Faults and associated karst collapse suggest conduitsfor fluid flow that influence hydraulic fracturing-induced seismicityElwyn Gallowaya, Tyler Haucka, Hilary Corlettb, Dinu Panaa, and Ryan Schultza,1

aAlberta Geological Survey, Alberta Energy Regulator, Edmonton, AB T8N 3A3, Canada; and bDepartment of Physical Sciences, MacEwan University,Edmonton, AB T5J 4S2, Canada

Edited by Paul Segall, Stanford University, Stanford, CA, and approved August 21, 2018 (received for review May 1, 2018)

During December 2011, a swarm of moderate-magnitude earth-quakes was induced by hydraulic fracturing (HF) near Cardston,Alberta. Despite seismological associations linking these two processes,the hydrological and tectonic mechanisms involved remain unclear. Inthis study, we interpret a 3D reflection-seismic survey to delve into thegeological factors related to these earthquakes. First, we document abasement-rooted fault on which the earthquake rupture occurred thatextends above the targeted reservoir. Second, at the reservoir’s strati-graphic level, anomalous subcircular features are recognized along thefault and are interpreted as resulting from fault-associated karst pro-cesses. These observations have implications for HF-induced seismicity,as they suggest hydraulic communication over a large (vertical) dis-tance, reconciling the discrepancy between the culprit well trajectoryand earthquake hypocenters. We speculate on how these newly iden-tified geological factors could drive the sporadic appearance of inducedseismicity and thus be utilized to avoid earthquake hazards.

hydraulic fracturing | induced seismicity | hydraulically active faults |dissolution karst

Induced seismicity is a phenomenon by which stress accumu-lating on a fault is suddenly released via an anthropogenically

triggered shear slip. This phenomenon is well documented, withnumerous cases related to mining, waste-water disposal, reser-voir impoundment, and geothermal development activities (1).More recently, induced earthquakes have also been recognizedas the direct result of hydraulic fracturing (HF) stimulation ofunconventional reservoirs (2–10). One such case occurred insouthern Alberta, Canada, near the town of Cardston (Fig. 1)(11). Starting in December, 2011, anomalous earthquakes oc-curred contemporaneously with the completion of a multistagehorizontal well in the uppermost Wabamun Group (also referredto as the “Exshaw play”). These induced events were of mod-erate magnitude (up to moment magnitude 3), with hypocenterslocated in the shallow crystalline basement (11).Induced earthquakes are thought to require three geo-

mechanical conditions: (i) the presence of a fault, (ii) nearlycritical slip-orientation of the fault, and (iii) a means to perturbstress on the fault past the critical condition (12). In general, thefirst two conditions are well-established requirements for faultrupture nucleation. However, the third point is contentious,since there are multiple anthropogenic means to communicatestress changes. For example, effective stress changes can becommunicated to the fault either through pore-pressure per-turbations transmitted along the damage zone of a mechanicallyactive fault (13) or poroelastically through a (impermeable) rockmatrix (14). For disposal cases, pore-pressure diffusion is oftenthe favorite speculative explanation, since continued injectioninto permeable formations allows plausible communication overgreat distances until the intersection with a critically unstablefault (e.g., refs. 15 and 16). However, for HF, this reasoning iscomplicated by additional conceptual hurdles. For example, HFwells inject into impermeable formations to stimulate reservoirproductivity (17); thus, reasonable pressure perturbations arerestricted to the stimulated reservoir zone proximal to the well

bore [up to hundreds of meters (18–21)]. Furthermore, geo-mechanical modeling reveals that poroelastic changes from HFare transmitted only locally (20). These results are in directcontradiction to HF-related earthquakes that are located kilo-meters away from the closest well bore (e.g., refs. 2, 6, 11, 22, and23). Specific to the Cardston case, earthquakes (located on re-gional arrays) were observed within the uppermost crystallinebasement, ∼1.5 km deeper than the target upper Stettler–BigValley Reservoir zone (SI Appendix, Fig. S1); this observation isfurther complicated by the fact that the fault slip response oc-curred immediately (i.e., ∼1.5–3.0 h) after well stage stimulations(11). Due to geological complexities/unknowns in the subsurface,seismological and geomechanical approaches to explaining thesediscrepancies remain speculative.Based on these complications, we instead look for geological

indications of past fluid migration in a 3D reflection-seismicsurvey that surrounds the Cardston earthquakes (Fig. 1). In thispaper, we describe the methods by which we analyzed this datasetand then bolster our interpretation with additional earthquake, drillcore, and well log data. Together, the intersection of these in-dependent datasets leads us to the conclusion that fault-associatedfluid flow caused the dissolution of the Stettler Formation anhy-drites (Wabamun Group) (SI Appendix, Fig. S1) and resulted inkarst features that affected accommodation patterns in the overlyingupper Stettler/Big Valley/Exshaw interval near the Cardston hori-zontal well (CHW) location. This interpretation has implications forunderstanding the induced seismicity caused by the CHW: The in-

Significance

Induced earthquakes can be caused by hydraulic fracturing(HF). However, the exact means by which stress changes aretransferred to seismogenic faults are unknown. This paperprovides evidence that a case of induced earthquakes insouthern Alberta responded to increased pore pressure on afault in hydraulic communication with the HF operation.Reflection-seismic and drill core data provide evidence thatfluid flow along this fault caused strata underlying the targetreservoir to dissolve, causing a karst collapse in the geo-logical past. We suggest that seismogenic and hydraulicallyactive faults are geologically rare and that the injectionof fluid directly into them is even rarer, potentially explain-ing the small percentage of HF wells that cause inducedearthquakes.

Author contributions: R.S. designed research; E.G., T.H., H.C., D.P., and R.S. performedresearch; E.G., T.H., H.C., and R.S. analyzed data; and E.G., T.H., H.C., D.P., and R.S. wrotethe paper.

The authors declare no conflict of interest.

This article is a PNAS Direct Submission.

Published under the PNAS license.1To whom correspondence should be addressed. Email: Ryan.Schultz@aer.ca.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1807549115/-/DCSupplemental.

Published online October 8, 2018.

www.pnas.org/cgi/doi/10.1073/pnas.1807549115 PNAS | vol. 115 | no. 43 | E10003–E10012

ENVIRONMEN

TAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

20,

202

0

dication of fault-associated fluid flow is inferential evidence of apotential permeable pathway through which pore-pressure pertur-bations can be transmitted quickly and over great distance. Wespeculate on the potential implications of this interpretation forHF-related earthquakes, including the paucity of seismogenicwells at a basin scale (24, 25), and for fault-associated fluid-flow features as indicators of areas which may be prone toHF-induced seismicity.

Tectonic SettingThe area under investigation is in the southern Alberta Plains,∼10 km east of the triangle zone of the foothills, which repre-sents the approximate eastern limit of the Cordilleran defor-mation front. The sedimentary succession comprises Cambrianto Paleocene strata of the Western Canada Sedimentary Basin(WCSB), which overlie the Precambrian crystalline basement ofthe North American craton. The strata of the WCSB were de-posited in three different general tectonic settings: a passivecontinental margin from the Neoproterozoic to Middle Devo-nian, a back-arc setting from the Late Devonian to Jurassic, anda foreland basin between the Jurassic and Eocene. Far-fieldstress related to the purported Late Devonian–Early Carbonif-erous Antler and the Jurassic-to-Early Eocene Cordillerancontraction to the west may be responsible for faulting in theinvestigated area. Postorogenic isostatic reequilibration of theCordilleran foreland system was accompanied by the removalof two kilometers of strata in the Alberta Plains (26), which mayhave triggered minor Late Cenozoic faulting or fault reactivation.Quaternary deposits cover the bedrock.First, we considered public 2D reflection-seismic transects

from Canada’s LITHOPROBE study, thereby placing the studyarea into a regional context (Figs. 1 and 2). Specific to southernAlberta, the Southern Alberta Lithospheric Transect (SALT)includes 290 km of reflection-seismic lines acquired in 1995 be-tween the Cordilleran deformation front and the western side ofthe Sweetgrass Arch in southern Alberta and northern Montana(27). This portion of the SALT consists of two east–west lines:line 30 (110 km) and line 31 (140 km), tied by the roughly north–south line 29 (40 km). The SALT data revealed a subhorizontallayer-cake structure of the WCSB stratigraphy affected by ex-tensional faults dipping steeply dipping down to the west (Fig. 2).Subvertical faults spaced 11–15 km apart cut through a

prominent near-basement reflector. These faults offset Cam-brian, Devonian, Mississippian, Jurassic, and Cretaceous strataand appear to be sealed by undeformed Campanian Belly RiverGroup strata (27). Lemieux et al. (28) inferred that the strike ofthese faults observable on the SALT data in the WCSB cover issimilar to the basement magnetic trends, and therefore theirpositioning would be basement-controlled.

SALT 30SALT 30

St. M

ary Rive

r

Belly R

iver

St. Mary

ReservoirS

3D SeismicStudy Area

WSO

FW

SOF

Cardston

49.6° N

49.4° N

49.2° N

113.4° W 113.2° W 113.0° W 112.8° W

0 5 10 15 km

CHWCHW

Fig. 1. Map of the study area. The location of the HF well responsible forthe induced earthquakes (yellow star), 3D reflection-seismic survey (blackbox), SALT 2D reflection-seismic line (black line), and a previously inferredfault (the WSOF) (red dashed line) are displayed alongside local municipal-ities (gray areas), roads (gray lines), and waterbodies (blue). (Inset) Mapshows the location of the study (white star) in North America.

lower Exshaw

Rundle

Precambrian

-1.5E+009-1E+009-5E+00805E+0081E+0091.5E+009

Reflectiv ity

0 5 10 15 km

3D seismicsurvey area(projectedonto line)

-500

-1000

-1500

-2000

TIME

(ms)

Fig. 2. LITHOPROBE seismic-reflection SALT line 30. The public reflection-seismic data from SALT line 30 have been marked with stratigraphic horizons(subhorizontal colored lines), previous fault interpretations (F1–F4 and red lines), and the projection of the southward 3D reflection-seismic survey (yellowbox) onto the east–west–trending transect along the strike of the WSOF.

E10004 | www.pnas.org/cgi/doi/10.1073/pnas.1807549115 Galloway et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

Evidence from a 3D Reflection-Seismic SurveyTo begin examining the geological controls of the Cardstonearthquakes, proprietary 3D reflection-seismic data were utilizedto characterize the subsurface around the CHW (Fig. 1). The 3Dreflection-seismic dataset, originally acquired in 1998, covers60 km2. This study used a version processed with surface-consistent deconvolution, finite difference migration, and noiseremoval by FXY deconvolution. The data were processed into50 × 50 m bins. The bandwidth of the interpretation volumedecreases with depth: the bandwidth at 750 ms is 10–70 Hz (witha dominant frequency of 55 Hz), while the bandwidth at 1,500 msis 11–65 Hz (with a dominant frequency of 40 Hz). Comparisonswith synthetic seismograms created from well logs suggest thephase of the reflection-seismic volume is reasonably stable andaverages near 0°.

Synthetic seismograms were generated for 10 nearby wellswith geophysical logs and were tied to the nearest observedreflection-seismic arrival. This process incorporated datathroughout the sedimentary sequence and resulted in a velocitymodel of the sediment down to the Precambrian basement.Below this, the crystalline basement lacks continuous interfacesof contrasting acoustic impedance; thus, the basement interval ispoorly suited for characterization by reflection-seismic data.Furthermore, the reflections that appear below the Precambrianhorizon on the 2D and 3D data are interpreted as being seismicmultiples and should be ignored. Based on horizon correlations,we examined stratigraphic surface attributes including coher-ence, curvature, and topography (see SI Appendix for details ofanalysis) (29, 30). Two significant structural features are de-tected by the 3D reflection-seismic data near the CHW: a linearfeature trending approximately south–southeast (SSE) and a

A A’ A A’

Dep

th (m

asl)

0

+

horiz

on c

urva

ture

A A’

Fig. 3. 3D reflection-seismic cross-section. Raw (Left) and interpreted (Middle) reflection-seismic cross-line data (gray area). Locations of earthquakes (redcrosses), the 14-21-4-25W4 vertical well (black line), stratigraphic markers (subhorizontal colored lines), and lineament interpretations (red/pink lines/arrows)are shown. Errors in earthquake locations are on the order of 600 m vertically and 300 m laterally, which is reflected in the variability of hypocenter locations.(Right) Corresponding maximum positive curvature attributes (blue area) for each horizon with callouts to locations of the cross-line data (A–A′, red line), thelineament interpretations (red/pink arrows), and the vertical (black circles) and lateral extent (black line) of drilled wells. Subcircular Exshaw Formation/Wabamun Group anomalies can be seen east of the CHW in the lower Exshaw Formation curvature horizon. An enlarged view of the Precambrian curvaturehorizon (dashed box) more closely shows the lineament interpretations with respect to the earthquake epicenters.

Galloway et al. PNAS | vol. 115 | no. 43 | E10005

ENVIRONMEN

TAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

20,

202

0

small number of anomalous subcircular depressions of the lowerExshaw Formation reflector localized on the linear feature (Figs.3 and 4).The appearance of a linear, approximately SSE-trending fea-

ture in the curvature attribute maps is persistent in numerousstratigraphic horizons from the top of the Belly River Group (theshallowest interpreted horizon) down to the Precambrian base-ment (Fig. 3). Due to its spatial overlap with the Cardstonearthquakes and CHW (Fig. 3), we interpret the linear feature asthe same West Stand-Off Fault (WSOF) that was previouslyconjectured to host the Cardston earthquakes (11). In the 3Dreflection-seismic dataset, we observe that the WSOF strikes at∼170° across the entire 3D reflection-seismic survey (∼4.8 km)and dips at ∼80° (to the west) through an interval of at least2,300 m. The reflection-seismic data have limited vertical reso-lution, so this feature does not create a detectable offset ofreflection-seismic horizons and therefore is not apparent in co-herence attribute volumes. Given the dominant frequencies ofthe interpretation volume and interval velocities measured inwell 14-21-4-25W4, the vertical resolution of the reflectionseismic ranges from 15 m at 750 ms to 35 m at 1,500 ms, as-suming the vertical resolution is approximately one-quarter ofthe dominant wavelength (31). Although distinct faults are not

imaged, a zone of strain manifested as flexure of strata along alinear trend is captured by the reflection-seismic horizons. Theanomalous trend of horizon flexure identified by the curvatureattributes is interpreted as the manifestation of the WSOF in thereflection-seismic data.The localized anomalous subcircular features visible on the

lower Exshaw Formation curvature and depth structure maps aswell as on the isopach maps for the Rundle-to-lower ExshawFormation and lower Exshaw Formation-to-Precambrian inter-vals vary in diameter and morphology (Figs. 3 and 4). The largestof these features is ellipsoid with major and minor axes of∼1,600 m and 600 m, respectively; the depression appears to beup to 40 m deeper than the adjacent area. We note that thelateral positioning of these anomalies is spatially coincident withan overall thinning of the interval between the lower Exshaw For-mation and Precambrian (Fig. 4). Given the limited vertical extentof the seismic anomalies, concomitant thickening and thinning isconstrained to the Exshaw Formation/Wabamun Group. Finally,these anomalies are clustered and represent only a small portion(∼2%) of the total area covered by the 3D reflection-seismic data.In fact, the clustered distribution of the Exshaw Formation/Wabamun Group anomalies and thinned strata underlying the

113°18'W 113°15'W 113°12'W 113°09'W

113°18'W 113°15'W 113°12'W 113°09'W

49°2

0'N

49°2

2'N

49°20'N49°22'N

765770775780785790795800805810815820825830835840845

340

350

360

370

380

390

400

410

420

-2560-2520-2480-2440-2400-2360-2320-2280-2240-2200-2160

-1760-1720-1680-1640-1600-1560-1520-1480-1440-1400-1360

0 2000 4000m-1400

-1360

-1320

-1280

-1240

-1200

-1160

-1120

-1080

-1040

-1000

14-05-005-25-W4

09-29-004-25-W4 (CHW)14-2

1-00

4-25

-W4

Depth Structure (masl) Isopach (m)Ru

ndle

low

er E

xsha

wPr

ecam

bria

n

N

Rundle - lower Exshaw

lower Exshaw - Precambrian

Fig. 4. Processed seismic-reflection horizons. Maps of the topography of three reflection-seismic horizons (Left) are shown alongside the thickness betweenthese intervals (Right). The locations of nearby wells (black circles/lines) and earthquake epicenters (red crosses) are displayed for geographic context.

E10006 | www.pnas.org/cgi/doi/10.1073/pnas.1807549115 Galloway et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

Exshaw Formation (coincident with the WSOF) is suggestive ofsome localized process being the causal mechanism.

Evidence from Well Logs and Drill CoresThe stratigraphic interval of interest for this part of the studyincludes the Devonian Wabamun Group and the Exshaw andBanff Formations (SI Appendix, Fig. S1). The Wabamun Groupis divided into the Stettler and Big Valley Formations, whichcomprise predominantly evaporites with lesser carbonates andcarbonates, respectively (SI Appendix, Fig. S1). The StettlerFormation can be further subdivided into a lower Stettler an-hydrite and a relatively thin upper Stettler carbonate. The latteris overlain by limestone of the Big Valley Formation, which maybe erosionally absent in the study area. The Wabamun Groupstrata are unconformably overlain by the black shales of theExshaw Formation, which can be subdivided into an informallower Exshaw Formation (black shales), and an upper ExshawFormation comprising a silty carbonate (SI Appendix, Fig. S1).This is in turn overlain by shale assigned to the Banff Formation(Banff A in SI Appendix, Fig. S1).The 3D reflection-seismic dataset revealed the presence of a

series of subcircular anomalies at the Exshaw Formation/WabamunGroup interval (Fig. 3). Because the anomalies are spatially limitedin extent (Fig. 4), it was conjectured that they may be features as-sociated with karst-related dissolution of the underlying StettlerFormation anhydrite (herein we refer to karst as being a processoperating only in the anhydrite of the Stettler Formation). Weinferred concomitant subsidence within the overlying carbonates ofthe upper Stettler and Big Valley formations through collapse of thekarst-related dissolution features in the Stettler anhydrite, resultingin increased accommodation and overthickening of the overlyingExshaw Formation, which is 22.4 m thick in well 14-24-4-25W4. Tosubstantiate this, we have considered geological drill core and welllog data. Unfortunately, no drill cores from this interval exist fromthe CHW area; in their absence, nearby analogs have been in-vestigated instead. To find these analogs, we have consideredstratigraphic correlations and isopach mapping of the Exshaw, BigValley, and Stettler Formations within 184 wells (Fig. 5). At least

18 analog wells where the Exshaw Formation is thicker than 10 mwere identified in the study area (Fig. 5) (a thickness of >10 m wasused as a metric to identify potential wells with analogous core)(Fig. 6). From these analog wells, four drill cores were availablewithin the study area (Fig. 5). In these overthickened ExshawFormation wells, there is commonly an associated decrease in thethickness of the underlying Stettler Formation anhydrite, althoughthe lack of deeper well control prohibits the thorough quantificationof the thickness of these evaporites within the study area [only 91 ofthe 187 total wells fully penetrate the Stettler Formation (SI Appendix)].First, to observe examples of an intact succession, areas out-

side (well 13-5-4-25W4) and on the edge (well 4-24-7-22W4) ofExshaw Formation overthickening were reviewed within two drillcores. The lower Stettler Formation in these areas comprisessabkha- and/or salina-type anhydrite with interbedded dolola-minites. The upper Stettler, typically 5–8 m thick, comprisesdololaminites and microbial laminites, with common organic-rich laminations and peloid grainstones. The Stettler Forma-tion is overlain by the mid- to distal-ramp deposits of the BigValley Formation, typically less than 2 m thick or absent beneaththe lower Exshaw Formation and comprising dark-colored shaleswith bioclastic nodular lime mudstone-wackestone and commoncrinoid columnals. The Big Valley Formation is abruptly overlainby the laminated organic-rich black shales of the lower ExshawFormation, which in turn are overlain by the silty dolostones ofthe upper Exshaw Formation. In areas outside the overthickenedExshaw Formation, the thickness of the combined lower andupper members ranges from 3.2 to 6.7 m, although this is alsovariable due to absence of the upper Exshaw Formation memberin parts of the study area (32). The interpretations in this study ofthe sedimentology and stratigraphy from these drill cores areconsistent with prior studies (32, 33).Four drill cores were examined in areas where the Exshaw

Formation thickness is greater than 10 m (e.g., well 6-16-6-22W4) (Fig. 6). In three of the drill cores, the upper Stettler andBig Valley Formation are dominated by intervals of intraclastbreccia (Fig. 6 B and C) or dipping and offset laminations whereunbrecciated. Many of the unbrecciated intervals are highly

St. Mary Reservoir

Oldman River

Milk River

De formation F ront Belly Rive

r

St. Mary

Rive

r

SALT 30

SALT 31

SA

LT 2

9

F1 F2

F3 F4

SA

LT 3

2

WSO

F

CHW

T1

T5

T10

T13

49º N

50º N

R17W4R20R25R29114º W 113º W

A

A'

N

6-16-6-22W44-24-7-22W4

8-24-11-28W4

1-20-1-24W4

F5

13-5-4-25W4

Lethbridge

Magrath

Cardston

Glenwood

4-21-4-25W4 (no core)Canada

Fig. 5. Map of drill core and well log data. Locationsof wells with logs reaching as deep as the WabamunGroup are plotted as gray squares; wells which havean Exshaw Formation logged thicker than 10 m arehighlighted in red around a gray square. Reviewedcores are labeled with their unique well identifier(e.g., 6-16-6-22W4), except for well 14-21-4-25W4,which had no core but exhibits Exshaw Formationoverthickening nearby the CHW. The CHW thatcaused the HF-induced earthquakes (star) is dis-played alongside the extent of the Lower Banff–Exshaw–Big Valley administrative pool boundaries(green area) and all other HF wells completed in thisplay (black dots). For geographic context, the 3Dreflection-seismic area (hatched area), SALT lines(black lines), and previously identified faults (F1–F5,red stars; and WSOF, dashed red line) have been in-cluded. The log section from A–A′ (blue line) is dis-cussed further in Fig. 6.

Galloway et al. PNAS | vol. 115 | no. 43 | E10007

ENVIRONMEN

TAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

20,

202

0

microfractured (Fig. 6A). Clasts within brecciated units of theupper Stettler Formation comprise faintly bedded to microbiallaminated dolostones and common dolomudstones. Clasts aredominantly angular, with some subrounded occurrences, andrange widely in size. Some breccias are poorly sorted (Fig. 6C),whereas others display more common clast size grading andpossible flow banding (Fig. 6B). Within drill cores where thelower Exshaw Formation was recovered, inclined laminationswere observed in the lower parts of the unit. Furthermore, onedrill core from an overthickened Exshaw Formation occurrence(well 8-24-11-28W4) (Fig. 5) displayed pervasive secondary do-lomite overprinting in the Big Valley Formation and possiblypart of the upper Stettler Formation. The dolomitization isrecognized by a medium-to-dark gray coloration relative to thebeige dolostones of other observed breccias. The dolomitizationresults in partially fabric-destructive textures that obscure theappearance of brecciation except where intraclast ghosts andpartially corroded rims of clasts are common (SI Appendix, Fig.S2). Overall, these observations are suggestive of fluid flowsalong faults and are responsible for (i) dissolution-related karst ofevaporites and associated accommodation patterns in the over-lying Exshaw Formation interval, which manifest as the subcircularreflection-seismic anomalies and (ii) localized secondary dolomi-tization of carbonates in the uppermost Wabamun Group.

Reflection-Seismic Modeling of the Exshaw Formation/Wabamun Group AnomaliesTo further explore our conjecture regarding the Exshaw Formation/Wabamun Group anomalies, we employed forward modeling to

compare (or reject) depositional case scenarios against the ob-served reflection-seismic waveform amplitudes. Using wirelinelogs for interval transit time (sonic logs) and bulk density, anacoustic impedance log is calculated for a well and convolved witha zero-phase wavelet to produce synthetic seismograms (see SIAppendix for details). The acoustic impedance log of well 14-21-4-25W4 was used as the basis for describing geological scenarios inthe forward modeling because of the proximity of the well to theExshaw Formation/Wabamun Group anomalies and the CHW.Well log interpretation indicates considerable thickening (22.4 m)of the Exshaw Formation at 14-21-4-25W4, which is supported bythe well’s apparent penetration of the edge of a seismic isopachanomaly (Fig. 4). The sonic and density logs of 14-21-4-25W4 weretherefore modified to better represent the background geologyby compressing the Exshaw Formation interval to a locally typicalvalue (5 m).Three scenarios and their seismic responses were considered

(Fig. 7). In each scenario, 35 m of extra accommodation is createdby removing the uppermost portion of the Wabamun Group. Theaspect that distinguishes the three scenarios is the interval which isthickened to fill the extra accommodation created by the disso-lution of underlying anhydrite: (i) syndepositional thickening ofonly the upper Exshaw Formation member, (ii) syndepositionalthickening of both the lower and upper Exshaw Formation, and(iii) syndepositional thickening of only the lower Exshaw Forma-tion member. The cases are differentiated by the forward seismicmodels. Although the Exshaw Formation/Wabamun Group transi-tion is marked by a strong amplitude peak in all cases, a significantamplitude increase is only associated with the Exshaw Formation/Wabamun Group anomaly in cases two and three. In addition, only

?

1750

1800

1750

1800

1850

1775

1800

1825

Depth

m

Depth

m

Depth

mGammaray

Gammaray

Gammaray

Density porosity(sandstone)

Density porosity(limestone)

Density porosity(limestone)

30 15 0 30 15 030 15 0

100/07-08-006-22W4 100/06-16-006-22W4 100/04-24-007-22W4

1825

1775

1825

1775

Stettleranhydrite

Big Valley

LowerExshaw

UpperExshaw

Banff

Stettler

A A'2 km 11.3 km20

00De

pth m

31.9 km

Gammaray

Density porosity(limestone)

100/14-21-004-25W4

30 15 0

2025

2050

2075

22.4 m26.2 m

7.8 m 11.3 m

2 cm2 cm 2 cm

A B C

ABC

Fig. 6. Cross-section A–A′ and images of drill cores.(Upper) The cross-section shown in Fig. 5 is displayedon the upper half of this figure, datumned on theupper Exshaw Formation. Well log 100/06-16-006-22W4 is labeled with the stratigraphic nomenclatureused in this study and displays overthickened ExshawFormation (>10 m) relative to nearby well 100/07-08-006-22W4. (Lower) Core photos show microfaultedbrecciation (arrows, A) with possible flow bandingrecognized by faint matrix laminations and subtlealignment of clasts (B) and vertical brecciated clasts(C) from the upper Stettler and Big Valley Forma-tions in well 100/06-16-006-22W4, which are inter-preted as resulting from dissolution-related karst inthe underlying Stettler Formation anhydrite.

E10008 | www.pnas.org/cgi/doi/10.1073/pnas.1807549115 Galloway et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

the forward reflectivity models for cases two and three exhibit acontinuity of the amplitude trough immediately above the ExshawFormation/WabamunGroup peak. Last, cases one and two exhibit anextra peak–trough of moderate amplitude above the Exshaw For-mation/Wabamun Group interface within the area of anomalouslythick Exshaw Formation.An equivalent portion of real reflection-seismic data exhibits

some of the characteristics identified in the forward seismicmodels (Fig. 7): The amplitude of the reflection from the ExshawFormation/Wabamun Group interface is consistent across thesection; the amplitude trough immediately above the ExshawFormation/Wabamun Group peak is continuous; and an extrapeak–trough of moderate amplitude above the Exshaw Forma-tion/Wabamun Group interface within the area of anomalouslythick Exshaw Formation is evident. From these qualitative ob-servations, case three can be dismissed as a viable model for theExshaw Formation/Wabamun Group anomalies. Although casesone and two both resemble the real reflectivity, they each dem-onstrate some aspect that differentiates it from the actual re-sponse. Case one lacks continuity of the amplitude trough abovethe Exshaw Formation/Wabamun Group peak. The amplitudeincrease along the Exshaw Formation/Wabamun Group horizonin the anomaly in case two is not evident in the real reflection-seismic field data. Qualitatively, cases one and two both showreasonable similarity to the reflection-seismic field data. Theforward modeling of the reflection-seismic data demonstratesthat dissolution of Stettler Formation anhydrite and continuousinfill of sediments during Exshaw Formation time can produceanomalies similar to those seen in the field data. Although themodeling does not discern the details of the timing, it suggeststhat overthickening occurs throughout Exshaw Formation time.These results are consistent with the dissolution timing suggestedby the brecciation of uppermost Wabamun Group carbonates inthe overthickened Exshaw Formation drill cores.However, we recognize that this simplistic approach to for-

ward seismic modeling has ignored any seismic acquisition illu-mination effects. Contrary to the zero-offset simplification usedfor this modeling, reflection-seismic data are acquired with someoffset between source and receiver. This can become an issuewhen the geometry of the source–receiver anomaly precludessource energy from reaching or reflected energy from returningfrom the anomalous structural feature. Should the reflection-seismic data be affected by these illumination complications,the magnitude of the reflections in the structural lows of the

Exshaw Formation/Wabamun Group anomalies would appearsmaller than expected. Therefore, caution should be taken whencomparing horizon amplitudes in the anomalies with those out-side the anomalies.

DiscussionEvidence of Faulting in Southwestern Alberta and the WSOF. Pre-vious studies in the southern Alberta Plains have identified (orinferred) extensional faults, which suggest subsequent accom-modation of Late Cretaceous compression. For example, clearevidence of extensional faulting in the study area was providedby the SALT (27, 28). Normal faults (Fig. 2) show obvious breaks(up to 450 m of vertical throw) in reflections in the Paleozoicstrata. The trends of these faults were inferred to be subparallel tothe local northwest–southeast grain of the crystalline basement(27), suggesting that at least some of these faults may be rooted inthe basement and thus may be reactivated basement structures.These previous studies anticipate the presence of extensional

tectonism in southern Alberta in general. Specific to the studyarea, inferential evidence points to the possibility of local ex-tensional faulting, interpreted as the WSOF, in the study area(Fig. 1): (i) basement lineaments derived from magnetic andBouguer gravity maps which may represent Precambrian base-ment discontinuities prone to reactivation (34–36), (ii) gradientsin maps of formation tops (11), and (iii) evidence of earthquakehypocenters induced by the CHW (11).Specific to the 3D reflection-seismic data analyzed here, our

confidence that the identified linear feature represents a genuinegeological structure is bolstered by several factors. First, the orien-tation of the anomaly is not aligned to the acquisition geometry or toany geographical factors. Second, the plane of flexure dips to thewest, translating the curvature anomalies westward for deeper hori-zons, suggesting that the anomaly has geological causes and is not aprocessing error. Finally, the lineament stretches across the entire 3Dsurvey, indicating that this feature is significant in at least the in-vestigated region. Considering these supporting pieces, the in-terpretation of the newly imaged ∼SSE lineament (Figs. 3 and 4) as abasement-rooted extensional fault is likely.Evidence for the commonly invoked Late Devonian–Early

Carboniferous Antler orogeny (e.g., ref. 37) is limited and con-troversial at this latitude in the Cordillera (38, 39). However,Paleozoic tectonism at the western margin of ancient NorthAmerica is recorded by multiple Famennian and Tournaisianunconformities (40). Therefore, the possibility that far-field

overthickenedupper Exshaw

overthickenedExshaw

overthickendlower Exshaw

u. Exshawl. ExshawWabamun

Fig. 7. Forward modeling of the Exshaw Formation/Wabamun Group anomalies. Forward reflection-seismic modeling results are shown for three sce-narios proposed for infill of the subcircular ExshawFormation/Wabamun Group anomalies. (Left) Thesonic log from well 14-21-4-25W4 has been modifiedto approximate regional Exshaw Formation thick-ness and is shown with elevations of the WabamunGroup and the upper and lower units of the ExshawFormation. (Center Left) Three geological models fordepositional infill of the anomalies are shown intime. (Center Right) Respective zero-offset seismicreflectivity sections. (Right) An actual reflection-seismic reflectivity section is shown for comparison.

Galloway et al. PNAS | vol. 115 | no. 43 | E10009

ENVIRONMEN

TAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

20,

202

0

stresses associated with the Antler tectonism triggered faulting ofthe Paleozoic stratigraphy in southern Alberta is not precluded.On the other hand, the west-dipping normal faults depicted bySALT that extend at least into the Upper Cretaceous strataconstitute undisputed evidence for downflexing of the NorthAmerican craton during the well-documented Cordilleran tec-tonic loading (41). The trend of the WSOF identified here issimilar to that of the extensional faults in the Cordilleran fore-land. Furthermore, the WSOF is subparallel to and just west of abasement magnetic lineament (11), a spatial relationship thatmay indicate a genetic relationship. We speculate that theWSOF may be the shallow expression of a Precambrian base-ment discontinuity, most likely reactivated during the forelandstage of the basin’s evolution. We infer this relationship becausethe fault can be traced upwards through the Campanian-agedBelly River Group.Given the presence of a fault, its orientation with respect to in

situ stress conditions dictates its degree of mechanical activity.The stress regime in the WCSB (42) likely supports a reac-tivatable stress state on the ∼SSE-orientated WSOF (Fig. 3), asevidenced by the induced earthquakes (11). Often, mechanicallyactive faults are also hydraulically active (43–45), since fluid flowis established along the damage zone of the shear slip (13, 46,47). For example, fault-associated fluid flow has significantlyinfluenced the spatial distribution of localized “pockmark” fea-tures in strata overlying faults in paleo-basins (48).

Interpretation of the Exshaw Formation/Wabamun Group Anomalies.Localized overthickening of the Exshaw Formation suggestssyndepositional accommodation (49–51). The creation of thisaccommodation space is associated with brecciation of the BigValley and upper Stettler Formations in the examined analogdrill cores (Fig. 6). There is no evidence of subaerial karst withinthe Big Valley and upper Stettler Formation carbonates in thecores examined. Rather, the brecciation is associated with col-located overthickening of the Exshaw Formation and decreasedthicknesses in the underlying anhydrite of the lower StettlerFormation. This suggests that dissolution of anhydrite by fault-associated fluids within this interval led to the collapse or sub-sidence of the overlying carbonates of the Big Valley and upperStettler Formations, resulting in the brecciated textures observed.This interpretation is consistent with the subcircular sinkhole-typemorphology of Exshaw Formation/Wabamun Group anomalies inthe 3D reflection-seismic dataset (Fig. 3).The interpretation of a solution–evaporite collapse origin for

the breccias of the Big Valley and upper Stettler Formations issupported by (i) the wide ranges in clast sizes, (ii) poor sorting,(iii) large rotated clasts, (iv) thickness of brecciated intervals, (v)high frequency of microfracturing and offset sedimentary struc-tures, and (vi) proximity to evaporate strata interpreted as beingpartially removed (Fig. 6) (52, 53). Although previous interpre-tations of some of these breccias includes formation in tidalchannels (51), the above observations coupled with a lack ofbrecciation outside overthickened Exshaw Formation areas ismore consistent with brecciation by subsidence through dissolutionof underlying lower Stettler Formation anhydrite. Furthermore,intraclast brecciation within the Big Valley Formation—interpretedas being deposited in a mid-to-distal ramp setting—precludes a tidalchannel origin for the brecciation. The lack of brecciation in theExshaw Formation and the presence of inclined bedding withinthe lower Exshaw Formation black shales above areas of thinnedanhydrite suggest that minor subsidence continued during ExshawFormation deposition (Devonian–Mississippian). These drill corefindings are consistent with the results of our waveform modeling(Fig. 7), which suggest collapse was initiated before and continuedthrough Exshaw Formation deposition. Based on the lack of de-formational features (brecciation, convolute bedding, and micro-fracturing) in the Exshaw Formation, it is unlikely that periods ofsignificant collapse in the underlying units continued during thedeposition of the overlying Banff Formation.The localized nature of the ellipsoid anomalies recognized on

the 3D reflection-seismic data (Figs. 3 and 4) and drill coreobservations (Fig. 6) suggest that undersaturated fluids may havebeen migrating along the WSOF. This fault-associated fluid flowlikely dissolved the Stettler Formation anhydrites local to theCHW, creating the Exshaw Formation /Wabamun Group sub-circular anomalies (Fig. 8). It is apparent from Figs. 3 and 4 thatthe ellipsoid anomalies can occur at some distance from theinterpreted faults associated with the CHW. This distance dis-crepancy may be explained by the nature of speleogenesis withinevaporites, whereby relatively narrow horizontal or subhorizontalkarst passages through more permeable beds (such as the dolo-laminites in the case of the lower Stettler) connect upward to largeperched caves at some distance from the potential source of fluids(Fig. 8B) (54, 55). Intervals of secondary dolomitization in theStettler–Big Valley Formation carbonates indicate that, in at leastsome parts of the basin (e.g., well 8-24-11-28W4 in Fig. 5), fluidsmigrated along faults. Both of these observations—dissolution anddolomitization—are consistent with hydraulic communication alongfaults. This type of dissolution and dolomitization may be similar tofracture-associated hydrothermal karst and fault-associated hydro-thermal karst and dolomite, respectively, as noted in the vicinity ofthe Peace River Arch in west-central Alberta (56). Devonian–Mississippian (Antler) tectonism has been documented in differentparts of the WCSB by fault-associated dissolution, dolomitization,and collapse of the uppermost Devonian strata (e.g., refs. 56 and57) and in the study area are inferred from unconformities in theuppermost Wabamun Group and Exshaw Formation (40). In

A B

C D

ELethbridge

LEGEND

V

V

V

VV

LimestoneShale

Exshaw Fm.Anhydrite

DolomiteBrecciaSandstone

Unconformity

N

Possible faultNormal fault

Ascending fluids

Descending fluids

Earthquake

Fig. 8. Proposed depositional and dissolution sequence. (A) Salina/sabkhadepositional setting of the Stettler Formation overlying Paleozoic carbonates(blue bricks) and Cambrian sandstones (deepest yellow layer). (B) Followingdeposition of the Big Valley Formation limestones, late Devonian extensionalfaulting provide a conduit for (possibly) ascending hydrothermal fluid (redarrows along faults) and fault-associated dissolution of the lower StettlerFormation anhydrite (cavity in pink layer). (C) Collapse of karst cavity in theStettler Formation anhydrites affecting the upper Stettler and Big ValleyFormations (purple and blue bricks, respectively) and causing brecciation andaccommodation. (D) Syndepositional overthickening of the Exshaw Forma-tion (black layer). (E) Present-day horizontal HF well (drill rig and black line)drilled in the Big Valley Formation and causing an earthquake (red circle).

E10010 | www.pnas.org/cgi/doi/10.1073/pnas.1807549115 Galloway et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0

contrast to the localized karst-related subcircular anomalies nearthe CHW, dissolution features related to basin-scale fluid floware often associated with more regionally pervasive karsting, forexample, as recognized in the Prairie Evaporite Formation innortheastern Alberta (58). Finally, the last supporting evidence forthe potential hydraulic conductivity of a deep-seated fault in thestudy area comes from induced earthquakes linked to HF of theoverthickened Exshaw Formation (11).

Implications for HF-Induced Earthquakes. Together, our findingssuggest the presence of fault-associated fluid flow in the geologicalpast. However, remineralization and cementation of fault-damagezones can eventually seal faults, drastically reducing their perme-ability (46). The stress regime in southern Alberta (42) has likelykept the WSOF mechanically/hydraulically active since the Lar-amide orogeny, as evidenced by natural earthquakes presentlyoccurring in southern Alberta (59). Additionally, repeated over-pressurization from HF stimulations has the potential to propa-gate shear/tensile failures along the fault pathway, reopening any(semi)sealed segments encountered (23, 60–62). In fact, shallow insitu measurements of fault hydraulic diffusivities computed duringreactivation experiments estimate values of 3,500 m2/s (63). Be-cause of these considerations, it is possible that the WSOF ispermeable enough to support the large hydraulic diffusivity esti-mated seismologically (4.5–64 m2/s) (11). We note, however, thatdenser arrays would have defined the seismologically estimatedhydraulic diffusivity more faithfully. Based on these findings, fu-ture geomechanical studies wishing to understand HF-inducedearthquakes should seriously consider the influence of fluid flowalong faults and fractures on modeled outputs.Given these prior arguments, it is natural to presume that

these indicators of fault-associated fluid flow have had a causalinfluence on the expression of induced earthquakes at the CHW(Fig. 8) and possibly for all HF wells in the upper Stettler–BigValley Reservoir play. By this reasoning, the relative paucity ofgeological conditions conducive to induced seismicity could re-flect the small fraction of seismogenic petroleum operations (24,25). To test this conjecture with available data for the ExshawFormation, we consider Fig. 5. In this plot, wells with a relativelythickened Exshaw Formation (>10 m) are potentially diagnosticof the conditions required for induced seismicity. On the otherhand, we note that very few regions indicating thick ExshawFormation overlap with HF development of the play; the CHWis one of the few nearby cases (Fig. 5). While we recognize thisassertion is hardly definitive for this case, it supports an estab-lished idea that geological factors influence the likelihood ofencountering induced earthquakes (62, 64–70). For example, HFoperations susceptible to induced seismicity in the Duvernay playdisplay a statistically significant spatial correspondence to themargins of a fossil reef (65). Some of these carbonate marginshave undergone hydrothermal dolomitization, a process relatedto structurally controlled fluid flow (68, 71, 72). These consid-erations are anecdotally supported by north–south strike-slipearthquakes (8, 9) and evidence of a transtensional flowerstructure (68, 73), a structure known to be conducive to verticalfluid flow (72). In this light, considering any features recognizedduring the reflection-seismic survey, wire-line log mapping, and

drill core analysis that are indicative of these processes couldprovide cost-effective insight to the geological susceptibility forthese types of earthquakes (69).We note that documenting these geological factors helps de-

velop a conceptual understanding of induced seismicity, withramifications for location models in forecasting induced seismichazard (65) and for identifying future sites (during exploration)that may be prone to these events. For example, pre-HF reflection-seismic site assessments may be unable to resolve the offsets/geom-etry of deep-seated faults to confidently establish their propensity forreactivation. The results of this paper suggest, instead, that geo-logical indicators of hydraulic connectivity may be utilized duringsite assessment to characterize fault influences and that proxies forfault-associated fluid flow may bolster confidence during pre-HFassessment or provide some level of input when reflection-seismicis either unresolvable or unfeasible. Scaled up to the basin level,the acquisition of 3D reflection-seismic surveys for the entire re-gion becomes unfeasible. Instead, based on our findings, it ispossible to utilize geological proxies at this scale to glean anypotential influence on the spatial distribution of induced earth-quakes (e.g., refs. 64, 65, and 69).

ConclusionsIn conclusion, we find that geological factors indicating fault-associated fluid flow have consequences for HF-induced earth-quakes. The coincidence of the CHW, evaporite karst-relatedoverthickening in the targeted Exshaw Formation/WabamunGroup interval, and deeper induced earthquakes share a commonhydraulic connection: a basement-rooted fault. The CHW oper-ation targeted the thickened reservoir, allowing an express routeof hydraulic communication from the stimulated fractures alongthe damage zone of the fault. This resulted in an increase in porepressure in the shallow basement fault segment, which causedfault slip nucleation. The spatial coincidence of the numerousgeological conditions required for HF-induced seismicity is rare,resulting in their sporadic documentation in the Exshaw Forma-tion play (and possibly also in other plays, globally). Recognitionof geological indicators of fault-associated fluid flow may help usunderstand induced seismicity potential in a given area.

Data AvailabilityAdditional earthquake-related data are available online throughthe Alberta Geological Survey Earthquake Catalogue at the In-corporated Research Institutions for Seismology (https://www.iris.edu/hq/) and in prior published material (11, 59). All well log anddrill core data used in this study are publicly available through theAlberta Energy Regulator (www1.aer.ca/ProductCatalogue/230.html;and https://aer.ca/providing-information/about-the-aer/contact-us/information-services-and-facilities/core-research-centre). Input3D reflection-seismic survey data from which our results werederived are proprietary. All data needed to evaluate the con-clusions of the paper are present in the paper.

ACKNOWLEDGMENTS. We thank ExxonMobil for the donation of the 3Dreflection-seismic dataset license that this study describes and Matt Grobeand two anonymous reviewers for critiques that helped improve this paper.

1. Foulger GR, Wilson M, Gluyas J, Julian BR, Davies R (2017) Global review of human-induced earthquakes. Earth Sci Rev 178:438–514.

2. Holland AA (2013) Earthquakes triggered by hydraulic fracturing in south-centralOklahoma. Bull Seismol Soc Am 103:1784–1792.

3. Clarke H, Eisner L, Styles P, Turner P (2014) Felt seismicity associated with shale gashydraulic fracturing: The first documented example in Europe. Geophys Res Lett 41:8308–8314.

4. Skoumal RJ, Brudzinski MR, Currie BS (2015a) Earthquakes induced by hydraulicfracturing in Poland Township, Ohio. Bull Seismol Soc Am 105:189–197.

5. Schultz R, Stern V, Novakovic M, Atkinson G, Gu YJ (2015b) Hydraulic fracturing andthe Crooked Lake Sequences: Insights gleaned from regional seismic networks.Geophys Res Lett 42:2750–2758.

6. Schultz R, Wang R, Gu YJ, Haug K, Atkinson G (2017) A seismological overview of theinduced earthquakes in the Duvernay play near Fox Creek, Alberta. J Geophys ResSolid Earth 122:492–505.

7. Babaie Mahani A, et al. (2017) Fluid injection and seismic activity in the northernMontney Play, British Columbia, Canada, with special reference to the 17 August 2015Mw 4.6 induced earthquake. Bull Seismol Soc Am 107:542–552.

8. Wang R, Gu YJ, Schultz R, Kim A, Atkinson G (2016) Source analysis of a potentialhydraulic‐fracturing‐induced earthquake near Fox Creek, Alberta. Geophys Res Lett43:564–573.

9. Wang R, Gu YJ, Schultz R, ZhangM, Kim A (2017) Source characteristics and geologicalimplications of the January 2016 induced earthquake swarm near Crooked lake, Al-berta. Geophys J Int 210:979–988.

10. Lei X, et al. (2017) Fault reactivation and earthquakes with magnitudes of up toMw4.7 induced by shale-gas hydraulic fracturing in Sichuan basin, China. Sci Rep 7:7971.

11. Schultz R, et al. (2015) The Cardston earthquake swarm and hydraulic fracturingof the Exshaw Formation (Alberta Bakken play). Bull Seismol Soc Am 105:2871–2884.

Galloway et al. PNAS | vol. 115 | no. 43 | E10011

ENVIRONMEN

TAL

SCIENCE

S

Dow

nloa

ded

by g

uest

on

June

20,

202

0

12. Raleigh CB, Healy JH, Bredehoeft JD (1976) An experiment in earthquake control atrangely, Colorado. Science 191:1230–1237.

13. Zhang Y, et al. (2013) Hydrogeologic controls on induced seismicity in crystallinebasement rocks due to fluid injection into basal reservoirs. GroundWater 51:525–538.

14. Segall P, Lu S (2015) Injection-induced seismicity: Poroelastic and earthquake nucle-ation effects. J Geophys Res Solid Earth 120:5082–5103.

15. Ellsworth WL (2013) Injection-induced earthquakes. Science 341:1225942.16. Schultz R, Stern V, Gu YJ (2014) An investigation of seismicity clustered near the

Cordel Field, west central Alberta, and its relation to a nearby disposal well. J GeophysRes Solid Earth 119:3410–3423.

17. King GE (2010) Thirty years of gas shale fracturing: What have we learned. SPE AnnualTechnical Conference and Exhibition, (Society of Petroleum Engineers, Florence, Italy),10.2118/133456-MS.

18. Davies RJ, Mathias SA, Moss J, Hustoft S, Newport L (2012) Hydraulic fractures: Howfar can they go? Mar Pet Geol 37:1–6.

19. Westwood RF, Toon SM, Styles P, Cassidy NJ (2017) Horizontal respect distance forhydraulic fracturing in the vicinity of existing faults in deep geological reservoirs: Areview and modeling study. Geomech Geophys Geo-Energ Geo-Resour 3:379–391.

20. Lele SP, Tyrrell T, Dasari GR (2017) Geomechanical Analysis of Fault Reactivation Dueto Hydraulic Fracturing. 51st US Rock Mechanics/Geomechanics Symposium, (Ameri-can Rock Mechanics Association, San Francisco).

21. Wilson MP, Worrall F, Davies RJ, Almond S (2018) Fracking: How far from faults?Geomech Geophys Geo-Energ Geo-Resour 4:193–199.

22. Kozłowska M, et al. (2018) Maturity of nearby faults influences seismic hazard fromhydraulic fracturing. Proc Natl Acad Sci USA 115:E1720–E1729.

23. Chen H, et al. (2017) Microseismic monitoring of stimulating shale gas reservoir in SWChina: 2. Spatial clustering controlled by the preexisting faults and fractures.J Geophys Res Solid Earth 123:1659–1672.

24. Skoumal RJ, Brudzinski MR, Currie BS (2015b) Distinguishing induced seismicity fromnatural seismicity in Ohio: Demonstrating the utility of waveform template matching.J Geophys Res Solid Earth 120:6284–6296.

25. Atkinson GM, et al. (2016) Hydraulic fracturing and seismicity in the western Canadasedimentary basin. Seismol Res Lett 87:631–647.

26. Kalkreuth W, McMechan M (1988) Burial history and thermal maturity, RockyMountain front ranges, foothills, and foreland, east-central British Columbia andadjacent Alberta, Canada. Am Assoc Pet Geol Bull 72:1395–1410.

27. Lemieux S (1999) Seismic reflection expression and tectonic significance of Late Cre-taceous extensional faulting of the Western Canada Sedimentary basin in southernAlberta. Bull Can Pet Geol 47:375–390.

28. Lemieux S, Ross GM, Cook FA (2000) Crustal geometry and tectonic evolution of theArchean crystalline basement beneath the southern Alberta Plains, from new seismicreflection and potential-field studies. Can J Earth Sci 37:1473–1491.

29. Chopra S, Marfurt KJ (2007) Seismic Attributes for Prospect Identification andReservoir Characterization (Society of Exploration Geophysicists and European Asso-ciation of Geoscientists and Engineers, Tulsa, OK).

30. Kington J (2015) Semblance, coherence, and other discontinuity attributes. LeadingEdge 34:1510–1512.

31. Kallweit RS, Wood LC (1982) The limits of resolution of zero-phase wavelets.Geophysics 47:1035–1046.

32. Johnston DI, Henderson CM, Schmidt MJ (2010) Upper Devonian to lower Mississip-pian conodont biostratigraphy of uppermost Wabamun group and Palliser formationto lowermost Banff and Lodgepole formations, southern Alberta and southeasternBritish Columbia, Canada: Implications for correlations and sequence stratigraphy.Bull Can Pet Geol 58:295–341.

33. Caplan ML, Bustin RM (1998) Sedimentology and sequence stratigraphy of Devonian-Carboniferous strata, southern Alberta. Bull Can Pet Geol 46:487–514.

34. Lyatsky H, Pana D, Olson R, Godwin L (2004) Detection of subtle basement faults withgravity and magnetic data in the Alberta basin, Canada: A data-use tutorial. LeadingEdge 23:1282–1288.

35. Lyatsky HV, Pana DI (2003) Catalogue of selected regional gravity and magnetic mapsof northern Alberta (Alberta Geological Survey, Edmonton, Alberta, Canada), SpecialReport 56, 43 pp.

36. Lyatsky HV, Pana DI, Grobe M (2005) Basement structure in central and southernAlberta: Insights from gravity and magnetic maps (Alberta Energy and Utilities Board/Alberta Geological Survey, Edmonton, Alberta, Canada), Special Report 72, 76 pp.

37. Root KG (2001) Devonian Antler fold and thrust belt and foreland basin developmentin the southern Canadian Cordillera: Implications for the western Canada Sedimen-tary basin. Bull Can Pet Geol 49:7–36.

38. Pana DI (2006) Unravelling the structural control of Mississippi Valley-type depositsand prospects in carbonate sequences of the western Canada Sedimentary basin.Potential for Carbonate Hosted Lead-Zinc Mississippi Valley-Type Mineralization inNorthern Alberta and Southern Northwest Territories: Geoscience Contributions,Targeted Geoscience Initiative, Bulletin of the Geological Survey of Canada, edHannigan PK (Natural Resources Canada, Ottawa, Ontario, Canada), Vol 591, pp 255–304.

39. Hauck TE, Panǎ D, DuFrane SA (2017) Northern Laurentian provenance for Famennianclastics of the Jasper basin (Alberta, Canada): A Sm-Nd and U-Pb detrital zircon study.Geosphere 13:1149–1172.

40. Pana DI, van der Pluijm GA (2015) Orogenic pulses in the Alberta Rocky Mountains:Radiometric dating of major faults and comparison with the regional tectonostrati-graphic record. Geol Soc Am Bull 127:480–502.

41. Johnston DI, Chatterton BDE (2001) Upper Devonian (Famennian) conodonts of thePalliser Formation and Wabamun group, Alberta and British Columbia, Canada.Paleontographica Canadiana 19:154.

42. Bell JS, Grasby SE (2012) The stress regime of the Western Canadian sedimentarybasin. Geofluids 12:150–165.

43. Barton CA, Zoback MD, Moos D (1995) Fluid flow along potentially active faults incrystalline rock. Geology 23:683–686.

44. Faulkner DR, et al. (2010) A review of recent developments concerning the structure,mechanics and fluid flow properties of fault zones. J Struct Geol 32:1557–1575.

45. Bense VF, Gleeson T, Loveless SE, Bour O, Scibek J (2013) Fault zone hydrogeology.Earth Sci Rev 127:171–192.

46. Sibson RH (1992) Implications of fault-valve behaviour for rupture nucleation andrecurrence. Tectonophysics 211:283–293.

47. Cox SF (2016) Injection-driven swarm seismicity and permeability enhancement: Im-plications for the dynamics of hydrothermal ore systems in high fluid-flux, over-pressured faulting regimes—An invited paper. Econ Geol 111:559–587.

48. Velayatham T, Holford SP, Bunch MA (2018) Ancient fluid flow recorded by re-markably long, buried pockmark trains observed in 3D seismic data, Exmouth Plateau,Northern Carnarvon basin. Mar Pet Geol 95:303–313.

49. Zaitlin BA, Berger Z, Kennedy J, Kehoe S (2010) The Alberta Bakken: A new, un-conventional tight oil resource play (BMO Capital Markets, Montreal). Available athttps://www.geoconvention.com/archives/2011/137-The_Alberta_Bakken.pdf. AccessedMay 1, 2018.

50. Zaitlin BA, Low WS, Kennedey J, Kehoe S (2011) BMO technical update: The AlbertaBakken petroleum system (ABPS) is “steady as she goes” an emerging unconventionaltight oil resource play (BMO Capital Markets, Montreal). Available at https://www.bmoaddeals.com/uploads/pdf/Alberta-ABPS-Bakken-Steady-As-She-Goes-April-6-2011.pdf. Accessed May 1, 2018.

51. Colborne J, Reinson G, Bustin RM, McIlreath I (2015) Stratigraphic framework anddepositional controls on reservoir occurrence, Big Valley formation, southern Alberta.Bull Can Pet Geol 63:192–223.

52. Flügel E (2010) Microfacies data: Fabrics. Microfacies of Carbonate Rocks (Springer,Berlin), pp 177–242.

53. Stanton RJ, Jr (1966) The solution brecciation process. Geol Soc Am Bull 77:843–848.54. Gutierrez F, et al. (2016) Sinkholes and caves related to evaporite dissolution in a

stratigraphically and structurally complex setting, Fluvia Valley, easter Spanish Pyr-enees: Geological, geomorphological and environmental implications. Geomorphology267:76–97.

55. Klimchouk A (1996) The typology of gypsum karst according to its geological andgeomorphological evolution. Int J Speleol 25:49–60.

56. Packard JJ, Al-Aasm I, Samson I, Berger Z, Davies J (2001) A Devonian hydrothermalchert reservoir: The 225 bcf Parkland field, British Columbia, Canada. AAPG Bull 85:51–84.

57. Mountjoy EWM, Halim-Dihardja MK (1991) Multiple phase fracture and fault-controlledburial dolomitization, Upper Devonian Wabamun Group, Alberta. J Sediment Res 61:590–612.

58. Hauck TE, Peterson JT, Hathway B, Grobe M, MacCormack K (2017b) New insightsfrom regional-scale mapping and modelling of the Paleozoic succession in northeastAlberta: Paleogeography, evaporite dissolution, and controls on Cretaceous de-positional patterns on the sub-Cretaceous unconformity. Bull Can Pet Geol 65:87–114.

59. Stern VH, Schultz RJ, Shen L, Gu YJ, Eaton DW (2013) Alberta earthquake catalogue,Version 1.0: September 2006 through December 2010, (Alberta Geological Survey,Edmonton, Alberta, Canada), Open File Report 2013-15.

60. Hummel N, Shapiro SA (2012) Microseismic estimates of hydraulic diffusivity in case ofnon-linear fluid-rock interaction. Geophys J Int 188:1441–1453.

61. Yang Y, Zoback MD (2014) The role of preexisting fractures and faults during mul-tistage hydraulic fracturing in the Bakken Formation. Interpretation 2:SG25–SG39.

62. Eaton D, Schultz R (2018) Increased likelihood of induced seismicity in highly over-pressured shale formations. Geophys J Int 214:751–757.

63. Guglielmi Y, et al. (2015) In situ observations on the coupling between hydraulicdiffusivity and displacements during fault reactivation in shales. J Geophys Res SolidEarth 120:7729–7748.

64. Schultz R, et al. (2016) Linking fossil reefs with earthquakes: Geologic insight towhere induced seismicity occurs in Alberta. Geophys Res Lett 43:2534–2542.

65. Schultz R, Atkinson G, Eaton DW, Gu YJ, Kao H (2018) Hydraulic fracturing volume isassociated with induced earthquake productivity in the Duvernay play. Science 359:304–308.

66. Shah AK, Keller GR (2017) Geologic influence on induced seismicity: Constraints frompotential field data in Oklahoma. Geophys Res Lett 44:152–161.

67. Skoumal RJ, Brudzinski MR, Currie BS (2018) Proximity of Precambrian basement af-fects the likelihood of induced seismicity in the Appalachian, Illinois, and Willistonbasins. Geosphere 14:1365–1379.

68. Corlett H, et al. (2018) Subsurface faults inferred from reflection seismic, earthquakes,and sedimentological relationships: Implications for induced seismicity in Alberta,Canada. Mar Pet Geol 93:135–144.

69. Pawley S, et al. (2018) The geological susceptibility of induced earthquakes in theDuvernay play. Geophys Res Lett 45:1786–1793.

70. Hincks T, Aspinall W, Cooke R, Gernon T (2018) Oklahoma’s induced seismicitystrongly linked to wastewater injection depth. Science 359:1251–1255.

71. Green DG, Mountjoy EW (2005) Fault and conduit controlled burial dolomitization ofthe Devonian west-central Alberta Deep Basin. Bull Can Pet Geol 53:101–129.

72. Davies GR, Smith LB, Jr (2006) Structurally controlled hydrothermal dolomite reservoirfacies: An overview. AAPG Bull 90:1641–1690.

73. Chopra S, et al. (2017) Seismic reservoir characterization of Duvernay shale withquantitative interpretation and induced seismicity considerations—A case study.Interpretation 5:T185–T197.

E10012 | www.pnas.org/cgi/doi/10.1073/pnas.1807549115 Galloway et al.

Dow

nloa

ded

by g

uest

on

June

20,

202

0