [e.g., surface casing vent, or “bradenhead,” pressures,
mechanical integrity tests (MITs), cement bond logs, production
volumes], and resolution, if any (SI Appendix). This contextual in-
formation is critical as geochemical evidence alone cannot pinpoint
specific pathways of gas migration (3, 21). In 10 of the 29 com-
plaints for which documentation exists, thermogenic stray gas was
attributed to barrier failures in nearby oil and gas wells and resulted
in Notices of Alleged Violation (NOAV) and/or remediation orders
issued by the COGCC (SI Appendix). In one case, wellbore failure
was suspected, but never confirmed by the COGCC. In three of the
complaints, the landowner settled with an oil and gas operator
privately, and no information about the cause of gas migration was
available. The remaining 15 complaints were unresolved or are still
under investigation by the COGCC.
Although low in number, the 11 cases of confirmed or suspected
oil and gas wellbore barrier failure inform discussions about path-
ways of thermogenic stray gas migration (10, 29–32) (SI Appendix).
Wellbore designs in the DJ Basin have evolved with the history of
COGCC regulations and operator practices. Wells with the highest
probability of failure have “short” surface casings, a legacy of an
earlier (pre-1993) regulatory era when surface casings were not set
deep enough to protect aquifers not in use at the time (20, 33) (Fig.
5). These wells also have uncemented sections of production casing
in which hydrostatic pressure of fluid in the wellbore annulus is
the only barrier to vertical migration of gas originating from
intermediate-depth formations (20) (Fig. 5). All 11 cases of
wellbore failure involved vertical wells drilled before 1993, 7 of
which were hydraulically fractured. All 11 wells had short surface
casings and uncemented intermediate sections. Six wells also had
casing leaks revealed by MIT failure and one well had a wellhead
seal leak; however, there likely would have been no impact to
groundwater had these wells been constructed with sufficiently
deep surface casings. These results highlight the importance of
“regulatory failure” to protect groundwater quality. All of the
failed wells underwent remedial cementing and were returned to
production (six wells) or plugged and abandoned (five wells) (SI
Appendix). Insufficient data were collected for assessment of
groundwater contamination levels after remediation.
Based on COGCC investigations, none of the oil and gas
wellbore failures involved horizontal, hydraulically fractured
wells, all of which have been drilled since 2010 (SI Appendix).
This result is likely due to the fact that, since 1993, surface
casings are required by COGCC regulations to extend at least 50
ft (15 m) below the deepest potable aquifer. The majority (89%)
of horizontal wells also have production casings that are either
fully cemented or cemented above the shallowest hydrocarbon-
bearing formation (20). This evidence supports the growing con-
sensus that wellbore barrier failure, not the process of high-volume
hydraulic fracturing itself, is the main thermogenic stray gas mi-
gration pathway (1, 3, 9, 10, 20, 27, 29, 30).
The 11 documented cases of wellbore failure also inform the
debate about setback distances. Impacted groundwater wells
were all located within 1 km of the known wellbore failure (SI
Appendix), a result generally consistent with thermogenic gas
occurrence in the Marcellus shale area of Pennsylvania (4, 6) and
the Barnett shale area in Texas (34). It is possible that the 15
unresolved cases were caused by wellbore failures beyond this
distance because COGCC investigations typically focus on oil
and gas development within a half mile (0.8 km) of an impacted
water well; however, the number of affected water wells drops off
rapidly within this distance (SI Appendix).
Reasons for not determining the source of stray gas include
missing or incomplete data or documentation in COGCC ar-
chives, inconclusive geochemical data (problematic when sam-
ples were not measured for δ13
CC2–3), lack of wellbore failures
identified within the COGCC-investigated search radius, and
failure to measure bradenhead pressure or conduct mechanical
integrity testing on all wells within the search radius. Based on
remedial cementing records and bradenhead pressure data,
Fleckenstein et al. (20) identified 388 of 11,617 wells with short
surface casings and uncemented intermediate sections in
Wattenberg field with “possible barrier failures.” Although there
were far fewer (n = 42) water wells with thermogenic stray gas, it
seems probable that wellbore barrier failures could have con-
tributed to unresolved cases of thermogenic stray gas migration
even though specific wellbores were never identified. For ex-
ample, case 1 (SI Appendix) from 1988 involved a suspected gas
well that underwent remedial cementing and was then aban-
doned without orders from the COGCC to conduct bradenhead
or MIT testing. The case was therefore never officially resolved,
despite a case narrative that strongly points to this well as the
source of stray thermogenic gas in a nearby water well.
Finally, there is the possibility of fault-assisted gas migration
beyond the apparent 1 km radius of impact, the probability of
which is difficult to assess. To our knowledge, there are no natural
seeps of thermogenic gas in the Wattenberg Field. Listric faults
cutting through the Laramie-Fox Hills formation are believed to
terminate in the upper 500 m of the Pierre Shale (14, 24), where
they could intersect with the uncemented production casings of
25002000150010005000
Depth(m)
−50−45−40−35−30−25
0.33 0.5 1
C3 C2 C1
1/Carbon Number
δ13
C(‰)
gas well
water well
Denver
Upr. Arapahoe
Lwr. Arapahoe
Laramie-Fox Hills
Pierre Shale
Sussex
Niobrara
Codell
J-Sand
cemented
surface casing
(0-65 m)
uncemented
production
casing
(0-2145 m)
cemented
production
casing
(2145-2453
casing leak
(71-99 m)
A
B
surface
casing vent
Gas Well
Water
Well
Fig. 5. Example of confirmed wellbore barrier failure (COGCC Complaint
200097544; SI Appendix). (A) Wellbore diagrams of gas well and water well
(lateral offset = 103 m). Mechanical integrity testing of gas well revealed
leaks in uncemented production casing at 71–99 m depth, below the surface
casing, which was set above the top of the Upper Arapahoe aquifer. Pro-
duction gases from the J-Sand formation thereby migrated to domestic
water well screened in the Upper Arapahoe aquifer. Horizontal scale is ex-
aggerated by 50×. (B) Natural gas isotope plot (45) showing identical δ13
CC1–3
signatures in the gas well and water well.
4 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1523267113 Sherwood et al.

5. older wellbores, thus establishing potential gas migration pathways.
Major wrench faults originating from basement rocks and believed
to be a heat source for the Wattenberg geothermal anomaly (12)
are another possible pathway, although less likely because of ter-
mination of these faults within the Pierre Shale and observation that
stray gas occurrence does not line up along the northeast-trending
wrench faults (Fig. 1). Moreover, stray gases have the same C1–C5
composition as production gases (SI Appendix), which suggests
relatively rapid transport via wellbores because geological seepage
tends to strip gases of heavier alkanes during migration (35).
The COGCC water quality dataset has the unique importance
of providing a 26-y sampling history that spans the time before
and after development of horizontal wellbores with high-volume
hydraulic fracturing beginning in 2010 (Fig. 6). In contrast to a
previous report (36), we find no increase in dissolved methane
concentrations over time (linear regression of dissolved methane
vs. date, excluding nondetects: slope = −0.22, P = 0.002; excluding
samples <1 mg/L: slope = 0.02, P > 0.05). The rate of thermogenic
stray gas occurrence, including both resolved and unresolved cases,
has remained steady at about two cases per year since 2001 (Fig. 6).
The apparent step change around the year 2000 probably reflects
changes in COGCC complaint investigation and reporting proce-
dures. Note that there is no change in the rate around the year
2010, when horizontal drilling with high-volume hydraulic fractur-
ing was introduced. These results further support the conclusion
that horizontal drilling with high-volume hydraulic fracturing has
not had a discernible impact on the origin and occurrence of
thermogenic stray gas in the DJ Basin.
With evidence that 42 separate water wells have been impacted
as of the year 2014, the risk of thermogenic stray gas is estimated
as 0.12% of 35,000 water wells in the DJ Basin (37) (lower esti-
mate) to 4.5% of 924 water wells sampled and analyzed for
methane (upper estimate). Assuming all 32 cases of thermogenic
stray gas originate from oil and gas wellbore failures, the rate of
wellbore failure is estimated at 0.06% of the 54,000 active and
abandoned oil and gas wells across the Colorado part of the DJ basin
(lower estimate) and 0.15% of the 20,700 active and abandoned oil
and gas wells in the Wattenberg field (upper estimate), where most of
the cases of stray gas occurred (Fig. 1). These rates are 1.6–4 times
lower than the fraction of shale gas wells in Pennsylvania that have
led to methane migration into groundwater (0.24%) (9). The contrast
in rates highlights differences in geology, regulations, operators, and
construction practices among different petroleum-producing basins
(29–31). It should also be noted that the occurrence of thermogenic
stray gas is likely underrepresented because of the spatial density of
sampled water, willingness of landowners to have their water tested,
and the possibility that stray hydrocarbons are rapidly oxidized under
certain redox conditions (38). Although the total number of cases of
thermogenic stray gas is relatively low, it is clear from COGCC
complaint reports that individual cases pose significant safety, fi-
nancial, emotional, and health risks to landowners (SI Appendix).
Evaluation, mitigation, and prevention of these impacts should
remain an ongoing high priority (32, 39, 40).
This study demonstrates the value of a large, continuously
updated, and publicly accessible groundwater geochemical da-
tabase from a petroleum industry regulator. Although focused
sampling campaigns can shed light on specific groundwater
contamination incidents and pathways, the true extent and na-
ture of industry-related impacts to groundwater is only revealed
through long-term, regional-scale monitoring. Thus, the COGCC
groundwater monitoring regulations and geochemical data ar-
chive may provide a useful template for other regulatory agen-
cies. The availability of isotopic measurements of ethane and
higher alkanes in particular provides a more robust forensic tool
than methane isotopes alone, which are less diagnostic of genetic
origins. Availability of wellbore integrity data (surface casing
vent pressure and mechanical integrity testing) is critical for
evaluating pathways of stray gas migration. COGCC data show
pervasive and naturally occurring microbial methane in coal-
bearing aquifers of the DJ Basin. A total of 42 water wells
contained thermogenic stray gas, representing 32 separate cases
of contamination, occurring at the rate of two cases per year
since 2001. None of the cases could be specifically attributed to
recent horizontal well drilling or hydraulic fracturing. Assess-
ment of the risk of thermogenic methane release should there-
fore address the full history and life cycle of both conventional
and unconventional oil and gas operations.
Methods
Groundwater geochemical data were downloaded from the COGCC online
Colorado Oil and Gas Information System (COGIS) system (dnrwebmapgdev.state.
co.us/mg2012app/) using custom computer scripts during October of 2014. A
subset of these data containing additional groundwater well metadata was
accessed via bulk download (cogcc.state.co.us/documents/data/downloads/
environmental/WaterWellDownload.html), and the two datasets were merged
based on “FacID” number. Data were limited to locations within the DJ Basin
that could be cross-correlated against Colorado Department of Water Re-
sources (DWR) water well permits (37) based on matching of at least two of
the following parameters: DWR receipt number, DWR permit number, loca-
tion, and depth. Data quality assurance/quality control protocols were modi-
fied from refs. 41 and 42 for large water quality datasets. Data screening
criteria, treatment of nondetects, and analysis of measurement repeatability
are described in SI Appendix. Production gas geochemical data are from 77
production wells in the Wattenberg Field (43). Additional production gas data
in Fig. 5 and in complaint narratives (SI Appendix) were downloaded from
A
B
Fig. 6. Time series plots. (A) Methane concentration vs. sample collection
date, colored by genetic origin. Line shows number of samples measured for
dissolved methane per year. Implementation dates of COGCC regulatory and
Colorado Oil and Gas Association (COGA) voluntary baseline water quality
monitoring shown in top margin (SI Appendix). (B) Cumulative number of
cases of thermogenic stray gas. Numbers indicate the number of impacted
water wells where n > 1.
Sherwood et al. PNAS Early Edition | 5 of 6
ENVIRONMENTAL
SCIENCES

6. the COGCC COGIS system. Sampled well American Petroleum Institute
numbers were cross-correlated to COGCC well production summaries (15)
to identify the source reservoir at the time of sampling. Complaint files
associated with thermogenic stray gas were accessed from the COGCC
Complaints database (cogcc.state.co.us/complaints2.html#/searchcomplaints).
Complaints were searched on public land survey system location (section,
township, range) or by well owner. Specifics to each case were interpreted from
the associated complaint documentation (SI Appendix). Because the COGCC pro-
vides limited functionality for filtering and querying online data, we built a data
portal with improved functionality for querying COGCC environmental samples,
facilities, inspections, complaints, and other information where all of the data and
information presented in this paper can be searched and viewed (https://data.
airwatergas.org/content/query-tools/cogcc-database).
Stable carbon isotope data are reported in delta notation, where
δ13
C = ½ð13
C=12
CÞsample=ð13
C=12
CÞVPDB − 1Š × 1,000 and δ2
H = ½ð2
H=1
HÞsample=
ð2
H=1
HÞVSMOW − 1Š × 1,000, Vienna Pee Dee Belemnite (VPDB) and Vienna
Standard Mean Ocean Water (VSMOW) are standards. Because of nonnormal
data distributions, summary statistics were computed using statistical boot-
strapping (n = 1,000) and are reported as the median +95% CI=−95% CI.
ACKNOWLEDGMENTS. We thank the Colorado Oil and Gas Conservation
Commission for providing access to data and Erica Wiener for assistance
with data management. The manuscript was improved by the comments
of three anonymous reviewers. This work was supported by the National
Science Foundation Sustainability Research Network program (Grant CBET-
1240584).
1. Vengosh A, Jackson RB, Warner N, Darrah TH, Kondash A (2014) A critical review of
the risks to water resources from unconventional shale gas development and hy-
draulic fracturing in the United States. Environ Sci Technol 48(15):8334–8348.
2. Vidic RD, Brantley SL, Vandenbossche JM, Yoxtheimer D, Abad JD (2013) Impact of
shale gas development on regional water quality. Science 340(6134):1235009, 10.1126/
science.1235009.
3. Baldassare FJ, McCaffrey MA, Harper JA (2014) A geochemical context for stray gas
investigations in the northern Appalachian Basin: Implications of analyses of natural
gases from Neogene-through Devonian-age strata. AAPG Bull 98(2):341–372.
4. Jackson RB, et al. (2013) Increased stray gas abundance in a subset of drinking
water wells near Marcellus shale gas extraction. Proc Natl Acad Sci USA 110(28):
11250–11255.
5. Molofsky LJ, Connor JA, Wylie AS, Wagner T, Farhat SK (2013) Evaluation of methane
sources in groundwater in northeastern Pennsylvania. Ground Water 51(3):333–349.
6. Osborn SG, Vengosh A, Warner NR, Jackson RB (2011) Methane contamination of
drinking water accompanying gas-well drilling and hydraulic fracturing. Proc Natl
Acad Sci USA 108(20):8172–8176.
7. Baldassare FJ, Laughrey CD (1997) Identifying the sources of stray methane by using
geochemical and isotopic fingerprinting. Environ Geosci 4(2):85–94.
8. Jackson RE, et al. (2013) Groundwater protection and unconventional gas extraction:
The critical need for field-based hydrogeological research. Ground Water 51(4):
488–510.
9. Brantley SL, et al. (2014) Water resource impacts during unconventional shale gas
development: The Pennsylvania experience. Int J Coal Geol 126(June 1):140–156.
10. Darrah TH, Vengosh A, Jackson RB, Warner NR, Poreda RJ (2014) Noble gases identify
the mechanisms of fugitive gas contamination in drinking-water wells overlying the
Marcellus and Barnett Shales. Proc Natl Acad Sci USA 111(39):14076–14081.
11. Humez P, et al. (2016) Occurrence and origin of methane in groundwater in Alberta
(Canada): Gas geochemical and isotopic approaches. Sci Total Environ 541:1253–1268.
12. Higley DK, Cox DO (2007) Oil and gas exploration and development along the Front
Range in the Denver Basin of Colorado, Nebraska, and Wyoming. Petroleum Systems
and Assessment of Undiscovered Oil and Gas in the Denver Basin Province, Colorado,
Kansas, Nebraska, South Dakota, and Wyoming (US Geological Survey, Reston, VA),
(USGS Digital Data Series DDS-69-P), Chap 2.
13. Clayton JL, Swetland PJ (1980) Petroleum generation and migration in Denver basin.
AAPG Bull 64(10):1613–1633.
14. Weimer RJ, Sonnenberg SA, Young GB (1987) Wattenberg Field, Denver Basin,
Colorado. Geology of Tight Gas Reservoirs, eds Spencer CW, Mast RE (AAPG, Tulsa,
OK), pp 143–164.
15. COGCC. Oil and gas facilities online database. Available at cogcc.state.co.us/data.
html#/cogis. Accessed March 29, 2016.
16. Sonnenberg S (2013) New reserves in an old field, the Niobrara resource play in the
Wattenberg Field, Denver Basin, Colorado. SPE Unconventional Resources Technology
Conference (Denver, CO). Available at: dx.doi.org/10.1190/URTEC2013-098. Accessed
March 30, 2016.
17. International Engineering Company, Inc. and Amuedo and Ivey, Inc (October, 1984)
Natural Gas Hazard Potential (International Engineering Company, Inc., Denver),
pp 1–57.
18. Denver Post Editorial Board (April 6, 1982) Fiddling while water burns. Denver Post,
Section B, p 2.
19. Rice DR, Threlkeld CN, Vuletich A (1984) Nature and Origin of Vent Gases in the
LaSalle Area, Northeastern Colorado (US Geological Survey, Denver), USGS Open-
File Report 84-220, 11 p.
20. Fleckenstein WW, Eustes AW, Stone CH, Howell PK (2015) An assessment of risk of
migration of hydrocarbons or fracturing fluids to fresh water aquifers: Wattenberg
Field, CO. SPE Kuwait Oil and Gas Show and Conference (Mishref, Kuwait). Available
at dx.doi.org/10.2118/175401-MS. Accessed March 30, 2016.
21. Gorody AW (2012) Factors affecting the variability of stray gas concentration and
composition in groundwater. Environ Geosci 19(1):17–31.
22. Whiticar MJ (1999) Carbon and hydrogen isotope systematics of bacterial formation
and oxidation of methane. Chem Geol 161(1-3):291–314.
23. Roberts SB (2007) Coal in the front range urban corridor—An overview of coal geology,
coal production, and coal-bed methane potential in selected areas of the Denver Basin,
Colorado, and the potential effects of historical coal mining on development and land-
use planning. Petroleum Systems and Assessment of Undiscovered Oil and Gas in the
Denver Basin Province, Colorado, Kansas, Nebraska, South Dakota, and Wyoming (US
Geological Survey, Reston, VA), (USGS Digital Data Series DDS-69-P), Chap 3.
24. Carroll CJ (2009) The coal geology and mining resources of the Boulder-Weld coal
field. Mt Geol 46(1):13–25.
25. Meissner FE (1984) Cretaceous and Lower Tertiary Coals as Sources for Gas Accumulations
in the Rocky Mountain Area, Hydrocarbon Source Rocks of the Greater Rocky Mountain
Region (Rocky Mountain Association of Geologists, Denver), pp 401–431.
26. Bates BL, McIntosh JC, Lohse KA, Brooks PD (2011) Influence of groundwater flow-
paths, residence times and nutrients on the extent of microbial methanogenesis in
coal beds: Powder River Basin, USA. Chem Geol 284(1-2):45–61.
27. Rowe D, Muehlenbachs K (1999) Isotopic fingerprints of shallow gases in the Western
Canadian sedimentary basin: Tools for remediation of leaking heavy oil wells. Org
Geochem 30(8):861–871.
28. Taylor S, Lollar SB, Wassenaar I (2000) Bacteriogenic ethane in near-surface aqui-
fers: Implications for leaking hydrocarbon well bores. Environ Sci Technol 34(22):
4727–4732.
29. Dusseault M, Jackson R (2014) Seepage pathway assessment for natural gas to
shallow groundwater during well stimulation, in production, and after abandon-
ment. Environ Geosci 21(2):107–126.
30. King GE, King DE (2013) Environmental risk arising from well-construction failure—
differences between barrier and well failure, and estimates of failure frequency
across common well types, locations, and well age. SPE Prod Oper 28:323–344.
31. Jackson RB (2014) The integrity of oil and gas wells. Proc Natl Acad Sci USA 111(30):
10902–10903.
32. Birdsell DT, Rajaram H, Dempsey DD, Viswanathan HS (2015) Hydraulic fracturing
fluid migration in the subsurface: A review and modeling results. Water Resour Res
9:7159–7188.
33. Llewellyn GT, et al. (2015) Evaluating a groundwater supply contamination incident
attributed to Marcellus Shale gas development. Proc Natl Acad Sci USA 112(20):
6325–6330.
34. Fontenot BE, et al. (2013) An evaluation of water quality in private drinking water
wells near natural gas extraction sites in the Barnett Shale formation. Environ Sci
Technol 47(17):10032–10040.
35. Clayton CJ, Hay SJ, Baylis SA, Dipper B (1997) Alteration of natural gas during leakage
from a North Sea salt diapir field. Mar Geol 137(1-2):69–80.
36. Li H, Carlson KH (2014) Distribution and origin of groundwater methane in the
Wattenberg oil and gas field of northern Colorado. Environ Sci Technol 48(3):
1484–1491.
37. Colorado Division of Water Resources (2016) Well permit search tool. Available at
www.dwr.state.co.us/wellpermitsearch/. Accessed March 28, 2016.
38. Van Stempvoort D, Maathuis H, Jaworski E, Mayer B, Rich K (2005) Oxidation of fu-
gitive methane in ground water linked to bacterial sulfate reduction. Ground Water
43(2):187–199.
39. Rogers JD, Burke TL, Osborn SG, Ryan JN (2015) A framework for identifying organic
compounds of concern in hydraulic fracturing fluids based on mobility and persis-
tence in groundwater. Environ Sci Technol Lett 2(6):158–164.
40. Adgate JL, Goldstein BD, McKenzie LM (2014) Potential public health hazards, ex-
posures and health effects from unconventional natural gas development. Environ Sci
Technol 48(15):8307–8320.
41. Dahm KG, Guerra KL, Xu P, Drewes JE (2011) Composite geochemical database for
coalbed methane produced water quality in the Rocky Mountain region. Environ Sci
Technol 45(18):7655–7663.
42. Jones D, Mayer B, Main C (2011) Baseline water well testing data assessment. Report
prepared for Alberta Innovates Technology Futures. Available at esrd.alberta.ca/
water/inspections-and-compliance/baseline-water-well-testing-for-coalbed-methane-
development/documents/BaselineWaterWellTestingData-Mar31-2011.pdf. Accessed
January 15, 2016.
43. COGCC (2007) Greater Wattenberg area baseline study. Report prepared for Colorado
Oil and Gas Conservation Commission. Available at cogcc.state.co.us/documents/library/
AreaReports/DenverBasin/GWA/Greater_Wattenberg_Baseline_Study_Report_062007.pdf.
Accessed January 15, 2016.
44. Paschke SS, ed (2011) Groundwater Availability of the Denver Basin Aquifer System,
Colorado (US Geological Survey, Reston, VA), USGS Professional Paper, Vol 1770.
45. Chung HM, Gormly JR, Squires RM (1988) Origin of gaseous hydrocarbons in sub-
surface environments: Theoretical considerations of carbon isotope distribution.
Chem Geol 71(1-3):97–103.
6 of 6 | www.pnas.org/cgi/doi/10.1073/pnas.1523267113 Sherwood et al.