structuresandconventionalseismicprospectingtechniqueshavebeenusedtoidentifypotentialgashydrateaccumulations,withsomedegreeofcertainty.Theintegrationofloggingandseismicdata,however,providesamorerobustinterpretationofgashydratepresenceanddistribution.Thelinkbetweenloggingdataandseismicdataisrockphysicsmodeling,whichwillbedevelopedinthefollowingsectionsofthereport.

3.3.1.IdentifyingsandsandclaysfromLWDlogs

Oneofthemainexplorationtasksinagashydratereservoircharacterizationstudyistodiscriminatesandsfromclaysand,moreimportantly,toseparatehighlyconcentratedhydrate-bearingsandsfromwater-bearingsandsandfree-gas-bearingsands.ClaysinterbeddedwithdiscretesandscomprisemostofthesedimentsinthedeepwaterGulfofMexico.Figures3e5illustratetypicalwell-logdatafortheGC955area.Ingeneral,eachoftheGC955wellsischaracterizebyarelativelythickstratigraphicsectionextendingfromthesea?oortoadepthbelow300mbsfthatischaracterizedbyrelativelyhighgammaraylogvaluesof70APIandhigher,whichsuggestsclay-dominatedsediments.Withinthemoredeeplyburiedlog-inferredgammaraysand-richsections,thegammaraylogdropstoabout25API.

Rockphysicsdepthtrendscanbecomplicatedbyvaryinglithology,mineralogy,?uidproperties,andporepressurecondi-tions(Avsethetal.,2005).Ingeneral,acousticimpedanceofshallowsedimentsincreaseswithdepthduetocompaction.Closetothesea?oor,sandshavehigherimpedancethanclaysbecauseclaystendtohavehigherwatercontentthansands.However,theporosityofclaysdecreasesfasterwithdepththanthatofsandsintheveryshallowsectionbecauseclaystendtocompactmoreeasilyduringearlyburial(Velde,1996).Theimpedanceofclaystendstoincreasemorequicklythanthatofsands.Thus,animpedancecrossoverofsandsandclaysmayoccurintheshallowsection.Belowthecrossover,clayshavehigherimpedance.NeidellandBerry(1989)observethattheimpedanceofshallowunconsoli-datedPleistocenesandsislowerthanthatofassociatedclays.Hilterman(2001)illustratesthattheimpedanceofsandsislessthanclaysfromveryshallowsectionto4000mbsfinGulfof

00

00

100

200Depth(mbsf)

300

4005006000

50100150

1

GammaRay(API)

2.2515002000250030001.51.75210

3

Resistivity(m)Vp(m/s)Density(g/cm)

Figure3.Gammaray,resistivity,compressional-wave(Vp)acoustic,anddensitylogsfromwellGC955-I,showingathicksandzone.Thelowdensityvaluesinthenon-hydrate-bearingportionofsand-richsectionareprobablyproducedbyboreholewashout(Guerinetal.,2009).

124Z.Zhangetal./MarineandPetroleumGeology34(2012)119e133

a

GC955-H

100200

Depth(mbsf)

300400

500600

501105010015015002000250030001.51.752

3

Resistivity(m)GammaRay(API)Vp(m/s)Density(g/cm)

b

GC955-Q

100100100100

Depth(mbsf)

200200200200

300300300300

400

450

50100150GammaRay(API)

400450

400400

450450

1500200025003000

1.51.752.02.25

3

Resistivity(m)Vp(m/s)Density(g/cm)

Figure4.Gammaray,resistivity,compressional-wave(Vp)acoustic,anddensitylogsfromwells(a):GC955-Hand(b)GC955-Q,showinghighlyconcentratedgashydratewithin

sand-richsections.Notethatthedensitylogsareofgoodqualityinthehydrate-bearingportionofsand-richsection.Thelowdensityvaluesinthenon-hydrate-bearingportionofsand-richsectionareprobablyproducedbyboreholewashout(Guerinetal.,2009).

Mexico.Figure5showsthenormalcompactiontrendfortheGC955area.Itisspeculatedthatthesedimentinducedimpedancecross-overoccursatashallowdepthintheGC955studyarea.Withinthegashydrate-bearingsedimentarysectionofourinterestinthisstudy,from350to700mbsf,theacousticimpedanceofsandsislessthanthatofclays.

Figure6showsthatclays,water-bearingsands,andhydrate-bearingsandsarewellseparatedinthecrossplotofcompressional-wavevelocityversusgammarayvalueintheGC955-Hwell.Water-bearingsandshaverelativelylowcompressional-wavevelocities,whereashydrate-bearingsandsarecharacterizedbyhighvelocities.3.3.2.Rockphysics-basedgashydratemodel

Impedancetrendsinhydrate-bearingsandscanbedescribedbyarockphysics-basedgashydratemodelthatderivesaphysics-basedrelationshipbetweenhydratesaturationandelasticpropertiesofsediments(Vp,Vsanddensity).Effectivemedia,contactmodels,and?uidsubstitutiontheoriesareallusedtocreatetherockphysicsmodelinthisreport.ThefundamentalprinciplesandequationsarepresentedbyMavkoetal.(2009)andhavealsobeenusedbyLeeetal.(2009)toconductsimilarinversionsofgashydrateaccumulationsinnorthernAlaska.Inthisstudythesedimentarysectionisassumedtobeanisotropiccompositemediaofporousrockwith

Z.Zhangetal./MarineandPetroleumGeology34(2012)119e133125

Vp(m/s)

Neutronporosity(%)

Figure5.Compressional-wavevelocityversusneutronporosityfromwellsGC955-H,GC955-I,andGC955-Q,showingnormalcompactiontrend(Hamilton,1971)andhighvelocitygas-hydrate-bearingsands.

isotropicmineraland?uidcomponents,andgashydrate.Themineralcomponentsincludemostlyclaysandsilica-richsands;the?uidcomponentsarewaterandgasinmostcases.Weassumethatgashydrateandfreegasgeneratedintheporespaceofsandsreducethewater-?lledporositybutthatthetotalporespacewouldnotchange(LeeandCollett,2001).Thehydratethat?llstheporespacepartiallyactsasacomponentof?uidandpartiallyactsasacomponentofthemineralframe.Thisbehaviorhasbeenobservedinlaboratorystudies(BuffettandZatespina,2000;Yunetal.,2005;Wintersetal.,2004;Priestetal.,2009).

TheHashin-Shtrikmanlowerboundwasusedtosimulateelasticmoduliofhydrate-bearingsediments(Mavkoetal.,2009).Thelowerlimitoftheboundforeffectiveelasticmoduliiswatersaturatedsedimentwithoutanyhydrate.ThehigherlimitoftheboundistheReussaverageofhydrateandmineralcomponents

(Helgerudetal.,1999;HanandBatzle,2004).Itissuggestedthatatlowgashydratesaturations,hydrate?oatsintheporespaceandisconsideredtobeinsuspension.Whenhydratesaturationincreases,thehydratebecomesgrainsupported.Gashydratehasbothpore?llingandgraincontactedbehaviorsatrelativelyhighgashydratesaturation.Weusetheparameter(ε)providedbyLeeandWaite(2008)todescribebothpore?llingandgraincontactedbehavior.

Weusedthe“Waltonsmoothmodel”topredicttheelasticmoduliofthe“dryrock”matrixandGassmann’sequationtopredicttheelasticmodulioffullywatersaturatedsediment(Mavkoetal.,2009).Table2includestheelasticconstantsusedforthecalculation.Ifgashydrateispresentintheporespace,theHashin-Shtrikmanlowerboundisusedtocomputetheelasticmoduliofthehydrated-bearingsands;iffreegasispresent,Gassmann’sequationisusedtocalculateelasticmoduliofthefree-gas-bearingsands.Uniformgasdistribu-tionisassumedandthe?uidbulkmodulusiscomputedfromtheReussaverageofwaterandgasbulkmoduli(Helgerudetal.,1999).Finally,thecompressional-andshear-wavevelocitiesarecomputedfromthewell-log-derivedmodulianddensities.

TheestimatesofhydratesaturationderivedfromtheJIPLegIIresistivitylogdataintheGC955wellsbyGuerinetal.(2009)wereusedtocalibratethemodel(Fig.7).Compressional-wavevelocityisoneofmainfactorstodetermineseismicacousticamplitudes.Figure7showsthatpredictedvelocitiesincreaseveryslightlyatlowgashydratesaturationslessthan0.12becausehydrateinthislow-saturationcase,hasastrongeffecton?uidsandaweakeffectonthematrix.Suchslightchangewouldnotcausetheamplitudeofhydrate-bearingsedimentstostandoutfromthebackgroundamplitudeofwatersaturatedsedimentsethusmakingthemdif?culttodistinguish.

Vp(m/s)

Table2

Elasticconstantsforcomponentsofsediments.ComponentSandClayHydrateWater

Gammaray(API)

Figure6.Compressional-wavevelocityversusgammarayfromwellGC955-H,showingclay-richsections,water-bearingsandsandgas-hydrate-bearingsands.A1-m-thickinterpolating?lterwasappliedtotherawvelocityandgammaraydatabetween350and437mbsf.

r,g/cm3

2.652.580.911.02

K,GPa3620.97.72.29

G,GPa456.853.20

126

Z.Zhangetal./MarineandPetroleumGeology34(2012)119e133

340032003000

)

s/2800m(p2600V2400220020001800

Gashydratesaturation(fraction)

Figure7.Compressional-wavevelocitiesversusmodel-derivedgashydratesatura-tions.Theredboxescorrespondtomeasuredpointsfromdepthinterval410e450mbsfinGC955-H.Thegreenlinecorrespondstopredictedcompressional-wavevelocitiesestimatedfromourrockphysicsmodelusingε?0.12whichsuggestedbyLeeandWaite(2008),andcoordinatenumber?3.5whichwasusedtopredictvelocitiesofshallowsedimentsinGulfofMexicobyDutta(2009).Aporosityof0.46wasusedinthepredictionofcompressional-wavevelocitiesbasedonloganalysisatthesection.(Forinterpretationofthereferencestocolourinthis?gurelegend,thereaderisreferredtothewebversionofthisarticle.)