integrating 3-d seismic data, field analogs, and mechanical models

AAPG Bulletin, v. 85, no. 7 (July 2001), pp. 1183–1210 1183

Integrating 3-D seismic data,field analogs, and mechanicalmodels in the analysis ofsegmented normal faults in theWytch Farm oil field, southernEngland, United KingdomSimon A. Kattenhorn and David D. Pollard

ABSTRACT

We propose a methodology for the analysis of normal fault geom-etries in three-dimensional (3-D) seismic data sets to provide in-sights into the evolution of segmented normal fault systems and toimprove recovery efforts in fault-controlled oil fields. Limited seis-mic resolution can obscure subtle fault characteristics such as seg-mentation and gaps in fault continuity that are significant for oilmigration and thus accurate reservoir characterization. Detailedseismic data analyses that incorporate principles of normal faultme-chanics, however, can reveal evidence of fault segmentation. Weintegrate seismic attribute analyses, outcrop analog observations,and numerical models of fault slip and displacement fields to aug-ment the use of 3-D seismic data for fault interpretation. We ap-plied these techniques to the Wytch Farm oil field in southern En-gland, resulting in the recognition of significant lateral and, to alesser extent, vertical segmentation of reservoir-scale faults. Slipmaxima on fault surfaces indicate two unambiguous segment nu-cleation depths, controlled by the lithological heterogeneity of thefaulted section. Faults initiated preferentially in brittle sandstoneand limestone units. Subsequent growth and linkage of segments,predominantly in the lateral direction, resulted in composite faultsurfaces that have long lateral dimensions andmultiple slipmaxima.Reservoir compartmentalization is greatest at the level of prevalentsegment linkages, which corresponds at Wytch Farm with the pre-dominant hydrocarbon-producing unit, the Sherwood Sandstone.At relatively shallower depths, fault segments are younger and lessevolved, resulting in a greater degree of segmentation with intactrelay zones.

Copyright �2001. The American Association of Petroleum Geologists. All rights reserved.

Manuscript received June 28, 1999; revised manuscript received July 26, 2000; final acceptance October10, 2000.

AUTHORS

Simon A. Kattenhorn � Department ofGeological Sciences, University of Idaho,Moscow, Idaho, 83844-3022;[email protected]

Simon A. Kattenhorn received B.Sc. and M.Sc.degrees in geology from the University ofNatal-Durban, South Africa, an M.S. degree ingeology from the University of Akron, and aPh.D. in geological and environmentalsciences from Stanford University. He iscurrently an assistant professor of geologicalsciences at the University of Idaho. Hisresearch interests include applying principlesof rock and fracture mechanics to fracturedevelopment, fault evolution, earthquakes,and planetary tectonics, using field mapping,numerical modeling, seismic reflection data,and seismological data.

David D. Pollard � Rock Fracture Project,Department of Geological and EnvironmentalSciences, Stanford University, Stanford,California, 94305;[email protected]

David D. Pollard received a B.A. degree fromPomona College, a Ph.D. from StanfordUniversity, and a D.I.C. from Imperial College,all in geology. He is a professor in theDepartment of Geological and EnvironmentalSciences at Stanford and is codirector of theRock Fracture Project, an industrial affiliatesprogram. His research interests focus onunderstanding rock fracturing and faultingusing applications to fluid flow inheterogeneous reservoirs using outcrop andsubsurface data, laboratory experiments, andnumerical modeling.

1184 Analysis of Segmented Normal Faults

INTRODUCTION

Recent work on normal fault systems has emphasized the impor-tance of the segmented nature of fault geometries in matters of faultevolution (Peacock and Sanderson, 1994; Trudgill and Cartwright,1994; Cartwright et al., 1995; Dawers and Anders, 1995; Childs etal., 1996; Marchal et al., 1998), sedimentary basin development(Anders and Schlische, 1994; Dawers and Underhill, 2000), geo-thermal fluid migration (Coussement et al., 1994; Martinez, 1998),and seismological behavior (Crone and Haller, 1991; de Polo et al.,1991; Machette et al., 1991; Wells and Coppersmith, 1994). Awidely recognized impact of segmented normal fault systems, andthe focus of this article, is with regard to fault-controlled hydrocar-bon traps in oil fields. Well placement and recovery efforts in manyoil fields have benefited significantly from highly detailed charac-terizations of segmented normal fault systems (Bouvier et al., 1989;Morley et al., 1990; Pegrum and Spencer, 1990; Knipe et al., 1998;Ottesen Ellevset et al., 1998; Maerten, 1999). Breaks in fault con-tinuity provide potential flow zones through which hydrocarbonscan migrate across a faulted region, therefore, a thorough meth-odology for the analysis of segmented fault systems is needed toenhance fault interpretations and thus recognize potential waterbreakthroughs and hydrocarbon escape points within fault-compartmentalized reservoirs.

Although distinct fault segments may be clearly visible at thesurface of the Earth, some large fault systems have evolved to apoint where evidence of initial segmentation has been eradicatedthrough fault segments linking together, allowing the accumulationof large amounts of slip over a resultant composite fault surface(Wesnousky, 1988; Peacock and Sanderson, 1991). Slip profilesalong normal fault traces at the Earth’s surface commonly exhibitheterogeneities in slip distributions (Cartwright and Mansfield,1998;Morley, 1999) that may imply a relict segmented nature. Thisprocess of initial segmentation and subsequent linkage is character-istic of normal fault system evolution (Cartwright et al., 1995;Dawers and Anders, 1995) and may be associated with geometricirregularities along fault strike at the points of linkage (Peacock andSanderson, 1994).

Fault traces at the Earth’s surface provide limited informationon the 2-D evolution of faults where they intersect a horizontalplane but cannot be used to elucidate the three-dimensional (3-D)evolution of the fault system. The 3-D characteristics of normalfaults that can be determined from 3-D seismic reflection data(Childs et al., 1995; Mansfield and Cartwright, 1996; Ottesen El-levset et al., 1998; Yielding et al., 1999; Dawers and Underhill,2000) are crucial for accurate reservoir characterization wherefaults act as barriers to hydrocarbon migration and may thus poten-tially compartmentalize the reservoir. Insights into 3-D fault ge-ometries are also important for formulating mechanical models thatexamine fault geometries, tip-line shapes, slip distributions, faultscaling laws, and mechanical interaction effects within segmented

ACKNOWLEDGEMENTS

Three-dimensional seismic reflection data forthis study were provided by BP ExplorationOperating Company Limited and the WytchFarm partnership companies: Arco British Ltd.,Premier Oil Plc., ONEPM Ltd., Talisman NorthSea Ltd., and Kerr-McGee Oil (UK) Plc. Inter-pretation software was provided to StanfordUniversity and the University of Idaho bySchlumberger GeoQuest. Fault rendering usedFAPS software provided by Badley Earth Sci-ences Ltd. Thanks to Richard Fox at BP forfield assistance in southern England and toGiles Watts at BP for detailed informationabout Wytch Farm oil field. We thank ScottYoung, Laurent Maerten, and Bashir Koledoyefor technical support. Funding for this projectwas provided by BP Exploration OperatingCompany Limited and the Rock Fracture Pro-ject at Stanford University. Simon Kattenhornacknowledges a McGee Grant Award from theSchool of Earth Sciences at Stanford Univer-sity. We are grateful to Bruce Trudgill, NancyeDawers, and Barry McBride for their construc-tive reviews of the manuscript.

Kattenhorn and Pollard 1185

BudleighSalterton

Sidmouth

CharmouthPoole

Weymouth

Vale of Pewsey Fault

Mere Fault

Cranborne Fault ZoneBereRegis Fault

WESSEX BASIN

WEALDBASIN

EnglishChannel

BristolChannel

0 50 km

N

Wessex Basin boundary

Purbeck fault

Isle of

Wight

Wytch Farm

Ladram Bay

DORSET

DEVONLymeRegis

WILTS

SOMERSET

WESTSUSSEX

HAMPSHIRE

SURREY

Figure 1. Map of the Wessex basin in southern England (inset) showing the location of the Wytch Farm oil field (modified afterEvans and Chadwick, 1994). Solid lines represent major faults in the basin.

fault systems. Fault interpretations from 3-D seismicdata, however, are limited by interpretation subjectiv-ity, structural complexity, processing artifacts, seismicresolution, and insufficient use of principles of fracturemechanics that can aid the interpretation. This mayresult in erroneous fault interpretations that incorpo-rate unrealistic fault geometries and overlook impor-tant geometrical features such as segmentation or faultlinkage zones.

This article documents the use of a good-quality3-D seismic data set from the Wytch Farm oil field insouthern England to characterize normal fault styles inthe reservoir. The effect of faulting on transmissibilitywithin the reservoir is evidenced by high-pressure dif-ferentials across faults within the Wytch Farm field(Smith and Hogg, 1997). Such differences in pressureindicate fault-sealing effects, which may be related toclay smearing along the faults (R. Knipe, 1994, per-sonal communication).

This impact on hydrocarbon flow by the faults inthe Wytch Farm oil field necessitates accurate charac-terization of the 3-D fault geometry, including theidentification of potential leakage points along faults inthe reservoir. To that end, our goal is to unravel thefault geometries and fault growth histories in the oilfield through the development of the following gen-erally applicable interpretation methodology. First,emphasis is placed on the use of seismic attribute char-acteristics to develop an initial fault interpretation.Second, we use outcrop-scale analogs of reservoir-scale

faults to describe the nature of fault geometry in crosssection, linkage tendencies, and other deformationcharacteristics pertinent to honing the seismic inter-pretations. Finally, we fine tune the initial interpreta-tion by integrating results of numerical models, whichexamine the relationships between fault geometries,slip distributions, and horizon displacements. Thecom-bination of these tools allows us to accurately constrainfault geometries in 3-D and to develop a hypothesis forfault evolution in the Wytch Farm field that augmentsprevious reservoir characterizations.

TECTONIC SETT ING

The Wessex basin in southern England is a late Paleo-zoic to Tertiary extensional basin approximately80,000 km2 in size (Figure 1). A prolonged period ofPhanerozoic extension resulted in distinct rhomboidaldepocenters confined by two sets of normal faults: east-west trending and northwest-southeast trending. Thebroad-scale tectonic history includes four distinctevents (Hawkes et al., 1998): (1) early Atlantic rifting(Late Permian to Early Triassic), (2) Atlantic riftingand opening (Early to Late Jurassic), (3) Biscay rifting(Early Cretaceous), and (4) the Alpine orogeny (LateCretaceous to Tertiary). Extensional events were ori-ented along approximately north-south axes, whereasthe Alpine compression was southeast to northwest di-rected (Miliorizos and Ruffell, 1998). Much of the


Figure 2. Stratigraphic column of Paleozoic and Mesozoicunits in the Wessex basin. Shaded units are hydrocarbon-producing units at Wytch Farm oil field. Hachured lines repre-sent unconformities. Horizontal bars on the right represent lo-cations of major seismic reflector horizons.

faulting bounding the depositional subbasins may havebeen inherited from reactivated basement faults asso-ciated with the Devonian to Carboniferous Variscanorogeny (Karner et al., 1987; Butler, 1998).

The Wessex basin stratigraphy encompasses a suc-cession of Permian through Eocene sedimentary rocksoverlying middle Paleozoic basement of molasse-typesediments (Underhill and Stoneley, 1998). Basin fill av-erages 1.5 km and has a maximum of 3.5 km (Karneret al., 1987) and consists of a variable accumulation ofsandstones, shales, limestones, and evaporites (Figure2). Depositional environments were closely linked tothe tectonic history of the basin. Basin emergence inthe Late Jurassic resulted in a prolonged period of Cre-taceous chalk deposition during a eustatic sea levelpeak, contemporaneous with a change from active rift-ing to thermal subsidence (Hawkes et al., 1998). Wes-sex basin sedimentation terminated during the Alpineorogeny, which culminated in the late Oligocene (Kar-ner et al., 1987), and resulted in internal basin defor-mation and inversion tectonics along normal faultssouth of the Wytch Farm oil field (Colter and Havard,1981; Stoneley, 1982; Selley and Stoneley, 1987; Un-derhill and Paterson, 1998).

WYTCH FARM FIELD

The Wytch Farm field is predominantly defined by a2.5 km–wide major horst block, internally dissected bynumerous east-west–trending conjugate normal faults(Figure 3). This field is the largest onshore oil field inwestern Europe, having reserves in excess of 428 mil-lion bbl (Underhill and Stoneley, 1998). The field isbounded to the south by the Wytch Farm fault, whichdips south, displacing down to the south, having offsetsof 100–300 m. The Northern Bounding fault dips tothe north, displacing down to the north by approxi-mately 50 m. The major faults within the interveninghorst block (Figure 3) are the Arne fault and theNorth-ern fault (NF) (Smith and Hogg, 1997).

Stratigraphic offsets indicate that faulting ceasedin the Early Cretaceous (Colter and Havard, 1981; Un-derhill and Stoneley, 1998), and postrift sedimentation


N S

Northern Bounding Fault

Arne Fault

Northern Fault

Wytch Farm Fault

? ??

northern boundaryof Wytch Farm oil field southern boundary

of Wytch Farm oil field

Sherwood Sandsto

ne

dip closures control easternand western boundaries of

Wytch Farm oil field

Figure 3. Schematic illustra-tion of the Wytch Farm oil field,defined by an internally dis-sected 2.5 km–wide horst. Theoil field is bounded to thesouth by the Wytch Farm faultand to the north by the North-ern Bounding fault. East andwest boundaries of the oil fieldare controlled by dip closuresof the oil-water contact.

continued into the Late Cretaceous. Stratigraphicthickness variations across Wessex basin faults indicateseveral episodes of synrift deposition between the Per-mian and Early Cretaceous (Stoneley, 1982; Chadwicket al., 1983; Selley and Stoneley, 1987; Jenkyns andSenior, 1991; Hawkes et al., 1998). In theWytch Farmfield, such syntectonic stratigraphic thickness varia-tions are most evident in the Upper Jurassic and LowerCretaceous strata. Inversion tectonics associated withthe Alpine orogeny reactivated several formerly exten-sional faults south of the Wytch Farm oil field (e.g.,Purbeck–Isle of Wight fault system) (Underhill andPaterson, 1998) but did not induce reverse motion onany of the normal faults at Wytch Farm (Underhill andStoneley, 1998).

Three formations currently produce hydrocarbonsat Wytch Farm (Figure 2) (Smith and Hogg, 1997;McKie et al., 1998): the Sherwood Sandstone (�1585m depth), the Bridport Sands (�925 m depth), andthe Frome Clay Limestone (�800 m depth). Sourcerocks for hydrocarbons are thought to be the Blue LiasMudstone (Ebukanson and Kinghorn, 1986; Selleyand Stoneley, 1987) in downdropped fault blockssouth of the Purbeck–Isle of Wight fault system,where structural burial depths were sufficient forsource rocks to reach thermal maturity for hydrocar-bon generation by the Late Cretaceous (Colter andHavard, 1981; Bowman et al., 1993; Hawkes et al.,1998). Hydrocarbons then migrated into the highly

faulted Wytch Farm region, accumulating within themain reservoir units, compartmentalized by a networkof normal faults. Hydrocarbon accumulation precededTertiary uplift associated with the Alpine collisionevent, during which time inverted normal faults mayhave begun acting as seals (Selley and Stoneley, 1987;Underhill and Stoneley, 1998).

Hydrocarbon traps in the Wytch Farm oil field arecontrolled by reservoir facies and dip closures to theeast and west and have sealing faults constraining thenorth and south extents of the reservoir (Figure 3)(Dranfield et al., 1987; Smith and Hogg, 1997); hy-drocarbon traps also influence hydrocarbon migrationacross the oil field. Approximately 100 wells have beendrilled at Wytch Farm, the trajectories of which werespecifically designed to avoid fault zones so as to reducewater breakthrough and drilling losses. Detailed faultcharacterization is thus a major initiative at WytchFarm to isolate potential fault-controlled traps and op-timize future well trajectories.

THREE-DIMENSIONAL SEISMIC SURVEY

To facilitate accurate characterization of the oil field,an approximately 66 km2 3-D seismic data set was ac-quired in 1994 by the Wytch Farm partnership com-panies. Coverage is predominantly onshore at the en-trance to Poole Harbour (Figure 1), extending into the


Figure 4. Normal fault appearance in the Wytch Farm 3-D seismic data. Faults appear approximately planar and are sharply resolved.The section is shown in two-way traveltime (TWTT) with approximate depth conversions (in meters) indicated in parentheses (using2800 m/s). Note that Upper Jurassic units are missing below the Lower Cretaceous unconformity in the main horst block north ofthe Wytch Farm fault.

harbor for a distance of about 3 km (30% of the dataset width). Normal fault locations are easily discerniblein the seismic data as a result of the shallow dips ofsedimentary layering, which result in clear offsets ofseismic reflectors (Figure 4). Most major normal faultsextend deeper than the maximum depth of good seis-mic resolution (�1250 ms two-way traveltime, orabout 1750 m depth), preventing fault characteriza-tion below the deepest resolvable unit in the field(Sherwood Sandstone). Resolution of all overlyingunits is excellent, with the exception of the on-landpart of the data set beneath Poole Harbour, where seis-

mic quality is greatly reduced. Seismic horizon iden-tification and correlation to the local stratigraphy wascarried out by geologists at Wytch Farm (G. Watts,1997, personal communication), using ties to wellsynthetics.

Fault interpretations were made using an uncon-verted two-way traveltime (TWTT) data set. Thedepth conversion algorithm developed for the regionspecifies a linear relationship between depth andTWTT plus correction factors for easting and northing.An approximate TWTT conversion is 1.4 m/ms(equivalent to an average velocity of 2800 ms–1).


Horizon interpretations were made on strong re-flectors throughout the faulted sedimentary section,facilitated by an autopicking tool. To ensure accuratematching of horizons across faults, interpretationswere confirmed using closed loops that circumscribedinterpreted faults at each horizon level (i.e., tracingout an unbroken horizon from footwall to hanging walland back to footwall without crossing the fault plane).This enabled accurate matching of horizons acrossfaults and the determination of reliable fault slipdistributions.

Faults were interpreted on every fifth inline (i.e.,every 62.5 m). Inlines strike approximately perpendic-ular to the east-west trend of fault strikes, allowing ac-curate determinations of true, rather than apparent,fault offsets. In addition, random seismic traverseswere used to constrain fault locations in three dimen-sions and to assess the occurrence of fault segmentationand linkage locations.

FAULT INTERPRETATIONS

Normal faults were investigated in the major horstblock between the northern and southern boundingfaults (Figure 3). The geometries of fault traces inmap view were obtained using hanging-wall and foot-wall cutoffs on a succession of seismic horizon mapsfrom the Sherwood Sandstone up through the faultedsection (Figure 5). The utility of such maps is thatthey can be compared and contrasted with maps offault traces documented from normal fault environ-ments at the Earth’s surface and they demonstratevariability in depth-dependent map patterns and faultgeometries in a simple and comprehensible manner.In the deepest horizons, faults appear continuouswith staggered or sinuous traces oriented along anoverall west-southwest–east-northeast or west-north-west–east-southeast strike trend. With decreasingdepth, faults become increasingly segmented andhave laterally stepping individual fault traces consis-tently aligned east-west.

The stratigraphy is compartmentalized into nu-merous horst and graben blocks (Figure 5) that inter-weave along fault strike. As slip decreases toward thelateral tips of graben-bounding faults, slip increases onadjacent horst-bounding faults, and the major struc-tural style transfers along strike from graben to horst.The amount of overlap of grabens and horsts may beas much as one-third of the individual fault lengths(Figure 5).

Fault interpretation and reservoir characterizationcan be augmented using 3-D fault rendering (Figure6), which illustrates 3-D variations in fault geometryand tip-line shapes. Fault surfaces at Wytch Farm ex-hibit flat tops and bottoms at the upper and lower ex-tents of the fault system, respectively. The flat bottomsare an artifact of the depth limitations of good seismicresolution. The flat tops are real and result from thefaults having approached or pierced the Late Jurassicto Early Cretaceous paleosurface associated with syn-tectonic sedimentation before subsequent burial bypostrift marine deposits. Also, some degree of erosionmay have removed parts of the upper tips.

One of the largest faults at Wytch Farm is thenorth-dipping NF/Horst East fault (HEF) system (Fig-ures 5, 6) that strikes across the entire survey area.Above the Lower Jurassic section (Bridport Sands),the NF and HEF appear in fault maps as separate en-tities striking approximately east-west (Figure 5) thatoverlap by about 20% of the mappable lengths andhave a spacing of about 10%.Mapped out in 3-D, how-ever, these faults can be linked to a common pre–Lower Jurassic basement fault system that strikes west-southwest–east-northeast in the western part of thedata set and east-west in the eastern part (SherwoodSandstone level map in Figure 5). Both the pre– andpost–Lower Jurassic parts of the fault system exhibitsignificant lateral segmentation and have only minoroverlaps between identifiable segments, perhaps im-plying that the segments are hard linked by smallerscale faults.

In the western part of the survey area, where theNF/HEF basement fault begins curving toward thenortheast, a south-dipping fault (North Graben fault[NGF]) to the north of NF apparently inhibited theNF from following the basement fault trend in thepost–Lower Jurassic section. The NF thus continuedstriking east-west where the basement part steps to thenorth (Figures 5, 6). In response, a second post–LowerJurassic system of faults (HEF) formed above the base-ment fault in the eastern area of the data set. Theserelationships suggest contemporaneous growth of theNGF and the NF/HEF system and indicates that con-jugate fault pairs impact greatly on each other’s devel-opment. The fault styles also indicate that faults in theshallower parts of the stratigraphy are spatially con-trolled by the locations and orientations of deeperfaults.

At the scale of seismic resolution, fault tip-lineshapes are variable (Figure 6) and are highly dependenton the presence and shape of adjacent fault segments


Figure 5. Fault trace maps for various horizons through the faulted section (Figure 4). Ball and stick symbols indicate downthrownhanging-wall blocks. Approximate depths are shown in the bottom left corner of each box. As depth decreases, there is an increasingdegree of segmentation and a tendency toward east-west fault orientations. AF � Arne fault; NF � Northern fault; NFB � Northfault basement; NGB � North Graben basement fault; NGF � North Graben fault; HEF � Horst East fault; HEU � Horst EastUpper; EGN � East Graben North; EGN2 � East Graben North 2; FA � Fault A; FB � Fault B; FC � Fault C; FD � Fault D.


Figure 6. 3-D rendering of fault geometries using fault segments interpreted from seismic data. Fault names are as in Figure 5.Lines on fault surfaces indicate respective dip directions. Flat bottoms of faults represent the maximum depth of seismic resolution,whereas flat tops are real and indicate that the faults pierced or approached the paleosurface. (A) Oblique view from above andtoward the southeast. (B) Oblique view from above and toward the northwest.

(e.g., Nicol et al., 1996). Most fault shapes can be ap-proximated by rectangular, elliptical, or semiellipticaltip lines. In general, faults have high aspect ratios (ratioof fault length to fault height) in the range of 2–4.Many faults exhibit steep tip lines along their lateraledges (Figure 6), generally at relay zones between ad-jacent fault segments. In addition to the flat tip lines atthe upper extent of the fault system, horizontal tiplines occur at vertical steps between fault segments(Figure 6) or where the antithetic fault of a conjugatefault pair approaches the through-going fault that cutsacross the entire faulted stratigraphy.

Although some faults are apparently continuousthroughout the pre-Cretaceous stratigraphic section,variations occur along strike where fault surfaces bi-furcate into separate segments (e.g., HEF and HEU inFigure 6). Such segments are separated from eachother both vertically and perpendicular to fault strike.In this manner, faults across a wide region collectivelyform a systematic patchwork of faults, which is similarto a concept described byWillemse and Pollard (2000)for the development of a single fault surface throughthe linkage of multiple segments. The patchwork ge-ometry is most compelling where viewed from directly


Figure 7. Systematic patch-work geometry defined by faultsurfaces within a faulted vol-ume, each fault surface filling ina gap in the patchwork. Viewdirection is from above. view from above

above, each fault filling a gap in the patchwork (Figure7). Of particular significance is that the pieces of thepatchwork at various levels of the stratigraphy do notnecessarily consist of faults that dip in the same direc-tion but may be conjugate to each other.

F IELD ANALOGS

Fault styles were examined in outcrop analogs of theWytch Farm units to provide potential insights intoseismic-scale fault architectures in the event of somedegree of scale independence and to provide con-straints on boundary conditions (e.g., slip vectors onfaults) in numerical models (see next section). Allfaults in the Wytch Farm field are constrained belowthe Aptian–Albian unconformity below the Creta-ceous Chalk and thus only occur in outcrop where theunits below this unconformity intersect the Earth’s sur-face. Along the southern coast of England, the regionaldips of the sedimentary units result in successivelyolder units cropping out with increasing distance westof Wytch Farm. Excellent exposures of the WytchFarm units occur along the high sea cliffs of westernDorset and eastern Devon. Outcrop-scale field analogsof Wytch Farm normal faults can be observed at leastas far west as Exmouth (Figure 1), approximately 100km west of Wytch Farm. These faults have similar ori-entations to those at Wytch Farm and developed dur-ing the same succession of regional tectonic events.They are thus potential analogs of the reservoir-scalefaults and may demonstrate the characteristics of fault-ing within the various sedimentary units at WytchFarm in terms of both broad-scale fault geometries andinternal fault architectures (R. Fox, 1997, personalcommunication).

A Sherwood Sandstone equivalent (Otter Sand-stone) at Ladram Bay, west of Sidmouth (Figure 1),provides numerous examples of conjugate fault styleson an approximately 20 m–high cliff face (about 10%

of the total thickness of the Sherwood Sandstone atWytch Farm) (Figure 8A). Fault zones are narrow(�15 m wide) and about 100 m apart, and no visibledeformation exists in the regions between the faultzones.

The outcrop exhibits a conjugate fault style, whichbears a strong resemblance to fault geometries at the3-D seismic scale. A southwest-dipping normal faultcuts through the entire outcrop and offsets horizontallayering by about 5 m. Minor northeast-dipping anti-thetic faults (�1 m offset), occur in both the hanging-wall and footwall blocks. The faults in Figure 8A arenarrow (�2 cm wide) deformation band-style faults(Aydin and Johnson, 1978; Antonellini et al., 1994)that strike approximately northwest-southeast, havedips of about 65�, and exhibit slip surfaces indicatingpure dip-slip motion. Although deformation is almostabsent outside of the region of faulting, the sandstonein the footwall juncture zone between conjugate faultscontains a network of conjugate deformation bands(Figure 8B). Thus, there is a hierarchy of deformationscales at the outcrop (subseismic) scale of faulting.

Along the main fault plane (Figure 8A), a thinlayer of gouge has developed from either finely com-minuted sandstone or entrained clay from shale hori-zons in the Sherwood Sandstone. East of Ladram Bay,near Sidmouth, a fault outcrop provides clear evidenceof shale being smeared into the fault zone from a shaleunit in the Sherwood Sandstone (Figure 9A). A similarexample occurs 50 m east of the mouth of the RiverSid, where a shale unit is associated with several ver-tically stepping echelon segments along a small normalfault (Figure 9B). This geometry provides a logicalstarting point for shale to become entrained into thefault zone as slip accumulates (Aydin and Eyal, 1996;Childs et al., 1996): slip is accommodated in segmentstepovers at shale horizons by smearing of the shaleinto the fault zone. In this way, the vertically steppingsegments do not have to be physically linked to behaveas a kinematically continuous fault.


Figure 8. (A) Normal faults in Otter (Sherwood) Sandstone atLadram Bay, Devon. Cliff section is approximately 20 m high.The conjugate style of faulting is similar to that observed inseismic data. (B) In the footwall block beneath the conjugatefaults (box in A), the sandstone exhibits a dense network ofconjugate deformation bands (thin lines).

MECHANICAL MODELING

Methodology

Numerical models of normal faults that examine slipdistributions and displacement fields can providemany

insights into fault evolution that can be applied to3-D seismic data where the fault geometries are similarto the models (Pollard et al., 1997; Maerten et al.,2000). This introduces a mechanical evaluation of in-terpreted slip distributions and displacement fields todetermine whether they are consistent with rock frac-ture mechanics principles for a given fault interpreta-tion. We use a boundary element computer programcalled Poly3D (Thomas, 1993) based upon linear elas-ticity in a half-space (i.e., containing a free surface,analogous to the surface of the Earth). Faults are ap-proximated as planar surfaces of displacement discon-tinuity (comprised of 400 boundary elements) that un-dergo a complete stress drop during slip under theinfluence of an effective remote tension perpendicularto fault strike. Such models simulate crustal extensionby normal faulting, with all accumulated shear stressesbeing relieved along the fault during slip. These con-ditions adequately approximate slip episodes on puredip-slip normal faults at Wytch Farm that were onceactive at or very near to the Earth’s surface. Poly3Dcan be used to calculate slip distributions on fault sur-faces and the displacement field in an elastic body con-taining faults of specified spatial geometry and shape.These results can be compared directly with seismicdata interpretations.

As described previously, the normal faults in theWytch Farm oil field are segmented above the scale ofseismic resolution. Fault segments have variablelengths and variable amounts of overlap and spacing inrelay zones. The amount of mechanical interaction be-tween adjacent fault segments is affected by these pa-rameters and is evidenced in the slip distributions anddisplacement fields (Willemse et al., 1996; Willemse,1997). Therefore, we examine slip and displacementfields for the simple cases of single faults, overlappingechelon faults, and conjugate faults. We approximateall faults as semielliptical slip surfaces intersecting theEarth’s surface and having an aspect ratio of 3, whichapproximates some tip-line shapes interpreted atWytch Farm. We thus illustrate the general character-istics of slip and displacements for such fault geome-tries and show how they may be used to improve theseismic interpretation of faults at Wytch Farm.

Model Results

Slip DistributionsSlip characteristics on fault surfaces vary with the geo-metrical arrangement of faults (Figure 10), whetheradjacent fault segments are partially overlapping, fully


Figure 9. (A) Example ofshale smear along a normalfault in a Sherwood Sandstoneoutcrop west of Sidmouth,Devon. A shale unit has beensmeared along the fault plane.The apparent jog in the faulttrace is a perspective view ef-fect. White scale card is 17 cmlong. (B) Smearing of shalewithin a zone of vertical seg-mentation along a normal faultin Sherwood Sandstone, east ofSidmouth.

overlapping, or conjugate to each other. Isolated faultsin an elastic half-space have maximum slip values atthe free surface (Figure 10A), halfway between thefault tips. Slip gradients steepen toward the lower faulttip line (Figure 10B). Faults that are not isolated fromeach other have different slip patterns because of themechanical interaction effect between adjacent faultsegments, resulting from deformation in the elasticbody being partitioned between the fault segments.Partially overlapping fault segments exhibit similar slipdistributions to isolated faults except that slip gradientsare steepest within relay zones between fault segments(Figure 10C) and have increasing prominence as over-lap increases. For fully overlapped faults (Figure 10D),the leading fault (as defined in Figure 10) has a morelocalized slip distribution peak than the trailing fault(situated in the footwall of the leading fault), whichhas a broader peak and steeper slip gradient toward itslower tip. Conjugate faults exhibit particularly steepslip gradients toward the lower tips (Figure 10E),where the oppositely dipping faults are closest to eachother.

Displacement FieldsThe pattern of displacement contours at the free sur-face (Figure 11) indicates the expected dip direction ofbeds deformed by slip on the faults (bed dips are per-pendicular to contour lines). Beds dip toward points ofslip maxima along faults. We illustrate how theamount of overlap affects the orientation of bedswithin a relay zone for faults spaced at 5% of the faultlength. Faults are more prone to mechanical interac-tion as the spacing between them decreases (Burgmannet al., 1994). Bed dip directions are accordingly af-fected; therefore, the overlap-to-spacing ratio becomes

the controlling factor when modeling bed dips. As theamount of overlap increases, bed dip directions rotateto progressively higher angles with respect to faultstrike (Figure 11A, B) until they are perpendicular tofault strike from the leading fault toward the trailingfault, for the case of fully overlapped faults. Models ofdisplacement fields in the vicinity of vertical stepsalong a fault (Figure 11C) can illustrate the effects ofvertical heterogeneities along a fault surface on horizondisplacements.

Comparison to Seismic DataSlip distribution models can be compared with con-toured horizon offsets across interpreted fault surfaces,whereas displacement field models can be comparedwith seismic horizon structure-contour maps (e.g., au-topicked horizons). For example, comparison betweennumerically modeled displacement fields and horizon-contour maps may enable the deduction of locationsof segmentation along fault strike. Because fault spac-ing can be more accurately measured than fault overlapin seismic data, it can be used as an input parameter innumerical models to then deduce the amount of over-lap of segments based on calculated bed dip directions(Figure 11). The magnitudes of bed dips are related tocumulative fault slip magnitudes; however, bed dip di-rections are a function of fault overlap geometry andmay thus be used to aid in fault interpretations in seis-mic data.

Figure 12 shows profiles of the displacement fieldalong a line through the relay zone oriented parallel tofault strike. Profile shapes vary with amount of faultoverlap and can thus be compared to profiles acrossseismic horizon contour maps along fault-parallel tra-verses to determine the amount of overlap between


Figure 10. Modeled slip distributions on semielliptical faults piercing the surface of the Earth. Slip distributions are noticeablyimpacted by fault configuration, as are slip gradients between the surface and the lower fault tips (B). Slip maxima (black dots) occurat the surface approximately halfway between the fault tips (A, D, E), except echelon faults (C), which have slip maxima skewedtoward relay zones. Fault spacings with respect to fault lengths (measured at the Earth’s surface) are as follows: (C) echelon faults� 5%; (D) fully overlapped faults � 16.7%; (E) conjugate faults � 33.3%. For echelon and fully overlapped configurations, thetrailing fault is situated on the footwall side of the leading fault. Conjugate faults meet at the lower tips and have mutually identicalslip distributions.

fault segments. Estimates can also be made of fault tip-line locations as projected onto the traverse line (Figure12). As overlap increases, the point of zero vertical dis-

placement within the relay zone (intersection pointwith the horizontal axis) moves progressively towardthe leading fault.


Figure 11. (A) Effect of fault configuration on displacement fields at the Earth’s surface. Contoured displacements are normalizedto the maximum downthrow in each example. Arrows between faults represent the direction of bed dip in the relay zone. Faults(thick lines) are all semielliptical and dip in the direction of the ball and stick symbols. Axes are in km; all faults are 3 km long and1 km high. Faults have spacings of 5% of the fault length and variable overlaps as labeled. Fully overlapped faults have a 16.7%spacing. (B) Plot of obliquity between fault strike and bed dip direction as a function of overlap. (C) Displacement field through thecenter of a vertical compressive step.


Figure 12. Profiles of vertical displacements along observation lines parallel to fault strike. Values are normalized to the maximumdownthrow. Positions of fault tips are projected onto the profile lines. Horizontal distances are in kilometers away from the midpointof the relay zone. (A) Profiles at 3.3, 10, and 16.7% of fault length away from an isolated fault as indicated in the inset figure. Profilesmidway across the relay zone for (B) �10% overlap , (C) 0% overlap, (D) 10% overlap, (E) 20% overlap, and (F) 50% overlap (faultspacing � 5%).

The slip distribution and displacement field mod-els described previously represent single-slip eventson faults that pierce the Earth’s surface. No assump-tions are made with regard to evolutionary history for

each fault configuration or the cumulative impact onthe final slip distributions, both of which may behighly complex in natural fault environments. Thenumerical models thus greatly idealize the complex


faulting process. Nonetheless, several normal faultstudies suggest that the distribution of cumulativeslip is somewhat systematic in nature over a range ofscales (Walsh and Watterson, 1988, 1989; Dawers etal., 1993) and bears a strong resemblance to single-slip event elastic model results (Willemse, 1997) de-spite the fact that natural faults accumulate slipthrough multiple slip events. Single-slip event nu-merical models may thus provide reliable analogs fordescribing slip distributions in nature. For example,comparisons between numerical models and seismichorizon contour maps have been successfully used topredict regions of subseismic deformation and faultlinkage points in a North Sea oil field (Maerten etal., 2000).

REFINING FAULT INTERPRETATIONS

First-order fault interpretations from 3-D seismic data,such as those described previously for Wytch Farm oilfield, can be improved through the use of seismic at-tribute tools in conjunction with fault mechanics prin-ciples and numerical model results. In addition, inte-grating detailed seismic data analyses with the salientfield analog observations (i.e., conjugate fault styles,interfault deformation characteristics, shale smearacross clay horizons) allows us to refine the interpretednormal fault characteristics and thus improve the res-ervoir characterization. We place emphasis on faultsegmentation, fault linkage, slip distributions, and tip-line locations.

Segmentation and Linkage

Sinuous fault traces mapped at the Sherwood Sand-stone level appear to be continuous along strike (Figure5). This interpretation, however, uses possibly erro-neous assumptions as to whether faults should be cor-related from one inline to the next. Whether or not theinterpretation is satisfactory is subjective. For example,if the feature of interest is the broad-scale mechanicalbehavior of the entire fault system (i.e., the cumulativebehavior of individual segments, either linked or be-having in a kinematically coherent fashion), the con-tinuous trace interpretation may be appropriate. If theparameter of interest is specific to fault architecture,however, such as the permeability across a fault, theexistence and nature of segments and relay zones maybe of particular importance (e.g., for determining wa-ter breakthrough or hydrocarbon leakage points along

a fault in an oil field), thus requiring a more finelyhoned interpretation.

The variation in fault geometries through thestratigraphic section (Figure 5) illustrates fault segmen-tation at a scale greater than the seismic resolution. Forexample, numerous east-west striking fault segmentsoccur at the level of the Bridport Sands, mostly be-tween 1 and 3 km long. At the deeper level of theSherwood Sandstone, the segmentation is lost in favorof sinuous faults (e.g., NF/HEF), suggesting that link-age of echelon segments may have occurred at depth.In light of interpreted segmentation at shallower levelsin the stratigraphy (Figure 5), we reexamine our initialfault interpretations at deeper levels in an attempt toisolate potential hydrocarbon leakage points using evi-dence of relict segmentation that may provide insightsinto the genesis and evolution of the deeper levelfaults.

To capture evidence for segmentation near to andperhaps below the limits of seismic resolution, use wasmade of dip magnitude maps and time slices. Dipmag-nitude maps reflect the maximum change in dip (inms/m) along a seismic reflector, computed as the dif-ference in TWTT between a particular peak or troughalong a seismic trace and all traces immediately sur-rounding it. Time slices are contoured maps of seismicamplitudes at a particular TWTT (depth). Both typesof display are very sensitive to sudden dip changes thatmay occur at a fault discontinuity or in the relay zonebetween overlapping fault segments.

A dip magnitude map of the D Anhydrite horizonof the Mercia Mudstone in the vicinity of the Arnefault (Figure 13A) corresponds in the fault trace mapsto a region of sudden direction change along the pre-dominantly west-northwest–east-southeast fault trend(Figure 5). The dip magnitude map (�13 m pixel res-olution) shows several east-west–oriented lineaments100–200 m long in a predominantly right-steppingechelon arrangement. This pattern provides an impe-tus to consider the Arne fault as being comprised ofseveral segments that may or may not be linked andprovides clues to the evolutionary history of the fault-ing in general.

A reexamination of the seismic data at an apparentsegment boundary in the dip magnitude map (i.e., afault step larger than the pixel resolution) suggests thepresence of a combined vertical and lateral step in theArne fault (Figure 13B). Seismic traverses in severalorientations through the stepover region indicate thatthe segments are linked across the relay from the trail-ing segment to the footwall of the leading segment via


Figure 13. (A) Dip magnitude map of the D Anhydrite horizon. Faults form lineaments of high dip magnitude and show evidenceof being comprised of numerous segments. (B) Initial and refined interpretations of the Arne fault in seismic section at the fault stepindicated in (A). The new interpretation introduces a vertical step in the Arne fault. (C) Conceptual illustration of segment linkagethrough an upper ramp breach. (D) 3-D representation of the stepover zone, indicating a possible window across the fault.

an upper ramp breach (using terminology of Crider[1998]) (Figure 13C). This geometry provides a pre-viously overlooked potential hydrocarbon leakagepoint along the Arne fault by providing a windowthrough the fault from footwall to hanging wall (Figure13D). The identification of such windows are impor-tant for delineating boundaries to fault-controlled hy-drocarbon traps (i.e., determining which faults effec-tively compartmentalize the oil field).

Abrupt variations in the general orientation of thepre–Lower Jurassic part of the NF/HEF system (Fig-

ures 5, 13A) may similarly signify segment linkage fea-tures. A seismic time slice at a 140 m–wide jog be-tween two east-west fault segments displays a distinctnortheast-southwest lineament (Figure 14) that sug-gests that the slightly overlapping segments werelinked from the tip of the leading segment to the hang-ing wall of the trailing segment, forming a lower rampbreach and producing an abandoned footwall splay(e.g., Trudgill and Cartwright, 1994). Time slices maythus be used to identify seismically resolvable faultlinkage sites, enabling an evaluation of the sealing


Figure 14. Interpretation of a fault linkage structure usingtime slice attributes. (A) Time slice for a region of the NF/HEFsystem (inset map). Closely spaced amplitude contours indicatefault locations. (B) Interpreted linkage geometry having a lowerramp breach and an abandoned footwall splay.

potential of the fault system and providing insights intothe fault growth history.

In addition to lateral segmentation atWytch Farm,3-D fault surface rendering (Figure 6), outcrop faultstyles (Figures 8, 9), and seismic attribute characteris-tics (Figure 13B) indicate the presence of vertical seg-mentation. Outcrop-scale analogs include areas wherefault continuity is broken by (1) a through-going faultof opposite dip (Figure 8) and (2) clay horizons (Figure9). A corollary of (1) is the large-scale vertical segmen-tation of the NF/HEF system associated with the op-positely dipping North Graben fault (Figure 6). To ex-amine the significance of observation (2) forseismic-scale faults, we reexamined our initial inter-pretations of faults cutting through clay-rich units inan attempt to locate evidence of vertical segmentation.We thus focused on the Mercia Mudstone (which alsocontains anhydrite layers), Lias Mudstone, and Fuller’sEarth horizons.

Initial fault interpretations showed irregularities infault geometries across clay-rich units in section view,possibly indicating vertical steps and implying a signifi-cant lithological control on large-scale vertical segmen-tation of normal faults. Childs et al. (1996) documenthow mechanically weak interfaces similarly control

step locations along outcrop-scale normal faults in Cre-taceous Chalk in Yorkshire, England. At Wytch Farm,contractional steps predominate, analogous to the out-crop example in Devon (Figure 9). Abrupt decreasesin shale thickness at the vertical steps suggest that me-chanically weak units accommodate the compressionat the steps through unit attenuation (e.g., Peacock andSanderson, 1992; Kattenhorn, 1994; Childs et al.,1996). In Figure 15A, the Mercia Mudstone is notice-ably attenuated at a vertical step in the HEF. In addi-tion, bed dips change suddenly in the region of thepostulated vertical step, similar to the Arne fault inFigure 13.

Analogous to shale attenuation at compressive ver-tical steps, clay-rich horizons appear to focus the pointsof convergence of all major conjugate faults at WytchFarm (Figure 15B). The fault trace patterns in Figure5 indicate that major graben-bounding faults convergeat either the Fuller’s Earth horizon (e.g., NF/Fault A;NF/Fault B) or the Mercia Mudstone horizon (e.g.,NF/North Graben fault; HEF/East Graben Northfault), emphasizing lithological control on fault ge-ometries and indicating that graben fault spacing at anyparticular depth may be a function of distance above aspecific horizon, such as a thick shale unit.

Based on the aforementioned criteria, we reinter-preted faults atWytch Farm to capture both lateral andvertical segmentation effects. For example, a refinedinterpretation of the NF/HEF system (Figure 16) re-vealed a greater degree of segmentation than was ini-tially interpreted (Figure 6).

Slip Distributions

Slip profiles along fault traces at the Earth’s surfacecommonly contain multiple slip maxima along a faultthat formed through the linkage of two or more faultsegments (Cartwright et al., 1995; Dawers andAnders,1995; Cartwright and Mansfield, 1998). Analogously,multiple slip maxima across a fault surface provide evi-dence of segmentation (Mansfield and Cartwright,1996) and linkage (Maerten et al., 2000), indicatingboth lateral and vertical loci of the original segments.We adopted this interpretation rationale to evaluate3-D segmentation at Wytch Farm, where multiple slipmaxima are present on all major faults (Figure 17).

The Arne fault slip maxima are consistently at thelevel of the top of the Sherwood Sandstone (Figure17), indicating that segmentation in faulted Triassicunits was a precursor to laterally continuous faults thatformed through subsequent linkage of the segments.


Figure 15. (A) Vertical seg-mentation along the Horst Eastfault. Compressional verticalsteps occur within Mercia Mud-stone that is attenuated withinthe relay. (B) Convergence ofconjugate faults within attenu-ated Mercia Mudstone.

edge

of d

atas

et

Figure 16. 3-D perspectiveview of segmented NF/HEF sys-tem. The refined interpretationaccounts for numerous in-stances of vertical and lateralsegmentation. Arrows representlines of linkage of two seg-ments. Tip lines are horizontalalong vertical steps betweensegments.

This observation is in agreement with both fault-tracemaps in the Sherwood Sandstone (Figure 5) and seis-mic attribute map analyses (Figure 13). Slip distribu-tions that have multiple maxima may thus be used toisolate potential locations of segment linkage sites,which would occur near the slip minima.

The NF/HEF system displays slip maxima in twoparts of the stratigraphy (Figure 17): Triassic units

(Sherwood Sandstone) and mid-Jurassic units (Brid-port Sands through Cornbrash limestone). This in-dicates two loci of fault nucleation in the WytchFarm field, apparently lithologically controlled by thelocations of brittle limestone and sandstone units.Propagation and mutual approach of faults in theserespective parts of the stratigraphy resulted in verticalsegmentation and partial linkages, consistent with


Figure 17. 3-D slip distributions for several normal faults at Wytch Farm, as labeled. White and red are regions of maximum slip.Black and purple are regions of lowest slip. Note multiple slip maxima in all examples. Horst East fault examples are spatially arrangedwith respect to each other as shown by arrows and fault tip-line outlines.

interpreted vertical steps at the level of the MerciaMudstone (Figure 15A). The twomain segments of theHEF have slip maxima and associated steep slip gra-dients close to the relay zone between them, in agree-ment with numerical predictions of slip gradients nearthe region of overlap between contemporaneous faults(Figure 10).

The North fault and North Graben fault both dis-play localized slip maxima at the level of the Mid Liashorizon (Figure 17). These represent locations whereunderlying Mercia Mudstone has been attenuated, re-sulting in increased apparent offset of faulted units.Analogous lithological contributions to localized slipprofile heterogeneity have been described in outcrop-


scale faults (Muraoka and Kamata, 1983; Kattenhornand McConnell, 1994). The NF example is analogousto the vertical step case in Figure 15A. The North Gra-ben fault example reflects a point of conjugate faultconvergence, as shown in Figure 15B. The spatial as-sociation of local slip maxima and clay-rich unitsshould thus be treated with caution and should not becasually classified as fault segment nucleation locations.

Tip-Line Locations

Seismic resolution constraints inhibit accurate locationof fault tip lines, where fault displacements decreaseto zero. Pickering et al. (1997) suggest the addition ofa length of fault beyond the seismically resolvable tipsusing throw and length scaling relationships. This ap-proach, however, may be inaccurate where slip gradi-ents are variable because of mechanical interaction ef-fects (Figure 10) or the impact of lithology (Muraokaand Kamata, 1983).

We applied our methodology for predicting faultgeometries at fault relay zones (based on displacementfield models) (Figures 11, 12) to the East GrabenNorth (EGN) and East Graben North 2 (EGN2) faults(Figure 5), both of which were active at the surface inthe Late Jurassic. First, we determined the dip direc-tion of beds in the relay zone using dip azimuth mapsand seismic lines intersecting the relay at a range oforientations. The so-determined obliquity betweenbed dip direction and fault strike suggests an overlapof 10–20% (Figure 11). The corresponding displace-ment profile (Figure 12) was then compared to the dis-placement of the Fuller’s Earth seismic horizon withinthe relay zone (Figure 18A). The predicted tip-line lo-cation using this method is located approximately 150m beyond the location deduced using horizon offsetsin seismic section view and corresponds to a dip mag-nitude perturbation in the tip-line vicinity (Figure 18B,C). A connecting segment between the faults is alsoevident, having formed an upper ramp breach (Figure18D).

Slip distributions on the EGN and EGN2 faults(Figure 18E) have maxima in different parts of the stra-tigraphy (Cornbrash and Bridport horizons, respec-tively). The shape of the slip contours on EGN2 ap-pears to reflect the projected shape of the EGN faulttip line in the region of overlap (Figure 18F), perhapsdue to mechanical interaction between the twosegments.

Application of numerical results to seismic inter-pretation may thus aid in constraining tip-line loca-

tions; however, care should be taken to not apply thesemodels indiscriminately. The displacement fields in thevicinity of relay zones (Figures 11, 12) are highly de-pendent on individual fault parameters such as faultshape, relative slip magnitude, slip sense, and slip dis-tribution, in addition to initial bed dips (Gibson et al.,1989; Peacock and Sanderson, 1994). Furthermore,horizon displacements produced by growth foldingabove an active fault should not be confused with relayzone deformation. Model estimates of fault tip-lineshape are thus case specific; however, the overall meth-odology described in this article is generally applicable.

DISCUSSION

A detailed characterization of normal fault architec-tures at the Wytch Farm oil field has been achievedthrough the integration of seismic data interpretation,outcrop-scale analog observations, and numericalmod-els. This methodology has enabled the recognition of asignificant amount of segmentation of the fault systemat Wytch Farm. This finding is particularly importantbecause segmentation affects both water breakthroughand sweep efficiency between wells (G. Watts, 2000,personal communication). Exact knowledge of seg-mentation characteristics is thus important for design-ing optimal well trajectories in the oil field. Breaksbetween fault segments provide potential points ofcommunication between otherwise compartmental-ized oil traps; therefore, the recognition of such breakscan provide petroleum geologists with an indication ofthe sealing effectiveness along a particular fault system,as well as isolating particular locations along faultswhere detailed mapping is necessary to determine evi-dence of fault segment linkages.

For example, the Arne fault in the Wytch Farmoil field is laterally continuous in initial interpretations(Figure 5) but shows numerous east-west linear pat-terns along the west-northwest–east-southeast faulttrend in the dip magnitude map (Figure 13) and mul-tiple slip maxima (Figure 17). The Arne fault maythus have been initially segmented, having the indi-vidual segments oriented east-west in a predominantlyright-stepping echelon arrangement. Linkages be-tween these segments then occurred through the de-velopment of northwest-southeast–oriented connect-ing faults, the existence of which has been suggestedby production-related data as well as previous analysesof the Arne fault (P. Kelly, 1997, personal commu-nication) that influenced subsequent decisions about


Figure 18. (A) Estimation of tip-line location using a displacement profile through the relay zone between overlapping faults (seeFigure 12). (B) The fault tip is predicted to occur 150 m beyond the initial interpreted tip location on the trailing fault. (C) Dipmagnitude anomalies near the fault tips suggest continuation of faults beyond the initial interpretation. (D) Reinterpreted faults withnew tip locations and addition of a linking fault segment forming an upper ramp breach. (E) Slip distributions on EGN and EGN2(oblique view toward the NW). Slip maxima (colors as in Figure 17) are in the Cornbrash (EGN) and Bridport (EGN2). (F) The contoursof slip on EGN2 trace out the tip-line shape of EGN across the relay zone.


well trajectories in the oil field (G. Watts, 2000, per-sonal communication).

Analogously, numerous jogs along the trace of theHEF imply an initially left-stepping arrangement ofeast-west fault segments having an overall west-south-west–east-northeast trend and subsequent northeast-southwest–oriented connecting faults (Figure 14).Staggered fault traces in 3-D seismic interpretationsmay thus result from closely spaced echelon segmentsthat were erroneously correlated, or perhaps a relictsegmented geometry that, through segment linkage,produced a through-going fault oriented along thetrend of the array of initial echelon segments. This sce-nario is consistent with both the east-west orientationsof smaller faults higher in the stratigraphy (Figure 5)and the postulated north-south major extension axis inthe region during the period of Atlantic opening in theJurassic and Cretaceous (Hawkes et al., 1998; Milior-izos and Ruffell, 1998).

Outcrop analogs demonstrate vertical segmenta-tion of faults in the vicinity of clay beds (Figure 9B),the significance of which was overlooked in initial seis-mic interpretations. At the seismic scale in the WytchFarm oil field, this outcrop characteristic is manifestedas vertical segmentation of large-scale faults withinclay-rich units such as the Mercia Mudstone, LiasMudstone, and Fuller’s Earth. Vertical segmentation isdifficult to capture in seismic sections in cases wherefault separation is small. If a through-going fault inter-pretation requires the insertion of a bend in the faultplane to match up the fault-induced breaks in horizoncontinuity from one part of the stratigraphy to another,then a segmented interpretation may be more appro-priate (for exceptions, see Peacock and Sanderson,1992; Childs et al., 1996). Beds have increased dips invertical step relay zones in seismic section (Figures13B, 15A) and in mechanical models of displacementfields around vertical fault steps (Figure 11C). Beddips can thus be used as indicators of the lateral extentof linkage between vertical segments. For example,vertically confined bed dip increases along a fault trace-able parallel with fault strike for a significant lateraldistance in time slice maps, are a possible indicatorof relay ramps between vertically stepping segments(Figure 19).

It may be difficult to determine the lateral extentof linkage of vertically stepping fault segments. In someinstances, fault-perpendicular spacing of verticallystepping segments increases with increasing distancefrom the point of linkage, providing unambiguous seg-mentation indicators in seismic section view (Figure

15A). Where segments are closely spaced, the point oflinkage across a vertical step may be approximated asthe location in the 3-D seismic data where the dippinghorizons in the relay are no longer resolvable in seismiclines or seismic attribute maps such as time slices (Fig-ure 19).

The consistency of vertical steps occurring in low-competency units implies a significant lithological con-trol on fault evolution, in agreement with outcrop-scale structures (Muraoka and Kamata, 1983; Childset al., 1996). Softer units respond to compressionalsteps by attenuating, resulting in thickness variationsacross faults that could erroneously be interpreted asbeing a result of syntectonic sedimentation. Slip dis-tributions further indicate the importance of lithologyon fault evolution. Fault segments initiated predomi-nantly at two levels in the stratigraphy (Figure 17): anearly nucleation event within or below the SherwoodSandstone and a later event in the shallower Bridportto Cornbrash part of the stratigraphy. Both faultingepisodes can be reconciled with the documented tec-tonic history of the region (Jenkyns and Senior, 1991;Hawkes et al., 1998). The deeper faults were mostlikely Permian early Atlantic rifting-related normalfaults that were reactivated during Early Jurassic At-lantic rifting. This conclusion is corroborated by thefact that synrift Permian units, such as the AylesbeareMudstone, show significant thickness variations acrossthe deeper faults, whereas postrift units, such as theSherwood Sandstone and Mercia Mudstone, do not(Butler, 1998). The shallower faults probably formedlater, during Late Jurassic Atlantic opening, nucleatingin Middle Jurassic postrift units that postdated theearly Atlantic rifting. These faults subsequently ap-proached or pierced the Late Jurassic paleosurface, re-sulting in thickness variations across faults in Late Ju-rassic (post-Cornbrash) units, whereas there are nothickness variations in the underlying Middle Jurassic(Bridport to Cornbrash) units.

Each of the two faulting episodes produced arraysof laterally stepping fault segments. Lateral propaga-tion and subsequent linkages between these segmentsresulted in composite faults longer than they are tall(aspect ratio� 1). Where linkages formed between thedeeper and shallower fault arrays, the resultant faultsare continuous through the entire stratigraphy, reduc-ing fault aspect ratios.

The evolution of normal faults at Wytch Farmthrough lateral and vertical segmentation and subse-quent linkage is conceptualized in Figure 20. Individualfault segments that are widely spaced from other faults


may be more likely to propagate upward toward thesurface than to propagate laterally, in response to thedecreasing lithostatic confining stress (Kattenhorn andPollard, 1999). This behavior preserves the segmentedgeometry as faults propagate upward through thestratigraphic section and is reflected in the fault-tracemaps at higher stratigraphic levels (Figure 5). Whereupward growth of the Wytch Farm faults was impededby ductile shale units, lateral propagation predomi-nated, probably facilitated by mechanical interactioneffects between laterally stepping faults, which havebeen shown to increase both lateral tip-line growthtendency (Kattenhorn and Pollard, 1999; Willemseand Pollard, 2000) and the likelihood of segment link-age (Trudgill and Cartwright, 1994; Cartwright et al.,1995; Crider and Pollard, 1998).

The staggered fault trends in the Triassic andLower Jurassic rocks imply a geometrically more ad-

vanced state of fault evolution produced by the linkageof east-west segments that formed at depth during theEarly Jurassic. Late Jurassic fault segments were syn-sedimentary faults at or near the surface. The youngerage of fault development is reflected in both the ab-sence of large-scale linkage geometries and the lowerspatial density of faulting, which reflects a shorter andlesser strain history.

Recognition of extensive fault segmentation atWytch Farm is important for evaluating hydrocarbonflow paths in the oil field. Unlinked fault segments pro-vide fluid migration pathways through the relay zonesconnecting hanging-wall and footwall blocks. Linkagezones may nonetheless provide leakage points throughcomposite fault surfaces. Fault segments in the Triassicrocks, such as the Sherwood Sandstone, are apparentlylinked across relays, resulting in laterally continuousfault geometries. Individual fault blocks are thus

Figure 19. Method for con-straining lateral extent of link-age between vertically steppingfault segments. (A) Sequence ofseismic lines (62.5 m apart) ina region where a vertical relayramp is eradicated through seg-ment linkage. Where segmentsare unlinked, a region of in-creased bed dip occurs be-tween the segments. (B) Timeslice through the stepover re-gion (horizontal line in A). Dipcharacteristics of beds are sug-gested by amplitude contourpatterns, which show an abruptchange where linkage is in-ferred. Locations of seismiclines in A are indicated by re-spective arrows.


Sherwood S

andstone

(A)

(B) linkage acrossvertical step

linkage acrosslateral step

faultscarp

abandonedhangingwall

splay

vertical step

lateralstep

echelon faultsegments

Triassicunits

LowerJurassic

units

Early Jurassicsurface

Mercia

Mudstone

Late Jurassicsurface

Figure 20. Conceptual evolu-tion of normal faults at WytchFarm. (A) Formation of east-west striking echelon segmentsin the Triassic section, possiblyabove reactivated basementstructures. A younger system ofsegmented faults subsequentlynucleated in the Middle Jurassicsection, forming vertical stepswith deeper segments acrossductile shale units. (B) Deepersegments linked together acrosslateral steps to form continuousfault traces that have curves orkinks at linkage points. Somelinkages occurred across verticalsteps. The shallower, less-evolved faults are segmentedand possibly pierced the LateJurassic paleosurface.


compartmentalized, in agreement withmeasured high-pressure differentials across faults (Smith and Hogg,1997). The identification of several segment bound-aries at the Sherwood Sandstone level, however, pro-vides justification for petroleum geologists consideringthese locations as potential leakage points across thefault system. Faults in Jurassic unit reservoirs (BridportSands and Frome Clay Limestone) aremore segmentedwith less linkages (Figure 5). Lateral continuity is thusreduced, suggesting greater connectivity between faultblocks and less compartmentalization of the reservoir.Well path trajectory planning may thus continue tobenefit from this fault characterization through thechoice of well paths that avoid fault segments andregions of potential hydrocarbon leakage or waterbreakthrough.

Vertical steps along faults may provide leakagepoints for hydrocarbons migrating vertically throughthe field and may complicate fault location iden-tification at the deeper levels of the oil field whereseismic resolution is poor. The predominantly contrac-tional vertical steps, however, are confined to low-permeability clay-rich units such as the Mercia Mud-stone, which may reduce the leakage tendency throughvertical-step relays in the absence of significant frac-turing. In outcrop, clay becomes entrained along faultzones at vertical steps. At the seismic scale, this phe-nomenon may result in a significant reduction of cross-fault permeabilities (� 0.01 md) (R. Knipe, 1994, per-sonal communication).

CONCLUSIONS

We have proposed a procedure for normal fault inter-pretation that integrates 3-D seismic data analyses withseveral techniques for the recognition of segmentationfeatures. The procedure involves (1) a standard inter-pretation of faults from 3-D seismic data, (2) mappingof outcrop-scale fault analog geometries, (3) numericalmodeling of slip and displacements associatedwith seg-mented fault geometries, and (4) the application of in-sights gained from outcrop observations and modelingto seismic interpretations to develop a more finelyhoned fault model with the aid of seismic attributetools.

A large volume of literature exists to characterizethe mechanics of normal fault systems, and such prin-ciples should be incorporated into the seismic inter-pretations. Outcrop-scale analogs may provide impor-tant insights into the nature of faulting at the seismic

scale, and some fault geometries exhibit a degree ofscale independence between the outcrop and seismicscale. Numerical models based on the principles of lin-ear elastic fracture mechanics provide a means of cal-culating displacement fields associated with overlap-ping faults and can be used to hone the locations offault tips below the level of seismic resolution. Seismicattribute maps are sensitive to faulting and bed dip var-iations and can thus be used to pinpoint deformationnear to and perhaps below the conventional limits ofseismic resolution.

The integration of all these techniques provides alogical basis for maximizing the utility of 3-D seismicdata to construct accurate fault interpretations, whichmay then be applied to oil field development such asthrough the design of optimal well trajectories into po-tential fault-controlled hydrocarbon traps or for pre-dicting leakage points along a fault zone. When thesemethods are collectively applied to Wytch Farm, in-terpreted faults are found to exhibit significant lateraland vertical segmentation, particularly in the shallowerhydrocarbon-producing units. This places emphasis onsegmentation and linkage effects in terms of develop-ing an instructive oil field characterization that isolatespotential flow zones across the faulted reservoir. Fu-ture development of theWytch Farm oil field will ben-efit from attempts to target traps in regions of con-firmed fault continuity rather than zones of potentialsegment boundaries, hydrocarbon leakage, and waterbreakthrough.

Tip-line shapes and slip distributions indicate thatsegments impact on the growth tendencies and slip be-haviors of each other and may be used to isolate loca-tions of potential fault segment boundaries. Steep tiplines and slip contours formed in zones of lateral faultoverlaps, whereas horizontal tip lines formed whereconjugate faults converged and along regions of verticalsegment overlap, resulting in a tendency toward rec-tangular fault shapes. Lithology is a major factor incontrolling both fault nucleation locations in brittlesandstones and limestones and the development of ver-tical steps across thick shale units.

Irregular fault traces in the deeper parts of the stra-tigraphy result from linkage of echelon fault segmentsacross relay zones. Higher in the stratigraphic section,such fault geometries are lost in favor of unlinked east-west–trending segments, implying that upward and lat-eral growth of segments is the first step of an evolu-tionary process in which fault continuity increasesthrough time as slip accumulates and segments me-chanically interact and link together. The resultant


fault orientation is thus controlled by the initial config-uration of echelon fault segments (e.g., left-steppingvs. right-stepping). This hypothesis for segmented faultevolution at Wytch Farm is consistent with the docu-mented north-south extension direction for southernEngland during the Mesozoic.

REFERENCES CITED

Anders, M. H., and R. W. Schlische, 1994, Overlapping faults, in-trabasin highs, and the growth of normal faults: Journal of Ge-ology, v. 102, p. 165–180.

Antonellini, M. A., A. Aydin, and D. D. Pollard, 1994, Microstruc-ture of deformation bands in porous sandstones at Arches Na-tional Park, Utah: Journal of Structural Geology, v. 16, p. 941–959.

Aydin, A., and Y. Eyal, 1996, Anatomy of a normal fault with shalesmear (abs.): AAPG Annual Convention Program, v. 5, p. A7.

Aydin, A., and A. M. Johnson, 1978, Development of faults as zonesof deformation bands and as slip surfaces in sandstone: Pure andApplied Geophysics, v. 116, p. 931–942.

Bouvier, J. D., C. H. Kaars-Sijpesteijn, D. F. Kluesner, C. C. On-yejekwe, and R. C. van der Pal, 1989, Three-dimensional seis-mic interpretation and fault sealing investigations, Nun Riverfield, Nigeria: AAPG Bulletin, v. 73, p. 1397–1414.

Bowman, M. B. J., N. M. McClure, and D. W. Wilkinson, 1993,Wytch Farm oil field: deterministic reservoir description of theTriassic Sherwood Sandstone, in J. R. Parker, ed., Petroleumgeology of northwest Europe: London, Geological Society, Pro-ceedings of the 4th conference, p. 1513–1517.

Burgmann, R., D. D. Pollard, and S. J. Martel, 1994, Slip distribu-tions on faults: effects of stress gradients, inelastic deformation,heterogeneous host-rock stiffness, and fault interaction: Journalof Structural Geology, v. 16, p. 1675–1690.

Butler, M., 1998, The geological history of the southern Wessex ba-sin—a review of new information from oil exploration, in J. R.Underhill, ed., Development, evolution and petroleumgeologyof the Wessex basin: London, Geological Society Special Pub-lication, v. 133, p. 67–86.

Cartwright, J. A., and C. S. Mansfield, 1998, Lateral displacementvariation and lateral tip geometry of normal faults in the Can-yonlands National Park, Utah: Journal of Structural Geology,v. 20, p. 3–19.

Cartwright, J. A., B. D. Trudgill, and C. S. Mansfield, 1995, Faultgrowth by segment linkage: an explanation for scatter in max-imum displacement and trace length data from the Canyon-lands grabens of SE Utah: Journal of Structural Geology, v. 17,p. 1319–1326.

Chadwick, R. A., N. Kenolty, and A.Whittaker, 1983, Crustal struc-ture beneath southern England from deep seismic reflectionprofiles: Journal of the Geological Society, v. 140, p. 893–912.

Childs, C., J. Watterson, and J. J. Walsh, 1995, Fault overlap zoneswithin developing normal fault systems: Journal of the Geolog-ical Society, v. 152, p. 535–549.

Childs, C., A. Nicol, J. J. Walsh, and J. Watterson, 1996, Growth ofvertically segmented normal faults: Journal of Structural Ge-ology, v. 18, p. 1389–1397.

Colter, V. S., and D. J. Havard, 1981, The Wytch Farm oil field,Dorset, in L. V. Illing and G. D. Hobson, eds., Petroleum ge-ology of the continental shelf of north-west Europe: London,Heyden, p. 494–503.

Coussement, C., P. Gente, J. Rolet, J. J. Tiercelin, M. Wafula, andS. Buku, 1994, The north Tanganyika hydrothermal fields, East

African rift system: their tectonic control and relationship tovolcanism and rift segmentation: Tectonophysics, v. 237,p. 155–173.

Crider, J. G., 1998, Mechanical interaction among normal faults: anumerical, field and seismological investigation: Ph.D. thesis,Stanford University, Stanford, California, 153 p.

Crider, J. G., and D. D. Pollard, 1998, Fault linkage: 3-Dmechanicalinteraction between echelon normal faults: Journal of Geo-physical Research, v. 103, p. 24373–24391.

Crone, A. J., and K. M. Haller, 1991, Segmentation and the coseis-mic behavior of Basin and Range normal faults: examples fromeast-central Idaho and southwestern Montana, U.S.A: Journalof Structural Geology, v. 13, p. 151–164.

Dawers, N. H., and M. H. Anders, 1995, Displacement-length scal-ing and fault linkage: Journal of Structural Geology, v. 17,p. 607–614.

Dawers, N. H., and J. R. Underhill, 2000, The role of fault interac-tion and linkage in controlling synrift stratigraphic sequences:Late Jurassic, Statfjord East area, northern North Sea: AAPGBulletin, v. 84, p. 45–64.

Dawers, N. H., M. H. Anders, and C. H. Scholz, 1993, Growth ofnormal faults: displacement-length scaling: Geology, v. 21,p. 1107–1110.

de Polo, C. M., D. G. Clark, D. B. Slemmons, and A. R. Ramelli,1991, Historical surface faulting in the Basin and Range prov-ince, western North America: implications for fault segmenta-tion: Journal of Structural Geology, v. 13, p. 123–136.

Dranfield, P., S. H. Begg, and R. R. Carter, 1987, Wytch Farm oilfield: reservoir characterization of the Triassic Sherwood Sand-stone for input to reservoir simulation studies, in J. Brooks andK. W. Glennie, eds., Petroleum geology of north west Europe:London, Graham and Trotman, p. 149–160.

Ebukanson, E. J., and R. R. F. Kinghorn, 1986, Oil and gas accu-mulations and their possible source rocks in southern England:Journal of Petroleum Geology, v. 9, p. 413–428.

Evans, D. J., and R. A. Chadwick, 1994, Basement-cover relation-ships in the Shaftesbury area of the Wessex basin, southernEngland: Geological Magazine, v. 31, p. 387–394.

Gibson, J. R., J. J. Walsh, and J. Watterson, 1989, Modelling of bedcontours and cross-sections adjacent to planar normal faults:Journal of Structural Geology, v. 11, p. 317–328.

Hawkes, P. W., A. J. Fraser, and C. C. G. Einchcomb, 1998, Thetectono-stratigraphic development and exploration history ofthe Weald and Wessex basins, southern England, in J. R. Un-derhill, ed., Development, evolution and petroleum geology ofthe Wessex basin: London, Geological Society Special Publi-cation, v. 133, p. 39–65.

Jenkyns, H. C., and J. R. Senior, 1991, Geological evidence for intra-Jurassic faulting in the Wessex basin and its margins: Journal ofthe Geological Society, v. 148, p. 245–260.

Karner, G. D., S. D. Lake, and J. F. Dewey, 1987, The thermal andmechanical development of the Wessex basin, southern En-gland, in M. P. Coward, J. F. Dewey, and P. L. Hancock, eds.,Continental extensional tectonics: London, Geological SocietySpecial Publication, v. 28, p. 517–536.

Kattenhorn, S. A., 1994, Outcrop-scale fault-related folds, Valleyand Ridge province, Appalachians: comparison to kinematicmodel predictions: M.S. thesis, University of Akron, Ohio,124 p.

Kattenhorn, S. A., and D. A. McConnell, 1994, Analysis of outcrop-scale fault-related folds, Eagle Rock, Virginia: SoutheasternGe-ology, v. 34, p. 79–88.

Kattenhorn, S. A., and D. D. Pollard, 1999, Is lithostatic loadingimportant for the slip behavior and evolution of normal faultsin the Earth’s crust?: Journal of Geophysical Research, v. 104,p. 28879–28898.


Knipe, R. J., G. Jones, and Q. J. Fisher, 1998, Faulting, fault sealingand fluid flow in hydrocarbon reservoirs: an introduction, inG.Jones, Q. J. Fisher, and R. J. Knipe, eds., Faulting, fault sealingand fluid flow in hydrocarbon reservoirs: London, GeologicalSociety Special Publication, v. 147, p. vii–xxi.

Machette, M. N., S. F. Personius, A. R. Nelson, D. P. Schwartz, andW. R. Lund, 1991, TheWasatch fault zone,Utah-segmentationand history of Holocene earthquakes: Journal of Structural Ge-ology, v. 13, p. 137–149.

Maerten, L., 1999, Mechanical interaction of intersecting normalfaults: theory, field examples and applications: Ph.D. thesis,Stanford University, Stanford, California, 167 p.

Maerten, L., D. D. Pollard, and R. Karpuz, 2000, How to constrain3-D fault continuity and linkage using reflection seismic data:a geochemical approach: AAPG Bulletin, v. 84, p. 1311–1324.

Mansfield, C. S., and J. A. Cartwright, 1996, High resolution faultdisplacement mapping from three-dimensional seismic data:evidence for dip linkage during fault growth: Journal of Struc-tural Geology, v. 18, p. 249–263.

Marchal, D., M. Guiraud, T. Rives, and J. van den Driessche, 1998,Space and time propagation processes of normal faults, in G.Jones, Q. J. Fisher, and R. J. Knipe, eds., Faulting, fault sealingand fluid flow in hydrocarbon reservoirs: London, GeologicalSociety Special Publication, v. 147, p. 51–70.

Martinez, M. C., 1998, Normal faulting in the Pathe geothermalarea, central Mexico: Geofisica Internacional, v. 37, p. 103–111.

McKie, T., J. Aggett, and A. J. C. Hogg, 1998, Reservoir architectureof the upper Sherwood Sandstone, Wytch Farm field, southernEngland, in J. R. Underhill, ed., Development, evolution andpetroleum geology of the Wessex basin: London, GeologicalSociety Special Publication, v. 133, p. 399–406.

Miliorizos, M., and A. Ruffell, 1998, Kinematics of the Watchet-Cothelstone-Hatch fault system: implications for the fault his-tory of the Wessex basin and adjacent areas, in J. R. Underhill,ed., Development, evolution and petroleum geology of theWessex basin: London, Geological Society Special Publication,v. 133, p. 311–330.

Morley, C. K., 1999, Patterns of displacement along large normalfaults: implications for basin evolution and fault propagation,based on examples from East Africa: AAPG Bulletin, v. 83,p. 613–634.

Morley, C. K., R. A. Nelson, T. L. Patton, and S. G. Munn, 1990,Transfer zones in the East African rift system and their rele-vance to hydrocarbon exploration in rifts: AAPG Bulletin,v. 74, p. 1234–1253.

Muraoka, H., and H. Kamata, 1983, Displacement distributionalongminor fault traces: Journal of Structural Geology, v. 5, p. 483–495.

Nicol, A., J. Watterson, J. J. Walsh, and C. Childs, 1996, The shapes,major axis orientations and displacement patterns of fault sur-faces: Journal of Structural Geology, v. 18, p. 235–248.

Ottesen Ellevset, S., R. J. Knipe, T. S. Olsen, Q. J. Fisher, and G.Jones, 1998, Fault controlled communication in the SleipnerVest field, Norwegian continental shelf: detailed, quantitativeinput for reservoir simulation and well planning, in G. Jones,Q. J. Fisher, and R. J. Knipe, eds., Faulting, fault sealing andfluid flow in hydrocarbon reservoirs: London, Geological Soci-ety Special Publication, v. 147, p. 283–297.

Peacock, D. C. P., and D. J. Sanderson, 1991, Displacements, seg-ment linkage and relay ramps in normal fault zones: Journal ofStructural Geology, v. 13, p. 721–733.

Peacock, D. C. P., and D. J. Sanderson, 1992, Effects of layering andanisotropy on fault geometry: Journal of theGeological Society,v. 149, p. 793–802.

Peacock, D. C. P., and D. J. Sanderson, 1994, Geometry and devel-opment of relay ramps in normal fault systems: AAPGBulletin,v. 78, p. 147–165.

Pegrum, R. M., and A. M. Spencer, 1990, Hydrocarbon plays in thenorthern North Sea: London, Geological Society Special Pub-lication, v. 50, p. 441–470.

Pickering, G., D. C. P. Peacock, D. J. Sanderson, and J. M. Bull,1997, Modeling tip zones to predict the throw and length char-acteristics of faults: AAPG Bulletin, v. 81, p. 82–99.

Pollard, D. D., A. Aydin, J. G. Crider, S. Kattenhorn, and L. Maer-ten, 1997, Testing 3-D numerical models of normal faults andassociated fractures using outcrop analogs and seismic data(abs.): AAPG Annual Convention Program, v. 6, p. A94.

Selley, R. C., and R. Stoneley, 1987, Petroleum habitat in southDorset, in J. Brooks and K.W. Glennie, eds., Petroleumgeologyof north west Europe: London, Graham and Trotman, p. 139–148.

Smith, G. S., and A. J. C. Hogg, 1997, Integrating static and dynamicdata to enhance extended reach well design: San Antonio,Texas, Society of Petroleum Engineers, SPE 38878, 12 p.

Stoneley, R., 1982, The structural development of theWessex basin:Journal of the Geological Society, v. 139, p. 543–554.

Thomas, A. L., 1993, Poly3-D: a three-dimensional, polygonal ele-ment, displacement discontinuity boundary element computerprogram with applications to fractures, faults, and cavities inthe Earth’s crust: M.S. thesis, Stanford University, Stanford,California, 97 p.

Trudgill, B., and J. Cartwright, 1994, Relay-ramp forms and normal-fault linkages, CanyonlandsNational Park, Utah:Geological So-ciety of America Bulletin, v. 106, p. 1143–1157.

Underhill, J. R., and S. Paterson, 1998, Genesis of tectonic inversionstructures: seismic evidence for the development of key struc-tures along the Purbeck–Isle of Wight disturbance: Journal ofthe Geological Society, v. 155, p. 975–992.

Underhill, J. R., and R. Stoneley, 1998, Introduction to the devel-opment, evolution and petroleum geology of theWessex basin,in J. R. Underhill, ed., Development, evolution and petroleumgeology of the Wessex basin: London, Geological Society Spe-cial Publication, v. 133, p. 1–18.

Walsh, J. J., and J. Watterson, 1988, Analysis of the relationshipbetween displacements and dimensions of faults: Journal ofStructural Geology, v. 10, p. 239–247.

Walsh, J. J., and J. Watterson, 1989, Displacement gradients on faultsurfaces: Journal of Structural Geology, v. 11, p. 307–316.

Wells, D. L., and K. J. Coppersmith, 1994, New empirical relation-ships among magnitude, rupture length, rupturewidth, rupturearea, and surface displacement: Bulletin of the SeismologicalSociety of America, v. 84, p. 974–1002.

Wesnousky, S. G., 1988, Seismological and structural evolution ofstrike-slip faults: Nature, v. 335, p. 340–343.

Willemse, E. J. M., 1997, Segmented normal faults: correspondencebetween three-dimensional mechanical models and field data:Journal of Geophysical Research, v. 102, p. 675–692.

Willemse, E. J. M., and D. D. Pollard, 2000, Normal fault growth:evolution of tipline shapes and slip distribution, in F. K. Lehnerand J. L. Urai, eds., Aspects of tectonic faulting: Berlin, Ger-many, Springer-Verlag, p. 193–226.

Willemse, E. J. M., D. D. Pollard, and A. Aydin, 1996, Three-dimensional analyses of slip distributions on normal fault arrayswith consequences for fault scaling: Journal of Structural Ge-ology, v. 18, p. 295–309.

Yielding, G., J. A. Øverland, and G. Byberg, 1999, Characterizationof fault zones for reservoir modeling: an example from theGull-faks field, northern North Sea: AAPG Bulletin, v. 83, p. 925–951.

integrating 3-d seismic data, field analogs, and mechanical models

Documents