VSP Data Interpretation and Processing PDF

Downloaded 06/29/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.
org/
VSP INTERPRETIVE PROCESSING'
Downloaded 06/29/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.org/
THEORY AND PRACTICE
RONALD C. HINDS
NElL L. ANDERSON
RICHARD D. KUZMISKI
WAVENUMBER (1/m* 10's)

-25,00-18,75-1•50-6,25 0 625 1250 1875 25•00
I I I
0
10
20 "n
30 z
50
60
I'AI
Edited by Bob A. Hardage
Society of Exploration Geophysicists

Post Office Box 702740 ! Tulsa, Oklahoma 74170-2740
Hinds, Ronald C. (Ronald Clifford), 1954•

VSP interpretiveprocessing:theoryand practice
Hinds, Neil L. Anderson, Richard D. Kuzmiski.
p. cm. • (Open file publications:no. 3)
Includesbibliographicreferences.
ISBN 1-56080-042-9(paper)
1. Verticalseismicprofiling.I. Anderson,Neil L. (Neil
Lennart),1954• . II. Kuzmiski,RichardD. (RichardDwight),
1957• .III. Title. IV. Series.
TN269.86.H56 1996
622'.1592•dc21
ISBN 1-56080-030-5 (Series)

ISBN 1-56080-042-9 (Volume)
Societyof ExplorationGeophysicists
P.O. Box 702740
Tulsa, Oklahoma 74170-2740
¸1996 by the Societyof ExplorationGeophysicists

All rightsreserved.This bookor partshereofmay not
be reproducedin any form without permissionin writing from the publisher.
Published 1996.
Secondprinting 1998.
Third printing 2001
Fourth printing 2002
Printed in the United States of America

Table of Contents
Introduction
AcquisitionConsiderations
Introduction
VSP configurations
Recorded wavefields
Far-offsetgeometries 14
IntegratedInterpretiveProcessing 21
Introduction 21
Processingrunstreams 24
Wavefieldseparation:near-offsetdata 26
- Median filtering 27
- Karhunen-Loeve(K-L)filtering 35
- Fourier transform(f-k) filtering 39
- •-p filtering 60
- Futuredirectionsfor •-p filtering in wavefield separation 69
VSP deconvolution and corridor stacks: near-offset data 71
- Deconvolution IPP 71
-Corridor stack IPPs 73
Far-offsetdata processing 76
Rotationsto isolatedowngoingP-waves:far-offsetdata 78
Rotationsto isolateupgoingP-waves:far-offsetdata 80
VSP-CDPand migration:far-offsetdata. 83
VSP deconvolution: far-offset data 86
Problematicfar-offsetinterpretiveprocessing 88
- Timevariant polarizationof Ricinuscasestudydata 88
- Far-offsetdeconvolutionof Ricinuscasestudy data 90
Integratedlog display (ILD) 93
Integratedseismicdisplay (ISD) 93
Integratedinterpretivedisplay(IID) 93
iii
VSP Interpretive Processing:Theory and Practice
Lanaway Field CaseStudy 97
Overview 97
Carbonatereef developmentin the WesternCanadian 97

SedimentaryBasin
Introduction 99
Well nomenclature 102
Lanaway Field 103
VSP data acquisition 105
VSP data processing 107
- P-wave separation 107
- VSP deconvolution 109
- Inside and outside corridor stacks 111
Integratedinterpretation 113
Summary 117
RicinusField CaseStudy 119
Overview 119
Introduction 119
Ricinus full Leduc reef 120
Near-offset(199m) VSP interpretiveprocessing 125
- VSP aleconvolution 127
Far-offset(1100m) VSP interpretiveprocessing 130
- Hodogram-basedrotation 132
- Time-variant model-based rotation 133
- Deconvolution of far-offset data 134
- VSP-CDPmapping 135
Summary 138
iv
Table of Contents
Fort St. JohnGrabenCaseStudy 141
Overview 141
Introduction 141
Geologicoverview 144
- Tectonicand depositionalhistoryof Peace 144
River Embayment
- Lower Carboniferousgeology 146
Original interpretationand well results 148
Near-offset(149 m) VSP interpretiveprocessing 151
Far-offset(FSJG1)VSP interpretiveprocessing 155
- VSP-CDP mapping and migration 159
Far-offset(FSJG2)VSP interpretiveprocessing 161
Summary 166
SimonetteField CaseStudy 167
Overview 167
Introduction 167
Simonette low-relief reef 171
Near-offset(252 m) VSP interpretiveprocessing 173

Far-offsetVSP interpretiveprocessing 179

- VSP-CDPmapping and migration 181
Summary 186
References 187
Appendix: Mathematical review 197
Median filter 197
Karhunen-Loeve(K-L) filtering 198
f-k filtering 200
•-p filtering 201
VSP deconvolution 202
Hodogram-based
singleanglepolarizations 203
Time-variantpolarization 204
vi
Symbols and Abbreviations
VSP Vertical Seismic Profile
CWS Crosswell seismics
RVSP Reversed Vertical Seismic Profile
IPP Interpretiveprocessing
panel
ILD Integratedlog display
ISD Integratedseismicdisplay
IID Integratedinterpretivedisplay
FRT Field recorded time
-TT Time-alignmentof VSPdata alongfirst-breaks

+TT time-alignmentof VSPdata alongupgoingevents
NMO Normal moveout
P Compressional
wave
SV Verticallypolarizedshearwave
SH Horizontallypolarizedshearwave
X Horizontallyorientedgeophonedata
Y Horizontallyorientedgeophonedata
Z Verticallyorientedgeophonedata
HMIN polarizedX and Y dataorientedperpendicularto plane
definedby sourceand well bore
HMAX polarizedX andY dataorientedtangentto planedefinedby
source and well bore
HMAX' polarizedZ and HMAX data orientedin a direction

towards source
! polarizedZ and HMAX orientedperpendicularto the

HMAX' data
time-variantpolarizedZ and HMAX dataorientedalong

upgoingwave arrivals
HMAX" time-variantpolarizedZ and HMAX data orientedperpen-
dicular to Z" data
MBT Model-based transform
vii
K-L Karhunen-Loeve
f-k temporalfrequency-spatialfrequency
•-p time intercept- slowness

VSP-CDP VSP commondepth point transformeddata
SVD Singularvalue decomposition
CDP Commondepthpoint
angularfrequency(2f)
kz spatialfrequencyof depth coordinate
v velocity
Hz cycles/second
offset
time
x2-t2 offsetsquared- time squared

w-p frequency- slowness
covariance matrix
data matrix (N time samples,M traces)

eigenvectormatrix
eigenvaluematrix
K principal componentmatrix
ki(t) row of principalcomponentmatrix
xi(t) row of data matrix
wi individual eigenvectors
individual eigenvalues
v(z,t) input (depth,time) data (f-ktheory)
Fourier transformed data
Dz depth interval
(z,t) data in •:-pdomain
H Hilbert transform
D(z,t) downgoingevents
U(z,0 upgoing events
viii
Introduction
The lowering of a recordingdevice down a boreholefor seismicinvestigation

purposeswas first reportedby F.A.Fessenden
(1917).This work was the basisfor bore-
holeseismicdevelopmentin the late 1920s(Barton,1929).Investigationof horizontallay-
ersand first arrivals(velocityanomalies)in the areaof salt domesfollowedin the 1930s
to 1950s (McCollum and LaRue, 1931; Dix, 1939, 1945, 1946; Navarte, 1946; Gardiner,
1949; Holste, 195.9).
Using the checkshot or velocitysurveys,Levin and Lynn (1958)analyzedthe
recordingsof later arrivalsbeyond the time of the first arrival (primary downgoing
wave). Their work was followedby a majorinvestigationby Gal'perin (1974).The verti-
cal seismicprofile(VSP)techniquesevolvedfrom theseearly seismic/borehole
studies.
Kennettet al. (1980)presentedoneof the earliercomprehensive
discussions
of the
processingand utility of verticalseismicprofiledata.In thispaper,higherfrequencyVSP
o
datawere comparedwith surface-seismic data for the purposesof seismiceventcorrela-

tion. Multiple reflectionidentificationusingthe downgoingwaves,sourcepulsedecon-
volution,and predictionof reflectionsaheadof the bit were alsoexamined.
More recently,Hardage(1985)preparedan excellentcomprehensivesummaryof
the utility of the seismicprofilingtechnique.Thisauthorestablished
thebasicguidelines
for VSP interpretation.The benefitof the VSP in termsof understandingcorresponding
geologiclogsand for providingadditionalseismicinterpretationalinsightwas reviewed
by Stewartand DiSiena(1989).DiSienaet al. (1981,1984)and ToksiSz
and Stewart(1984)
detailedthe utility of recordingthree-component
data in the borehole.They discussed
thepolarization(hodogram)methodof usingtimeinvariantrotationanglesto extractthe
partitionedP-, SH-, and SV-waveenergyfrom the X, Y, and Z channeldata.
In the literature,emphasishas been placedmostly on the developmentof new
acquisitiontoolsand processingmethods;consequently,
thereis a paucityof published
integratedVSP and surface-seismic
interpretations,however,severalexcellentarticles
havebeenwritten. For example,Noble et al. (1988)presentedthe interpretationof three
componentVSPdatafrom the Vulcangasfield and discussed the impactof the interpre-
tation of the VSP data on the developmentof the field. Dupal et al. (1993)reviewed the
useof threecomponentVSP data from a horizontalwell in the Gullfaksfield. Jonesand
Fung(1988)discussed
the useof VSPdatafor theexplorationof a Winnipegosis
reef,and
Stewart(1989)presenteda VSP acquiredfor a Keg River carbonatebuildup play.Dupal
and Miller (1985)usedmultiple offsetVSP data for reef interpretation.Geis et al. (1990)
presentedan interestingexample of the utility of convertedshear wave VSP data.
Coulombeet al. (1991,1992)and Coulombe(1993)illustratedthe processing

and analy-
sis of AVO effectsin VSP interpretation.Recently,Hinds et al. (1989,1993a,b,c,and
1994a,b)prepareda suiteof VSPcasestudies.In theselatterpapers,theprocessing
and
interpretation
of VSPdatahavebeenintegratedto produceseismicimagesthatbetter
illustratethe subsurfacegeology.
Our goalin writing thisbookis to demonstrate the utility of VSPsurveysand
the necessityfor integratingVSPprocessing andinterpretation. With thisobjectivein
mind, we have subdividedthe book into six chapters.In Chapter 1, we describehow
VSPsurveysdiffer from otherboreholeand surface-seismic methods,and definethe
geometriesandnomenclatures usedthroughout thebook.In Chapter2, we describe the
interpretiveprocessing methodology and processing procedures appliedto the VSP
data incorporatedinto the casehistoryChapters3 through6. The basicmathematics
behindthe wavefieldseparation,deconvolution, and far-offsetprocessing of VSPdata
are reviewedin moredetail in the Appendix.Chapters3 through6 are casestudiesof
carbonate reef and sandstone exploration plays from the western Canadian
Sedimentary Basin.Foreachcasestudywe discuss ther•levantgeologyandtheinter-
pretationof theexistingseismiccoverage priorto thedrillingof theVSPwell, thewell
resultsand the rationalebehindrecordingthe VSPdata,theprocessing
and interpretation
of the VSP data,the re-evaluationof the surface-seismic
coveragebasedon the VSP and
associated
well control,and the utility of the respectiveVSPsurvey.
The Lanaway case study (Chapter 3) is based on a situationwhere an
exploratorywell wasdrilledintotheupdipraisedrim of a LeducFormationreef.The
well was expectedto encounteran anomalous late-stageaccretionarybuildupat the
Leduclevel, and it was anticipatedthat the Leducat the VSP well locationwould be
as muchas 80 rn higherthan at adjacentwell sites.To the consternation of the geo-
physicists,
theenvisionedaccretionary growthwasnot present.The top of theLeduc
in the VSP well was consistent with other rim wells and inconsistent with the seismic
interpretation..
Fortunately,
however,
theLeducwasstructurally
closed
andtheVSP
well wascompletedasan oil producerfromboththe Nisku and Leducformations.
Toresolvetheapparentdiscrepancy betweentheinterpretedsurface-seismic data
andgeologyat thewell, a verticalseismic profilewasconducted. Theinterpretationof
theVSPdatawasexpectedto elucidatethe geological originof the misinterpreted seis-
micanomaly. TheVSPsurveywasrelativelysuccessful in thatthesedataconfirmed that
the originalinterpretation
of thesurface-seismic
data(with respectto theNisku,Ireton,
andLeductop)wasincorrect, andthattheanomalyobserved on thesurface-seismicline
was not a processing artifact.The misleadingsurface-seismic imagewas ultimately
attributedto severalsuperposed effects:
anomalousstructuralreliefat thepre-Cretaceous
Introduction
subcrop,stratigraphicanomalieswithin the WinterburnGroup, multiple interference,

and tuning effectsassociated
with the thinningof the Ireton alongthe lengthof seismic
profile.
The secondcasestudy (Chapter4) describesa situationwhere, on the basisof the

crediblemisinterpretation
of conventionalsurface-seismic
data, an exploratorywell was
drilled into an off-reeflocationand ultimatelyabandoned.The prognosisprior to drilling
was that the explorationwell had a reasonablechanceof encounteringgas-bearingcar-
bonatesin the updip, raisedrim of the LeducFormation(DevonianWoodbendGroup)
reef complexat RicinusField. Unfortunately,the well was drilled to the northeastof the
Ricinusreef complexand encounteredonly off-reefshales.
Prior to abandonment,two verticalseismicprofiles(near-offset:199m; far-offset:
1100m) were conductedat the well site.Thesedata were acquiredin an effort to resolve
the apparentdiscrepancybetweenthe interpretedsurface-seismic
data and the geology
at the well and to evaluatethe feasibilityof whipstockingthe well in the directionof the
reef complex.The VSP data were definitiveand allowed a more confidentand geologi-
cally consistentreinterpretationof the surface-seismic
data, and clearly indicatedthat
whipstockingwas not an economicallyviable option.
In Chapter 5, the VSP well was drilled into the deltaic sandstonesof the basal
KiskatinawFormation(StoddardGroup, upper Mississippian).Thesesandstones
were
preferentiallydepositedwithin structurallowsin a regimecharacterized
by faultingand
structuralsubsidenceand form reservoirswhere they are laterally sealedagainstthe
flanksof upthrown fault blocks.Explorationfor basalKiskatinawreservoirsgenerally
entailsthe acquisitionand interpretationof surface-seismic
data prior to drilling. These
data areusedto map the grabensin which thesesandstones
were depositedand the horst
blocksthat actaslateralseals.Subsequentto drilling, verticalseismicprofile surveyscan
be run. Thesedata supplementthe surface-seismic
and well log controlin that:
1) the VSP data can be directly correlatedto surface-seismic
data and, as a result,the
surface-seismic
controlcan be accuratelytied to the subsurfacegeology;
2) multiples,identifiedon VSP data, canbe attenuatedin the surface-seismic
data;and
3) the subsurface,in the vicinity of the borehole,can be more clearly resolvedon the
VSP data than on surface-seismic control.
On the Fort St.JohnGrabendatasetof Chapter5, faultsthat are not well resolved

on surface-seismic
data arebetterdelineatedon VSPdata.The interpretiveprocessing
of
thesedata illustratethe utility of the VSP techniquein the searchfor hydrocarbonsin
structurallycomplexareas.
Chapter6 is basedon a situationwhere, on the basisof conventionalsurface-seis-
mic data, an exploratorywell (13-15-63-25W5M)was drilled into a low-relief Leduc
Formationreef. The well was expectedto intersectreef crestand to encounterabout50-

60 rn of pay. Unfortunately,thewell wasdrilledinto a flankpositionand abandoned.The
decisionto abandonthe well, as opposedto whipstockingin the directionof the reef
crest,was made after the acquisitionand interpretiveprocessingof both near- and far-
offset(252and 524 m, respectively)VSPdata and a re-analysisof existingsurface-seismic
data.
The near- and far-offsetVSPswere recordedand interpretedwhile the drill rig

remained on-site.The immediate objectiveswere to determinean accuratetie between
the surface-seismic
data and the subsurfacegeology,and to map the relief alongthe top
of the reef over a distance of 150 rn from the 13-15 well location in the direction of the
adjacentproductive16-16well with a view to whipstocking.Thesesurveysprovedto be

cost-effective
in that the operatorswere able to determinethat the crestof the reef was
out of the targetarea,and thatwhipstockingwasnot a viablealternative.The useof VSP
surveysin this situationallowedthe operatorsto avoid the costsassociated
with whip-
stocking,and to feel confidentwith their decisionto abandonthe well.
Thesecasestudiesare intendedto providethe readerwith a greaterappreciation
of the utility of VSP surveysand the necessityfor integratingVSPprocessingand inter-
pretation. They alsoprovideinsightinto when, and where,VSPsurveysare useful.
We would like to extendspecialthanksto Prof. Dr. C. E Snymanof the University
of Pretoria,Dr. R.J.Kleywegt of the Geoscience
Councilof SouthAfrica, Mo Crous and
Phil Pelletier of TalismanEnergy Inc., CharlesWelsh of Husky Oil, Tom Andrews of
SkyhawkExploration, Ron Tinline of Integra Geoservices,
Inc., CharlesPretoriusand
Billy Steemkempof AngloAmericanProspecting
Services,LouisDupal of Schlumberger-
GeoQuestLibya, Didier Belaud of SchlumbergerEurope-Africaand GeorgeSmith of
Soekorfor the useof their data, assistance
and teachingsduring the tenureof the work
on thistext.Enlighteningdiscussions
and usefulmaterialfromKarstenK6hlerandJesper
Smit were appreciated.The detailedwork doneby the VolumeEditor Dr. BobHardage,
SEGManagerof Salesand PromotionsJohnvan Gundy,SEGPublicationsManagerTed
Bakamjian,and SeriesEditor Mike Cooperis much appreciated.Their editingand sug-
gestionswere extremelyhelpful.
1- Acquisition Considerations
Introduction
Verticalseismic
profile(VSP)surveysdifferfrom othertypesof boreholeand surface-
seismicmethodsin that they utilize surfacesourcesand boreholereceiversand recordboth
upgoingand downgoingwavefields.Strongandconsistent
sources
areavailablefor useon
or near the surface,and the boreholeprovidesa relativelynoise-freeenvironmentfor VSP
recording.
ReverseVSP andcrosswell
(CWS) surveysdescribedby Hardage (1992)are similar
to VSPsin that theyrecordboth upgoingand downgoingwavefields,but.theydiffer with
respectto sourceand receiverconfigurations.The reverseVSPutilizesdownholesources
and surfacereceivers,and like surface-seismic
recording,is affectedby variable ground
coupling and all the noise inherent in surface-recordeddata. Downhole sources,
althoughimproving,do not yet createa good signal-to-noise
ratio for the reverseVSP
method in comparisonto the VSP (Zimmerman and Chen, 1993). The CWS technique
usesboth boreholesourcesand receiversin adjacentwellbores. The crosswellsurveycan
give a high resolutionseismicpictureof the reservoir(Khalil et al., 1993;Li, 1994;Li and
Stewart,1994)but is not yet a widespread,low-costimaging tool.
Surface-seismic
surveys,with whichwe aremostfamiliar,areconductedusingboth
ß
surfaceor near-surfacesourcesand receiversand recordonly the upgoingwavefields.
VSP configurations
VSP surveys utilize surface (or near-surface)sourcesand borehole receivers.
Dependingon the relativealignmentof the sourceand receiverconfigurations,
VSPscan
be broadly categorizedas either near- or far-offset. A few of the more frequently
employedVSP surveyconfigurationsare illustratedin Figure1.1.
The term near-offset
VSP or zero-offset
VSP (Cassell,1984)refersto a geometrical
configurationwhere the sourceand receiverlocationsare vertically aligned, or can be
treatedas suchduring processing.A near-offsetconfigurationis illustratedin part A of
Figure1.1. Near-offsetVSPsare typicallyusedto tie surface-seismic
data to the subsur-
facegeologyand to differentiateprimary reflectionsfrom surface-generated
and interbed
multiples.
The termfar-offset
VSP or offsetVSP (Cassell,1984)refersto a geometricalconfig-
uration where the sourceand receiverlocationscannotbe treated as vertically aligned
S3 S4
(A) (B)
Fig. 1.1 The field layout of the (zero) near- and far-offset VSP surveys. In (A), the sur-
face source at S• is at a zero-incidence locationwith respect to the geophone borehole
sonde in the vertical non-deviated borehole. The up- and downgoingwaves travel verti-
cally down to the reflectorand back to the receivergeophone sonde in the borehole. The
source at S2 is a non-zero offset locationas the source does not lie directlyover the bore-
hole geophone. In (B) for the deviated borehole,source S3 is in a zero-offsetconfigura-
tion for the upper vertical (shallow depths) part of the borehole and (non-zero) far-offset
for the deviated remainder of the borehole (from Hinds et al., 1989).
during processing.The configurationsusing sourceS2(Figure 1.1A) and sourceS3with

receiversat either B or C (Figure1.lB) illustratefar-offsetgeometry.Our preferreddefi-
nition is that far-offsetgeometryexistswheneverit is necessaryto partition a wavefield
in orderto isolatea wavefieldcomponentontoa singledatachannel(followingan angle-
basedrotation;seeFar-offsetgeometries,this chapter). As illustratedin Figure 1.2, on
far-offsetVSP data, the relativewave amplitudesof eachchannelare a functionof the
incidenceangleat eachgeophone.Far-offsetVSPsare typicallyrecordedto seismically
image the subsurfaceaway from the boreholein more detail, or when conditionspre-
clude running a near-offsetVSP.
Far-offsetgeometriesare commonfor deviatedboreholes. A typical exampleis
the situationat an offshorewellsitewhere an air gun sourceis placednear the drilling
platform and the wellborehasbeendeviatedto intersecta reservoirsomelateraldistance
away. In the casewhere the boreholeis deviatedbut a near-offsetgeometryis desired,
the sourcecanbe movedalongthe surfacedirectlyabovethe receiversonde(Figure1.lB;
sourceat S4and receiverat C). Thisis usuallyan expensiveprocedurebecausethereceiv-
er sonde location must be monitored and the source moved so that the source-receiver
Acquisition Considerations
,•
II
II
Sourcea
II
ACOUSTIC IMPEDANCE 1
P ACOUSTIC
IMPEDANCE
2
ii
REF H
ACOUSTIC IMPEDANCE I
ACOUSTIC
IMPEDANC••.//
• TRANSMI'I-FED
SH
(c)
Fig. 1.2 The up- and downgoingwaves propagatingto the wellbore geophone sonde can
be compressionalP-waves that vibrate in the directionof travel or shear SV- or SH-waves
that vibrate normal to the direction of travel, either in the plane of the source and receiv-
er, or out of the plane. Reflections from dipping and contorted geological strata can
enable all three of these wavefields to be recorded on the X, Y and Z geophones. As
shown in (B), the P- and SV-waves can reflect and transmitas P- or SV-waves at imped-
ance interfaces. The SH incident wave in ( C ) reflects and transmits as an SH-wave
(after Belaud et al., 1988).
geometrymeetsthe near-verticalcriteria. In land-basedVSP surveys,retainingthe zero-

offsetgeometrycanbe difficultbecauseof terrainconditionsand the immobilityof some
sources.
Recorded wavefields
VSP surveysrecordboth upgoingand downgoing

waves (Figures1.2 and 1.3) of
variousorigins. Figure 1.3A illustratesraypathsfor both primary and selectedmultiple
upgoingand downgoingwaves. Figure1.3Bdepictsthe locationof theseVSPeventson
a depth versustraveltimediagram.
Throughoutthe text, the abbreviations
FRT,-TT, and +TT (Hinds et al., 1989)are
usedto referto specificVSPdisplays(Figure1.4). FRT, the abbreviation
for field-recorded
PRIMAR• •?•_
IIA /SURFACE
EOING
Z•
INTERBED
DOWNGOING
MULTIPLE
Z•
(A)
INCREASING DEPTH
Zl Z2
r•
(B)
Fig. 1.3 Examplesof the up- and downgoingraypaths(A) and depth-traveltime diagrams(B)
for bothprimaryand multiplereflections.The upgoingraysare illustratedon the leftside of
(A) whereasthe downgoingraysare illustratedon the rightside. The geometryfor the near-
offsetsourcesare displayedwithexaggeratedsourceoffsetsfor clarity.The surface-gener-
ated downgoingmultipleswillbe recordedat all subsurfacegeophonelocations,whereasthe
interbeddowngoingmultiplegeneratedbetweenlayersZ• and Z2willbe recordedonly when
the geophoneis below layer Z•. Upgoingreflectionsfrom layer Z• will be recordedonly at
geophonelocationsabove layer Z•. The up- and downgoingprimarieswill merge when the
geophonesare locatedat the generatinginterface. A traveltimecurve plot for the up- and
downgoingrays is seen in (B). The downgoingprimaryis the first-breakcurveincreasingin
traveltimefrom left to rightin the diagramand, withthe exceptionof head waves, is the first
recordedsignalon each VSP trace. The downgoingwave multiplefor the near-offsetcase
parallelsthe downgoingprimary.An upgoingprimary(B1) generatesa reflecteddowngoing
multipleat interfaceZ•, whichin turncan generatean interbedupgoingmultipleat interface
Z2 (from Hinds et al., 1989).
(A)
DEPTH INCREASING DEPTH INCREASING
,,q (B) <•

• • 2'TFB
8
Fig. 1.4 The near-offset data correction used to place the data into pseudo-two-way
traveltime (+TT). (A) shows the geometry of the surface-seismicwith the well superim-
posed on the raypaths. To the right, the rays are terminated at the well are different
from the surface-seismic rays by a traveltime equivalent to the zero-offset first-break
times of the downgoing rays. In (B), when the first-break traveltimes are added back
onto the traces, the time axis changes from field-recorded time (FRT) to pseudo-two-
way traveltime (+TT) that should be equivalent to the surface-seismictwo-way travel-
time (from Hinds et al., 1989).
time, is used to describethe time-depth display of the raw VSP field records. FRT
includestime shiftsassociatedwith air gun delaysor Vibroseisphasedelays. The terms
-TT and +TT referto specificdataconfigurations.
The descriptor,-TT,is usedin refer-
enceto displayswhere the first-breaks(first arrivals)and downgoingP-wavesare time
aligned. On thesedisplays,the first-breaktimeshave been subtractedfrom eachtrace,
and the modified traceshave been bulk-shiftedto an arbitrary time-datum (usually to
100 to 200 ms) so that the onset of the first-break wavelet is retained. The term +TT is
usedin referenceto displayswhere the first-breaktime of eachtracehas been added to
that trace(plus possiblenormal moveoutor "NMO" corrections).On +TT displays,the
upgoing eventson near-offsetVSP data are aligned at the same time coordinateand
shouldbe positionedin pseudo-two-waytraveltime(Figure1.4).
Fornear-offset
VSPdata,oneof themoreimportantstepsin theprocessing
andinter-
pretationof VSP data is the identification
and attenuationof the surface-generated
and
interbedmultiples(Figure1.3). The differentiation
of primaryand multipleeventson VSP
datahasbeenreviewedby Wuenschel
(1976),Kennettet al. (1980),Hampsonand Mewhort
(1983),Hinds et al. (1989),and manyothers.An overviewof the methodology
employedin
thistextbookfor the identificationand differentiation
of primaryand multipleVSPeventsis
presented
below.
Forthesurface-generated
multiple,thedowngoing
wavereflects
at a subsurface
inter-
face,travelsback to the surface,and then reflectsfrom the surfaceback down to the borehole
geophone.
Thekeytotheinterpretation
ofthesurface-generated
multipleisthatit mustreside
on all of theVSPtraces.In Chapter2, whentheVSPdataareplacedin the-TT configuration,
the surface-generated
multipleis readilyinterpreted.The simplecorollaryis that the first-
breakcurveshownin Figure1.3istheonlyprimarydowngoingP- waveeventontheVSE We
notethatif themode-converted
P-SVdowngoingwaveis createdat interfaceA in Figure1.3,
thenthemode-converted
downgoingwavewill alsobea primarydowngoingwave. Sincethe
primarydowngoing
P-eventis thefirst-break,
laterarrivingdowngoing
P-events
recorded
at
the sondeare multiples,by definition.Multiplesare moredifficultto interpreton surface-
seismicdatabecause
downgoingwavesarenot recorded
by thesurface-seismic
geophones.
DisplayingVSPdatain +TT and -TT time canassistin the recognition
of multiple
eventson VSP data. As shownin Figures1.3 and 1.5,downgoinginterbedmultiplesare
recordedonly at depthsbelow their uppergenerating
subsurface
interface
and, on near-offset
data,effectively
parallelthefirst-break
arrivals.Theupgoinginterbedmultiples,in contrast,
are recordedonly at sondedepthsabovetheir bottom
generating
interface
and, on near-offset
data,effectivelyparalleltheupgoingprimaryevents.
Thesetwo observations
canbecombinedin a simpleway to assistin theinterpretation
of theinterbedmultiple.In Figure1.5,theupgoingwave(+TT) VSPdatashowthattheupgo-
ing interbedmultiple(U-M) parallelstheupgoingprimary(U-P)fromtheinterface
at Z2but
is delayedby thetwo-waytraveltimebetweentheinterfaces
at Z1andZ2. Figure1.3shows
that the upgoinginterbedmultiplecanbe mistakenfor a surface-generated
multipleif a
downgoingsurface-generated
multiplewascreatedwith thesametimedelayasthetimedif-
ferencebetweentheupgoingprimaryassociated
with Z2andthemultipleseenbelowtheZ2
primary(U-M in Figure1.5). Theinterpretation
is clarifiedby viewingthe(-TT) downgoing
eventsdepictedin Figure1.5.A downgoinginterbedmultiple(D-M) isrecorded
afterthepri-
mary downgoingwave (D-P) and existsonlybelowthe Z1 interface.
The otherobservation
usedin designing
corridorstacks
in Chapter2 isthattheinterbed(oranyother)multipledoes
not intersect the first-break curve.
10
DEPTH DEPTH
Zl Z2 Z, •'2
I I I I
D-P
U-M T,
D-M
Fig. 1.5 Upgoing wave events in (+TT) time (left) and downgoingwaves in (-TT) time
(right) are time-shifted to align the upgoing primary event generated at depth Z2 (the
upgoingprimaryevent being labeled U-P on the left) with the horizontallyaligned event
first-break curve (downgoingprimary labeled D-P on the right). This illustratesthe equal
time separation (Ti) between U-P and U-M (upgoing multiple) events from interface Z2
and the D-P and D-M (downgoingmultiple)events (from Hinds et al., 1989).
The downgoingVSPwavefieldcanbe particularlyhelpful for processingsurface-

seismicdata. More specifically,Hubbard (1979) and Gaiseret al. (1984) showed that a
deconvolutionoperatorthat couldcollapsethe downgoingeventwavetraininto a single
pulsecouldbe usedto attenuatethe upgoingmultipleson surface-seismic
data also(Up
over Down deconvolution). In Figure 1.6 (Hardage, 1985),there is, on near-offsetVSP
data, effectivelya constanttime shift TE between the upgoing primary and its related
upgoingmultiple, and betweenthe downgoingprimary and its relateddowngoingmul-
tiple. The upgoing and downgoing multiples are linked to their respectiveprimary
eventsby thisconstanttime shift. Sincethe upgoingwave is the sameasthe downgoing
wave at an interface,the operatordesignedusingthe downgoingwavetrainwill attenu-
ate the correspondingupgoing multiples. In practice,this procedureof using a single
deconvolutionoperator works well with surface-generatedmultiples, but not so well
with interbedmultiples.
Interbedmultiplesare not recordedon all VSP traces(Figure 1.5). This implies
that independentdeconvolutionoperatorsmustbe calculatedfor eachVSP traceto atten-
uate all interbedupgoingmultiple. Multiples that occurafter the first-breaktime of the
deepestgeophoneare alsodifficult to attenuate(onereasonis that we cannoteven con-
firm whether theseeventsare multiple or primary).
For the interpretiveprocessingof near-offsetVSPdata,suchimportanceis placed
on identifyingmultiples,that three-quartersof the interpretiveprocessingpanelshave
beendesignedto evaluatethe performanceof the Up overDown VSPdeconvolutionpro-
11
INCREASING DEPTH INCREASING DEPTH

Zl Z2 Z3 Z2
I
c•
m w
/
(A) (B)
Fig. 1.6 The constant time shift between the up- and downgoingprimaries and multiples
for (A) surface-generatedand (B)interbed multiples(Hardage, 1985). The layers labeled
Z• in (A) and Z3 in (B) are the multiple generating interfaces for the surface-generated
and interbed multiple,respectively. Part (A) is the depth versus FRT configurationfor the
surface-generated multiple. The numerical operator (inverse filter) that attenuates the
downgoingmultiple(the downgoingevent that is time shiftedfrom the primarydowngo-
ing event by time Tm) will attenuate the upgoing surface-generatedmultiple. In (B), the
operator that will attenuate the upgoing interbed multiple produced between Z2 and a
nearby layer labeled Z3 can be designed from any trace beneath this second, deeper
depth location Z3 (from Hinds et al., 1989).
cedure. As illustratedon the modeledVSP data presentedlater in this chapter,the key

to multiple attenuationis the constant(effectively)time delay,at all sondelocations,of
the upgoingand downgoingmultiplesrelativeto their respectiveprimaries. As the VSP
geometryapproaches
far-offsetconditions,the constanttime lag assumptionis no longer
valid. As a result,the deconvolutionof far-offsetVSP data is generallylesseffectivethan
the deconvolution of near-offset data.
Figure 1.7 depictssurface-generated

multiple (panel 1) and interbed multiple
(panel2) contaminateddata for a four-layermodel. Comparethe modeledresultsto the
schematicVSP eventsshownin Figure1.3B. Below,a shortreview is presentedto show
how an interpreter/processor
pulls apart the differenteventsdisplayedon the modeled
(simulatingreal) VSP data.
Considerpanel I of Figure1.7. The horizontalaxisshowsthe depth of the record-
ing sonde(in m); and the verticalaxisis a FRT time configuration(in s). On everytrace,
the first downgoingwave constitutesthe first-break. As the sondedepth increases,the
traveltimeof the downgoingeventalsoincreases
(thebasicassumptionfor a downgoing
VSP event). The downgoingprimary wave encountersan acousticimpedanceinterface
12
DEPTH (m)
(1) (2)
Fig. 1.7 Synthetic VSP seismograms illustratingsurface-generated (panel 1) and interbed

(panel 2) up- and downgoingmultipleevents. The model used in the ray tracing consists
of four layers with the half space being at the bottomof the borehole. The trace spacing
is 20 m. The geometry consistsof a.near-offset source and a vertical non-deviated bore-
hole. The amplitudes have been equalized, and the first arrival on the 0 m trace has been
muted for clarity. Note the differences in the locationof the up- and downgoing multiple
events and the polarity reversals between the various up- and downgoing wave events
(from Hinds et al., 1994c).
at 120 m depth, and an upgoing primary event is generatedbecauseof reflection. The

upgoingprimary intersectsthe primary downgoingeventon the 120m traceat approx-
imately 0.05 s. The upgoing wave traveltime increaseswith decreasingdepth. This
showsthe familiar relationshipwhere, for the near-offsetVSP,the upgoingand down
going eventshave equal but oppositeapparentvelocities(Hardage, 1985). At 0.09 s on
ß
the surfacetrace,the upgoingevent is turned into a downgoingsurface-generated

mul-
tiple becauseof the reflectionof the upgoingwave at the surface.The downgoingmul-
tiple travelsbackdown to the interfaceat 120m, transmitsdownwardsas a downgoing
multiple and reflectsbackup to the surfaceas the first upgoingsurface-generated
multi-
ple. The other upgoingmultiplescanbe tracedsimilarly.
The surface-generated
downgoingmultiple in Figure1.7 (panel1) is time delayed
from the primary (first-break)downgoingeventby the propagationtime the downgoing
wave tookto reachthe firstinterfaceat 120m depthand travelback(asan upgoingwave)
13
to the surface.As notedabove,the multipleis recordedon all of the VSPtraces.Visually,

it is clearthat the constanttime shiftbetweenthe downgoingprimary and the downgoing
multiplecorresponds
to thesametimedelaybetweenall of theupgoingmultiplesandtheir
respectiveupgoingprimaries.
In panel2, theinterbedupgoingand downgoingmultiplesaregeneratedat inter-
faces2 and 3 (320and 480 rn depth,respectively).As notedpreviously,the downgoing
interbedmultipleis recordedat sondedepthsbelowthe uppergeneratinginterface(at
320 rn depth). The upgoinginterbedmultiple is presenton all tracesabovethe 480 rn
interface(thebottomgeneratinginterface).Thesekey interpretationaspectscanbe diffi-
cult to recognizeif downgoingwavefieldseparationtechniques
havesmearedthe down-
goingmultiple eventsacrossadjacenttraces.
As far asmultiplesareconcerned,
our objectiveis to identifyand attenuatethem
as effectivelyas possible.If attenuationeffortsare not overlyeffective,we needto con-
siderthe implicationsof multiplecontamination
duringour interpretationof the data.
Far-offset geometries
A triaxialgeophone
sondeisusuallyemployedin a far-offset
survey.In specialcases
whererock-properties
areevaluated
usingmode-converted
shearwaves(Figure1.8A;
Ahmed, 1989,1990),or where a near-verticalrocksaltfaceor someothersteeplydipping
reflectorisimaged(Figure1.8B),a triaxialsondemightbeusedin a near-offset
geometry.In
thesecases,the geometryof thetarget,or the rockpropertiesthemselves,
causeorientation
changeswithin the transmittedor reflectedwavefields.At any givensondelocation,three
channelsof dataarerecorded;namely,the Z(FRT),X(FRT),and Y(FRT)data. The X and Y
channelsrecorddatafromthe two orthogonalhorizontalgeophones.
As thesondeis raisedup theborehole,thehorizontalgeophones
rotate.TheX, Y,
andZ datachannels
containvaryingamountsof vertically(SV)andhorizontally(SH)polar-
ized shear(BullenandBolt,1985),andcompressional
P-waveenergy.Oneobjective
in far-
offsetVSPdataprocessing
is to isolatethe downgoingP-waveevents.
This is done usinga two-stepprocedure.The first stepisolatesthe downgoing
P-waveeventsfrom the X and Y channelsonto a singledata panel.Hodogramanalysis
(Hardage,1985;DiSienaet al., 1984)performedon thefirst-break
waveletsof the two hori-
zontaldatasets(fromthesamesondelocation)is usedto polarizetheX and Y dataontotwo
principalaxesthatarenormal(HMIN) andtangential(HMAX) to theplanedefinedby the
sourceand the well (Figure1.9).The resultantHMIN and HMAX datacontainpolarized
SH- and combinedP- and SV-energy,
respectively.
The assumption
for SH- and SV- wave-
14
SOURCE: EAST-WEST
I • CRACK
ORIENTATION
ø T
T SLOW
SHEAR
5t S/: NORTH
45øEAST
(A) -•- FAST SHEAR

(S)
Sll: NORTH
45øWEST
Fig. 1.8 A schematicdiagramof shear wave splittingcausedby the propagationof a SV-

wave throughfracturedrock (fromAhmed, 1989, 1990) is shown in (A). This type of
wave propagationwould complicatethe waveforms recorded at the triaxial geophone
sonde. Part (B) illustratesreflectedrays (from a near-offsetsource) propagatingto the
boreholealongcomplicatedpaths becauseof the steeplydippinggeology. This situation
requiresthat the triaxialdata undergospecial polarizationanalysis in order to interpret
the VSP reflections.
field separation
is that thesetwo wavespropagatealonga raypathsimilarto that of the
downgoingP-wave(Hardage,1985).
The hodogrammethodpolarizesthe horizontalaxisdatausingthe downgoing
P-wave energy in the first-break wavelet. A time window around the first-break is
selected,
andthedatafromthetwo horizontalchannels
areplottedonan orthogonal
axis.
For any giventime samplein thewindow,the corresponding
X and Y valuesareplotted
on the graph to createa hodogram(Hardage,1985;Figure 12.24of Hinds et al., 1989).
Thehodogramusuallyresembles
an ellipsepointingin thedirectionof theazimuthof the
downgoingwave.The anglefoundin thehodogramanalysisis usedin a time-stationary
rotationmatrix equationwhere the input tracesplus the rotationmatrix are usedto cal-
culate
theHMIN andHMAXtraces
(seeAppendix).
Theoutput
dataistheHMAXpanel
(containingSV- and P-waveenergy).For conventional
VSPinterpretationpurposes,the
HMIN data are discarded.The HMAX and Z data (Figure1.9) are assumedto be ori-
entedin the plane definedby the well and the source.
15
KELLY Bt
SOURCE
MAX'
8Z HMAX
HMAX
x,y
HMIN
l
Fig. 1.9 The orthogonalcoordinatesystem of the local x, y, and z geophonesalong with

the coordinateaxis (HMIN, HMAX, Z' and HMAX') that will be used as the principalaxis
in the hodogram (nontime variant) analysis (from Hinds et al., 1989).
The secondstepin the isolationof the downgoingP-waveeventsuseshodogram-

basedpolarizationon the Z and the HMAX data to output the Z' and HMAX' data. The
rotationanglesusedfor the polarizationare found usinga hodogramanalysisof the Z
and HMAX input data. In Chapter 2, we show the inputs and outputsof the various
polarizationsand developan interpretiveprocessing
panel(IPP) to ensuredataprocess-
ing quality for the procedures.
Source-receiver
geometryplaysan importantrole in the polarizations.In Figure
1.10, we show the sonde axes in a deviated borehole and three different source locations.
The amountof wavefieldpartitioningonto the variousgeophoneaxescanvary signifi-

cantly dependingon the source-receiver
geometrywithin the deviatedborehole. Our
objectivewith the two rotations,however,is achievedusingnormalhodogramanalysis.
The HMAX' data panel containspredominatelydowngoingP- and upgoingSV-wave
energy. The Z' axis data containpredominatelydowngoingSV-and upgoing P-wave
energy(Figure1.11).
16
x Y Z geophone tool
reference frame
WAVEMODE CONTRIBUTIONS:
SOURCE S1
DOWNGOING P } _ Z
UPGOING SV
DOWNGOING SV } _ X
UPGOING P
SOURCE S3
DOWNGOING P } _ X
UPGOING SV
DOWNGOING SV } _ Z
UPGOING P
B
Fig. 1.10 For the case of a deviated borehole, the triaxial geophone package receives
contributions of P-, SV-, and SH- waves on all geophones. Even for the near-offset
source location at 8 2 , the Z channel data will contain wave modes other than just P-
waves. To separate the various wavefield contributions, the data on the three geophones
are numerically rotated using hodogram polarization analysis. This is the "theoretical"
geometry used in the analysis of the upgoing rays for the deviated borehole. Note that
the upgoing rays travel very different directions when the subsurface layers are not hori-
zontal as in Figure 1.1 (from Hinds et aI., 1989).
If the upgoing and downgoing wave particle motion were exactly orthogonal, this
stage of our processing would be complete. However, as is discussed in more detail in
Chapter 2, the HMAX' data contain residual upgoing P-wave energy. Generally, the angle
of the upcoming P-wave originating at an interface just below the sonde is larger than the
angle of an upgoing P-wave originating at a deeper interface. Indeed, the deeper reflections
may approximate near-vertical incidence angle reflections (as in the near-offset VSP case).
A time-variant polarization of the upgoing events on a given set of Z and HMAX
traces can be performed by windowing the upgoing events and performing a hodogram
analysis of each event. However, because of the low amplitudes involved in the upgoing
events, a model-based system is used to evaluate the time-variant polarization angles
(time and angle pairs).
17
VSP Interpretive Processing: Theory and Practice
SOURCE
HMAX'
Y-GEOPHONE
X-GEOPHONE
Z'
HODOGRAM-BASED POLARIZATION:
HMAX' - ENHANCES "FIRST DOWNGOING" P
AND UPGOING SV-WAVES
Z' - ENHANCESUPGOING P AND
DOWNGOING SV-WAVES
Fig. 1.11 The geometry for stationary rotations and introduction to time-variant rotations.
To use the classical stationary rotations, it is assumed that the HMAX' data contain down-
going P- and upgoing SV-wave events only, and that the Z' data contain upgoing P- and
downgoing SV-wave events only. For the Z', HMAX' rotation, the polarization analysis is
performed on the downgoing primary wavelet (a single angle is chosen at each VSP
depth trace). The time-variant analysis consists of calculating the changing incidence
angles of the upcoming P-wave (or SV) with increasing traveltimes at a single geophone
location. The angle for a single layer is shown in the diagram, whereas an entire earth
model is input into the calculations, which necessitates the time-variant calculation
because the upgoing reflection angles change with each new layer used. For the single
VSP trace, the angles are interpolated between the calculated model-based (time, angle)
pairs to give a (time, angle) pair for each time sample following the first break for that
trace (from Hinds et aI., 1989).
For far-offset data, a model of the interfaces is constructed using the near-offset first-
break times, and ray-tracing is performed to given sonde model locations. The reflections
from the various interfaces at a given geophone location will produce parameter pairs (time,
incidence angle) from the modeled upgoing raypaths. These time-angle pairs are interpo-
lated to provide a time-variant rotation angle for every time sample for our recorded trace
at the same depth as the modeled geophone location. A time-variant polarization of the
upgoing wave data from the Z and HMAX panels yields the Z' and HMAX' data panels
illustrated in Chapter 2. The Z' data are then interpreted and correlated with the surface-
18
seismic. The interpretive processing described in the following chapters dictates the
quality control decisions used to evaluate every step of the processing.
For far-offset data, the downgoing waves separated from the HMAX' panel are
used to deconvolve the final Z" data. A simplistic time-delay relationship between the
upgoing and downgoing multiples is assumed; however, some caution must be used in
applying this assumption to far-offset data. Figures 1.12 and 1.13 show the modeled
results from a 980-:-m deep borehole and the Z-axis data for 0, 200, 400, 600, and 800-m
source offsets. At far-offsets, both the upgoing and downgoing surface-generated and
interbed multiple and primary events lose the simple time-delay relationship that exists
between the primaries and multiples of the near-offset VSP data. In Chapters 2 and 6, a
data example is presented for a situation where a 524-m offset VSP was deconvolved
with satisfactory results.
0.2
--l
0.4 ~
m
~
~
0.6 m
:§:
0.8
=~===="'"'- 1 .0
(1) (2) (3) (4)
Fig. 1.12 Synthetic VSP seismograms for the up- and downgoing primary and surface-
generated multiple events for offsets of 0, 200, 400, 600, and 800 m (panels 1-5,
respectively). The panels illustrate that the simplistic near-offset primary-multiples time
delays become complicated and invalid with increasing offset. The simple time delays
between primary and multiple events (for far-offset geometries) assume that the up-
and downgoing propagation paths are vertical. This relationship breaks down for far-
offset geometries.
19
(1) (2) (3) (4) (5)
Fig. 1.13 Synthetic VSP seismograms for the up- and downgoing primary and interbed
generated multiple events for offsets of 0, 200, 400, 600, and 800 m (panels 1-5, respec-
tively).
20
2 - Integrated Interpretive Processing
Introduction
The term interpretive VSP processing is used to refer to a generalized iterative pro-
cessing methodology where the interpretations of the output of a processing step and the
input data are used to modify preceding steps and/ or to constrain subsequent process-
ing parameters. Early interpretive processing methodologies focused largely on using
VSP results to guide surface-seismic processing (Stone, 1981; Hardage, 1985) and on the
determination of specific processing variables such as reflector dip (Noponen, 1988).
More recently, VSP and intermediate surface-seismic outputs have been interactively
analysed so that the interpreter can continuously monitor the effectiveness of processing
steps such as multiple attenuation. One example of such an application is described by
Nsess (1989) in a paper on model-based transform (MBT) processing. A second example,
using marine seismic data, is presented by Hinds and Durrheim (1993). These authors
used a Karhunen-Loeve (K-L) based multiple attenuation scheme (Jones, 1985; Jones and
Levy, 1987) to create an output seismic section. The latter was used as input to MBT pro-
cessing. In the vicinity of the well site, VSP results could be used to further constrain the
MBT processing by supplying a refined definition of primary reflections.
Interpretive processing of VSP data involves the continuous monitoring and
interpretation of the data during processing to constrain the various processing stages.
Flowchart 1 illustrates interpretive processing for median-filter-based wavefield separa-
tion. Median filter wavefield separation processing involves the amplitude balancing of
raw data, static shifting, median filtering to isolate the downgoing events, amplitude bal-
ancing of the separated downgoing event data, the subtraction of the data containing
upgoing and downgoing events from the panel containing only downgoing events, stat-
ic shifting into pseudo-two-way traveltime, and upgoing event enhancement.
An interpretive processing panel (IPP) is constructed to examine the input and
output of the various processing steps used to isolate the upgoing events on the final dis-
play, Zup(rned)(+TTl. Here, the subscript denotes the type of events within the data (plus
filtering), and the term in the bracket refers to the data time-configuration.The upgoing
events in Zup(rned)( +TTl are interpreted; however, the interpretation is also focused on the
previous outputs along the processing route, such as the Zup(+TTl, Zup(-TTl, Zdown(-TTl,
Z(-TTl, Zgain(FRTl, and Z(FRTl data. If any of the processing steps, from the raw data
Z(FRTl to the median-filtered upgoing event data in pseudo-two-way traveltime
Zup(rnedl(+TTl, have introduced artifacts that adversely affect the interpretation, then the
processing steps are modified to eliminate the artifacts. This procedure of generating the
21
- Process data; Z(FRT). z,.,"(FRT). Z(-TT) ,

Z._(-TT), z,,(-TT). Z",(+TT). z"...,,(+TT)
- incorporate results into wavefleld separation IPP

- Interpret upgoing events on on all panels

- Interpret downgoing multiples
- Identify noise or undesired VSP events
DESIGN PROCESSING PARAMETERS

TO CORRECT FOR INTERPRETATION
- are upgoing events unaffected by processing artifacts? NO

- are downgoing events interpretable?
- have noise and undesired events been satisfactorily
attenuated ?
- have downgoing events been isolated without retaining
unacceptable amounts of residual upgoing events
YES
Finished IPP
Flowchart 1: An example of an interpretive processing decision-making flowchart for

median filter. plus subtraction wavefield separation, processing (from Hinds 1994c).
IPP and then determining the effect of the processing on the data is repeated until the
interpretation is clear of processing artifacts. Using the example of wavefield separation,
interpretive processing continues by utilizing the Zup(med)l+TTl, Zupl+ TTl, and
Zdownl-TT) data in the deconvolution processing and within the deconvolution IPP.
The utility of interpretive VSP processing is greatest where a single step within a
processing sequence has introduced processing artifacts that are detrimental to the inter-
pretation. One such example is the situation where Rieber mixing during frequency-
wavenumber (f-k ) wavefield separation causes an upgoing event to extend laterally
beyond its corresponding downgoing primary (Hardage, 1985). In this scenario, the
interpretive processing of the VSP output data allows us to determine that a problem has
occurred. Back-tracking the problem to the processing step of origin (such as the f-k
22
Integrated Interpretive Processing
domain filtering parameters in the case of Rieber mixing) allows the problem to be cor-
rected, and ultimately allows for more robust processing and introduces confidence into
the interpretation. Interpretive VSP processing is directed by the continued interpreta-
tion of the processing output.

To facilitate the interpretive processing of the VSP case studies incorporated into
this text, we have designed a standardized suite of processing output panels. These pan-
els allow interpretive analysis of the output from sequential processing steps. Modern-
day processing systems facilitate interpretive processing even further by enabling the
interpreter to create, display, and compare output panels interactively. Hard copies are
not necessarily required, and interpretations can be made without leaving the processing
workstation.
This chapter is focused on the processing of near- and far-offset VSP data. The
interpretive processing methodology is illustrated through the use of IPPs and interac-
tive screen processing. Where alternative processing routes can be used, as in the case of
wavefield separation, the various processing options are reviewed. The results of using
the various processing techniques are evaluated through the use of interpretive process-
ing panels. Where a single method can be used in more than one way to perform a sin-
gle task, such as t-p wavefield separation processing, the different processing routes are
explored. The advantages and disadvantages of using the various routes are evaluated
in terms of the effect on interpretation.
One of the far-offset VSP processing studies (Ricinus data; Chapter 4) presented
in this chapter illustrates a situation where routine processing produced contaminated
results (mode-converted downgoing and upgoing SV-events). These events were ini-
tially identified on the Z'up(FRT) and HMAX'up(FRT) data. To rectify the problem, we
filtered these Z(FRT) and HMAX(FRT) data to attenuate the mode-converted events at
that point in the processing procedure. The data were then successfully processed using
wavefield separation and time-variant polarization to output more interpretable results.
The success of obtaining interpretable data following far-offset VSP deconvolution pro-
cessing is shown to be data dependent. In this chapter (see VSP deconvolution: far-off-
set data), the far-offset VSP data of the Simonette (Chapter 6) and Ricinus carbonate reef
case studies are deconvolved to show the positive and negative effects of far-offset VSP
deconvolution.
The use of the median,f-k (Hardage, 1985), K-L (Freire and Ulrych, 1988), and t-p
(Devaney and Oristaglio, 1986; Moon et al., 1986; Hu and McMechan, 1987) filters for
wavefield separation processing are demonstrated in this chapter. We show how the var-
ious methods are combined to perform individual tasks within another method's wave-
field separation IPP. In one example, surgical f-k muting is used to attenuate the aliased
23
tube wave events that contaminate the Zup(+TTl data. These data were originally wave-
field separated using the median filter wavefield separation technique. Surgical f-k muting
is used to enhance the upgoing P-wave event.
Interactive screen processing is shown to assist interpretive processing in the eval-

uation of event identification and the effect of processing parameter selection on event iso-
lation. This type of processing is considered to be a part of interpretive processing.
In the latter sections of this chapter, the completely processed VSP data are used
in conjunction with other exploration data in the integrated log display (ILD), integrated
seismic display (ISO), and the integrated interpretive display (IID). This allows an addi-
tional interpretive check on the processing since the geological log and surface-seismic
data can be compared to the VSP results. These displays are used within integrated geo-
logical/ geophysical interpretation.
Processing runstreams
In the following sections, the main components of VSP processing are described
and illustrated. Topics include: 1) the separation of upgoing and downgoing wavefields
on near-offset data; 2) deconvolution of near-offset VSP data (Hubbard, 1979); 3) the iso-
lation of downgoing P-waves on far-offset data; 4) the isolation of upgoing P-waves on
far-offset data; 5) VSP-CDP transformations (Dillon and Thomson, 1984) and migration
of far-offset data; 6) deconvolution of far-offset data; 7) problematic far-offset interpretive
processing; and 8) integrated log, seismic and interpretive displays. These topics are pre-
sented to familiarize the reader with the processing and interpretational steps referred to
in the case studies presented in Chapters 3 through 6. The related mathematics are pre-
sented in the Appendix.
To place these selected topics within the context of the overall processing flow, we
present below two example processing runstreams (Hinds et al., 1989). In the first exam-
ple, a median filter wavefield separation technique is applied to near-offset VSP data
(Flowchart 1). In the second example, two hodogram-based trace data rotations, a time-
variant rotation, NMO correction, VSP-CDP transformation and/or migration are
applied to the far-offset data. These runstrearns, and those presented elsewhere in VSP
literature, should be considered as guidelines because actual runstreams are data depen-
dent. An example of near-offset processing flow using the median filter wavefield sepa-
ration could include:
1) Geometry design and completion of trace headers;
2) Editing of the unstacked Z(FRTl data;
3) Compensation for source timing errors (as an example, the picking of near-source
airgun first-breaks and subsequent static shifting of the corresponding VSP data);
24
4) Common sonde location gathering and summation;

5) First-break picking of first-break compensated Z(FRT) data;
6) Editing of VSP stacked data;
7) Application of a spherical divergence correction for downgoing waves using a

velocity model defined by the VSP first-break times;
8) Static shifting of Z(FRT) data to output Z(-TT) data;
9) Application of a median filter to the Z(-TT) data to isolate downgoing waves and
output Zdown(-TT) data;
10) Scaling of Zdown(-TT) to match the Z(-TT) data;
11) Subtraction of Zdown(-TT) data from Z(-TT) data to output Zup(-TT) data;
12) Multiwindow frequency analysis to design and perform time-variant filtering on
Zup(-TT) data;
13) Designing deconvolution operator(s) from Zdown(-TT) data and application to
z.,« TT) data to output Zup(de<on)(-TT) data;
14) Statically shifting Zup(-TT) and Zup(de<on)(-TT) data to output Zup(FRT) and
Zup(decon)(FRT) data, respectively;
15) Application of spherical divergence correction for upgoing waves (model-based) to
Zup(FRT) and Zup(decon)(FRT) data;
16) Application of normal moveout correction (and simultaneously static shift) to out-
put Zup(+TT) and Zup(decon)(+TT) data;
17) Application of median and band-pass filter to enhance upgoing events on Zup(+TT)
and Zup(decon)(+TT) data;
18) Muting of upgoing wavefield data for use in the outside and inside corridor stack
IPPs of Zup(+TT) and Zup(de<on)(+TT) data.
The wavefield separation route and analysis of multiple contamination are data
dependent. The far-offset reference runstream assumes that the X(FRT), Y(FRT), and
Z(FRT) channel data have been edited, common-receiver-location gathered, stacked, and
first-break picked. An example runstream could include:
1) Hodogram analysis in a window centered around the first-break of the X(FRT) and
Y(FRT) data;
2) Rotation of the X(FRT) and Y(FRT) data to output the HMIN(FRT) and
HMAX(FRT) data;
3) Hodogram analysis about a window centered around the first-breaks of the
HMAX(FRT) and Z(FRT) data;
4) Rotation of the HMAX(FRT) and Z(FRT) data to output the Z'(FRT) and
HMAX'(FRT) data;
5) Application of the spherical divergence correction for downgoing waves to the
Z'(FRT) and HMAX'(FRT) data;
25
6) Statically shifting the HMAX'(FRT) to output HMAX'(-TT) data;

7) Wave field separation using the median filter to output HMAX'down(-TT) data;
8) Scaling of the HMAX'down(-TT) data using a factor calculated from a window
around the aligned first-break wavelets in the HMAX'(-TT) data;

9) Subtraction of scaled HMAX'down(-TT) data from HMAX'(-TT) data to output the
HMAX'np(-TT) data;
10) Wavefield separation of the Z'(FRT) data usingf-k masking to output the Z'up(FRT)
data;
11) Application of inverse operation of step (4) on Z'up(FRT) and HMAX'up(FRT) data
to output Zup(derot)(FRT) and HMAXup(derot)(FRT) data;
12) Calculation of the time-variant polarization (time, rotation angle) for every
depth trace using ray-tracing of upgoing waves (from a model created using the
first-break times and the interval velocities of the near-offset VSP data);
13) Application of the time-variant model-based rotations to the Zup(derot)(FRT) and
HMAXup(derot)(FRT) data to output the Z"up(FRT) and HMAX" up(FRT) data;
14) Application of a spherical divergence correction to the Z"up(FRT) data using mod-
eled upgoing traveltimes;
15) Application of model-based NMO correction to Z"up(FRT) data;
16) Time-variant band-pass filtering of Z"up(FRT) data;
17) Statically shifting Z"up(FRT) data to output Z"up(+TT) data;
18) Application of median and band-pass filters to Z" up(+TT) data;
19) Application of VSP-CDP transform or VSP migration to the Z"up(+TT) data
to output the Z"up(+TT) data in the (+TT) versus offset distance domain;
20) Application of a deconvolution operator calculated from the HMAX'down(-TT)
data to the Z" up(FRT) data to output to Z"up(dewn)(FRT) data;
21) Replication of steps (14) to (19) to the Z"up(deconj(FRT) data to output the VSP-CDP
or migrated Z" uptdecon) data in the (+TT) versus offset distance domain.
Wavefield separation: near-offset data

In this section, we demonstrate wavefield separation using median, K-LJ-k and t-p
filtering. For each method, an IPP is designed to reflect the processing steps of the wavefield
separation process. The methods and the associated IPPs are described with interpretation
in mind. The example data for each method were selected to highlight potential interpreta-
tional problems.
The median filtering method is used to isolate the downgoing events and transform
the Z(-TT) data into Zdown(-TT) data. The Zdown(-TT) data are amplitude-balanced to match
the Z(-TT) data and then subtracted from the Z(-TT) data to output the Zup(-TT) data.
26
The K-L filtering similarly isolates the Zdown(-TTl data from the Z(-TTl data. Fil-
tering transforms the Z(-TTl into its principal components with the first (and largest)
principal components representing the time-aligned downgoing events. The data are
inverse transformed using only the first few principal components. Subtraction of the
Zdown(-TT) data from the Z(-TTl data yields the separated upgoing events.
The f-k method transforms Z(FRTl data into the f-k domain. In this domain, the
upgoing events reside in the negative wavenumber quadrant and the downgoing events
in the positive quadrant. The downgoing events are muted from the f-k domain using
quadrant or surgical j-k muting. The muted f-k data are then inverse transformed to out-
put the isolated upgoing events, Zup(FRTl. In the f-k filtering section, f-k wavefield sepa-
ration IPP and interactive f-k screen processing are reviewed. Problems encountered in
f-k filtering because of Rieber mixing (Hardage, 1985), spatial aliasing, and events cross-
ing within the f-k domain are addressed.
The t-p-filtering-based wavefield separation method relies on the different slow-
nesses of the upgoing and downgoing events to perform the separation (Hu and
McMechan, 1987; Hardage, 1992). In this section, the t-p filter is used to perform velocity fil-
tering and spatial interpolation (Hu and McMechan, 1987). Spatial interpolation is used in
the separation step and as a pre-processing step for other wavefield separation methods.
One way to appreciate the case study examples is to recognize the types of prob-
lems (such as random noise, tube wave and SV-event contamination) present within each
dataset, and then to look for similarities between your industry data and our example
VSP data. A second way is to compare the Chapter 2 data to the data interpreted in the
case histories. The ways in which the various filtering methods discussed in Chapter 2
have affected the interpretability of VSP data are shown in Chapters 3 to 6. Interpreta-
tions are not presented in depth in this chapter; rather our focus is on the development
of interpretive processing (IPP) and associated displays.
Median filtering
In this section, median filtering (Gallagher and Wise, 1981; Nodes and Gallagher,
1982; Arce and McLoughlin, 1984; Fitch et al., 1984; Stewart, 1985; Arce et al., 1986) is used
for wavefield separation (Hardage, 1985; Kommedal and Tjostheim, 1989). The proce-
dure is shown as Flowchart 2. The median filter isolates the downgoing events from the
Z(-TT) data into the downgoing event data displayed in the Zdown(-TTl panel. The iso-
lated downgoing events are scaled and subtracted from the upgoing and downgoing
events of the Z(-TTl data, resulting in the upgoing event data displayed in the Zup(-TTl
data panel. Following static shifting and further median filtering, the Zup(med)(+TTl and
the other data panels are interpreted.
27
Z(FRD Normalize on first-break
Z•••[FRD •
Pick first-breaks
SIalic shill data to (-TT) OR
Z(-TD
Z.~(-TT)
+ Normalize ~(-TT) first-break wavelet
e using Z(-TT) dala.

Subtract normalized Zdown(-TT) from Z(-TT}
~
to outpul Z..(-TT) data.
I z,,(-TT)
I
I •
Z..(+TT)
I
StaUc shill dala 10 (+TT)
Determine cause of
I •
~_(+TT)
I
apply median filter
artifacts and redesign
parameters
! YES
Interpret up- and downgoing events on the data displays.

Has the choice of median filter length or first-breaks
alignment created processing artifacts?
1 NO
Use Z.,.,,(-TT). z,,(+TT), and ~...)[+TT) in the

deconvolution processing and IPP design.
Use :z..p(lnIiId)(+TT) in integrated interpretive display (110)
Flowchart 2: An example of an interpretive processing flowchart for median filter, plus

subtraction wavefield separation, processing (from Hinds 1994c).
The effect of each processing step on the interpretation of the upgoing events must be
analyzed. When using the median filter for wavefield separation, the two concerns are first-
break picking and the length of the median filters. If improper first-break picks or unsuitable
median filter lengths have created processing artifacts, then the first-break times need to be
repicked or the lengths of the applied median filters need to be modified. The processing is
then repeated. The use of the median filter for wavefield separation of multiple and tube wave
contaminated data is discussed below.
The input to the median filter is a selected window of the data (data points from
a series of depth traces at a common time or within a data square). The length of the win-
dow can be an even or an odd number of points (2N or 2N+1). Each of the two end-points
28
of the time series (shallowest and deepest traces) are padded with N additional points to
accommodate the center location of the window. The values within the input window are
sorted according to magnitude, starting with the smallest and progressively increasing to
the largest. The center value of the sort is the median value. For the odd-numbered-point
filter, the median value at the center of the sorted windowed time series becomes the new
value of the output series. When N is even, the mean of the two middle median values is
the output of the filter. This new point of the output data is placed at the location (time
and depth trace) of the center of the window. A new output time series is generated, one
point at a time, as the window slides across and down the 2-D time series. Median filter-
ing equations are discussed in the Appendix.
In Figure 2.1, the median-filter-based wavefield separation IPP for the multiple-
contaminated near-offset data of the Fort St. John Graben case history (Chapter 5) is pre-
sented. The normalized (on the first-break wavelets) Z(FRT) data are shown in panel 1 of
Figure 2.1. Note that the downgoing waves are orders of magnitudes greater than the
upgoing waves, and that the strong repetitious wavetrains follow the first-breaks in
panel 1.
The Z(FRT) data in panel 2 have been gained (a form of PGC; Yilmaz, 1987,page 59)
and show the inter-relationship between the upgoing and downgoing multiples. We want
to track the multiple generated at an interface at about 690 to 710 m (Spirit River Formation).
The upgoing event (colored blue) intersects the primary downgoing "curve" at about 0.115
s FRT traveltime between the 690 and 710 m traces. The upgoing event can be traced
(increasing in traveltime with decreasing depth) back to the shallowest trace (at approxi-
mately 0.24-0.25 s); however, note that this event intersects the surface-generated downgo-
ing multiple at 0.19 s on the 470 m trace. The downgoing surface-generated multiple (col-
ored yellow) is opposite polarity to the primary downgoing wave (the same polarity as the
upgoing primary). The intersection of the two waves is noted only to guide the reader; there
is no physical significance to the intersection. Follow the highlighted portion of the down-
going multiple deeper in time to 0.22 s on the 690-710 m trace. The downgoing multiple
reflects at the 690 m interface and creates an upgoing multiple (a peak). The upgoing multi-
ple (colored purple) exists only on traces shallower than 690 m (this is shown more clearly
in the absence of downgoing events as in panelS).
The lateral extent of the multiple event should be preserved after the wavefield
separation processing. For the multiple event discussed above, the upgoing multiple
event must not extend to traces deeper than 690 m.
The time delay between the downgoing primary event (peak) and the first strong
surface-generated multiple (trough colored yellow) on panel 2 is the same time delay as
between the Spirit River event (trough at 0.28 s on colored blue) and the upgoing multi-
29
..........,.
ot t
...•••
llCL
• LS
cu.
OUI
...........,
- Di l l
DL£I
Olt .
OU.
...,
OU I
..II
...
QU.
•••
• L~
~
D,.
llCL
oco
DU
0<0,
...,
llClI
OI ~1
...,
ou r
....
llC~'
Oltl
DLL'
~,E3~~~~1::
=JE::q=J= F= C
....(:) !:l ~ 0 p ~m
CXl C) ~ '" O"C
-l
(S) 3~I.l13AVl:U ::I:
-
3'
p p 0
(:) Cl:l C) A
( s ) 3~I.l13"~.l
"3
.....
30
pIe (a peak at 0.4 s colored purple). This time delay is a fundamental assumption in VSP
deconvolution (Hardage, 1985).
A second upgoing primary event of interest (highlighted in orange) is generated
at about 1250 m depth (Nordegg Formation). An associated set of upgoing multiples are
generated by this primary (similar to the multiples from the 690-m interface). The up-
going multiple from the Nordegg Formation (resulting from the surface-generated
downgoing multiple) is highlighted in purple (at approximately 0.725 s on panels 6 and
7). On panel 3, the Z(FRT) data are statically shifted to Z(-TT). The first-break is aligned
at 100 ms so as not to lose any part of the first-break wavelet. If we had picked the first-
break as the dominant trough of the first-break (following the strong peak) and shifted
that picked first-break to the time origin, the peak would be lost to further processing.
In panel 4, an ll-point median filter isolates the time-aligned downgoing waves.
The median-filter-wavefield separation technique is really a three-part exercise. In (-TT)
time, the method consists of:
1) separating the downgoing waves Zdown(-TT) from the combined wavefield data
Z(-TT);
2) balancing the amplitudes of the separated downgoing waves Zdown(-TTl with the
combined wavefield data Z(-TT); and
3) subtracting the amplitude-balanced downgoing waves Zdown(-m from the combined
wavefield data Z(-m to obtain the upgoing events Zup(-m.
The combined wavefield data, downgoing events, and separated upgoing events
(-Tf'I are shown in panels 3, 4, and 5, respectively. The data are displayed as functions of
(+ TT) coordinates by first-break time shifting (Figure 1.4). Now we see a second multiple
problem that alerts our interpretive processing. Superimposed on the upgoing multiples
(starting from 0.6 - 0.8 s), we have an upgoing primary that intersects the first-break curve
in panels 2, 5, and 6 on the 1470-1490 m traces at 0.7 s. This primary (highlighted in green on
panel 7 of Figure 2.1) must be preserved during deconvolution.
The conclusions reached using interpretive processing are that the median-
filter-wavefield separation method has preserved both the depth starting point of
the upgoing multiples, and the distinction between the primary from 1470 m and the
multiples (in which that primary is superimposed). The Zup(FRT) data in panel 6 are
enhanced using a short-length median filter and redisplayed in panel 7. This pro-
cessing has revealed a classic multiple example that will now be used for other
wavefield separation methods.
An example of the median-filter-based wavefield separation IPP for tube wave
contaminated data is shown in Figure 2.2. [For a thorough description of the tube wave
and its appearance on VSP data, refer to Hardage (1981, 1985)]. The shallowest trace was
31
-0
-
.!::
0
Q)
E
c
0
+=
o
TRAVELTIME ( s )
'0;1" <0 CIO
N
TRAVELTIME (s)
'0;1" <D CIO
-
.0
~
::J
CJ)
o 0 0 o c:i c:i c:i

~ y
"0
c
C13
>-
Q)
== r-..:
;;::::
T"""
@C\J
:a
Q)
~
::J
E 0>
Q)U::::
£ o~
0>"0
C
0- >. Q)
~ C13
C13Q..
_rJ)
~:a
"O.~
Q) -
(iioo
oc
- -
Q)
Ec
.$[
c-
o 0>
U c
Q)"S
~ E
~
~
Q)
-got:»
_ ::J
- >-
o rJ)
0.. Q)
a.. £
cO
go
~ a..
[o~
Q) C13
rJ) E
-0 0
0)"0
;;::::..lc
Q)"':'"
> Q)
~F
C\j
C\I
0>
u::::
32
recorded at 410 m depth. By following the tube wave event from deeper depths to the 410
m trace, we can extrapolate this noise pattern to the surface. Since its origin is at (or near)
the surface, we can assume that the tube wave was created by the Rayleigh wave which
transferred energy into the fluid column at the wellhead and caused vertical motion
within the borehole fluid column. This would be categorized as Hardage's tube wave
Mode 2. The question is whether we need to, or can, attenuate the tube wave. On panel
2, the gained data reveal that the tube wave is spatially aliased (Hardage, 1985; Yilmaz,
1987). The depth increment of 30 m was not adequate to sample the low-velocity tube
wave. The downgoing tube wave events are highlighted in yellow on panels 1, 2, and 3.
The Zdown(-TT) data in panel 4 show that the downgoing tube wave has not been
effectively subtracted from the combined wavefield data, Z(-TT). This is because some of
the aliased tube wave is not identified as downgoing energy. The gained, separated,
upgoing Zup(-TT) data in panel 5 still contain high amplitude tube wave. Remnant
aliased tube wave events are highlighted in yellow on panels 5-8. A primary upgoing
event (colored green) located at 0.9 s on panels 7 and 8 (Figure 2.2) is partially masked by
the highlighted remnant tube wave event.
Panel 8 shows the upgoing events after surgical f-k muting of some of the aliased
tube wave. The f-k plot of the data following the surgical muting is shown in Figure 2.17.
The events surrounding the tube wave have been raised in amplitude; however the pres-
ence of the tube wave event still makes the interpretation of the upgoing events difficult.
Processors can choose to enhance the upgoing primary events using a very long median
filter (panel 7), but this procedure is mostly cosmetic and does not accurately preserve
amplitudes.
Some intermediate steps can be used to select the length of the median filter
that best separates downgoing and upgoing wavefields. For example, in panels 1-5 of
Figure 2.3, a series of Zdown(-TT) data panels for a marine near-offset VSP have been
generated using a 3-, 5-, 7-, 9-, and ll-point median filter, respectively, on the origi-
nal Z(-TT) data. The separated downgoing waves appear similar with minor
amounts of residual upgoing waves left in panels 1 and 2. In panels 6, 7, 8, and 9, the
ll-point median Zdown(-TT) data have been subtracted from the 3-, 5-, 7-, and 9-point
median filter panels, respectively. The amount of residual upgoing energy in the
downgoing wavefield panels can be qualitatively estimated using this technique. The
3-,5-, and 7-point median filters are not suitable for wavefield separation of this par-
ticular data set. This exercise is important because it shows that any residual up-
going wave events in the Zdown(-TT) data are subtracted out of the Z(-TT) data dur-
ing wavefield separation and can never be recovered.
33
-
E
:c
f-
o,
L.LJ
,
o
...
IIiC
I ~O O
1""1 0
Ilil O
1"· 0
'.' 0
lle~
raee
IllO
g
,.."'.
"' 0
, eGO I
Ii:
TRAVELTIME (s)
.,...
,,~ o
L.LJ
o
...."".
" lioO
.180 I'"
IfiBO
""
I"ZO
....".
'li lO
I~"O
I JOO
1' 2'0
1 ~·O
'''3' 0.
'0
...
m. 'JOO
...
.,.......
'221>
,
1: "0
,
114 0
•.00••
..'"I"., oz.
,...
TO•
•••
leeo
li ce
1710
....
, ItO
,,..
l ~O
:· t o
171 0
,,..
•lila
I1AO ''''0
lJ80
"00 IJ OO
..,
Il Z'O
12'20
.........
1140
1140
lOGO
101i0
SO,
IIl 0
,.,
'" TO.
,...
•••
....
'no
,,,.
I BI D
1710
111 0
litO 1120
" ":1
1'600
13·0
1)00
....
IUO
1'40
1180
......,...
IJOO
.....
I UO
11 40 mo
1 14 0
'000
Ol.
".
•••
0<.
.
..."
,...
19'0
18fJO IliC
1810
liO O
,oz.
'7;'0
11'20
11;2 0
I ~ "O
''''0 1'''0
IJeD ."'0e
ne
1300
!Joe
....2,....
It10
1140 11 20
....2.'''''
'00. 1140
,
TOO
. .....
'IIO
34
Karhunen-Loeve (K-L) filtering

In this section, the K-L transform is used for wavefield separation (Flowcharts 3A
and B). The gained Z(- TT) data are K-L transformed; the K-L domain consists of princi-
pal components that are related to the eigenvectors of the data autocovariance matrix (see
Appendix for details). The eigenvalues associated with the eigenvectors are indicative of
aligned phase coherency within the data.
Z(FRT)
+ Normalize on first-break data
Z,••(FRT)
..
..
Pick first-breaks, align (-TT)
--- OR
Z(-TT)
. I" Perform K-L transform

----
- Oecide which eigenvalues and
eigen vectors (principal components)
repres ent P-weve downgoing events
-Invers e K-L transform Z(-TT) data using
prine ipal components of upgoing VSP
even ts
+
I z...,(-TT)
I
I •
Z",,(+TT)
I
Static shift data to (+TT)
Determine cause of
I •
z"P(mod)(+TT)
I
Apply median filter
parameters
~ YES

Hes the choice of principal components or first-break
! NO
Use Z."",,(-TT), Z",(+TT). and Z"P(mod'(+TT) in the

Use z"P(mod,(+TT) In Integrated interpretive display (110)
Flowchart 3A: An example of an interpretive processing flowchart for the K-L transform
wavefield separation using the inverse K-L transform to isolate the upgoing events.
35
Z(FRT)
•
Normalize on first-break data
-----
~
Pick first-breaks OR
Static shift data to (-TI)

Z..,,(-TT)
• -----
... Perform K-L transform
- Decide which eigenvalues and eigenvectors (principal

components) represent P-wave downgoing events
- Inverse K-L transform Z\llJlro(-TT) data using principal
components of downgoing VSP events
.
e Subtract Z,_(-TI) from Z(-TI)
to output z,.,(-TI) data.
+
Z...(-TT)
Determine cause of
parameters
Static shift data to (+TI)

apply median filter
J
Z",(mod,(+TT)
YES

Has the choice of principal components or first-breaks
.. NO
Use Z_(oTT), z,.p(+TT). and z..cmodl(+TT) in the

Use Z""modl(+TI) in integrated interpretive display (110)
Flowchart 38: An example of an interpretive processing flowchart for the K-L transform
using the inverse K-L transform to isolate the downgoing events. The downgoing events
are then subtracted from the Z(-TTl data to output the upgoing events.
The simplest assumptions involving the Zl-TTl data are that the downgoing P-
wave events are time aligned, and that each downgoing multiple is time-delayed rela-
tive to its associated primary downgoing event and also similarly time aligned during
the (oTT) time shifting. These assumptions have enabled processors to use the
Karhunen-Loeve transform (Hotelling, 1933; Karhunen, 1947; Loeve, 1948, 1955; Kramer
and Mathews, 1956; Watanabe, 1965) to decompose aligned data via eigenanalysis
(Hinds et al., 1986; Freire and Ulrych, 1988).
36
Two equivalent procedures, called the singular value decomposition method

(5VD; Freire and Ulrych, 1988) and the K-L transform, use the magnitude of eigenvalues
to select a reconstruction based on the trace-to-trace coherency of the data. Following
eigenanalysis (Jones, 1985; Jones and Levy, 1987; Jackson et al., 1991), the largest eigen-
values (Hardage, 1992) represent the principal components having the greatest amount
of coherent energy (Jones and Levy, 1987). The output data are reconstructed using
selected principal components (see Appendix). Some of the principal components asso-
ciated with the largest eigenvalues are not used in the Z(-TT) data reconstruction. As
shown in Flowchart 3A (K-L method A), the data are reconstructed (inverse K-L trans-
formed) excluding the principal components that represent the aligned downgoing
events. Leaving these out during the inverse K-L transform results in the Zup(- TT) data.
The data within the K-L 1PP are interpreted, and if processing artifacts have been intro-
duced into the data, the processing steps are iteratively modified.
This method of inverse K-L transformation using selected principal components
is used when data can be separated according to lateral coherency. In surface-seismic pro-
cessing, multiple attenuation can be effected by applying NMO corrections using the
multiple velocity, and then using the K-L transform to reconstruct prestack events that are
not linearly coherent (Jones, 1985; Jones and Levy, 1987; Hinds and Durrheim, 1993).
Alternatively, interpreted primary events can be horizontally aligned using an NMO cor-
rection with the "best" primary velocity function. Following the K-L transform, the prin-
cipal components that define the largest eigenvalues can be isolated and used during the
inverse K-L transformation to enhance the primary events and attenuate the multiple
events. In VSP processing, the -TT downgoing waves are aligned to make the down
going energy those events that have the greatest linearly coherent energy (Hinds et al.,
1986; Freire and Ulrych, 1988). Following the eigenanalysis in K-L method B (Flowchart
3B), the data are reconstructed to yield the separated downgoing events using only those
principal components associated with the largest eigenvalues. The separated downgoing
data are then subtracted from the Z(-TT) data to output the Zup(-TT) data. An excellent
review of the use of the K-L transform on VSP (and CWT) data is given in Hardage (1992).
Jones (1985) shows that the K-L transform can preserve the edges of events as well as sep-
arate events based on their linear coherency.This is important when dealing with multiple up-
going and downgoing events. Recallthat the interpretation of the interbed downgoing multiple
depends on showing that the downgoing multiple does not exist on all the depth traces.
The wavefield separation 1PP for K-L method B is displayed in Figure 2.4. The
data that have been normalized using a calculation window around the primary down-
going event are shown in panel l. This normalization of the input data to the SVD step is
crucial (Freire and Ulrych, 1988; Hardage, 1992). Another step imbedded in the method
can be the normalization of the eigenvalues following the SVD step.
37
E
I TRAVELTIME (s)
.....
n,
w C!
o 0
,"".
..,
P")
'~""l
- Q)
rJ)
,,,.
1"0
"0
t:
"
,n"
Il IQ
". I
'-"
II ' 0
10'0
". S
11. I TRAVELTIME (5)
II.
! )O
6::w
...".". o
0 N
o
"<t
a
co
o
".00. I ~ ~O
0
. ,.,
I n~
11 0:0
....
18" 0
., ' 0
1"1 0
""'0
.i1 0
I' »)
' '\)0
. ..... 0 1"'0
IUO
I PO
11t 3
m. 1110
.... 10
I UO
,.,.
IU O
.."..
I t ') !)
,'. ,,
'"
.".,'",. "..
OlD
."
".
....""
n.
.""
.,
JI.
,no ",.
,no
, ..,
1"70
1610
. In o
177 0
.n o
1•• 0
,.,p,
I '))C ,.,.
1!t ) O
,.."
1110
~
I-,
I
"
no
11 :0
" ,...
,.
..
"'"
".,.
.->.
N
.'.
"
.,'..,....
01 •
• te
.,.
""
1).
"'
".
...
...
·to
])
,no
" , no
1 ' '' (1
1" 0
....
" '0
. ,...
" ~O
, rc "1 0
I ~ JO "
I ...n
1] ' 0 1 1'0
...
,no
"""
1;'10 111 0
,.,.
..
,,)0
,1 11~
--...... ".,,..
,". ,'.
...,'.....".
,
01 •
'
.. ......
".
".
, ..
38
The Z(- TT) data are displayed in panel 2. The downgoing events Zdown(-TT) in
panel 3 display a more uniform appearance than the separated downgoing events
obtained using the median filter wavefield separation (panel 3; Figure 2.1). Panel 3 (Fig-
ure 2.4) was constructed using the first two eigenvalues only (related to the first two prin-
cipal components) created during the K-L transformation. Compare panel 3 of Figure 2.4 to
panel 4 of Figure 2.1 (median filter method). The K-L reconstructed downgoing events show
slightly more realistic amplitude changes along single downgoing events across the traces.
The median filter tends to smear these amplitude variations.
The separated upgoing events Zup(- TT) are output (panel 4) by subtracting
the data in panel 3 from the Z(-TT) data in panel 2. In panelS, the Zup(+TTl data are
shown. An enhanced version using a second application of a K-L filter is presented
in panel 6. For a comparison, a 5-point median was also used to enhance the upgo-
ing events contained in the Zup(+TTl data. The output is shown in panel 7. The
enhanced upgoing events in panels 6 and 7 are similar (since a 5-point median filter
does not severely affect the data). The primary upgoing event, embedded in the con-
taminating multiples appearing beyond the 1210 m depth trace, is adequately pre-
served for interpretation purposes.
Compare panel 4 of Figure 2.4 to the median filter separated events in panelS of
Figure 2.1. The K-L separated upgoing event panel still contains remnant downgoing
event (for example, at 0.43 s between the 710 and 1050 m traces; at 0.525 s between the
1090 and 1550 m traces). Fine tuning the K-L reconstruction by using additional principal
components should eliminate these downgoing event residuals.
The K-L transform method (either Aor B) should be used when amplitude preser-
vation is critical and when interbed multiples within the separated downgoing events are
being investigated.
Fourier transform (f-k) filtering

In this section, f-k domain filtering within wavefield separation is discussed.
In the depth-time domain, the upgoing and downgoing events overlay and interfere
with each other; however, in the f-k domain, the upgoing events are separate from
the downgoing events (Hardage, 1985). If user selected portions of the f-k domain
Z(FRT) data are attenuated, the downgoing events can be effectively eliminated. The
resultant depth-time data, Zup(FRTl, contain the separated upgoing events. Two
types of f-k domain techniques are currently available within interpretive process-
ing:
1) designing the f-k mute parameters, implementing the parameters within a process-
ing runstream, and evaluating the f-k domain wavefield separation IPP; and
39
10
20 "'T1
;0
m
oc
30 ~
Q
50
60
(A)
DEPTH (m)
330 530 730 930 1130 1330 1530 1730
0.2
+"+t~'Ti+i--l 0.4
1.0
1.2
1.4
(8)
Fig. 2.5 Depth-FRT time and t-k domain plots for the Fort
St. John Graben case study data. Note that the up- and
downgoing P-wave events separate in the f-k domain.
40
2) using interactive screen processing to compare the input and outputs of the J-k
filtering step, evaluating the success of the single step of processing, and mod-
ifying the J-k filtering parameters until processing artifacts are minimized.
The term J-k muting describes the lowering of J-k domain values within a specific
mute window to attenuate the depth-time events therein. The boundary between the
mute window and the desired J-k domain data is usually not sharp (March and Bailey,
1983), rather it is sloped to minimize the 2-D version of Gibb's phenomena (Bracewell,
1986). Before the introduction of software packages that accepted polygon definitions of
the J-k mute zones, two input velocity values were used to specify the familiar pie-slice
(velocity filter) mute (Kanasewich, 1981).
Z(FRT) data contain upgoing and downgoing events that map onto the negative
and positive quadrants of the J-k domain after 2-D Fourier transformation, respectively.
A linear event in the depth-traveltime (FRT) domain maps as a linear event in the J-k
domain. As shown in Hardage(1985), the upgoing and downgoing P- and SV-events
become separate linear events in the J-k domain and therefore are easily subjected to J-k
masking (attenuation).
The relationship between frequency (w), spatial wavenumber (k,), and apparent
velocity (V) is w = k, . V. This implies that linear events in the Z(FRT) data map into linear
events that pass through the origin in the J-k domain and have a slope equal to their appar-
ent velocity. The relationship between frequency, spatial wavenumber, and velocity
(inverse of p) highlights the relationship of J-k filtering to the t-p filtering-based wavefield
separation. Excellent overviews ofJ-k domain processing are presented in Hardage (1985,
1992), Hatton et al. (1986), and Yilmaz (1987).
To illustrate the different VSP events in both the depth-time (FRT) and f-k
domains, the Fort St. John Graben (FSJG) case study near-offset Z(FRT) data are shown
in Figure 2.5. Take some time to review both domain presentations and look for;
1) downgoing P-wave events (positive k quadrant);
2) upgoing P-wave events (negative k quadrants);
3) downgoing tube wave events; and
4) upgoing tube wave events.
For clarity, only the upper 1.5 s of data are plotted in Figure 2.5A. The upgoing
tube wave J-k event results from the reflection of the downgoing tube wave at the bottom
of the borehole; however this is not evident on the time domain plot. With respect to the
downgoing tube waves, the nonaliased data are between 0-35 Hz in the positive k quad-
rant. The aliased downgoing energy wraps around between 35 and 60 Hz in the negative
k quadrant (Figure 2.5B). Downgoing tube wave energy with frequencies above 35 Hz
appears to be upgoing events in the depth-time domain because of spatial aliasing (the
41
WAVENUMBER ( 11m • 10. 5 )

-2500 -1875 -1250 -625 0 6 5 1 50 1875
10
20
50
60
(A) QUADRANT MUTE (B) POLYGON REJECT
5
WAVENUMBER (11m' 10. )
-2500 -1875 .1250 -625 0 625 1250 1875 2500
o
10
20
."
;z:J Fig. 2.6 Categorization of the types of f-k
~ based mutes.
30 Iiiz
~
40 if
50
60
(C) POLYGON ACCEPT
5)
WAVENUMBER (11m' 10. WAVENUMBER (11m' 105 )
-2500 -1875 -1250 -625 0 625 1250 1875 -2500 -1875 -1250 -625 0 625 1250 1875
10 10
20 20
-n ."
30~ 30 ~
o
c:
m
szm
z
40 ~ 40~
:I: :I:
N N
50~ 50~
60 60
70 70
(0) SURGICAL MUTE (E) SURGICAL MUTE
42
depth recording interval was too coarse to faithfully record all of the tube wave energy).
Similarly, for the upgoing tube wave events, the nonaliased energy lies between 0-35 Hz
in the negative k quadrant, and the aliased energy wraps around between 35-60 Hz in the
positive k quadrant. An excellent review on tube wave aliasing is given in Hardage

(1981).
The aliased downgoing tube wave event intersects the upgoing P-wave events at
approximately 55 Hz in the negative k quadrant (note that the P-wave events do not wrap
around in f-k space). This interference is a source of noise. In Figure 2.6, different types of
f-k mutes are shown. These are reviewed later in more detail. The muting procedures
used are:
1) f-k quadrant attenuation (the data in one of the k quadrants are muted; Figure 2.6A);
2) polygon or narrow reject zone muting (similar to pie-slice filtering discussed in
March and Bailey, 1983; Figure 2.6A);
3) polygon or narrow accept zone muting (Figure 2.6C); and
4) nteractive polygon muting (more than one polygon within the f-k domain is speci-
fied; Figures 2.6D and E).
The interpretation of the separated up going events are discussed below
for multiple- and tube wave contaminated Z(FRT) VSP data (Figures 2.7 through
2.16). The f-k mute affects the isolation of primary, multiple, and tube wave (aliased
and nonaliased) events. We demonstrate this procedure using the f-k domain IPP.
The results of various types of f-k filtering on the FSJG Z(FRT) multiple-contam-
inated data are shown in Figures 2.7 through 2.13. These displays serve as visual intro-
ductions to data processing in the f-k domain. The Z(FRT) data and corresponding f-k
plot are presented in Figures 2.7 (panel 1) and 2.8, respectively. The horizontal axis in Fig-
ure 2.8 is the spatial frequency (multiplied by 10-5). The vertical axis is the frequency val-
ues in Hertz.
The dominant events in the f-k plot are the downgoing P-waves (shown in the
positive quadrant). The downgoing tube wave is shown in panel 1 of Figure 2.7 as the
dispersed wavetrain emerging on the 890 to 1610 m traces at 1.2 s (FRT). On the f-k
plot, follow the downgoing and upgoing tube wave events (highlighted in red in the
positive and negative quadrants) that have a slope of 1400 and -1440 mys, respectively.
Aliasing of the tube wave occurs above 35 Hz. The spatially alia sed part of the
downgoing tube wave reappears above the negative Nyquist spatial frequency
(1/2 D z = -1/40 m- 1 = -0.025 m- 1) as a positive slope in the negative f-k quadrant.
An upgoing tube wave originates at the bottom of the borehole (panel 3 and 5 of
Figure 2.1). The nonaliased portion of the upgoing tube wave event lies between 0 and
35 Hz in the negative f-k quadrant; the aliased (higher frequency) portion wraps back into
43
......
.s
J:
TRAVELTIME ( 5 ) -E
J:
N
TRAVELTIME ( 5 )
CXl
6:w ~ ...C"! I-
a. c:i c:i
o o Q)
I I~ ~
">O~~~~~~~~~~'~
1BSOJ ~
1"0
E
:"30~:;~-~~-;-~-I-i·llill~lll~ 111111111111111
~ I~ §
~
E ~
1510 C)
~
.-... ... w
"I~ 0
14>O~iii;!~
1J70
Il'JO
1210
~
~ ~ 1m
t6 mo • i=' ~
" U. .~
g- ;:l IIJO '""Q. E'
:: J =========i'fi.~t~ N
_~L. 0" 112101
0'01 1 1 11
~ rx::]
C'Il
N~
;:l
9 70 j. ~~§~~~ e~ no -~
810-b==~~~~
' 90
110
730
~~
73 0
" 0
' 50
570 "0
490 4'0
410
330
1950~1I
I'OO ~
1170
"C ..~
'IO
jliiiiil
1
Q.
Q)
C'Il 1530
.0 1400 8
C'Il
-.
I- .$ Ina
3:l
I - 1370
c:
rx: ~ IllO
o
u.0)
LL
.......... C IllO
-~
0.-
:c I050j~~11
:]0
Ngo C'Il 970 NO.
---Q) no
:'tD.'Wl"-=~:" ~ E 110 -..:Ie
e~
50
730
"0
410
410
no
1850
1770
I s>ojlliiilllllll•••11
I' SO
II,.
1770
1510
1590
")0
1'10
1400
11°°ll l l l l l ill
1400
1370
137 0
IllO
1290
121 0
12 10
IIJO
1130
lO5Oj~~~~~
1050
.........
17 0
890
810
7 )0
- 170
t90
1101==
730
'50
=
B111l1111i1l
500
"0
490
490
4 10
070
410
330
330
Fig. 2.7. Z(FRT) and Zup(FRT) data wavefield separated using the median filter (plus sub-
traction) and various f-k operations (mute designs in Figure 2.6) for the Fort St. John
Graben near-offset multiple contaminated data.
44
10
20 ."
:;0
m
oc
30 m
z
o
-<
40 ~
50
60
Fig. 2.8 The f-k plot of the Z{FRT} shown in panel 1 of Figure 2.7.
The non-aliased up- and down going tubewave f-k events are high-
lighted in red.
WAVENUMBER ( 11m * 10- 5 )

-2500 -1875 -1250 -625 0 625 1250 1875 2500
o
10
20."
:;u
m
oc
30~
o
-<
40~
50
60
Fig. 2.9 The f-k plot of the median filter and subtraction method
derived Zup{FRT} data (shown in panel 2 of Figure 2.7).
45
the positive f-k domain. The aliased downgoing tube wave f-k event overlays the upgoing
P-wave events at 55-60 Hz in the negative f-k quadrant; however, the bulk of the upgoing
P-wave energy resides at lower frequencies. The upgoing P-wave f-k events in the negative
domain are low amplitude compared to the associated downgoing events (Figure 2.8).
The output of the median filter wavefield separation (panel 2; Figure 2.7) are
included alongside the f-k filtering output (panels 3 to 6) for comparison purposes. Note
how the median-filtered data appear in the f-k domain. The remnant tube wave in panel 2
is shown in the correspondingf-k plot (Figure 2.9). Poor tube wave attenuation is obtained
using the median filter method. Only a narrow zone immediately around the downgoing
P-wave events has been attenuated; this means that most of the tube wave events remain.
The performance of the f-k filtering method in wavefield separation depends on
how the f-k mutes are defined and applied. Panel 3 of Figure 2.7 displays the result of the
simplest type of f-k filtering; namely, the masking of the entire positive quadrant. To eval-
uate the effect of the f-k quadrant elimination processing on the Z(FRT) data (panel 1 of Fig-
ure 2.7), compare the f-k plot of the Z(FRT) data (Figure 2.8) to the f-k plot of the Zup(FRT)
data (Figure 2.10). In panel 3 (Figure 2.7), the downgoing tube waves with dominant fre-
quencies below 35 Hz are attenuated; however, the aliased downgoing tube wave (appear-
ing between 36-60 Hz) and unaliased upgoing tube wave energy (between 0-35 Hz)
remain in the Zup(FRT) data. There is a band of energy associated with a broad k, (hori-
zontal axis of the f-k plot) range centered at 30 Hz that could be sonde resonance.
In panel 4 of Figure 2.7, the result of attenuating the content of a surgical f-k mute
polygon zone around the downgoing P-wave f-k events is illustrated. The f-k mute zone is
highlighted in the f-k domain plot (shown in Figure 2.11) of the Zup(FRT) data of panel 4
(Figure 2.7). Downgoing mode-converted shear waves are evident in the Zup(FRT) data
(highlighted in blue in panel 4) and in the f-k plots of Figure 2.11. The downgoing mode-
converted shear wave events (shown in blue on panel 4) display a similar dip or apparent
velocity to that of the downgoing tube wave events (displayed in yellow on panel 4). The
upgoing mode-converted shear wave events plot between the upgoing P-wave and tube
wave events. On panel 4 of Figure 2.7, the bulk of the mode-conversions originate at the
690-730 m interface. The surgical mute zone seen in the f-k plot of Figure 2.11 does not
encompass the tube wave and mode-converted SV-wave events. As a result, these events
appear in the Zup(FRT) data after the surgical mute wavefield separation procedure. Little
if any attenuation of the tube waves occurred during the f-k downgoing P-wave reject fil-
tering.
The results of applying a narrow acceptf-k mute filter to the Z(FRT) data from panel
1 are shown in the Zup(FRT) data in panel 5 of Figure 2.7. The f-k response showing the
highlighted boundaries of the zone within the f-k domain that was retained is displayed in
46
WAVENUMBER ( 11m * 10-5 )

-2500 -1875 -1250 -625 0 625 1250 1875 2500
10
20 "'TI
:::0
m
oc
m
30 z
o
-
-<
40 i:f
-
50
60
Fig. 2.10 The t-tc plot of the Zup(FRT) data shown in panel 3 of Figure 2.7. The positive
spatial wavenumber (k) quadrant has been attenuated. The upgoing and aliased down-
going f-k tube wave events have not been attenuated since these events reside in the
retained negative k quadrant.
-2500 -1875 -1250 -625 0 625 1250 1875 2500
10
20 "'TI
:::0
m
oc
30~
o
-
-<
::I:
40N
-
50
60
Fig. 2.11 The t-k plot of the Zup(FRT} data shown in panel 4 of Figure 2.7. A spatially lim-
ited reject polygon has been used to attenuate the downgoing P-wave i-k events. The
applied mute does not appreciably affect the up- and downgoing tube wave f-k events .
47
Figure 2.12. This is a visual example of Rieber mixing as explained in Hardage (1985). The
narrow range of spatial frequencies in the f-k pass zone results in a broad expression of the
events in the depth-time domain (also termed f-k smearing).
This data smearing effect can be seen by examining the Spirit River and Nordegg
Formation primary and multiple events of the Zup(FRT) data in panelS of Figure 2.7.
Compared to the Zup(FRT) in either panels 4 or 6, upgoing events in panelS are smeared
(trace-to-trace amplitude changes are not preserved) due to Rieber mixing. The Spirit
River primary event (colored green) intersects the first-breaks on the 690-710 m traces. A
multiple of the Spirit River primary event cannot exist on traces deeper than the traces
where the primary event intersects the first-breaks; however, the Spirit River multiple
(colored orange) event extends onto deeper traces. A primary event (shown in blue)
superimposed onto a zone of multiples (colored yellow) originating from the 1250 m
interface (Nordegg Formation event colored pink) is barely distinguishable as a separate
event.
The Zup(FRT) data in panel 6 of Figure 2.7 are a result of applying surgical f-k
polygon zone muting (Figure 2.13) to the Z(FRT) data of panel 1 (Figure 2.7). The surgi-
cal mutes were designed to attenuate the downgoing P-wave and the upgoing and
downgoing tube wave events. Two examples of the lateral truncation of the multiples
below their respective primaries are shown in panel 6 of Figure 2.7. The primaries are
highlighted in yellow and the multiple events are shown in red. The multiple events are
not smeared onto traces beyond the depths at which the primary events associated with
the multiples were generated. To illustrate this, vertical lines have been drawn (in pink)
for both sets of primaries and multiples. These lines indicate the expected lateral extent
of these events. The tube waves have been effectively eliminated, the truncation of the
multiples below their respective primaries is acceptable, and minimal amplitude trace-to-
trace smoothing occurs.
The most acceptable types of f-k filtering for the FSJG data are positive-quadrant
attenuation (Figure 2.10) and surgical f-k muting (Figure 2.13). The concern for this type
of data is the proper truncation of the multiple events and the trace-to-trace preservation
of amplitude variations.
The use of f-k surgical muting on the tube wave contaminated Z(FRT) data is
demonstrated in the wavefield separation IPP (Figure 2.14). Panel 1 shows the normal-
ized Z(FRT) data with a dominant downgoing alia sed tube wave. Panel 2 contains the
Zup(FRT) data resulting from the use of the median-filter-based wavefield separation
method. The upgoing wave data in panel 2 are low amplitude compared to the tube wave
amplitudes. Panel 3 contains the separated upgoing events created by the surgical mut-
ing of the downgoing events and tube wave events in the f-k domain. The aliased tube
48

-2500 -1875 -1250 -625 0 625 1250 1875 2500
10
20 "T1
:::0
m
oc
30~
o
40N
--<
:::I:
50
60
Fig. 2.12 The f-k plot of the Zup(FRT} data shown in panel 5 of Figure
2.7. A spatially limited f-k accept polygon has been used to isolate and
retain the upgoing P-wave f-k events. Due to the spatially narrow accept
zone, the filtering results in Reiber mixing in the depth-time domain.
WAVENUMBER (11m * 10-5 )

-2500 -1875 -1250 -625 0 625 1250 1875 2500
10
20 "T1
:::0
m
oc
30~
o
-<
:::I:
40N
50
60
Fig. 2.13 The f-k plot of the Zup(FRT} data shown in panel 6 of Figure 2.7. The f-k poly-
gons outlined in red show the surgically muted reject zones in the f-k domain designed
to attenuate the down going P-wave f-k events and the aliased and non-aliased tube wave
t-k events.
49
vsp Interpretive Processing: Theory and Practice
(1) Z(FRT) (2) Zup(FRT) (3) Zup(FRT) (5) Zup(+TI)

median filter based '·k surgical polygon 3-point median 5-point median
mute filter filter
Fig. 2.14. The f-k based wavefield separation IPP of the tube wave contaminated data
using surgical muting of both the downgoing P-wave and tube wave t-k events. Panel 2
is Zup(FRT) data resulting from the median filter and subtraction wavefield separation
(included for comparison purposes).
(1) Z (FRT) (2) Zup(FRT) (3) Zup(+IT) (4) Zup(+TT) (5) Zup(+TT)
'·k quadrant mute 3-point median filler 5-poinl median filler
Fig. 2.15. The f-k based wavefield separation IPP of the tube wave contaminated data
using f-k quadrant attenuation.
50
(1) Z(FRT) (2) Zu.(FRT) (3) Zu.(FRT)

median filter based '-Ie quadrant mute
li i~ l! i jj~ i ~ ~! !' l! : ~ i ~ il ~ i ~ l ~~ i j i;. I. " ~~ii ~ ~~~ k l~iii !lll ;' DEPTH(m)
-i
. 0 .4~
§. ~
~ O.8~
. i:
. 0 .8 III
~
Fig. 2 .16. Z(FRT) (panel 1) and Zup(FRT) wavefieJd separated data using the median fil-
ter plus subtraction method (panel 2) and various t-k muting methods (panels 3-6) on the
Z(FRT) tube wave contaminated data .

-1666 -1250 -833 416 0 416 833 1250
"'T'\
:::0
20 m
oc
m
z
o
30~
::I:
N
40
50
Fig. 2.17 The t-k plot of data resulting from the application of f-k surgical muting to the
median filter-based wavefield separation results (unfiltered data is shown in panel 6 of
Figure 2.2).
51
wave f-k events that intersect the upgoing P-wave events were not eliminated by the sur-
gical muting. The Zup(+TT) and median filter (3 point) enhanced Zup(+TT) data are pre-
sented in panels 4 and 5, respectively. The upgoing P-waves in the data are more easily
recognized on the f-k surgical muted wavefield separation data.

The muting of the downgoingf-k quadrant of the time-depth data (panel 1; Fig-
ure 2.15) results in the retention of the aliased portion of the tube wave (panel 2). The
Zup(+TT) data are depicted in panel 3. To further enhance the upgoing events with
respect to the aliased tube wave, 3- and 5-point median filters (plus appropriate post-
median band-pass filter to eliminate median filter whiskers; Hardage, 1985) were
applied to the data from panel 3. The output is displayed in panels 4 and 5, respectively.
For interpretive purposes, the quadrant attenuation f-k method does not solve the prob-
lem of the aliased tube wave.
The results of using various types of f-k filtering on the tube wave contaminated
data (panel 1; Figure 2.16) are shown on panels 3-6 in Figure 2.16. Compare the result of
the median-filter-based wavefield separation method shown in panel 2 with the panels
resulting from f-k quadrant attenuation (panel 3), narrow pie-slice downgoing P-wave f-k
reject filter (panel 4), narrow pie-slice upgoing P-wave f-k accept filter (panel 5), and the
surgical mute f-k filtering (panel 6). The surgical muting method (results in panel 6) is con-
sidered to be the optimum wavefield separation method for the input Z(FRT) data
because the upgoing events are easily interpretable and amplitude differences between
the upgoing and tube wave events are minimal.
Median-filter-based wavefield separation of the tube wave contaminated Z(FRT)
data was reviewed when we examined Figure 2.2. In panels 6 and 7 of Figure 2.2, the
Zup(+TT) data resulting from the use of the median filter are shown. Surgical f-k muting
was applied to the Zup(FRT) following the median filter wavefield separation, and these
f-k filtered data are shown in panel 8 of Figure 2.2. Figure 2.17 is the f-k plot of the resul-
tant data. The surgical mute used during the filtering was designed to attenuate the tube
wave; however, the muted zone of the f-k domain did not include the portion of the
aliased tube wave that intersected the upgoing P-wave f-k events. This is an example of
how the f-k filter surgical muting can be combined with other wavefield separation
methods.
Interactive screen processing for surgical mute processing in the f-k domain is
reviewed with the aid of Figures 2.18 to 2.25. The examples illustrate single-step inter-
pretive processing. The input data are shown on one of the bit planes of the interactive
monitor display. (A bit plane is one of the screen images in memory within modern
graphics systems that can be chosen for display using mouse or keyboard commands.)
On another bit plane, the f-k domain plot of the input data is stored, and, when displayed,
52
mute zones are interactively selected within the f-k plot. The result of applying the cho-
sen f-k mutes to the input data is then stored on yet another bit plane which is displayed
immediately after the filtering has been completed. By comparing the input and j-k fil-
tered data, the effect of the application of the surgical muting on interpretation can be
evaluated. Once any f-k filtering-induced artifacts are minimized, the associated f-k mute
parameters are incorporated into the wavefield separation runstream.
Two examples of interactive interpretive processing using interactive f-k filtering
to interpret upgoing and downgoing event content on Z(FRT) data and to enhance
Figure 2.18 1 , data on screen 1

- display Z(FRn _
- Fourier transform Z(FRT) data into

Figure 2.19 the '-Ie domain
• display on screen 2
w design surgical '-Ie mute of the 11+-- "l

Figure 2.19
downgolng '-Ie events on screen 2 I
- apply surgical '-Ie

mute to the
'-Iedata and inverse transform
Figure 2.20
to obtain ~(FRT).
- display .z,...(FRT) on screen 3 .
• compare Z(FRT) data on screen 1 with the z...,(FRT) data

Figure 2.18 on screen 3. - NO
Figure 2.20 - are the upgolng events on screen 3 interpretable?
1 YES
• output f-k surgicalmute parameters(polygon comers)

for use In batch processing or to retain the interactive
processIng z..,(FRT) results for the IPP
Flowchart 4. An example of the interpretive processing flowchart

of the interactive f-k processing used for event identification.
Zup(FRT) data are discussed below. The processing decisions used for wavefield separa-
tion by interactive f-k processing are shown in Flowchart 4.
The FSJG near-offset Z(FRT) data displayed on the interactive monitor screen are
shown in Figure 2.18. The f-k transformed plot of the same data is shown in Figure 2.19.
A surgical f-k polygonal mute zone is designed using on-screen menu commands. The
53
350 550 750 950 1150 1350 1550 1750 2010 DEPTH (m)
0.0
0.2
0.4
..,
0.6 ~
..,r-m
~
0.8 m
til
1.0
1.2
1.4
· " 1.
.
Fig. 2.18. Single-operation interpretive processing initial screen dis-
play showing the input Z(FRT) data. The objective of the interactive
processing (seen in Figs. 2.18 to 2.20) is to understand the different
types of up- and downgoing wave events within the data.
10
"'T1
20~
oc
m
z
30Q
:I:
N
40
50
AO
Fig. 2.19. The t-lcplot of the data shown in Figure 2.18. This is the next
interactive processing display following the screen presentation of Fig-
ure 2.18. A surgical mute polygon is interactively designed and the f-k
events within the polygon can either be attenuated or preserved.
54
Fig. 2.20. The "results" interactive processing display following the muting (reject filtering)
of the f-k data inside the mute polygon shown in Figure 2 .19. One can flip between the
screen images using the mouse-controlled menu to evaluate the application of the poly-
gon surgical mute. The upgoing P-wave and tube wave content of the Z(FRn data seen
in the initial screen display (Figure 2.18) can now be interpreted.
designed mute zone is highlighted in red in Figure 2.19. The f-k data within the mute
zone can now be rejected or accepted. The resultant Zup(FRT) data are shown in Figure
2.20. The resultant data are plotted onto another bit-plane within the graphics system.
Through the use of on-screen menu commands, the screen can alternatively dis-
play the Z(FRT) shown in Figure 2.18 or the filtered data, Zup(FRT), shown in Figure 2.20.
A preliminary interpretation of the upgoing events can be done using the Zup(FRT) data.
Moreover, by changing the vertices of the polygon f-k mute within the f-k plot bit-plane
display, the effect of the mute design on the interpretation of the upgoing events can be
analyzed. This familiarizes the processor/interpreter with the wavefield content of the
input Z(FRT) data.
The effect of mute design on a particular segment of the VSP data can be examined
through f-k interactive processing. The processing decisions made during interactive f-k pro-
cessing are shown in Flowchart 5. The f-k domain segment of interest encompasses the portion
of the aliased downgoing tube wave that crosses the upgoing P-wave events (Figure 2.9).
The input data to the second example of f-k interactive processing are Zup(FRT)
data (Figure 2.21). These data resulted from median filter wavefield separation and the
downgoing tube wave is still evident within the Zup(FRT) data. The f-k representation of
this data is shown in Figure 2.22. The upgoing and downgoing tube waves are within the
55
display Z,.(FRT) data fromthe medianfitterand

subtraction wavefield separation method on screen
1 (Figure 2.21)
f-ktIanSfonnthe Z;..(FRT) data anddisplayon

screen 2 (Figure 2.22)
- design surgical (-Ie mutezoneson the - design surgiCal (-Ie mutezoneson the
up- and downgoing a1iased and 000- up- anddowngolng allesed and ncn-
allasedtube wave f-k svants(ALLOW aliasedtube wave '-Ie events (DO NOT
the muting of the upgOing P-wave (-Ie AllOW the muting of the upgoing P-W8VEl
eventsthat are intersected by the allased (-Ie eventsthat are intersected by the aUssed
upgolng tubewave f-k avents): mute A upgolng tube wave (-Ie events): mute B
- apply muteA to the inputdata - apply muteB to the input data
- displaythe filtereddata on screen3 • displaythe filtereddata on screen 4

(Figure 2.23) (Figure 2.25)
! !
- comparethe z....(FRT) data on screen3 withthe Z,,(FRn
data on screen4
NO_
- does the interpretation sufferbecauseof the omission of f-------+ YES
1
the upgolng P-waveeventsattenuated through the use of
1-
mute A ?
output muteA parameters 1- output mute B parameters

1
I I
I
- use the chosen f-lf surgical muteparameters in betch
processing or use the interactive results to refinethe
z...., data In the f-k wavefleld separatiOn IPP
Flowchart 5. An example of the interpretive processing flowchart

of the interactive t-k processing used for the attenuation of tube
wave events on Zup(FRT) data.
first selected mute zone polygons (outlined in red). The aliased part of the downgoing
tube wave events in the f-k domain crosses the upgoing P-wave events at approximately
-1250 * 10-5 m! on the spatial wavenumber axis (k,) and at 55 Hz on the frequency axis.
The objective for this example is to design the surgical mute f-k polygon that best
attenuates this tube wave. The f-k surgical mute polygon in the negative k, quadrant (Fig-
ure 2.22) intersects upgoing P-wave f-k events between 50-60 Hz. Muting the f-k data
within these reject zones eliminates the associated upgoing P-wave events. This
approach illustrates how f-k polygon muting can adversely affect the interpretation of the
upgoing P-waves in the depth-time domain. The filtered Zup(FRT) data are shown in fig-
ure 2.23. The downgoing and alia sed tube wave events have been attenuated.
The f-k plot of the original data (Figure 2.22)can be used to redefine the f-k mute zone
such that it does not intersect the upgoing P-wave data. By interactively redefining the fil-
ter, the effect on the interpretation on the upgoing P-wave data by the previous f-k filtering
(Figures 2.22 and 2.23) can be ascertained. The surgical mute polygons are selected to be the
56
350 550 750 950 1150 1350 1550 1750

Fig. 2.21. The Fort St. John Graben Zup(FRT) data resulting from median fil-
ter-based wavefield separation (plus subtraction). The downgoing tube wave
is aliased back into the data as an upgoing event. A question that the single-
operation interactive processing can answer is "how does one design the
surgical f-k mute for the attenuation of the tube wave events and what
effects, if any, does the design have on the upgoing P-wave events? "
-------
-25 0 -1875 -1250 -6 5 6 5 1 50 1875 ..•
10
,,'>/.(/'?:~.. <~&
. /'
:, ' " ", \. '
20
, ./ ~ ' / : 4~~. . " <>. . ""
",/'"h...:~~,)~, '/,',' '" ~~_". '" ' " , 'Tl
. , .... .
" \.~~-40' " i, ~
" ., / . _
" ,
¥ , ,
... 30
oc:
:-' ~ I · '. ~ ,
s'" :'<." .: <, ' : > .; m
-, .~ / /
40 ~
-<
,,.~~
- ,~, . .
,
,/:'; /
/ /
:t
50~
60
. ~ / ' /
""" ././ /. -':~ '/ -

"
'. ' . . /
. / 70
Fig. 2.22. The f-k plot of the data shown in Figure 2.21 illustrating the
surgical mute reject polygons enveloping both the up- and downgoing
t-k events (aliased and non-aliased portions included). The polygon in
the negative k quadrant intersects the upgoing P-wave '-k events.
57
0.6
~
""""
<
m
0.8 ~
=::
--
m
en
Fig. 2.23. The filtered output of the data in Figure 2.21 following the application of the
f-k polygon surgical mutes shown in Figure 2.22. The aliased and nonaliased portions
of the tube wave events have been attenuated. The next question is "what effect would
the application of t-k mute polygons which do not intersect the upgoing t-k events have
on the filtered output?".
same as those shown in Figure 2.22 except that the negative kz quadrant f-k polygon mute
will not attenuate any upgoing fk P-wave events (Figure 2.24). This new set of surgical
mutes yields the results shown in Figure 2.25. Figures 2.23 and 2.25 can be compared to
evaluate the effect of muting the upgoing event f-k data that lie in the 50-60 Hz range
of the f-k domain. The upgoing events in Figures 2.23 and 2.25 are similar. The inter-
pretation of the upgoing events are not degraded using the f-k mute shown in Figure
2.22. The fk procedures shown in Figures 2.18 through to 2.25 are other types of inter-
pretive processing that the interpreter Iprocessor can use.
58
------
-25 0 -1875 -1250 -6 5 6 5 1250 1875 II
•
10
. ) . e-
---I / /~~ 10.. ; - - , ,
/I . ~. ~.
- ,.' - ./.,:' ,I . ~.0: "-
.. ..'.
• I I , " "1 ' : . , '
-~~
. ,/./;..,/ : ~ :.~'S . ~.: _,~. ~ ~. 20
• ' .4 '! / ' . - :~.' '" "," .-, \' :' ,."
.,: 1i/ ,', ~-: .~~~~', , : .' ~ ..:'
- , :-~' ~'»
. - "'-';'.;'''
. .' '
"'T1
/; ~... .. . ' - 30 ~
oc:
,I:·
?
,"" ,
(
.' .• , ~
'
,
-
, ,,
~- '
.
" i.
,;-
.
40 ~
m
" 'Y'-
"
\. ~~ :.""':1" " " .- :

, "' _ . '. .
.-Co ' _ • /
I
- , _
-<
:I:
:V-..' :.- ", . / . 50~
," j '- . ' .'
.·l :.. ,-,~ '/

' I' ' / /
'/>' . 60
(,<. ~.- ;I/ ::~. '. .
70
Fig. 2.24. The f-k plot of the data shown in Figure 2.21 illustrating the
surgical mute reject polygons enveloping both the up- and downgoing
tube wave f-k events (aliased and nonaliased portions included) , how-
ever, excluding the muting of any upgoing P-wave f-k events.
350 550 750 950 1150 1350 1550 1750 2010 DEPTH (m)
'. ' : 0.0
0.2
Fig. 2.25. The filtered output of the data in Figure 2.21 after the
application of the polygon surgical mutes shown in the f-k plot
of Figure 2.24. A minor amount of the aliased downgoing tube
wave has been retained in the filtered output.
59
t -p filtering
In this section, the use of the t-p or Radon transform (Deans, 1983; Robinson, 1983;
Durrani and Bisset, 1984; Moon et al., 1986; Hu and McMechan, 1987; Carswell and
Moon, 1989; Turner, 1990; Miao et al., 1995) in VSP wavefield separation processing is dis-
cussed. We examine four routes through which the t-p transform can be used for wave-
field separation. These options are summarized in Flowchart 6 and consist of:
1) Route A: t-p transform the Z(FRD data with upgoing event slownesses ( p is the
inverse of velocity), and inverse t-p transform to output Zup(FRT) data;
2) Route B: t-p transform the Z(FRT) data with downgoing event slownesses, inverse
t-p transform to output Zdown(FRT) data, and subtract the Zdown(FRT) from the
Z(FRT) data to output the Zup(FRT) data;
3) Route C: t-p transform the Z(FRT) data with upgoing event slownesses, and inverse
t-p transform using a depth interval parameter of half the original depth spacing to
spatially interpolate the Zup(FRT) data; and
4) Route D: t-p transform the Z(FRT) over both up- and downgoing event slownesses,
inverse t-p transform using half the original depth increment, and wavefield sepa-
rate the Z(int"pl(FRD data using the median-filter-based wavefield separation to
output Zup(int"pl(FRT) data.
These t-p transform wavefield separation routes are applied to both multiple- and
tube wave contaminated Z(FRT) data. Multiple contaminated data are used to illustrate
Route A processing. Rieber mixing can be observed on the Zup(FRT) data when using pro-
cessing Route A. The lateral preservation of multiple events on the Zup(FRT) data is exam-
ined below. The three other routes of example t-p processing use tube wave contaminated
Z(FRT) data. The tube wave events in these data examples are spatially aliased. The effect
of the processing schemes on the spatially aliased tube wave is discussed.
A conceptual explanation of the t-p transform is illustrated in Figure 2.26. The t
variable refers to the time intercept of the lines of integration in the time-depth (or time-
offset for CDP gathers) domain. The p variable refers to the slowness (inverse of velocity)
values associated with the lines of integration. In Figure 2.26A, two lines with different
slopes are shown intersecting the example time-depth data point. Integration of the time-
depth data along each line, I=pjx+tj and I=P2X+t2, would yield two values in the t-p
domain as seen in the lower half of Figure 2.26A. The different slopes of the lines refer to
different p or slowness values.
To perform the inverse t-p transform (Kappus et al., 1990), the same procedure of
integrating along straight lines is done within the t-p domain. As shown in the top part
of Figure 2.26B, three lines of integration through a single point in the t-p domain result
in the values for three points along a straight line in the time-depth domain.
60
Flowchart 6. Example interpretive processing flowcharts using the t-p transform. The t-p
processing in Route E uses data that undergo the transform of x-z to t-p and then from
t-p to w -p using the Fourier transform on the t variable.
ROUTE A ROUTE B ROUTE C

Z(FRD Z(FRD Z(FRD
.. normalize data normalize data .. normalize data ..
- determine p range of .----- Z.~(FRD
- determine p range of both
upgoing events
- r-p transform the dala ~ upgolng and downgoing
events
within this prange - determine p rangeof the - t-p transronn the data :
- inverse '["-p transform downgoing events upgoing event prange
- r-p transform the data - inverse r-p transform and
within this prange simultaneoustyperform
- inverse r-p transform depth Interpolation (use
1/2 l::.z of original forward
~ transform)
Z_(FRD
~
~ Z,....,,(FRTj
ROUTE D
e-
Z(FRT) ~
z",(FRD
ROUTE E
~ normalize data
- determine p range of both up- and downgoing

events (1) interactive r- p
- r-p transformthe data within this prange surgical muting
- inverse r-p transform and simultaneously (2) UJ- P interactive
perform depth Interpolation (use 1/2 l::.z of processing
original forward transform)
First break pick

Align to ( -IT)
Z,_,(-TD--------,
I Apply median filter
t Normalize data
z.-..(I>lletpl(-TT)
~
e+---------'
~
ZUP(lnl.rp~FRT)
Whereas the J-k transform mapped a line in the time-depth domain into a line
intersecting the origin of the f-k domain, the t-p transform maps a line in the time-depth
domain into a point. The two transforms are linked through the p, or inverse velocity,
parameter. Rieber mixing that can occur during f-k filtering also occurs in t-p filtering.
For surface-seismic data, the hyperbolic-shaped reflection events on a CDP gather
map into ellipses in the t-p domain (Stoffa et al., 1981). For multiple attenuation in surface-
seismic data, the CDP gather can be transferred into the xL t2 domain and t-p transformed.
Using interpolation in the xLt2 domain, the hyperbolic reflection events of the x-t data
become linear events in the xL t2 domain and "points" in the t-p domain. The points in the
t-p domain are smeared because of truncation effects resulting from finite events and the
61
Fig. 2.26. The concept of the forward and inverse top transform. The forward transform
(part A) maps a line into a point and the reverse transform (part B) maps a line in the top
domain back into a point in the time-offset (depth) domain (after Hardage, 1992).
t (5)
.(s)
'2
x x (m) P2 p(sfm}
!
FORWARD r- P
TRANSFORM
Xi= -tan4l;
!INVERSE,-p
TRANSFORM
Ij = P2 x/ t 2
'2
----.....JL-----_+ P
. P....
x (rn)
(A) (B)
representation of the events with near-zero spatial frequencies (near-offset events at the
apex of the reflection hyperbolas). Muting in the top domain results in multiple attenua-
tion (Hinds and Durrheim, 1993).
Instead of appearing as hyperbolic events (such as those seen in the CDP gather
of surface-seismic data), the depth-time events of the Z(FRT) data are linear. The purpose
of the x2.t 2 processing of the CDP gathers is to transform the hyperbolic events into lin-
ear events. Similarly, VSP linear events are transformed into "points" following a top
transformation.
For our interest, the top transform is able to separate upgoing and downgoing
VSP events into oppositely signed top quadrants and also to separate the events based on
their time-intercepts with the FRT time axis (Hu and McMechan, 1987; Hardage, 1992).
Route A: x-p transformation of upgoing VSP events

Rieber mixing can be a factor in top wavefield separation because of the equivalence
of the slownesses used to define the top filtering and the velocities that define the J-k pie-
slices. To illustrate this and the importance of being able to use interpretation to guide the
processing, the FSJG Z(FRn data are used for Route A processing. The result of the top fil-
tering is shown in the f-k domain to illustrate the equivalence of the two methods. The top
wavefield separation IPP for Route A is shown in Figure 2.27. The processing runstream for
the Route A top processing is shown in Flowchart 6. The input data to the forward top trans-
form are limited to the VSP data that have velocities similar to the upgoing events.
62
The normalized and gained Z(FRT) data are shown in panels 1 and 2, respectively,
of Figure 2.27. The Zup(FRTI data shown in panel 3 are a result of the inverse t-p transfor-
mation of the isolated upgoing events from the forward t-p transform of the Z(FRT) data.
The forward t-p transform of the Z(FRTI data do not include the slowness values of the
downgoing VSP events. The slowness range in the x-p masking is between -0.0004 s /m
(-2,500 m/s) and -0.0000465s/rn (-21,500 m/s). The t-p transform is accomplished by per-
forming 101 t-p increments of 0.000035 s/m. The f-k plot of the Zup(FRT) data from panel
3 of Figure 2.27 is shown in Figure 2.28. Recall that the sonde location interval is 20 m,
which makes the spatial Nyquist frequency equal to -2,500*10- 5 m! (0.025 mI), The two
slowness values define linear cut-offs in f-k space filtering.
Compare the Zup(FRT) data in panel 3 with the f-k narrow pass pie-slice Zup(FRTI
data in panel 5 of Figure 2.7. Except for global gaining differences, both panels appear
similar. Note, however, that panel 5 of Figure 2.7 shows Rieber mixing that degrades the
interpretation.
The data in the two IPP panels (Figures 2.7 and 2.27) are opposite polarity. In pan-
els 4 and 5 of Figure 2.27, the highlighted (orange) isolated trough at 0.755 s appearing
on the depth traces between 1270 m and the first-break curve is the primary event
embedded within a series of multiples. The same primary upgoing event appears as a
peak (highlighted in blue) in Figure 2.7 (panel 5), but the trace-to-trace amplitude con-
trasts in Figure 2.7 have been smeared by Rieber mixing. The "equivalent" narrow pass
zone used in the t-p processing (Figure 2.28) is larger than the zone used for the f-k wave-
field separation (Figure 2.12). Here the problem is not so much the extension of the mul-
tiple, but the lateral amplitude smearing of the separated upgoing waves in panel 5 of
Figure 2.7.
Route B: Subtraction of Zdown(FRT) from Z(FRT) data

In Route B, the downgoing events are separated from the Z(FRTI data using the
t-p transform. The Zdown(FRTI data are then subtracted from the Z(FRTI data to yield the
Zup(FRT) results.
The IPP of Figure 2.29 demonstrates this approach (using the tube wave
contaminated Z(FRT) data example; panel 1 of Figure 2.29). The Zdown(FRT) data,
shown in panel 3, are isolated using t-p filtering. Following scaling, the data are
subtracted from the Z(FRT) data (shown in panel 2 of Figure 2.29) to output the
separated upgoing event data Zup(FRT) in panel 4. This is similar to the results
obtained using the median-filter-based wavefield separation method shown in
Figure 2.2. Note that the aliased downgoing tube wave (appearing as upgoing
events and highlighted in panels 4 and 6) are not attenuated. This Route B
63
0.2
-I
~
m
~
:!:
m
Cii
,:.-1.0 -
1.4
(1) Z(FRT}
Fig. 2.27. t-p (Route A) based wavefield separation of the Fort St. John Graben
multiple contaminated data.
10
20
'TI
~
m
o
30 ~
z
Q
40 'i
N
50
60
Fig. 2.28. The equivalent t-k surgical mute pass zone to the t-p filter used to
create the resultant data in panel 3 of Figure 2.27. This can be compared to
the t-k narrow pass zone shown in Figure 2.12.
64
approach avoids possible Rieber mixing because the effect of the applied t-p filter
is similar to that of a narrow reject f-k filter with respect to the attenuation of the
upgoing events.
Route C: t-p spatial interpolation during wavefield separation

In Route C, upgoing events are extracted from the Z(FRT) data during the for-
ward transformation to the t-p domain on the basis of slowness. Trace depth interpola-
tion to a finer depth sampling is performed during the inverse transform. The effect that
spatial interpolation has on the inverse t-p transformed aliased tube wave is examined.
Figure 2.30 shows the f-k plot of the tube wave contaminated Z(FRT) data. The
downgoing P-wave events are shown in the negative k, quadrant. The depth interval for
the traces is 30 m. The downgoing tube wave aliases back into the upgoing events. The
center of the intersection is approximately at 37 Hz and -0.00833 m- I on the k, axis
(labeled as -833 *10-5 rrrl). The intersection of the aliased downgoing tube wave with the
upgoing P-wave f-k events is highlighted in Figure 2.30 by the cursor location.
The Z(FRT) data are forward t-p transformed using the slowness range of the
upgoing events only. The aliased downgoing tube wave f-k events are also retained dur-
ing the forward transformation because, as a result of aliasing, these events appear in the
same slowness range as the upgoing P-wave events. During the inverse transform, the
trace spacing sampling interval is redefined to be 15 m. The Zup(in,.,pj(FRT) data at the
interpolated trace spacing of 15 m are displayed in panel 2 of Figure 2.31. The interpola-
tion performed during the inverse t-p transform has faithfully interpolated even the
aliased tube wave (as is shown in panels 2, 3, and 4 in Figure 2.31 and in the
ZupUnt.,pi(FRT) f-k plot in Figure 2.32).
The upgoingf-k events are shown in the negative f-k quadrant of Figure 2.32. Note
that the intersection of the interpolated aliased tube wave with the interpolated upgoing
f-k events is still at 37 Hz and at -833 (*10-5) m-I . This is the same location as the tube wave
event intersection with the upgoing events in the f-k domain before interpolation. The
major change in the f-k plot is that the k, Nyquist values have doubled in size from 1666
to 3333 *10-5 m- I
It is important to determine if the Zup(FRT) data are markedly different without
trace interpolation. To evaluate this possibility, the tube wave contaminated Z(FRT) data
were wavefield separated using t-p filtering without trace interpolation. The resultant
Zup(+TT) data are shown in the IPP of Figure 2.33. By inspection, it is evident that the
ZupUnt.,pl(+TT) interpolated data in panel 4 of Figure 2.31 are similar to the noninterpo-
lated Zup(+TT) data in panel 4 of Figure 2.33.
65
-l
~
m
~
~
m
'~~~_ 1.0 :E
(3) Zdown(FRT) "r,~: 1.2
T-p based
:; • i 3 i ~ i ~ ~ i ~ i ~ ~ ~ ~ ; ~ ~ : fll ! iii i ~ i i i I i ~ ii ~ : I; i i ~ i ~ ~ ~ ~ i ~ ~ i ~ i ~ i DEPTH (m)

,::,I~.'I ·i ~ · 0.2
t , ' , .'
,..-'
(4) Zup(FRT) (5) Zup(+TT) (6) Zup(+TT)

median filtered
Fig. 2.29. top (Route B) based wavefield separation IPP of the tube wave
contaminated data using top downgoing event separation and then subtrac-
tion of the Zdown(FRT) from the Z(FRT) to output the Zup(FRT) data .
10
20
"T1
::n
m
30 0
c
m
z
o
40 -<
I
N
50
60
70
Fig. 2.30. The f-kplot of the tube wave contaminated data showing the aliased tube wave
f-k event intersecting (at the crossing of the red lines) the upgoing P-wave t-k event at
0.00833 rrr! spatial frequency (k) and 37 Hz.
66
,...
,,,::, --1.2
Fig. 2.31. t-p (Route C) wavefield separation IPP using t-p upgoing event separation on
Z(FRT} with 30-m trace separation and inverse t-p transformation using trace interpola-
tion to create a 15-m trace spacing.
-3333 -2500 -1666 -833 0 833 1666 2500 3333
o
10
20 "
;;c
m
o
c
30 ~
o
-<
40 ~
50
60
Fig. 2.32. The f-k plot of the interpolated tube wave contaminated Zup(interp)(FRT} data
(Figure 2.31) where the aliased tube wave f-k event intersects the upgoing P-wave t-k
event at 0.00833 rrr ' spatial frequency (k) and 37 Hz. The range of the wavenumber (k)
has doubled in value (due to the 15-m spacing); however, the intersection of the aliased
tube wave and upgoing P-wave f-k events remains the same as in Figure 2.30.
67
0.2
0.4
-l
0.6 ~
<
m
r-
0.8 ::::l
s:::
m
1.0
1.2
-
(i)
Fig. 2.33. t-p based wavefield separation IPP of the tube wave contaminated
data using t-p upgoing wave isolation (filtering out p values outside of the
range of the upgoing P-wave events) .
Route D: t-p spatial interpolation of Z(FRT) data before wavefield

separation
In Route D, both the upgoing and downgoing events of the Z(FRT) data are pre-
served during the forward t-p transform. Spatial interpolation is performed during the
inverse transform, and wavefield separation is performed using median filtering on the
Z(jnterpl(FRTI data. This procedure is useful when processing a dataset that has irregular
depth spacing and requires interpolation. Although the tube wave contaminated Z(FRT)
data are regularly spaced, the IPP shows a median filter wavefield separation that is
based on t-p trace interpolated data.
We begin with the spatially interpolated data shown in panel 1 of Figure 2.34. The
results of using the Z(interp)(FRTI data in a median-filter-based wavefield separation pro-
cedure method are shown in panels 2, 3, 4, and 5. In panelS, the Zup(interpl(-TI) data with
1S-m depth spacing are shown. The Zup(interpl(+ m m
and Zup(interp;medl(+ data are
shown in panels 6 and 7, respectively.
The f-k plot of the interpolated Z(FRT) data is shown in Figure 2.35. The interpo-
lation method used on the Z(FRTI data was a two-part operation where the upgoing and
the downgoing waves were separated, interpolated, and merged to create the
Z(interpl(FRT) data. This procedure is equivalent to imposing cutoff slownesses for both
the upgoing and downgoing wavefields during the forward t-p transformations. This
translates into the two "accept" f-k zones shown in Figure 2.35.
68
Future directions for t-p filtering in wavefield separation

The t-p filtering and IPPs shown previously have familiarized the reader with the
many roles that the t-p filtering can play within wavefield separation. When interactive
t-p muting becomes a common processing tool, one can routinely try to examine and
eliminate multiple reflections using the t-p transform (Hu and McMechan, 1987). If we
are limited to slowness filtering (specifying slowness cutoffs during the forward t-p
transformation), the method will suffer the same limitations as pie-slice fk filtering.
t-p data can be transformed into the t-p domain (the results from Fourier
transforming along, holding p constant). In this domain, tube wave events can be
muted easily using interactive processing. In addition to the t-p filtering discussed
above, Miao et al. 1995 used hyperbolic velocity filtering (Kelamis and Mitchell, 1989)
to optimize the t-p wavefield separation technique.
0.2
0.4 --t
"::-->- - "'0.6 ~
m
r-
0.8 --t
s:::
1.0 m
1.2 ~ -
1.4
..... ... "" '"

o~~8~
0.2
0.4 --t
0.6 ~
m
0.8 r
--t
1.0 s:::
m
1.2 --
fJ)
1.4
Fig. 2.34A. Panels 1 to 4 of the t-p based wavefield separation IPP using the interpolated
Z(FRT) tube wave contaminated data.
69
0.2
0.4 ....
~
0.6
m
0.8 ~
~
1.0 m
en
1.2 ....,
o::::=~=~1.4
0.2
0.4 ....
0.6 ~
m
r-
0.8 ....
~
1.0 m
1.2 ~
1.4
Fig. 2.348. Panels 5 to 7 of the t-p based wavefield separation IPP using the interpolated
Z(FRT) tube wave contaminated data.
10
"T1
20 fil
cc
m
z
30 Q
:I:
N
40
50
60
Fig. 2.35. f-k plot of the interpolated Z(FRT) data shown in panel 1 of Figure 2.34 .
70
VSP deconvolution and corridor stacks: near-offset data

VSP, or Up over Down, deconvolution was introduced in Chapter 1. The term,
Up over Down, refers to the computational procedure of the temporal frequency spec-
trum of Zup(-TT) data being divided by the temporal frequency spectrum of Zdown(-TT)
data to perform deconvolution in the Fourier domain. If the spectrum of Zdown(- TT) is
divided by itself (in the band-limited sense), a zero phase, band-limited, multiple-free
delta function will result (Hubbard, 1979; Gaiser et al., 1984).
For interpretive processing, the following questions should be posed:
1) Are the multiples attenuated;
2) Are the underlying primary events preserved;
3) Has the deconvolution processing added unwanted noise; and
4) What are the origins of the multiple events?
In this section, the following IPPs are developed:

1) deconvolution IPP: the Zup(+TT) and Zup(de<on)(+TT) can be compared directly to
evaluate the effect of the deconvolution process on the upgoing events;
2) Zup(+ TT) data inside and outside corridor stacks: the Zup(+TT) inside and outside
corridor stacks and the muted data that go into the stacks can be compared to facil-
itate multiple identification; and
3) Zup(decon)(+ TT) data inside and outside corridor stacks: the deconvolved data and
the corridor stacks can be used to evaluate the effects of deconvolution on multiple
suppression.
Deconvolution IPP
A VSP deconvolution IPP is presented in Figure 2.36 (using the FSJG near-offset
data). Panels 1 and 2 are the Zup(+TT) and Zup(med)(+ TT) data, respectively. These non-
deconvolved data clearly show the multiples (highlighted in yellow). The Zdown(-TT)
data in panel 3 show the downgoing event multiple pattern. To evaluate any increases
in the background noise level as a result of the deconvolution process, the Zup(-TT) and
Zup(decon)(-TT) data shown in panels 4 and 5 (Figure 2.36) can be compared. The
Zup(decon)(+TT) and Zup(decon;medl(+TT) data are shown in panels 6 and 7.
The first step in this interpretive processing procedure is the detailed inter-
pretation of similar events in panels 2 and 7. A primary event (peak highlighted in
orange at 0.595 s) in panel 2 is severely affected by the series of multiple events asso-
ciated with Spirit River event. The primary event is preserved and continuous across
the entire suite of deconvolved traces shown in panels 6 and 7 after the interfering
multiples are attenuated.
71
g TRAVELTIME ( s )
J:
1- 0 C'! ~ <0 CXJ ~ C'\I
a.... o 0 0
WO 0 or- or-
....
Cl
It,.
I . :>i
I
"..
...",..
1710
1110
C\J
1)70 (/J
I l. . Q)
.'",.,'"""...
11'10
.u. c
co .
no
.,..,.... -
E TRAVELTIME (s)
c...C\1
~
co c
Q)
.,.
J'"
~
o
<0
o
o
..... or-
c...CO
E c...
o c
,,..
,....
I _"CoO
....
mo I- 0)
I-
L..
+ ~ g ~
• C
...."'"
15 10
mo
""::::.c;::::{gcn
.no gee ~
Gl .as
I!I O U Q) -c..
_ .0
...."'".
""".
0 ~~ ~:;
....'"...
'''' - N' E" E
- C Q)
to-- ..c..c
".
)JO
- 0 ....
,,.. "C "'0
r:
0)
,... U5 oC
,0>0
L..
1710
.!
...."..
1510
,...
1"'0 -
It l O
~;:
I- c
I-as
+ .-
-
+
c
8Gl
t::~
o
LL C
.$
Q)
::::l
...",.""
,0>0
ro
-"C
CoO) "'C
£Cii
-- 0:: "- Q)
........''''. NE .!2..c ....

--
::::l
N 0.. Q)
C'\I
o..-ro
,-,-
<0
•••
J'" ::::l
co
gC::::l >
~
' 0>0
17'70 Q)
...
' 510
,"'"
u'o
,no -c I
0,9
>
c""'::'
81:
......,'"",.,...
12'10
11 )0 8Gl Q) I
"'C o:=::
0:: c
....,.... <.0
e0
--
::::l
N C')
C\I Q)
".
uo
LO
. c:i>~
_
LLN
:::l"
72
Corridor stack IPPs

A more detailed investigation into the success of the VSP deconvolution process
can be achieved using the corridor stack IPF. The basic premise behind the use of the
inside and outside corridor stacks is shown in Figure 2.37. The time delays between a pri-
mary event and associated upgoing multiples are key factors in the interpretation panels
designed to facilitate the identification of multiples in the Zup(+ IT) data.
Interpreted primary upgoing events intersect the first-break curve. Upgoing mul-
tiples are recorded later in time at sonde locations above the bottom generating interface
because of the added traveltime delay present in the surface-generated or interbed mul-
tiples. In the Zup(+ TT) data (Figure 2.37), the multiple of a primary reflection is recorded
later in time than the primary event. An upgoing multiple is recorded on all data traces
recorded above the bottom generating interface.
A corridor of predominately multiple-free upgoing event data exists in the immediate
area of the first-break curve of the Zup(+IT) data since upgoing multiples do not intersect the
first-break curve. Primary events will dominate on a stack of the outside corridor of Zup(+IT)
data. In contrast, both multiples and primary events (if not destructively interfered with by
multiples generated at shallower interfaces) will be preserved on a sum performed from the
corridor line in the opposite direction from the first-break curve (inside corridor).
Z, DEPTH (m) ""7<-,-----,-

Fig. 2.37. The schematic definition of the
outside and inside corridor stack. The
depth/(+TT) traveltime plot shows the
U -M downgoing primary (D-P), upgoing primary
(Up-P), and upgoing multiple (Up-M)
---"-I-''------z, events with corresponding raypaths shown
i
Inside Corridor
Outside
Corridor
in the raypath diagram. In the inside corri-
dor stack, both the primary and multiple
events are retained. For the outside corridor
~
R Up-p
Up-M
R stack, only the primary event (Up-P) is seen
after summing the events in the corridor
stack (after Hinds et aI., 1989).
The upgoing and downgoing multiple patterns reveal the top and bottom generating
interfaces of interbed multiples, respectively The downgoing waves in (-IT) time show the top
generating interface, and the upgoing waves in (+IT) time will reveal the bottom generating
interface. This is illustrated in Figure 2.38. The simplistic Zdown(-IT) and Zup(+IT) diagrams in
Figure 2.38B and C illustrate the trends of the upgoing and downgoing primary and multiple
events. The two data panels can be used to determine the origin of the multiple events.
73
Up Mult /Dgprim
ray diagram
Top generating
interface
Dg Prim
Bottom generating
interface
Dg Mult Dg Prim
(A)
Z1 Z2 DEPTH (m) Z, Z2 DEPTH (m)
en Dg Prim '"
W
W ::;;
::;; st
>=
...J
Dg Mult ~ Up Prim
W
W > st
~
f--
~
f--
Up Mult
Dg Prim
Zdown(-TT) Zup(+TT)
(B) (C)
Fig. 2.38. Ray and depth/time plots, Zdown(-TT) and Zup(+TT), showing
the top and bottom generating interfaces for interbed multiples.
The corridor stack IPP for the FSJG Zup(+ TI] data is shown in Figure 2.39. The near-
offset Zup(+TTl data are displayed in panel 1. Two sets of multiple events are highlighted in
yellow. The highlighted multiples on the muted Zup(+ TI) data in panel 2 will affect the inside
corridor sum shown in panel 3.
The inside corridor sum contains multiples and primaries. To illustrate this, the out-
side corridor sum data of panel 4 are placed next to the inside corridor sum data of panel 3.
The multiple events highlighted in yellow in panel 3 do not appear in the outside corridor
stack of panel 4; however, the primary events (shown in orange) are cornmon to both corridor
stack data in panels 3 and 4. The outside corridor mute data shown in panelS contain only pri-
mary events (highlighted in orange at 0.63 s, 0.9-1.0 s, 1.05 s, 1.1-1.2 s, and at 1.25 s). The events
below 1.3 s cannot be interpreted (using this teclmique) to be either multiple or primaries
because there is no longer a first-break curve at those recording times. To interpret the multi-
ple and primary events, compare panels 3 and 4 in Figure 2.39.
74
1~~~- 0 . 6
-t
~
0.8 <
m
r
-t
- 1.0 s::
m
1.2
-- en
- 1.4
~~~'--
(3)Zup(+TT) (4) Zup(+TI)

inside corridor stack outside corridor stack
Fig 2.39. Corridor stack IPP of the Fort. 81. John Graben Zup(+ TT) data. Note the high-
amplitude multiple events at 0.7-0.8 sand 1.0-1.1 s on the inside corridor stack .
0.6
-t
- 0.8 ~
m
r
-t
1.0 s::
--
m
en
- 1.2
- 1.4
(5) Zup(decon)(+TTj (6) ZUp(deconj(+TT)
muted outside corridor median filtered
(3) ZUP(decon}(+TT) (4) Zup(deconj(+TT)
inside corridor stack outside corridor stack
Fig. 2.40. Corridor stack IPP of the Fort. 81. John Graben Zup(decon)(+TT) data. Residual
multiple events are present on the corridor stacks; such as the peak at 0.74 s on panel 3.
75
The effect of VSP deconvolution processing on Zup(+TT) data can be determined

by interpreting a corridor stack IPP that uses the Zupldecon)(+TT) as input. Ideally, if
deconvolution is successful, the data in both the inside and outside corridor stacks of the
Zupldecon)(+TT) data will be multiple-free. These stacks can be compared with the
Zup(+TT) outside corridor stacks of Figure 2.39 to determine if any deconvolution-gen-
erated noise was introduced into the data.
The corridor stack IPP for Zupldecon)(+TT) data is presented in Figure 2.40. The cor-
ridor stacks (panels 3 and 4) differ where deconvolution did not effectively attenuate
multiples. Such remnant multiple events are highlighted (in yellow) on panels 1 and 3 of
Figure 2.40. A prominent primary highlighted (at 1.05 s on Figure 2.40) has been
enhanced by deconvolution. On Figure 2.39, the same highlighted primary event (high-
lighted in orange) is detected at 1.05 s on the 1270 m trace and out to the first-break curve.
The remainder of the primary event at 1.05 s in Figure 2.39 is destructively interfered
with by an upgoing multiple event (colored yellow) occurring at the same time. The out-
side corridor stack for the Zup(+TT) and Zupldecon)(+TT) data are similar, indicating that
the deconvolution process did not add appreciable amounts of noise to the outside data
corridors. The Zupldecon)(+TT) inside corridor stack is different from the outside corridor
stack of the same data between 1.2 to 1.3 s. This indicates that the shallow traces were not
appropriately deconvolved; however, the deeper-trace near-offset data can still be confi-
dently interpreted as multiple-free.
The usefulness of the corridor stack IPP does not end at this point. In a production
processing run, one could take the two IPP paper displays, fold them between panels 3 and
4, and overlay them on the corresponding surface-seismic section. A comparison between
the VSP corridor stacks and surface-seismic events can assist in determining where the COP
stacking method has not effectively attenuated multiples. Later in this chapter, inside and
outside corridor stacks are used as part of the integrated interpretation display (lID). This
latter display brings together a variety of related exploration data.
Far-offset data processing

In this section, far-offset processing is reviewed and case studies showing the
use of interpretive far-offset processing are presented. Normal far-offset processing
consists of:
1) polarization of the X(FRT) and Y(FRT) data into HMAX(FRT) and HMIN(FRT)
data [the HMAX(FRT) data is the projection of the X(FRT) and Y(FRT) data into a
plane defined by the well and the source location];
2) polarization of the HMAX(FRT) and Z(FRT) data into HMAX'(FRT) and Z'(FRT)
data [the HMAX'(FRT) data are polarized in the direction of the source, and the
Z'(FRT) data will be orthogonally polarized to HMAX'(FRT) ];
76
3) wavefield separation of the Z'(FRT) and HMAX'(FRT) data into Z'up(FRT) and
HMAX'up(FRT) data;
4) wavefield separation of the HMAX'(FRT) data into HMAX'down(FRT) data [this step
is used to deconvolve the final output, Z"up(FRT) into Z"up(decon)(FRT) ];

5) derotation of the Z'up(FRT) and HMAX'(FRT) data into Zup(dewu(FRT) and
HMAXupldew,)(FRT) data;
6) time-variant polarization of the Zup(derou(FRT) and HMAXuplderot)(FRT) data into
Z"up(FRT) and HMAX"up(FRT) data;
7) VSP-CDP or migration of the Z"up(+TT) data into the +TT-versus-offset from the
well domain.
In the interpretation of the far-offset IPPs, a number of critical observations are
made. Some of the questions that need to be considered are:
1) Do the X, Y, and Z data obey the assumptions behind time-invariant polariza-
tions (as reviewed in Chapter I)?
2) What are the origins of noise?
3) How is the noise being distributed onto the output data panels following each pro-
cessing stage?
4) What is the extent of multiple contamination in the VSP-CDP or migrated data?
5) Do the multiples on the VSP-CDP or migrated data interfere with the interpretation
of primary events?
6) Can the primary and multiple events be differentiated (recall that reflectors imaged
away from the well may not intersect the first-break curve)?
These questions are asked repeatedly during the interpretive processing of the
far-offset IPPs. The partitioning of the energy of the various wavefields onto the polar-
ization axis and the time-variant axis is illustrated using far-offset data (from Chapters 4,
5, and 6). The strength and weaknesses of far-offset deconvolutions are illustrated. A
review of the Ricinus carbonate reef case study (Chapter 4) polarization IPPs demon-
strates:
1) the output data following normal processing runstreams (see Processing Run-
streams);
2) an analysis of noise on polarized data; and
3) modifications to the processing runstream to enhance the interpretation of the
Z"up(+TT) data.
77
Rotations to isolate downgoing P-waves: far-offset data

In this section, Z(FRT), X(FRT) and Y(FRT) data are polarized to isolate the down-
going P-wave events onto a single data panel, HMAX'(FRT). Until recently, the Z'(FRT)
data were considered to contain predominantly upgoing P-wave events, and interpreta-
tion was performed on the wavefield separated Z'up(+TT) data and on the VSP-CDP
transformed upgoing events. The polarization results presented in this section use
hodogram-derived rotation angles derived in a time-invariant sense. The hodogram
analysis is performed on a single window of data around the first-breaks of the two input
VSP data to yield a single polarization angle. In the following section, the polarization
angle for a pair of input traces is allowed to vary in time along the traces.
The time-invariant polarization IPP for the far-offset data (700 m offset source
location, FSGl) of the Fort St. John Graben case study is shown in Figure 2.41. The
X(FRT), Y(FRT), and Z(FRT) data are shown in panels 1, 2, and 3, respectively. Some of
the downgoing P-wave events in panel 3 are highlighted in yellow. In comparison to pan-
els 1 and 2, the Z(FRT) data contain significant downgoing P-wave energy.
On panels 1 and 2, the primary downgoing P-wave energy is divided in varying
amounts on both the X(FRT) and Y(FRT) data (highlighted in yellow). The P-wave ener-
gy in panels 1 and 2 indicates the sonde was rotating. However, the first-break wavelet
on the two panels show relative consistency over the bottom two-thirds of the borehole,
indicating that the tool was rotating slowly during acquisition at these depths.
The Z(FRT) data shown in panel 3 contain a diffraction event (highlighted in pur-
ple in the lower left-hand comer of the panel). This diffraction is transferred to the dif-
ferent polarization panels in interesting ways (look at the highlighted purple events in
panels 5-7). Since these diffraction events are not polarizable in the same way that the
upgoing and downgoing P-wave events are, interpretation is performed over these noise
events.
The result of hodogram-based polarization analysis on the X(FRT) and Y(FRT)
data and subsequent single-angle rotation yields the HMIN(FRT) and HMAX(FRT) data
shown in panels 4 and 5, respectively. The minor amount of downgoing P-wave energy
evident in the X(FRT) and Y(FRT) data shown in panels 1 and 2 has been isolated on the
HMAX(FRT) data (panelS). The rotation of the X(FRT) and Y(FRT) data was performed
using an interactive hodogram analysis routine. A window of input data (in this case, the
X(FRT) and Y(FRT) data), centered around the first-break of two input traces at the same
depth recording level, is shown in one portion of the screen. The trace data are color
coded so that hodogram points in other portions of the screen can be correlated to the
traveltime of the trace data. This enables an interpretation of the hodogram plot. A least-
squares routine estimates the best straight line through the cloud of hodogram points.
78
- rn
c
-- --
Q)
>
Q)
E E Q)
TRAVELTIME (s) TRAVELTIME (s) >

ttl
:I: ~
CD CD o ....n, ~ CD CD 0 ~
d.
OJ
o o w 0 0, 0 T""" c
·0
S!iill
1170
- -
en
c
~
o
' 0.
17911
PIO -0
IUO I===F==;:S Q)
.c
147 0 I-
UIO
13 10
g
ttl
-
I t 30
11'0 "C
1070 Q)
..0
110 Jg
0>0 o
no ~
"' 0 $
",0
4. 0
c
Q)
4 10 .0
330 ttl
'-
1.,0 <!J
c
.c
1780
1710 ...,o
.....
1130
1.70 Ci5
139ll t
ula
123 0 ~
-
Q)
"'"-1':==::1==
-ti::
10 70 ..c
no '-
.. 0 o
no -c..-
ai
....
no
->-
LL )70
c..~
--
C\I
4'"
4 10
330
1070
§i='
.';:::; a:
ttlLL.
N-
·'::::X
ttl
et
1710
1710 8.:!:
..' ''"'1
1030
111111
-o::I:
Q)
-~
Q)
Ina rn.c
1300
ttl-
.0 C
IJIO
"3O ~~~~
E0
-x
LL 11:10 -1
1010
~-g
gCii
--
. .0
.. 0 "C 0
no o rn
T """ 730 I ·~
"'0 ..... Q)
'<:l:~
)70
410
4 10 C\J Q)
330
.2> ~
LL..c
79
The HMAX(FRT) and HMIN(FRT) windowed results, using the rotation angle deter-
mined by the least-squares fit, are displayed on another portion of the monitor. The user
can now interactively alter the angle of the straight line through the hodogram cloud of
points and the displayed window of HMAX(FRT) and HMIN(FRT) output traces (also
centered around the first-break) immediately change according to the updated rotation
angle. The calculations that produce the HMIN(FRT) and HMAX(FRT) traces use the
rotation matrix shown in the Appendix.
During this exercise, we want to maximize the amount of downgoing P-wave
first-break energy on the HMAX(FRT) traces and maintain a consistent output polarity.
The polarity of the HMAX(FRT) data in panelS is consistent, and minimal downgoing
P-wave energy is seen on the HMIN(FRT) data (panel 4). Mode-converted downgoing
SV events are highlighted (in orange) on the HMAX(FRT) in panelS.
The Z'(FRT) and HMAX'(FRT) data calculated using a second hodogram analy-
sis and subsequent rotation of the HMAX(FRT) and Z(FRT) data are shown in panels 6
and 7, respectively. The downgoing P-wave data are isolated onto the HMAX'(FRT)
panel. The Z'(FRT) data contain mode-converted SV-energy (highlighted in orange).
Mode-converted SV-downgoing events on the Z'(FRT) data appear to originate at the
impedance boundaries at 690 m (Spirit River Formation; Chapter 5) and the 1260 m
(Nordegg Formation). An upgoing P-event from the Nordegg reflector (highlighted in
green) is identified on panels 3 and 6. The diffraction event seen on the Z(FRT) data is
now partitioned onto both the HMAX'(FRT) and Z'(FRT) data. Interestingly, the
HMAX'(FRT) data contain the upgoing part of the diffraction, while the Z'(FRT) data
contain the downgoing part of the event.
Rotations to isolate upgoing P-waves: far-offset data

In this section, time-variant polarization is applied to the HMAX'up(FRT) and
Z'up(FRT) data to isolate the upgoing P-wave events onto the output Z"up(FRT) data.
The final interpretations are done on the Z"up(+TT) and the VSP-CDP transformed
and! or migrated Z"up(+TT) data.
The time-invariant polarizations performed in the previous section assumed
that a single angle was adequate to polarize the upgoing and downgoing Z(FRT)
and HMAX(FRT) event onto separate panels, Z'(FRT) and HMAX'(FRT), respective-
ly. The polarization angle needed to isolate (polarize) the upgoing P-wave onto the
Z"up(FRT) data is time-variant. The angle of reflection from interfaces below the
sonde changes with depth as shown in Figure 2.42. This implies that the incident
angle of the upgoing P-wave events at a single geophone location changes with
recording time.
80
Time-variant polarization concept

increasing offset ____ source
V1
Z,
V2
Zz 5·
o
a
III
U>
V3 5·
cc
C.
I1l
Z3 "S.
zr
V4
Z4
!
Vs
Zs
Fig. 2.42. The reflection angle for upgoing raypaths emerging at the geophone at A from
deeper interfaces decreases in comparison to the reflections from shallower interfaces.
The fi are the reflection angles, Vi are the layer velocities, Zi are the layer depths and Ii
are the raypath traveltimes to the sonde at A.
The Z'up(FRT) and HMAX'up(FRn data are a result of performing wavefield sepa-
ration on the Z'(FRn and HMAX'(FRn data, respectively. The IFP for the time-variant
polarization of the FSJGl far-offset data as shown in Figure 2.43. The Z'up(FRT) and
HMAX'up(FRT)data shown in panels 1 and 2, respectively, contain upgoing P-events (high-
lighted in green), and mode-converted downgoing shear-wave events (highlighted in blue).
The Zupldero,,(FRn and HMAXuplderotl(FRT) data shown in panels 3 and 4 are cal-
culated from Z'up(FRT) and HMAX'up(FRT) data by applying a rotation operation oppo-
site to that used in calculating the Z'(FRn and HMAX'(FRT) data. Panels 3 and 4 indi-
cate that this simple derotation partially isolates the upgoing P-wave events (highlight-
ed in green; panel 3).
We use the interval velocities and first-break times from the near-offset VSP data
of the Fort SI. John Graben case study to construct a model for ray tracing. The first-break
times (when the downgoing P-wave should reach an interface) and the velocity of the
various interfaces (which are used to propagate the upgoing reflections) are derived
81
~ A AO~~.~. ~ A 4~~~ • •• ------------ w44 o~ ~.~~------------D E PT H (

"' -~""g iW~-~O-"4WW.l.Icn ........ :t ~-~"" 4,I~~-~O- "'W t:~ """' CD::i .... W-~"'g:!:~-:80-f'tIWW"l"Icn"''''Q)1.O m)
0 0 0 0 0 ~~~O~~~~O~~O 0 0 0 0 ~~~O~~O~O~ ~O 00 0 0 ~g ~o~~~~O~;~
liI ' 111'!111 1111~ II . !. ; 11II ! I I 11,1! 11!.llllll!111I.-111,1'1111111"I.:II tilliLl 111 ,11 1111 1111111111 1.11~ i h!J iililiti -: II It ,,111111II:" I: 111ll,l: IiIfI li!Ii .:1, ; 1:,1: lil'",IIII1-11 11I I !; tiLII !II I LII:llll.1111I.II' I I! I ll '" .1111!1;.!Ill !. ,!I
Il::;!'!,::l Iilll:l:l!!::::!:f#;- 0.4
"Ii"'II""
~, lh ulil 0 •6 il>
<
m
r-
0.8 :::!
s::: <:
m (IJ
-. '"CI
1.0 ~ ....
::s
it
'"I
"'l:S
'"I
1.2 rtl
(1) Z'up(FRT) (2) HMAA'up(FRT) (3) Zup(derot)(FRT) ..<'
rtl
'"CI
'"I
on
rtl
QC <Il
<Il
N
S'
~
;t
rtl
o
-<
III
::s
Q..
'"CI
liln
......
n
rtl
1,2
(4) HMAXuPlderot)(FRT) (5) HMAX~p(FRT) (6) Z'~p(FRT}
Fig. 2.43. Time-variant polarization IPP for the Fort 81. John Graben (FSJG1) far-offset data. The upgoing P-wave events
have been predominately partitioned onto the Z" up(FRT) panel.
through simple calculations. The ray tracing to the defined geophone locations (given the
source offsets) will result in the time-angle pairs to be used in time-variant polarization
(see Appendix). The results of the time-variant polarization are the HMAX"up(FRT) and
Z"up(FRTI data shown in panels 5 and 6 (Figure 2.43), respectively.

Compare panels 3 and 6 (Zup(dew') and Z"up). The residual downgoing SV- events
on panel 3 are now mapped jointly onto panels 5 and 6 (colored in blue). The upgoing
P-event (colored pink) originating near the 1750-m depth is more interpretable on the
Z"up(FRTI data in panel 6. The shallow Spirit River event (colored green on panel 6) at
690 m, which was not resolved on panels 1 to 4, can now be interpreted. Some of the other
up going events on the Z"up(FRT) data in panel 6 are also colored green.
VSP-CDP and migration: far-offset data

In this section, the Z"up(FRTI data are VSP-CDP transformed or migrated, and the
relevant IPP is presented. The term VSP-CDP or VSP-CDP mapping, refers to mapping
the VSP data into a pseudoseismic section that displays coverage starting from the bore-
hole out to the farthest reflection defined by the VSP data geometry and input velocity
model (Wyatt and Wyatt, 1982, 1986; Millahn et al., 1983). The product of the VSP-CDP
mapping has been called the VSPCDP (Hardage, 1985).
The VSP-CDP and migration far-offset IPP for the FSJGl data (Chapter 5) is
shown as Figure 2.44. The Z"up(+TTl and Z"up(med)(+TTl data are shown in panels 1 and
2, respectively. The VSP-CDP (Dillon and Thomson, 1984) mapped Z"up(+TTl data are
shown in panel 3. Note that the horizontal axis for panel 3 (and 4) is not depth, but off-
set distance from the well. We are looking at the image of the reflectors on the VSP-CDP
or migrated data as one would see them on a seismic section; meaning lateral coverage.
The IPP is completed with a Kirchhoff migrated (Wiggins and Levander, 1984;
Wiggins, 1984; Wiggins et al., 1986; Dillon, 1990) version of the Z"up(+TTl data (panel 4).
The migrated data are smoother than the VSP-CDP mapped data of panel 3. The events
between 1.2 and 1.35 s are faulted. An example fault is highlighted by the offset event (in
green). Interpreted faults are shown in pink.
One of the most important interpreted events is the dominant peak (in blue) at 1.27
s that exists laterally from the well out to a distance of 85 to 105 m. Its amplitude attenuates
rapidly at farther offsets. We are looking at seismic evidence of a possible facies change in
the basal Kiskatinaw that is very important in terms of hydrocarbon exploration.
Certainly the exploration picture can change locally around a well. The other far-
offset data for the Fort SI. John Graben case study (source location offset of 741 m; loca-
tion of offset source is East of the well) are referred to as the FSJG2 data. Figures 2.45,
2.46, and 2.47 display the far-offset IPP's for the FSJG2 data.
83
.4
0.6
~._1'4
Fig. 2.44. VSP-CDP and Kirchoff migration IPP for Fort St. John Graben (FSJG1) far-off-
set data. Note the termination of the peak at 1.27s (panels 3 and 4) on the offset traces
0-55 m.
-i
~
m
~
s:::
m
-
1.0 "iii'
0.8
1.0 _
~
(5) HMAX(FRT) (6) Z' (FRT) (7) HMAX' (FRT)
Fig. 2.45. Hodogram-based polarization IPP for the Fort St. John Graben (FSJG2) far-
offset data. The downgoing P-wave events have been isolated onto the HMAX'(FRT)
data in panel (7).
84
0.4
--i
0.6 ~
<
m
r-
::!
3:
m
~
1.2
Fig. 2.46. Time-variant polarization IPP for the Fort St. John Graben (FSJG2) far-offset
data. The upgoing P-wave events have been predominately partitioned onto the
Z" up(FRT) data.
OFFSET(m) OFFSET(m)
~ ~ ; ;3J ~)~ ;~;;; 'J ~'. "Z&Sliiiiiiziii l ! !
+;:.--:- 0.4
0.6
1.4
Fig. 2.47. VSP-CDP and Kirchhoff migration IPP for the Fort St. John Graben (FSJG2)
far-offset data. Note the continuity of the event (peak) at 1.27 s (panels 3 and 4).
85
The rotation processes illustrated on these figures should be followed carefully.

We see that the event at 1.25 - 1.27 s on the VSP-CDP and migrated Z"up(+TTl data (col-
ored blue) in panels 3 and 4 of Figure 2.47 does not terminate near the well. The event is
laterally more continuous than the same event on the FSJGl data seen in panels 3 and 4
of Figure 2.44. Faulting (shown in pink) is also evident on the VSP-CDP and migrated
data in Figure 2.47 between 1.1 and 1.35 s.
VSP deconvolution: far-offset data

In this section, the deconvolution of the far-offset VSP data is reviewed, and far-offset
deconvolution IPPs are presented. The process of far-offset deconvolution is the spectral divi-
sion of Z"up(FRT) data by HMAX'down(FRT) data to output Z"upldewnj(FRTl data.
VSP (far-offset) deconvolution has not received a lot of attention in the literature
because the decision to deconvolve the far-offset is usually made after interpretation.
Whereas deconvolution of near-offset VSP data is done as a matter of course, far-offset
deconvolution decision-making involves searching for possible interfering multiples on
the far-offset Z"up(+TT) data, confirming that the event is not a terminating reflector
located away from the borehole, utilizing the Z'down(-TT) data to deconvolve the
Z"up(+TT), and then reconfirming the interpretation by comparing the data before and
after deconvolution. These steps illustrate the arduous task that is often presented to the
interpreter /processor.
The VSP-CDP /migration IPP for the far-offset data of the Simonette carbonate
reef case study of Chapter 6 is presented in Figure 2.48. From the previous section of this
chapter, one might think that our job is over; but we still need to hunt for problematic
multiples because the data do contain destructively interfering multiples (colored green
for reference). On panel 2, the off-reef platformal carbonate event (peak colored purple)
starts at 1.96 s, rises onto low-relief reef mid-way across the display, and then rises up
onto the peak event at 1.92 s (the transition being between depth levels 2470 - 2570 m).
There is an anomalous event (peak) at 1.8 s that does not correspond to the geol-
ogy as interpreted from the sonic log. The Wabamun Formation event (highlighted in
red) intersects the first-break curve at 3020 m as a peak at 1.735 s in the Z"up(+TTl data.
Note that the anomalous event (the peak colored green for reference) between 1.77 - 1.81s
immediately underlies the Wabamun event and does not intersect the first-break curve.
This could be a real geological event that phases-out before the borehole and subse-
quently is not on the borehole logs; however, it is probably a good candidate to be a mul-
tiple of the Wabamun primary event.
The Z'down(-m data were used to deconvolve the Zup(+TT) data from panel 1 of
Figure 2.48 to output the Zupldewn)(+TT) data in panels 1 and 2 of Figure 2.49. Deconvo-
86
--
en
.I.R
~1.8
Fig. 2.48. VSP-CDP and migration IPP for the Simonette far-offset non-deconvolved
data. A possible multiple event occurs at 1.8 s (peak).
1.0
1.2
-;
~
~ if.tH'j~~_~i'f,- . 1.4 ~
r-
-;
1.6 ~
--
en
= "'-"- 2.0
(3) Z'~p(decon)(+TT) (4) Z~p(decon)(+n)
VSP-CDP migrated
Fig. 2.49. VSP-CDP and Kirchhoff migration IPP for the Simonette far-offset deconvolved
data. The suspected multiple at 1.8 s has been attenuated; the reef interpretation
(detailed in Chapter 6) remains.
87
lution has successfully attenuated the interpreted multiple. The interpretation of the reef
events (purple) can proceed based on the VSP-CDP and migrated data shown in panels
3 and 4. The two IPPs shown in Figure 2.48 and 2.49 can be interpreted cooperatively to
minimize the effect of possible deconvolution noise.
Problematic far-offset interpretive processing

Textbooks often showcase success stories and ideal textbook data; but in this sec-
tion, we will attempt to show a problematic data example and the slightly convoluted
processing pathway taken in response to interpretive processing. In other words, we will
look at a difficult, real-life example. The far-offset polarization and deconvolution pro-
cessing of the Ricinus carbonate reef case study (Chapter 4) is presented.
Time-variant polarization of Ricinus case study data

Figure 2.50 shows the time-variant polarization IPP for the Ricinus carbonate reef
case study far-offset (1100 m) data after "normal" processing. The processing followed
the guideline far-offset processing runstream presented in the section "Processing Run-
streams" in this chapter.
The Z"up(FRT) data (panel 6) are contaminated with mode-converted upgoing
(orange) and downgoing SV-events (highlighted in green in panels 2, 3, 4, and 5). A
clearer view of the mode-converted upgoing SV-events (orange) is shown in panel 1 of
the VSP87CDP IPP in Figure 2.51. On panel 1, two upgoing P-events (highlighted in
blue) intersect the first-break curve between the 3300 and 3510 m traces. Linear events,
highlighted in orange, trend from the left-hand bottom corner of panel 1, cross the data,
and merge with the highlighted (blue) P-wave events. The orange highlighted events
represent the upgoing SV-events that contaminate some of the underlying P-wave data.
Upgoing P-wave (blue) and mode-converted SV-wave (orange) events are similarly high-
lighted in panels 1, Z'up(FRT), and 2, HMAX'up(FRT), of Figure 2.50.
The downgoing mode-converted SV-events are better imaged on panels 2, 3, 5,
and 6 of Figure 2.50 as linear events (highlighted in green) in the same direction, but at a
steeper dip, than the P-wave first-break curve. The first-break curve no longer exists, fol-
lowing the downgoing wavefield separation processing. The curve refers to the mute
limit on the data chosen using the first-breaks. We have now accomplished the second
stage of interpretive processing. By interpreting the noise (mode-converted upgoing and
downgoing SV-events), we (the interpreter/processor) can redesign the processing to
minimize unwanted noise.
Our decision was to go back to the Z(FRT) and HMAX(FRT) data and perform
the wavefield separation on these two data panels, taking into account that our added
88
-i'-'-"~":;+-- O. B -I
~
m
1.2 !:i
3:
m
Vl
- 1 .6 ~
2.0
(1) Z'up (FRT) (2) HMAX'up (FRT) (3) HMAXup(derOl)(FRT)
-I
~
m
r-
-I
3:
m
~
1.6 Vl
2.0
(4) Zup(derol)(FRT) (5) HMAX*up(FRT) (6) Z·up (FRT)
Fig. 2.50. TIme-variant polarization IPP for the Ricinus data using the example far-offset
processing initially presented in the "processing runstreams" section of Chapter 2.
1.2
1.4
, 1.6
~
I ~
1.8 <
rn
r
~
2.0 rn
3:
- 2.2
2.4
2.6
Fig. 2.51. VSP-CDP IPP for the Ricinus data using the Z" up data shown in Figure 2.50.
The upgoing P-wave events are difficult to interpret due to the up- and downgoing SV-
wave events .
89
aim is to filter out the mode-converted 5V-events. The new time-variant IPP showing
this approach is shown in Figure 2.52. The input HMAX{FRT) and Z{FRT) data are
shown in panels 1 and 2. The wavefield separated HMAXup{FRT) and Zup{FRT) data in
panels 3 and 4 can now be compared to the original HMAXupldew,){FRT) and

Zupldewtl{FRT) data in panels 3 and 4 of Figure 2.50 (the guideline runstream processing).
Bypassing the (Z, HMAX) to (Z', HMAX') rotation (and possible noise generation caused
by the polarization), the upgoing P-events (some are highlighted in blue in panels 3 and
4 in Figure 2.52) are more interpretable on the new processing. Time-variant polarization
of the HMAXup{FRT) and Zup{FRT) data produces the HMAX"up{FRT) and Z"up{FRT)
data shown in panels 5 and 6.
The VSP-CDP IPP display (Figure 2.53) for the modified processing run-
stream shows that the upgoing P-events are easier to follow and to interpret than the
results presented in the VSP-CDP IPP using the original processing (Figure 2.51).
This is illustrated by the upgoing events highlighted in blue on panels 1 of Figures
2.51 and 2.53.
The Z"uplmed){+TT) data in panel 2 of Figure 2.53 was created by applying an
ll-point median filter to the data seen in panel 1. Once again, the interpreter/proces-
sor would derive an interpretation from Figure 2.53; however, one would also verify that
the same events exist on Figure 2.51. That is what interpretive processing does; it includes
every piece of available information during the interpretation stage.
Far-offset deconvolution of Ricinus case study data

In this section, the far-offset deconvolution of the Ricinus carbonate reef case study
(Chapter 4) is considered. The deconvolved far-offset results are also presented in Chapter
4 in order to evaluate deconvolution-generated noise within the context of a case study. The
far-offset deconvolution IPP for the Ricinus HOO-m data is shown in Figure 2.54. The
HMAX'{-TTl and HMAX'down{-TT) data are shown in panels 1 and 2, respectively. The
HMAX'down{FRT) data do not show evidence of significant multiple contamination.
The deconvolution verification panel contains the deconvolved downgoing events
(panel 3). The Z"up{-TT) data of the new interpretive processing (Figure 2.52) are shown in
panel 4. The Z"upldemn){-TTl data are shown in panelS. A comparison of panels 4 and 5 indi-
cates that deconvolution has added undesirable noise to the upgoing event data.
The VSP-CDP IPP of the Z"up(deconl(+TT) data is shown in Figure 2.55, and veri-
fies that the data were degraded by the addition of deconvolution noise. One must bear
in mind that the simple time-delay relationship between upgoing (Z"up) and downgoing
(HMAX'down) events may not be satisfied for the far-offset data (HOO-m offset).
90
mumimmmmmmmi~mumHmmmmml~~Uii~m~~mmm~~~imii~~mmDEPTH (m)
, i i ' . . I .... I( . I ' .,,' , I '. 0.4
. ....' ~ _ ~ I I • , -
I I I "
0.8 ~
m
~
r
1.2 ~
~
m
CI>
1.6 ~
2.0
0.8 ~
~
m
r
1.2 -t
3:
m
rn
1.6 ~
2.0
Fig. 2.52 . Modified time-variant polarization and wavefield separation IPP for the Ricinus
data following decis ions made during interpretive processing .
....~- 1 .4
. 1.6
~
1.8 ~
<
m
~
• 2.0~
2.2
, 2.4
2.6
Fig. 2.53. VSP-CDP IPP for the Ricinus data using the Z"up data shown in Figure 2.52.
The upgoing P-wave events are more interpretable although the diffraction and downgo-
ing SV-wave events have not been completely eliminated.
91
0.2 ~
0.4 m ~
~
0.6 ~
- 0.2
~
- 0.4 rn~
~
f
- 0.6 m
III
~
- 0.8
1.0
Fig. 2.54. Far-offset deconvolution IPP for the Ricinus data (Hinds et aI., 1993c) . Note the
deconvolution induced noise by comparing panels 4 and 5.
1.6
92
Integrated log display (ILD)

This display (Figure 2.56) combines the surface-seismic data around the well, the
geological log data (in time), and the synthetic seismogram. This display is included
because the VSP survey is usually the final borehole survey run at a well site (following
the acquisition of the geophysical logs). By the time the VSP is processed, the sonic log
should have been digitized and ready for a check-shot type calibration. As the VSP is
being processed, the near-offset VSP first-break times can be used to calibrate the sonic
log by correcting for sonic log drift and tying the integrated sonic log to the VSP first-
break times. The seismogram and calibrated sonic log are merged onto the seismic line
on the ILO to provide interpreters with a first-hand look at the location of the seismic
reflectors that were intersected by drilling.
Integrated seismic display (ISD)

The ISO for the Ricinus carbonate reef case study is shown as Figure 2.57. The dis-
play has merged the VSP-COP (or migration) results with the surface-seismic. The VSP-
COP is a lateral offset display of the reflectors away from the well and usually provides
higher-resolution than on surface-seismic data.
Integrated interpretive display (lID)

At the final presentation stage of the project, we create a display that shows a
large part of the geoscience exploration results, the integrated interpretive display (110).
The IIO shown in Figure 2.58 displays the integrated results from the Simonette carbon-
ate reef case history (Chapter 6).
The 110 can illustrate the interpretation procedure and the ties between different
datasets. For example, the Wabamun Formation interface was interpreted from the sonic
and gamma log to be at 3020-m well depth. Starting with the sonic and gamma logs in
depth on Figure 2.58, a line can be drawn through the sonic and gamma high (at 3020 m)
down to the upgoing (peak) event that intersects the 3020-m depth trace on the far-offset
Z"up(+ TTl data. This line is shown in green on Figure 2.58. The Wabamun seismic mark-
er is now tied to the logs through the common-depth axis of the far-offset VSP data (cen-
ter panel) and the logs (located above the center panel).
The Wabamun VSP event can be tied to the surface-seismic event at the 13-15
well, using the time tie from the geologic logs. To the right of the VSP data are the out-
side corridor stack after deconvolution (to show primary events), the 13-15 synthetic seis-
mogram, and the 13-15 sonic log displayed in two-way traveltime. Using the tie of the
VSP to the geologic logs, an interpreted Wabamun event can be further confirmed by
tying the event (on the 3020-m trace) to the corridor stack, sonigram, and sonic log (in
93
NORTHEAST
230
.,
Fig. 2.56. Integrated log display (ILO) for the Ricinus data.
SOUTHWEST NORTHEAST
COP # OFFSET (m) COP #
~!! !: ::;:: 5:;5:: ;~ ; ;; :! ~!! .: ~; i : i $ ~ ::ti ' ~ i r~ l ~a i a l ~ ; ~3 ~ ~ : ~ 3 ~ ~ :~ ;: ;3
1.7
_ _ 2.2
2.3
2.4
SEISMIC
Fig. 2.57. Integrated seismic display (ISO) for the Ricinus case study data (Hinds et aI.,
1993c). The VSP-CDP transformed Z" up(+TT) data are inserted at the well location onto
the seismic and replace the seismic trace data at the overlapping COPNSP-CDP offset
locations.
94
I:
'-<\'. i·. JO ~
'i t ).";}',. I j\ ~
I I ,J• •ij
I#'-.J.
l l~ I ~1ltr.rl"
, ,~'1• \~~~~J"¥"
ii " ''! . "
I I
•8
I~ ~
1), 15
'-~
... (;
,,I:!: §i
,rfl)(~..... I~'" ~'~'{; j""W"r
I 'I i i f;
, I
1
'8-
DEPTH ( m )
· ~ ·i i i ~f~ ~ i i i i ~ ~ ~ ~~ ! 5~ ~ E ~ 5 ; Z ::s
....
-
ttl
'" ~
t;'
l:l-
\t) ~- " ,
-
....:ttl:s
"1
"0
VI ;;;
~ ....
~
<'
ttl
'"0
1"~ "1
.. o
I"J
'' '-, ttl
Ul
Ul
....::s
OQ
, 1. 1
Fig. 2.58. Integrated Interpretive Display (liD) of the Simonette case study data (Hinds et aI., 1993b). This display includes sonic and
gamma logs (both in time and depth), the Z" up(+ TT) VSP data , corr idor stack results, and surface seismic data . This panel incorporates
all of the exploration data for the area onto one single display for easy integrated interpretation.
time). This is shown by a blue line in Figure 2.58. This tie can be drawn on the 13-15
sonigram, outside corridor stack of the Zupldeconj(+ TTl data, and inside corridor stack
of the Zup(+ TTl data that are spliced into the surface-seismic at the 13-15 well location.
This is seen on the left panel of Figure 2.58. The nondeconvolved inside corridor stack
enables an interpretation of multiples on the surface-seismic at the 13-15 well location.
The interpretation of the Wabamun event (shown in red) is extrapolated onto the surface-
seismic on either side of the well. The 16-16 sonic log (in time) and 16-16 sonigram are
spliced into the seismic section at the 16-16 well location to enable the interpretation of
the reef at the 16-16 location.
The interpreted basal reef marker rising onto the pinnacle reef is depicted on the
far-offset VSP data panel. The VSP-CDP IPP displays complete the presentation (Figures
2.48 and 2.49) by showing the reef interpretation at the well and laterally offset for 200 m
away from the well towards the "known" reef (at the 16-16 well). Questions about the
effects of multiples are answered, and the tie between the geology and the geophysics is
established. Interpretive processing has driven the processing, and the various IPP and
displays confirm the interpretation.
96
3 - Lanaway Field Case Study
Overview
On the basis of conventional surface-seismic data, an exploratory well (referred to
as the VSP well) was drilled into the up-dip, raised rim of the Devonian Leduc Formation
reef complex at Lanaway Field, south-central Alberta, Canada. The VSP well was expected
to encounter an anomalous late-stage carbonate accretionary buildup at the Leduc level. It
was anticipated that the Leduc at the VSP well location would be about 80 m higher than
at adjacent rim well sites. The envisioned accretionary growth was not present; the top of
the Leduc in the VSP well was consistent with other rim wells in the vicinity (Figure 3.2)
and inconsistent with the seismic interpretation. Fortunately, however, the Leduc was
structurally closed, and the VSP well was completed as an oil well (producing both from
the Nisku and Leduc formations).
To resolve the apparent discrepancy between the interpreted surface-seismic data
and geology at the VSP well, a near-offset vertical seismic profile (VSP) was recorded at
the well site. The interpretation of the VSP was relatively successful in that these data con-
firmed that the original interpretation of the surface-seismic data, with respect to the
Nisku, Ireton, and Leduc tops, was incorrect, and also that the anomaly observed on the
surface-seismic line was not a processing artifact. Our interpretation is that the surface-
seismic anomaly is caused by several superposed effects, including anomalous structural
relief at the pre-Cretaceous subcrop, stratigraphic anomalies (thicker sections of reefal car-
bonate) within the Winterburn Group, and seismic focusing caused by draping of the
Ireton over the raised rim of the reef along the traverse of the seismic line. Interbed mul-
tiples within the Wabamun/Nisku interval could be a minor contributing factor.
Carbonate reef development in the Western Canadian Sedimentary Basin

As an introduction to the geology of the carbonate reef case studies presented in
this chapter (Lanaway Field), Chapter 4 (Ricinus Field), and Chapter 6 (Simonette Field),
the depositional sequence history for the Hume-Dawson, Beaverhill, and Saskatchewan
subsequences (Moore, 1988 and 1989a) in the Western Canadian Sedimentary Basin
(WCSB) is reviewed below. The case studies focus on the Leduc reefs of the Saskatchewan
subsequence.
Three major subsequences in the WCSB encompass oil-producing carbonate reefs:
Hume-Dawson (Upper Keg River, Rainbow, and Upper Winnipegosis reefs), Beaverhill
97
(Swan Hills Formation reefs), and Saskatchewan (Leduc Formation reefs). Sea level in the
first reef-building subsequence, Hume-Dawson (mid-Eifelian to mid-Givetian) occurred
contemporaneously with the deposition of the Lower Keg River Member in Northern
Alberta and the Lower Winnipegosis unit in southern Saskatchewan (Brown et al., 1990).
As illustrated in Moore (1988, 1989a), the Keg River barrier reef (or Keg River - Pine Point
barrier reef; Moore, 1989a) formed northeastward of the Peace River Arch in the shallow
marine Elk Point Basin environment. Behind this barrier complex, isolated pinnacle reefs
up to 200 m in height developed (Keg River and Winnipegosis Formations). Transgression
towards the end of Winnipegosis/Keg River time resulted in basin shallowing, hyper-
saline conditions, and the termination of reef growth. The rock salts of the Prairie
Formation and the Black Creek Member of the Muskeg Formation were deposited in the
inter-reef areas. Anderson et al. (1989d) show examples of Elk Point carbonate reservoir
signatures on conventional seismic data.
The post Hume-Dawson reef building subsequences are the Beaverhill (late
Givetian to early Frasnian) and Saskatchewan (mid-Frasnian to end Frasnian). The depo-
sitional histories of both subsequences involve the development of a carbonate platform,
reef growth, basin filling first by carbonates, then by shales and siliciclastics, and finally
by progradation of carbonate platforms (Moore, 1989a).
The Beaverhill subsequence consists of Swan Hills platform carbonates, Swan
Hills Formation reefs, and Waterways Formation shales. The Swan Hills reefs are
described seismically in Anderson et al. (1989a) and Ferry (1989). The growth of the Swan
Hills reefs was pulsatory, and they have been described in terms of several cycles of reef
"layering" (Wendte and Stoakes, 1982). The progradation of the carbonate platform at the
end of the subsequence resulted in the deposition of the basal part of the Cooking Lake
Formation.
The platform facies for the Leduc Formation (Woodbend Group, Saskatchewan
subsequence) is either the platformal carbonates of the Cooking Lake Formation (in the
eastern WCSB), or in their absence, an argillaceous ramp of the Waterways Formation
(western WCSB). Leduc reefs have been divided into geographical realms by Stoakes and
Wendte (1987). The Leduc platform reefs in the southeastern realm include the Bashaw-
Duhamel and Rimbey-Meadowbrook reef chains (Ricinus and Lanaway examples;
Chapters 3 and 4) which overlie Cooking Lake Formation platform facies. The reefs in the
western WCSB region (Simonette example; Chapter 6) developed on the Waterways
Formation. These reefs are encased in the Duvernay and Ireton formation shales.
Anderson et al. (1989c) show examples of Leduc reef seismic signatures.
Nisku Formation (Upper Frasnian) reefing occurred in the West Pembina Shale
98
Lanaway Field Case Study
Basin area. Bioherms developed mostly as Zeta Lake Member pinnacle reefs. The basi-
nal infill consists predominately of carbonates and shales (Moore, 1988).
Introduction
The Upper Devonian Woodbend Group (Saskatchewan subsequence) in central
Alberta is subdivided into four formations: Cooking Lake, Leduc, Duvernay, and Ireton.
The Cooking Lake is the platform facies, the Leduc is reefal, and the Duvernay and Ireton
are inter-reef shales (Klovan, 1964; Mossop, 1972; Mountjoy, 1980; Stoakes, 1980; Stoakes
and Wendte, 1987; Moore, 1988; Anderson et al., 1989a, band c; McNamara and Wardlaw,
1991; Moore, 1989a and b; and Hinds et al., 1994a). The stratigraphy of the Central Plains
area of the Western Canadian Sedimentary Basin is noted in Figures 3.1A, B, and C (rep-
resenting the stratigraphic sequence from the Quaternary through Cambrian.)
The Leduc buildup at Lanaway (Figures 3.2 and 3.3) is classified as a large atoll. It
towers some 200 m above the Cooking Lake platform and exhibits a seismically mappable
(peripheral) raised rim and a structurally lower central lagoonal area. Such raised rims are
described in Mossop (1972) in his study of the isolated Leduc Formation limestone reef
complex at Redwater. In Mossop's study, the raised rim is attributed to the greater com-
paction of the finer-grained sediment in the central lagoonal area in comparison to the
coarser-grained sediment that comprise the rim facies. The updip edge (northeast) of the
rim at Lanaway Field is productive where the reef is structurally closed and effectively
sealed by the Duvernay and Ireton formation shales. The geologic cross-sections shown in
Figures 3.4, 3.5 and 3.6 (from wells shown in Figure 3.3) and the seismic section (Figure
3.7, for the seismic line indicated on Figure 3.3) illustrate the interpreted morphological
relationships between the Leduc, and Ireton and Duvernay shales in the Lanaway area.
On surface-seismic data, the Lanaway Leduc reef is readily differentiated from
inter-reef shales. The carbonate build-up is characterized by an appreciable velocity pull-
up (25 ms), time-structural drape at the top of the Devonian (25 ms), and character varia-
tions within the Woodbend Group (associated with the abrupt transition from shale to
reefal facies). Back from the steeply dipping edge, the top of the reef is defined usually on
seismic data, being manifested as a high-amplitude peak on normal polarity seismic data.
In this chapter, we discuss the interpretations of the 2-D surface-seismic and VSP
surveys (performed on VSP well at C' shown in Figure 3.3). The surface-seismic data
were acquired prior to drilling the VSP well which intersected the reef some 80 m below
prognosis. The VSP survey was run in an attempt to resolve the apparent discrepancy
between the initial seismic interpretation and the well results.
99
"- c: ~ AGE "- e
ERA ~ SERIES GLOBAL STAGES ~G: CENTRAL PLAINS a> "- ffi ~
SERIES GLOBAL STAGES MY CENTRAL PLAINS a> "-
0. a> 0. a>
0. ..... TITHONIAN 0. .....
~~ PLEISTOCENE LAURENTIDE DRIFT
:::>~ PORTLANDIAN
:::>~
~ 1.6 KIMMERIDGIAN
a> ~ a>~
U UPPER OXFORDIAN
PLIOCENE CALLOVIAN GREY BEDS ..c
£a> Ci5 ..... a>
f - - - - - - - - - 1 5.3 (/) BA~£,~JAN_
~ >- ..c o I o..c
..........
MIOCENE
~ MIDDLE f---- ROCK CREEK
...
AALENIAN -
<li 2~ 237 .....0 .....
:::l
TOARCIAN POKERCHIP SH
0
....uU o i= OLIGOCENE
() CO
0- 0
N
""')
PLEINSBACHIAN
SINEMURIAN NORDEGG I 0
.- -CO
~ ffi 36.6 .- CO CO 0
LOWER o COCO
tS () I- EOCENE o a> 0> HETTANGIAN Na>o>
~ 57.6,1- -, N"-T"""
oe CO (/)• en
W
208
g~T"""
"'t:l PALEOCENE 1••. PASKAPOO I a> e(/) .;::
a> ~
NORIAN a>(/)cn
~
4
::2:ea>
oe
MAASTRICHTIAN
f-=~~=~---174.5 EDMONTON
SCOLLARD
BATILE
O
,-,,'(ij 0 U
Ci5
UPPER
CARNIAN
: ; '(ij
.- a.. cu
s
CAMPANIAN GROUP ~~~~H~.-----=""'AR='=AWc-----'---, ~a.. '§ (/)
-c (/) "-
<li BELLYRIVER {J~~g:S~~R) VICTO~"" RIBSTONE
CO- 0 LADINIAN (/) - 0
...t:: 1--------164.0 LEAPARK eCO..o ii: MIDDLE ~~..o
E.- ::::l ..... CO
CD~~ I-
SANTONIAN ANISIAN
J~....J
FIRST WHITE SPECKLED HALE
f----:===,,-----1S7.5 ..... e
...bb
t::
rIJ
UPPER CONIACIAN
TURONIAN
88.5!5
o
l3
CARDIUM
COLORADO SHALE
CO a> I-:
::::lO«
0a>CJ
LOWER SCYTHIAN
KAZAN1AN
245 CDO~
o.a>CJ
o
o
,...;
rIJ
en g o...c
:::> ..... «
a> ..c«
91.0 SECONDWHITE SPECKLEDSHALE PERMIAN
<li
~ ..... "- KUNGURIAN
ARTINSKIAN
sU:::l
U
@ s0 ..c: a>
SAKMARIAN
a> '+- CD
£ ~
ASSELIAN
N U
CENOMANIAN
----?; FJ~H_S_C~l:E_Z9~E
- ----------- ?? -----
..... 0 .:::: GlElIAN
KASIMQVIAN ..cO.::::
0« ..... (/)CO
~
U
97.5 •
E (/)
.......> UPPER MOSCOVIAN
<li
~ ~ o c E '0 '-"
::iE 0:
I au BASAL COLORADO
'0
"- .0- .- (/)
:::l
BASHKIRIAN 320
o .g .!::
~a>~
Q. '+- (/)
~ U ALBIAN 5 cr; UPPER MANNVILLE >.CDCO 0
~
e- ~ &
~
CLEARWATER
::>1-_ _-::-:-:-====-
..c: 0. CO
-1 0 . _ ~ o
0:
w
u..
SERPUKHOVIAN
>. 0. CO
..c -
0..2 ~
-
t::
~
rJ)
:> LOWER
f - - - - - - - 113
APTIAN
BARREMIAN
HAUTERIVIAN
119
124::;
131
"
~
~
~
~
g
GLAUCONITIC
OSTRACOD
RSUE BASAL UARTZ
DETRITAL -:
~ .2 CO
0)0 .....
.- N e
cuoa>
~(/)E
(/) a> .-
0 Z
N 0
0 m
W ~
LOWER
VISEAN
c,
::J
0
0:
o
~
c
z
~ELKTON
SHUNDA
PEKISKO
-,
-,
~
O)N'E
cua>E
0 CO
.- 0 a>
"- a..
..... CO .-
...J U
so '-" -g
::J
«'~~ «a.. 0:
VALANGINIAN
BERRIASIAN
136
-" (/)
T""" () CO
TOURNASIAN
360
---------E~~;W- ? --------~ a::ie(/)
T""" .5!! CO
C,)
r;;
TITHONIAN
PORTLANDIAN
144
. 'w '0
C'0(/)CO
z
< WABAMUN BIG VALLEY
M. eO '0
Z CO
~ UPPER K6"x~~~~I;'~N . CO e 0 FAMENNIAN . > e
~ CALLOVIAN FERNIEGROUP GREY BEDS
O)"-co >
l1J CROSSFIELD \ STEITLER .Q'l a> CO
i.L-=50 c 367 LLOO
~ SERIES GLOBAL STAGES AGE

RANGES ~ w ~ MY CENTRAL PLAINS
11 10 9 8 7 6 5 4 3 2 1 ~ 27 26 25 24 23 22 21 en
::)
UPPER SERPUKHOVIAN
0
x 47 0::
I I I I I I I w
u,
lip 46 Z VISEAN
~~~
~0 SHUNDA"""
0 OWER =>'"
al 0::: C!J PEKISKO -,
45 0::
~ A I~ ttl -c
U TOURNASIAN 360 -----------'---?Mtl",4?-------------4
44 WABAMUN BIG VALLEY
~ II! 1# / FAMENNIAN
CROSSFIELD I STETILER I
4"1>/ 43 367
f>'/p} WINTER
BURN NISKU
IJ1 /ZET~
'1 CAMROSE
ff 42 b UPPER 0
tfJ FRASNIAN ~~
F\\» 41 :iE :go
0'" LEDUC
DUVERN ...r
~" BEY z U 0'"
en ;: ~, _ _ _ ? ____ COOIOOO
A o I''i >)" 40 ;;;

0 Z BASAL COOKING L A K E ' LAKE
374 i-I ~ WATERWAYS
t"""
Ii } 39
"'C
en N sZ
~ _~? ~::" ? SLAVE POINT ~
2h STR,",CH N
tv » 0 ~~ ::s
GllWOOD WAIT MOUNTAIN
W 0 GIVETIAN I
tj 38 >
...J PRAIRIEIMUSKEG GROUP
~ IF w 0.. I ~
::)
"'" ~"" \)
<C Cl MIDDLE
fI 37 0 ~
f8Y(' l,f' 1/ a.. 0:: KEG RIVER (WINNIPEGOSIS)
~
iff ~~~ 380 "rj
o I
f-
I-'
"'< ~ ,-
j,; l>-r
I 'liLA AWA 36 Z CONTACT RAPIDS I
...
rD
o ~Iff ~ EIFELIAN 5 l:l.o
I-' !J' 0..
o C 35
-
<\ ~'9""c: ,'\: Ii IT)dl
ARF INGT ~ ("')
...J
"
Q' EMSIAN
387 W COlD LAKE ~
34 SIEGENIAN 401 ERNESTINA
'IJ
\, OWEn rD
~
F'l '10 """ I"'G re';' 1)7'
,,~).• jJJ GEDINNIAN 408 ~ rJ'}
i, 33
LUDLOVIAN
SILURIAN
...
(b 32 LLANDOVERIAN 4'"
=l:l.o
'-<:l
,i ~ z ASHGILLIAN
< UPPER
31 ti CARADOCIAN
P s
a MIDDLE LLANDEILAN
IV i\ 0
T if III·. LLANVIRNIAN
Q) 30 AR I IAN
0 LOWER
'z" T
.li1······ 29 -c
iii UPPER FRANC NIAN PINNEGAN
lfii Ii i DEADWOOD
'" MIDDLE
'" PIKA
EARLIE
8ASAL SANDSTONE
5 LOWER 5
PRECAMBRIAN
Fig. 3.2. Regional location map of the Lanaway study area. The Fig. 3.1 C. Stratigraphy from the Upper Carboniferous
shaded regions represent the areal extent of the major Leduc (Paleozoic) period to the Precambrian of the Central
Formation carbonate reefs of the area (with permission of Talisman Plains area of Western Canada Sedimentary Basin
Energy Inc.) (after AGAT Laboratories, 1988).
Well nomenclature
As shown in Figures 3.2 and 3.3, the basic geographical areal unit used in the
WCSB is the township (denoted by a township and range number). Township numbers
increase from south to north; range numbers increase (until a meridian is crossed) from
east to west. Townships are six miles square and are subdivided into 36 sections. Each
section is one mile square and numbered sequentially from 1 to 36, starting in the south-
east corner of the township. Each section is subdivided into 16 land survey divisions
(LSD), each of which is a quarter mile square. LSD numbering starts in the southeast cor-
ner of the section.
Well nomenclature within most of the WCSB is denoted by a five-number sequence
(e.g., 1-2-3-4W5M). The first number (1) refers to the LSD in which the well is located. The
second, third, and fourth numbers refer to the section, township, and range of origin,
respectively. The fifth number refers to the meridian. For example, well 11-1-36-4W5M
(Figure 3.2) lies in LSD 11, section 1, township 36, range 4, west of the fifth meridian. Often,
the well name is presented in a shortened format in which only the LSD and section are
noted. Using this format, 1-11-36-4W5M becomes 1-11.
1 MILE
SCALE
TWP 36 RGE4 TWP 36 RGE 3 W5M

W5M
16~~.~. 16-~-Q
.
11-1 <r"
.... '
·····~·1·1~~··· . . .
A
I~ 6-6
Fig. 3.3. Detailed map of the I..~
Lanaway study area displaying the ..
! :
11-36 : :
seismic section traverse for the
seismic data shown in Figures 3.7, B"¢;. i··.ysp
...... C' .......
3.8 and 3.14 and the locations of
the wells used in the geological
cross-sections shown in Figures
..
'2;-3.&"
. '
-,
' .
....
....
".
3.4, 3.5 and 3.6. -<>
15-25
". ...."A'
~
10-29
·6-30
TWP 35 RGE4
TWP 35 RGE 3W5M
W5M
GEOLOGIC CROSS-SECTION
SEISMIC SECTION ••••••••
102
Lanaway Field
The Leduc reef at Lanaway Field rises 200 m above the Cooking Lake platform,
and exhibits a structurally elevated reef rim and a lower interior lagoon (Figures 3.4 and
3.5). Production from the Leduc is restricted to the updip eastern edge (raised rim) of the
reef complex (up to 50 m of pay). The eastern and western limits of production are
defined by the fore-reef slope and the hydrocarbon/water interface, respectively.
The geologic cross-sections depicted in Figures 3.4, 3.5 and 3.6 illustrate the mor-
phology of the Lanaway complex (Figure 3.3). Well 10-29 (Figure 3.4) is off-reef and
encountered a full section of inter-reef shale (Ireton and Duvernay); 11-1 penetrated the
structurally low and wet interior lagoon; 16-1 and 11-6 were drilled into the up-dip raised
rim of the Lanaway reef and are productive. Well 11-36 (Figure 3.5) penetrated the struc-
turally low and wet interior lagoon; 2-36 and the VSP well were drilled into the raised
rim of the Lanaway complex. The VSP well is productive. Well 2-36 penetrated the Leduc
below the hydrocarbon/water contact; however, the well is classified as a Nisku oil well,
with extended productivity from gas from the Cretaceous Viking Formation and oil from
the Cretaceous Basal Quartz (Figures 3.1A-C).
11-1 16-1 11-6 16-6 10-29 EAST
w~~s~T.-..-Lj=-__-i-----~let--l-------r-
"' a.
0
l-
[' 1650
1700
~
z WABAMUN
f
Q. 1750
o w «
Z
...J
al ::!:
o S ::!:
en «
~
CJ
s
0
z
§! -c
o
1800
1850 ~
CJ)
C
OJ
»
NISKU
IRETON
LEDUC REEF
DUVERNAY} f
~~ Z-
COOKING LAKE
Fig. 3.4. West-east geologic cross-section A-A' (refer to Figure 3.3 for well locations). The
Leduc Formation reef in well 11-1 is structurally low and wet; wells 16-1 and 11-6 are pro-
ductive (Leduc oil producers); and wells 16-6 and 10-29 are off-reef and abandoned (from
Hinds et aI., 1994a).
103
WEST EAST
B 11-36 -36 VSP C' 10-29
A'
1700
« o -e
~
«
o ~«
~il~
:::i! :::i! ::::i!
:::i!
«
z
0
:::i!
«
z
0
::::i!
« WABAMUN
t? (/) t? (/) o
1
l~
NISKU 3
S
IRETON
f
---:-7
L~
s;.'__
-l-
t
J..
DUVERNAY"",
.-;t ~5.-.'.
-;f;
.~~ -r
COOKING LAKE
Fig. 3.5. West-east geologic cross-section B-A' (refer to Figure 3.3 for location). The
Leduc Formation reef in wells 11-36 and 2-36 (producing oil from the Nisku) is structurally
low and wet; the VSP well is productive (Nisku and Leduc Formation oil reservoirs); and
well 10-29 is off-reef and abandoned.
SOUTH NORTH
c' VSP 6-6
Vi
a.
0
f-
1700
~
z WABAMUN
e.w
...J
1750
< l!l
:::;;
:::;; :s<:
<
C) (/)
:iC C
en
C) tn
(/)
0 m
...J
0 »
z
NISKU
IRETON
2000
LEDUC
2050
Fig. 3.6. North-south geologic cross-section C-C' (refer to Figure 3.3 for location). The
Leduc Formation reef in well 11-6 and the VSP well is productive; and well 6-6 is wet and
classified as an abandoned oil well.
104
The VSP well has a different oil/water contact than 11-6 (to the north) and is
assigned to a separate pool. The hydrological barrier between these two wells is related to
structural relief at the Leduc level (Figure 3.6). Relief could be a result of several processes
or features, including surge channels, shale tongues, differential compaction, or original

reef morphology (Klovan, 1964; Mossop, 1972; Anderson et al., 1989a, b, and c).
The VSP well (Figures 3.5 and 3.6), although ultimately productive (from both the
Nisku and Leduc), was 80 m low at the Leduc level relative to the original seismic-based
prognosis. This pre-drill structural estimate was based on the interpretation of the con-
ventional seismic data displayed as Figure 3.7. In the original interpretation, the Leduc
event is 30 ms (80 m) higher between CDP traces 110 and 140 than elsewhere. Geologically,
this anomaly was initially envisioned as localized, late-stage, accretionary reef growth.
The drilling of the VSP well, however, established that on the original seismic interpreta-
tion, the top of the Leduc was miscorrelated by one cycle (between COP traces 110 and
140). Well-log data established that at the VSP well, the Leduc top was more-or-less time-
structurally consistent with the top of the Leduc elsewhere on the seismic line.
VSP data acquisition

In an effort to resolve the apparent discrepancy between the surface-seismic data
(as originally interpreted; Figure 3.8) and the Leduc top at the VSP well, a VSP survey
was conducted at that well site. The interpretation of the VSP data was expected to elu-
cidate both the geological significance (if any) of the misinterpreted seismic anomaly and
the extent of multiple interference on the seismic data.
The VSP data were acquired using a single surface offset, located 200 m west of the
VSP well on the lease access road. The source consisted of two tandem Vibroseis units uti-
lizing an 8-to-96 Hz input sweep (compatible with the frequency content of the existing
surface-seismic data). The Vibroseis sweep was 13 seconds in duration, and the recording
length was 16 seconds, resulting in a 3 s-long cross-correlated output time series. On aver-
age, six sweeps were summed for each subsurface geophone sonde location. The ground
at the source site was partially frozen and suitable coupling was achieved. The Vibroseis
pad location at the offset was shifted periodically by a minor amount to ensure ground
coupling stability and to minimize damage to the surface of the lease access road.
The total depth (TD) of the VSP well was 2990 m below the Kelly Bushing (KB) at
the time of the running of the VSP survey. The well was drilled slightly deeper following the
VSP survey for stratigraphic evaluation purposes. The KB elevation was 968 m above sea
level (asl); the source elevation was 963 m asl. The geophone sonde was lowered to TD, then
raised at 20 m intervals to 2590 m below KB (30 m above the top of the Wabamun Group).
105
SOUTH NORTH
eo rc
'"
~
'"
~ '" '"
~ ~ ~ ~ '"
~
~
- '" '"
:: a>
" '" '" ~ ~ eo
- 0 ~ ~ ~ ~ ~ ~COP
0.0
-i
0.5 :E
o
-i
~
1.0 ~
- <
m
r
-i
s::
- 1.5 m
..-..
2.0
1 KM
SCALE 1 KM
SOUTH NORTH
::; ::;
'" - ;;;
'"
en ;:;;
'"
<Jl :;;:
'"
A ;;;
'" - '" -'"
;;; ;;; ;;; - ;;
'"
0 '"
'"
~
CD
'" -
a>
OJ COP
1.0
CARDIUM 1.2 -i
:E
0I
. VIKING
:E
. GLAUCONITIC
1.4 ~
. PEKISKO
-i
~
BANFF
• WABAMUN
1.6 m
r
. NISKU -i
. IRETON
LEDUC s::
m
..-..
1.8 rn
-
Fig. 3.8. Enlarged interpreted version of the apparent time-structural anomaly at the
Leduc level. The location of the VSP well site is at COP number 116.
--_ii:.. 2 .0
106
Between 2590 and 1315 m (shallowest processed sonde depth location), the sonde recording
interval was increased to 25 m. Data were recorded at only three levels above 1315 m (at 600,
1000, and 1150 m). These data were acquired principally for sonigram calibration purposes.
At each sonde location, the three-component geophone tool was coupled to the borehole
(using a locking arm). The data were recorded at a 2-ms sample interval using the sse 1078
micro-Vax based system. The recording filter was set at OUT lOUT.
VSP data processing

As an aid to the interpretation of the VSP data, a suite of interpretive processing
panels (IPPs) were generated for these data. These panels are used in the progressive
interpretation of the data to facilitate quality control of the processing sequence. More
specifically the panels display:
1) upgoing and downgoing P-wavefield separation;
2) deconvolution of the separated upgoing P-waves using an inverse filter calculated
from the separated downgoing P-waves; and
3) inside and outside corridor stacks of both the deconvolved and nondeconvolved
upgoing P-waves.
P-wave separation
In the initial phase of processing, the upgoing and downgoing P-waves of the ver-
tical geophone data, Z(FRT), were separated. This separation procedure is illustrated in
the wavefield separation IPP of Figure 3.9.
The trace normalized Z(FRT) and the gained Z(FRT) data are displayed in panels
1 and 2 of Figure 3.9. Several primary upgoing reflections can be identified on these data:
Viking (at a depth of 2100 m); Mannville (the Glauconitic at 2158 m); and Banff (2293 m).
Note that the series of strong reflections originating below the TD of the VSP well do not
intersect the first-break curve, and therefore these events cannot be confidently classified
as either primary or multiple events.
The gained Z(-TT) data in panel 3 illustrate that the downgoing wavetrain (the
multiples) is comprised mostly of high-amplitude, surface-generated multiples and less
prominent interbed multiples. Surface-generated multiples are recorded on all of the
traces and are manifested as laterally continuous events arriving after the first-break
events. Interbed multiples, in contrast, do not extend over the entire depth range. If pre-
sent, they would be present on the deeper traces only.
The downgoing waves contained in panel 3 were separated from the gained Z(-IT)
data using an ll-point median filter and displayed in panel 4, as Zdown(-TT). The residual
107
DCARDIUM
VIKING
0.25 -I GLAUCONITIC
~ PEKISKO
m DBANFF
0.50 ~ WABAMUN
3: DNISKU
m IRETON
DLEDUC
0.75 -;;
1.00
(1) Z(FRT) (2) Z(FRT) (3) Z( -TT)
gained
i j iii iii iii i i ~ i i ~ ~~ 3~ ~ ~ ~ ~ ~ ~ g! ~ ~ Iii i i i iIi! iii iii i i ~ i ~ i !! j ! a! ! ! ~ ! ! ! ! i ~~ i! DEPTH ( m )

II II I I 111111 III IIIi 0.00
0.25 ~
rn
0.50 ~
3:
m
~
0.75 en
(6) Zup(+TT) (7) Zup(+TT)

1.00
median filtered
Fig. 3.9. Interpretive processing panel depicting the wavefield separation of the near-
offset VSP data.
DCARDIUM
0.25 -I VIKING
GLAUCONITIC
~
m
PEKISKO
DBANFF
0.5 ~ WABAMUN
3: NISKU
m IRETON
0.75 :E DLEDUC
1.0
(1) Zup(+TT) (2) z.,(+ TT) (3) Z(-TT)
median filtered
(5) Z(decon)(- TT) (6) Z up(decon)( +TT) (7) Zup(decon)(+ TT)

median filtered
Fig. 3.10. Interpretive processing panel depicting the deconvolution of the near-offset
VSP data.
108
upgoing wave content in the Zdown(-TT) data (panel 4) is minimal. The upgoing waves,
Zup(-TT) in panelS, were separated by subtracting a scaled version of the downgoing
wavefield, Zdown(-TT) of panel 4, from the combined wavefield, Z(-TT) of panel 3, fol-
lowing methodologies described by Balch and Lee (1984), Hardage (1985), Hinds et aI.,
1989, and others. The two final panels (6 and 7) in the wavefield separation IPP show the
separated upgoing waves Zup(+TT) before and after the application of a 3-point median
filter. The Cardium, Viking, Glauconitic, Pekisko, Banff, Wabamun, Nisku, Ireton, and
Leduc events in +TT time configuration are interpreted on the final panel and confirmed
on the previous panels.
Successful wavefield separation is critical to the interpretation of the VSP data,
because any residual upgoing energy in the Zdown(-TT) data will be subtracted out of the
upgoing wavefield data. Note that the median filtering process (panels 4 and 5) was not
particularly effective in separating those upgoing waves that arrive after the first-break
time on the deepest sonde location trace (panel 4). This is acceptable, however, only
because these noise events are below the time interval of interest.
VSP deconvolution
Surface-generated and interbed multiples are represented on the separated down-
going wavetrain shown in panel 4 of Figure 3.9. The initial downgoing pulse (except in the
case of head wave contamination) is the primary downgoing P-wave; any downgoing
waves that arrive later are multiple reflections. These multiple events can be effectively
removed by deconvolving the upgoing wave data with an inverse filter derived from an
analysis of the downgoing wavetrain. The deconvolution IPP shown in Figure 3.10, enables
the interpreter to control the quality of the VSP (Up over Down) deconvolution process.
Panels 1,2,6, and 7 in Figures 3.9 and 3.10 are bulk time-shifted to improve the IPP display.
The first two panels (Figure 3.10) are the unfiltered and median-filtered Zup( + TT)
data. Panels 3 and 4 are the nondeconvolved Z(-TT) and Zup(-TT) data. The fifth panel is
the deconvolved upgoing combined wavefield data, Z(decon) (-TT), and represents an
example of downgoing wavefield deconvolution applied to the combined (total) wave-
field as first reported in Smidt (1989). Following deconvolution, the data were wavefield
separated, normalized around the first breaks, corrected for spherical divergence, and
shifted to pseudo-two-way traveltime. The resultant Zup(decon)(+TT) data are displayed in
panel 6. A comparison of the median-filter-enhanced, nondeconvolved (Zup(+TT) in
panel 2) and deconvolved upgoing (Zup(decon)(+TT) in panel 7) wavefield data indicates
that deconvolution has enhanced the frequencies of the upgoing waves and yet pre-
served the integrity of the primary reflections.
109
1.0
.....
1.25 m
~
!:i
3:
m
III
1.50 ~
2.00
(1) Zup(+TT) (2) Zup(+TI) (3)(4) (5) Zup(+TI) (6) Zup(+TI)
median filtered muted outside corridor muted inside corridor median filtered
o CARDIUM (3)Zup(+TI) (4) Zup(+TI) o BANFF
VIKING WABAMUN
GLAUCONITIC outside corridor stack inside corridor stack o NISKU
PEKISKO IRETON
o LEDUC
Fig. 3.11. Interpretive processing panel illustrating the utility of the nondeconvolved inside
and outside corridor stacks for the near-offset VSP data.
1.0
.....
~
m
1.25 !:i
3:
m
.-..
III
1.50
1.75
(2)ZuP(decon)(+TT)
muted outside corridor
(3) Zup(decon)(+TT) (4) Zup(decon)(+TT)
outside corridor stack inside corridor stack
Fig. 3.12. Interpretive processing panel illustrating the utility of the deconvolved inside
and outside corridor stacks for the near-offset VSP data.
110
The effect of multiple interference and the relative success of deconvolution can
be appreciated by an analysis of the VSP Glauconitic event. On the Zup(+TT) data in
panel 2 (Figure 3.10), the nondeconvolved Glauconitic event is relatively continuous at
sonde depths below the Cardium (1825m). At shallower depths, the Glauconitic event is
partially masked by a probable interbed multiple that has as its lower generating surface,
the top of the Cardium. On the Zup(decon)(+IT) data in panel 7, the Glauconitic event is
relatively continuous at all the sonde depths, implying that the interfering multiple has
been effectively attenuated.
Note that the Nisku and Leduc events are relatively low amplitude (in compari-
son to the Nisku and Leduc events on the seismic line away from the VSP well site), prob-
ably because of destructive multiple interference. On the Zup(+TT) data (panels 1 and 2),
the strong peak from the shallowest trace out to the 2140 m trace visually merges with
the Nisku peak (seen clearly from depths 2140 to TD). For clarity, the multiple-contami-
nated portion of the Nisku event is not highlighted in Figures 3.10 and 3.11. (Note that
the Nisku primary is lower amplitude in comparison.) This is an example of a multiple
more-or-less constructively interfering with a primary. After deconvolution, the Nisku
event on the Zup(decon)(+TT) data is more laterally continuous and uniform amplitude.
The anomalous multiple-induced peak that existed before deconvolution was about 5-10
ms higher than the underlying Nisku primary.
Inside and outside corridor stacks

Multiple contamination of the VSP upgoing waves can be examined on the inside
and outside corridor stacks of both the Zup(+TT) and Zup(decon)(+TT) data. The inside cor-
ridor stack of the Zup(+TT) data contains both primary and multiple events, whereas the
outside corridor stack data is relatively free of multiples. If the deconvolution is success-
ful in removing the multiples, then both the inside and outside corridor stacks for the
Zup(decon)(+IT) data will be dominated by primary reflections.
In Figure 3.11, the outside and inside corridor stacks are shown as panels 3 and 4,
respectively. The input data that are stacked to create the inside and outside corridor
stacks are shown as panels 2 and 5, respectively. The Zup(+TT) data shown in panels 1
and 6 are placed beside the corridor muted Zup(+TT) data of panels 2 and 5 to show what
has been muted.
The inside and outside corridor stacks (panels 4 and 3, respectively) differ below
the Cardium, Banff, Wabamun, Nisku, and, to a lesser degree, at the Leduc events. The
most plausible explanation is that multiple contamination is significant on the inside cor-
ridor stack but not on the outside corridor stack.
111
o- CARDIUM PEKISKO 0 NISKU

VIKING o BANFF IiiI IR E T O N L.....
, /I/-I....rc-
GLAUCONITIC WABAMUN 0 LEDUC
<
V'J
""'d
-=....
I'D
""l
"e
""l
I'D
.......
<:
I'D
""'d
~
r
1.
3:-
an
I'D
[IJ
~ [IJ
~
N
S·
~
\ ;i
I'D
o
~
~
p..
=
""'d
~
n
.......
n
I'D
,. ... . . ... k, · · <'"';I -~ .•
.,'...:-. .t..;-. "", :.o:~~ui~~·~~(:\~'i ~~ '
Fig. 3.13. Integrated interpretive display (110) showing the interpretation of the available exploration data for the Lanaway Field case
study. The synthetic seismogram was generated using a zero-phase, 30 Hz center frequency Ricker wavelet.
A comparison of the Zup(decon)(+IT) data inside and outside corridor stacks (Figure
3.12) indicates that multiple interference was substantially attenuated by deconvolution.
More specifically, note that the inside and outside corridor stacks of the Zup(decon)(+IT) data
are similar, suggesting that deconvolution has effectively attenuated the multiples so evi-
dent on the inside corridor stack of the Zup(+IT) data (panel 1; Figure 3.11).
Integrated interpretation
The reinterpreted conventional surface-seismic line incorporating the VSP results
(Figure 3.14) is displayed on the left-hand side of Figure 3.13. A synthetic seismogram for
the VSP well, the Zup(+ TI) data inside corridor stack, and the Zup(decon)(+ TT) data out-
side corridor stack are time-tied to the seismic line at the VSP well site (CDP number 127).
The inside and outside corridor stacks and the synthetic seismogram are inserted
between the two separated parts of the seismic section at the VSP well location.
On the right-hand side of Figure 3.13, the VSP data are time-tied to the seismic line, the
Zup(decon)(+ TI) data outside corridor stack, the VSP well synthetic seismogram, the VSP
well velocity log, and the VSP well gamma ray log. The horizontal scale (depth axis) of
the VSP display, Zup(decon)(+ TT), and the scale used for the horizontally oriented VSP well
sonic and gamma ray log depth displays are the same. The outside corridor stack (con-
taining predominantly primary events), the synthetic seismogram, and the two well-logs
(converted to time) allow the interpreter/processor to compare the corridor stack, the
synthetic seismogram, and the well-log data. Because the synthetic seismogram was gen-
erated from the sonic log using a high-frequency wavelet, the synthetic seismogram has
higher resolution than the corridor stack. Since the range of the wavelengths contained
in the Zup(decon)(+ TT) data are the same as for surface-seismic, in most cases the outside
corridor stack will tie to the surface-seismic data reasonably well.
The VSP well sonic and gamma logs are displayed in depth and plotted immedi-
ately above the Zup(decon)<+ TT) data in Figure 3.13. The correlation between these data
can be illustrated by considering the top of the Wabamun. Lithologically, the top of the
Wabamun in the VSP well is represented by the shale/carbonate contact at a depth of
2616 m. On the VSP data, there are recorded VSP traces at 2610 m, and the Wabamun for-
mation top is immediately below them at 2630 m. The Wabamun event is therefore iden-
tified as the trough located in time between the first breaks for the 2610 and 2630-m sonde
depths (approximately 1450 ms). In a similar manner, the other VSP upgoing events can
be identified and then correlated directly to the surface-seismic line at CDP number 127,
the nondeconvolved inside corridor stack, the deconvolved outside stack, the synthetic
seismogram, and the velocity and gamma logs (displayed in time) for the VSP well.
113
The location of the Leduc event is the principal problem faced by the interpreter.
Note that this reflection, on both the surface-seismic and VSP data, is a cycle lower than
initially interpreted (pre-VSP interpretation shown in Figure 3.8). The revised surface-
seismic interpretation is presented in Figures 3.14. The correlation of the VSP and the sur-
face-seismic data indicates that the Leduc top was incorrectly identified at the VSP well
site on the pre-VSP interpretation. This miscorrelation explains why the Leduc top came
in 80 m low relative to prognosis, but it does not explain why the surface-seismic line was
originally misinterpreted. Anomalous Nisku, Ireton, and Leduc surface-seismic signa-
tures between CDP numbers 90 to 140 may have resulted in a misleading interpretation
(Figure 3.14).
This seismic interval from the Wabamun to seismic basement at the VSP well site
is characterized by:
1) positive time-structural relief (on the order of 5 ms) along both pre-Leduc and deeper
post-Leduc events;
2) a low-amplitude Leduc event; and
3) an anomalously high amplitude event immediately above the Nisku event (this was
initially interpreted to be the Nisku reflection).
These three seismic features were initially interpreted to be diagnostic of anomalous
accretionary reef growth.
The VSP well dataset conclusively established that the Leduc was misidentified
on the original interpretation (Figure 3.8) and that the observed seismic anomalies are not
related to late-stage accretionary reef growth. There are several possible causes of this
surface-seismic anomaly; including:
1) structural relief at the Pekisko and Shunda pre-Cretaceous subcrops;
2) stratigraphic anomalies (possible patch reef) within the Winterburn Group (Figure 3.1c);
3) structural thinning of the Ireton near the reef crest (Figure 3.6); and
4) multiple interference within the Wabamun/Nisku interval.
Let us deal with the first possible explanation. It is conceivable that the observed
anomaly is caused, at least in part, by erosional relief at the pre-Cretaceous subcrop. As
illustrated by the velocity log (Figure 3.13), the velocity of seismic wave propagation
within the Upper Mississippian interval is higher than within the overlying Cretaceous.
Positive erosional relief at the Mississippian level would cause pre-Cretaceous events to
be time structurally "pulled-up". The overlying seismic events would be draped as a
result of differential compaction (Anderson et al., 1989a, 1989d). However, the Shunda
Formation was not encountered at the VSP well, as compared to 14 m of Shunda at the 6-
6 well location (Figure 3.15). The combined Shunda/Pekisko thicknesses at the VSP well
114
SCALE 1 KM
SOUTH NORTH...,
::;
en
... ;;;en .., ..,;:;;
'"
:;:
en
A ;;;
en
;;;
-
;:;
en - ;;;
'"
-
:: ;;en 0 "'
en ~
CIl
en -
CIl
en COP
1.0
1.2
~
CARDIUM -
0I
:E
. VIKING
1.4 -<
»
. GLAUCONITIC -l
. PEKISKO
BANFF
~
. WABAMUN <
m
1.6 r
. NISKU -I
• IRETON
LEDUC s::
m
1.8
- 0
-~2.0
Fig. 3.14. Enlarged version of the post-VSP interpretation of the north-south oriented
seismic line (uninterpreted version displayed in Figure 3.7). The location of the VSP well
site is at COP number 116.
1 MILE
SCALE
TWP 36RGE 4 TWP 36 RGE 3 W5M
W5M
11-1 ~o
<>
11-36
Fig. 3.15. Shunda isopach map showing
the absence of Shunda at the VSP well. *
2-36
e
15~5
10-29
I
20 e6-30
TWP35 RGE4 ~o
W5M TWP 35 RGE 3 W5M
SHUNDA ISOPACH (m) ---------
115
and 6-6 are 43 and 55 m, respectively.

A second explanation for the observed time-structure and amplitude anomalies is
a lithologic variation within the Winterburn Group. It is possible that a localized Nisku
Formation (Winterburn Group) patch reef developed above the crest of the underlying
Leduc reef (Rennie et al., 1989). The envisioned carbonate build-up could be character-
ized by a velocity pull-up, time-structural drape, and a character change within the
Winterburn Group. Unfortunately, because of the absence of core control within the
Winterburn Group, this possible explanation cannot be confirmed or negated; however,
it should be noted that the VSP well and well 6-6 intersected 47.4 and 39.3 m of Nisku,
respectively. The additional Nisku build-up and its corresponding porosity could con-
ceivably contribute to the observed character and time-structural anomaly.
A third possibility is that variation in the thickness of the Ireton above the reef
crest at the VSP well (see geologic cross-section; Figure 3.6) could be the cause of the mis-
interpretation. The isopach map of the Ireton in Figure 3.16 shows that the north-south
oriented surface-seismic line (Figures 3.7 and 3.8) overlies thick Ireton shales at well 6-6
(66m). The Ireton thins over the reef crest (43 m at the VSP well) and thickens again
towards well 6-30 (Figure 3.16). The anomalous Nisku to Ireton anomalous isochron val-
ues, evident at the VSP well location, return to normal ranges approximately 640 m south
of the VSP well along the seismic line (Figure 3.16). This location corresponds to COP
number 200 in Figure 3.7 (the location of the intersection of the seismic line and the 100
m Ireton isopach contour; Figure 3.16). The correlation between the seismic anomaly at
the Nisku, Ireton, and Leduc levels to the Ireton isopach suggests that the thickening of
the Ireton (away from the VSP well site) caused an anomalous tuning effect that con-
tributes to the surface-seismic event anomaly (drape and pullup) at the VSP well.
The initial misinterpretation of the seismic events at the VSP well site is probably
caused by a combination of the thinning of the Ireton Formation and an increasingly
thick and porous Nisku interval. Additionally, the inside and outside corridor stacks of
both the Zup(decon)(+ T'T) and the Zup(+ TT) data (Figures 3.11 and 3.12) show that there is
multiple contamination immediately below the Wabamun event that could have con-
tributed to the amplitude anomaly.
Other interesting observations can be made from the integrated sonic log, VSP
corridor stacks, and the seismic data in the immediate vicinity of the VSP well. The
Glauconitic-to-Banff sequence and top of the Leduc are poorly resolved on the surface-
seismic data (relative to the outside corridor stack). One possible explanation is that the
surface-seismic signatures of events within the Glauconitic-to-Banff interval may be
degraded as a result of multiple (interbed) interference.
116
1 MILE
SCALE
TWP 36 RGE 4 W5M
16-1 -¢-
11-1
-¢-
11-36
*
2-36
•
IRETON ISOPACH (m)
Fig. 3.16. Ireton isopach map showing the drape of the Ireton
shales along the example seismic line (shown in Figure 3.14).
The data presented in Figure 3.13 allow for the confident reinterpretation of the
surface-seismic data in the vicinity of the VSP well and offers possible origins for the
observed anomaly. These reinterpreted seismic data are displayed as Figure 3.14. (The
anomaly in the immediate vicinity of the VSP well is enlarged in scale and displayed in
Figure 3.14.)
Summary
The VSP well was drilled into the Leduc reef at Lanaway Field, south-central
Alberta, in order to evaluate an anomaly observed on surface-seismic data. The pre-well
interpretation of these seismic data suggested that the VSP well would encounter up to
117
80 m of anomalous accretionary reef growth at the Leduc level. However, drilling con-
firmed that the Leduc at the VSP well was more or less regional and indicated that the
initial interpretation was inaccurate as the seismic events were a response to something
other than anomalous Leduc buildups and corresponding anomalous Ireton thinning. To
explain the discrepancy between the pre well seismic interpretation and the drilling
results, a near-offset VSP was run at the VSP well site.
The interpretively processed VSP data provided invaluable information regard-
ing the seismic anomaly. On the basis of these data, it was possible to:
1) establish that the Leduc event had been miscorrelated on the pre-well surface-seismic
interpretation;
2) correctly identify the Leduc event in the surface-seismic data at the VSP well site; and
3) further elucidate the nature of the misinterpreted anomaly on the surface-seismic line.
On the basis of VSP data and the integrated interpretation, the observed seismic
anomaly at the VSP well can be attributed (at least in part) to:
1) anomalous erosional relief at the pre-Cretaceous subcrop;
2) localized patch reef development within the Winterburn Group;
3) tuning resulting from thinning of the Ireton Formation in the vicinity of the VSP-well;
and
4) multiple interference within the Wabamun/Nisku interval.
The most likely effects are Winterburn patch reef development and tuning effects.
118
4 - Ricinus Field Case Study
Overview
On the basis of the interpretation of conventional surface-seismic data, an
exploratory well (referred to as the VSP well) was drilled in the Ricinus Field, southern
Alberta, Canada. Prior to drilling, the prognosis was that the VSP well had a reasonable
chance of encountering gas-bearing Leduc Formation reef (the northeastern updip mar-
gin of the known full reef). The known full reef had been defined by existing wells as
shown in Figure 4.1. However, the VSP well encountered only off-reef shale and was ulti-
mately abandoned. The final interpretation was that the VSP well had been drilled some
800 m northeast of the full reef build-up.
Prior to the abandonment of the VSP well, two VSP surveys were run at the VSP
well site. These data were acquired to resolve the apparent discrepancy between the inter-
preted surface-seismic data and the actual geology at the VSP well site, and to evaluate the
feasibility of whipstocking the VSP well to the southwest in the direction of the known full
reef complex. One of the VSP surveys had a source offset of 199 m (near-offset), the other
had a source offset of 1100 m (far-offset). The VSP data were definitive and allowed for a
more confident and geologically consistent interpretation of the surface-seismic data, and
clearly indicated that whipstocking was not an economically viable option.
Introduction
The Devonian stratigraphy at the Ricinus Field is similar to that at Lanaway Field
(Figures 3.1A, H, and C). The Leduc Formation at Ricinus Field (Figures 3.2 and 4.1)
developed as a large atoll that towers some 350 - 400 m (based on well 6-9-34-8 W5M)
above the Cooking Lake platform. It exhibits a mappable raised rim and a structurally
lower central lagoon. As at Lanaway Field, the updip edge (northeast) of the raised rim
at Ricinus is productive where structurally closed and effectively sealed by the inter-reef
shales of the Duvernay and Ireton. The schematic cross-section shown in Figure 4.2 illus-
trates the interpreted morphological relationships between the Leduc and the inter-reef
shales of the Ireton and Duvernay in the Ricinus area.
In those areas of western Canada where the Devonian and/or overlying rock units
are relatively undisturbed structurally, full Leduc reefs and inter-reef shales can usually be
differentiated on reflection seismic data. The carbonate build-ups are typically character-
ized by appreciable velocity pull-up (up to 25 ms), significant time-structural drape at the
119
TWP 34 RGE 8
32 33 34 35 36
,..,.'.'.'.'.'
29 28 27 ~ .... 26 25 Fig. 4.1. Detailed map of the Ricinus
'" o>.~!: p "0
0 study area showing the location of the
\.-:?v
"'IS'o ....-<> ~
6-23 6-24 wells used in the geological schemat-
>. 5-22 : <>- -o- ic sections shown in Figures 4.2 and
20 ..··21 : 22 23 24
...... 4.4, the seismic data shown in
.... Figures 4.3, 4.4 and 4.17, and loca-
........., 7-15;¢-
."
6-14
n- tions for Leduc Formation level wells
./.. 17 16 15 14 13 in the Ricinus Field area (from Hinds
/
et aI., 1993c).
p.
6-9
8 9 10 11 12
V
5 4 3 2 6-1 1
GEOLOGIC CROSS-SECTION •••.••• 1 MILE
SEISMIC SECTION •••••••-. SCALE
top of the Devonian (up to 70 ms at Ricinus), and reflection waveform variations within the
Woodbend Group (Anderson, 1986; Anderson and Brown, 1987). Back from their steeply
dipping margins, the tops of these reefs are generally manifested as high-amplitude
troughs on reverse-polarity seismic data. The seismic signatures of low-relief reefs (Leduc
buildups with thicknesses on the order of 100 m) are significantly more subtle. The exam-
ples of low-relief reefs described by Anderson et al., (1989a, 198%) and Hinds et al., (1993b)
are characterized by less than 10 ms of velocity pull-up and less than 20 ms of time-struc-
tural drape. Additionally, the reflections from the tops of such reefs are often difficult to dif-
ferentiate from inter-shale events. In those areas where extensive subsurface structural
deformation has occurred, even the seismic image of the full Leduc reef can be effectively
masked by the superimposed seismic signature of the structural complexities. Thrust
faulting within Mesozoic strata in the general area of the Ricinus reef, for example, can
significantly affect the seismic signature of Leduc reefs.
Ricinus full Leduc reef

The geologic section incorporating wells 6-9, 7-15, and the VSP-well (Figure 4.2),
summarizes the geological and geophysical interpretation at the VSP well site prior to the
drilling of the VSP well. This cross-section was based on the initial (pre-VSP well) interpre-
tation of the example seismic data (Figure 4.3) and on well control available at that time.
120
Ricinus Field Case Study
SOUTHWEST NORTH
6-9 7-15 vsP
~ ~
~
.3
BANFF
E 2500
Ql
~
«
w
en
rn
::::l
en 300
3500
Fig. 4.2. Schematic section depicting the envisioned subsurface geology at the VSP well
site prior to the drilling of the VSP well. This interpretation is based on downhole data
from wells 6-9-34-8 W5M and 7-15-34-8 W5M and the seismic interpretation shown in
Figure 4.3. The preferred, current interpretation is shown in Figure 4.4.
1.8
~
0I
- 1.9
~
-<
-l
-2.0
~
<
m
roo
- 2.1 -l
~
m
en
2.2
-2.3
2.4
BLAIRMORE COALSD IRETON • COOKING LAKE .
SCALE NORDEGG • INTER-IRETON 0 ACOUSTIC •
WABAMUN • LEDUC 0 BASEMENT
Fig. 4.3. Interpretation of the example seismic data prior to the drilling
of the VSP well (at COP 258). The data are normal polarity.
121
The location of the seismic sections used in the initial interpretation of the area
(Figures 4.3 and 4.5) is shown in Figure 4.1. The interpreted 12-fold surface-seismic data
displayed in Figures 4.3 and 4.5 were acquired using a source pattern consisting of five
1-kg charges spread over 60 m (at a single shotpoint). The shotpoint location interval was
120 m; the average shot depth was 9 m. The geophone groups consisted of nine in-line
14-Hz geophones over 30 m; the group interval was 30 m; 96 traces were recorded using
DFS-V recording equipment and a split-spread geometry with a near-offset of 30 m. The
field anti-aliasing filter for the surface-seismic was OUT 1128 Hz. The surface-seismic
datum in the area was 1400 m ASL. The refraction statics replacement velocity used to
reduce the surface-seismic data to the seismic datum was 3350 m/s.
In the initial interpretation of the surface-seismic data, the Leduc reef was thought
to be fully developed at the site where the VSP well was drilled. Ultimately, drilling con-
firmed that the VSP well site was off-reef and that this initial seismic-based interpretation
was incorrect. On the initial interpreted version of the seismic data (Figure 4.3), the north-
ern edge of the Leduc reef is located near trace 242, and full reef is mapped as present at
the VSP well site at trace 259. This plausible, though incorrect interpretation is supported
by the patterns of time-structural relief observed along the more prominent seismic events.
For example, the events referred to as near-Cambrian and Cooking Lake are pulled up by
15 ms immediately south of trace 242. The Ireton, Wabamun, Nordegg, and Blairmore
drape by up to 25 ms across the interpreted northern edge of the reef. Note that the time-
structure at the Blairmore and Nordegg events to the south of trace 314 in Figure 4.3 has
been incorrectly attributed to thrust faulting within the Mesozoic section. In this interpre-
tation, the Blairmore Coals and Nordegg events to the south of trace 290 are shallower in
time compared to the same events to the north of trace 290. The Ricinus Field is situated
near the eastern limit of Mesozoic thrust faulting in the Western Canadian Basin, so this
misinterpretation is consistent with the regional geology (see Figure 5 of Moore, 1989a).
The geologic section of Figure 4.4 illustrates the morphology of the Ricinus Field
as interpreted after the drilling of the off-reef VSP well. This geologic section is con-
strained by well control. It is based on the VSP-data, the post-VSP interpretation of the
example surface-seismic line (Figure 4.5), and the drilling information from the VSP well.
Wells 6-9 and 7-15 (Figure 4.1) were drilled into a fully developed reef and are produc-
tive; 6-9 encountered approximately 22 m of gas pay within the Leduc; 7-15 encountered
140 m of pay. The VSP well (Figures 4.1 and 4.4) is off-reef and encountered a full section
of inter-reef shale (Ireton and Duvernay; Figure 3.1A, B and C). Well 6-9 and the VSP well
encountered Cooking Lake; however, 7-15 was not drilled deep enough to penetrate the
Cooking Lake Formation.
122
SOUTHWEST NORTH
,
)"
6-9
'-".
S
~.;..
.J
.:i
~
7-15
~
VSP
ELK
SHUNOA
K K
250 f
1
BANFF
.,E
'l'
«UJ .\ WABAMUN
-,
rn
l:ll
:>
rn
"
3000
LEDUC
BEAVERH'LL LAKE
SCALE
~ ELK PO,NT
350
Fig. 4.4. Schematic section depicting the subsurface geology at the VSP well site, and
the relationships between wells 6-9 and 7-15 (locations shown in Figure 4.1) and the VSP
well. The geologic section is consistent with available well log control and with the seis-
mic interpretation displayed in Figure 4.5.
SOUTHWEST
COP 330 310
~) m:'rl"'T'lm~+rJr.~'::":":'I"~~IT-rr~-r-F--.~
1.8 ~
o
1.9 ~
~
-t
2.0 ~
m
~
2.1 ~
2.3
BLAIRMORE COALSD IRETON • COOKING LAKE .

SCALE
NORDEGG • INTER-IRETON D ACOUSTIC •
500 M WABAMUN • LEDUC D BASEMENT
Fig. 4.5. Current, preferred interpretation of the exampie seismic data shown within the
integrated log display (ILD; Chapter 2). This interpretation is consistent with the 6-9, 7-15,
and VSP wells (as shown in Figure 4.4). The VSP well log sonic (displayed in time) and
synthetic seismogram are inserted to display the correlations used in the interpretations.
123
In the post-VSP interpretation of the seismic data (Figure 4.5), the northeastern
edge of the Leduc complex is located near trace 314, and the VSP well site (trace 259) is
off-reef. This interpretation is supported by the patterns of time-structural relief observed
along the more prominent seismic events. For example, the Acoustic Basement event is
pulled up by about 40 ms to the south of trace 314; the Ireton, Wabamun, Nordegg, and
Blairmore events drape by as much as 70 ms across the interpreted northeastern edge of
the reef as a result of the differential compaction of reef and off-reef sediments (not as a
result of Mesozoic thrust faulting).

After the analysis of the well-log data and prior to abandonment, two vSP sur-
veys were recorded at the VSP well site. These two VSP surveys were designed to more
accurately tie the surface-seismic to the subsurface geology, determine if the reef crest
was within 500 m west of the VSP well (with a view to possible whipstocking), and to dif-
ferentiate primary reflections from both surface-generated and interbed multiples.
The near-offset source was 199 m from the VSP well, the far-offset was 1100 m.
Both sources were on-line with respect to the surface-seismic profile (Figure 4.1) and in
the direction of the 5-22 well. Two vibroseis units were operated in tandem at each source
location. The 16-s sweep ranged from 8 to 80 Hz, the recording length was 20 s, and the
cross-correlated output was 4 s. Six to eight sweeps were summed for each geophone
sonde location. An sse 1078 VSP recording system with a sampling rate of 2 ms and a
recording filter setting of OUT lOUT was used.
The total depth of the VSP well was 4528 m below KB (KB was 1317 m asl). The
source elevation of both offset source points was 1304 m asl. Data were recorded at
selected depths as the sonde was lowered down the borehole; these sonde locations were
reoccupied during the production run. The use of these dual recording locations facili-
tated the detection of cable stretch or cable depth counter malfunction, and provided
information regarding the gain amplification that was required during the VSP produc-
tion run. The first production VSP recordings were 4260 m below KB for the near-offset
VSp, and 4350 m for the far-offset VSP survey. During the production recording, the geo-
phone sonde was raised at intervals of 30-m intervals to a depth of 420 m below KB. The
sonde interval spacing was 60 m for the near-offset and far-offset VSP from 420 m to the
surface (KB). A wall-damped 3-component geophone tool was used.
124
Near-offset (199 m) VSP interpretive processing

During the processing of the near-offset VSp' a series of interpretive processing
panels were generated to illustrate:
1) upgoing and downgoing P-wave separation;

2) deconvolution of the Zup(+TT) data using an inverse filter calculated from the
Zdown(+TT) data; and
3) inside and outside corridor stacks of Zup(+TT) and Zup(decon)(+TT) data.
P-wave separation
The separation of upgoing and downgoing P-waves from the Z(FRT) data is
depicted in the wavefield separation interpretive processing panel (IPP) of Figure 4.6.
Panel 1 displays the Z(FRT) data after trace normalization. In panel 2, these Z(FRT) data
have been gained to highlight prominent primary upgoing events such as the Blairmore
Coals, Nordegg, Ireton, and Acoustic Basement (crystalline Precambrian) events. Note
that the upgoing event identified as the Acoustic Basement does not intersect the first-
break curve and cannot be positively identified as a primary reflection (see Chapter 1).
The exact traveltime and depth to crystalline basement cannot be accurately determined
without VSP control at that depth. Indeed, it is possible that the downgoing surface-gen-
erated multiple (that appears deeper in time than each first-break event by approximate-
ly 0.3 s) may have generated the interpreted basement reflector. On the Z(FRT) data in
panel 2, the downgoing multiple event that intersects the deepest trace at 1.05 s coincides
with the upgoing wave that begins at 1.05 s and ends on the shallowest trace at 1.65 s.
In panel 3 (Figure 4.6; Z(-TT) data), the first-breaks and downgoing P-wave mul-
tiple events are time aligned. These Z(-TT) data illustrate that the downgoing wavetrain
consists of the primary downgoing wavelets plus high-amplitude, surface-generated
multiples and less prominent interbed multiples. The surface-generated downgoing mul-
tiples are recognized as those time aligned, post-first-break arrivals that are recorded on
all of the traces. As discussed in Chapter I, if a downgoing multiple event does not
extend over the entire depth range but is evident on the deeper traces only, then that mul-
tiple is an interbed multiple.
In the next processing step, an Tl-point median filter was used to remove the
upgoing P-waves. The output, consisting of separated and scaled Zdown(-TT) data, is dis-
played in panel 4. Note that the residual upgoing wave content in the Zdown(-TT) data is
minimal. This panel is one of the most important panels for the interpretive processing of
VSP data. If residual upgoing events remain in the Zdown(-TT) data, then that amount of
residual upgoing event is subtracted out of the Z(-TT) data (paneI3) during wavefield
125
mmmmmmummmmmmmmumm mmmmmmummDEPTH (m)

"i I ' ,I~ rl'I"I'iI'I''il-~;:"~ ----- 0.0 BLAIRMORE COALS D
NORDEGG
0.5 -i WABAMUN •
~
IRETON •
miNTER-IRETON D
~ LEDUC D
~ COOKING LAKE •
ACOUSTIC
1.5 ~ BASEMENT
2.0
0.5 -i
~
m
r-
-i
~
m
(4) Zdown(-TT) (6) Zup(+TI)
Fig. 4.6. Interpretive processing panel depicting the wavefield separation of the near-
offset VSP data.
mmmmmmmmu mmmmmmmnm mmmmmmmmu OEPTH ( m )

.:t. 0.0
• I BLAIRMORE COALSD
NORDEGG •
0.5 :il WABAMUN •
~ IRETON •
m
1.0 ~ INTER-IRETON D
~ LEDUC D
m COOKING LAKE •
1.5 en ACOUSTIC
BASEMENT
(1) Z( +TT)
2.0
(2) Z( +TT)
median filtered
0.5
1.0
~1'5 .,
(7) Z uP(deoon)(+TT )
median filtered
Fig. 4.7. Interpretive processing panel depicting the deconvolution of the near-offset
VSP data.
126
separation. The prominent multiples that appear at traveltimes less than 0.455 s are promi-
nent surface-generated multiples. At greater traveltimes, surface-generated and interbed
multiples are present (as seen between 0.75 and 0.9 s; panel 4).
In the next step of the wavefield separation, the Zdown(-TT) data of panel 4 were
subtracted from the Z(-TT) in panel 3 to yield the output Zup(-TT) data of panel 5. The
upgoing P-waves and downgoing shear waves (both primary and multiples) are shown
in panel 5. The downgoing shear waves (5V) were generated in the shallow subsurface
(possibly at the bottom of the surface casing). One downgoing 5V-event appears on the
shallowest trace on panel 5 at 0.2 s (-TT time) and trends opposite to the upgoing events
(deeper in time from left to right).
The Zup(+TT) data before and after the application of a 3-point median filter are
shown in panels 6 and 7, respectively. The equalized amplitude events of the time aligned
upgoing Blairmore Coals, Nordegg, Wabamun, Ireton, inter-Ireton, Cooking Lake, and
Acoustic Basement reflections are interpreted in panel 7. The downgoing 5V-events in
panel 5 dip more steeply in the +TT display; however, these events have been effectively
attenuated by the application of the median filter. As discussed below, because of multi-
ple interference, the Wabamun and Cooking Lake events can be identified only on traces
deeper than the depth of the Nordegg interface.
VSP deconvolution
The deconvolution IPP (Figure 4.7) was designed to monitor the deconvolution of
the Ricinus Zup data. The incorporated panels reveal information (about multiples) that
was difficult to determine from the wavefield separation IPP (Figure 4.6) alone. The first
two panels (Figure 4.7) are the nonfiltered and median-filtered Zup(+TT) data, respec-
tively. Panel 3 contains the Z(-TT) data and allows the examination of the downgoing P-
wave multiple pattern of the near-offset data. Panels 4 and 5 contain the Zup(-TT) and
Zup(decon)(-TT) data, respectively. Both panels 4 and 5 are contaminated with downgoing
shear waves. A comparison of these panels illustrates that deconvolution has enhanced
the higher frequencies and preserved the primary reflections, without appreciably
decreasing the signal-to-noise ratio.
The last two panels (6 and 7) contain the nonfiltered and median-filtered
Zup(decon)(+TT) data, respectively. A comparison of panels 2 and 7 (Figure 4.7), illustrates
how Blairmore Coals and Nordegg multiples affect the continuity of primary reflections.
Multiple contamination most noticeably affects the Wabamun and Cooking Lake events.
On the Zup(+TT) data in panel 2, the Wabamun is relatively unaffected on traces recorded
below the Nordegg depth of 3466 m (-2151 m ASL). The upgoing multiple reflection from
127
the Nordegg is not detected on traces deeper than the bottom-generating layer of the
multiple (the Nordegg). Examination of panel 3 reveals a series of interpreted surface-
generated multiples [a series of time-delayed events parallel to the first-break primary
(-TT)]. Multiple contamination persists for 0.4 s or more. Both surface-generated and
interbed multiples could be contaminating the Zup<+ TT) data at the Wabamun level. On
traces recorded at depths shallower than 3466 m, the multiple events generated at the
Nordegg Formation interface destructively interfere with the Wabamun event.
The Cooking Lake event is located immediately below the inter-Ireton reflector
(panel 2; Zup(+ IT) data) on the traces recorded below depths of 3466 m. The identifica-
tion of the Cooking Lake event is supported by the sonic log/synthetic seismogram cor-
relation to the surface-seismic profile (ILD; Figure 4.5). The sonic log was used to create
a synthetic seismogram (corrected using the VSP first-break times).
On the deconvolved data (panel 7; Figure 4.7), the Wabamun event is laterally con-
tinuous and can be correlated across all depth traces. The Cooking Lake event (trough) can
also be confidently correlated across the entire Zup(decon)(+IT) dataset. Multiple contamina-
tion is minimal in the zone of interest (around the Ireton event). The inter-Ireton reflector is
continuous and does not exhibit significant time structural relief either before or after VSP
deconvolution. The inter-Ireton event truncates against the edge of the Ricinus reef and is
regarded as key to the interpretation of the VSP and surface-seismic data.

Inside and outside corridor stacks and associated displays for the nondecon-
volved Zup(+ IT) data are presented in Figure 4.8. A comparison of the Zup(+TT) outside
and inside corridor stacks (panels 3 and 4, respectively) illustrates the utility of these dis-
plays. For example, the interbed multiple interference generated by the Blairmore Coals
or Nordegg is present on the inside corridor stack but not on the outside corridor stack.
On the outside corridor stack of the Zup(+ TT) data (panel 3), the Wabamun trough dom-
inates (at 1.85 s); however, because of multiple interference, this event is masked on the
inside corridor stack.
On the unmuted input data for the corridor stack displays (Figure 4.8, panels 1
and 6), the Cooking Lake event is a trough. Prior to deconvolution, the event was so weak
that it could not be confidently mapped even on the outside corridor stack. This illus-
trates the reason we display the unmuted input data in the corridor stack IPP during
interpretive processing. Plotting the unmuted data shows us where the corridor mute
line is located. On the Zup(decon)(+ IT) data corridor stack panels of Figure 4.9, the
Wabamun and Cooking Lake events can be confidently interpreted.
128
(1) Zup(+TT) (2) Zup(+TI) (3)(4) (5) Zup(+TT) (6) Zup(+TT)

median filtered muted outside corridor muted inside corridor median filtered
BLAIRMORE COALSD (3)Zup(+TT) (4) Zup(+TT) LEDUC 0
NORDEGG • COOKING LAKE •
outside corridor stack inside corridor stack .
WABAMUN • ACOUSTIC •
IRETON • BASEMENT
INTER-IRETON 0
Fig. 4.8. Interpretive processing panel illustrating the utility of the nondeconvolved inside
and outside corridor stacks for the Ricinus near-offset Zup(+ TT) data.
1.75
-l
~
m
r-
2.0 -l
~
-
m
1J>
2.25
(2)ZuP(deconl(+TT) (3)(4) (5) Zup(deconl(+ TT) (6) ZUP(decon)(+TD

muted outside corridor muted inside corridor median filtered
(3) Zup(deCOn)(+TT) (4) Zup(decon)(+TT)
Fig. 4.9. Interpretive processing panel illustrating the utility of the deconvolved inside and
outside corridor stacks for the Ricinus near-offset Zup(decon)(+ TT) data .
129
If deconvolution is successful, the Zup(deconl(+ TT) outside and inside corridor

stacks (panels 3 and 4; Figure 4.9) should be similar. The inside and outside deconvolved
corridor stacks show that the surface-generated multiples have been substantially attenu-
ated by deconvolution; however, there is enough difference (as in the zone immediately
below the Nordegg event at about 1.75 to 1.8 s) between the two stacks to suggest that
interbed multiples have been attenuated but not totally removed.
The deconvolved data and the corridor stacks show that the Wabamun and
Cooking Lake primaries are flat lying, continuous, upgoing primary events across the
entire VSP panel. At this point, the VSP and surface-seismic data can be confidently cor-
related at the well site.
Far-offset (1100 m) VSP interpretive processing

On the far-offset VSP data, the vertical (Z) and both horizontal (X and Y) axes con-
tain nonpartitioned elements of the upgoing and downgoing P- and SV-wavefields. We
present an examination of the IPPs that reveal the partitioning of the wavefields (both
upgoing and downgoing P-and SV-events) has significant implications with respect to
interpretation. In Chapter 2, the Ricinus far-offset data were presented as an example of
the interpretive processing of problematic VSP data. These data showed how interpretive
processing aids in the determination of the origins of problematic "noise" wavefields
within far-offset data. In this section, interpretive processing will be discussed in more
detail. The reader is referred to Chapter 2 for information pertaining to the differences in
runstream approaches.
As demonstrated in panels 5 and 6 of the near-offset wavefield separation IPP
(Figure 4.6), a downgoing SV-wave (trending in the opposite direction to the upgoing
waves in panelS) exists in these data. Normally, downgoing and upgoing shear waves
are not seen on near-offset data because near-offset surveys usually contain near-vertical
incidence reflections. According to the Zoeppritz equations, the mode-conversion of
downgoing P- waves into upgoing and downgoing SV-waves should be minimal under
conditions of near- vertical incidence reflections. SV-events should be more noticeable on
far-offset data. The runstreams used for the partitioning of the wavefields play an impor-
tant role in isolating the upgoing P-wave events (from downgoing P-wave events and
from upgoing and downgoing SV-events). As illustrated in Chapter 2, interpretive pro-
cessing assists in the modification of the processing runstreams.
Far-offset IPPs were designed to monitor the following procedures:
1) hodogram-based rotation of the X, Y, and Z data (based on windowed data that are
centered on the P-wave first arrival);
130
BLAIRMORE COALS 0
NORDEGG •
0.50 WABAMUN •
IRETON •
-t INTER-IRETON D
~
1.00 LEDUC 0
m
COOKING LAKE •
~
~ ACOUSTIC
m BASEMENT
1.50
2.00
(1) Z(FRT) (2) Z(FRT) (3) Z(-TT)
gained
0.5
-t
1.0 ~
~
~
m
1.5 -;;
2.0
(4) HMIN(FRT) (5) HMAX(FRT) (6)Z' (FRY) ( 7) HMAX' (FRT)
Fig. 4.10. Interpretive processing panel depicting the hodogram-based rotation of the
far-offset Ricinus VSP data.
~';"";;::-+.I O.B -t
~m
1.2 ~
~
m
2.0
(1) Z'up(FRT) (2) HMAX'up(FRT) (3) HMAXup(derot)(FRT)
-t
~m
1.2 ~
~
m
2.0
(5) HMAX"up(FRT) (6) Z';,p(FRT)
Fig. 4.11 . Interpretive processing panel depicting the time-variant model-based rotation
of the far-offset Ricinus VSP data resulting from the suggested processing runstream in
Chapter 2.
131
2) time-variant model-based rotations applied to the HMAXup(FRT) and Zup(FRT) data

in response to interpretive processing results as highlighted in Chapter 2 (within the
section "problematic far-offset interpretive processing"); and
3) VSP-CDP mapping of the data.
Hodogram-based rotation
The X(FRT), Y(FRT), and Z(FRT) data for the far-offset VSP are displayed in
Figure 4.10 as panels I, 2, and 3, respectively. The X(FRT) and Y(FRT) data contain both
P- and SV- downgoing waves plus recognizable upgoing SV-events. These upgoing SV-
events originate at depth levels from 2790 m and lower (colored purple in panels 1 and
2). Up going SV-events could be present on shallower traces but are masked by the down-
going mode-converted SV-events. The HMAX(FRT) data in panelS show upgoing and
downgoing SV-events. These events (highlighted in panels 5 and 6) differ in slope rela-
tive to the upgoing P-wave events of the Z(FRT) data displayed on panel 3.
The partitioned downgoing primary P-waves (first-breaks wavelet) are consistent
on both the X(FRT) and Y(FRT) data in panels 1 and 2 on traces recorded in the upper
two-thirds of the borehole. They are less consistent at deeper depth traces which indi-
cates that the geophone tool was rotating. The Z(FRT) data first-breaks in panel 3 are
phase consistent.
The hodogram-based rotation of the X(FRT) and Y(FRT) data onto the
HMIN(FRT) and HMAX(FRT) data is illustrated using panels I, 2, 4 and 5, respectively.
This rotation corrects for first-break inconsistencies (because of the rotation of the tool
during the movement of the sonde up the borehole) by projecting data from both of the
input channels onto an axis which lies in the plane defined by the borehole and the
source. The rotated data are referred to as the HMAX(FRT) data. HMIN(FRT) and
HMAX(FRT) data are assumed to be aligned perpendicular to, and in the plane contain-
ing the source and wellbore, respectively. Note that the HMIN(FRT) data (comprised of
horizontally polarized shear (SH) wave events and out-of-the-plane reflections) contain
possible downgoing SV-waves (highlighted in panel 4 and appearing on the shallow
traces below 0.6 s) that could originate at the casing joints and reflected downgoing
waves that originate from out-of-the-plane of the well.
The remnants of a mode-converted downgoing SV-wave with components within
the HMIN(FRT) data appears at the Nordegg and Blairmore levels. The upgoing SV-
events from the Nordegg and Blairmore Coals (first noticed on both of the X(FRT) and
Y(FRTI data) are accentuated after redistribution onto the HMAX(FRT) data. These upgo-
ing SV- events are highlighted on panelS. Unfortunately, the downgoing SV-events on the
132
HMAX(FRT) data have nearly the same amplitude as the downgoing P-waves. The
HMAX(FRT) data contain consistent, polarized, downgoing P-wave first-breaks because
the input X(FRT) and Y(FRT) channel data have been rotated into the plane containing
the downgoing P-wave.

The Z'(FRT) data in panel 6 and the HMAX'(FRT) data in panel 7 were generat-
ed by rotating the Z(FRT) and HMAX(FRT) data using polarization angles estimated
from a hodogram analysis of data within a window around the P-wave first arrival. This
technique is designed to polarize the data so that the downgoing P-waves are effectively
isolated on a single channel, HMAX'(FRT). On the HMAX'(FRT) data in panel 7 of
Figure 4.10, downgoing P-wave energy is dominant. The HMAX'(FRT) panel contains
residual downgoing shear (SV) waves (converted to SV-waves at the Blairmore coal
interface) and upgoing P-wave energy. This indicates that the polarization required to
delineate the downgoing P-wave data onto a single panel is not adequate to separate the
upgoing P-wave data entirely onto the orthogonally aligned Z'(FRT) data.
The Z'(FRT) data shown in panel 6 contain downgoing shear wave (highlighted in
panel 6) and upgoing P-wave energy that fit the assumption that the primary upgoing P-
waves are orthogonal to the primary downgoing P-waves (isolated on the HMAX'(FRT)
data). The downgoing shear wave is more robust than the lower amplitude upgoing P-
wave. The time-variant polarization of the data (modeled on the upgoing P-waves) should
detach the contaminating upgoing and downgoing shear waves from the upgoing P-
waves and isolate interpretable upgoing P-wave events onto a single output data panel,
ZItup(FRT). However, this "routine-type" processing did not produce optimal results for
interpretation, and the use of a modified processing runstream was necessary.
Time-variant model-based rotation

The normal time-variant rotation runstream recommended in Chapter 2 was used
for the Ricinus far-offset data. The output time-variant polarization IPP is shown in
Figure 2.50 and interpreted in Figure 4.11. However, the ZItup(FRT) data (in panel 6 of
Figures 2.50 and 4.11) contain unacceptable contamination due to diffractions, and mode-
converted SV upgoing and downgoing events.
The time-variant model-based polarization runstream (as suggested in Chapter 2)
was changed to eliminate the second polarization rotation. In the modified runstream, the
ZItup(FRT) output data are generated through wavefield separation and time-variant rota-
tion of the Z(FRTI and HMAX(FRT) data. The HMAX'(FRT) data displayed in Figure 4.10
are used in far-offset deconvolution of the ZItup(FRT) data. The updated, interpretively
processed, time-variant, model-based, polarization results are shown in Figure 4.12.
133
The HMAX(FRT) and Z(FRT) data (panels 1 and 2; Figure 4.12) are wavefield sep-
arated using the f-k based surgical muting (see Chapter 2) to output the HMAXup(FRT)
and Zup(FRT) data of panels 3 and 4, respectively. The f-k mute was designed to attenu-
ate the upgoing and downgoing 5V-events. The lack of the second hodogram-based rota-
tion on the Z(FRT) and HMAX(FRT) input data (used in the wavefield separation) elim-
inated the possibility of excessive rotation-induced noise.
These data (panels 3 and 4 of Figure 4.12) are compared to the final two panels (5
and 6) containing the time-variant model-based polarized HMAX"up(FRT) and
ZItup(FRT) data, respectively. An improvement is noted in the quality of the upgoing P-
wave events on the ZItup(FRT) panel in Figure 4.12 compared to the ZItup(FRT) generat-
ed through the use of the normal processing runstream (without interpretive processing)
shown in Figure 4.11. The upgoing events on panel 6 of Figure 4.12 are readily inter-
pretable; however, the same events on panels 5 and 6 of Figure 4.11 are difficult to inter-
pret because of interfering noise. The updated ZItup(FRT) contains residual diffractions
(highlighted between 0.7 and 0.9 s) which manifest themselves as horizontal events.
On the Zup(FRT) data in panel 4 of Figure 4.12, upgoing P-waves generated by
shallow reflectors are improperly aligned (because of the choice of time-stationary rota-
tion angles). The deeper events do not suffer much misalignment because deep-event
raypath geometries satisfy the near-vertical incidence angle assumption better than do
the raypaths of shallower events. The time-variant, model-based rotation corrects for
such misalignment. The output upgoing wave displays, HMAX"up(FRT) and ZItup(FRT),
are shown on panels 5 and 6 in Figure 4.12, respectively. Note that the shallow events dis-
play better alignment than on the Zup(FRT) in panel 4. The rotation angle required for the
Blairmore Coals event on a particular trace was different than the rotation angle for deep-
er events on the same trace.
Deconvolution offar-offset data

A realistic approximation to far-offset VSP deconvolution utilizes the separated
polarized downgoing waves, HMAX'down(-TT), and the separated time-variant polar-
ized upgoing P-waves contained in the ZItup(FRT) data (panel 6 of Figure 4.12). A far-off-
set deconvolution IPP was designed to assist in the evaluation of the effect of far-offset
deconvolution on the ZItup data (Figures 2.56 and 4.13). The process involved designing
the inverse wavelet for the HMAX'down(-TT) data (panel 2 of Figure 4.13) and subse-
quently applying the operator to the ZItup(-TT) data (panel 4 of Figure 4.13).
Deconvolution was performed on the ZItup(-TT) far-offset data (see Figures 2.56
and 2.57); however, the process added unacceptable noise to the data. This can be seen by
134
-l
~
m
r
1.2 -l
3:
m
~
1.6 '"
2.0
0.8 -l
1.2
~~
i:
m
1.6
2.0
(4) Zup{FRT) (5) HMAX" up(FRT) (6) Z"up{FRT)
Fig. 4.12. Interpretive processing panel depicting the time-variant model-based rotation
of the far-offset Ricinus VSP data resulting from interpretive processing.
-0.2
-I
- 0.4 rn ~
r-
-I
3:
- -0.6 m
C/l
-- 0.8 ~
~' 'j ;,
(1) HMAX'(-TI) 1.0
(3) HMAX'down(decon)(-TT)
BLAIRMORE COALS D
NORDEGG II
-l WABAMUN II
0.4 m ~ IRETON
INTER-IRETON
II
D
!:i LEDUC D
3:
0.6 m COOKING LAKE II
ACOUSTIC II
0.8 ~
BASEMENT
1.0
(5) Z"up(de<on){-TT)
Fig. 4.13. IPP depicting the far-offset deconvolution of the Ricinus VSP data.
135
comparing the Z"up(-IT) and Z"up(decon)(-IT) data shown in panels 4 and 5 of Figure 4.13.
The upgoing P-wave events are difficult to interpret on the deconvolved data.
In the final analysis, the non deconvolved far-offset VSP data were determined to
contain minimal multiple contamination at the zones of interest on the Z" up(+TT) data (on
the Blairmore Coals through Acoustic Basement interval shown in panel 6 of Figure 4.12).
On the far-offset VSP data, the Blairmore or Nordegg multiples have a longer travel path to
the Wabamun (compared to the near-offset VSP data). For this reason, the Wabamun is less
affected on the far-offset VSP data by the Blairmore coal and Nordegg multiple than on the
near-offset VSP data. Far-offset data deconvolution is not essential for the far-offset data.
VSP-CDP mapping
The Z"up(+TT) data from panel 6 of Figure 4.12 are used for the interpretation of
the off-reef markers. Two interpretation panels are used for this purpose. The first focus-
es on the transformation of the Z"up(+IT) data in time (+IT) and depth into the VSP-
CDP domain of time (+TT) and offset-distance from the well. The interpretive processing
of the VSP-CDP mapping is displayed in the VSP-CDP IPP of Figure 4.14. For compari-
son, the VSP-CDP IPPs resulting from the normal time-variant runstream suggested in
Chapter 2 and from the far-offset deconvolution are presented in Figures 4.15 and 4.16,
respectively. The second interpretation panel is the integrated seismic display shown in
Figure 4.17 of the merged VSP-CDP display and the surface-seismic data.
The Z"up(FRT) data in panel 6 of Figure 4.12, placed in +IT time, are shown as
the first panel of the VSP-CDP IPP in Figure 4.14. Amedian-filtered Z"up(+TT) and the
VSP-CDP mapped data (pseudo-two-way traveltime versus offset) are shown as panels
2 and 3, respectively. The Blairmore Coals, Nordegg, Wabamun, Ireton, inter-Ireton,
Cooking Lake, and Acoustic Basement markers are interpreted on these presentations.
The median filtering (panel 2), as well as the VSP-CDP mapping, processing (panel 3)
have not appreciably distorted the interpretability of the original data (Z"up(+IT) data in
panel 1 of Figure 4.14).
The Ireton and inter-Ireton events are interpreted to be continuous and effective-
ly parallel, indicating that only off-reef events (as opposed to reefal events) have been
imaged on the far-offset VSP data. In contrast, the displays in Figures 4.15 and 4.16 are
difficult to interpret (compared to the data in Figure 4.14) because of the interfering noise
and deconvolution processing artifacts. Interpretive processing improved the original
mode-converted, upgoing and downgoing 5V-contaminated Zup(FRT) and
HMAXup(FRT) data sufficiently enough (as shown in the final results of Figure 4.14) to
enable a confident interpretation during each processing step.
136
1.6
-l
! 1.8 ~
m
~
2.0m
3:
2.2
2.4
2.6
BLAIRMORE COALS0 IRETON • COOKINGLAKE .

NORDEGG INTER-IRETON 0 ACOUSTIC
Fig. 4.14. Interpretive processing panel showing the VSP-CDP transformation of the
nondeconvolved far-offset Ricinus VSP data.
1.4
1.6
-l
!I 1.8~
.. m
r-
I -l
·2.0 ~
2.2
2.4
2.6
BLAIRMORE COALS0 IRETON • COOKING LAKE .

NORDEGG INTER-IRETON 0 ACOUSTIC •
Fig. 4.15. IPP showing the VSP-CDP transformed results for the far-offset Ricinus VSP
data resulting from the suggested "normal" runstream in Chapter 2.
137
In the integrated seismic display (lSD, Figure 4.17), the surface-seismic line is
merged with the VSP-CDP (panel 3; Figure 4.14). The integrated seismic display shows
that in the initial surface-seismic interpretation (Figure 4.3) the edge of the reef was
extended too far to the northeast, but supports the current interpretation (Figure
4.5).These data also illustrate that the off-reef inter-Ireton event (Figure 4.5) extends about
500 m to the southwest of the VSP well and allow for the confident correlation of the
inter-Ireton marker.
This final display shows that the surface-seismic processing effectively attenuat-
ed the 5V-events (through normal moveout correction and CDP stacking). Interpretive
processing similarly enabled the effective attenuation of the 5V-events on the VSP data.
Since few raypaths between the VSP source and the borehole receiver locations were
affected by any possible faulting, and normal CDP summing is not done in VSP process-
ing, the lateral coverage of the far-offset VSP can be interpreted with a high level of con-
fidence, and allows for the independent interpretation of the seismic data.
Summary
The VSP offsets for the Ricinus VSP survey were selected with the following con-
siderations in mind:
1) the inter-Ireton marker would terminate against the reef flank;
2) if the reef edge was further than 500 m (approximately half the 1100 m VSP offset) from
the borehole, or in a different direction than the selected offset, the inter-Ireton mark-
er would be relatively flat and continuous;
3) surface-seismic multiples would be identified on the zero-offset VSP data; and
4) exact geological ties to the surface-seismic profile would be provided by the zero-
offset VSP data.
The 199-m near-offset VSP data enabled the identification and correlation of the
surface-seismic events at the VSP well site. The VSP interpretation was focused on the
Ireton, inter-Ireton, and Cooking Lake events rather than the Leduc event. One interpre-
tational concern was that the Wabamun event was affected by Blairmore Coals and
Nordegg multiples.
On the nondeconvolved near-offset VSP data, multiple interference at the
Wabamun level occurred except where the borehole geophone was below the Nordegg.
The upgoing VSP events tied the surface-seismic section well only after VSP deconvolu-
tion was applied. The CDP stacking enabled sufficient multiple attenuation to facilitate
138
1.6
-f
~iIill-"llll
1.8 ~
,....
-f
~
2.0~
en
(3) Z"up(decon) (+TT)

VSP-CDP
BLAIRMORE COALS D IRETON • COOKING LAKE .
NORDEGG • INTER-IRETON El ACOUSTIC •
WABAMUN • LEDUC D BASEMENT
Fig. 4.16. Interpretive processing panel showing the VSP-COP transformed results for the
deconvolved far-offset Ricinus VSP data.
500m
SOUTHWEST SCALE NORTHEAST
CDP # OFFSET (m) CDP #
H!~;~5E;5~5::~;;;;~=HH~:B i m m i~m ;i= . ; zm .::~~:u:::=m:=:
1.6
1.7
1.8
~
0
1.9
~
-f
2.0
~
m
,....
-f
~
2.1 m
~
en
2.2
2.3
2.4
Fig. 4.17. Integrated seismic display showing the VSP-COP transformed far-offset
Z' up(+ TT) data merged with the surface seismic data. These VSP data replace the sur-
face seismic data at COP's 258-293 (from Hinds et aI., 1993c).
139
interpretation of the Wabamun on the surface-seismic data. The far-offset VSP upgoing
events did not require deconvolution (using the separated HMAX'down(-TT) data to
design the operator) and were readily correlated to the surface-seismic data.
The far-offset VSP provided lateral subsurface coverage up to a distance of 500 m

from the VSP well site. Each trace represents approximately 6 m of subsurface coverage.
The updated geologic model based on the far-offset VSP results is shown in Figure 4.4. In
this figure, wells 6-9 and 7-15 are reefal. The VSP well is off-reef and at least 500 m away
from the reef edge.
The ISD in Figure 4.17 shows that the inter-Ireton event, within the range of the VSP-
CDP coverage, is continuous and relatively flat. Within the 500 m imaging range, the event
does not terminate against the flank of the reef. The Wabamun event on the far-offset VSP
data has not been affected appreciably by the multiples from the Blairmore Coals or
Nordegg. The Leduc reef event dips relatively steeply to the north of well 7-15 (Figures 4.5
and 4.17), suggesting that an abrupt edge may be present farther than 500 m to the south-
west of the VSP well. The seismic/geologic tie has been confidently interpreted from the
IPPs, and the interpretation has been carried through the merged far-offset VSP and seismic
dataset.
The mapping of the off-reef sediments enabled the final whipstock decision to be
made. The decision was to abandon the well because the reef was more than 500 m away
from the well location.
140
5 - Fort St. John Graben Area Case Study
Overview
The deltaic sandstones of the basal Kiskatinaw Formation (Stoddart Group, upper
Mississippian) were preferentially deposited within structural lows in a regime character-
ized by faulting and structural subsidence. These sandstone facies can form reservoirs
where they are laterally sealed against the flanks of upthrown fault blocks. Exploration for
basal Kiskatinaw reservoirs generally is accompanied by the acquisition and interpreta-
tion of surface-seismic data prior to drilling. These data are used to map the grabens in
which these sandstones were deposited and the location of horst blocks which act as lat-
eral seals. In the case study of the Fort S1. John Graben area, northwest Alberta, Canada,
three vertical seismic profile (VSP) surveys were conducted at the 9-24-82-11 W6M
exploratory well site subsequent to drilling. These data supplemented the surface-seismic
and well-log control such that:
1) direct correlation was made with the surface-seismic data, ensuring that the surface-
seismic control was accurately tied to the subsurface geology;
2) multiples were identified on the VSP data, and their effect on the interpretation of the
surface-seismic data was determined; and
3) the subsurface geology, in the vicinity of the borehole, was more dearly imaged on the
VSP data than on the surface-seismic control and reveals amplitude anomalies and
faults which are not evident on the surface-seismic data.
Introduction
On the basis of the interpretation of conventional surface-seismic data, an
exploratory well (9-24-82-11 W6M) was drilled into the basal Kiskatinaw Formation
(Stoddart Group, upper Mississippian; Figure 5.1) on the downthrown side of a fault block
in the Fort S1. John Graben area, Peace River Embayment (Figures 5.2 and 5.3). It was
expected that gas-prone sandstones of the basal Kiskatinaw would be laterally truncated
and sealed against shales of the Golata Formation. Contrary to expectations, the well
encountered unproductive tidal-flat shaly sandstone facies in the basal Kiskatinaw and a
commercial gas-bearing zone in the upper Kiskatinaw, and is now shut-in. To obtain a high-
er resolution image of the subsurface in the vicinity of well 9-24, and to map any fault fea-
tures that were not resolved on the surface-seismic data, three VSP (vertical seismic profile)
surveys were recorded at the 9-24 well site. These profiles were used in conjunction with
141
SUBSURFACE, PEACE RIVER EMBAYMENT

[2 SERIES GLOBAL STAGES AGE NW. ALBERTA AND N.E. BRITISH COLUMBIA
UJ
MY NW. ~.E.
KUNGURIAN
Z l1. ::;
-c ARTINISKIAN Cl UPPER CARBONATE MEMBER u,
:E ...J
--_? -------------
>
0:: lOWER SAKMARIAN UJ 0
MIDDLE SANDSTONE MEMBER ...J
UJ
l1.
III
J: --- - -------------
lOWER CARBONATE MEMBER
...J
UJ
ASSELIAN 286 C/)
III
GZHELIAN
KASIMOVIAN
UPPER MOSCOVIAN
KSITUAN MEMBER
BASHKIRIAN
320
n,
C) I t
~":OR c",[.\
~
C/) SERPUKHOVIAN
0:::
U
::::) «0
a ::Jrn
0 0::
N w
a
0
a ___1_ ?tJl ___
u, 0::: ~
0 w rn KISKATINAW FM
W Z LL. '"
w
.:7
...J a z
c,
o, GOlATA
:::>
<: rn a ----
0:: rn
a.. <C 0:::
VISEAN I--
-c w a.. DEBOLT FM
U o
...J
0
0
:i!
C) "-
---
I-- W
0::: ...J ?
W w
l:'" 0
~ 0
...J Z ELKTON FM
a...J ::J
0::: ~ SHUNDAFM
'"
w
n,
"-
FORMATION F
:::>
'"L;;"""
...J
__ ~I~KOFM
TOURNAI SIAN 0 ILL ? MEMBER 0
0 MEMBERB
:i! LL ?
~ Z s~----
1"'""::;;
w
l: -c MEMBER A
~g; ~
., u,
0
...J 360 rn ,.
." ~ ~
;j X
UJ ~
I
en
Z BLACK SHALE MBR x
sZ ?
UJ
0 UPPER FAMENNIAN
[ij ?
C WABAMUN FM
Fig. 5.1. Stratigraphy of the Fort St. John Graben study area
(modified from Richards, 1989).
surface-seismic coverage to image the fault system in the area and to determine the seismic
signature of the upper Kiskatinaw reservoir.
This chapter is a case history of the Fort St. John Graben 9-24-82-11 W6M well.
An overview of the stratigraphy (Figure 5.1) and the geologic history of the Lower
Carboniferous in the study area is included. The acquisition and interpretive processing
of the VSP data is described, and an integrated interpretation of the well-log, surface-
seismic, and VSP data is presented. On the Fort St. John Graben dataset, faults that are
not well resolved on the surface-seismic data are better delineated on the VSP data. The
142
Fort St. John Graben Area Case Study
Fig. 5.2. Map of Western Canadian Sedimentary Basin tectonic elements showing the
Peace River Embayment, Prophet Trough, Sukunka Uplift, cratonic platform, and Fort S1.
John Graben (Barclay et aI., 1990).
,
(
( I m
\ ;;:
.
...J ....
00::
IV
I...J
III
m
-c ! I~ ?1" -'" ,/~ "'' 0 88
)IR
~ ,,~ / r-r' / ' "-a 7
,Iv~~~ ['--->\ .~'l-~6
Fig. 5.3. Detailed area map
(/ r-.
of the Fort S1. John Graben
Vl/i-''' (~~I '" ~r7 I' \(.\~~ ~,,~ 85 showing the Bear Canyon,
Josephine, Bonanza and
~~~~ ~~!!~UL ( , , , 'J \j~~~Y)83 George faults. The study
\ _
~\l-c=~K~:Y (,", 1 _ ~nL..

""\ ~ ~(>::,_,./ ... ~2 area is centered in Township
82 and Range 11 west of the
6th Meridian in Alberta,
Canada, near the Alberta
and British Columbia provin-
cial border (from Richards et
aI., 1994).
75~
/'
) " " HytM \
143
interpretive processing of these data illustrates the utility of the VSP technique in the
search for hydrocarbons in structurally complex areas.
Geological overview
Tectonic and depositional history of Peace River Embayment
This section will review the tectonic and depositional history of the study area.
Unlike the case studies of Chapters 3, 4, and 6 which deal with carbonate reef explo-
ration, this case study involves hydrocarbon entrapment in channel sandstones.
CAMBRIAN TO EARLY DEVONIAN

(d) UPPER CARBONIFEROUS
U. Carboriferous
Fig.5A. Diagrammatic summa-

(b) MID TO LATE DEVONIAN ry of the depositional and tec-
(e) MESOZOIC
tonic history of the Peace River
Embayment area (Cant, 1988)
from the Cambrian to the
(e) LOWER CARBONIFEROUS Cretaceous periods.
Investigations of the Peace River Arch and Peace River Embayment have revealed
a depositional regime characterized by faulting (O'Connell, 1990) and structural subsi-
dence. The Peace River Arch is related to a regional, uplifted, crystalline basement struc-
ture (Figure 5.4a; Cant, 1988) that was formed or deformed in three main phases:
1) Theformation of a structural high during the LateProterozoic that was progressively onlapped
by Middle to Late Devonian siliciclastics, evaporites, and carbonates. The basement was
uplifted 800 to 1000 m during the early Paleozoic (mid-Cambrian) and formed an
asymmetrical structure with the northern flank dipping steeply and the southern
flank dipping more gently (Cant, 1988; O'Connell et al., 1990). During the Middle to
Upper Devonian, a diachronous siliciclastic unit (lithozone as defined in Trotter, 1989)
called the Granite Wash (sediments derived from plutonic and metamorphic base-
ment relics) carbonate and shale deposits of the Elk Point (along with Gilwood sand-
stone), Beaverhill Lake (Slave Point, carbonate reefs and Waterways Formation),
Woodbend (fringing reefs, basal ramp to stacks of carbonate shelves and overlying
carbonate ramp), and Winterburn (Nisku and Blueridge Formation; Moore, 1988) sed-
iments onlap the emergent arch. The carbonates of the Wabamun Formation carbon-
ates eventually buried most of the Arch (Richards, 1989, 1990; Richards et al., 1994).
144
Fort S1. John Graben Area Case Study
1
T GOLATAF~t-=:;::;=t-==~=""IrI
NEEDGEOF
FilL BYEHD
1 km (SHALE) SUKVNKAUPLIFT(S,U,)
GOLATADEPOSITION
1 DEBOLT FM. (LIMESTONE)

(UPPERVISEAN)
T_70 . 26W.8 NaiiiiiJ

lL-'L---------;e--------;--;c"'cc:km:=..---"_-':":"1
FORT ST. JOHN GRABEN (F S J)AND
HUDSONHOPELOW (H H L)
Fig. 5.5A. Block diagram showing the Fort

St. John Graben by the end of the Golata FilL BY END
KISKATlNAW DEPOSITION
Formation time (from Barclay et al., 1990) (UPPER VISEAN)
N .,;;;;;;;;7
Fig. 5.5B. Block diagram showing the Fort St.

John Graben by the end of the Kiskatinaw
Formation time (from Barclay et al., 1990)
Fill BV END
TAYlOR FLAT DEPOSITION
(SERPUKHOVIAN OR YOUNGER)
<.
Fig. 5.5C. Block diagram showing the Fort
St. John Graben by the end of the Taylor
Flat Formation time (from Barclay et al.,
1990) FilL BYEND
BELLOY DEPOSITION
(LEOHARDIANIPERMIAN)
N .,;;;;;;;;7
Fig. 5.5D. Block diagram showing the Fort St.

John Graben by the end of the Belloy
Formation time (from Barclay et al., 1990)
2) The formation of a series of embayments resulting in anomalously thick Carboniferous, Permian

and Triassic deposits. Normal faulting occurred during the deposition of the Middle
Devonian Elk Point Group. These normal faults were rejuvenated in the late Famennian
during the deposition of the upper Wabamun Formation and the overlying Exshaw
Formation (Figure 5.4b).
3) The development of a deep basin component of the foreland basin during the Jurassic and
Cretaceous. The arch ceased to be emergent by the beginning of the Carboniferous.
The Peace River Embayment, a structural inversion consisting of northeast to
southwest trending interlinked graben and half-grabens (Barclay et al., 1990), started to
develop during the latest Devonian (late Famennian) and earliest Carboniferous while
145
the Exshaw and overlying lower Banff Formation were deposited (Richards, 1989;
Richards et al., 1994). The major tectonic features surrounding the Embayment were the
Prophet Trough (Richards, 1989; Richards et al., 1994), Sukunka Uplift (Richards, 1989), and
cratonic platform (Figure 5.2). The northwest-trending narrow pericratonic Prophet Trough
(O'Connell et al., 1990) resulted from the downwarping and downfaulting of the western
margin of the North American plate during the latest Devonian (late Famennian) and
Carboniferous and is interpreted as a back-arc basin (Richards, 1989). In the detailed strati-
graphic correlation chart (Figure 5.1), the Famennian-to-Tournaisian boundary coincides
with the Devonian-to-Carboniferous boundary. The Carboniferous Peace River Embayment
opened northwestward into the Prophet Trough. During the middle-to-late Tournaisian, the
Embayment became better defined as a separate entity. The northern Sukunka uplift on the
southwestern side of the Embayment formed a low rim that restricted the access of the
Embayment sediments into the southeastern part of the Prophet Trough.
Regional subsidence continued, associated with extensive block faulting (Figure
5.4d) along normal faults (Cant, 1988). During the period of block faulting, anomalously
thick Lower Carboniferous successions of siliciclastics and ramp-to-platform carbonates
were deposited along the axis of the embayment (Richards, 1989; 1990).
Lower Carboniferous geology

Within the Carboniferous succession in the Fort S1. John Graben study area
(Figure 5.3), four principal episodes of block faulting have been recognized (Richards
1989; 1990):
1) Episode 1 - after deposition of the latest Devonian to earliest Carboniferous deposits
of the Exshaw Formation, but prior to the deposition of the overlying Banff Formation;
2) Episode 2 - during deposition of the lower Banff Formation (early middle Tournaisian);
3) Episode 3 - during deposition of the Golata and Kiskatinaw Formations (late Visean);
and
4) Episode 4 - after deposition of the upper Visean to Serpukhovian (?) Taylor Flat
Formation but prior to that of the Permian Ishbel Group.
Marked regional subsidence accompanied the second and third episodes of block
faulting, whereas subaerial erosion and local uplift accompanied the first and last phases.
Pronounced regional subsidence, at least locally, accompanied by block faulting also occurred
during deposition of the middle and upper Tournaisian Pekisko and Shunda Formations. In
the middle Tournaisian, the developed embayment can best be termed a deep re-entrant into
the western cratonic platform, and the Prophet Trough is interpreted as a back arc basin
(Richards, 1989). Deep-water environments existed in the embayment area during periods of
146
pronounced subsidence. In the general study area, the Fort St. John and Hines Creek Grabens
were developing prior to the deposition of the Stoddart Group (Barclay et al., 1990).
Lower Carboniferous deposition spans the upper Exshaw through Taylor Flat
Formation. Carbonate and fine-grained terrigenous clastics were deposited during the
Exshaw-Debolt interval. Basal shale, carbonate ramp deposits, shallow-water shelf car-
bonates, intertidal to supratidal carbonate, clastics, and evaporites were deposited in the
Banff Formation. Shales are also present in the Pekisko Formation, a coeval correlative for
the Pekisko and Shunda (defined in Richards, 1989), that overlies the Banff Formation. A
well-developed basin, slope, and shelf environment in the Peace River Embayment was
formed during the deposition of the lower Banff to Pekisko/Shunda succession. Apart
from these times, sedimentation in the area kept pace with the subsidence, resulting in
normal topographic relief.
The Golata, Kiskatinaw, and Taylor Flat Formations were deposited in the late
Visean and Serpukhovian, with the Golata overlying the Debolt Formation conformably
in the central part of the embayment. The Golata consists of fissile grey mudstone, silt-
stone, and shale (Barclay, 1988).Towards the basin margins, the Golata and Debolt con-
tact becomes disconformable (O'Connell et al., 1990).
The Kiskatinaw consists of sandstone with minor quantities of fine-grained silici-
clastics, limestones, dolostones, and coals. At least eight depositional cycles make up the
formation, but three main depositional cycles are referred to informally as the lower, mid-
dle, and upper Kiskatinaw. The Kiskatinaw Formation represents a marine-dominated
deltaic system which may have resembled an estuary in the study area during deposition
(Richards, 1989; Richards et al., 1994). The study area was a semienclosed body of coastal
water. Block faulting occurred, accompanied by deep subaerial erosion, which resulted in
substantial thickness changes in the Kiskatinaw and Taylor Flat Formation deposits.
The Kiskatinaw and Golata Formation contact is unconformable. This prodelta
sequence (Golata Formation) can serve as a reservoir seal in those places where, as a
result of faulting, it is juxtaposed against the sandstones of the basal Kiskatinaw. The
uppermost and basal Kiskatinaw Formation sandstones deposits are exploration targets.
Sequential deposition in the study area is shown in the models for the Stoddart
Group deposition (Figure 5.5A-D). The development of hydrocarbon traps occurred prior to
the deposition of the Permian Belcourt and Belloy Formation. The study area is in the Fort
St. John Graben area defined by east-west faulting to the north (Bear Canyon and Josephine
faults) and to the south (Bonanza fault) as shown in Figure 5.3. The Fort St. John Graben is
part of the Dawson Creek Graben Complex (Barclay et al., 1990; Figures 5.6 and 5.7).
Basal Kiskatinaw channel-fill (Barclay, 1988) was the target of the 9-24 well. The
147
hypothesis was that thick sections of basal Kiskatinaw would be preferentially deposited
in the structural lows on the downthrown side of early Kiskatinaw faults. The basal
Fig. 5.6. Location of the Fort

St. John Graben as outlined
by the Taylor Flat Formation
isopach map. The Belloy
marked the end of the subsi-
dence in the graben (from
Barclay at aI., 1990)
TAY LOR FLAT FM

ISOPACH
SOkm
Fig.5.? Peace River

Embayment stratotectonic
elements including the Fort
St. John Graben (from
Barclay et aI., 1990).
Kiskatinaw encountered by the 9-24 well was not of reservoir quality; however, the sand-
stones of the upper Kiskatinaw contained commercial gas.
Original interpretation and well results

The split-spread, 96 trace data used to create the stacked section (Figure 5.8) were
acquired using a patterned dynamite source (3 X 3 kg at 15 m) and DFS4 recording equip-
ment (12/18-124 Hz filter; notch (60 Hz) filter out). The groups consisted of nine inline
148
14-Hz L25D geophones spaced at 6.1 m. The geophone group, shot, common mid-point
(CMP), and near-offset intervals were 67.1, 134.1,33.5, and 201.2 m, respectively.
The seismic section (normal polarity display; Figure 5.8) was migrated using a
prestack partial migration scheme which consisted of applying a common-offset domain,

dip-moveout (DMO) correction (Hale, 1984) to the prestack data. Updated velocities
were determined using the DMO-corrected CMP gathers. Closely spaced velocity analy-
ses were performed since the stacking velocities changed rapidly along the line. This
prestack processing was followed by poststack phase-shift migration. The poststack
migration method is reviewed in Gazdag (1978) and Gazdag and Squazzero (1984).
The surface-seismic data were used to clarify the structural setting at the Kiskatinaw
level. The interpreted events on the seismic line are the Nordegg, Halfway/Doig, Belloy,
basal Kiskatinaw, and Debolt (stratigraphically listed in Figure 5.1). The 9-24-82-11 W6M
well was located approximately 200 m east of the seismic line and has been projected onto
the seismic line as shown in Figure 5.8. Note that well 9-24, as projected, is on a downthrown
fault block relative to wells 2-25-82-11 W6M and 7-36-82-11 W6M (Figure 5.8).
Well 2-25 encountered approximately 25 m of relatively clean, wet basal
Kiskatinaw sandstone and was abandoned. Well 7-36 penetrated 35 m of basal Kiskatinaw
sandstone that tested gas plus salt water and is currently classified as a shut-in gas well.
Well 9-24 was drilled in the expectation that basal Kiskatinaw gas would be stratigraphi-
cally entrapped against the flank of the upthrown flank block; however, the basal
Kiskatinaw proved to be shaly-sandstone that was not of reservoir quality. Well 9-24 was
shut-in as an upper Kiskatinaw gas well.

Three VSP surveys were recorded at the 9-24 well site (one near-offset and two
far-offset sources). The vSPs were planned with the following objectives in mind:
1) to provide a confident surface-seismic tie to:
a) the upper Kiskatinaw Formation reservoir;
b) the lower Kiskatinaw Formation; and
c) the Debolt Formation.
2) to determine if multiple reflections were a significant problem at the Kiskatinaw level
and to design an effective deconvolution filter.
3) to map the lateral extent of the shaly sandstone of the basal Kiskatinaw and upper
Kiskatinaw reservoir in the direction of the far-offset VSPs.
4) to provide a higher resolution seismic image of the Debolt fault in the vicinity of the
9-24 well.
149
SCAlE: 1 kin
0.8 ~
~
~
__ ~1-1.0 ~
!:j
ff
m
In
SPIRIT RIVER • BELLOY • BASAL KISKATINAW .

NORDEGG • TAYLOR FLAT 0 GOLATA •
HALFWAYIDOIG . KISKATINAW • DEBOLT •
Fig. 5.8 Example surface seismic data displaying the original interpre-
tation of the owners of the 9-24 well (preceding the drilling of the well).
1I
_- 0'2 ~ -0.4 ~
~
0.6 !:j
~'!!!!~ 3:
0_8~
III
1.0
1.2
_ _ 1.0
1.2
(5)Zup(+TT) (6) Zup(+TT) (7)Zup(+TT)
median filtered
Fig. 5.9. Interpretive processing panel depicting the wavefield separation of

the near-offset Fort St. John Graben VSP data.
150
The near-offset VSP source was located 149 m from well 9-24 and in the direction
of well 2-25. One of the far-offset source locations (referred to as FSJGl in Chapter 2) was
700 m from the 9-24 well site in the direction of 2-25. The other far-offset source location
(referred to as FSJG2) was situated 741 m east of well 9-24. Two Vibroseis units operated
in series at each offset using a 12-second sweep of 10-90 Hz. The recording length was 15
seconds, resulting in a 3-s cross-correlated output. On average, six sweeps were summed
at each geophone sonde location. The total depth of well 9-24 was 2126 m below the Kelly
Bushing of the drilling rig (KB at 644 m asl). All three offset sources were at 639 m asL Data
were recorded at a sampling rate of 1 ms using an MDS-lO system. The recording filter,
OUT/250, was designed to prevent aliasing.
The triaxial sonde vertical spacing was 20 m (from a depth of 2030 m up to 350 m).
As a precautionary measure, calibration records were acquired at several depths as the
sonde was lowered down the borehole. These depth stations were reoccupied during the
production runs to detect possible depth errors or cable stretch.

During the processing of the near-offset VSP data, a series of interpretive pro-
cessing panels (IPPs) were designed. These panels were designed to display the follow-
ing interpretive processing steps:
1) upgoing and downgoing P-wave separation;
2) deconvolution of the separated upgoing P-waves; and
3) inside and outside corridor stacking of both the nondeconvolved and deconvolved
upgoing waves.
P-wave separation
The separation of the upgoing and downgoing P-waves from the Z(FRT) data is
displayed in the wavefield separation interpretive processing panels of Figure 5.9. Panel
1 displays the Z(FRT) data after trace normalization. The upgoing P-wave events are dif-
ficult to discern until the Z(FRT) data are gained (panel 2). On these panels, the tube
wave is visible below 0.9 s as a high-frequency, downgoing wavetrain with a velocity of
propagation of approximately 1435 m/s. The tube wave also reflects from the bottom of
the well borehole and travels back up.
Several upgoing primary events and multiple reflections can be identified on
panel 2. Consider, for example, the Spirit River and Nordegg events. Both of these events
are followed by a trailing surface-generated multiple with a lag time of about 110 ms.
151
0.2
-I
- 0.6 m ~
r
-I
0.8 ~
«s m
, 1.0 -(II
(2) Zup(+IT)
1.2
median filtered
SPIRIT RIVER •
-¥il2.Hr:-:-h-HHit-~;jj-!ffifHT-rl+h+fHt-r.;- 0.2 NORDEGG •
;d . HAlFWAY/DOIG •
~ BELLOY •
m TAYLOR FLAT D
!:j KISKATINAW •
~ BASAL KISKATINAW .
~:f+-!il-- 0 . 8 m GOLATA •
:§: DEBOLT •
1.0
(5) Zup(decon) (-IT) (6)Zup(deCOn)(+TT) (7)ZuP(deCOn)(+TT) 1.2

median filtered
Fig. 5.10. Interpretive processing panel depicting the deconvolution of the near-offset Fort
S1. John Graben VSP data.
!iimiiiUiimmim mmmmmmm
:1
~m~~~mmm.~!i! ~~ ~~~mtiirri.im DIrI H
(m)
,I' J I •
0.6
-I
0.8 ~
<
m
r-
-I
1.0 ~
1.2
-
~
Fig. 5.11. Interpretive processing panel illustrating the inside and outside corr idor stacks
of the Zup(+TT) VSP data .
152
This multiple pattern (both upgoing and downgoing waves) is highlighted in panel 2 of
Figure 5.9.
In panel 3 (Figure 5.9), the Z(-TT) data are displayed. An Ll-point median filter
was used to remove the upgoing P-waves; the Zdown(-TT) data are displayed in panel 4.
In the next step, the Zdown(-TT) data of panel 4 were subtracted from the Z(-TT) data to
yield the Zup(-TT) data (panelS). Note that a residual tube wave is visible within the
Zup(-TT) data panel. The Zup(+TT) data before and after the application of a 3-point
median filter are shown in panels 6 and 7, respectively. The two panels have been time-
shifted to facilitate plotting. In panel 7, the Spirit River multiple (highlighted in panel 2)
is observed as high amplitude events which lie directly below the Spirit River primary
and can be interpreted for several cycles (from 0.3 to 0.7 s). This multiple is not observed
at sonde depths deeper than the top of the Spirit River (at 730 m).
VSP deconvolution
Deconvolution IPPs were designed for the Zup(+TT) data (Figure 5.10) to enable
the monitoring of the deconvolution process for the near-offset data. The incorporated
panels reveal information about multiples that was difficult to determine from the wave-
field separation IPP (Figure 5.9) alone. The first two panels (Figure 5.10) are the nonfiltered
and median-filtered Zup(+TT) data, respectively. The interpreted multiple events have
been highlighted in purple on panels 1 and 2. Panel 3 is the gained Zdown(-TT) data.
Panels 4 and 5 are the nondeconvolved and deconvolved Zup(-TT) data, respec-
tively, which enable an evaluation of the effect of deconvolution. The last two panels (6
and 7) are the nonmedian and median-filtered Zup(decon)(+TT) data, respectively. An
inspection of panels 2 and 7 (Figure 5.10) shows that deconvolution has effectively atten-
uated multiple reflections. The Spirit River multiple wavetrain from 0.5 to 0.9 s on panel
2 (colored purple), for example, has negligible amplitude on panel 7. Note also that
deconvolution has increased the high frequency content of the data, allowing for better
resolution at the Kiskatinaw level.

Nondeconvolved, inside and outside corridor stacks, and associated displays
were designed for the near-offset data (Figure 5.11). A comparison of the inside and out-
side corridor stacks (panels 3 and 4, respectively) illustrates the utility of these displays.
For example, the Spirit River multiple (between 0.65 and 1.0 s) is present on the inside
corridor stack and absent on the outside corridor stack. Note also, that basal Kiskatinaw
and Golata events can be resolved on the outside corridor stack. On the inside corridor
153
-l
~ - 0.8 ~
m
r
-l
'- 1.0 §:
m
1.2
~~~~f~ .-.
11 """""---.__
~!!I:
-1.4
(1) Zup(decon)(+TT) (2)ZUP(deconjC+TT) (3)(4) (5) Zup(decon)(+TT) (6) Zup(decon)(+TT)
median filtered muted inside corridor muted outside corridor median filtered
SPIRIT RIVER •
(3) Z (+TT) (4) Z (+TT) KISKATINAW •
NORDEGG • . . uptdecon) . uptdecon) BASAL KISKATINAW .
Inside comdor stack outstde corrtdor stack GOLATA •
HALFWAYIDOIG •
BELLOY • DEBOLT •
TAYLOR FLAT 0
Fig. 5.12. Interpretive processing panel illustrating the inside and outside
corridor stacks of the Zup(decon)( +TT) data.
0.4
-f
0.6 ~
m
r-
0.8 -f
s:
-1.0 ~
-
rn
(1) X(FRT) (2) Y(FRT) (3) Z (FRT) (4) HMIN(FRT) 1.2
s~i ~ ~ i il ~ i ~ ~ ~ ~ i ~ ~ ~ ~ i ~ ~ D E PTH (m)

.! I;' 04
I' •
I :
Iii
0.6
- 1.0 -rn
-
Fig. 5.13. Interpretive processing panel depicting the hodogram-based rota-
tion of the Fort 81. John Graben FSJG1 far-offset V8P data.
154
stack these reflections are masked by a high-amplitude multiple.

If deconvolution is successful, the deconvolved inside and outside corridor stacks
(panels 3 and 4; Figure 5.12) should be similar. At the zone of interest just above 1.35 s, the
Debolt and the Golata are similar (panels 3 and 4), however, the basal Kiskatinaw is not ade-
quately represented on the inside corridor stack of the deconvolved data. An exam-
ination of the input data to the inside corridor stack (panel 2 of Figure 5.12) reveals that the
multiple contamination has not been completely eliminated on the shallow traces (from 1290
m to the surface). This results in the broad peak in the zone immediately above the basal
Kiskatinaw (on panel 3 at 1.26 s), and the observation that the high-frequency basal
Kiskatinaw trough has been overpowered by residual multiple contamination. Note that the
Spirit River multiple, which is highlighted in purple between 0.65 and 1.0 s, has been sig-
nificantly attenuated on all of the traces except for the traces recorded between 490 m and
the surface. The deconvolution process has not achieved all of the imaging objectives; how-
ever, the data are adequate for correlating the geology and the surface-seismic.
Far-offset (FS/Gl) VSP interpretive processing

On the far-offset FSJGl VSP data, the vertical (Z) and both horizontal (X and Y) axis
data contain nonpartitioned elements of the upgoing and downgoing P- and SV-wavefields.
Examination of the IPPs (Figures 5.13-5.15) reveals that the partitioning of the wavefields
has significant implications with respect to interpretation. The far-offset IPPs for the FSJGl
data were designed with the aim of displaying the following major processing steps:
1) hodogram-based rotation of the X(FRT), Y(FRT), and Z(FRT) data (based on win-
dowed data enveloping the P-wave first arrival);
2) time-variant model-based rotations applied to the HMAXup(derot)(FRT) and
Zup(derot)(FRT) data; and
3)VSP-CDP mapping and Kirchhoff migration processing of the Z"up(+TT) data.
Hodogram-based rotation (offset FSJGl)

The X(FRT), Y(FRT), and Z(FRT) data for the FSJGl far-offset VSP are displayed
in Figure 5.13 on panels I, 2, and 3, respectively. The horizontal axis data (X and Y) are
extremely noisy and contain only minor amounts of P-wave events. The Z(FRT) data con-
tain strong downgoing P-wave events plus lower amplitude upgoing P-wave events.The
hodogram-based rotation technique is designed to polarize these data so that the down-
going P-waves are presented on a single channel, HMAX'(FRT).
The first step illustrated in Figure 5.13, is the hodogram-based rotation of the
X(FRT) and Y(FRT) data (correcting for phase changes as a result of tool rotation during
155
(3) Zup(derot)(FRT)
-I
0.6 ~
<
m
r-
- 0.8 :::!
.
~~ .
-- 1.0
~
~
of the Fort S1. John Graben FSJG1 far-offset VSP data.
0.6
-l
0.8 ~
m
"' \ r-
-l
- 1.0 3:
m
~~f- 1 .2 ~
;~ " ""!' ,'I,i,

(2) Z'~p(+TT) (3) Z'~p(+TT)
median filtered VSP-CDP
SPIRIT RIVER • BELLOY • BASAL KISKATINAW •
Fig. 5.15. Interpretive processing panel depicting the VSP-CDP transformed and
Kirchhoff migrated results of the Fort S1. John Graben FSJG1 far-offset VSP data.
156
the movement of the sonde up the borehole). The output HMIN(FRT) and HMAX(FRT)
data are displayed as panels 4 and 5, respectively. HMIN(FRT) and HMAX(FRT) data are
assumed to be aligned perpendicular to and tangent to the plane formed by the well and
the source, respectively. Note that the HMIN(FRT) data (comprised of horizontally
polarized shear (SH) wave events and out-of-the-plane reflections) contain the dominant
portion of the diffraction that appears at LOs on the 650 to 800 m traces, suggesting that
the diffraction is side-swipe energy originating from a feature such as a fault out of the
plane defined by the wellbore and seismic source.
The Z'(FRT) and HMAX'(FRT) (panels 6 and 7) data were obtained by rotating the
Z(FRT) and HMAX(FRT) data using polarization angles estimated from a hodogram
analysis of a window of data centered around the P-wave first arrivals of the Z(FRT) and
HMAX(FRT) data (DiSiena et al., 1984). Downgoing mode-converted SV-events can be
interpreted on all three axis data, X(FRT), Y(FRTI, and Z(FRT). These SV-events were
described in Chapter 2. The SV data can be used for the quality control of the second polar-
ization; namely, HMAX(FRT) and Z(FRT) to HMAX'(FRT) and Z'(FRT), respectively.
The analysis of the second polarization is focused on the detection of SV-events
(mode-converted or from any other origin) on the HMAX'(FRT) data. In panel 6 and 7 of
Figure 5.13, the mode-converted SV-events have been completely polarized onto the
Z'(FRT) data and, at first inspection, do not appear on the HMAX'(FRT) data.
Upgoing VSP events can be up to 100 times lower amplitude than the downgoing
events (Hardage, 1985). The time-variant polarization IPP (Figure 5.14) that has been
designed and presented in the next section contains the wavefield separated Z'up(FRTI and
HMAX'up(FRT) data in panels 1 and 2, respectively. It is only after the wavefield separation
processing on the Z'(FRT) and HMAX'(FRTI data that the underlying upgoing event com-
ponents of the Z'(FRT) and HMAX'(FRTI data can be confidently evaluated.
The two sets of rotations have polarized the X(FRT), Y(FRT), and Z(FRT) data so
that the downgoing P-waves are effectively isolated on a single channel, HMAX'(FRT),
as shown in panel 7 of Figure 5.13.
Time-variant model-based rotation

The IPP for the time-variant model-based polarization analysis for the FSJGl off-
set data is shown in Figure 5.14. The Z'up(FRT) and HMAX'up(FRT) are shown in panels
1 and 2 (Figure 5.14), respectively. On both panels, upgoing P-waves from the Debolt,
Golata, and Kiskatinaw can be identified, indicating that the hodogram-based rotations
were unsuccessful in isolating the upgoing P-wave events on the Z'(FRT) panel.
To remove the undesirable effects of the Z(FRT) to Z'(FRT), and the HMAX(FRT)
157
.m .m~mmmm
• • • •
m.iiiiimmmm im.mimmmm mHm m ~ im ~m DEPTH [rn)
, . . . . " , . , •• f " " "
0.4
-l
0.6 ~
m
r
0.8 ~
m
1.0 :§:
-l
~
<
m
r
-l
~
m
1.0
:§:
1.2
(6) Z' (FRT) (7) HMAX' (FRT)
Fig. 5.16. Interpretive processing panel depicting the hodogram-based rotation of the Fort
S1. John Graben FSJG2 far-offset VSP data .
i i" i.
I,
ii. i! ~ ~ i ~ ~ i ! i ~ i i ~ i DEPTH (m)
I
0.4
-l
0.6 ~
m
r
-l
~
m
(3) Zuplderot)(FRT)
of the Fort S1. John Graben FSJG2 far-offset VSP data (see Figure 5.18 for color legend) .
158
to HMAX'(FRT) transformations, the Z'up(FRT) and HMAX'up(FRT) were derotated

(using the inverse operation of the second polarization rotation). The Zup(derot)(FRT) and
HMAXup(derot)(FRT) data are shown as panels 3 and 4 of Figure 5.14, respectively. By
inspection, the upgoing P-wave events have been effectively distributed back onto a Z-
type axis data, Zup(derot)(FRT). Unlike the upgoing wave events in the Z(FRT) data (panel
3 of Figure 5.13), where the downgoing P-waves were dominant, the separated upgoing
P-wave events in the Zup(derot)(FRT) data are dominant and interpretable.
On the Zup(derot)(FRT) data (panel 3), upgoing P-waves generated by the shallow
reflectors are improperly aligned (because of the choice of a single rotation angle per
trace). See for example the upgoing event resulting from the reflection from the Spirit
River interface. These data have been derotated, but the upgoing P-wave events are still
partitioned on both output data {Zup(derot)(FRT) and HMAXup(derot)(FRT)} because of the
nonzero offset of the source. The deeper events do not suffer much misalignment because
the deep event raypath geometries satisfy the near-vertical incidence angle assumption
better than do the raypaths of shallower events. The time-variant model-based rotation
corrects for this misalignment. The output of the time-variant polarization, the
HMAX"up(FRT) and Z"up(FRT) data, are shown in panels 5 and 6, respectively. Note that
the shallow events display more alignment than on the Zup(derot)(FRT) of panel 3. The rota-
tion angle required for the Spirit River and Nordegg events on a particular trace are dif-
ferent to the rotation angle needed for a deeper event (such as the Debolt) on the same
trace. The time-variant rotation technique generated these different rotation angles.
The Spirit River event is barely discernable on the Z"up(FRT) data because the
reflected raypaths from this shallow event are near the critical angle. The surface-gen-
erated multiple from the Spirit River interface (observed on the nondeconvolved near-
offset data of panel 7 in Figure 5.9) are significantly lower amplitude on these far-offset
data (panel 6; Figure 5.13).
VSP-CDPmapping and migration

Nonfiltered and median-filtered Z"up(FRT) data are displayed (pseudo-two-way
traveltime versus depth) in Figure 5.15 as panels 1 and 2, respectively. The VSP-CDP
mapped (pseudo-two-way traveltime versus offset) and Kirchhoff migrated Z"up(+TT)
data are shown as panels 3 and 4, respectively.
The FSJGl offset source was located 700 m away from the well and in the direction
of well 2-25. The interpretation of the surface-seismic data (Figure 5.8) suggests that a major
Debolt fault (displacing Debolt to the Belloy Formation material) was located between well
9-24 and well 2-25. On panels 3 and 4 (Figure 5.15), the Debolt event on the VSP data is inter-
159
1.4
(3}Z'up(+IT)
VSP-CDP
SPIRIT RIVER • BElLOY • BASAL KISKATINAW .
HALFWAY/DOIG . KISKATINAW • DEBOLT •
Fig. 5.18. Interpretive processing panel depicting the VSP-CDP transformed and
Kirchhoff migrated results of the Fort St. John Graben FSJG2 far-offset VSP data.
SCALE:1km
0.6
1tJ_t- a ~
0. :E
~
-I
~
m
1.0 !:j
i:
m
-
III
1.4
SPIRIT RIVER • BELLOY • BASAL KISKATINAW •
NORDEGG TAYLOR FLAT D GOLATA •
Fig. 5.20. Current, preferred surface seismic interpretation. There are now two interpret-
ed faults between wells 9-24 and 2-25. The VSP (FSJG1 offset) data seismically image
the fault nearest well 9-24 (shown in this figure) as two separate faults.
160
preted to be faulted at two locations between these wells; however, it is shown in "far-offset
VSP interpretive processing" (see previous section) that these VSP data image two faults that
are not shown on the interpretation on Figure 5.8. A second interesting feature on panel 3 is
that the signature of the basal Kiskatinaw event changes abruptly in proximity to the fault
nearest well 9-24. On the migrated version, this event is smeared. The basal Kiskatinaw
event, as interpreted, is continuous at greater offsets, but is substantially decreased in ampli-
tude. This character change could indicate a change in lithology (increase in shale content)
or porosity (increase in water content as in well 2-25).
The upper Kiskatinaw event (upper surface of the location of the hydrocarbon
reservoir at well 9-24) is interpreted on panels 3 and 4 to be laterally continuous, but
faulted. Vertical displacement is interpreted across two faults. The upper Kiskatinaw is
present in both wells 9-24 and 2-25. At well 9-24, the upper Kiskatinaw forms a gas reser-
voir; in the structurally higher 2-25 well this unit is nonproductive.
Far-offset (FS/G2) VSP interpretive processing

The FSJG2 far-offset source was 741 m east of well 9-24. These data were acquired
to map the lateral extent of the upper Kiskatinaw reservoir to the east of well 9-24 and to
provide a higher resolution seismic image of faults that displace Debolt strata. The offset
survey was designed in an effort to explore a possible area of future exploration. This is
a case where the VSP data capture precedes a possible surface-seismic survey in an area.
The recording and processing of these VSP data were similar to that described for the
FSJGl survey.
In the first stage of interpretive processing, a hodogram-based rotation technique
polarized the X(FRT), Y(FRT), and Z(FRT) data, so that the downgoing P-waves were
presented on a single channel, HMAX'(FRT), as shown in Figure 5.16. The filtered output
data, Z'up(FRT) and HMAX'up(FRT), were derotated and time-variant, model-based rota-
tions were applied to the output data as shown in Figure 5.17. In the final stage, the
Z"up(FRT) data were displayed in pseudo-two-way traveltime versus depth, and in pseu-
do-two-way traveltime versus offset.
The Z"up(+TT) data after the application of the VSP-CDP mapping are displayed
in Figure 5.18. Nonmedian and median-filtered versions of the Z"up(+TT) data are dis-
played (pseudo-two-way traveltime versus depth) as panels 1 and 2, respectively. The
VSP-CDP mapped (pseudo-two-way traveltime versus offset) and Kirchhoff migrated
Z"up(+TT) data are shown in panels 3 and 4, respectively. The data in panels 3 and 4
image two faults, both of which are of the same magnitude as those on the FSJGl data
(Figure 5.15). The basal Kiskatinaw event exhibits a laterally continuous seismic signa-
161
'" GAM~~~~~~~~lr~~1
§I
"'8
~I l!
8_ ~ ~
c..>
~
SHOTPOINT # (/)u
.... <
.~
....
\,I
ali; "
v-~~ ','
Ii ;; ( i1)[ '"
l':
100
l:l..
~
I::
l':
e-
0 ~0 .4
~
i=
~ ...J
wO .6
...bh.
I::
til ~ 0.8 N
100
til I- .-I
~
~
\,I
0
~ 1.0
100
l:l..
~
.::.... ~
I-
~
100
Q..
1.2
100
....
~
- I::
l:l..
CI}
;>
SURFACE SEISMIC
SPIRIT RIVER •
SURFACE SEISMIC
BELLOY
VERTICAL SEISMIC PROFilE
[] BASAL KISKATINAW .
NORDEGG TAYLOR FLAT 0 GOlATA •
HALFWAY/DOIG • KISKATINAW • DEBOLT •
Fig. 5.19. Integrated interpretive display of the Fort St. John Graben exploration data.
ture. This is unlike the character of the basal Kiskatinaw in a direction away from the VSP
well and towards the FSJGl offset source, which decreases in amplitude beyond the first
imaged fault. The upper Kiskatinaw event (hydrocarbon reservoir at well 9-24), as inter-
preted on panel 3, is laterally continuous but faulted.
The integrated interpretive display (lID) is shown in Figure 5.19. On the left-hand
side of the lID in Figure 5.19, gamma ray logs for well 9-24, well 2-25, and well 7-36 are time-
tied to the current interpretation of the surface-seismic data. On the right-hand side, near-
offset VSP data are time-tied to the gamma ray and sonic logs acquired in well 9-24. These
correlated data allow a confident interpretation of the surface-seismic line and the identifi-
cation of the Spirit River, Nordegg, Halfway/Doig, Belloy, Taylor Flat, upper Kiskatinaw,
basal Kiskatinaw, Golata, and Debolt events. The nondeconvolved version of the VSP data
is presented to facilitate an analysis of multiple contamination. As evidenced by a compari-
son of the surface-seismic line and the corridor stacks, the multiple-contaminated inside cor-
ridor stack provides a poor tie to the data at the zone of interest (Kiskatinaw). This suggests
that multiples on the surface-seismic data have been attenuated effectively.
The interpretation of the surface-seismic section that incorporates the VSP information
(normal polarity display; Figure 5.20) differs slightly from the prewell interpretation (Figure
5.8). Of particular significance is that on the updated version, the Taylor Flat and Kiskatinaw
events are confidently correlated. Note that in the post-VSP interpretation, the Taylor Flat
event is absent at well 7-36. This interpretation is supported by the well-log control.
The geologic cross-section of Figure 5.21 was constructed on the basis of the
post-VSP interpretation of the surface-seismic line and the analysis of the far-offset VSP
data. It is consistent with the well-log, surface-seismic, and VSP control. Note that three
faults have been placed between wells 9-24 and 2-25 on the geologic cross-section
(Figure 5.21), whereas only two faults are interpreted on the corresponding seismic-sec-
tion (Figure 5.20). The two faults interpreted on the VSP-CDP data (Figure 5.15) do not
correlate with the interpreted faults shown on Figure 5.8 (between wells 9-24 and 2-25).
It should be recalled that the surface-seismic line and the VSP profiles (subsurface cov-
erage) have different orientations.
A plan view of the FSJGl source offset position along with the location of the two
VSP interpreted faults and surface-seismic located fault (near well 2-25) is shown in
Figure 5.22. This presentation of the VSP and surface-seismic line geometry helps us
understand the subsurface coverage of the VSP and surface-seismic data. In Figure 5.22,
well 9-24 is located due east of CDP number 49 of the seismic line shown in Figure 5.19.
163
9~ 9-24 SOUTH 2-25 NORTH 7-36
1 { ( } ,
<C
::!:
~! j~
10001 ::!:
<C
C>
C>
'U~
.....
~
I.J
I.J
~
11 J i
"L BLY
~ELLOY
BLYI
c::.. L BELLOY
""
I:l-.
"'tj
j1211~
KSITUAN MBR
=
f
~
TAYLOR . I
~ TF FLAT KISKATINAW
0
~
..s=
E-<
be
...=
[/l
[/l
-
<
W
12
KISKATINAW
BASAL I BASAL
KISKATINAW
[
( """
1.0
.-I
0
~
I.J en BASAL
KISK
KISK
[C 13 KISK ~~ .. ~. I GaLATA
""
I:l-. ::J GaLATA
.....
~
;;- en BK DEBOLT
~ 13 DEBOLT
..""""
p. GTA
~
-=
I:l-.
CJ)
~
1400
,.::;... --':=.-1 DB
Fig 5.21. Geologic cross-section incorporating information form the surface seismic, VSP and geologic well log
results. Note that three faults have been postulated to be between wells 9-24 and 2-25.
The far-offset source location is marked by the circle to the northwest of the well 9-2410ca-
tion. The subsurface coverage of the VSP-CDP data is in a direction toward the offset
source location, starting at well 9-24 and extending out to half the distance from well 9-24
to the source. The two VSP-CDP imaged faults near well 9-24 project onto the seismic line
at CDP numbers 47 and 48. These faults are shown as a single fault on the reinterpreted
seismic line in Figure 5.22. The layout in Figure 5.21 shows that the fault displayed on the
seismic line immediately south of well 2-25 is not imaged on the VSP-CDP data; only on
the surface-seismic data.
Fig. 5.22. Plan map of the FSJG1 offset

2-25 source area showing the fault locations
--e i
as interpreted from the VSP and surface-
seismic data. The VSP data (FSJG1 far-
offset) imaged two faults along the
surface-seismic line nearest to well 9-24,
and the surface-seismic data imaged the
FSJG10 I fault nearest to well 2-25. These two
I datasets complement each other in the
I construction of the geological interpreta-
P- tion shown in Figure 5.21.
9-24
---interpreted fault
------seismic line
SCALE: 1km
165
The two interpreted faults on the VSP-CDP data (Figure 5.15) correlate with the
single interpreted fault nearest to well 9-24 on the reinterpreted seismic data (Figure
5.20). The surface-seismic data could not resolve the seismic expression of the faults into
two separate images as seen on the VSP-CDP data (Figure 5.15). In this case, the VSP-
CDP interpretation has brought new information into the integrated interpretation.
Summary
The exploratory well 9-24 was drilled on the downthrown side of a fault block on
the basis of conventional surface-seismic data. The expectation was that the gas-prone
sandstones of the basal Kiskatinaw Formation would be truncated laterally and sealed
against the upthrown fault-block. Contrary to expectations, the basal Kiskatinaw was
unproductive; however, well 9-24 did encounter commercial gas within the upper
Kiskatinaw.
To obtain a high-resolution seismic image of the subsurface in the vicinity of
well 9-24, and to map any faults that were not resolved on the surface-seismic data,
three VSP surveys were recorded at the 9-24 well. This seismic profiling information,
in conjunction with surface-seismic coverage, was used to image the subsurface fault
pattern, and to elucidate the seismic signature of the upper Kiskatinaw Formation. The
profile data supplement the surface-seismic and well-log control in that VSP data can
be directly correlated to the surface-seismic data. As a result, the surface-seismic con-
trol was accurately tied to the subsurface geology, multiples could be identified on the
VSP data and evaluated on the surface-seismic data, and the subsurface, in the vicin-
ity of the borehole, was better resolved on the VSP data than on the surface-seismic
control.
The information provided by the VSP surveys allowed a refinement of the inter-
pretation of the surface-seismic data and enabled the construction of a detailed geologic
cross-section (Figure 5.21). These interpretations provide information with respect to the
subsurface in proximity to well 9-24, and perhaps more importantly, further elucidate the
geologic history of the structurally complex Fort St. John Graben area (Richards et aI.,
1994).
166
6 - Simonette Field Case Study
Overview
On the basis of conventional surface-seismic data, the 13-15-63-25 W5M
exploratory well was drilled into a low-relief Leduc Formation reef (Devonian
Woodbend Group) in the Simonette area, west-central Alberta, Canada. The prognosis
was that the well would intersect the crest of the reef and encounter 50 to 60 m of pay.
Unfortunately, it was drilled into a flank position of the reef and was abandoned. The
decision to abandon the well, as opposed to whipstocking in the direction of the reef
crest, was made after the acquisition and interpretive processing of both near- and far-
offset (252 and 524 m, respectively) VSP data, and after the re-analysis of existing surface-
seismic data.
The near- and far-offset VSPs were recorded and interpreted while the drill rig
remained on-site, with the immediate objectives of determining an accurate tie between
the surface-seismic data and the subsurface geology, and mapping relief along the top of
the reef over a distance of 150 m from the 13-15 well in the direction of the adjacent pro-
ductive 16-16 well (with a view to whipstocking). These surveys proved to be cost-effec-
tive in that the operator was able to determine that the crest of the reef was out of the tar-
get area, and that whipstocking was not a viable alternative. The use of VSP surveys
allowed the operators to avoid the costs associated with whipstocking and to feel confi-
dent with their decision to abandon the welL
Introduction
In this chapter, the 2-D surface-seismic and VSP signatures of the low-relief reef
in the Simonette study area are discussed.The 2-D data were acquired prior to drilling the
13-15 exploratory well, which intersected the reef in a flank position and below the
oil/water contact.
The Simonette Reef lies within the western Woodbend depositional realm
(Stoakes and Wendte, 1987). The Woodbend Group in central Alberta (Figures 3.1A, B,
and C) is subdivided into four formations: Cooking Lake, Duvernay, Leduc, and Ireton.
In the Simonette study area (of the western Woodbend depositional realm), the Cooking
Lake Formation is depositionally absent, and the Leduc Formation conformably overlies
the Waterways shale (Beaverhill Lake Group; Moore, 1989a). The Leduc reef in the
Simonette study area developed as both full reef and low-relief reef. The areal extent of
167
Fig. 6.1. Ireton isopach map of

the main and low-relief Simonette
reef within the eastern Woodbend
depositional realm. The locations
of the wells used in the construc-
tion of the geologic cross-section
in Figure 6.2 are shown in this fig-
ure. The exploratory well 13-15 is
also posted.
- - - GEOLOGIC SECTION
,...-'0- CONTOUR INTERVAL, 30
• WELL
m
-
1 km
SOUTHWEST NORTHEAST
2100 10-34 0~-06 10-06 04-16 10-16 16-16 04-22 08-34
« ::; «
::;
::; ::;
220
::;
C5
«
(!)
~
__- r - -
~
o o
Z Z
aen aen
230
E 240
~
:( 2500"'1--1..---
W
C/)
a:l
~ 2600
I~~~
LEDUC REEF
280
2900
I__ ------?
ELK POINT
Fig. 6.2. Geologic cross-section A-A' traversing the wells shown in Figure 6.1. The tra-
verse starts in the southwest of the main Simonette reef at the back-reef lagoonal facies
well, 10-34; continues to the northeast on the main reef for wells 4-6, 10-6, 4-16; into
inter-reef shales for well 10-16; onto the low-relief reef with wells 16-16 and 4-22; and into
the shale basin in well 8-34.
168
Simonette Field Case Study
the full reef, which towers up to 230 m above the Beaverhill Lake platform, is defined
roughly by the 70 m Ireton isopach contour interval in Figure 6.1. Where thick reef is pre-
sent, the encasing shales thin, resulting in lower shale isopach values. The low-relief reef
in the study area attains a maximum thickness on the order of 120 m. Its approximate
areal extent is defined by the 130-m contour interval (Sections 15, 16, 21, and 22 of
Township 63, Range 25 W5M; Figure 6.1).
In Figure 6.1, the locations of the 8 wells incorporatedinto the geologic cross-
section of Figure 6.2 are shown. The borehole logs consist of neutron, gamma ray and
sonic logs. The cross-section shows the morphology of the Wabamun, Winterburn,
Ireton, Leduc, Beaverhill Lake, and Elk Point Formations and displays the placement of
the study area (comprising of wells 16-16,04-22 and the VSP-weIl13-15) with respect
to the main Simonette atoll reef. The reef of interest in this chapter is the low-relief car-
bonate buildup that developed basinward of the main Simonette complex. The updip
edges of this type of reef can be productive where structurally closed and sealed by the
inter-reef shales of the Duvernay and Ireton Formations.
Well 10-34 (Figure 6.2) penetrated the calcareous muds of the interior lagoon
facies of the Leduc reef and was abandoned. Well 4-6 was drilled as a development well
and is now a suspended oil well. Well 10-6 was drilled into the northeastern portion of
the Simonette atoll reef and is a flowing oil well. Well 4-16 was drilled into the raised
peripheral rim and is a flowing oil well. Well 10-16 and 8-34 are off-reef (penetrating only
Ireton and Duvernay shales) and were abandoned. Wells 16-16 and 04-22 were drilled in
the period 1986-89 and are described in the next section.
The term "full reef" was used in Anderson et al. (1989a and 1989b) to describe
atolls and adjacent pinnacles that attain maximum heights of over 200 m (Figures 6.1
and 6.2). Such reefs are usually easy to map on good quality 2-D seismic data. They are
characterized by appreciable velocity pullup (on the order of 30 ms), time structural
drape of about 30 ms at the top of the Devonian (Wabamun Formation), and reflection
waveform variations within the Woodbend interval (the Leduc, Ireton, and Duvernay
Formation seismic events). The seismic signature of the low-relief reef (the target of the
VSP-well; Figure 6.3) is more subtle, being manifested by less than 15 ms of pullup and
less than 15 ms drape of the Wabamun event from reef to off-reef. In addition, the reflec-
tion from the top of the low-relief reef can be difficult to distinguish from the Z-marker
(an inter-reef shale event).
169
WEST EAST
16-16 13-15 250m
1.2
D WABAMUN
IRETON
•
•
Z-MARKER
LEDUC ••
Fig. 6.3. Example surface-seismic data over the low-relief reef displaying the original
interpretation of the owner of the data. The interpretation was done prior to the drilling of
well 13-15.
Fig. 6.4. Ireton to Leduc isochron

map resulting from the original
interpretation of the seismic lines
within the area of the low-relief
reef. The oil pay-zone of the low-
relief reef is interpreted to be as
productive at the proposed well
13-15 as at the existing well 16-
• WELL
- SEISMIC LINE
16.
0.5 km
_ 5 - CONTOUR INTERVAL, 5 m
170
Simonette low-relief reef

Three wells that penetrate the low-relief carbonate buildup in Figure 6.1:
16-16-63-25 W5M, 4-22-63-25 W5M, and 13-15-63-25 W5M. Well 16-16 encountered
72 m of net pay; well 4-22 encountered 24 m of net pay but watered out after 12
months of production; and well 13-15 is the abandoned exploratory well for which
the two VSP surveys were acquired (Figure 6.4). This latter well was drilled with the
expectation that it would be similar to well 16-16 .
• WELL
0.5 km - - - - SEISMIC LINE
_ 5 - pONTOUR INTERVAL, 5 m
Fig. 6.5. The preferred Ireton to Leduc isochron resulting from the updated interpretation
(using the VSP results) of the seismic lines in the area of the low-relief reef, the VSP
results, and geologic borehole data. This interpretation shows the reef gradually rising out
to 120 m from 13-15 and then steeply rising towards well 16-16.
The original (pre-VSP well) contour map of the Ireton-to-Leduc isochron is shown
in Figure 6.4. The prognosis was that well 13-15 would encounter 50-60 m of net pay. The
contour map of Figure 6.5 summarizes the final preferred interpretation of the 13-15 well
location. It is based on drilling results, the interpretation of the far-offset, nondeconvolved
and deconvolved VSP data, and the reinterpretation of the suite of existing surface-seismic
171
lines within the area. The example surface-seismic line (normal polarity display; Figure 6.3)
is nominally 20-fold, split-spread, 120 trace data, acquired using a patterned dynamite
source (2 X 0.5 kg at 9 m depth; shotholes spaced 25 m apart) and DFS-5 recording equip-
ment (8-128 Hz filter). The seismometer groups consisted of nine, inline, 14 Hz geophones
spaced at 3.75 m. The geophone group and shot intervals were 25 and 75 m, respectively.
The seismic markers of principal interest correspond to the Mannville, Nordegg,
Debolt/Belloy, Wabamun, Ireton, Z-marker (a regional marker within the inter-reef shales),
and Leduc (Figures 6.2 and 6.3). The Z-marker corresponds to the top of the Duvernay
shales (Figure 6.2). The seismic image of the subsurface at well 13-15 (Figure 6.3) was inter-
preted initially to be comparable to that at the productive well 16-16 location; hence it was
drilled. The high-amplitude event at about 1.9 s, at the 13-15 location, was incorrectly inter-
preted to represent the reef top. In retrospect, it is interpreted to be the off-reef Z-marker.
Well 13-15 intersected the low-relief reef in a flank position below the oil/water contact and
encountered 134 m of inter-reef shale (in comparison with 75 m of shale encountered in
well 16-16). On the basis of the well results, it was suggested that the low-relief reef could
rise abruptly to the west, and that whipstocking in the direction of the productive 16-16
well should be considered.
The operators of well 16-16 had two alternatives: abandon the well or whipstock
in the direction of well 16-16, bearing in mind that the further well 13-15 deviated from
the original bottomhole location, the greater the production penalty imposed by the
Alberta Government. To ascertain the cost-effectiveness of whipstocking, the operators
ran two VSP surveys: a near-offset (252 m) and a far-offset (524 m) VSP survey. It was on
the basis of these data and the reinterpretation of the existing surface-seismic, that the
decision was made to abandon well 13-15.

After an analysis of the 13-15 well geophysical logs and prior to abandonment,
two VSP surveys were recorded in order to:
1) more accurately tie the surface-seismic data to the subsurface geology (in particular to
the top of the low-relief Leduc reef);
2) map the top of the reef over a distance of 150 m in the direction of the 16-16 well (with
a view to whipstocking); and
3) differentiate primary reflections from both surface-generated and interbed multiples.
The near- and far-offset locations were chosen to be 252 and 524 m, respective-
ly, from welll3-15. Both were online with respect to the surface-seismic profile (Figure
6.4). Two vibroseis units were operated in series at each offset. The 12 s sweep ranged
172
from 10 to 70 hz, the recording length was 15 s, and the cross-correlated output was 3 s.
On average, six sweeps were summed for each geophone sonde location. MDS-lO
recording instruments and a sampling interval of 1 ms were used. The recording filter
was OUT/250; the instrument filter served as an anti-aliasing, low-pass filter with a ramp
rolloff starting at 250 hz.
Well 13-15 extends 3620 m below the Kelly Bushing (at 878 m asl). Both source loca-
tions were 868 m asl. The geophone sonde was lowered to the bottom of the well and raised
at 20-30 m intervals. At each sonde location, the three-component geophone was locked in
place. The two offsets were designed to be in the direction of the producing 16-16 well, with
the following interpretation considerations in mind:
1) if the slope of the reef was steep, a diagnostic seismic signature would be seen in the
252-m offset data; however, a diagnostic shadow zone (null data) would be seen on
the 524-m data;
2) if the crest was not steep, but gradual, then both offset datasets would image the reef
slope (the 252-m offset data would effectively be a subset of the 524-m offset VSP
data); and
3) if the slope was gradual but became steeper beyond the range of the near-offset (252
m) VSP, then the 252-m and 524-m offset data would image only the gradual slope,
whereas the longer offset data would image both slopes.

During the processing of the near-offset VSP, a series of interpretive processing
panels (IPPs) were designed to display the following:
1) separation of Zup(FRT) and Zdown(FRT) data from the Z(FRT) data;
2) deconvolution of the Zup(FRT) data to output the Zup(decon)(FRT) using an inverse fil-
ter calculated from the Zdown(FRT) data; and
3) inside and outside corridor stacks (Hinds et al., 1989) of both the Zup(+TT) and
Zup(decon)(+TT) data.
P-wave separation
The separation of the upgoing and downgoing P-waves on the vertical (Z) geo-
phone data is illustrated in Figure 6.6. Panel 1 (Figure 6.6) displays the Z(FRT) data after
gain adjustment. In panel 2, high-amplitude surface-generated multiples and less promi-
nent interbed multiples are seen in the Z(-TT) data. The surface-generated downgoing
multiples are recognized as continuous events arriving after the first-break primary
173
(D - - - -f\)NP\,) I'\,J f\) WW W ..... - - - - - r \ ) l\) l'\,) l\) www

--W~~OIl,).~~ON.(DO~.""'~I\).~""'O-W
o~mo.m(D(D~(D(D~(Dow ..... m l\)~OO(D(DO(D~
00 VI UI 00 UI eft
(1) Z(FRT) (2) Z(-TT) (3) Zdown(-TT) (4) Zup(- TT) (5) Zup(+TT)
••
MANNVILLE D median filtered
WABAMUN • Z-MARKER
NORDEGG II IRETON • LEDUC
DEBOLT/BELLOY •
Fig. 6.6. Interpretive processing panel depicting the wavefield separation of the near- offset
(252 m) Simonette VSP data.
(D - - - - N I\)N N N WW W .....- - - - -I\)NNNW W W

-- W ~ ~ O N . ~ m OI\) . (D OI'\,) . "", (D~ . ~ "", O - W
o ~mO.~(D~~~~~(Do w .....
m I'\,)UlOO~(DO~~
-l
O.5~
<
m
~
i::
m
~ :IJ-;;
(3) Zdown(-TT) (4) Zup(deconkTT) (5) ZUP(decon)(+TT)

median filtered
MANNVILLE D WABAMUN • Z-MARKER •
NORDEGG • IRETON • LEDUC •
DEBOLT/BELLOY •
Fig. 6.7. Interpretive processing panel depicting the deconvolution of the near-offset (252 m)
Simonette VSP data.
174
downgoing wavetrain on all of the traces, from the deepest to the shallowest receiver
location. A downgoing multiple associated with the Mannville Formation interface is
interpreted to start at the 2080 m trace (between 0.3 and 0.32 s; panel 2) and to continue
onto the deeper traces. This event is probably an interbed multiple, since on panel 2, it
does not appear to extend onto the shallower depth traces. The Zdown(-TT) data in panel
3 do not assist in the interpretation since the median filter has smeared the event onto the
shallower traces. An inspection of the Zup<+TT) data in panelS confirms that multiple
interference is definitely originating from the top of the Mannville Formation.
The Zdown(-TT) data were obtained from the Z(-TT) data using an Tl-point medi-
an filter and are displayed in panel 3. The residual Zup(+TT) data content in panel 3 are
interpreted to be minimal since the data consist predominately of time-aligned downgo-
ing events. The Zup(-TT) shown in panel 4 were separated from the Z(-TT) data by sub-
tracting the Zdown(-TT). The Mannville, Nordegg, Belloy/Debolt, Wabamun, Ireton, Z-
marker, and reefal Leduc events are interpreted in the median filtered Zup<+TT) data and
shown in panelS of Figure 6.6.
VSP deconvolution
The initial downgoing pulse (except in the case of head wave contamination) is the
primary downgoing P-wave; later downgoing arrivals are multiples. The deconvolution IPP
shown in Figure 6.7 enables us to monitor the deconvolution process.
The first panel of Figure 6.7 is the median-filtered Zup(+TT) data. Panels 2, 3, and
4 are the Z(-TT), Zdown(-TT), and Zup(decon)(-TT) data, respectively. The median-filtered
Zup(decon)(+TT) data are displayed in panelS.
A comparison of the Zup(+TT) with the Zup(decon)(+TT) data illustrates the effect
of multiple contamination on the continuity of primary events. In panel 1, for example,
the Debolt/Belloy event is high amplitude and continuous at sonde depths below the
Mannville (from 2080 m to 2570 m). At shallower recording depths, the Debolt/Belloy
event and a Mannville interbed multiple interfere. More specifically, the peak at 1.275 s
on the left-hand side of panel 1 is the interfering multiple and can be correlated across
the panel to the 2080-m depth trace. At that point, the Debolt/Belloy event (of lower fre-
quency than the multiple) dominates. The Debolt/Belloy is continuous to the 2570-m
depth trace, where it disappears (since this is where the "first-break" for that depth level
occurs). On panelS, deconvolution has substantially reduced multiple interference and
the Debolt event is much more continuous. Also the series of multiples immediately
below the Mannville are strongly attenuated.
Within the Wabamun to the Leduc Formation interval, multiple contamination is
175
... U) "'f\)I'\)~"'WWW
.... - -w cn Cl O~ · cn ClO O...., ·

OO ""ClO.Ot~YtUtUlUlUlu)
o o o o o V..,.oJIUUIJl CJI
~ 2 .0
(1) Zup(+TI) (3)(4) (5) Zup(+TI) (6) Zup(+TT)

median filtered muted inside corridor median filtered
MANNVILLE
NORDEGG
D (3) Zup(+TT) (4)Zup(+TI)
WABAMUN
IRETON ••
••
DEBOLT/BELLOY • Z-MARKER
LEDUC
Fig. 6.8. Interpretive processing panel illustrating the utility of the inside and outside cor-
ridor stacks of the near-offset (252 m) Simonette Zup(+TT) data.
ffitHlttb.fII1-1.0 -t
~
m
!::;
~
m
1.5~
I
II i
I I
I - 2.0
I
(2) Zup(decon)(+TI) (3)(4) (5) Zup(decon)(+TI)
muted outside corridor muted inside corridor
(3) Zup(decon)(+TI) (4) ZUp(decon)(+TI)
Fig. 6.9. Interpretive processing panel illustrating the utility of the inside and outside cor-
ridor stacks of the near-offset (252 m) Simonette Zup(decon)(+TT) data .
176
minimal as evidenced by the observation that events in this zone are continuous and rel-
atively unchanged by deconvolution. The Z-marker and Leduc events are continuous
and do not exhibit either significant time-structural relief or appreciable character varia-
tions. The description of the 524-m VSP interpretation will show that the far-offset data
contain a Wabamun multiple, in contrast to the near-offset results (presented above) on
which Wabamun multiple interference is minimaL

Multiple contamination on the Zup(+ TT) data can be re-examined using the inside
and outside corridor stacks of both the Zup(+ TT) and Zup(decon)(+ TT) data. Zup(+ TT)
inside and outside corridor stacks, along with the input data to the muting and stacking
processes are shown in Figure 6.8. The amplitude of the Debolt/Belloy primary is weak
on the Zup(+ TT) inside corridor stack (panel 4; Figure 6.8) in comparison to the same
event on the outside corridor stack (panel 3). The Debolt/Belloy event at sonde depths
shallower than 2080 m (above the Mannville), on the muted input data shown in panelS
for the Zup(+ TT) inside corridor stack, is masked (destructively interfered with) by a pos-
sible interbed multiple. At sonde depths greater than the Mannville (2080 m), the
Debolt/Belloy (primary) event is unaffected by the multiple contamination. On the traces
in this depth range, this primary event is much higher in amplitude and 5-10 ms deeper
in time than the multiples. This confirms that the bottom generating interface of this mul-
tiple is the top of the Mannville Formation.
On the inside and outside Zup(decon)(+ TT) corridor stacks (Figure 6.9), the upgo-
ing multiples appear to have been substantially attenuated by deconvolution. For exam-
ple, the Mannville associated multiples that lie immediately beneath the Mannville pri-
mary event on the Zup(decon)(+ TT) data are greatly attenuated with respect to the
Zup(+ TT) data. This can be seen by comparing panel 1 of Figure 6.8 to panel 1 of Figure
6.9 in the time window of 1.35 to 1.5 s. In Figure 6.8, the time window is contaminated
with multiple reflection events (colored turquoise blue in panels 1 and 4), whereas in
Figure 6.9, the same time window is relatively free of multiples. One multiple reflection
event of note (at 1.48 s) lies immediately above the Nordegg primary event. On panel 1
of Figure 6.8, the event is dominant; however, in panel 1 of Figure 6.9 (after deconvolu-
tion), the event is attenuated.
Some primary events recorded between the Mannville and Nordegg primary
reflection events are seen only on the few traces beyond the deepest trace recording the
Mannville primary event. These few peaks are included in the outside corridor stack. One
177
.........
E
"<t
a>
TRAVELTIME ( s)
-- :r
C\I
U')
Ol I TRAVELTIME (s)
~
o
~
~ l- .E
o e..
~
'"
' _ _...;;.L_ _...;;.L_ _..;;.a._ _~~:-:,-..I:~:--:-'-":"""":"""~':I
i'0;,j Ol
~
.E
Ol
3545
33 ?0
3 1?O'
? 9?0 -
:5
o
--
s: ?7?0 C
?5?0
? 3? 0 o
:;:::;
2 110
o 1880 .rg
C
o
1&30
1400
1180
e
:;:::; 940 "0
ttl 700 Ol
e 3395
3 195
3000 -
27 95
rn
ttl
.0
"0 25 95
I
Ol 2 400 Ol
rn 2200 "0
ttl
.0
1950 o
-E
J72 0
E 1480
1270
10 30 C
~ 7 90 ttl
0> 34 70 'i::
o 3? 70
30 70 ~,
"0
o
~
28 70
2&70 Ol
2470
Ol
2270 E
:;:::;
2050
ca: :5
-
18 ! 0
1570 Ol
0> 13bO ..c
u. 1120
Z .!: 880 0>
s .2
:r
3570
33 45
,5
a. 3 145 "0
Ol
"0
29 45
27 45 '0..
25 40 Ol
23 30 "0
Ol 21 40
C 19 00 Q)
ttl 1&60 C
a. 1420
12 10 ttl
0> 970 a.
7 30 0>
C
'w
rn
3420
322 0
C
Ol
301 5
28 15
'~ cci
o 2&20 Ol1U
ea. 2 4Z 0
222 0
199 0 ea.(,)"0a..
1750
Ol 1510 OlOO
> 1300
»
:;:::;cci 1060
Ol-
..... ttl
820
3495 ~2
e-"O 32 95 Cia>
2Coo a.. ~~= ~ ..... C
Ol 0
,>
2 695
~~
u,
~ c E
R~ ~
-U5
OOl
.....
<cia> - 20 80
18 40
1&00
i-
.....
~E
r~~~~I~~~~~~~~~~~~~~~~~:r
C 1150
9 10 ,'V
0,0 570
1380 O>C\J
u::: .S --
LL_l{)
178
such event, at 1.45 s, is preserved across the entire depth range of 670 to 2295 m after decon-
volution is applied. Since the depth interval between these primary events (between the
Mannville and Nordegg interfaces) represents only a few traces, corridor stacks alone are
difficult to interpret unless the unmuted and muted corridor stack data are on the same IPP.
Far-offset VSP interpretive processing

The Z(FRT), X(FRT), and Y(FRT) data contain nonpartitioned elements of the
upgoing and downgoing P- and SV-wavefields. Hodogram-based rotation and time vari-
ant polarization rotation IPP's used in the processing of the far-offset for the Simonette
524-m offset data are presented below to reveal that wavefield partitioning has signifi-
cant implications for interpretation.
Hodogram-based rotation
The Z(FRT), X(FRT), and Y(FRT) data are presented in Figure 6.10 as panels 1, 2,
and 3, respectively. The hodogram-based method initially polarized the components of
the X(FRT) and Y(FRT) data, outputting the HMIN(FRT) and HMAX(FRT) data shown
in panels 4 and 5, respectively. HMAX(FRT) data display consistent, primary, downgo-
ing P-wave first-breaks indicating that the first set of rotations were relatively successful.
The HMAX(FRT) data also contain mode-converted SV-events (colored purple) that
originate in the region of the Mannville reflector.
The Z(FRT) and HMAX(FRT) data were input to the second rotation that polar-
ized the downgoing P-waves. Since the source offset distance is 524 m and the depth of
the well is 3270 m, the HMAX' and Z' polarization axes were near vertical and horizon-
tal, respectively. The resultant Z'(FRU and HMAX'(FRT) data are shown in panels 6 and
7, respectively. The downgoing P-wave events are resident on the HMAX'(FRT) data;
however, both the HMAX'(FRT) and Z'(FRT) data contain upgoing P-wave events. This
indicates that time-variant polarization is necessary to separate the upgoing P-wave
events onto a single data panel (Z"up).
The Z'(FRT) data (panel 6) contain mode-converted downgoing shear wave pri-
maries, downgoing shear wave multiples, and upgoing P-waves. The first-break on the
2080-m trace (representing the depth of the Mannville) can be approximated to be at 0.68 s
on panel 6. From this point, the mode-converted downgoing SV-event (colored purple) can
be traced down to 1.12 s on the 3570-m trace. Two visible, downgoing SV-multiples (also
colored purple but arriving later in time than the primary downgoing SV primary event)
parallel the primary and arrive within a 150 ms window on the 3570-m trace.
179
-- 1.0
--
en
~~ 2.0
(2) Z'~p(+TT)
median filtered
MANNVILLE 0 WABAMUN •
NORDEGG IRETON •
DEBOLT1BELLOY •
Fig. 6.12. Interpretive processing panel showing the VSP-CDP transformed and
Kirchhoff-migrated far-offset (524 m) Z" up( +TT) data.
DEPTH (m) DEPTH (m) OFFSET (m) OFFSET (m)

i ~ ~ i ~ ~ i ~~
. i ~ ~ ~ ~ ~ ~- ~ ~- ~ ~ ~ ii ~i~~§~~!
- _.
~~~~~~~~~~~
- - -
~33s ; ~ ~5 s ' ~~~~ ~;3;j
,
~i~:1;J ~~~
Fig. 6.13. Interpretive processing panel showing the VSP-CDP transformed and
Kirchhoff-migrated far-offset (524 m) Z" up(decon)( +TT) data.
180
Time-variant model-based
In the first stage of the time-variant model-based rotation, Z'(FRT) and
HMAX'(FRT) data were wavefield separated (usingf-k filtering) into HMAX'down(FRT),

HMAX'up(FRT), Z'down(FRT), and Z'up(FRT) data. The HMAX'down(FRT) data are
retained and used in the deconvolution of the Z"up(FRT) data in the following section.
The HMAX'up(FRn and Z'up(FRT) data, shown as panels 1 and 2 in Figure 6.11,
respectively, were derotated to output the HMAXup(derot)(FRT) and Zup(derot)(FRT) data
shown in panels 3 and 4, respectively. Most of the upgoing P-wave events are distributed
back onto the Z-type axis, Zup(derot)(FRT). In the Z(FRT) data shown in panel 1 of Figure
6.10, the dominant downgoing P-waves are much higher in amplitude than the upgoing
events, and the upgoing events are more difficult to interpret. Following wavefield separa-
tion, the separated upgoing P-wave events in the Zup(derot)(FRT) data are easily correlated.
The upgoing P-wave events on the Zup(derot)(FRT) data (panel 4) are improperly
aligned, particularly those generated by shallow reflectors, because of the use of a single
rotation angle per data trace. These data have been derotated, but the upgoing P-wave
events are still partitioned on both output data sets, Zup(derot)(FRT) and
HMAXup(derot)(FRT), because of the nonzero source offset.
To correct for misalignment, time-variant rotation angles were calculated (using
paraxial ray-tracing; Beydoun, 1985a and 1985b), and applied to every pair of traces in
panels 3 and 4. The resultant HMAX"up(FRT) data (panel 5) contain the residual down-
going SV-wave energy that were left following wavefield separation. The Z"up(FRT) data
contain robust upgoing P-wave events. On the Z"up(FRT) data, shallow events (originat-
ing at 0.58 s between the traces for 1630 to 1750 m) are better isolated (onto a single panel)
than on either the Zup(derot)(FRT) or HMAX'up(FRT) data panels.
VSP-CDP mapping and migration

A VSP-CDP and migration IPP for the far-offset Z"up<+TT) and Z"up(decon)<+TT)
were designed to facilitate the interpretation of the interfering multiples and subsurface
structure. The Z"up(+TT) data shown in Figure 6.12 were used for the interpretation of
the low-relief Leduc reef. The appearance of many anomalous events arising from multi-
ple contamination on the Z"up(+TT) data (such as the event at 1.8 s) makes the interpre-
tation of this plot, along with the VSP-CDP (panel 3) and Kirchhoff migrated Z"up(+TT)
data, difficult.
The interpretation of the Leduc reef event is that the reef slope rises at 145 m away
from the well; however, the multiple events may be influencing the interpretation
through destructive and constructive interference at the Leduc level.
181
Because of the relatively small offset of the source (524 m) in comparison to the
overall depth of the borehole, far-offset VSP deconvolution was attempted on the
Z"up(+TT) data. Nonmedian-filtered Z"upldecon)(+TT), median-filtered Z"upldecon)(+IT),
VSP-CDP mapped Z"upldecon)(+IT), and Kirchhoff migrated Z"upldecon)(+TT) data are

shown in panels 1-4 of Figure 6.13, respectively.
Two different sets of upgoing P-wave multiples are prevalent in Figure 6.12. One
set of multiples is associated with the Mannville event. The nondeconvolved VSP-CDP
and migrated data (Figure 6.12) show Mannville (possibly interbed) multiples interfering
with deeper P-wave primaries between 1.4 and 1.7 s. (specifically, 1.43, 1.56, and 1.68 s).
Significant multiple contamination is interpreted between the shallowest trace and the
2080-m trace (top Mannville). Three multiple reflections are evident; the first Mannville
multiple can be seen at 1.45 s, the second at 1.56 s (in between the Nordegg and Debolt
primaries), and the third at 1.68 s.
The second set of multiples proved to be the most troublesome with respect to the
reef interpretation. There is a strong event below the Wabamun in the Z"up(+TT) data at
1.8 s. It exists on the shallower traces down to the Wabamun, but is absent on the deeper
traces. The interpreted reef flank shown in Figure 6.12 rises gradually (145 m away from
the borehole) from 1.95 s to 1.92 s. On the depth and time plots, the rise does not corre-
spond with the truncation of the multiple seen above at 1.8 s (panels 1 and 2). This gives
confidence in the interpretation that the interpreted rise in the reef event is a result of
properly imaging the sloping reef edge.
As evidenced by Figure 6.13, deconvolution has severely attenuated the
Mannville and Wabamun multiples. The Mannville associated multiple events at 1.45,
1.56, and 1.68 s are attenuated, and the Wabamun associated multiple event at 1.8 s is
reduced from a dominant event to a low-amplitude event. The Mannville, Nordegg,
Debolt/Belloy, Wabamun, Ireton, Z-marker, and Leduc events can now be more confi-
dently correlated. The Leduc event, as interpreted on both the VSP-CDP and Kirchhoff
migrated displays of Figure 6.13, verifies the interpretation of the nondeconvolved
VSP-CDP and migrated data in the panels of Figure 6.12. The Z-marker (Figure 6.13)
rises gradually away from the location of the 13-15 well. The Leduc, although less dis-
cernable, rises gradually away from the borehole (basal Leduc) and toward the reef top
about 130 m from the borehole. The reflection from the top of the reef merges visually
with the Z-marker event.
These interpretations assisted in the construction of the Ireton to Leduc isochron
map shown in Figure 6.5.
182
MANNVILLE D WABAMUN Z-MARKER

NORDEGG IRETON LEDUC •
DEBOLT/BELLOY
ri~rJA"'J"'-J\.J4.;~\.~~~
II 1 I
1·l..;t'ptI'/"r---""'/'~\
••
1).. 15
"'IlL 0
._.u.
f' F .. I f III f'\ §
'/">'''I''''I'~\
I ("'>r1"1J !! [ilit
~ .-J~~ 8- 'J)
__ _ __ ....
51
0
DEPTH (m)
i i;i!i~~~~iii ! ~~~~3 ! S~ g~£ 5 ; ~!i~~js~ t l)

....
....=
tl)
"rj
I-'
....
tl)
co
r..J Q..
-
(j
~
[/l
tl)
'J)
....
r::
-~: Q..
~ '<
~
-~
~
--
Fig. 6.14. Integrated interpretive display showing the interpretation of the available exploration data for the Simonette reef case study.
Integrated interpretion
On the left-hand side of the integrated interpretive display shown in Figure 6.14,
sonic logs for wells 16-16 and 13-15, nondeconvolved, near-offset, and inside and outside
corridor stacks are time-tied to the post-VSP interpretation of the surface-seismic data.
On the right-hand side, the Z"up(decon)(+TT) data is time-tied to both the welll3-15 sonic
log and to the deconvolved, near-offset outside corridor stack. The horizontal (depth
axis) of the Z"up(decon)(+1T) data corresponds to the same scale used for the well 13-15
sonic and gamma ray log depth display.
The correlated data shown in Figure 6.14 allow a confident interpretation of the sur-
face-seismic line and the identification of the Mannville, Nordegg, Debolt/Belloy,
Wabamun, Ireton, Z-marker, and Leduc events. The Leduc event is tied exactly at well 16-16
using the integrated sonic log from that well. Note that within the inter-reef shale interval,
the sonic-log-based synthetic seismogram is a poor fit to the VSP and surface-seismic data
for the Ireton, Z-marker, and Leduc events, exhibiting up to a 5-ms mistie. The sonic mea-
surements could be misleading. For example, wellbore effects (such as washouts), or the
increased concentration of heavy drilling fluids injected into the borehole (intended to pre-
vent a blowout) could have changed the sonic character of strata in the vicinity of the well-
bore. Since the well was left open during the interpretation of the VSP survey data, it is rea-
sonable to assume that the well engineers took precautionary steps to stabilize the fluids
within the well. Alternatively, these misties could be related to wavelet variability or phase
problems with the data. Whatever the source, such discrepancies between sonic-log-based
synthetic seismograms and seismic data provide additional justification for acquiring VSP
data.
The preferred interpretation of the surface-seismic section (shown as a normal
polarity display in Figure 6.15) differs slightly in several respects from the prewell inter-
pretation (Figure 6.3). Of particular significance is that on the updated version, the Z-
marker is present at the 13-15 well location, indicating that the well is situated in a flank
position. The reinterpretation of the seismic line exhibits a flat reef extending from the
borehole out to 100-120 m from the borehole. This is in agreement with the interpreted
near- and far-offset VSP data. Beyond the coverage of the VSP, the reef crests. With
respect to lateral variations in the thickness of the inter-reef shale isochron values (shale
thinning is indicative of reefal thickening), the following interpretations can be made:
1) the inter-reef shale isopach, as derived from the VSP data, is 136 m (isochron value of
63 ms) on the trace nearest the 13-15 well;
2) the shale is 120 m thick (55 ms) at distances on the order of 100 m laterally from the
well; and
184
WEST EAST
16-16 13-15 250 m
_1.4
-
t/)
<. c. ' ):
i'i.
MANNVILLE
NORDEGG
0
DEBOLT/BELLOY •
WABAMUN
IRETON •• Z-MARKER
LEDUC ••
Fig. 6.15. Current. preferred interpretation of the example seismic section . The Z-marker
is laterally continuous at the 13-15 well location and the reef slope has been reinterpret-
ed to be at least 150 m west of well 13-15 .
185
3) the shale is about 100 m thick (47 ms) at traces around 150 m from the well.
On the basis of the reinterpreted surface-seismic and VSP data, a revised inter-reef
shale isochron map (incorporating available well and surface-seismic data) was drafted
(Figure 6.5). The 13-15 exploratory well is shown to be in a reef flank position and indi-
cates that the crest of the low-relief reef is more than 150 m west of well 13-15.
Summary
The 13-15-63-25 W5M exploratory well was drilled into low-relief Leduc reef in the
Simonette area, Alberta. The well was expected to intersect the crest of the reef and to
encounter about 50-60 m of pay. Unfortunately, according to the interpretation of all of the
well data, it appears that the well was drilled into a flank position and was ultimately aban-
doned. The decision to abandon the well, as opposed to whipstocking, was made after the
acquisition and interpretive processing of near- and far-offset VSP data, and after the
reanalysis of existing surface-seismic data. The VSP data were acquired and interpreted
while the drill rig remained on-site, awaiting the decision to whipstock or abandon.
On the basis of VSP data, the operator was able to determine an accurate tie between
the surface-seismic data (Figure 6.15) and the subsurface geology, identify a mistie between
the surface-seismic data and sonic log seismogram, determine that the reef was not close
enough to well 13-15 to make whipstocking a viable option (given the production penalties
involved in drilling out of the target area), identify surface-generated and interbed multi-
ples, and ascertain their effect on the surface-seismic.
186
References
AGAT Laboratories, 1988, Table of formations of Alberta: AGAT Laboratories, Calgary.

Ahmed, H., 1989, Application of mode-converted shear waves to rock-property estima-
tion from vertical seismic profiling data: Geophysics, 54, 478-485.
Ahmed, H., 1990, Investigation of azimuthal anisotropy from offset VSP data - a case
study: First Break, 8, No. 12,449-457.
Anderson, N.L., 1986, An integrated geophysical!geological analysis of the seismic signa-
tures of some western Canadian Devonian reefs: PhD. thesis, University of Calgary.
Anderson, N.L., and Brown, R]., 1987, The seismic signature of some western Canadian
Devonian reefs: J. Can. Soc. Expl. Geophys., 23,7-26.
Anderson, N.L., Brown, RJ., Hinds, RC and Hills, L.V, 1989a, Seismic signature of a
Swan Hills (Frasnian) reef reservoir, Snipe Lake, Alberta: Geophysics, 54, 148-157.
Anderson, N.L., Brown, RJ. and Hinds, RC, 1989b, Low- and high-relief Leduc forma-
tion reefs: A seismic analysis: Geophysics, 54, 1410-1419.
Anderson, N.L., Brown, RJ., Gendzwill, D.]., Hinds, RC and Lundberg, RM., 1989c,
Elk Point carbonate reservoirs, in Anderson, N.L., Hills, L.V and Cederwall,
D.A., Eds., Geophysical atlas of western Canadian hydrocarbon pools: Can. Soc.
Expl. Geophys./Can. Soc. Petro Geol., 27-66.
Anderson, N.L., White, D. and Hinds, RC, 1989d, Woodbend group reservoirs. tn
Anderson, N.L., Hills, L.V and Cederwall, D.A., Eds., Geophysical atlas of west-
ern Canadian hydrocarbon pools: Can. Soc. Expl. Geophys./Can. Soc. Petro Geol.,
101-132.
Arce, G.R and McLoughlin, M.P., 1984, Theoretical analysis of the Max/Median filter:
Proceedings of the 22nd Annual Allerton Conference on Communications,
Control, and Computing, program with abstracts.
Arce, G.R, Gallagher, N.C and Nodes, N.A., 1986, Median filters, theory for one- and
two-dimensional filters, in Huang, T.5., Ed., Advances in computer vision and
image processing: JAL Press Inc., 2, 89-166.
Balch, A.H., and Lee, M.W., Eds., 1984, Vertical seismic profiling - techniques, applica-
tions and case histories: Internat. Human Res. Dev. Corp.
Barclay, J.E., 1988, The Lower Carboniferous Golata Formation of the Western Canada
Basin, in the context of sequence stratigraphy, in James, D.P. and Leckie, D.A.,
Eds., Sequences, stratigraphy, sedimentology: Surface and subsurface: Can. Soc.
Petro Geol., Memoir IS, 1-14.
187
Barclay, J.E., Krause, EE, Campbell, RL and Utting, J., 1990, Dynamic casting and
growth faulting: Dawson Creek Graben Complex, Carboniferous-Permian Peace
River Embayment, Western Canada, in O'Connell, S.c. and Bell, J.5., Eds., Bull.,
Can. Petro Geol., 38A, 115-145.

Barton, D.C., 1929, The seismic method of mapping geologic structure: Geophys. Prosp.,
1,572-624.
Belaud, D., Christie, P., de Montmollin, v., Dodds, K, James, A, Kamata, M., and
Schaffner, J., 1988, Detecting seismic waves in the borehole: Technical Review, 36,
18-29.
Belaud, D., and Leaney, W.s., 1991, P-S wave separation using parametric inversion: an
offset VSP case study: 53rd Mtg. Eur. Assoc. Expl. Geophys., Abstracts, 516-517.
Beydoun, W.B., 1985, Asymptotic wave methods in heterogeneous media: PhD. thesis,
MIT.
Beydoun, W.B., and Keho, T.H., 1987, The paraxial ray method: Geophysics, 52, 1639-
1653.
Bracewell, R, 1986, The Fourier transform and its applications, 2nd Ed.: NcGraw-Hill,
New York.
Brown, RJ., Anderson, N.L., and Hills, L.V, 1990, Seismic interpretation of Upper Elk
Point (Givetian) carbonate reservoirs in Western Canada: Geophys. Prosp., 38,
719-736.
Bullen, KE., and Bolt, B.A, 1985, An introduction to the theory of seismology. 4th edi-
tion, Cambridge Univ. Press.
Cant, D.J., 1988, Regional structure and development of the Peace River Arch, Alberta: A
Paleozoic failed-rift system?: Bull., Can. Petro Geol., 36, 284-295.
Carswell, A, and Moon, W.M., 1989, Application of multioffset vertical seismic profiling
in fracture mapping: Geophysics, 54, 737-746.
Cassell, B.R, 1984, Vertical seismic profile - An introduction: First Break, 2, No. 11,9-19.
Cerveny, V, 1985, The application of ray tracing to the propagation of shear waves in com-
plex media, in Dohr, G.P. Ed., Seismic shear waves, 15A, Part A: Theory; in Helbig,
K and Treital, S., Eds., Chapter 1, Seismic exploration: Geophysical Press, 1-124.
Cerveny, v., Molotkov, LA and Psencik, 1., 1977, Ray method in seismology: Univerzita
Karlova.
Cerveny, V, and Hron, E, 1980, The ray series method and dynamic ray tracing for three-
dimensional inhomogeneous media: Bull. Seis. Soc. Am., 70,47-77.
Cerveny, V., Popov, M.M., and Psencik, LA, 1982, Computation of wave fields in inhomo-
geneous media - Gaussian beam approach: Geophys. J. Roy. Astr. Soc., 70, 109-128.
188
References
Coulombe, CA, Stewart, RR, and Jones, M.J., 1991, AVO analysis using the VSP: 61st
Ann. Intern. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1080-1083.
Coulombe, CA, Stewart, RR, and Jones, M.J., 1992, Elastic wave AVO using borehole
seismic data: 62nd Ann. Intern. Mtg., Soc. Expl. Geophys., Expanded Abstracts,
864-866.
Coulombe, CA, 1993, Amplitude-versus-offset analysis using vertical seismic profiling
and well-log data: M.5c. thesis, Univ. of Calgary.
Deans, S.R, 1983, The Radon transform and some of its applications: Wiley Interscience.
Devaney, AJ. and Oristaglio, M.L., 1986, A plane-wave decomposition for elastic wave
fields applied to the separation of P-waves and S-waves in vector seismic data
(short note): Geophysics, 51, 419-423.
Dillon, P.B. and Thomson, RC, 1984, Offset source vertical-seismic-profile (VSP) surveys
and their image reconstruction: Geophys. Prosp., 32, 790-811.
Dillon, P.B., 1990, A comparison between Kirchhoff and GRT migration on VSP data:
Geophys. Prosp., 38, 757-778.
DiSiena, J.P., Gaiser, J.E. and Corrigan, D., 1981, Three-component vertical seismic profile
- orientation of horizontal components for shear wave analysis: 51st Ann.
Internat. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 1990-2011.
DiSiena, J.P., Gaiser, J.E. and Corrigan, D., 1984, Horizontal components and shear wave
analysis of three-component VSP data, in Toksoz, N.M. and Stewart, RR, Eds.,
Vertical seismic profiling: Advanced concepts: Geophysical Press, 177-188.
Dix, CH., 1939, The interpretation of well-shot data: Geophysics, 4, 24-32.
Dix, CH., 1945, The interpretation of well-shot data II: Geophysics, 10, 160-170.
Dix, CH., 1946, The interpretation of well-shot data III: Geophysics, 11, 457-461.
Dupal, L. and Miller, D. E., 1985, Reef delineation by multiple offset borehole seismic pro-
files: 55th Ann. Intern. Mtg., Soc. Expl. Geophys., Expanded Abstracts.
Dupal, L., Ekmann, G., Hansen, R, and Opedal, A, 1993, Triaxial offset VSP in a hori-
zontal well in the Gullfaks field: 55th Mtg. Eur. Assoc. Expl. Geophys., Abstracts.
Durrani, T.S. and Bisset, D., 1984, The Radon transform and its properties: Geophysics,
49, 1180-1187.
Ferry, RM., 1989, Beaverhill Lake group carbonate reservoirs, in Anderson, N.L., Hills,
L'V. and Cederwall, D.A, Eds., Geophysical atlas of Western Canadian hydrocar-
bon pools: Can. Soc. Expl. Geophys./Can. Soc. Petro Geol., 67-99.
Fessenden, RA, 1917, Method and apparatus for locating ore bodies: U.S. Patent No.
1,240,328.
189
Fitch, E.P, Coyle, E.J. and Gallagher, N.C., 1984, Median filtering by threshold decom-
position: IEEE Transactions Acoustics, Speech, Signal Processing, ASSP-32, 1183-
1188.
Freire, S.L.M. and Ulrych, T.J., 1988, Application of singular value decomposition to ver-
tical seismic profiling: Geophysics, 53, 778-785.
Gaiser, J.E., DiSiena, J.P and Fix, J.E., 1984, Vertical seismic profiles: Fundamentals of the
downgoing wavefield and applications that improve COP data interpretation, in
Toksoz, N.M. and Stewart, RR, Eds., Vertical seismic profiling: Advanced con-
cepts: Geophysical Press. 87-112.
Gallagher, N.C. and Wise, G.L., 1981, A theoretical analysis of the properties and conver-
gence rates of median filters: IEEE Transactions Acoustics, Speech, Signal
Processing, ASSP-33, 230-239.
Gal'perin, E.L, 1974, Vertical seismic profiling, in White, J.E., Ed., Special Publication no.
12: Soc. Expl. Geophys.
Gardiner, L.W., 1949, Seismograph determination of salt-dome boundaries using well
detection deep on zone flanks: Geophysics, 14, 1100-1107.
Gazdag, J., 1978, Wave equation migration with the phase-shift method: Geophysics, 43,
1342-135l.
Gazdag, J. and Squazzero, P, 1984, Migration of seismic data by phase-shift plus inter-
polation: Geophysics, 49, 124-13l.
Geis, W.T., Stewart, RR, Jones, M.J., and Katopodis, PE., 1990, Processing, correlating,
and interpreting converted shear waves from borehole data in southern Alberta:
Geophysics, 55, 660-669.
Hale, D., 1984, Dip-moveout by Fourier transform: Geophysics, 49, 741-757.
Haldorsen, J., Miller, D., Walsh, J., and Zoch, H.-I., 1992, A multichannel approach to sig-
nature estimation and deconvolution for drill bit imaging: 62nd Ann. Intern. Mtg.,
Soc. Expl. Geophys., Expanded Abstracts, 181-184.
Haldorsen, J., Miller, D., and Walsh, J., 1994, Multichannel Weiner deconvolution of ver-
tical seismic profiles: Geophysics, 59, 1500-15
Hampson, D., and Mewhort, L., 1983, Using a vertical seismic profile to investigate a
multiple problem in Western Canada: J. Can. Soc. Expl. Geophys., 19, 16-33.
Hardage, B.A., 1981, An examination of tube-wave noise in vertical-seismic-profiling
data: Geophysics, 46, 892-903.
Hardage, B.A., 1985, Vertical seismic profiling, 2nd Ed.: Geophysical Press, London.
Hardage, B.A., 1992, Crosswell seismology and reverse VSP: Geophysical Press.
Hatton, L., Worthington, M.H., and Makin, J., 1986, Seismic data processing: Blackwell.
190
References
Hinds, RC, Levy, S., Stinson, K. and Hajnal, Z., 1986, The Karhunen-Loeve transform as
a method of wavefield separation in VSP processing: Can. Soc. Expl. Geophys.
Ann. Mtg., abstract with program.
Hinds, RC, Kuzmiski, RK., Botha, W.J. and Anderson, N.L., 1989, Vertical and lateral
seismic profiles: in Anderson, N.L., Hills, L'V, and Cederwall, D.A., Eds.,
Geophysical atlas of western Canadian hydrocarbon pools: Can. Soc. Expl.-
Geophys./Can. Soc. Petro Geol., 319-344.
Hinds, RC, Kuzmiski, RD. and Anderson, N.L., 1991a, Clastic reservoir and fault delineation:
Fort St. John Graben area: Can. Soc. Expl. Geophys. Ann. Mtg., abstract with program.
Hinds, RC, Kuzmiski, RD. and Anderson, N.L., 1991b, Delineation of a low-relief reef:
An integrated seismic perspective: Can. Soc. Expl. Geophys. Ann. Mtg., abstract
with program.
Hinds, RC, 1991c, Seismic signatures and integrated interpretation: invited paper for the
South African Geophysical Association Nuusbrief, 2, July issue.
Hinds, RC, Kuzmiski, RD., Anderson, N.L., and Richards, B.R, 1993a, An integrated
surface and borehole seismic case study: Fort St. John Graben area, Alberta,
Canada: Geophysics, 58, 1662-1675.
Hinds, RC, Anderson, N.L., and Kuzmiski, RD., 1993b, An integrated surface seismic/seis-
mic profile case study: Simonette area, Alberta: Geophysics, 58, 1676-1688.
Hinds, RC, Kuzmiski, RD., and Anderson, N.L., 1993c, An integrated surface seis-
mic/seismic profile case study of the Leduc Formation reef, Ricinus Field,
Alberta, Canada: J. Can. Soc. Expl. Geophys., 29, 440-451.
Hinds, RC, and Durrheim, RJ. , 1993, The suppression of sea-floor multiples: a case
study from the Pletmos Basin: South African Geophysical Association third annu-
al technical meeting, Cape Town, extended abstracts, 90-94.
Hinds, RC, Anderson, N.L., and Kuzmiski, RD., 1994a, An integrated surface seis-
mic/ seismic profile case study of a misinterpreted seismic anomaly associated
with Leduc Formation reef, Lanaway Field, Alberta, Canada: Comput. and
Geosci., 20, 53-73.
Hinds, RC, Kuzmiski, RD., Anderson, N.L., and Richards, B.R, 1994b, On: "An integrated
surface and borehole seismic case study: Fort St. John Graben area, Alberta, Canada"
(Ronald C Hinds, Richard Kuzmiski, Neil L. Anderson, and Barry Richards), by M.
M. Roksandic (with reply from authors), Geophysics, 59, 1171-1172.
Hinds, RC, 1994c, Interpretation of vertical and lateral seismic profiles: Some case his-
tories: PhD. thesis, Univ. of Pretoria.
191
Hinds, RC., and Durrheim, RJ., (in review), Comparison of methods to attenuate sea-
floor multiples: A case study from the Pletmos Basin, South Africa: Southern
African Geophysical Review.
Holste, W., 1959, Problems and results with reflection and refraction seismics in bore-
holes: Geophys. Prosp., 7, 231-240.
Hotelling, H., 1933, Analysis of complex statistical variables into principal components:
J. Educ. Psychol., 24, 417-438, 498-520.
Hu, L.-Z. and McMechan, GA, 1987, Wave-field transformations of vertical seismic pro-
files: Geophysics, 52, 307-321.
Hubbard, T.P., 1979, Deconvolution of surface recorded data using vertical seismic pro-
files: Presented at the 49th Ann. Internat. Mtg., Soc. Expl. Geophys.
Jackson, GM., Mason, I.M., and Greenhalgh, S.A, 1991, Principal component transforms
of triaxial recordings by singular value decomposition: Geophysics, 45, 528-533.
Jones, I.F., 1985, Applications of the Karhunen-Loeve transform in reflection processing:
Ph.D. thesis, Univ. of British Columbia.
Jones, I.F., and Levy, S., 1987, Signal-to-noise ratio enhancement in multichannel seismic
data via the Karhunen-Loeve transform: Geophys. Prosp., 35, 12-32.
Jones, M.J., and Fung, M., 1990, Case history using a multioffset VSP to map Devonian
reef: 58th Ann. Internat. Mtg., Soc. Expl. Ceophys.. Expanded Abstracts.
Kanasewich, E., 1981, Time sequence analysis in geophysics: Univ. of Alberta Press.
Kappus, M.E., Harding, AJ., and Orcutt, J.A, 1990, A comparison of tau-p transform
methods: Geophysics, 55, 1202-1215.
Karhunen, K., 1947, Uber lineare methoden in der Wahrscheinlichkeitsrechnung: Ann.
Acad. Sci. Fenn., 37, 1-79.
Kelamis, P.G and Mitchell, AR, 1989, Slant-stack processing: First Break, 7, 43-54.
Kennett, P., Ireson, RL. and Conn, P.J., 1980, Vertical-seismic-profiles - their applications
in exploration geophysics: Geophys. Prosp., 28, 676-699.
Khalil, AA, Stewart, RR, and Henley, D.C., 1993, Full-waveform processing and inter-
pretation of kilohertz cross-well seismic data: Geophysics, 58, 1248-1256.
Kirlin, RL., 1987, Image coding for seismic wavefront analysis, in Aminzadah, F., Ed.,
Pattern recognition and image processing: Geophysical Press, 157-186.
Klovan, J.E., 1964, Facies analysis of the Redwater reef complex, Alberta, Canada: Bull.
Can. Petro Geol., 12, 2260-2281.
Kommedal, J.H. and Tjostheim, B.A, 1989, Tutorial: A study of different methods of
wavefield separation for application to VSP data: Geophys. Prosp., 37, 117-142.
192
References
Kramer, H.P., and Mathews, M.V., 1956, A linear coding for transmitting a set of correlat-
ed signals: IRE Trans. Inform. Theo., IT-2, 41-46.
Lee, M.W., 1984, Processing of vertical seismic profile data, in Simaan, M., Ed., Advances
in geophysical data processing: 1, JAI Press Inc., 129-160.

Lee, M.W. and Balch, AH., 1983, Computer processing of vertical-seismic-profile data:
Levin, EK., and Lynn, RD., 1958, Deep hole geophone studies: Geophysics, 23, 639-664.
Li, G., 1994, Crosswell seismic processing: automatic velocity analysis, filtering, and
reflection imaging: M.sc. thesis, Univ. of Calgary.
Li, G., and Stewart, RR, 1994, Common-reflection-point stacking of crosswell seismic
data: Friendswood, Texas: 64th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Extended abstracts, 42-45.
Loeve, M., 1948, Functions aleatories de second ordre: Chapter 8, 299-352, Hermann,
Paris.
Loeve, M., 1955, Probability theory: D. van Nostrand, New York.
March, D.W. and Bailey, AD., 1983, A review of the two-dimensional transform and its
use in seismic processing: First Break, 1, No.1, 9-21.
McCollum, B., and LaRue, W.W., 1931, Utilization of existing wells in seismograph work:
Bull. Am. Assn. Petro Geol., 15, 1409-1417.
McNamara, L.B., and Wardlaw, N.C., 1991, Geological and statistical description of the
Westerose reservoir, Alberta: Bull. Can. Petro Geol., 39, 332-351.
Miao, X.G., Moon, W.M., and Milkereit, B., 1995, A multioffset, three-component VSP
study in the Sudbury Basin: Geophysics, 60, 341-353.
Millahn, K., Zerouk, K., and Tufekcic, D., 1983, VSP in deviated wells or with a moving
source: 54th Ann. Internat. Mtg., Soc. Expl. Geophys., Expanded abstracts, 587-
589.
Moon, W., Carswell, A, Tang, Rand Dilliston, c., 1986, Radon transform wave-field sep-
aration for vertical seismic profiling data: Geophysics, 51, 940-947.
Moore, P.E, 1988, Devonian geohistory of the western interior of Canada, in McMillan,
N.J., Embry, AE and Glass, D.J., Eds., Devonian of the world, Proc. Internat.
Symp. Devonian System, Calgary, Canada, 1, Regional synthesis: Can. Soc. Petro
Geol., Memoir 14, 67-87.
Moore, P.E, 1989a, Devonian reefs in Canada and some adjacent areas. in: Geldsetzer,
H.H.J., James, N.P. and Tebbutt, G.E., Eds., Reefs, Canada and adjacent areas: Can.
Soc. Petro Geol., Memoir 13, 367-390.
193
Moore, EE, 1989b, The Kaskaskia Sequence: Reefs, platforms and foredeeps, The lower
Kaskaskia Sequence - Devonian, in Ricketts, B.D., Ed., Western Canada sedimen-
tary basin, a case history: Can. Soc. Petro. Geol., 139-164.
Mossop, GD., 1972, Origin of the peripheral rim, Redwater reef, Alberta: Bull. Can. Petro.
Geol., 20, 238-280.
Mountjoy, E., 1980, Some questions about the development of Upper Devonian carbon-
ate buildups (reefs), Western Canada: Bull. Can. Petro. Geol., 28, 315-344.
Naess, O.E., 1989, Model-based transformations of common midpoint gather: Geophys.
Prosp., 37, 781-808.
Navarte, EE., 1946, On well velocity data and their applications to reflection shooting:
Noble, MD., Lambert, R.A, Ahmed, M., and Lyons, J., 1988, The application of three-
component VSP data on the interpretation of the Vulcan gas field and its impact
on field development: First Break, 6, No.5, 131-149.
Nodes, T.A and Gallagher, N.C., 1982, Median filters: Some modifications and their
properties: IEEE Transactions Acoustics, Speech, Signal, Processing, ASSP-30, 739-
746.
Noponen, I., 1988, Application of slant moveout on VSP data to display reflector dips:
58th Ann. Internal. Mtg., Soc. Expl. Geophys., Expanded Abstracts, 809-811.
O'Connell, S.c., 1990, The development of the Lower Carboniferous Peace River
Embayment as determined from the Banff and Pekisko formation depositional
patterns: Bull. Can. Petro Geol., 38A, 93-114.
O'Connell, S.c., Dix, G.R. and Barclay, J.E., 1990, The origin, history and regional struc-
tural development of the Peace River Arch: Bull. Can. Petr. Geol., 38A, 4-24.
Rennie, W., Leyland, W. and Skuce, A, 1989, Winterburn (Nisku) reservoirs, in Anderson,
N.L., Hills, L'V, and Cederwall, D.A, Eds., Geophysical atlas of western Canadian
hydrocarbon pools: Can. Soc. Expl. Geophys./Can. Soc. Petro Geol., 133-153.
Richards, B.C., 1989, Upper Kaskaskia Sequence, uppermost Devonian and Lower
Carboniferous, in Ricketts, BD., Ed., The Western Canadian Sedimentary Basin, a
case history: Can. Soc. Petro Geol., 165-201.
Richards, B.C., 1990, Tectonic and depositional history of the Early Carboniferous Peace
River Embayment, Alberta and British Columbia, in Basin perspectives: Can. Soc.
Petro Geol., abstracts with program, 118.
194
References
Richards, B.C., Barclay, J.E., Bryan, D., Hartling, A., Henderson, C.M., and Hinds, RC.,
1994, Carboniferous strata of the Western Canada Sedimentary Basin, in Mossop,
G. and Shetson, 1., Eds., Geological Atlas of the Western Canada Sedimentary
Basin: Chapter 14, Alberta Research Council, 221-250.

Robinson, E.A., 1967, Multichannel time series analysis, with digital computer programs:
Holden Day.
Robinson, E.A., 1983, Multichannel time series analysis, with digital computer programs,
2nd Ed.: Goose Pond Press.
Smidt, J.M., 1989, VSP processing with full downgoing wavefield deconvolution applied
to the total wavefield: First Break, 7, No.6, 247-257.
Stewart, RR, 1985, Median filtering: Review and a new F/K analogue design: J. Can. Soc.
Expl. Geophys., 21, 54-63.
Stewart, RR, 1989, Integrated seismic analysis: Kidney area, northern Alberta, Canada:
Geophysics, 54, 1240-1248.
Stewart, RR, and DiSiena, J.P, 1989, The values of VSP in interpretation: The Leading
Edge, 8, No. 12, 16-23.
Stoakes, EA., 1980, Nature and control of shale basin fill and its effect on reef growth and
termination: Upper Devonian Duvernay and Ireton Formations of Alberta,
Canada: Bull. Can. Petro Geol., 28, 345-410.
Stoffa, EL., Buhl, P., Diebold, J.B., and Wenzel, E, 1981, Direct mapping of seismic data to
the domain of intercept time and ray parameter - a plane wave decomposition:
Geophysics, 46, 3, 255-267.
Stokes, EA., and Wendte, J.c., 1987, The Woodbend Group, in Krause, EE and Burrows,
O.G., Eds., Devonian lithofacies and reservoir styles in Alberta: 2nd Internal.
Symp. Devonian System, 153-170.
Stone, D. G., 1981, VSP- the missing link. Paper presented at the VSP short course spon-
sored by the Southeastern Geophys. Soc. in New Orleans.
Tariel, E and Michon, D., 1984, On vertical-seismic-profile processing: Geophys. Prosp.,
32,775-789.
Toksoz, M.N., and Stewart, RR, 1984, Vertical seismic profiling- part B- Advanced top-
ics: Geophysical Press.
Trotter, R, 1989, Sedimentology and depositional setting of the Granite Wash of the
Utikima and Red Earth areas, north-central Alberta: M.5c. thesis, Dalhousie
University.
Turner, G., 1990, Aliasing in the Tau-p transform and the removal of spatial aliased coher-
ent noise: Geophysics, 55, 1496-1503.
195
Watanabe, S., 1965, Karhunen-Loeve expansion and factor analysis. Theoretical remarks
and applications, in J. Sklansky, Ed., Pattern recognition: Stroudsburg, Penn.,
1973, 146- 17l.
Wendte, J.c., and Stoakes, EA, 1982, Evolution and corresponding porosity distribution
of the Judy Creek reef complex, Upper Devonian, central Alberta, in Cutler, W.G.,
Ed., Canada's giant hydrocarbon reserves: CSPG-AAPG Core conference, Can.
Soc. Petro Geol., 63-8l.
Wiggins, J.W., 1984, Kirchhoff integral extrapolation and migration of nonplaner data:
Geophysics, 49, 1239-1248.
Wiggins, J.W., and Lavander, AR., 1984, Migration of multiple offset synthetic vertical
seismic profiles in complex structure, in Simann, M., Ed., Advances in
Geophysical Data Processing, 1, 269-290, JAL Press Inc.
Wiggins, J.W., Ng.P., and Mazur, A, 1986, The relation between the VSP-CDP transfor-
mation and VSP migration: 55th Ann. Internat. Mtg., Soc. Expl. Geophys.,
Expanded Abstracts, 565-568.
Wuenschel, P.c., 1976, The vertical array in reflection seismology - some experimental
studies: Geophysics, 41, 219-232.
Wyatt, K.D., and Wyatt, S.B., 1981, Determination of subsurface structural information
using the vertical seismic profile: 51st Ann. Internat. Mtg., Soc. Expl. Geophys.,
Extended abstracts, 1915-1949.
Wyatt, K.D. and Wyatt, S.B., 1982, Downhole vertical seismic profile survey reveals struc-
ture near borehole: Oil and Gas Journal, 80, 77-82.
Yilmaz, 0., 1987, Seismic data processing. Investigations in Geophysics: 2, Soc. Expl.
Geophys.
Zimmerman, L.J., and Chen, S.T., 1993, Comparison of vertical seismic profiling tech-
niques: Geophysics, 58, 134-140.
196
Appendix: Mathematical Review
In this Appendix, the median, K-L, f-k and t-p filtering, VSP deconvolution, and
the matrix equations involved in the hodogram-based and time-variant polarizations are
reviewed. These processes represent some of the fundamental processes involved in VSP
data processing.
Median filtering
One wavefield separation method utilizes a 1-D median filter combined with a
band-pass filter. The band-pass filter eliminates the median filter "whiskers" (Hardage,
1985) resulting from the nonlinear operation. The theoretical basis of the median filter has
been reviewed in Gallagher and Wise (1981), Nodes and Gallagher (1982), Fitch et al.,
(1984), Arce and McLoughlin (1984), and Arce et al., (1986).
The input to the median filter is a selected window of data. The length of the win-
dow can be an even or an odd number of points (2N or 2N+ 1). The two ends of the time
series are padded with N additional points in order to accommodate the center location
of the window. The windowed data are sorted according to magnitude with the center
value of the sort being termed the median value. For the odd point filter, the median
value at the center of the windowed time series becomes the new value of the output
series. When N is even, the mean of the two middle median values is the output of the
filter. This new point of the output data is placed at the location of the center of the win-
dow of the input series. For the 1-D median filter application, a new output time series is
generated as the window slides across the input series, one point at a time.
The median filter can be defined as the rearrangement of the windowed time
series according to size. The output of the nonrecursive median filter, y(t), is given by
(Arce et al., 1986) as
y(t) =median [xii-N), ...,x(t-1)~x(t),x(t+1), ...,x(t+N)},
and the recursive filter is given as
y(t) = median (y(t-N), ...,y(t-1),x(t),x(t+1),...,x(t+N)}
for a window of length 2N+l samples centered at location t of the input time series.
The type of median filter used for the wavefield separation operation is a nonre-
cursive median filter. The operation of a recursive filter differs from the operation of a
nonrecursive filter, where the previously determined output data is used to compute fur-
ther output data. The options that are available for the nonrecursive median filters
197
include normal and tapered median filters. The tapered filters filter the sorted values
using a boxcar, triangular, or cosine filter. The sum of a prespecified number of the cen-
tral tapered values is the new median value. A tapered median filter eliminates the need
for postmedian band-pass filtering.

When the median filter operation is applied to Z(-TT) data, the divergent
upgoing waves appear as a triangular anomaly. The triangular anomaly can be filtered
out using a suitable length median as shown in Figure 5-27 of Hardage (1985). The
median-filter operation smooths the amplitude (phase) variations of the first-break
downgoing event over several traces. A scaling program is applied to restore the
Zdown(-TT) data to an amplitude similar to a selected window in the Z(- TT) input
data.The window usually is comprised of a zone surrounding the first-break wavelet.
This zone is restored to the amplitude range of a similar window around the Z(-TT)
first-break wavelet using a multiplicative factor determined using a least-squares ratio
fit, or a ratio computed as the inverse of the absolute amplitudes over the window.
Following the scaling of the Zdown(-TT) data, subtraction of the Zdown(-TT) data
from the Z(-TT) data yields the Zup(-TT) data.
Karhunen-Loeve (K-L) filtering

The rationale for using the K-L transform for wavefield separation is explained in
Chapter 2. The eigenanalysis will dissect the VSP data with N traces into N eigenimages
(N eigenvalues and N corresponding eigenvalues). The degree of linear coherency is
reflected in the magnitude of the eigenvalue. An excellent review of the method as it is
used in VSP work is given in Jackson et al., (1991) and Hardage (1992). The method has
also been referred to as eigenvector coding (Kirlin, 1987).
For VSP data, we can have N traces and M time sample points. In the derivation,
we assume that there are more time samples than traces (depth recordings). The data
matrix is formed by placing the VSP depth traces to be the rows of the data matrix, {x/O,
i=l, ...,N; 1£ t£MJ. We form the cross-energy or covariance matrix as the outer product of
the data matrices
- - - T
M=XX.
The covariance matrix can be spectrally decomposed (using singular value decomposi-
tion or SVD) to form
198
The Karhunen-Loeve or the principal components are
K- = W- T X,
-
where
is the eigenvector matrix calculated using SVD of the cross-energy matrix. The eigenvec-
tors are
The eigenvalue matrix L has as the trace of the matrix, the corresponding eigenvalues
(also calculated during the SVD of the cross-energy matrix)
The eigenvalues are examined to determine the number of eigenvectors to be

included in the reconstruction. In the case of the isolation of the downgoing events from
the Z(-TT) data, the first few (corresponding to the largest eigenvalues) eigenvectors are
chosen. The downgoing events are reconstructed by performing the inverse transform
using only those eigenvectors selected to represent the downgoing events. If the first J
eigenvectors are chosen, then
grecon = WK ,
where only the J-chosen principal components and eigenvectors are used. This means
that some of the columns of the eigenvector matrix and some of the rows of the principal
component matrix are not used or zeroed, depending which eigenvalues were chosen to
represent the downgoing events only.
In summation form, this would appear as
J
recon ~ •
Xi (t) =.{-t Wij k/t) 1=1 ,... ,N; 1 s t~M ,
]=1
and in general, partial reconstruction can be seen as (for a misfit analysis; Jones, 1985)
199
p
xl,recon( t ) = ~
L W ti kj ()
t l. =1 ,... , N ; m s p s N; Ist s M
j=m+l
for integers m and p whose choice are dependent on the aim of the reconstruction.
f-k filtering
In 2-D wavefield transformations, a linear event in the z-i domain becomes a lin-
ear event in the f-k domain. The transformation equations for the forward and reverse f-
k transforms are
for the forward transform and
for the reverse transform.

The term f-k is used loosely since the Fourier transform is usually expressed in
terms of and kz (Hu and McMechan, 1987). The up- and downgoing linear events con-
tained in the VSP data are mapped into linear events in the positive and negative k quad-
rants of the f-k domain. It can be shown to be the case from the following brief derivation.
From Robinson (1967), the equation for a line will be
Z
t = + to '
V
and we can form a delta function
If we insert this function into the 2-D Fourier transform, then
-
V( kz , W ) = IJ 0 Z
(t- V - to) e i(CA>t-kZz) dZ dt
200
is the equation of a line in the f-k domain passing through the origin. The function has a
magnitude given by the real part of the equation (the delta function) and a phase equal
to -w to. The phase is associated with the location of the line within the z-t domain and is
linked to to which will be discussed later in the t-p domain.
In Chapter 2, numerous examples of the downgoing events being clustered in
a tight linear group in the positive k quadrant are shown. The slope of the linear events
in the f-k domain yields the apparent velocity Von the z-i plot (VSP FRT display) since
w = k z • V. The concept of spatial aliasing was discussed in Chapter 2 along with a
numerical example (also see DiSiena et a1., 1984; and pages 104-114 of Hardage 1985).
To avoid aliasing in thef-k domain, a depth increment (for the sonde locations) should
be used that satisfies the constraint (Hardage, 1985)
Vmin
2 f max
where V nzin is an estimation of the minimum strata velocity expected during the VSP run
(check the sonic log which is usually run before the VSP), andfnzax is the maximum expected
frequency of the data (what band-pass filter will be used in the final IPP panels?).
t-p filtering
The t-p filtering technique is related to the f-k filtering method (Figure 2.25). In
Figure 2.25, the slowness limits used in the t-p filtering are shown in the f-k domain as a
"pie-slice" accept zone, which is logical since the equation w =kz• V translates into kz=w .p.
The t-p filter is also called the "slant stack" since in the z-t domain (the VSP data),
the slant stack domain is calculated by performing individual sums along lines defined by
t=poz+to. po is related to the slope of the line of integration, which is tan a (a being the angle
that the line of integration makes with the z axis) and to being the t-intercept. This maps a
linear event into a point in the t-p domain. The calculation continues to include lines of
integration of all slopes (both positive and negative relating to the down- and upgoing
events, respectively) and t-intercepts (t's) and all z values. The depth (z) values do not need
201
to increase by a constant increment, which means that unequally spaced sonde locations
are not a problem. This would cause a problem with the f-k filter since the fast fourier
transform, FFT, desires both constant Dz and Dt.
The t-p transform is defined (Robinson, 1967; Deans, 1983; Hu and McMechan,
1987; Carswell and Moon, 1989; Kappus et al., 1990; Turner, 1990; and Hardage, 1992) to be
U( p , 't ) = J v( z .pz + 't ) dz .
The inverse transform is
V(Z,t) = f .!!...dt H[ U(p, t-pZ) ]dp

where Hi-] is the Hilbert transform operator (Durrani and Bisset, 1984).
In the inverse transform (described in detail in Robinson, 1967 and Hu and
McMechan, 1987), the trace increment Dz can be respecified to another value other than
the input value for the forward transform. This enables global or local trace interpolation,
which can infill missing depth levels (Hu and McMechan, 1987).
Since the up- and downgoing VSP events are opposite in sign with respect to
apparent velocity on the FRT display, specifying either the positive or negative p-range
as input to the forward t-p transform will, in effect, perform the wavefield separation. As
shown in Chapter 2, trace interpolation can be performed, and then another method can
be used to perform up- and downgoing event separation.
The use of the transform is data dependent, and we favor using all of the above
methods (median, K-L,f-k included) to create the IPPs. We recommend analyzing the out-
put before deciding which method, or combination of methods, is best for the data.
VSP deconvolution
The use of the downgoing events, Zdown(-TT), to design a deconvolution operator
has been called downward-traveling wave train deconvolution (Balch and Lee, 1984), Up
over Down deconvolution, and special VSP deconvolution among other names. The con-
cept is discussed in Balch and Lee (1984), Hardage (1985), and Hubbard (1979). In the
simplest case of the VSP data containing only primary and surface-generated multiple
events, the downgoing events represent all that is needed for the deconvolution of the
upgoing events. An alternative approach derives a multichannel deconvolution opera-
tor from the average downgoing waveform incorporating frequency-dependent sem-
blance weighting (Haldorsen et al., 1992, and Haldorsen et al., 1994).
202
From Gaiser et al., 1984, the reason for the name Up over Down deconvolution
can be seen. If we consider a VSP recording at a single level which has upgoing events
U(z,t) (originating from reflections below the sonde), and downgoing events D(z,t) (the
primary downgoing event plus surface-generated multiples), then the composite wave-
field seen on the trace from the Z(FRT) display is
V(Z,t) = D(z,t) ® [l-RC(z,t)] ,
where RC(z,t) is the reflectivity coefficient series in time, and the symbol f denotes a con-
volution operation. If we design a deconvolution operator from the downgoing waves,
namely D-l(Z,t), and convolve this with the v(z,t), then the operation will produce
D-1(Z,t) Q9 v(Z,t) =1 - RC(z,t) .
However, we can also do this in the Fourier domain where convolution with an
inverse operator of the downgoing events is equivalent to division. The process would
then be
U(Z, co ) _
U(Z, W)decon •
D(Z,w)
The Fourier transform of the Zup(-TT) is divided by the Fourier transform of the Zdown
(-TT) data; hence the name Up over Down deconvolution.
The downgoing multiple event resulting in an upgoing interbed multiple exists
on the sonde locations starting from the top generating interface (the interface that
reflects the primary upgoing wave back down) to deeper sonde locations. The upgoing
interbed event exists on traces from the lower generating interface sonde location and
upwards to the shallowest level. What this means is that the downgoing interbed events
needed to evaluate the corresponding upgoing interbed multiple are not present on the
traces recorded shallower than the top generating interface; the upgoing interbeds at
these levels may not be attenuated.
Hodogram-based single angle polarizations

The far-offset VSP data recorded on the X(FRT), Y(FRT), and Z(FRT) channels are
polarized to isolate the downgoing P-wave (or SV) events onto a single panel,
HMAX'(FRT). Polarization is done through two series of data rotations using hodogram
analysis (DiSiena et al, 1981; Balch and Lee, 1984; Gaiser et al., 1984; DiSiena et al., 1984;
Hardage, 1985; Hinds et al., 1989). The series of rotations are designed on the primary
203
downgoing wavelet since it is that type of event that we desire to isolate. Our assump-
tion is that the first-break wavelet is not contaminated by other wavefields such as upgo-
ing P- and 5V-events and downgoing 5V-events.
As reviewed in Chapter 2, the hodogram is constructed using a window of data

around the first-break wavelet. This is done interactively using a color-coded display that
enables the interpreter/processor to understand what part of the hodogram relates to
individual portions of the windowed data. The angle used in the rotation matrix is cho-
sen using a line through the hodogram display that can be rotated interactively, plus the
output data window is redisplayed each time the line is rotated. When the operator is
satisfied with the polarization, the angle is saved automatically. Many papers suggest
using least-squares fitting routines to make the angle decision; however, the essence of
interpretive processing is to make decisions based on viewing in detail the effect of the
processes on the data. This interaction cannot be done if "black-box" methods are used.
Once the rotation angle ql is chosen, then all of the time samples of the X(FRT)
and Y(FRT) data are rotated into the HMAX(FRT) and HMIN(FRT) output data accord-
ing to the matrix equation
The polarization of the HMAX(FRT) and Z(FRT) data into the HMAX'(FRT) and
Z'(FRT) data follows a similar procedure. The important aspect to note is that a SINGLE
angle is used to matrix rotate the entire trace, and that the angle is based on the primary
downgoing P-wave event.
Time-variant polarization
We could estimate a pseudotime-variant polarization using the same software as
the hodogram analysis if we could track all of the upgoing wave events on the Z'up(FRT)
and HMAX'up(FRT) data. For a given upgoing event that spans the traces beginning at
the trace for the interface to the shallowest trace, a hodogram analysis is done on every
trace for that particular upgoing event. This is done for all of the upgoing events, and an
angle-versus-time function is built for all of the traces. Interpolation of the angles
between the given times for the angles on a single trace gives us the time-variant angles
for that trace. The problems with this approach are that the upgoing events are being dis-
sected by other types of events and the signal-to-noise ratio of the upgoing events and
the background may not be high. This suggests that ray-tracing and modeling should be
done using all available velocity and model information from the zero and far-offset data.
204
The ray-tracing algorithm used in the time-variant polarization method for the
far-offset data in Chapters 4, 5, and 6 was paraxial ray tracing (Cerveny et al, 1977; Cer-
veny and Hron, 1980; Cerveny et al., 1982; Cerveny, 1985; Beydoun, 1985; Beydoun and
Keho, 1987; and Beydoun et al., 1990). The method is suitable since the ray tracing could
be sparsely performed, and then curvature corrections could be used to estimate ray trac-
ing at nearby locations.
After ray-tracing was performed through a model designed from the zero-offset
derived velocities and incorporating model restrictions given by interpreted far-offset
first-break times and upgoing reflection times, a series of polarization angles for various
reflections arriving on a single trace are computed. The single polarization angle q for a
trace is replaced by a time-variant angle, q(t), for a given trace. No polarization calcula-
tion are done prior to the first-break, and a constant angle is used for upgoing reflections
below the bottom hole interface. The matrix equation for the time-variant polarization
would now include a time-varying rotation angle
[
ZII(t)
.
1 = (ZDERor(t) HMAXDERor(t»
COS( O(t» -SIN(O(t») .
( SIN(O(t»
HMAX"(t) COS(O(t»
The Z"up( +TT) data are then used for interpretation and input into the VSP-CDP trans-
form and Kirchhoff migration algorithms.
205

VSP Data Interpretation and Processing PDF

Transféré par

Informations du document

Titre original

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

VSP Data Interpretation and Processing PDF

Transféré par

Droits d'auteur :

Formats disponibles

Downloaded 06/29/14 to 134.153.184.170. Redistribution subject to SEG license or copyright; see Terms of Use at http://library.seg.

THEORY AND PRACTICE

WAVENUMBER (1/m* 10's)

Edited by Bob A. Hardage

Society of Exploration Geophysicists

Hinds, Ronald C. (Ronald Clifford), 1954•

ISBN 1-56080-030-5 (Series)

¸1996 by the Societyof ExplorationGeophysicists

Printed in the United States of America

- Fourier transform(f-k) filtering 39

- Futuredirectionsfor •-p filtering in wavefield separation 69

VSP deconvolution and corridor stacks: near-offset data 71

-Corridor stack IPPs 73

VSP deconvolution: far-offset data 86

- Timevariant polarizationof Ricinuscasestudydata 88

- Far-offsetdeconvolutionof Ricinuscasestudy data 90

Integratedlog display (ILD) 93

Lanaway Field CaseStudy 97

Carbonatereef developmentin the WesternCanadian 97

Well nomenclature 102

Lanaway Field 103

VSP data acquisition 105

VSP data processing 107

- P-wave separation 107

- VSP deconvolution 109

- Inside and outside corridor stacks 111

RicinusField CaseStudy 119

Ricinus full Leduc reef 120

VSP data acquisition 124

Near-offset(199m) VSP interpretiveprocessing 125

- P-wave separation 125

- VSP aleconvolution 127

- Inside and outside corridor stacks 128

Far-offset(1100m) VSP interpretiveprocessing 130

- Time-variant model-based rotation 133

- Deconvolution of far-offset data 134

Fort St. JohnGrabenCaseStudy 141

- Tectonicand depositionalhistoryof Peace 144

Original interpretationand well results 148

VSP data acquisition 149

Near-offset(149 m) VSP interpretiveprocessing 151

- P-wave separation 151

- VSP deconvolution 153

- Inside and outside corridor stacks 153

Far-offset(FSJG1)VSP interpretiveprocessing 155

- Time-variant model-based rotation 157

- VSP-CDP mapping and migration 159

Far-offset(FSJG2)VSP interpretiveprocessing 161

SimonetteField CaseStudy 167

Simonette low-relief reef 171

VSP data acquisition 172

Near-offset(252 m) VSP interpretiveprocessing 173

- P-wave separation 173

- VSP deconvolution 175

- Inside and outside corridor stacks 177

Far-offsetVSP interpretiveprocessing 179

- Time-variant model-based rotation 181

- VSP-CDPmapping and migration 181

Appendix: Mathematical review 197

Median filter 197

Karhunen-Loeve(K-L) filtering 198

f-k filtering 200

•-p filtering 201