Processing ground penetrating radar (GPR) data - …

1 GPRP rocessingProcessinggroundpenetratingrada r(GPR) *, , (GPR)profilesprovidedbytheUniversityofCa lgary, ,100 MHz, , ,gainrecovery,spikingdeconvolution,bandp assfiltering, ,50 MHz, 'sDeltaonLakeAthabaska,Saskatchewan, , (CMP) ~ ,theWilliam' , ' , (GPR)hasgainedpopularityasashallowsubsur face, ,engineering,andtheearthsciencessuchas"d elineatingthewatertabledepth,frozen-unfr ozeninterfaces,mappingsoilstratigraphy,s ubsurfacebedrocktopography,peatdeposits, geologicalstructure,andlocatingburiedpip es,cables,"(JolandSmith,1991)andordnance s(Hogan,1988).GPRisatechniquethattransmi tspulsedelectromagneticwaves(10-1000 MHz)whichcanberefractedand/orreflectedof fsubsurfacefeatures,received, , ,assuming electromagneticwavespropagateanalogously toelasticenergy, *UniversityofCalgary,geophysicsundergrad uatestudenttUniversityofCalgary,Departme ntofGeographyCREWESR esearchReportVolume4(1992)11-1 Fisher,Stewart, (1991).SlaveRiver(topleft)andWilliam'sDe lta(topright).

2 DiagrammodifiedfromJolandSmith(1991).11- 2 CREWESR esearchReportVolume4(1992)GPRP rocessingEQUIPMENTT hedataforthisprojectwereprovidedbytheUni versityofCalgary,DepartmentofGeography,w hoownandoperateanadvanced,high-fidelity( orhighsignal-to-noise),digitallyrecordin gpulseEKKOIVGPR acquisitionsystem(manufacturedbySensorsa ndSoftwareInc.,Mississauga,Ontario). (DCoffset/signalsaturationcorrection) ,120tracelinefromSlaveRiverDelta,FortSmi th,NorthwestTerritories(Figure1).Thegeom etryconsistedofasingle-fold,constantlman tennaseparation,withlmshotinterval(orste psize). (1991) #40801200, :SlaveRiverDelta(JolandSmith,1991).CREWE SR esearchReportVolume4(1992)11-3 Fisher,Stewart,andJolTheseconddatasetisa 50 MHzprofile,withCMPgather,fromWilliam'sDe ltaonthesouthshoreofLakeAthabaska,Saskat chewan,showingsteeplydippingforesetbedsi nawatersaturatedenvironment, ,rawdata(Figure4) , , , , (a) ,groundwave,andnearsurfacereflectors, ,slowlydecaying"wow"onthehigherfrequenci esofthesignaltracearrivals, (1989)(Figure5(b)).

3 GainrecoveryDuetogeometricalspreadingoft ransmittedwavefields,laterarrivalsonasig naltraceshownoticeablyloweramplitudestha nearlierarrivals(Figure4).Torecoverrelat iveamplitudeinformation,atime-variant,tr aceequalizationfunctionsuchassphericaldi vergence(equation1)orautomaticgaincontro l(AGC)isapplied(Figure6).Thesphericaldiv ergenceexponentialgainconstant(n)usedher eis50:.,,dtIn1oooobi[j]=i*ai[j]*e(1)Wher eai[J]istheithsampleoftheoriginalamplitu detracea[j],bi[j]istheithsampleofthegain recoveredtraceb[j],anddtisthesamplerate( ).SpikingdeconvolutionSpikingwaveletdeco nvolution(40nsoperatorwindowand1%prewhit ening)vszero-phasedeconvolutionwasfound, bytrialanderror,tobestenhancetheresoluti onofthedatawhenfollowedbyabandpassfilter (Figure7).Predictivedeconvolutionwasalso attemptedbutfailedtoremovetheprimarymult ipleat-850ns(Figure6).11-4 CREWESR esearchReportVolume4(1992)GPRP rocessingRAWDATAISIGNALSATURATIONCORRECT IONSSPIKINGJDECONVOLUTIONBANDPASSFILTER] lVELOCITYANALYSIS]ELEVATIONANDISTATICCOR RECTIONSJlNORMALMOVEOUT] (1992)1I-5 Station#100?

4 00300400500O--rn300--Ill, "_t,-__"'" }[ ,oN,App',EoUIJ,,,,T,300600 FREQUENCY(MHz)(a)(b) (a)DCoffsetandsignalsaturationcorrection s(Moorman,1990).(b)Signalsaturationcorre ctionlow-cutfi] (William'sDelta)and100 MHz(SlaveRiverDelta)GPRdatasets(Figure8) .AccordingtoDavisandAnnan(1989)" "Forthe100 MHzdatarapidattenuationoffrequenciesgrea terthan75 MHzwasobserved(Figure8(a)).Thefrequencyb andschosenforfilteringwere20/30-70/100an d20/30-100 (CMP)gather(Figure10)wasaquiredbyJolandS mith(1991)atWilliam'sDeltaadjacenttowher etheprofilingsurveywasacquired(Figure4) ,theantennaconvergedtowardstheCMP(trace# 39) ,crossedattheCMPandseperateduntilthefina lrecordtrace# , ' ( (Figures11and12)).Knowingthisisusefulfor fightingGPRsystemfibre-opticrecordingpro blemsbyflatteningtheairwavetothetime:ant ennaseparationt= (2)CREWESR esearchReportVolume4(1992)11-7 Station# ,airwaveflattening, #[002003004005000300-- (20/30-70/100).Fisher_Stewart_andJol11 UJa22<O300600O300600 FREQUENCY(MHz)FREQUENCY(MHz)(a)(b) (a)50 MHzWilliam'sDeltaand(b) "_TjF_Tx\\\\AI////\\\\XI////\X\\XI////\X //\\//\\\XI///\\\k\AII////.]

5 ,,,, i///\ (CMP) (1992)GPRP rocessingStation#GMP119395978,.] ' (Tx)_=' (Rx)r-7-----"--IIIINDIRECTGROUNDWAVEX/TR ANSMITTED\/SURFACESIGNAL""/REFLECTED\/SI GNAL\/_j, (iesecondarrivalonFigure6).It'svelocityi sdependentonsurfacelithologiesandisdeter minedby:CREWESR esearchReportVolume4(1992)11-I1 Fisher,Stewart,and,1olV-source-receivero ffset(3)airwavearrivaltime+(airwave-grou ndwavetraveltimes)givingthefirstradarwav evelocitypickonthevelocityspectra(Table1 ).VelocitysemblancepicksfortheFigure10 CMPtraces# (thespeedoflight). (Table1)andJolandSmith's(1991)observatio nthatWilliam'sDeltawasawatersaturatedenv ironment,Table2suggeststhelithologyheret obeawatersaturated, (ns)velocity(x10-3m/ns) # (offsetsgreaterthantrace#19and#59andprio rto200ns) (DavisandAnnan,1989).MATERIALDIELECTRICC ONDUCTIVITYVELOCITYATTENUATIOnCONSTANTmS /mm/ (1992)GPRP rocessingTrace# (m/ns)Offset(m) (1992)11-13 Fisher,Stewart,andJolDielectricconstants Analogoustoacousticimpedanceinseismic,di electricconstants(K)determinethereflecti oncoefficientsforGPRsignalreflections(Da visandAnnan,1989):R=acFffi-_]-_(4)_K-1+_ /K2 Thus,assumingalow-lossgeologicalenvironm ent,thedielectricisrelatedtoelectromagne ticvelocity(V)accordingto:Cv-(5)wherecis thespeedoflight(c= ).

6 Foraporousmediacontainingfluid,theporosi ty(d_)isdeterminedfrom: _=_/-Kfluid+(1-_b)-fK-grain(6)whenq-Kflu idand_K-grainareknownexperimentally, 'sDelta,theearliervelocityanalysisfoundt heaverageradarwavevelocity(V) ,thedielectricconstantisfromequation(5): K=(_)2-{0"30" )=18.(7)Todetermine(_,notefromTable2that Kfluid=80(water),andKgrain-=4(drysand).T hus,assumingWilliam'sDeltaisawatersatura tedenvironment:q'-ffgrainqT8"-,/qq_-q-ff fluid+"]Kgrain-8,/80+ (8) , ,theelevationcorrectionforatopographiclo wis:At-elevation(m)(9) (upwardsforahigh).11-14 CREWESR esearchReportVolume4(1992) (Moorman,1990). 'sDeltacase,twovelocitystaticswererecogn izedbetweenstations#80-120and#315-350(Fi gure6and15). ,thestaticswerewater-tableflattened(inte rpretedasthesecondgroundarrival-Figure15 ) ' (1991)itwaslearntthattheWilliam' ' ,Table2 ,thestaticinfluencedgroundwavearrivedear lierthaninthesurroundings(Figure15).Sugg estingadriertopographichighsincetheradar velocityofdrysandissubstantiallyhigherth anwetsand(Table2).}

7 Therefore, , ,xistheantennaCREWESR esearchReportVolume4(1992)11-15 Station#100200300400500o_, (Killedtrace#252from675-850ns).GPRP rocessingseparation(2m),z1istheelevation ,z2isasubsurfacefeature'sdepthbelowthewa tertable,V1istheradarvelocityabovethewat ertable(< (Table2)foradrysand)andV2istheapproximatevelocitybelowthewaterlevel( ).Station#5075100125 ._IIIIII_l_/# 'r_l/I//ll#ltt/_l/I/gl//l///Hll/l/llt_Jt_d-"50 Station#300325350375f___=.. ' ,themajorvelocitystaticshiftsbetweenstations#80-120and#315-350onFigure7,showthegroundwave(the2ndmajorarrival) "lens"isactuallyareflectionoffthewetsand/drysandinterface(thewatertable).The1stlayervelocity(V1)canthenbedeterminedfromthedifferenceinarrivaltimesbetweenthedirectgroundwavearrivalandthewatertablereflection(Atw).(Zl2x0o1_,tw=2x(2_1)2+_,V1/V1 Arbitrarilytakingz1=2mandfromFigure7themaximumAtw=25ns:CREWESR esearchReportVolume4(1992)11-17 Fisher,Stewart,andJolDRIERSANDX=2rn--(Vl)V1> ,v_AwPx__-j__WA R"_C_IZIv_" )SAND('_)_tISIGNALI_,_, ' +Zl2 Atw__2x+(2m)2(11)25ns= (whichTable2suggestsisadriersand).

8 Thesevelocitystaticcorrectionsshouldbepe rformedbeforenormalmoveout(NMO) (seediagrambelow).Therefore,thebulkstati cshiftforatraceofelevation(Zl)= (Figure14).Theidealvelocitystaticandelev ationcorrectionwillaccomplishthefollowin gbeforeNMO(Figure17).01and02aretheincide ntandrefractedanglesforelectromagneticwa vestravellingatthevelocitiesV1andV2respe ctively, 'sLaw:sin01sin02=(12)VIV2 Xlx2 sin02.(13) +Zl2'-Jx22+z2211-18 CREWESR esearchReportVolume4(1992)GPRP rocessingb2m_IX1X2 TxP2m%,\ \81/DRIERSANDZlV= \IApplicationofStaticand\/Z2I---\/eIElev ationCorrections(Ats)_\_2/2I/WATERSATURA TEDSAND\Iv= \t/STATICINFLUENCEDSIGNALDESIREDOUTPUT(B EFORENMO) +2x2(Figure17)andthusx2=Im-xl:x1v,,,nO,_ V2-sin02-l-x,(z__2-_2(14)_/(1-xl)2+z22wh ichcanbesolvedforx1analyticallygivenV1,V 2,z1, ,01and02canbederivedfromequations12,13, ,thestaticcorrection(Ats)ofthetotaltwo-w aytraveltimestaticrefractedray(ts),tosim ulateawatertablereflectedray(tx)withante nnaseparation(x)priortoNMOcorrectionbase donFigure17is:2z12z2ts-+;(15)Vlcos01V2co s02.)

9 Z2.,2tx=2x+\V2].(16)CREWESR esearchReportVolume4(1992)11-19 Fisher,Stewart,andJolTherefore:Ats=ts+tx =+-VlCOSO1V2cos02_,Zx+Z2(17)gApositive(n egative)(Ats)signifiesanupward(downward) , (Zl)of~2m,thebulkstaticcorrection(Atsf)f orthattracewasabout25nsupwards(Figures14 and15).Using,equations12,13,and14,tosolv eforx1,01and02forthearbitrarydepthsz2(=V 2xt2wheret2isthe2-wayverticaltraveltimet hroughthesecondlayerwithvelocityV2), ,V1= ,V2= ,andz1= (Ats)approximatedusingequation16,choosin gV1= ,V2= ,andz1= ,flatteningthewatertablereflectionisonly aroughestimate(givennoelevationdata)forc orrectingthehighersurfacelayervelocityst aticsseenontheWilliam' , (1992)Station# (0-375ns). , 'sDeltadatawerefrequency(f)-wavenumber(k )migratedFigure18from0to375ns, ,GPRdataprocessingcanbetterthequalityofs ignaltracesandenhancethedam'scorrelatabi lity,continuity, , , ,andAnnan, ,1989, ground -penetratingradarforhigh-res olutionmappingofsoilandrockstratigraphy: GeophysicalProspecting,37, ,G.

10 ,1988, , ,Golden, , , , ,D., :CanadianJournalofEarthSciences,28, , , , , ,1989,PuI_EKKOIV operationsmanual, (1992)