Fie!d, laboratory, and mode!ing studies of faulted rock yield insight into the hydraulic character of thrust faults. Late-stage faults comprise foliated and subparallel ...
GEOPHYSICAL RF•EARCH LETTERS, VOL.18,NO.5,PAGES 979-982, MAY1991 HYDROGEOLOGY OFTHRUST FAULTS ANDCRYSTALLINE THRUSTSHEETS: RESULTS OF COMBINED FIELDANDMODELINGSTUD•S CraigB. Forster
Department ofGeology & Geophysics, University ofUtah JamesP. Evans
Department ofGeology, UtahState University Abstract.Fie!d,laboratory, andmode!ing studies of faulted obtained forsystems thatspana widerangeof spatial scales, rockyieldinsightinto the hydrauliccharacterof thrustfaults. aidinestablishing appropriate scales ofobservation. Although
Late-stage faultscomprise foliatedandsubparallel faults,with ournumerical modelscannotprovidea uniqueanswerthat clay-rich gouge andfracture zones, thatyieldinterpenetrating appliesdirectlyto a givenfield situation, thenumerical results layers of low-permeability gougeandhigher-permeability doprovide insight intothisclass of groundwater flowsystem. damage zones.Laboratory testingsuggests a permeability contrast of two ordersof magnitude betweengougeand damage zones.Layersof differingpermeability leadto overall FaultZoneHydrogeology permeability anisotropy withmaximum permeability withinthe plane of thefaultandminimumpermeability perpendicular to thefaultplane. Numericalmodelingof regional-scale fluid Precambrian granitesand gneissesfoundin our field area flowandheattransportillustratesthe impactof fault zone (northwest Wyoming) werethrusted overadjacent sedimentary hydrogeology on fluid flux, fluid pore pressure,and rocksalongmoderately-dipping thrustfaults.About150m of temperature in thevicinityof a crystalline thrustsheet. exposed faultwereexamined through large-scale fieldmapping anddetailedsampling bothacrossandalongstrike. Wellfoliated,induratedgougeand very fine-grained,foliated Introduction cataclasite formthecoreof thethrustfauk [Evans,1990]. The surrounding damagezonecomprises faultedandfractured granitic gneiss.Totalthickness of catalasite anddamage zone Faultzonescanform complexconduit-barrier featuresthat ranges from5 to20 m. Faultslocatedwithinthedamage zone maylargelydetermine theregional-scale patterns andratesof aresubparallel to themainfault;impartinga relativelyhigh massand heat transferin the uppercrust[Smith et el., in permeability withintheplaneof thefault. Late-stage quartz press]. Faultscantransmit meteoricwaterto depthsin excess veinsparallel tothefaultsareevidence forhigherpermeability of !0 km and water flowing within faults can dissolve, transport, and depositmineralsin the fault [Kerrich, 1986].
Asa consequence, patterns andratesof groundwater flowin theuppercrustmustultimatelyinfluencethe rheologic properties of rocksin faultsandshearzones[Kerrich,1986]. Thedynamic interaction of faultevolution andflowsystem development is a transient problemof coupled fluidflow,solutetransportwith chemicalreactions,heattransfer,anddeformation thatis computationally intractable. Previous model-
ingstudies address different subsets ofthisproblem (e.g.,Shi
duringfault evolution[Evans,1990]. Fracturenetworksthat
actedasfluidpathways atdifferentstages in faultevolution are mapped in outcrop structures, polished handsamples, andthin sections.
Finite elementmodelingof isothermalfluid flow within
digitized versions of the2-dimensional fracture networks (e.g., Figurela) providesestimates of the averageflux of fluid throughthe model domain. Each fracturesegmentis represented asanopenconduit withparallelbounding surfaces
surrounded by impermeable matrix blocks. First,a uniform andWang,1987;Deminget al., 1990;Ge, 1990). We use hydraulic gradient of 2 x 10-3is oriented parallelto themain combined fieldandmodeling studies to assess howhigh-relief faulttoestimate theequivalent permeability (ice)along thefault. surface topography andthepermeability structure of thrust Next, thegradientis reorientedperpendicular to the faultto
faults mightinfluence fluidfluxes, pressures, andtemperatures inaninactive thrustfaultandthroughout theoverlying thrust sheet.Althoughwe modelonly idealizedhydrogeologic regimes, themodeling results helpustounderstand howfault- (a) related flow systems are modifiedduringthe changes in
(b)
topography andpermeability associated withfaukevolution.
Although thispaperaddresses onlythefluidflowsystems that maydevelop oncefaulting hasceased, theoverallaimof ourresearch is to examinethe evolutionof fluid flow systems
during thrusting. Weposethreequestions: (1)howdofluids reach andleavefaultzonesin a givenstructural setting?, (2)
how dofluidsmigrate within,andinteract with,fauk rocks?, and(3) howdo spatialandtemporal variations in hydraulic
0 lorn i
!
properties of faultzonesandprotolithinfluence bothlocaland
0 20cm N'-• I
I
Down
regional-scale flowsystems andthermal regimes? Weusefield rocks:a) line drawingof a thin section mapping andlaboratory testingto collectthedatarequired to Fig. 1. Fault-related to the address thesequestionsacrossscalesrangingfrom cut parallelto slip direction(S) and perpendicular microscopic to megascopic. Numerical modeling results, dominantfoliationfroma damagedzonewith transgranular fractures that,in turn,connect zonesof intragranular fractures, b) rocksamplecontaining a gougezone(G), a zoneof fo!iated cataclasite (FC), anda fractureddamagezone(DZ). Oriented cores cut bothperpendicular (.L)andparallel ( Ii1 and•2)tothe
Copyright1991 by the AmericanGeophysicalUnion. Paper number 91GL00950
0094-8534 /91/ 91GL-00950503 . O0
foliation are indicated. 979
980
Forster andEvanS'Hydrogeology of Thrust FaultsandThrustSheets
Table 1. Numerical andexperimental estimates ofthepermeability (m2)offault-related rocks Method ofEstimating k
Nume•6al'EStirhate ' G'øu•e (Ice for'fgrn'&pertures) .....
Parallel toFault(11)
2x 10-17 ............
PerDendicu! .artoFault(J_)
3x 10-18
Numerical Estimate - Gouge (kefor10gmapertures) 2 x 10-14 LabTestsonGouge 1.0x 10-17(111),6.0x 10-17(112) LabTests onDamaged Zone 1.2x 10-15and3.0x 10-15 Lab Testson Protolith
estimateke acrossthe fault. In eachcase,domainboundaries
parallelto the directionof the appliedhydraulicgradientare assumed impermeable.Computedvaluesof ke (Table 1) represent the permeabilityof the equivalentporousmedium that,whensubjected to the samehydraulicgradient, yieldsa volumetric flow rate matching that computed for the corresponding fractured medium.Two casesaresimulated; eachwith uniform,butdifferent,fractureapertures (1 !.tmand 10 [xm). Modelingresultssuggestthatthe geometryof the fracturenetworkalone(Figure 1a) canimpartan anisotropic characterto the permeability structurewithin the fault; computed valuesof keparalleltothefaultareabout6 timesice perpendicular tothefault(Table1). Steady-state permeabilitytestsof 2.5 cm diametercores fromgouge,damaged zone,andprotolith(Figurelb) yie!dk
1.4 x 10-16 and 2.3 x 10-16
modelingresultsprovideinsightintotheunderlying hydraulic regime that would exist where steeptopographicrelief is maintainedthroughout fault evolution.
A shallowwatertablecoincides withtheupperboundary
andverticalboundaries arebothinsulated andimpermeable. A
uniform regional heatfluxof 60 mWm-2is applied along the basal,impermeable boundary.Temperatures on theupper boundary arespecified asa functionof elevation, exceptatthe thrustfault outcropwhere a free-temperature boundary condition is assigned to allowa thermalspringto develop. Valuesof bulkk (Figure2a) are controlledby fracturing ,
within the rock mass. At the regionalscaleeachunit is treated
as an equivalent,isotropicporousmedium. The crystalline thrustsheetis assigneda value of k typical of relatively
valuesrangingfrom 1 x 10-17to 3 x 10-15m2 (Table!).
LEGEND
Althoughpermeabilities arelargerthanexpected duringfault development,the relative variation in permeability from protolith(intermediatek), throughdamagezone(highk), to gougezone(low k) is likely representative of conditions within manyfaults. Alternatingzonesof highandlow k withina fault impart an effective anisotropyto the overall permeability structure of thefault;maximumpermeabilityparallelsthefauk planeandminimumpermeability is perpendicular to thefault. Valuesof fault gougek are within the rangereported for otherlaboratorytests(10-21to 10-12m2) [seeSmithet al., in press].Valuesof damageandgougezonepermeablity arealso within the range (10-17to 10-11m2) obtainedfrom in situ testingof k within a thrustfault in crystallinerocks[Davison
3 x 10-15 2.0x 10-17 --
log k(m 2) Porosity
Crystalline ThrustShet E]ZZ]]-16 Sedimentary Rocks :•..... '15
{Crystalline Seaimentary 6r, d '• Rocks •.... -18 -22 FootwellLayer ThrustFoult
_ll!-m -15 -II
.05 .I 5
.05 .05} .15 .05
,-., 4(a) . '-
•'- ',.,,,.. •-,-',,- ,-'' '., -',_'.' '-)>;',3;'z
'"
7••-'.,('.-';i
FaultZoneHydrogeology:HydraulicandThermalRegimes
Regional-scale numerical modeling Studies illustrate how faultzonehydrogeology mightinfluencefluid flux, fluid pore pressure, andtemperature in the vicinityof a thrustfault. The idealizedmodeldomainandthedistribution of permeability and Fig.2. Regional-scale modeling of thereference case(Run 1 porosityin ourreferencecase(Figure2a) represents oneof a domain andpermeability structure, b) seriesof range-basintopographicfeaturestypical of the in Table2): a) model andcontours of fluidporepressure (MPa), deformed forelandof Wyoming. Verticalboundaries markthe fluidfluxvectors andcontours oftemperature (øC). Plus approximate locations of thesub-vertical groundwater divides andc)fluidfluxvectors signs (+) shown in (b)and(c)indicate regions wherefluidflux expectedbetweeneach adjacentthrustsheet. Althoughwe consideronly the present-daycaseof an inactivefault the is less than 10-12 m s-1.
Forster andEvans:Hydrogeology of ThrustFaultsandThrustSheets
981
unfractured crystallinerock (10-16m2). The higher-k Pressureperturbations that developduringfault evolution sedimentary package isassigned a valuerepresentative ofmany would be superimposedon the topographically-induced carbonate andclasticrocks(10-!5m2). Twobasallayersare pressureregime. For example, pore volume reductionand incorporated in themodelto represent a generaldecline,with increasing depth,in permeabilityof both crystallineand scdLmentary rocks. Thermalconductivity andspecific heatof fluidandrockare assumed constant. Porosityis variedfrom 0.05to 0.15 (Figure2a). Fluid densityand viscosityare computed asa function of temperature andpressure. The thrustfault is representedas a discretefeaturewith
permeability kf andwidthb. Bothkf andb maybevaried independently alongthefauk.A valueof 10-11 m2 isselected for
thermal pressurizationwithin an active fault could cause temporarylithostaticporefluid pressures within 1 or 2 meters of thefault [MaseandSmith,1987]. In addition,compression of the sedimentary packageduringa thrusteventcouldcause regional-scaleperturbationsof the groundwaterflow system [Ge, 1990; Deming et al., 1990]. Additional studiesare requiredto assess therelativeimportanceof topographicallyandtectonically-driven flow neara thrustfault.
kfinourreference caseto demonstrate themaximum impacta
Fault Zone Scale ParametricStudy
permeable thrustfaultmighthaveontheregional-scale fluid flowandthermalregimes. This is the upperlimit of fault
permeability measured in situwithina thrust fault[Davison and Kozak,1988]. Fault width (b) is 1 m.
Coupled equations of steady-state fluidflow andheattransferaresolved usingthefiniteelementmethod.Thefaukisrepresented by superimposing a seriesof line elementson a field of triangularelements. Forsterand Smith [1988] outlinethe ½6rriplete boundary valueproblemandmodeling approach. Regional-Scale ModelingResults
A seriesof simulations (Table2) illustratetherelativeimpact of fault andprotolithpermeabilityon thetemperature (T), pore fluidpressure (P), andfluid flux (q) conditions encountered by fluid movingthroughthe fault shownin Figure2a. Variations in q areplottedasa functionof d/stancealongthefauk (Figure 3) with 0.0 located where the fault intersects the basal
boundary(Figure 2). The permeabilitystructureshownin Figure2a is thereferencecase. Computedvaluesof T, P, and q vary along the fault in a nonlinear fashion that cannotbe obtainedusingsimpleanalyticalsolutions. Table2. Summaryof ModelingRuns
Groundwater rechargesat high elevationsand discharges bothatthefaultoutcropandat low elevations neareachendof themodeldomain (Figures2b and 2c). In unfaultedterrain groundwater wouldbe transmitted morepervasivelythrough shallower region. s of the model domain. Focusingof fluid flowwithinthe fault increasesthepotentialfor localizedfluid-
SCENARIO
assisted chemical reactions andpromotes temporal andspatial variations in faultrheologyandpermeability. Changes in fluid flux alongthe fault are closelyrelatedto variationsin permeability of the surroundingprotolith. Although fault permeabilityis high at the baseof the model domain,the lack of recharge from surrounding lowpermeability basalunitsyieldslimitedfluid flow deepwithin
CHARAd'•RISTICS
1 2
Reference case(RC)
k strucmr• of Figure3a'
Unfaulted case
RC minus fault
3
Ard'sotropic fault
RC k footwalllayer10-18m2
4
kf on 2 km of fault
RC, kf of 10-12m2
5
Reducedthrustsheetk RC, k thrustsheet10-!7 m2
,,
Modeling resultsindicatethat the fault has a pervasive impacton temperatureandpressuredistributionsbothwithin thefault andthroughoutthe modelingdomain. In the absence thefault. Maximum values of fluid flux are found at shallow of a permeable fault (Run2), fluid pressures nearthebaseof depths wherethefaultis adjacentto thehigher-permeabilitythe domainrise as high as 3 MPa abovethoseobtainedat the crystalline thrustsheetandthefull impactof topographically- corresponding locationin the referencecase (Run 1). The induced hydraulicgradients is transmitted tothefault. permeablefault transmitstopographically-derived hydraulic
A three-orderof magnitudeincreasein fluid flux is computed asfluid flowsfromthethrustsheetto theunderlying fault(Figures 2b and2c). Thisresultsuggests thatwater-rock ratios derivedfor the fault, usingestimates of thevolumeof watertransmitted througha unit volumeof fault rock,could greatly exceedthosederivedfor the overlyingthrustsheet. Thus, cautionshouldbe exercisedwhenestimating fluidflux andfluidtransittimeswithinan overlyingthrustsheet[e.g.,
Region oft•educed kf.(œun 4)
Losh,1989]. Computedtransittimesare 105 yearsfor meteoric waterrecharging at theridgetoptoreachthefaultand
......I......--....•'-'_. ,-J
anadditional 300 yearsfor waterto dischargeat the ground
!i
surface aftermigrating through about3.7kmof thefauk. Thetopographically-driven flow system contributes to non-
L.J
hydrostatic vertical pressure gradients, therefore directions and
Run
rates of fluid flow cannotbe inferredsolelyon thebasisof a
fewpointestimates of fluidpressure obtained fromfluid
inclusion studiesor in situ measurements. For example, computed directionsof fluid flux (Figure 2b) indicate groundwater can moveeitherdownor up local gradientsin pore pressure. Fluidpressures throughout thecentralpartof
themodel domain areunderpressured byasmuchas12MPa relative tothose thatmightbecomputed assuming hydrostatic conditionswithin the thermal regime of Figure 2c.
Underpressuring occurs primarily because non-hydrostatic
-
0
Up
1 2,
..............
5
.....................
Down
. .
.
ø
.....
DISTANCE ALONG FAULT ZONE (kr'n)
pressure gradients develop withinthelow-permeability thrust Fig. 3. Fluid flux computedalong the thrustfault shownin
sheet whilepressure g•adients in theunderlying more- Figure2a for a seriesof runssummarizedin Table2. Except
permeable sedimen• package areapproximately hydrostatic. whereindicated,fluid flow is generallyupwardalongthefault.
982
ForsterandEvans:Hydrogeology of ThrustFaultsandThrustSheets
gradientsdeep into the flow systemand causesregional variationsin fluid porepressures thatultimatelydependupon theoverallhydrauliccharacterof thefault. The fault exertsa significantimpacton temperatures computed withinthemodel domain;temperatures alongthe upper5 km of the fault are elevatedby up to 10 øCabovethoseof theunfaultedcase. Thepermeability of thecrystalline thrustsheetoverlyingthe faultcontrolstheamountof groundwater entering or exitingthe fault and
the rate of fluid flux within the fault.
Reduced
groundwater rechargeto thefault allowslocal,topographicallyinducedhydraulicgradientsto causea reversalin fluid flow, from upflowto downflow,at shallowdepthsin thethrustfault
groundwater flow system mustbe considered. Ratesof fluid
flowwithinaninactive tkrust faultarestrongly influenced by
thetopography andpermeability of the overlyingthrustsheet. Althoughthe detailedhydrauliccharacterof the fault can influencefluid transittimeswithin the fault and adjacent
protolith, localvariations in faultpermeability havelittleimpact on theregional-scale distribution of temperature andpore pressure. On theotherhand,thebulkpermeability of thefault caninfluence regional-scale fluidflow andthermal regimes.
Acknowledgements. Fundingprovidedby NationalScience
grant#EAR-9005087,Utah StateUniversity (Run5 in Figure3). Reducing thethrustsheetk from10-16to Foundation grams,andtheUniversityof Utah. Two anonymous 10-I7m2reduces q in theupper10.5kmofthefaultby'atleast research oneorderof magnitude(Figure3). The sensitivityof fluid reviewsimprovedthe manuscript. fluxeswithin the fault to the permeabilityof the thrustsheet indicatesthat the hydrauliccharacterof the fault cannotbe divorcedfrom the surroundingregional groundwaterflow system. Reasonableestimatesof bulk permeabilityof the thrustsheetmustbemadebeforethemagnitudeandvariability of q withinthe faultcanbe estimated. Earlier we notedthat: 1) a fault can exhibit strongspatial variationsin permeability, and 2) that faults developedin crystalline rock may exhibit anisotropic permeability. Reducing thepermeability of thefault(kf) from10-11to 10-!2 m2alongthe2 km fault segment(Figure2a) leadsto reducedq withinandadjacentto the regionof lowerkf. Little changein T and P along the fault is computedbecausethe overall distributionof P, T, and q within the regional-scalemodel domainis little affectedby thelocalizedvariationin kf. The effective anisotropyassociatedwith layeringof gouge anddamagezoneis simulatedby reducingthepermeability of a segmentof the footwall (footwall layer shownin Figure2a) from 10-15to 10-18m2. Thisrelativelythickgougezonelies immediatelyadjacentto theoverlying,more-permeable damage
zone(k = 10-!1m2). Althougha significant increase in fluid flux is indicatedalongtheuppermorepenncablesectionof the
References
Chester,F. M., andLogan,J. M., Implicationsfor mechanical properties of brittle faults from observationsof the Punchbowlfault zone, Calif.,Pageoph., 124, 80-106, 1986. Davison,C. C., and Kozak, E. T., Hydrogeologicalcharacteristicsof major fracturezonesin a large granitebatholithof theCanadianShield,Proc. 4th Can./Am.Conf.Hydrogeol, 52-60, 1988.
Deming, D., Nunn, J.A., and Evans, D.G., Thermal effectsof compaction-driven groundwaterflow from overthrust belts, J. Geophys.Res., 95,6669-6683, 1990. Evans, J. P.,Textures, deformation mechanisms and the role
of fluids in cataclastically deformed granitic rocks,in DeformationMechanisms, Rheology,and Tectonics, edited by R. J. Knipe and E. Rutter, Geol. Soc.Lond. Sp.Pub. 54, 29-39, 1990.
Forster, C.B., and Smith, L., Groundwaterflow systems in mountainous terrain!. Numericalmodelingtechnique, Water
Resour.Res.,24, 999~!0•0, 1988. Ge, S., On the mechanicsof tectonically-driven fluid flowin forelandbasins,Ph.D. Diss., JohnsHopkinsUniv., 215 theupper3 km of the fault areelevatedby up to 5 øCabove p., 1990. those of the reference case. At first glance, theseresults Kerrich, R., Fluid infiltration into fault zones: Chemical,isosuggestthat it may be more importantto determinethe bulk topic,andmechanical effects,Pageoph.,124,225-268,1986. permeabilityof the thrustsheetratherthanexpendsignificant Losh, S., Fluid-rockinteractionin an evolvingductileshear effort assessing the magnitudeof anisotropyin fault k. Fluid zone and across the brittle-ductile transition, central migration into the fault would be significantlyreduced, Pyrenees,France,Am. J. Sci.,289, 600-648, 1989. however,if the low k zone were placed at the baseof the
fault, fluid pressurealong the fault is little affectedby the changein permeabilitystructure.Computedtemperatures in
hangingwall, ratherthanat thetopof thefootwall.
Mase, C. W., and Smith, L., Effects of frictional heatingon
thethermal,hydrologic, andmechanical response of afault, J.Geophys.Res., 92, 6249-6272, 1987.
ConcludingRemarks
Shi,Y., andWang,C-Y, Two-dimensional modeling ofthePT~tpathsof regionalmetamorphism in simpleoverthrust terrains, Geology,15, 1048-1051, 1987.
We envision thrust faults in crystalline rocks as anastomosingnetworks of relatively high-permeability
fractured rockinterspersed withpinching andswelling podsof low permeabilitygougeand unfracturedprotolith.Preferred orientationof interleavedpodsof lower-permeability gouge
andhigher-permeability damaged rockyieldsanisotropy in the permeablitystructureof the fault. Deformationcausesfault
Smith,L., Forster,C. B., andEvans,J.P., Interactions of fault zones,fluid flow andheattransferat thebasinscale, in
Hydrogeology ofLowPermeability Environments, edited by
S.P. Neumanand I. Neretnieks,!nt.. Assoc.Hydrogeol.,
HydrogeologySelectedPapers,2, in press.
C. B. Forster,Department of GeologyandGeophysics,
Universityof Utah, SaltLake City, UT, 84112-1183 permeability to become increasingly anisotropicwith J.P.Evans, Department ofGeology, UtahStateUniversity, development of through-going inter-andintragranular fractures subparallel tothemainfaultandselective plugging of fractures Logan,UT 84322-4505 in clay-andmica-richgougezones[ChesterandLogan,1986; Evans, 1990]. (ReceivedJanuary10, 1991; Modelingresultsrevealthatthehydraulicinteraction of both revisedMarch 7, 1991; thethrustfaultandthetopographically-driven, regional-scale acceptedMarch 12, 199!.)
