Publishedin:Nieuwland,D.A.(ed.)NewInsightsintoStructuralInterpretationandModelling.

GeologicalSociety,London,SpecialPublications,212,101‐116.

101

TectonicstressintheEarth’scrust:advancesintheWorldStressMapproject

BlankaSperner1,BirgitMüller1,OliverHeidbach1,DamienDelvaux2,3,JohnReinecker1,KarlFuchs1

1GeophysicalInstitute,KarlsruheUniversity,Hertzstrasse16,76187Karlsruhe,Germany

(e‐mail:blanka.sperner@gpi.uni‐karlsruhe.de)2RoyalMuseumforCentralAfrica,Leuvensesteeweg13,3080Tervuren,Belgium

3presentaddress:VrijeUniversity,Amsterdam,TheNetherlands

Abstract:TectonicstressisoneofthefundamentaldatasetsinEarthsciencescomparablewithtopog‐raphy,gravity,heatflowandothers.Theimportanceofstressobservationsforbothacademicresearch(e.g.geodynamics,platetectonics)andappliedsciences(e.g.hydrocarbonproduction,civilengineering)provesthenecessityofaprojectliketheWorldStressMapforcompilingandmakingavailablestressdataonaglobalscale.TheWorldStressMapprojectoffersnotonlyfreeaccesstothisglobaldatabaseviatheInternet,butalsocontinuesinitsefforttoexpandandimprovethedatabase,todevelopnewqualitycriteria,andtoinitiatetopicalresearchprojects.Inthispaperwepresent(a)thenewreleaseoftheWorldStressMap,(b)expandedqualityrankingschemes forboreholebreakoutsandgeologicalindicators,(c)newstressindicators(drilling‐inducedfractures,boreholeslotterdata)andtheirqualityrankingschemes,and(d)examplesfortheapplicationoftectonicstressdata.

Tectonic stress is feltmostextremelyduring its re‐leaseincatastrophicearthquakes,althoughlessspec‐taculartectonicstressisakeysafetyfactorforunder‐groundconstructionssuchastunnels,cavernsforgasstorageordepositsofnuclearwaste(Fuchs&Müller2001). Furthermore, the economic aspect is of in‐creasing importance in hydrocarbon recovery.Knowledgeofthestressfieldisusedinthehydrocar‐bon industry to foresee stability problems of bore‐holesandtooptimizereservoirmanagementthroughtectonic modelling in combinationwith correlationwith other data sets such as 3D‐structural infor‐mation,e.g.locationandorientationoffaults.

Geodetic measurements (e.g. Global PositioningSystem, GPS; Very Long Baseline Interferometry,VLBI; Satellite Laser Ranging, SLR) have becomeavailableinincreasingnumbersandenhancedqual‐ity.Theyprovidesurfacedisplacementvectorsfromwhichthestrainratefieldcanbededuced.Thecom‐bined interpretation of stress and strain rate dataprovidesuniquechallenges forEarth scientists in anumberoffields.Intectonics,plateboundaryforcesconfine the kinematics of platemotion and the dy‐namicsofplatedeformationresulting,forexample,inmajor differences between intraplate and plateboundarydeformationzones.Thewidthofthelatterevenvariesforthedifferenttypesofplateboundaries(Gordon&Stein1992).Thesedifferencesexpressthe

laterally and vertically heterogeneous deformationpatternofthecrustduetoitscomposition,mechani‐calpropertiesandtectonicsetting.Thecomparisonofthe strain rate field at the surfacewith earthquakedatafromtheseismogenicpartofthecrustwillgiveinformationonthedepthvariationofstrain.There‐lationshipofstrainandstressisofspecialimportanceatactivefaultzones,whereaccumulatedstressisab‐ruptlyreleasedinearthquakes.

Thescaleofstressimpactrangesfromacontinent‐widescaleinordertoexplaintectonicprocesses,toametrescaleatreservoirandconstructionsites.Theglobal databaseWorld StressMap (WSM) providesinformation on the contemporary tectonic stress inthe Earth's crust in a compact and comprehensiveway.TheWSMprojectwasinitiatedasataskforceofthe International Lithosphere Program under theleadershipofM.L.Zoback.Thedatabaseisnowmain‐tainedandexpandedat theGeophysical InstituteofKarlsruheUniversityasaresearchprojectoftheHei‐delberg Academy of Sciences and Humanities. TheWSMteamregardsitselfas'brokers'forthesefunda‐mentaldata:dataofdifferenttypesfromallovertheworldareintegratedintoacompactdatabasefollow‐ingstandardizedprocedures forqualityassignmentand fordata formatconversion.Theresultingdata‐base is available to all via the Internet (http://ww.world‐stress‐map.org)forfurtherinvestigations.

WorldStressMapproject

102

Weareindebtedtonumerousindividualresearch‐ersandworkinggroupsallovertheworldforprovid‐ingdataforthedatabaseandwearelookingforwardtoreceivingnewdataforupgradingthedatabaseinthe near future. For its successful continuation theWSM project is dependent on further data releasefromindustryandacademia.

SincethepresentationoftheWSMprojectinaspe‐cialvolumeoftheJournalofGeophysicalResearchin1992(Vol.97,no.B8),relevantchanges tookplace,notonlyconcerningtheamountofdataavailableintheWSMdatabase,butalsoregardingnewmethodsfor stress investigations (e.g. drilling‐induced frac‐tures,boreholeslotter)aswellas improvedqualitycriteria for geological indicators and for boreholebreakouts.Here,wegiveanoverviewaboutthestate‐of‐the‐artofthedatabasecomplementedbyanout‐lookonthecapabilityofstressdataforpracticalap‐plications in industry and academia. Thereby, weconcentrateonthenewfeaturesoftheWSM;forbasicinformationabouttheprojectwereferreaderstotheJournal of Geophysical Research special volume (asmentionedabove).DatabaseandInternetaccessInformation about the tectonic stress field in theEarth'scrustcanbeobtainedfromdifferenttypesofstress indicators, namely earthquake focal mecha‐nisms,wellborebreakouts,hydraulicfracturingandovercoringmeasurements, and young (Quaternary)geologicalindicatorslikefault‐slipdataandvolcanicalignments.The reliabilityandcomparabilityof thedataareindicatedbyaqualityrankingfromAtoE,withAbeingthehighestqualityandEthelowest(forfurther investigations only the most reliable datawithaqualityofA,BorCshouldbeused).Anover‐viewaboutthequalityrankingschemeisgiveninZo‐back (1992) as well as on the WSM website(http://www.world‐stressmap.org).

TheWSMwebsitegivesadetaileddescriptionofthedatabasecomprisingnotonlydataontheorien‐tationofthemaximumhorizontalcompression( ),whichareusuallyplottedonthestressmaps,butalsoinformationsuchas thedepthof themeasurement,themagnitude,thelithologyandsoon.ThecompletedatabasecanbedownloadedfromthewebsiteeitherasdBaseIVorasanASCIIfile.Inaddition,thewebsiteprovidesinstructionsforstressregimecharacteriza‐tion,abstracts fromthe twoWSMEuroconferences,interpretationsofthestressfieldinselectedregions,

guidelines for data analysis (which until now areavailableforboreholebreakouts),andmuchmore.

NumerousstressmapsfordifferentregionsoftheworldareavailableontheIntemetaspostscriptfiles.AsanadditionalservicetheprogramCASMO(CreateAStressMapOnline)offersthepossibilitytocreateuser‐definedstressmaps,forexamplebyaddingto‐pography, plotting only data of a specific type (e.g.only earthquakes), or plotting data from a certaindepthrange.CASMOisavailableonrequestandtakeslessthantwohoursforareturnofthestressmapviae‐mail.AllstressmapsareplottedwithGMTprovidedbyWessel& Smith (1991, 1998; http://www.soest.hawaii.edu/gmt/); plate boundaries come from thePLATESproject(http://www.ig.utexas.edu/research/projects/plates/plates.html).ThenewreleaseoftheWSMThenewreleaseoftheWSM(Mülleretal.2000;Fig.1)encompasses10920datarecords,eachwithupto56detail entries. As in the earlier releases (e.g.Zoback1992;Mülleretal.1997)mostdatacomefromearth‐quakes(63%;withHarvardUniversitybeingthema‐jor 'data contributor'; CMT solutions; http://www.seismology.harvard.edu/projects/CMT/) and bore‐holebreakouts(22%;Table1,Fig.2).Abouttwo‐thirdofthedata(68%)haveaqualityofA,BorCandthuscanbeusedforfurtherinvestigations(DandEqualitydataaretoounreliable,butarekeptforbook‐keepingpurposes in order to inform future researchers thatthese data have already been analysed). New datamainlycamefromEurope,AustraliaandAmerica.

TheWSMdatabaseisdesignedasatooltoworkwithstressdata.Duetothestructureofthedatabasethedatacanbeselectedaccordingtoanumberofcri‐teriasuchastype,location,regime,depth,andsoon.Additionally,datafrommid‐oceanridgeswhichmaybedirectlyrelatedtoplateboundaryprocessesandwhichhadsofarbeenexcludedfromtheWSMdata‐basearenowincluded.Thesedatawithlessthan2°distance to thenextplateboundary aremarkedby'PBE'(PossiblePlateBoundaryEvent)inthelastfieldofthedatabase(fieldPBE),sothattheycaneasilybefilteredifnecessary.NewdatabasestructureAthree‐lettercountrycodeisusedforthenumberingofthedata;itisbasedonISO3166providedbytheUnited Nations (http://www.un.org/Depts/unsd/

B.Sperneretal.

103

methods/m49alpha.htm).Weoptedforanewcoun‐trycodetoeliminateconfusionoccurringwiththeoldcountrycode.The lattersubdividedsomecountriesinto several (tectonic)units.However, this subdivi‐sionwasnotdoneconsistently(somecountrieslikeMexico were subdivided into numerous subunits,whileotherlargecountrieswerenotsubdivided)wedecidedfortheoptiononecountry=onecodeaccord‐ingtointernationalstandards.Informationaboutthetectonicunitcanbegiven/foundinthefields'LOCAL‐ITY'or'COMMENT'.Forthesamereasons,theoceansareno longersubdivided intodifferentregions,butallgetthecode'SE'.Thethree‐lettercountrycodeisnotedinthenewfield'ISO'attheverybeginningof

eachdatarecord; the field 'SITE'with theoldnum‐beringremainspartofthedatabasetoguaranteethecomparabilitywitholderreleases.Toavoidthemil‐lennium bug (Y2K problem) all fields containing adate,i.e.thefields'DATE','REFI'to'REF4',and'LAST‐‐MOD',wereextendedbytwocharactersandtheyearisnowshownasfour‐digitnumber.NewqualityrankingschemesDuringtheoperationoftheWSMdatabaseandfromdiscussionwithotherresearchers,theoriginalqual‐ity ranking schemes have been modified and im‐proved. This is sometimes related to clarity of the

Table1.DistributionofdatatypesintheWSMdatabase

Datatype(abbreviations) Number(A‐E) Percentage Number(A‐C) Percentage

Focalmechanism(FMS,FMA,FMC) 6893 63.12 4894 66.30Boreholebreakouts(BO,BOC,BOT) 2414 22.11 1683 22.80Geological:fault‐slip(GFI,GFM,GFS) 373 3.42 321 4.35HydroFrac(HF,HFG,HFM,HFP) 314 2.88 205 2.77Overcoring(OC) 615 5.63 104 1.41Geological:volcanicalignments(GVA) 223 2.04 101 1.37Drilling‐inducedfractures(DIF) 44 0.40 36 0.49Boreholeslotterdata(BS) 33 0.30 29 0.39Petalcentrelinefracture(PC) 9 0.08 9 0.12Shearwavesplitting(SW) 2 0.02 0 0

Sum: 10920 100.00 7382 100.00

Fordetailsaboutthedatatypesseehttp://www.world‐stress‐map.org

Figure1.ThenewreleaseoftheWorldStressMap(WSM2000).Linesshowtheorientationofthemaximumhorizontalcompression( );differentsymbolsrepresentdifferentdatatypes,differentsymbolsizescharacterizedifferentdataqualities;differentcoloursandfillingsindicatedifferentstressregimes(red/unfilled,normalfaulting(NF);green/half‐filled,strike‐slipfaulting;(SS);blue/filled,thrustfaulting(TF);black,unknown(U)).

WorldStressMapproject

104

qualityassignmentinstructionsandsometimestotheuseofdifferenttoolsforrecordingofthedata.Uptonow,themodificationiscompleteforthreetypesofstress indicators, geological indicators (especiallyfault‐slipdata),boreholebreakouts,andovercoringdata.Otherswillfollow.Expandedqualityrankingschemeforgeologicalindicators(GFI,GFM,GFS)Manymethodsforanalysing fault‐slipdata(e.g.All‐mendingeretal.1989;Angelier1984)andnumerouscomputer programs have been developed (e.g.Sperner et al. 1993; Sperner 1996; Delvaux et al.1997). Nevertheless, some disagreement prevailsconcerning themeaningof thedeterminedorienta‐tions, whether they represent strain rate or stress

orientations;forareviewseeTwiss&Unruh(1998).Wedonotintendtointerferewiththatdiscussionorto describe all thesemethods and the related soft‐ware. Insteadwewant to provide a useful tool forqualityranking.

Qualityrankingforfault‐slipdatahassofarbeendefinedthroughthefollowingcriteria:Aquality: inversionof fault‐slipdata forbest fittingmeandeviatoricstresstensorusingQuaternaryagefaults(GFIdata).Bquality:slipdirectiononfaultplane,basedonmeanfault attitude andmultiple observations of the slipvector;maximumstressat30°tofault(GFMdata).Cquality:attitudeof faultandprimarysenseofslipknown,noactualslipvector(GFSdata).Dquality:(a)offsetcoreholes,(b)quarrypop‐ups,or(c)postglacialsurfacefaultoffsets.Due to this quality ranking GFI data (inversion offault‐slipdata)arealwaysassignedqualityA.How‐ever,thequalityofthesedataisdependentonseveraldifferentcriteriasuchasthenumberofdataandtheaverage misfit between calculated and theoreticalslip direction.We use these criteria to introduce anewquality ranking forGFIdata.Foreachof thesecriteria 'subqualities' ranging from 'a' to 'e' are de‐fined(Table2).Theoverallqualityoftheanalysisisgivenbyitslowestsubquality.QualitycriteriaforGFIdata(a)Numberofdata.Thenumberofdatausedforthecalculationofthestressaxes.(b)Percentageofdata.Thepercentageofdatarela‐tivetothetotalnumberofdatameasuredinanout‐crop.Withmorethanoneeventrecordedinanout‐crop,thequalitywill(inmostcases)decreasereflect‐ing thedifficulty tocorrectly separate thedata intosubsets.(c)Confidencenumber.Thereliabilityoftheresultsof

Table2.Qualityrankingschemeforfault‐slipdata(GFI)

Subquality Numberofdata

Percentageofdata

Confidencenumber(slipsensedeterm.)

Fluctuation(°)(averagemisfit)

Datatype

a ≥25 ≥60 ≥0.70 ≤9 ≥0.90b ≥15 ≥45 ≥0.55 ≤12 ≥0.75c ≥10 ≥30 ≥0.40 ≤15 ≥0.50d ≥6 ≥15 ≥0.25 ≤18 ≥0.25e <6 <15 <0.25 >18 <0.25

Theoverallqualityisdefinedbythelowestsubqualityofanyofthesecriteria.

Figure2.DistributionoftheWSM2000datawithrespect to(a)datatype(seeTable1forabbreviations)and(b)quality(A,best,E,worstquality).Mostdatacomefromearthquakes(63%;FMS,FMA,FMC) andboreholebreakouts (22%;BO,BOC,BOT).ThemostfrequentqualityisCduetothefactthatmostearthquakedataareHarvardCMTsolutionswhichareassignedqualityC(fordiscussionseeZoback1992).

B.Sperneretal.

105

fault‐slipanalysisisverysensitivetothereliabilityofslip sense determination. Hardcastle (1989) intro‐ducedfourconfidencelevels(1‐4)forslipsensede‐terminationwhich are assigned to the single faultsduring field measurements. We gave these confi‐dencelevelsdifferentweightsinthefollowingway:1,absolutelycertain: 1.002,certain: 0.753,uncertain: 0.504,veryuncertainorunknown: 0.25The'confidencenumber'isdefinedastheaverageofthesevaluesforalldatausedinthefinalanalysis.Ifnoconfidencelevelswereassignedduringfieldmeas‐urementstheoverallqualityassignedduetothe'rest'ofthecriteriaistakenanddowngradedbyoneclass(e.g.fromqualityBtoC).(d) Fluctuation. The fluctuation is defined as thearithmeticmeanbetweenthemeasuredandthethe‐oreticalslipdirections(thetheoreticalslipdirectionisassumedtobeparalleltothedirectionofmaximumshearstressalongthefaultplane).(e)Typeofdata. Information fromother structuresliketensionjointscangiveadditionalconstraintsonthecalculatedstresstensor.ThecomputerprogramTENSORwrittenbyDelvaux(Delvauxetal.1997;Del‐vaux& Sperner 2003) includes these data into thestressanalysisbyusingdifferentweightsforthedif‐ferenttypesofstructures:Slickenside 1.00Tension/compressionjoint 0.50Twoconjugatedplanes 0.50Movementplane+tensionjoint 0.50Shearjoint 0.25Theaveragevalueforallstructuresusedinthestressanalysisgivesthe'data‐typenumber'forTable2.Forpure fault‐slip data sets this number (1.00) has norelevance.(f)Diversityoforientationoffaultplanesandoflinea‐tions.Fault‐slipanalysisisbasedonanequationsys‐temwith four unknowns (Angelier&Goguel 1979)due to the search for the orientation of the threestressaxesandforthestressratio.Thusatleastfourdifferentlyorientedfaultplanesarenecessarytore‐veal an unambiguous result. Unbalanced data setswithfaultplanesandlineationswithsimilarorienta‐tionswillnotgiveclearresultsforboththeorienta‐tionofthestressaxesandthestressratio.Datasets

including only two fault plane orientations (conju‐gatedfaultsets)givebetter,butstillnotoptimumre‐strictionsconcerningthestressaxes,whilethestressratio is still undefined. Thus, the diversity of themeasureddataensuresthenon‐ambiguityofthecal‐culatedresult.Ameasureofthediversityisgivenbythenormalizedlengthofthemeanvectorofthefaultplane poles and of themean slip line vectors. Thisnormalizedlengthiscloseto1forparallelplanesorparallellineationsandcouldbeusedasanadditionalqualitycriterion.Butincontrasttotheotherqualitycriteria(whichareallstandardmeasuresinfault‐slipanalysis)thediversitycannotbeeasilyobtainedfrompublisheddata.Itcanonlybeestimatedfromthedataplots,butforanaccuratecalculationtheoriginaldatamustbeavailable indigital form.Thus,wegaveupourinitialintentiontoincludethediversityintothequalityranking,butwepleadforacarefulhandlingoffault‐slip data concerning this problem. Data setswhichshowaclearlyunbalancedfaultplaneorienta‐tionmightbedowngradedbytheWSMauthorities.

Inthiscontextthequalityrankingoftwootherge‐ologicalindicators,i.e.GFMandGFSdata(seedefini‐tionabove),hastobecriticallyreviewed.Theirqual‐ityassignment is justified for recentlyactive, large‐scaledstructuresasdescribedbyZoback(1992),i.e.for earthquake fault scarps. But the fact that thesedata build one data class together with GFI data(fault‐slip inversion; see above),which are derivedfrommicrostructures,brings thedanger thatsmall‐scaled structures areused in the sameway. In thiscase,theGFMandGFSdataaredowngradedtoqual‐ityCandD,respectively(Table3);themotivationforthisisgiveninthefollowingparagraphs.

GFMdataarederivedfromsinglefaultplanesun‐dertheassumptionofanangleof30°betweenmaxi‐mumcompressionaxesandfaultplane(withthefaultplanepole,thelineationandthecompressionaxisly‐inginoneplane).Fornewlyformedfaults,e.g.inQua‐ternaryrocks,thisassumptionmightbereasonable,butforreactivatedfaultsotheranglesareasprobableastheusuallytaken30°(e.g.SanAndreasFaultwithc.80°;Zobacketal.1987).ThisuncertaintymakesitnecessarytodowngradedataofthistypetoqualityC,insteadofBasintheoldqualityrankingscheme.This

Table3. Additionalconstraintsforthequalityrankingof GFMandGFSdata

Datatype GFM GFS

Largescale(earthquakerelated) B CSmallscale(microstructures) C D

WorldStressMapproject

106

isadditionallyjustifiedwhentakingintoaccountthatdatafromfault‐slipinversion(GFIdata)withqualityBarebasedonatleast15fault‐slipdatawithanav‐erage misfit angle less than 12°. GFM data do notreach this level of accuracy and thus represent alowerquality(i.e.qualityC).

ThenextlowerlevelofqualityisrepresentedbyGFSdatawhicharederivedfromfaultplaneswithun‐known slip vector, but with 'primary sense of slipknown'(duetotheoldqualityrankingscheme).Usu‐allythis'primarysenseofslip'isderivedfromtheoff‐setofplanarstructures(e.g.beddingplanes).Aslongasnootherplanarstructure,obliquetothefirstone,givesadditionalinformationontheslipdirection,nostatementabouttheslipdirectioncanbemade.Theoffset of one‐dimensional features (e.g. longish fos‐sils)givesanunambiguousindicationfortheslipdi‐rection, but they are rarely found offset in nature.Thus,theoffsetalongafaultusuallydoesnotclearlyshowtheslipdirectionfromwhichtoderivetheori‐entationoftheprincipalstressaxes.Consequentlywedowngraded thequalityassignment for this typeofdata(GFSdata)toqualityD.Expandedqualityrankingschemeforboreholebreakouts(BO,BOC,BOT)Breakoutsdevelopwhenthestressconcentrationattheboreholewallexceedstherockstrength.Theyoc‐curinthedirectionoftheminimumhorizontalstress,90°offtheorientationofthemaximumhorizontal

stress .Inthelasttwodecadesofbreakoutinvesti‐gationloggingtoolshaveimprovedandthehorizon‐tal aswell as the vertical resolution (a) of the rec‐ordeddatahaveincreased.Avarietyofbreakoutin‐terpretationmethodsbasedonthesehighresolutiontools have beendeveloped. To assist the individualresearcher and to keep data comparable the WSMteamsetupguidelinesofdataanalysisforboreholebreakoutinvestigation.(a)Criteriaforbreakoutidentification.Numerousau‐thorshavetreatedtheproblemofbreakoutidentifi‐cation (e.g. Fordjor et al. 1983; Plumb & Hickman1985;Peska&Zoback1995;Zajac&Stock1997).FortheWSMdatabasewesuggestthefollowingcriteriafortheidentificationofbreakoutsfromboreholege‐ometrytoolssuchasBGT,SHDT,FMS,FMI.Thesecri‐teriaaremodifiedafterPlumb&Hickman(1985)andZajac&Stock(1997).(1) Tool rotates freely below and above the

breakout interval.Rotationhas to stop in thebreakoutinterval.

(2) Caliperdifferencehastoexceed10%ofbitsize.(3) The smaller caliper has to range between bit

sizeand110%ofbitsize.(4) The length of the breakout zone has to be at

least1m.

Figure3. (a)Distributionof four‐armcaliperdata(Pad‐l,3azi‐muth:darkgrey;Pad‐2,4azimuth:lightgrey).DuetothePad‐l,3maximumorientation'A'seemstorepresentthebreakoutorien‐tation. (b)Contourplot of caliperdata.Thedistributionof thedataindicatesthatthetoolstucktoonesideofthebreakout(dueto cable torque), so that orientation 'A' doesnot represent thecorrectbreakoutazimuth.

B.Sperneretal.

107

(5) The direction of elongationmust not consist‐entlycoincidewiththeazimuthofthehighsideoftheborehole.Thus,forboreholeswhichde‐viatemore than 1° from vertical, the angulardifferencebetweenthedirectionoftheholeaz‐imuth and the azimuth of the greater caliperhastoexceed15°.

(b)Recommendationsforbreakoutanalysis(1) Usebreakoutidentificationcriteria(seeabove).(2) Usevisualcontrol(forcaliperdata):iftheanal‐

ysisispurelybasedonthestatisticsofPad‐Iaz‐imuth(Fig.3a),directionAwouldbeobtainedasbreakoutazimuth.But,duetocabletorque,thetoolmaysticktoonesideofthebreakoutand Amay be off the true breakout azimuth.Contourplotshelptoavoidthissourceoferror(Fig.3b).

(3) Usecircularstatistics(e.g.Mardia1972).(4) Correctthe azimuthaccordingtomagnetic

declinationatthetimeoflogging.(5) Checkforkeyseats:ameanstocheckforkey

seateffectsistoplotbreakoutorientationwithdepth including thedirectionofhole azimuth(Fig.4).

(c)Qualityrankingschemeforboreholebreakouts.Forboreholebreakoutsthequalityisultimatelylinkedtothestandarddeviation,thenumberofbreakoutsandthetotallengthofbreakouts(Zoback1992)(Table4).Theoverallqualityoftheanalysisisgivenbyitslow‐estsubquality.Expandedqualityrankingschemeforovercoringdata(OC)The previous quality ranking for overcoring meas‐urementscontainedsomeunclearformulations(e.g.'multiple consistent measurements') which will beremoved.Forthesakeofconsistency,overcoringdatawillbetreatedlikeboreholeslotterdatabecausetheyfacesimilarproblemsinqualityranking.(FordetailsseesectiononBoreholeslotterdataandTable6.)NewmethodsTwonewtypesofstressindicatorsarenowincludedin the WSM database: drilling‐induced fractures(DIF) and borehole slotter data (BS). Their qualityrankingsaredescribedindetailbelow.Drilling‐inducedfractures(DIF)Theavailabilityofhighresolutionloggingtoolssuchas FormationMicroScanner (FMS), FormationMi‐croImager(FMI)orBoreholeTeleviewer(BHTV)en‐ablestheanalysisoftensilefailurephenomenaattheborehole wall. In theWSM database tensile failurefrominducedhydraulictesting(hydrofrac)isalreadyincluded. In addition to the stimulated hydrofracs,

Table4. Qualityrankingschemeforboreholebreakouts(BO,BOC,BOT)

Subquality Numberofdata

Totallength(m)

Standard deviation(°)

a ≥10 ≥300 ≤12b ≥6 ≥100 ≤20c ≥4 ≥30 ≤25d <4 <30 ≤40e ‐ ‐ >40

Theoverallqualityisdefinedbythelowestsubqualityof anyofthesecriteria.

Figure4. Identificationof key seats. In the deeper section thebreakoutsfollowthedirectionoftheholeazimuthindicatingpos‐siblekeyseats.Deviation:anglebetweenboreholeandverticaldirection;holeazimuth:dipazimuthofnon‐verticalborehole.

WorldStressMapproject

108

unintentional tensile failure created during drillingmayoccur(Fig.5a).Thesedrilling‐inducedfractures(DIF)canbeusedforstressdetermination(Wiprutetal. 1997; Fig. 5b). Identification criteria have beenproposed,forexamplebyBrudy&Zoback(1999)andAsk (1998). We basically follow their criteria withsomeminormodifications.(a)CriteriaforDIFidentification(1) Fracturesthataresubparallel totheborehole

axis, occur in pairs on opposite sides of theboreholewall and are not connected to eachother (not to bemisinterpreted as sinusoidalsteepfractures).

(2) Short en‐echelon fracture traces (also occur‐ring inpairs)but inclinedwithrespect to theboreholeaxis.Theyresultfromstressdistribu‐tion at the borehole wall when the boreholeaxisdoesnotcoincidewithoneoftheprincipalstressaxes.

(3) ThelengthofDIFsrangesbetween0.1and2m.(4) IfbreakoutobservationsareavailableDIFand

breakoutorientationsshoulddifferby90°.(5) The shape of DIFs should be irregular since

perfectly straight features could result frommechanicalwearoftheborehole.

(b)QualityrankingschemeforDIFs.Thequalityrank‐ingschemeforDIFsisbasedonthenumberoffrac‐tureswhichhadbeenidentified,thestandarddevia‐tionoftheiraverageandthetotalfracturelength(Ta‐ble5).Thequalityrankingschemeissimilarinstruc‐turetothebreakoutqualityranking.Boreholeslotterdata(BS)Boreholeslotterdataareobtainedbyastressreliefmethod(noovercoring)whichisdescribedindetailbyBock&Foruria(1983),Bock(1993),andBecker&Werner(1994).Resultsofboreholeslottermeasure‐mentsagreewithdatafromothermethods(e.g.bore‐holebreakouts;Becker&Werner1994).Becker&Pal‐adini(1990)developedaqualityrankingschemeforin situstressdatawhichweusedas abasis for the

Table 5. Quality ranking scheme for drilling induced fractures(DIF)

Subquality Numberofdata

Totallength(m)

Standard deviation(°)

a ≥10 ≥300 ≤12b ≥6 ≥100 ≤20c ≥4 ≥30 ≤25d <4 <30 ≥40e <4 <30 >40

Theoverallqualityisdefinedbythelowestsubqualityof anyofthesecriteria.

Figure5.(a)Boreholeteleviewerimagewithverticalfracturetraceswhichareinterpretedasdrilling‐inducedfractures.Colourscor‐respondtotheamplitudeofthereflectedacoustic(ultrasonic)pulseandthustothereflectioncoefficientoftheboreholewall(lowamplitude=rough,fracturedboreholewallsuchasinbreakoutintervals).(b)Sketchofaboreholewithdrilling‐inducedfracturesparalleltothemaximumhorizontalcompression( )andboreholebreakoutsperpendicularto .

B.Sperneretal.

109

(new)WSMqualityranking.Weadaptedittotheal‐ready existing criteria for similar stress indicatorslikeovercoringorbreakoutdata. Importantcriteriaare the depth ofmeasurement, the number of dataand the standard deviation of the principal stressaxesorientation(Table6).Boreholeslottermeasure‐mentsnormallyarewithinthefirst30mofabore‐hole,sothattopographiceffectsorstressdeflectionsaround underground openings might influence theresult.Thusthedistancetotopographicfeatures(e.g.quarrywalls)andexcavationwallsinthecaseofun‐derground openings is important: the distance totopographicfeaturesmustbelargerthanthreetimestheheightofthetopographicfeatureandthedistancetoexcavationwallsmustbelargerthantwotimestheexcavation radius (Fig. 6). All measurements atsmallerdistancesareassignedqualityE.QualityassignmentanddepthofmeasurementWeareawarethatthequalityassignmentcannotbetransferredfromonemethodtoanother.ResearchersmayarguethattheassignmentofqualifiesAorBtoboreholeslotter (BS)orovercoring (OC)data is re‐stricted to measurements at depths greater than100m,whichisinmostcasesequivalenttomeasure‐mentsintunnelsorminessincethedrillholesforBSandOCdatausuallyarelessdeep.Incontrasttobore‐holeslotterandovercoringdata,thequalityrankingfor geological fault‐slip data (GFI), drilling‐inducedfractures (DIF), and borehole breakouts (BO) doesnotcontainsuchadepthcriterion.Forthelattertwoitisconsideredtobeunnecessarybecausethesedataresultfrommeasurementsinindustrialwellsatsev‐eral hundred metres depth. In addition, the depthconditionforBOsandDIFsissubstitutedbythecon‐ditionsoftotallengthofBOorDIFdata,respectively.

GFI data are surface data in terms of measure‐ments. However, their originmay be from greater,

but unknown, depth. To account for the missingdepthconditioninthequalityrankingoftheGFIdataa larger number of data is required. For example,qualityAcanbeobtainedwithaminimumof25dataforGFImeasurements,whileonly11areneededforBSandOCmeasurements.ApplicationoftectonicstressdataNumerous fields of applications exist for tectonicstressdata.Theyarenotrestrictedtopure'academic'research projects, but also include investigationswith large economic impact, e.g. in hydrocarbonorgeothermal energy production. In the followingwegiveexamples forbothtypesofapplication.Forthe'academic'exampleweselectedtheeasternMediter‐raneanregionwherenumerousstressdataareavail‐able andwhere geodeticmeasurements (GPS, SLR)providethepossibilitytocomparebothdatasets.The'economic'examplesdealwiththestabilityofunder‐ground openings and with hydrocarbon reservoirmanagement.StressandvelocityintheeasternMediterraneanregionTectonic overview. The Neogene to recent tectonicevolution of the eastern Mediterranean region is

Table6. Quality ranking scheme for borehole slotters (BS) andovercoring(OC)data

Subquality Numberofdata

Totallength(m)

Standard deviation(°)

a ≥11 ≥300 ≤12b ≥8 ≥100 ≤20c ≥5 ≥30 ≤25d ≥2 ≥10 ≤40e <2 <10 >40

Theoverallqualityisdefinedbythelowestsubqualityof anyofthesecriteria.

Figure6.Minimumdistancedofstressmeasurementsfrom(a)topographicfeatures(threetimestheheightofthetopographicfea‐ture)and(b)undergroundopenings(twotimestheexcavationradius).

WorldStressMapproject

110

drivenbythenorth‐southconvergencebetweentheAfrican and Arabian plates, and the Eurasian plate.TheAfricanplatemovesatavelocityof9mma‐1andtheArabianat25mma‐1relativetotheEurasianplate(DeMetsetal.1994).Anassemblageof lithosphericblockstrappedinthisconvergencezonemovesinde‐pendentlybetweenAfrica,ArabiaandEurasiaresult‐ing inobliquecontinentalcollisionsandthus in theclosevicinityofcollisionandsubductionzones(e.g.McKenzie1972).SubductionoccursbeneaththeHel‐lenicArc,whilecontinentalcollisionalreadystartedin its lateral continuations towards theNW(Dinar‐ides)andtowardstheeast(CypreanArc)(Fig.7).Thesouthwestward retreat of the Hellenic Arc subduc‐tionzonecausesbackarcextensionintheAegeanandenables the westward extrusion of the Anatolianblockoutofazoneofcontinentalcollisionattheeast‐ernendoftheAnatolianblock(e.g.McKenzie1970;LePichon&Angelier1979;Royden1993).Thiscollisionistheresultofthenorth‐northwestwardmovementoftheArabianplatewhichisseparatedfromtheAfri‐canplatebytheDeadSeafaultandtheRedSearift.Stress data. The majority of the stress data comesfromfault‐slipanalysis(foracompilationseeMercieretal.1987)andfromearthquakedata(e.g.Jackson&

McKenzie1988).IntheAegeanandwesternAnatolianormalfaultingisthedominantstressregimewith orientationstrendingNWtoNEindicatinganapprox‐imatelynorth‐southextension(Fig.8).Inthenorth‐ernpartofcentralAnatoliastrike‐slipfaultingistheprevailingstressregime.

North‐southextensioninthenorthernAegeanSeaandwesternAnatoliahadbeeninterpretedtobetheexpression of backarc extension due to subductionzone retreat of the Hellenic Arc (e.g.Müller et al.1992).Thefan‐shapedarrangementof (minimumhorizontalstress; )mightresultfromthesuc‐tionforceactingradiallyoutward(i.e.southwest‐tosoutheastward)causedbyslabrollbackasshowninthe numerical models of Meijer & Wortel (1996).Strike‐slipmovementsinthisregionandinnortherncentral Anatolia are most probably related to thewestwarddirectedlateralextrusionoftheAnatolianblockwithdextralmovementsalongtheNorthAna‐tolianfault(Fig.7).Velocitydata(GPS,SLR).Geodeticobservationswereinitiated in the mid‐1980s, so that today accuratedataaboutthevelocityvectorsareavailableformostof the observation sites (e.g. Noomen et al. 1996;Reilingeretal.1997;Kaniuthetal.1999;Fig.8).They

Figure7.TectonicmapoftheeasternMediterraneanregion.LateralextrusionoftheAnatolianblocktowardsthewestistriggeredbycontinentalcollisionoftheArabianplatewiththeEurasianplateincombinationwithsubductionzoneretreatattheHellenicarc.

B.Sperneretal.

111

indicate west‐directed movements of Anatolia andsouthwestwardmovementsoftheAegean.Thetran‐sitionzonebetweenbothblocks(westernTurkey)ischaracterized by westsouthwestward movements.Theinfluenceofplateboundaryforcesanddifferentmaterialpropertiesonthesemovementshavebeeninvestigatedbyfiniteelementmodelling.FordetailsandreferencesseeHeidbach&Drewes(2003).Othermodels reconstructing the velocity vectors in theeastern Mediterranean assume two individuallymoving rigid lithospheric blocks: (1) the clockwiserotating Aegean block encompassing the southernandcentralpartsoftheAegeanandthesouthernpartofGreece(southoftheGulfofKorinth);and(2)thecounterclockwise rotating Anatolian blockwhich isbordered by the North and East Anatolian faults(Fig.7;e.g.Drewes1993;LePichonetal.1995).Thisapproach fits with palaeomagnetic datawhich alsoshowopposite rotation senses for both blocks (e.g.Kissel et al. 1987; Kissel & Laj 1988). In contrastReilingeretal.(1997)modelledtheirsetofvelocityvectorswithasingleAegean‐Anatolianblock.Ingen‐eralagoodagreementexists,butsignificantdiscrep‐ancies occur in western Anatolia and the southernAegean.Comparisonofstressandvelocitydata.FortheeasternMediterranean both types of data are explained byidenticalprocesses (seeabove).Nevertheless, thereis no straightforward relation between them and anumber of aspects have to be taken into account

whencomparingstressandvelocitydata.(1)Velocityfieldsreflectnotonlystrain(i.e.inter‐

nal deformation), but also translation and rotation(somepeopleuse theexpression 'strain' inawidersense including translationandrotation,but in thispaperweuseitinthemorerigorouswaywhichex‐cludes these twoprocessesand is restricted todis‐tancechangesbetween thepointsofabody).Thus,firstrigidbodymovementshavetoberemoved.Forlarge‐scalevelocityfields,whichhavetotakeintoac‐count the curvature of the Earth, all rigid plate orblockmovementsarerotationsaroundaEulerpole.Accordingly,Reilingeretal.(1997)calculatedthein‐ternal deformation of the eastern MediterraneanfromgeodeticdataandfoundthatitissmallincentralAnatolia.Incontrast,theplateandblockboundariesarecharacterizedbybroaddeformationzones.

(2)Axesofprincipalstrainrateandstressareonlyin somecasesparallel to eachother (coaxialdefor‐mation,e.g.pureshear); inmostcasestheyareori‐ented obliquely to each other (non‐coaxial defor‐mation,e.g.simpleshear).Additionally,therelation‐shipbetweenstressandstrainratemightbenon‐lin‐ear(e.g.strainhardeningorsoftening).

(3)Stressdataarecompiledthroughoutthewholecrustwhilegeodeticdataexclusivelyreflectsurfacemovements. Decoupling between different layersmightresultindifferentorientationsofstressaxesatdifferentdepths(e.g.Brereton&Müller1991;Jarosin‐ski1998). IntheeasternMediterranean orienta‐tionsfromtheseismogenicupper15kmofthecrust

Figure8.VelocityandstressdatafortheeasternMediterraneanregion.GPSdatacomefromReilingeretal.(1997)andKaniuthetal.(1999),SLRdatafromNoomenetal.(1996),andstressdatafromtheWorldStressMapdatabase(Mülleretal.2000).ForfurtherexplanationsofthestresssymbolsseeFigure1.

WorldStressMapproject

112

areingoodagreementwith orientationsfromthenumeroussurfacedata(geologicalindicators;Fig.8).Incontrast,thestressregimeshowsadepthdepend‐enceinthecentralandnorthernAegean:surfacedataindicate a normal faulting regime, while data fromdepthsofmorethan5kmshowstrike‐slip faulting.Thechangefromnormalfaultingtostrike‐slipfault‐ingregimeisincontradictiontotheregimechangesobservedelsewherewithdepthwhicharecausedbyincreasingoverburdenandthusarecharacterizedbyachangefromstrike‐sliporthrustfaultingtonormalfaulting.Oneexplanationcouldbeagravitationalcol‐lapseofthethickercrustwithitshighertopographyinthenorthernmostpartoftheAegeantowardsthecentralAegeanwithitsthinnercrustanditslowerto‐pography.Inthiscaseadecouplingofthesouthwardsliding uppermost crust from the underlying rockswouldbenecessary.

(4)Bothdatatypesrepresentdifferenttimeperi‐ods.While geodetic data reflectmovements duringthe lastdecadeorevenshorterperiods, stressdatacome from several decades (earthquake data) orevenfromseveraltenstohundredsofthousandsofyears(geologicalindicatorswhichcomefromQuater‐nary rocks). Deformation of the Earth's crust takesplaceinanon‐continuousway,implied,forexample,by the locally observed clustering of earthquakesthroughtime(aseismicallyactiveperiodisfollowedbyaninactiveperiod)orbyprogressivefailurealongafault(Steinetal.1997).Theseismiccycleofstrongearthquakes can be as long as several hundreds ofyears(Jacksonetal.1988).FortheeasternMediter‐raneanregionthemajorityofthestressdatacomesfromearthquakesandgeological indicatorsthusre‐flectingamuchlongertimeperiodthanthegeodeticmeasurements in Figure 8 which cover the period1988‐1994 (Reilinger et al. 1997). Nevertheless,McCluskyetal.(2000)foundthatthegeodeticdatafortheNorthandEastAnatolianfaultsareinagreementwithslipratesforthelast4‐5Marevealedfromgeo‐logical data. However, the data of McClusky et al.(2000) do not include observations after the Izmitearthquake in August 1999. Earthquakes have astrongeffectonthevelocityfieldevenatdistancesofhundreds of kilometres or more depending on themagnitudeoftheevent.Forthe 8.0earthquakenearAntofagasta(northChile)Klotzetal.(1999)ob‐servedacoseismicslipof10cmatadistanceof300km.

(5) The processing of geodetic data has not yetbeenstandardized.Thus, the resultsdependon the

choiceofthereferencesystemandofthemodelsin‐cludedintotheprocessing.Othercriticalfactorsarethenumberofobservationsitesandtheobservationtime.Thismakesacomparisonoftheresultsofdiffer‐ent research groups difficult or even impossible.McCluskyetal.(2000)didacompilationofGPSdatafor the eastern Mediterranean which encompassesmoreobservationsitesandalongerobservationpe‐riod (1988‐1997) compared to earlier compilationsby Reilinger et al. (1997). The differing results ofthese two compilations led to different interpreta‐tionsoftheblockmovementsinthisregion.Reilingeretal.(1997)'saw'onlyoneblock,theAnatolian‐Ae‐geanblock,whileMcCluskyetal.(2000)'needed'twoblocks, theAnatolian and theAegeanblocks, to ex‐plainthevelocityvectors(seeabove).ConclusionsThislistof'complications'aswellasthedatafromtheeasternMediterraneanregionindicatethatmuchmore extrawork is necessary to under‐standtherelationshipbetweenstress,strainandve‐locity,andhowtheseparametersdependonchangesin plate tectonic processes andmaterial properties(e.g.changeofviscosityduetoenhancedheatflow)aswellasonlocalinfluencesbyseismicactivity.OnaglobalscaletheILPproject'GlobalStrainRateMap'under theguidanceofW.Holt (SUNYStonyBrook)and J.Haines (CambridgeUniversity) and theWSMareprerequisitesforthiskindofstudy.StabilityofundergroundopeningsAnyundergroundconstructioncausesseveremodifi‐cationsoftheinitialinsiturockstressatthelocationof the site: artificial openings create fresh surfaceswhichmodifytheboundaryconditionsatthelocationofthewalls(mostlyzerosurfacestress).Thisleadstoaspatialdistortionof thestresspatternwithstressaccumulationsaroundtheopening.Thestressredis‐tribution for circular holes in an anisotropicallystressed rock is classically described by the Kirschequations(Kirsch1898).Stressredistributionaroundtunnels and shafts has been treated analytically aswell as numerically for more complicated designs(numerous examples are summarized inAmadei&Stephansson1997).

Sincerock is lessstable in tension, tensilestressconcentrations may open pre‐existing fractures orevenfracturetherock,thusincreasetheweatheringand leading to further instability. Compressivestressesinthevicinityofundergroundopeningscan

B.Sperneretal.

113

result in different kinds of rock failure, e.g. rockburstsorside‐wallspalling.Inboreholesthestress‐induced compressive failure is called boreholebreakout(Fig.5b).

Thestabilityofconstructionsitesdoesnotdependonlyontherocktypebutalsoontheorientationandmagnitudeofthetectonicstressincombinationwiththedesignandorientationoftheundergroundopen‐ing.Innormalfaultingregimes,boreholes(shaftsandmines)aremostlikelytofailundercompressionside‐wayswhentheyaredrilledparalleltothe direc‐tion(Fig.9).Thisis inagreementwiththeobserva‐tion that inplaceswherevertical stresses arehigh,

problemsinthesidewallsofminesoccur.Especiallyinareaswithhightopographyandsteepvalleyslope(forexampleNorwegianfjords)thetunnelsthatpar‐allelthevalley(parallelto )showrockburstsinthetunnelwall (Amadei&Stephansson1997). In thrustfaultingregimesboreholesfailatthetopandbottomifdrilledparallelto .Againthisisinagreementwithmining observations where problems in roofs andfloorsoccurforareasofhighhorizontalstress.Understrike‐slip conditions vertical boreholes and shaftsaremostendangeredforfailurebecausetheyexperi‐encehighstressconcentrationsinthedirectionpar‐allelto .

Figure9.Stabilityofundergroundopeningsduetotheirorientationrelativetotheprincipalstressaxes.Inallthree stressregimes,openingsparalleltotheintermediatestressaxis(greenarrows; innormalfaulting, instrike‐slipfaulting,and inthrustfaultingregime)aremostendangeredforrockfailurewiththefailureoccurringinthedirectionoftheminimumstressaxes(bluearrows;innormalandstrike‐slipfaulting, inthrustfaultingregime).Discsshowstressconcentrations(yellowtoredcolours)aroundtheborehole.

WorldStressMapproject

114

HydrocarbonreservoirmanagementInthefuture,thediscoveryofnewgigantichydrocar‐bonreservoirswillberare,thuscompanieswillfocuson the optimization of the production of existingfields.Thiscoversexploitationofsmallerreservoirsor marginal fields and activities in the deeper off‐shoreareas.Tectonicstressincombinationwiththeactualfracturenetworkcontrolsthefluidflowanisot‐ropyandthusthedrainagepatternwithinthereser‐

voir.Numerousexamplesofastrongcorrelationbe‐tweenthepreferentialdirectionoffluidflowinares‐ervoir and the orientation of are described byHeifer& Dowokpor (1990),Heffer& Koutsabeloulis(1995), and Heifer et al. (1995). Thus, knowledgeabouttheorientationof canbeusedforanopti‐mizedarrangementoftheproductionwells(Fig.10).

One distinguishes the fracture‐induced permea‐bility from the intrinsic permeability of the rockwhich is connected with the available pore space.Both together form the effective permeability (e.g.Connolly&Cosgrove1999).Inparticular,oilorgasre‐coveriesfromreservoirswithlowintrinsicpermea‐bilities require widespread networks of intercon‐nectedfractures(Lorenzetal.1988;Hickman&Dun‐ham1992).Oilrecoveryisenhancedbythecreationof new (artificial) fractures (pathways). A fracturedevelopsawayfromtheboreholebypumpingwaterorgelsintothetargetarea.Itpropagatesperpendic‐ularly to the least principal stress. Thus, advanceknowledgeofthestressorientationsenablespredic‐tionoftheorientationsofstimulationfracturesandthusenhancestheeconomicsuccessoffieldactivities.OutlookTheWSMisdesignedasasecurerepositoryofstressdata fromacademicresearchandeconomic investi‐gations,andasatooltoworkwiththedata.Ithasalong‐term perspective (continuation until at least2008)andthusenablesoilcompanies,forexample,tocomebacktohydrocarbonfieldswhichhadbeencon‐sideredasuneconomicearlier.Thedataarereadilyavailable via the Internet so that time‐consumingsearching in archives can be avoided. Additionally,withCASMOweprovideauser‐friendlytool forthecreationofstressmaps.Thegrowthofthedatabaseby c. 20% comparedwith the 1997 release (1997:9147datasets,2000:10920datasets)justifiesthehopethatthedatacoverageonaglobalscalewillfur‐therimproveduringthenextfewyears,sothatsomeofthelargedatagapscanbeclosed.

Stress observations are fundamental Earth sci‐ence datawhich are used to relate global dynamicmodellingof lithospheric stresseswithplatemove‐ments.Otherscientificchallengesarethecorrelationbetweencrustaldynamicsandvelocityanisotropiesin the uppermantle, regional investigations of tec‐tonicmotions,andthecomparisonofstressdatawithmodem geodetic observations as described for theMediterranean. The increase of stress observations

Figure10.Stress‐controlledflowanisotropy(modifiedafter Bell1996).Duetothepreferredorientationoffracturesparallelto ,thedrainageareaofeachwellisellipticalwiththelongaxispar‐allelto .(a)Thearrangementofwellsduetogeometricaspects(north‐southandeast‐westgrid)requires12wellsfordraining65%ofthereservoir.(b)Thearrangementduetoreservoirflowcharacteristicrequiresonly11wellsfor66%drainageandisthusmoreeconomic.

B.Sperneretal.

115

will promote further research projects leading to abetterunderstandingofthesourcesandtheeffectsoftectonicstresses.

Investigationoftectonicstressisnotonlyofscien‐tific interest, but also has an economic impact byproviding information on the stability of under‐groundopeningslikeboreholes,tunnels,orradioac‐tivewastedepositoriesandthusenhancesthevalueof the data gained by the industry. Moreover, pre‐ventingrockfailurebychoosinganappropriateori‐entation of underground openings and precautionsforstabilizationcanbeconsideredasacrucialsecu‐rityaspectforpeopleworkinginminesandtunnels.TheWorldStressMapproject is financedby theHeidel‐berg Academy of Sciences and Humanities.We thank A.BeckerandD.Nieuwlandfortheirhelpfulandconstructivereviews.ReferencesALLMENDINGER,R.W.,GEPHARDT,J.W.&MARRETT,R.A.1989.NotesonFaultSlipAnalysis.NotestotheGeologicalSocietyofAmericashort course 'Quantitative interpretationof jointsand faults'.CornellUniversity,Ithaca.

AMADEI,B.&STEPHANSSON,O.1997.RockStressand itsMeasure‐ment.Chapman&Hall,London.

ANGELIER,J.1984.Tectonicanalysisoffault‐slipdatasets.JournalofGeophysicalResearch,89,5835‐5848.

ANGELIER,J.&GOGUEL,J.1979.Suruneméthodesimplededéter‐minationdesaxesprincepauxdescontraintespourunepopu‐lationdefailles(Asimplemethodfordeterminingtheprincipalaxesofstressforafaultpopulation).ComptesRendusdel'Aca‐démiedesSciences,Paris,SérieD,288,307‐310.

ASK,M.V.S.1998.In‐situandlaboratorystressinvestigationsusingboreholedatafromtheNorthAtlanticOcean.PhDthesis,Depart‐mentofCivilandEnvironmentalEngineering,RoyalInstituteofTechnology,Stockholm.

BECKER,A.&PALADINI,S.1990.InsituSpannungeninNord‐undMitteleuropa. Schriftenreihe Angewandte Geologie Karlsruhe,10,1‐63.

BECKER, A. & WERNER, D. 1994. Strain measurements with theboreholeslotter.TerraNova,6,608‐617.

BELL,J.S.1996.Insitustressesinsedimentaryrocks(part2):Ap‐plicationsofstressmeasurements.GeoscienceCanada,23,135‐153.

BOCK,H.1993.Measuringinsiturockstressbyboreholeslotting.Comprehensive Rock Engineering ‐ Principles, Practice& Pro‐jects,3,433‐443.

BOCK,H.&FORURIA,V.1983.Arecoverableboreholeslottingin‐strumentforinsitustressmeasurements,notrequitingover‐coring. Proceedings of the International Symposium on FieldMeasurementsinGeomechanics,Zürich,15‐29.

BRERETON,R.&MÜLLER,B.1991.Europeanstress:contributionsfrom borehole breakouts. Philosophical Transactions of theRoyalSocietyofLondon,SeriesA,337,165‐179.

BRUDY,M.&ZOBACK,M.D.1999.Drilling‐inducedtensilewall‐frac‐

tures:implicationsfordeterminationofin‐situstressorienta‐tion and magnitude. International Journal of RockMechanicsandMiningSciencesandGeomechanicsAbstracts,36,191‐215.

CONNOLLY,P.&COSGROVE, J.1999.Predictionof fracture‐inducedpermeability mad fluid flow in the crust using experimentalstressdata.AAPGBulletin,83,757‐777.

DELVAUX,D.&SPERNER,B.2003.Newaspectsoftectonicstressin‐versionwithreferencetotheTENSORprogram.In:NIEUWLAND,D.A.(ed.)NewInsightsintoStructuralInterpretationandMod‐elling.GeologicalSociety,London,SpecialPublications.

DELVAUX,D.,MOEYS,R., STAPEL,G.etal.1997.Paleostress recon‐structionandgeodynamicsof theBaikal region,CentralAsia,Part2,Cenozoicrifting.Tectonophysics,282,1‐38.

DEMETS,C.,GORDON,R.G.,ARGUS,D.F.&STEIN,S.1994.Effectofre‐centrevisionstothegeomagneticreversaltimescaleonesti‐mates of current platemotions.GeophysicalResearchLetters,21,2191‐2194.

DREWES,H.1993.AdeformationmodeloftheMediterraneanfromspacegeodeticobservationsandgeophysicalpredictions.Inter‐nationalAssociationofGeodesy,Symposium,112,373‐378.

FORDJOR,C.K.,BELL,J.S.&GOUGH,D.I.1983.BreakoutsinAlbertaand stress in the North American plate. Canadian Journal ofEarthSciences,20,1445‐1455.

FUCHS,K.&MÜLLER,B.2000.WorldStressMapoftheEarth:akeytotectonicprocessesandtechnologicalapplications.Naturwis‐senschaften,88,357‐371.

GORDON,R.G.&STEIN,S.1992.Globaltectonicsandspacegeodesy.Science,256,333‐342.

HARDCASTLE,K.C.1989.Possiblepaleostresstensorconfigurationsderivedfromstrike‐slipdataineasternVermontandwesternNewHampshire.Tectonics,8,265‐284.

HEFFER,K. J.&DOWOKPOR,A.B.1990.Relationshipbetweenazi‐muthsoffloodanisotropyandlocalearthstressesinoilreser‐voirs.ProceedingsoftheNorthSeaoilandgasreservoirconfer‐ence,251‐260.

HEFFER,K.J.&KOUTSABELOULIS,N.C.1995.Stresseffectsonreser‐voirflow;numericalmodellingusedtoreproducefielddata.In:DEHAAN,H.J.(ed.)NewDevelopmentsinImprovedOilRecovery.GeologicalSociety,London,SpecialPublications,84,81‐88.

HEFFER,K.J.,FOX,R.J.,MCGILL,C.A.&KOUTSABELOULIS,N.C.1995.Noveltechniquesshowlinksbetweenreservoirflowdirection‐ality,Earthstress,faultstructureandgeomechanicalchangesinmature waterfloods. Society of Petroleum Engineers, AnnualTechnicalConference&Exhibition,Dallas,SPE30711.

HEIDBACH,O.&DREWES,H.2003.3DFiniteelementmodelofmajortectonicprocessesintheEasternMediterranean.In:NIEUWLAND,D.A.(ed.)NewInsightsintoStructuralInterpretationandMod‐elling.GeologicalSociety,London,SpecialPublications.

HICKMAN,R.G.&DUNHAM,J.B.1992.Controlsonthedevelopmentof fractured reservoirs in theMonterey Formationof centralCalifornia.In:LARSEN,R.U.,BREKKE,U.,LARSEN,B.T.&TELLERAAS,E.(eds)StructuralandTectonicModellinganditsApplicationtoPetroleumGeology.NorwegianPetroleumSociety,SpecialPubli‐cations,1,343‐353.

JACKSON, J.&MCKENZIE,D.1988.Therelationshipbetweenplatemotions and seismic tremors, and the rates of active defor‐mationintheMediterraneanandMiddleEast.GeophysicalJour‐naloftheRoyalAstronomicalSociety,93,45‐73.

JACKSON,J.,HAINES,J.&HOLT,W.1988.ThehorizontalvelocityfieldinthedeformingAegeanSearegiondeterminedfromthemo‐menttensorsofearthquakes. JournalofGeophysicalResearch,97,17657‐17684.

WorldStressMapproject

116

JAROSINSKI,M.1998.Contemporary stress fielddistortion in thePolishpartof theWesternOuterCarpathiansand theirbase‐ment.Tectonophysics,297,91‐119.

KANIUTH, K., DREWES, H., STUBER, K. et al. 1999. Crustal defor‐mationsinthecentralMediterraneanderivedfromtheWHATACATGPSproject.Proceedingsofthe13thWorkingMeetingonEuropeanVLBIforGeodesyandAstronomy,Wettzell,192‐197.

KIRSCH,G.1898.DieTheoriederElastizitätunddieBedürfnisseder Festigkeitslehre. Zeitschrift des Vereines Deutscher Inge‐nieure,42,797.

KISSEL,C.&LAJ,C.1988.TheTertiarygeodynamicalevolutionofthe Aegean arc: a paleomagnetic reconstruction.Tectonophy‐sics,146,183‐201.

KISSEL,C.,LAJ,C.,SENGOR,A.M.C.&POISSON,A.1987.Paleomagneticevidenceforrotationinoppositesensesofadjacentblocks inthenortheasternAegeaandWesternAnatolia.GeophysicalRe‐searchLetters,14,907‐910.

KLOTZ,J.,ANGERMANN,D.,MICHEL,G.E.etal.1999.GPS‐derivedde‐formationofthecentralAndesincludingthe1995AntofagastaMw=8.0earthquake.PureAppliedGeophysics,154,709‐730.

LEPICHON,X.&ANGELIER,J.1979.TheHellenicarcandtrenchsys‐tem:akeytotheneotectonicevolutionoftheeasternMediter‐raneanarea.Tectonophysics,60,1‐42.

LEPICHON,X.,CHAMOT‐ROOKE,N.,LALLEMANT,S.,NOOMEN,R.&VEIS,G. 1995. Geodetic determination of the kinematics of centralGreecewithrespecttoEurope:ImplicationsforeasternMedi‐terranean tectonics. Journal of Geophysical Research, 100,12675‐12690.

LORENZ,J.C.,WARPINSKI,N.R.,TEUFEL,L.W.,BRANAGAN,P.T.,SATT‐LER,A.R.&NORTHROP,D.A.1988.ResultsoftheMultiwellExper‐iment: insitustresses,natural fractures,andothergeologicalcontrolsonreservoirs.LOSTransactions,AmericanGeophysicalUnion,69,817,825‐826.

MCCLUSKY,S.,BALASSANIAN,S.,BARKA,A.etal.2000.GlobalPosition‐ingSystemconstraintsonplatekinematicsanddynamicsintheeasternMediterraneanandCaucasus.JournalofGeophysicalRe‐search,105,5695‐5719.

MCKENZIE,D.P.1970.PlatetectonicsoftheMediterraneanregion.Nature,226,239‐243.

MCKENZIE,D.P.1972.Active tectonicsof theMediterraneanre‐gion.GeophysicalJournaloftheRoyalAstronomicalSociety,30,109‐185.

MARDIA,K.V.1972.StatisticsofDirectionalData.AcademicPress,NewYork.

MEIJER,P.Th.&WORTEL,M.J.R.1996.TemporalvariationinthestressfieldoftheAegeanregion.GeophysicalResearchLetters,23,439‐442.

MERCIER,J.L.,SOREL,D.&SIMEAKIS,K.1987.Changesinthestateofstressintheoverridingplateofasubductionzone:theAegeanArcfromthePliocenetothePresent.AnnalesTectonicae,1,20‐39.

MÜLLER,B.,ZOBACK,M.L.,FUCHS,K.,MASTIN,L.,GREGERSEN,S.,PAVONI,N.,STEPHANSSON,O.&LJUNGGREN,C.1992.Regionalpatternsoftectonicstress inEurope. JournalofGeophysicalResearch,97,11783‐11803.

MÜLLER,B.,WEHRLE,V.&FUCHS,K.1997.The1997releaseoftheWorld StressMap.WWWaddress: http://www.world‐stress‐

map.org.MÜLLER,B.,REINECKER,J.,HEIDBACH,O.&FUCHS,K.2000.The2000release of the World Stress Map. WWW address:http://www.world‐stress‐map.org.

NOOMEN,R.,SPRINGER,T.A.,AMBROSIUS,B.A.C.etal.1996.Crustaldeformations in theMediterranean area computed from SLRandGPSobservations.JournalofGeodynamics,21,73‐96.

PESKA,P.&ZOBACK,M.D.1995.Compressiveandtensilefailureofinclinedwellboresanddeterminationofinsitustressandrockstrength.JournalofGeophysicalResearch,100,12791‐12811.

PLUMB,R.A.&HICKMAN,S.H.1985.Stress‐inducedboreholeelon‐gation:Acomparisonbetweenthefour‐armdipmeterandtheboreholeteleviewerintheAuburnGeothermalwell.JournalofGeophysicalResearch,90,5513‐5522.

REILINGER,R.E.,MCCLUSKY,S.C.,ORAL,M.B.etal.1997.GlobalPo‐sitioningSystemmeasurementsofpresent‐daycrustalmove‐mentsintheArabia‐Africa‐Eurasiaplatecollisionzone.JournalofGeophysicalResearch,102,9983‐9999.

ROYDEN,L.1993.Thetectonicexpressionofslabpullatcontinen‐talconvergentboundaries.Tectonics,12,303‐325.

SPERNER,B.1996.ComputerprogramsforthekinematicanalysisofbrittledeformationstructuresandtheTertiarytectonicevo‐lutionoftheWesternCarpathians(Slovakia).TübingerGeo‐wis‐senschaftIicheArbeiten,ReiheA,27,1‐120.

SPERNER,B.,RATSCHBACHER,L.&OTT,R.1993.Faultstriaeanalysis:aTurboPascalprogrampackageforgraphicalpresentationandreducedstresstensorcalculation.Computers&Geosciences,19,1361‐1388.

STEIN,R.S.,BARKA,A.A.&DIETRICH,J.H.1997.ProgressivefailureontheNorthAnatolianfaultsince1939byearthquakestresstriggering.GeophysicalJournalInternational,128,594‐604.

TWISS,R.J.&UNRUH,J.R.1998.Analysisoffaultslipinversions:Dotheyconstrainstressorstrainrate?JournalofGeophysicalRe‐search,103,12205‐12222.

WESSEL,P.&SMITH,W.H.F.1991.Freesoftwarehelpsmapanddisplaydata.LOSTransactions,AmericanGeophysicalUnion,72,441.

WESSEL,P.&SMITH,W.H.F.1998.New,improvedversionoftheGenericMapping Tools released.EOSTransactions,AmericanGeophysicalUnion,79,579.

WIPRUT,D., ZOBACK,M.D.,HANSSEN,T.H.&PESKA,P. 1997.Con‐strainingthefullstresstensorfromobservationsofdrilling‐in‐ducedtensile fracturesand leakoff test:Applicationstobore‐holestabilityandsandproductionon theNorwegianmargin.InternationalJournalofRockMechanicsandMiningSciences,34,417.

ZAJAC,B.J.&STOCK,J.M.1997.Usingboreholebreakoutstocon‐strainthecompletestresstensor:ResultsfromtheSiljanDeepDrillingProjectandoffshoreSantaMariaBasin,California.Jour‐nalofGeophysicalResearch,102,10083‐10100.

ZOBACK,M.D.,ZOBACK,M.L.,MOUNT,V.S.etal.1987.NewevidenceonthestateofstressontheSanAndreasfaultsystem.Science,238,1105‐1111.

ZOBACK,M.L.1992.First‐andsecond‐orderpatternsofstressinthelithosphere:TheWorldStressMapProject.JournalofGeo‐physicalResearch,97,11703‐11728.