AUTHORS

Keumsuk Lee � Bureau of Economic Geolo-gy, John A. and Katherine G. Jackson Schoolof Geosciences, University of Texas at Austin,Austin, Texas 78713-8924;Keumsuk.Lee@beg.utexas.edu

Keumsuk Lee received his B.Sc. degree in math-ematics (1994) and his M.Sc. degree in geo-logical oceanography (1999) from KunsanNational University, South Korea, and his Ph.D.in geophysics from the University of Texas atDallas in 2005. He is currently working with theBureau of Economic Geology, University ofTexas at Austin. His main research interests arebasin analysis based on seismic sequencestratigraphy and reservoir characterizationusing ground-penetrating radar data.

M. Royhan Gani � Energy and GeoscienceInstitute, University of Utah, 423 Wakara Way,Suite 300, Salt Lake City, Utah 84108;rgani@egi.utah.edu

M. Royhan Gani is a research assistant professorat the University of Utah. His research inter-ests are clastic sedimentology, surficial pro-cesses, and tectonics. He specializes in faciesarchitecture and sequence stratigraphy of clas-tic depositional systems. He received his B.Sc.and M.Sc. degrees in 1997 and 1999 fromthe University of Dhaka, Bangladesh, and hisPh.D. in 2005 from the University of Texasat Dallas.

George A. McMechan � Center for Litho-spheric Studies, University of Texas at Dallas,P.O. Box 830688, Richardson, Texas 75083-0688; mcmec@utdallas.edu

George A. McMechan received a B.A.Sc. de-gree in geological engineering from the Univer-sity of British Columbia in 1970 and an M.Sc.degree in geophysics from the University ofToronto in 1971. He is currently the Ida GreenProfessor of Geosciences at the University ofTexas at Dallas. He has published more than220 technical articles and, in 1997, receivedthe Virgil Kauffman gold medal from the So-ciety of Exploration Geophysicists. His mainresearch interests are wavefield imaging, three-dimensional seismology, reservoir character-ization, ground-penetrating radar, and parallelcomputing. He is a member of the Society ofExploration Geophysicists, American Geophys-ical Union, Seismological Society of America,

Three-dimensionalfacies architecture andthree-dimensional calciteconcretion distributions ina tide-influenced delta front,Wall Creek Member, FrontierFormation, WyomingKeumsuk Lee, M. Royhan Gani, George A. McMechan,Janok P. Bhattacharya, Stephanie L. Nyman, andXiaoxian Zeng

ABSTRACT

Ground-penetrating radar (GPR) has been used to image the three-

dimensional (3-D) internal structure (and, thus, the 3-D facies archi-

tecture) of a top-truncated delta front in the topmost parasequence

in the Wall Creek Sandstone Member of the Frontier Formation in

Wyoming and to estimate the distribution of low-permeability

concretions throughout the 3-D GPR volume.

The interpretation of the 3-D GPR data is based both on cor-

relations with outcrop and on calibration with core data from holes

within the survey grid. Two main radar facies (RF) are identified.

Radar facies 1 corresponds to tide-influenced mouth bars formed

by a unidirectional flow during delta progradation or bidirectional

flow during tides, whereas RF2 is correlated with laterally migrat-

ing channels developed on previous bar deposits. The delta-front

foreset beds dip in the same direction as the dominant paleocurrent

indicators. The GPR interpretation is consistent with the outcrop

interpretation that, following a regressive period, bars and channels

were developed at the Raptor Ridge site before subsequent trans-

gressive ravinement. The individual 3-D deltaic facies architectures

were reconstructed from the 3-D GPR volume and indicate that the

depositional units are larger than the survey grid.

GEOHORIZONS

AAPG Bulletin, v. 91, no. 2 (February 2007), pp. 191–214 191

Copyright #2007. The American Association of Petroleum Geologists. All rights reserved.

Manuscript received July 21, 2005; provisional acceptance September 30, 2005; revised manuscriptreceived January 26, 2006; final acceptance August 31, 2006.

DOI:10.1306/08310605114

Cluster analysis of the GPR attributes (instantaneous ampli-

tudes and wave numbers) calibrated with the cores and the outcrop

was used to predict the distribution of near-zero permeability con-

cretions throughout the 3-D GPR volume; clusters of predictive

attributes were defined and applied separately in the bars and chan-

nels. The predicted concretions in the bars and the channels are

14.7 and 10.2% by volume, respectively, which is consistent with

those observed in the cores (14.7 and 10.5%, respectively), and their

shape and thickness are also generally in consonance with those in

the outcrop and cores. The estimated concretions are distributed

in an aggregate pattern with irregularly shaped branches within the

3-DGPR volume, indicating that the cementation does not follow a

traditional center-to-margin pattern. The concretions and 3-D geo-

logical solid model provide cemented flow baffles and a 3-D struc-

tural framework for 3-D reservoir modeling, respectively.

INTRODUCTION

Ground-penetrating radar (GPR) has been used for detailed sub-

surface imaging of deltaic deposits since the early 1990s; however,

most studies have focused on modern deltas associated with uncon-

solidated fluvial deposits (Jol and Smith, 1991; Pepola and Hickin,

2003; Roberts et al., 2003), instead of ancient marine environments

exposed in outcrops. Although GPR has revealed high-resolution

structural information in fluvial sediment bodies and has been used

to build reservoir models (McMechan et al., 1997; Corbeanu et al.,

2001; Zeng et al., 2004), few examples document detailed three-

dimensional (3-D) delta-front facies architecture (Lee et al., 2005),

and there are no studies of 3-D cement distribution. Preserved delta

fronts in outcrops are commonly characterized by prograding

clinoforms and systematic variations in sedimentary facies (Bhat-

tacharya and Walker, 1992; Snelgrove et al., 1996, 1998).

Calcite cement is a common diagenetic feature in sandstone

reservoirs. Pervasive pore-filling calcites can be found in spheroi-

dal, elongate, tabular, or irregular forms (McBride, 1996). Calcite-

cemented sandstone can occur over a range of burial depths, depend-

ing on the supply of the cementing materials. For example, calcite

cement zones are characteristic within the Middle Jurassic delta-

front sandstones in the North Sea (Walderhaug and Bjorkum, 1992).

Because concretions fill up the pore spaces as their volume ex-

pands (Dutton et al., 2002), the permeability and porosity distribu-

tions in sandstone reservoirs may be significantly affected (Hassouta

et al., 1999). This has been demonstrated by two-dimensional (2-D)

reservoir simulations (Dutton et al., 2002; White et al., 2003).

Because three-dimensional models of concretion-bearing sand-

stone reservoirs are derived from the limited information avail-

able from outcrop and borehole sedimentology, fluid behavior is

difficult to predict accurately. Although boreholes can provide sta-

tistically defined 3-D information, uncertainty remains in interwell

interpolation. In contrast, GPR can provide detailed 3-D structural

Environmental and Engineering GeophysicalSociety, and the Association of ProfessionalEngineers and Geoscientists of British Columbia.

Janok P. Bhattacharya � Geosciences De-partment, SR1 Rm. 312, University of Houston4800 Calhoun Rd., Houston, Texas 77204-5007; jpbhattacharya@uh.edu

Janok P. Bhattacharya is the Robert E. SheriffProfessor of Sequence Stratigraphy at theUniversity of Houston. His research interests in-clude deltaic sedimentology and sequencestratigraphy, the local control of structure onstratigraphy, and reservoir architecture ofclastic depositional systems. He received hisB.Sc. degree in 1981 from the Memorial Uni-versity of Newfoundland, Canada, then workedat ESSO Resources Calgary, before completinghis Ph.D. in 1989 from McMaster University,Hamilton, Ontario, Canada. Janok worked forthe Bureau of Economic Geology at Austin,ARCO Research in Plano, Texas, and the Uni-versity of Texas at Dallas before joining theUniversity of Houston last fall. He has receivednumerous awards for his service to geologicalorganizations and for papers he has presented.He is an associate editor for both the Journalof Sedimentary Research and AAPG Bulletinand has authored more than 100 abstracts and40 technical articles. He is an active memberof AAPG, SEPM, the Geological Society ofAmerica, and the International Association ofSedimentologists.

Stephanie L. Nyman � Department of Earthand Ocean Sciences, University of Waikato,Private Bag 3105, Hamilton, New Zealand;snyman@waikato.ac.nz

Stephanie L. Nyman received her M.S. degreein geology from the University of Texas atDallas. Currently, she is working on a Ph.D.at the University of Waikato, Hamilton, NewZealand. Her geologic specialty is clastic dia-genesis, with research interest in carbonatecement within subsurface hydrocarbon seepsystems.

Xiaoxian Zeng � Center for LithosphericStudies, University of Texas at Dallas, P.O.Box 830688, Richardson, Texas 75083-0688;zeng@utdallas.edu

Xiaoxian Zeng received his B.S. degree in geo-physics from Peking University and his Ph.D.in geosciences from the University of Texas at

192 Geohorizons

Dallas. He is currently a research associateat the University of Texas at Dallas. His researchinterests include three-dimensional imagingand velocity model building, with applicationto both seismic and ground-penetratingradar data.

ACKNOWLEDGEMENTS

This research is funded by the U.S. Departmentof Energy under Contract DEFG0301-ER15166with supplementary support from the PetroleumResearch Fund of the American Chemical So-ciety under ACS-PRF Grant 35855-AC8 and thesponsors of the University of Texas at DallasQuantitative Sedimentology Consortium. Winpics(Divestco, Inc.) and Gocad (Earth Decision Sci-ences) software were used for three-dimensionalground-penetrating radar interpretation andthe corresponding visualization of the solidmodel, respectively. Constructive commentsby reviewers Dennis Harry, Bradford Macurda, Jr.,and Jory Pacht were appreciated. This articleis Contribution 1094 from the Department ofGeosciences at the University of Texas at Dallas.

information within the sandstone reservoirs and also provide

estimates of permeability and porosity through calibration with

core measurements.

The distribution of obstacles to fluid flow in a reservoir may

be extracted from GPR data by analyzing GPR attributes (Lemke

and Mankowski, 2000). Corbeanu et al. (2002) and Hammon et al.

(2002) estimated the spatial distribution of mudstone flow baf-

fles and barriers in fluvial deposits by calibrating the GPR attri-

butes to the borehole measurements via cluster analysis. This tech-

nique is also applicable to deltaic sandstone reservoirs to provide a

3-D image of the diagenetic features (cements), as demonstrated

below.

This study has two objectives. The first is to image the internal

structure of a top-truncated lowstand delta front from 3-D GPR

data acquired at the Raptor Ridge reservoir analog site inWyoming;

this allows for the 3-D visualization of the deltaic sedimentary bod-

ies. The second objective is to estimate the distribution of low-

permeability cements that are flow baffles, by calibrating GPR at-

tributes (instantaneous amplitudes and wave numbers), with core

permeability measurements. The resulting models are suitable for

input to fluid-flow simulation (Tang, 2005).

GEOLOGICAL SETTING

The study area is located between the Powder River andWind River

basins in Wyoming (Figure 1) and contains deltaic environments

influenced by rivers, waves, and tides (Bhattacharya and Willis,

2001). During the Late Cretaceous, sediments were transported

eastward from the topographic highs of the Sevier orogenic belt into

the western margin of the Western Interior seaway, forming the

Frontier Formation (Barlow and Haun, 1966; Dyman et al., 1994).

The Wall Creek sediments form the uppermost member of the

Frontier Formation. They were deposited into a foreland basin and

were eroded significantly during subsequent transgression (Bhatta-

charya and Willis, 2001). Continuous, north-south–oriented, Wall

Creek outcrops are well exposed in a series of sandstone cliffs along

the Frontier outcrop belt (Figure 1). Winn (1991) interpreted the

sandstone bodies as storm-dominated offshore shelf deposits;

however, a recent study suggests that the formation contains top-

truncated deltas (Bhattacharya and Willis, 2001; Lee et al., 2005).

The Wall Creek Sandstone Member is interpreted to consist of sev-

eral delta lobes, indicated by different sandstone bodies (in para-

sequences PS 1 to PS 6 from oldest to youngest, Figure 2) (Howell

et al., 2003). The exposed topmost parasequence (PS 6) at Raptor

Ridge, near the southern end of the Frontier outcrop belt, was chosen

for this study.

The Raptor Ridge site contains a mixed-influenced delta lobe

capped by a transgressive ravinement surface (Gani, 2005). The site

provides favorable conditions for a GPR survey; it has near-planar

Lee et al. 193

surface topography that is cut by a northwest-southeast–

oriented valley (Figure 1). The Raptor Ridge vicinity

contains exposed sedimentary columns of the outcrop

showing the six parasequences; only the topmost del-

taic sand body (PS 6) can be clearly seen along the cliff

face (Figures 3–5). Parasequence 6 is interpreted as a top-

eroded lowstand delta front (Gani, 2005). In the delta-

front sandstone, calcite cements commonly show a red-

dish color in the cliff face (Figures 3, 4).

DATABASE

The database for this study consists of a 3-D GPR data

volume, outcrop sedimentology conducted on a 300-m

(1000-ft)-long cliff face, 10 cores drilled behind the

cliff to a depth of 11 m (36 ft), digital cliff face photos,

and the present-day topography. The GPR data are in-

tegrated with the geologic data for calibration and in-

terpretation. The 3-D data used here (Figure 5) were

Figure 1. The RaptorRidge site is located nearthe west margin of thePowder River Basin ineast central Wyoming inthe United States. Thelower panel shows thetopography of the studyarea (with 20 ft [6.096 m]contour interval), the lo-cations of the GPR surveygrid, and the boreholes.Oil and gas fields are froma public domain Web site(WyGISC, 1990).

194 Geohorizons

3 km

50 mAB

EMIGRANTGAP

WALLCREEK

RA

PT

OR

RID

GE

PA

RA

SE

QU

EN

CE

1 (

PS

1)

PA

RA

SE

QU

EN

CE

2 (

PS

2)

PA

RA

SE

QU

EN

CE

3 (

PS

3)

PA

RA

SE

QU

EN

CE

4 (

PS

4)

PA

RA

SE

QU

EN

CE

5 (

PS

5)

PA

RA

SE

QU

EN

CE

6 (

PS

6)

MU

DS

TO

NE

MA

RIN

E E

RO

SIO

NA

L S

UR

FA

CE

BE

NT

ON

ITE

PE

BB

LE C

ON

GLO

ME

RA

TE

N

E

S

W

A

B(a

)(b

)

Figure

2.StratigraphicsectionalongtheFrontieroutcropbeltcontaining

sixparasequences.TheRaptor

Ridgesiteliesinthetopm

ostparasequence

(PS6)

near

thesouthern

end

ofthesectionAB.

Modified

from

How

ellet

al.(2003).

Figure

3.Photom

osaicof

thecliff

face

inthedepositionaldipdirection(a)andits

beddingdiagram

(b)show

ingthealternationof

southw

ard-dippingclinoformsof

the

distributary

mouth

bars

separatedby

laterally

migratingchannels.SeeFigure

5forlocation.

In(a),concretions

arethereddishareaswithin

thebarandchannelsandstones.

Lee et al. 195

collected in the context of a larger project, which also

includes two larger 2-D grids. The 2-D data provide

some constraints at the edges of the 3-D grid and help

to interpret the radar facies within a larger scale frame-

work. Detailed analysis of the 2-D data will be pre-

sented elsewhere.

The GPR reflection data were collected at the Rap-

tor Ridge site in the summer of 2002 using a Pulse-

EKKO IV system (manufactured by Sensors & Soft-

ware, Inc.). The 3-D GPR data were acquired on a

rectangular survey grid of 30 � 80 m (98 � 262 ft)

(Figures 1, 5). The 3-D volume had 1-m (3.3-ft) GPR

line spacing and 0.25-m (0.82-ft) trace spacing along each

line; the lines run approximately parallel to the paleo-

current direction (Figure 5). A common-offset geom-

etrywith 3m (9.8 ft) transmitter-to-receiver spacingwas

used to obtain the GPR data, at a frequency of 50 MHz.

The time window and sampling interval were set at 300

and0.8ns, respectively, and traceswere stacked128 times

at each survey point to improve the signal-to-noiseFigure

4.Photom

osaicof

thecliff

face

inthedepositionalstrike

direction(a)andits

beddingdiagram

(b)show

ingthealternationof

subhorizontalbedsetsof

thedistributary

mouth

bars

separatedby

laterally

migratingchannels.SeeFigure

5forlocation.

In(a),concretions

arethereddishareaswithin

thebarandchannelsandstones.

Figure 5. Detailed map of the Raptor Ridge site showing thelocations of the sedimentary cores, the photomosaics, and the3-D survey grid. Paleocurrent direction measurements conductedon the topmost parasequence are indicated in the circle. SeeFigure 1 for location.

196 Geohorizons

ratio. Topographic data collected with a global posi-

tioning system (GPS) with accuracy of approximately

0.03 m (0.098 ft) were used to define the datum for

depth migration of the GPR data.

Detailed outcrop analysis of the exposed cliff face

was done in 2002 and includedmapping of facies, shales,

cements, bedding and bounding surfaces, paleocurrent

directions, and traditional vertical logged sections com-

bined with high-resolution digital photomosaics (Gani,

2005). Clinoform surfaces traced along the outcrop al-

low correlations of the sedimentary facies and struc-

tures with the 3-D GPR data.

Ten cores (Figure 5) were drilled to provide inde-

pendent sedimentologic information (including con-

cretions) away from the cliff faces and unweathered

samples of the entire section for permeability and po-

rosity measurements and thin sections. Two of the cores

(cores 8 [10.9 m {35.7 ft} depth] and 9 [10.7m {35.1 ft}

depth]) in the 3-D GPR volume are used for depth con-

trol for processing and interpretation of the 3-D GPR

data. These two cores are also used for calibration of the

GPR attributes to predict the concretion distribution.

The other eight cores are used to constrain analysis of

a 2-D GPR grid (which is not presented here).

GROUND-PENETRATING RADARDATA PROCESSING

Preprocessing of the 3-D GPR data (Table 1) is similar

to that described by Lee et al. (2005). The average trace

amplitude (Figure 6a), determined over a user-specified

time and trace window after air and direct wave re-

moval, shows northwest- to southeast-trending linear

anomalies (parallel to the survey lines), which are acqui-

sition artifacts (Brown, 2004; Cordsen, 2004) caused by

antenna coupling changes from line to line. These un-

desired amplitude artifacts were effectively reduced by

performing amplitude balancing (Figure 6) by scaling

average absolute amplitude of each survey line to the

same reference level.

Prestack Kirchhoff depth migration (Epili and Mc-

Mechan, 1996) was performed on the GPR data with a

3-D migration velocity model obtained by matching

GPR reflection depths with the corresponding bound-

aries in the cores. The migration code was modified

to analytically compute the traveltimes instead of ray

tracing, by assuming locally constant velocity, resulting

in a significant reduction of the computation time (Lee

et al., 2005). The migration used the topographic sur-

face as the datum and, thus, needs no elevation statics.

The grid increments in the migrated volume are 0.1 m

(0.33 ft) in the depth and in in-line directions and 0.2m

(0.66 ft) in the crossline direction. An automatic gain

control (AGC) with a 0.5-m AGC window is applied

to the migrated profiles to compensate for attenua-

tion to make the deeper parts of the images more

visible for interpretation (Table 1). For the concre-

tion calibration, the migrated GPR volume is used

with no AGC because instantaneous amplitudes are

used.

GROUND-PENETRATING RADAR DATAANALYSIS AND INTERPRETATIONS

This article follows the facies architectural model sug-

gested by Gani (2005) based on an outcrop study at

Raptor Ridge. The model consists mainly of five basic

facies-architectural elements: prodelta fines, frontal

splay, tidally modulated deposit, channel, and bar ac-

cretion. Their corresponding sedimentary facies are fa-

cies 1–5, respectively in Figure 7 and are bounded by

five orders of bounding surfaces (zero to fourth order

from lowest to highest), which are defined on the prin-

ciple of superposition and crosscutting relationships

(Gani, 2005).

For a more confident interpretation, the outcrop

and two cores (8 and 9) were directly compared with

the GPR data to establish radar facies type and bound-

aries (Figures 8, 9; Table 2). The radar facies were

recognized from the GPR volume to map facies ar-

chitectures, and then the depositional setting was

inferred from the corresponding radar facies, corre-

lated to core and outcrop data, which are associated

with the depositional setting. However, the inherent

resolution limitation (a quarter wavelength) allows

50 MHz GPR to detect no lower than third-order

surfaces.

The third-order bounding surfaces bind channels

and sets of most of the subordinate surfaces down-

lapping and/or onlapping onto the third-order surfaces

(Figure 3). The top transgressive marine ravinement

surface TES (Figure 3) is fourth order, truncates lower

order surfaces, and erodes delta-plain deposits; it is a re-

gionally traceable surface strewn with granules and peb-

bles. The GPR penetration was limited in most places

from 10 to 12m (33 to 39 ft) depth, which is the depth

to the top of the prodelta mud deposits that highly at-

tenuate the GPR waves.

Lee et al. 197

Table 1. Ground-Penetrating Radar Data-Processing Steps Applied to the GPR Volume

198 Geohorizons

Ground-Penetrating Radar Facies Analysis

Radar facies were identified using reflection amplitude,

continuity, and configuration. The radar facies were

compared with the sediment textures in the cores and

were correlated with the facies architectures seen in

the outcrop. The calibration and correlation provide

the crucial information to correctly identify the depo-

sitional features, which produce the radar reflections,

and to confirm the identity of the sedimentary struc-

tures detected in the subsurface.

Two radar facies (RF1 and RF2) are recognized in

this study (Table 2), although the tide influence and

diagenetic features prevent discernment between them

in some places within the GPR volume. RF1 is charac-

terized by moderate- to high-amplitude and continuity

reflections both in the dip and strike directions. The

corresponding radar facies show approximately east-

west–striking, repetitive, wide, subparallel reflection pat-

terns on planar slices (i.e., Figure 10a). RF1 is subdi-

vided into two subordinate radar facies (RF1a and RF1b)

in the dip direction, although being similar to each

other in the strike direction (Table 2).

RF1a consists of oblique reflection configurations

with in-situ dip angles of about 5j, which is consis-

tent with that of the third-order bounding surfaces

of bars (and, thus, is consistent with prograding delta-

front foreset beds; Lee et al., 2005). RF1b is charac-

terized by low-angle (1–2j), subhorizontal, landward-

dipping reflectors, with occasional steps. Because of

difficulty identifying the corresponding sedimentary

facies, RF1b can be interpreted to represent two dif-

ferent depositional facies of the delta-front bars (up-

stream accretion and/or tidal modulation of the mouth

bars), unless otherwise constrained by other evidence

(Jol and Bristow, 2003). These two facies can both

produce landward-dipping bedding geometry, but the

depositional environments are different. Although the

landwardmigrations of the delta-front bars are observed

in a river-dominated delta (van Heerden and Roberts,

1988), tidally modulated bars can be found in tide-

influenced river deltas (Willis et al., 1999). Tidal sand

bars are also commonly associated with herringbone

cross-bedding. Siegenthaler (1982) showed herring-

bone cross-bedding surfaces with less than 1 m (3.3 ft)

vertical separation, between which are the foreset beds

Figure 6. Amplitude balancing removes the linear stripes indicated by arrows in (a) that are artifacts of the data acquisition. Eachpanel is a view, from above, of the survey grid, and the amplitude plotted as each survey location is the average absolute amplitudeof the trace at that location. (a) is before balancing and (b) is after.

Lee et al. 199

of tidal sand bars. The ebb-flood–induced surfaces are

subhorizontal or seaward dipping, but they are still

lower angle than the foreset beds. Similar features

were observed in the sandstone outcrop at Raptor

Ridge (Gani, 2005). Thus, RF1b would be more rea-

sonably interpreted as the internal architecture of

tidal sand bars.

RF2 is characterized by short, complex, low- to high-

amplitude, and continuity reflections both in the dip and

strike direction profiles; oblique and sigmoidal accretion

Figure 7. A representative stratigraphic log showing sedimentary facies of the Raptor Ridge (Gani, 2005). Facies 1 = thinlylaminated, very fine-grained, sandy mudstone; facies 2 = interbedded mudstone and thick, very fine- to fine-grained, rippledsandstone; facies 3 = sharp- to erosional-based sand fine- to medium-grained sandstone with massive to parallel lamination;facies 4 = medium-grained, channelized sandstone; facies 5b = fine- to medium-grained, mostly landward-directed, trough cross-stratified sandstone with mud drapes; facies 5a = fine- to medium-grained, mostly seaward-directed, trough cross-stratifiedsandstone.

200 Geohorizons

Figure

8.A

northw

est-southeast–trending,dip-orientedGPR

profile

throughcores8and9,illustratingthecorrelationofboreholelithofacies

toGPR

facies:(a)uninterpretedradar

line;(b)interpretedradarline.Therectangularbox(5

�10

m;16

�33

ft)in(b)representsdip-view

radarfacies

show

ninTable2;thethicksolid

lineintheboxindicatesthe

correspondingstrike-viewradarfacies.SDR=seaw

ard-dippingreflection;LDR=landward-dippingreflection;andGU1toGU3d

areGPR

units,asdescribedinthetext.See

Figures5

and12

forlocationanddetailedlithofacies,respectively.

Lee et al. 201

events that downlap theU-shaped reflection are observed

in the strike direction, suggesting a channel migration

(Table 2). Because of mud clasts and accretionary fea-

tures in the channel facies, the corresponding GPR re-

flection configurations are irregularly oriented,mottled,

or chaotic on planar slices (i.e., Figure 10b).

Ground-Penetrating Radar Boundaries andGround-Penetrating Radar Architectural Elements

The core-calibrated GPR bounding surfaces and as-

sociated GPR units (GUs) are identified within the

3-D volume and are well correlated with the outcrop

Figure 9. A northeast-southwest– trending, strike-oriented GPR profile through core 9, illustrating the correlation of outcroplithofacies to GPR facies: (a) uninterpreted radar line; (b) interpreted radar line; (c) part of the dip-view outcrop. PB = proximalmouth bar; CH = channel; DB = distal mouth bar; and GU1 to GU3d are GPR units, as described in the text. The images in (a) and(b) are vertically exaggerated 2.3 times. See Figures 3 and 5 for outcrop section and radar profile, respectively.

Table 2. Ground-Penetrating Radar Facies Recognized in This Study and Their Geological Interpretation*

202 Geohorizons

interpretation (Figures 3, 4). The GPR boundaries

are traceable through the whole 3-D GPR volume

(Figures 11, 12), suggesting that the interbar bound-

aries are substantially longer than the survey grid. The

GPR units are referred to as GU1 toGU3, respectively,

from oldest to youngest. To reconstruct the individ-

ual 3-D facies architectures, the (picked) horizon in-

terpretations were imported into Gocad. Figure 12

shows both the shapes of the erosional surfaces and

the 3-D geometry of the bodies; to the best of our

knowledge, this is the first time that delta-front fa-

cies architectures can be readily seen at the submeter

scale in 3-D.

3-D Facies Architecture Descriptions

GU1 is the thinnest GPR unit (0.5–1.5 m; 1.6–4.9 ft)

and forms a single architectural element, extending

throughout the 3-D volume (Figure 12).GU1 is charac-

terized by seaward- and landward-dipping reflections

in RF1 (Figure 8), but the reflections are weak, with

low continuity because of the presence of the interbed-

ded mud deposits as well as the inherent GPR resolu-

tion limitations. The top of GU1 forms three east-west–

trending troughs that are overlain by GU2.

GU2 is composed of four elements (GU2a to

GU2d in Figure 12), and the thickness of its elements

increases upward, ranging from 3.5 to 5 m (11.5 to

16.4 ft). The subunits are characterized by RF2 and

are vertically stacked through about 5 m (16.4 ft). The

top surface of the stacked radar unit is relatively flat

throughout the GPR volume.

GU3 also consists of four elements (GU3a to GU3d

in Figure 12) and is characterized by RF1; the three

subunits (GU3a to GU3c) are composed of seaward-

dipping reflections (RF1a), whereas the top subunit

(GU3d) contains amixture of RF1a and RF1b in which

each radar facies is locally lapping against the other; a

gradational transition occurs between them, both ver-

tically and laterally. Thus, it is difficult to delineate in-

dividual elements from theGPRboundaries (i.e., onlap,

downlap, and toplap).

Figure 10. Slices showing clipped amplitude variations on planes at about (a) 3 m (10 ft) and (b) 8 m (26 ft) depth below, andparallel to, the topographic surface ([a] is in the bars, and [b] is in the channels). The bar facies (RF1 in [a]) shows east-west–trending, banded, high amplitudes with spotted patterns interpreted as tidal modulation. The channel facies (RF2 in [b]) exhibitsirregularly oriented, mottled, or chaotic patterns caused by mud clasts and accretionary events.

Lee et al. 203

Interpretations

GU1 is interpreted as distal delta-front bar deposits

(Figure 13) consisting of facies 2 and 3 (Figure 7),

which is consistent with the outcrop interpretation

(Figures 3, 4). The coexistence of seaward- and landward-

dipping reflections (Figure 8) suggests that the distal

mouth bars are slightly tidally modulated at the delta

front. The trough structures of GU1 are the result of

lateral erosion by the overlying GPR unit (GU2).

GU2 is interpreted as channel-fill deposits (Figure 13),

as a result of intermittent fluvial flood events (Gani,

2005), as seen in the outcrop (facies 4 in Figure 7). The

stacked radar units represent the product of an accumu-

lation by a high-energy series of channelizing events,

each separated by scour surfaces. The scour and depo-

sition events repeat until the top channel fill formed a

boundary between the channel and the overlying bar-

growth elements (GU3).

GU3 is interpreted as progressively tidallymodulated,

proximal river-mouth bars at the delta front (facies 5

in Figure 7) because it contains a higher sand:mud ratio

than the distal bars (GU1) (Figure 13); the top delta-

front body corresponds to the bar deposits overlain by

the channels in the outcrop (Figure 9). The mixed bar

Figure 11. Three-dimensional perspectiveview of the migrated GPRvolume (a) and its 3-Dsolid geological model(b) showing distributary-bar deposits separatedby laterally migratingchannels. A verticallyseparated view of thesolid model is shown inFigure 10. The upper-most ravinement surface(the TES) is not shown,because the plot startsslightly deeper.

204 Geohorizons

facies suggest a change in depositional environment from

a river-dominated to a tide-influenced river delta. The

single radar-facies architectures (GU3a to GU3c) rep-

resent dominantly prograding river-mouth bars depos-

ited by a unidirectional flow (i.e., a river flow or an ebb

tide current). The presence ofmixed radar facies in the

subunit (GU3d) is indicative of highly tide-influenced

mouth-bar deposits, whose internal structures may be

highly modified by migrations of dunes and bars dur-

ing bipolar tides (Willis, 2005).

In summary, distal delta-front mouth bars with rela-

tively small tide influence were deposited on a prodeltaic

layer at the Raptor Ridge site, consisting of very fine-

grained frontal splay deposits with mudstone rip-up

Figure 12. A vertically separated version ofthe 3-D GPR interpreted volume in Figure 9showing nine facies architectural elementswithin the GPR units.

Lee et al. 205

clasts (GU1, Figure 12), and they were subsequently

eroded by laterally migrating channels (GU2) that are

interpreted to be river-flood-induced deposits. The four-

stacked channel deposits are capped by more proxi-

mal, increasingly tidallymodulated, mouth-bar deposits

(GU3) that were dominant until subsequent transgres-

sive ravinement eroded the top of GU3. The lower

(GU1) and upper (GU3) bar sediments were deposited

in the paleoflow direction. The delta-front development

is consistent with the outcrop analysis by Gani (2005).

OBSERVED DISTRIBUTION OFCALCITE CONCRETIONS

Calcite Concretions in Cliff Face and Cores

Calcite cementation is a major diagenetic event at Rap-

tor Ridge and is confined to the middle of the deltaic

sand body (Figures 3, 4). Cemented concretions were

mapped along a 100-m (330-ft)-long outcrop in both

the dip and strike directions (Figures 3, 4). The cement

Figure 13. Lithofacies of core 9 and the per-meability profile showing a coarsening-upwardsequence of alternating bar and stacked channelfacies. Cemented concretions are indicated bynear-zero permeability.

206 Geohorizons

is patchy, forming elongate, tabular concretions with-

in the host sandstone. The pattern of the calcite ce-

ment is typical of that in sandstones (Walderhaug

and Bjorkum, 1992; McBride, 1996; Klein et al., 1999;

Dutton et al., 2000). The concretions are apparently

more abundant, more rounded, and larger in the chan-

nels than in the bars. The larger concretions in both

bars and channels possibly contain multiple nucle-

ation centers (Dutton et al., 2002). The concretions in

general do not always follow the stratigraphic geom-

etries; they terminate within a bed or cut across bed-

ding surfaces.

Differences in concretion sizes exist between the

outcrop and the cores. The thickness of the concre-

tions in the cliff face ranges from 0.2 to 1.7 m (0.66 to

5.57 ft), and they attain a maximum length of about

10 m (33 ft); the thicknesses in the cores range from

0.02 to 1.22 m (0.065 to 4.00 ft). These two types of

measurements also show inconsistency in their aver-

age occurrence of concretions. In the outcrop, the bars

contain less cement (5.5% in area) than the channel

deposits (9.7% in area); in contrast, in the cores, ce-

ment is more abundant in the bar facies (14.7% in

volume) than in the channel facies (10.5% in volume).

For the prediction of concretions by calibration of GPR

attributes, we use the outcrop values as constraints in

size and shape because the distance between individ-

ual one-dimensional cores are longer than the maxi-

mum length of the concretions; otherwise, the predic-

tion will result in overestimation of the concretion

dimensions (i.e., size and shape). The distributed cores,

however, are probably better constraints in estimat-

ing the percent volume of the concretions because

they provide broader spatial sampling. The differences

in concretion fractions between the cliff face and the

cores may be affected by weathering; neither are truly

3-D.

Petrophysical Analysis

The Raptor Ridge sandstones that contain the calcite-

cemented concretions are feldspathic litharenites to

lithic arkoses with an average composition of quartz

(51%), feldspar (21%), and rock fragments (28%). The

cemented regions have no distinct difference in grain

composition from the rest of the sandstone. Concretions

in the outcrop are entirely cemented by calcite (with

minor chlorite and kaolinite). The grain texture of the

cemented concretions is upper fine grained, moder-

ately sorted, and subangular to subrounded (Figure 14).

The main components of the concretions are a low-

magnesium, ferrous calcite.

Permeability profiles were constructed from Hassler

cell measurements on the cores (e.g., Figure 13) for con-

structing a 3-Dpermeability distribution for input to flow

modeling. The permeability in the bars is 325 ± 176 md;

the permeability in the channels is 234 ± 149 md. The

concretions have much lower permeability, which is

locally superimposed on the higher permeability back-

grounds in the bars and channels.

The permeability of the cemented sandstones ranges

from 0.01 to 0.1 md (e.g., Figure 13). The mouth-bar

sands show porosity ranging from less than 10 to 32%

(Figure 14a), but near-zero porositywhen calcite cements

fill the intergranular pores (Figure 14b). The petro-

physical relationships observed in the cores and thin

sections suggest that the sharply bounded concretions

Figure 14. Thin-section photomicrographs of (a) uncementedand (b) cemented sandstones (see Figure 13 for sampledepths). The uncemented sandstone (a) shows 0.0% concretion,permeability = 270.8 md, and porosity = 21.33%. The cementedsandstone (b) shows 24.0% concretion, permeability = 8.2 md,and porosity = 0.0%.

Lee et al. 207

can be good GPR reflectors because porosity contrasts

are associatedwith changes in electrical properties (Davis

and Annan, 1989).

ESTIMATED DISTRIBUTION OFCALCITE CONCRETIONS

Predicting the occurrence and distribution of concre-

tions in the sandstone reservoirs is an important factor

in predicting porosity and permeability.We use twoGPR

attributes (instantaneous amplitude and instantaneous

wave number) to separate the concretions from the other

background lithologies (i.e., sandstone) within the 3-D

GPRvolumebetween theTESand themud layer. For the

estimation, cores 8 and 9 were used to define the depths,

thicknesses, and boundary sharpness of the concretions.

Ground-Penetrating Radar Attribute Analysis

The procedure consists of five steps (see Corbeanu et al.,

2002; Hammon et al., 2002, for discussion): (1) GPR

traces in the vicinity of each hole are extracted; (2) cor-

relation of these traces with the lithology of the cores

is used to identify depth windows for reflections corre-

sponding to concretions; (3) two GPR attributes, instan-

taneous amplitude (IA) and instantaneous wave number

(IW) (which is the instantaneous spatial frequency),

are crossplotted to quantify the behaviors of the con-

cretions relative to the background sediments (Figure 15);

(4) IA and IW are calibrated by identifying clusters cor-

responding to the concretions (separately for the bars

and the channels); and (5) the criteria are applied to esti-

mate the distribution of the concretions throughout the

GPR volume. Themost important step is the calibration.

Figure 15. Crossplots of the GPR attributes on which the criteria for the concretions were built. The boxes in (a) indicate the criteriathat are defined and applied separately to the bar and the channel parts of the GPR data volume. The black dashed lines in (a) aresecond-order polynomial trends of the concretions. The concretions are superimposed on the background data points (b). The thindashed line in (b) represents a separation between the bar and channel facies. C1W8 = top concretion in core 8; C1W9 = topconcretion in core 9; C2W8 = middle concretion in core 8; C3W8 = lowest concretion in core 8; C2W9 = middle concretion in core 9.See Figure 16 for location of the real and predicted concretions at the cores.

208 Geohorizons

Figure 15a shows the IA-IW trajectories of the

five main concretions identified in cores 8 and 9. The

reflection associated with each individual concretion has

its maximum amplitude at approximately 50 MHz (the

nominal dominant frequency of the GPR data) and then

has a well-defined trend of progressively lower ampli-

tude toward the higher frequencies (see the dashed black

lines). The most salient prediction of each concretion is

the group of points at the 50-MHz end of its trajectory,

where it is flattened, and at higher amplitude. This sub-

set of points also has minimal overlaps with the (more

evenly distributed) background points (Figure 15b).

Assuming that the five concretions in Figure 15a

are characteristic of those throughout the volume (in

terms of thickness and degree of cementation, which

determine the reflection amplitudes), it is sufficient to

use the boxes in Figure 15a as the criteria for identi-

fying concretions. It is not necessary to use the higher

frequency tails in the criteria because all the concre-

tions produce reflections with higher amplitudes at

frequency near 50 MHz (so none will be missed in the

predictions), and the possibility of overprediction will

be minimized (Figure 16). The tails do not add any

new information because they are redundant with the

clusters at approximately 50 MHz.

Figure 15b shows the IA-IW point distributions

for the concretions and the background points in both

the bars and the channels. The sloping dashed line gives

an almost complete separation between the bar and

channel background values (the ‘‘6’’ and ‘‘+’’ loca-

tions, respectively). This also shows that it is very dif-

ficult to separate concretions from the backgrounds

if the IW windows described above were not used, but

it is straightforward if they are.

Distribution of Predicted Concretions

A reasonable match exists between the predicted bod-

ies and the corresponding concretions in the cores

(Figure 16); concretions that are only lightly cemented

Figure 16. Comparisonbetween concretions pre-dicted from the 3-D GPRdata (the white and yellowfeatures) and those ob-served in the two cores8 (a)and 9 (b). All the largerpredicted concretionsmatch with the real con-cretions (the red zones)observed in the two cores.Predicted concretions areplotted only at one sideof the bodies for visibil-ity. The thin concretionsnear the bottom of core 9are not predicted becauseit is below the resolutionof the GPR data.

Lee et al. 209

and/or are thinner than the GPR resolution are not ex-

pected to be detected. Reliability of predictions is higher

in the bars because there is little overlap between the

cemented and uncemented sandstones (Figure 15b).

Overlap leads to prediction of concretions where there

are none; conversely, concretion data points that lie

outside the selected concretion bounds lead to non-

prediction of concretions where there are some. Thus,

the observed and predicted concretions at the core lo-

cations (Figure 16) match well if we simply count the

number of predicted concretions, but not so well if the

concretion width and locations are considered in de-

tail. The separation of cemented and uncemented re-

gions in the IA-IW plane (Figure 15b) in the channels

is not as good as the bars, so there is a higher uncer-

tainty associated with the corresponding individual pre-

dictions (although the volume percent is forced to be

accurate via the calibration to the cores).

The 180 and 100 largest predicted concretions, in

the bar and channel facies, respectively, are incorpo-

rated into the 3-D geological solid model (Figure 17).

These numbers of predicted bodies were chosen by

matching the predicted volume percentage of concre-

tions (14.7% and 10.2% in the bars and channels, respec-

tively) with the observed volumes (14.7% and 10.5%, re-

spectively). This is further supported by two independent

observations: first, the predicted distributions of concre-

tions confined to the middle of the two-facies model

(Figure 17) are very similar, and second, both of these sets

of predicted concretions have a lower bound in volume

of about 130model grid points (0.26m3; 9.2 ft3), which

are also (fortuitously) approximately the lower bound

of resolution of the GPR data. Any smaller concretions

that are present would have no detectable influence on

fluid flow through themodel, so not being able to detect

them has no net detrimental effect.

Figure 18 contains two additional types of displays

to provide more complete information on the 3-D con-

cretion distribution and geometry. Figure 18a and b show

the predicted concretion distributions on the same two

planar slices on which the amplitudes are shown in

Figure 10 (in the bars and channels, respectively). The

bar concretions (Figure 18a) consist of roughly east-

west–trending linear features on the planes and approxi-

mately follow the larger amplitudes on the same plane

(Figure 10a). The channel concretions show much more

heterogenous distributions of positions and shapes, as

do the corresponding amplitudes (Figure 10b).

Figure 18c and d contain views, from above, of the

shapes and thicknesses of a few representative predicted

concretions in (1) the bars and (2) the channels. The bar

and channel concretion bodies attain maximum thick-

nesses of 1.0 and 1.4 m (3.3 and 4.6 ft), respectively

(Figure 18c, d), which are consistent with the observed

thicknesses both in the cores and the outcrops. In the

bars, the predicted concretions have relatively uniform

Figure 17. Predicted concretions both in the dip and strike directions are incorporated into a 3-D geological solid model. Thearrows at the edge of the model indicate the locations of the planes in Figure 18a and b.

210 Geohorizons

Figure 18. Slices showing thepredicted concretions (a) in thebars on a plane at about 3 m(10 ft) depth below and parallelto the topographic surface and(b) in the channels on a planeat about 8 m (26 ft) below andparallel to the topographic sur-face. Transparent views of fiverepresentative (the 2nd, 4th, 6th,8th, and 20th largest) concre-tions lying (c) in the bars and(d) in the channels. The slicesin (a) and (b) are at the samelocations as those in Figure 10.

Lee et al. 211

thickness and tend to be elongated in the strike (east-

west) direction. In the channels, the predicted concre-

tions are more equidimensional, with more rapid varia-

tion in thickness. These differences are consistent with

the expected spatial variations in porosity and perme-

ability in the bar and channel facies (and, thus, the

concretion-filling patterns). It is clear, from compar-

ing Figure 18a and b with Figure 18c and d, that 2-D

exposures (whether model slices or outcrop) underes-

timate the size, and have almost no information on the

shape, of 3-D concretions.

The modeled concretions appear to be aggregately

distributed in the bar and channel facies (Figure 18).

This distribution pattern is typical of calcite concretions

in shallow-marine sandstones (Bjorkum and Walder-

haug, 1990a). The geometry and spatial distribution

of predicted concretions are consistent with a detailed

study of diagenesis at the Raptor Ridge. The growth

patterns of the Raptor Ridge concretions do not always

follow the traditional concentric center-to-margin pat-

tern; they typically show in a complicated aggregate

pattern related to the direction of fluid flow. Similar

results were found in theMcBride et al. (2003) study of

concretions in the Ferron, Frontier, and Second Frontier

sandstone units of Utah and Wyoming.

Although calcite cements are commonly thought

to be pervasive permeability barriers (Bjorkum and

Walderhaug, 1990b; Walderhaug and Bjorkum, 1992),

we find an irregular concretion distribution that would

be a baffle to fluid flow. Geochemical analysis sug-

gests that the concretions nucleated from carbona-

ceous and calcareousmud clasts, which originated from

d18O-rich, marine pore waters, precipitating from shal-

low (�100 m; �330 ft) to deeper (�900 m; �2952 ft)

parts of the delta front, reaching a maximum tem-

perature of approximately 40j. The major growth of

the calcareous concretions occurred between 400- and

800-m (1312- and 2624-ft) depths, filling up the deltaic

sandstone volume.

A 3-D reservoir fluid simulation based on geostatis-

tically derived Raptor models (Tang, 2005) concludes

that inclusion of the GPR information increases the

estimated average of geometrical permeability by 7.04%,

resulting in an enhanced sweep efficiency from 0.79 to

0.81, compared to the models based on the core and

outcrop data alone.

The lithofacies solid model (Figure 17) is a proxy

for the corresponding deterministic permeability vol-

ume. From the measurements cited above, we can di-

rectly substitute the permeability values into the cor-

responding part of the solid model; 325 ± 176 md in

the bars, 234 ± 149 md in the channels, and 0.01–

0.1 md for the concretions. This permeability volume

can be input to flow modeling (Tang, 2005). An addi-

tional contribution to permeability, which may domi-

nate at large scale, is the near-vertical fracturing that

is seen in the outcrop; thus, the permeabilities esti-

mated here for the bars and channelsmay be considered

as lower bounds. These fractures are not imaged by the

(nearly vertically propagating) GPR signals and, thus,

must be estimated independently. Only the lithologic

contribution to permeability has been studied here.

IMPLICATIONS FOR PETROLEUM PRODUCTION

The results of this characterization of a delta-front res-

ervoir analog are useful to petroleum geologists and

engineers seeking to increase recovery from reservoirs

in similar environments (i.e., theNorth Slope ofAlaska,

the Gulf ofMexico, and the RockyMountains). The 3-D

GPR-derived detailed internal geometry within a sand-

stone body contributes to estimation of reservoir vol-

umes, and the predicted concretions embedded in the

GPR stratigraphic model may provide improved pro-

duction performance prediction via upscaled reservoir

models. In particular, the study of the delta-front reser-

voir model will be immediately applicable to improving

production because oil and gas fields are located in the

Wall Creek and other Frontier sandstones included in

the National Petroleum Reserve (NPR-3) to the east of

the study area.

SUMMARY

Ground-penetrating radar has been used to image the

detailed internal structures of a top-truncated delta

front in the topmost parasequence of the Wall Creek

Sandstone Member. The Raptor Ridge site was chosen

because of its favorable conditions for GPR data acqui-

sition. TheGPR reflection data were acquired, together

with GPS data, over a grid of 30� 80 m (100� 262 ft)

at a frequency of 50 MHz in the summer of 2002. The

outcrop was mapped and photographed for calibration

and correlation. Ten cores were drilled both inside and

outside the 3-D survey grid.

The GPR data were processed with the GPS data

and interpreted by correlation with bedding surfaces

observed in the cliff face and in the two cores within

the grid. From the 3-DGPR data, GPR boundaries and

associated radar units are recognized. The GPR units

212 Geohorizons

are characterized by two main GPR facies: RF1 and

RF2. RF1 is interpreted as tidally reworked mouth-bar

deposits, whereas RF2 is interpreted as fluvially domi-

nated laterally migrating channel fills. The deltaic se-

quence was developed by two main types of depo-

sitional episodes (channelization and bar growth). The

3-D GPR interpretation shows that the sequence is

composed of three GPR units (GUs) including nine

facies architectural elements.

Three-dimensional visualization of the architec-

tural units indicates that the deltaic sequence begins

with the sand-bar deposits (GU1) prograding in the

same direction as the paleocurrent indicators. The dis-

tal delta-front deposits were scoured by the subsequent

four channels (GU2). After channel migration, prox-

imal sand bars are deposited on the channel fills. The

uppermost GPR unit (GU3) starts with northwest-

southeast–elongated mouth bars formed by unidirec-

tional flow. As time went on, the prograding sand bars

were progressively more influenced by tidal modula-

tion, producing the bidirectional-flow deposits at the

delta front, until the mouth bars were truncated by

subsequent transgressive ravinement.

The volume and distribution of calcite cements,

the main diagenetic features in the Raptor Ridge sand-

stones, are estimated by a core-constrained cluster analy-

sis of GPR attributes (instantaneous amplitudes and

wave numbers). The GPR-derived criteria (concretion

clusters) for separation of cemented and noncemented

regions were determined and applied separately to the

bars and channels. The predicted concretions in the del-

taic sandstone are distributed in a dendritic, aggregate

manner and account for 14.7% of the bar volume and

10.2% of the channel volume. The overall estimated con-

cretion pattern matches reasonably with that observed in

the cliff face particularly in terms of shape and thickness.

The predicted spatial geometries are consistent with the

difference in porosity and permeability distributions be-

tween the bars and the channels. The formation of the

concretions do not always proceed from the centers to

the edges. The modeled near-zero permeability baffles

are incorporated into a 3-D digital geological solid model,

which has been used as a framework for subsequent

reservoir fluid-flow modeling by Tang (2005).

REFERENCES CITED

Barlow, J. A., and J. D. Haun, 1966, Regional stratigraphy of Fron-tier Formation and relation to Salt Creek field, Wyoming: AAPGBulletin, v. 50, p. 2185–2196.

Bhattacharya, J. P., and R. G., Walker, 1992, Deltas, in R. G.WalkerandN. P. James, eds., Facies models: Response to sea level change:St. John’s, Newfoundland, Geological Association of Canada,p. 157–177.

Bhattacharya, J. P., and B. J. Willis, 2001, Lowstand deltas in theFrontier Formation, Powder River Basin, Wyoming: Implica-tions for sequence stratigraphic models: AAPG Bulletin, v. 85,p. 261–294.

Bjorkum, P. A., and O. Walderhaug, 1990a, Geometrical arrange-ment of calcite cementation within shallow marine sandstones:Earth-Science Reviews, v. 29, p. 145–161.

Bjorkum, P. A., and O. Walderhaug, 1990b, Lateral extent ofcalcite cemented zones in shallow marine sandstones, in A. T.Buller, E. Berg, O. Hjelmeland, J. Kleppe, O. Torsaeter, andJ. O. Aasen, eds., North Sea oil and gas reservoirs II: London,Graham and Trotman, p. 331–336.

Brown, A. R., 2004, Interpretation of three-dimensional seismicdata: AAPG Memoir 42, SEG Investigations in GeophysicsNo. 9, 541 p.

Corbeanu, R. M., K. Soegaard, R. B. Szerbiak, J. B. Thurmond,G. A. McMechan, D. Wang, S. Snelgrove, C. B. Forster, andA. Menitove, 2001, Detailed internal architecture of a fluvialchannel sandstone determined from outcrop, cores, and 3-Dground-penetrating radar: Example from the middle CretaceousFerron Sandstone, east-central Utah: AAPG Bulletin, v. 85,p. 1583–1608.

Corbeanu, R. M., G. A. McMechan, R. B. Szerbiak, and K. Soegaard,2002, Prediction of 3-D fluid permeability and mudstone dis-tributions from ground-penetrating radar (GPR) attributes: Ex-ample from the Cretaceous Ferron Sandstone Member, east-central Utah: Geophysics, v. 67, p. 1495–1504.

Cordsen, A., 2004, Acquisition footprint can confuse: AAPG Ex-plorer, v. 25, p. 26–27.

Davis, J. L., and A. P. Annan, 1989, Ground penetrating radar forsoil and rock stratigraphy:Geophysical Prospecting, v. 37, p. 531–551.

Dutton, S. P., B. J. Willis, C. D. White, and J. P. Bhattacharya, 2000,Outcrop characterization of reservoir quality and interwell-scalecement distribution in a tide-influenced delta, Frontier Forma-tion, Wyoming, U.S.A.: Clay Minerals, v. 35, p. 95–105.

Dutton, S. P., C. D. White, B. J. Willis, and D. Novakovic, 2002,Calcite cement distribution and its effect on fluid flow in adeltaic sandstone, Frontier Formation, Wyoming: AAPG Bul-letin, v. 86, p. 2007–2021.

Dyman, T. S., E. A. Merewether, C. M. Molenaar, W. A. Cobban,J. D. Obradovich, R. J. Weimer, andW. A. Bryant, 1994, Strati-graphic transects for Cretaceous rocks, Rocky Mountains andGreat Plains regions, in M. V. Caputo, J. A. Peterson, and K. J.Franczyk, eds., Mesozoic systems of the RockyMountain region,U.S.A.: SEPM Rocky Mountain Section, p. 365–392.

Epili, D., andG. A.McMechan, 1996, Implementation of 3-D prestackKirchhoff migration, with application to data from the Ouachitafrontal thrust zone: Geophysics, v. 61, no. 5, p. 1400–1411.

Gani, M. R., 2005, 3-D facies architecture of river deltas: Lessonsfrom the Turonian Wall Creek Member, central Wyoming,U.S.A.: Ph.D. dissertation, University of Texas at Dallas, Dallas,Texas, 148 p.

Hammon, S. W., X. Zeng, R. M. Corbeanu, and G. A. McMechan,2002, Estimation of the spatial distribution of fluid perme-ability from surface and tomographic GPR data and core, witha 2-D example from the Ferron Sandstone, Utah: Geophysics,v. 67, p. 1505–1515.

Hassouta, L., M. D. Buatier, J. L. Potdevin, and N. Liewig, 1999,Clay diagenesis in the sandstone reservoir of the Ellon field(Alwyn, North Sea): Clays and Clay Minerals, v. 47, p. 269–285.

Lee et al. 213

Howell, C. D., J. P. Bhattacharya, A. B. Robinson, and W. R. Griffin,2003, Topographically controlled, mixed-influence top-truncateddeltas: Turonian Wall Creek Member, Frontier Formation, Pow-der River Basin, Wyoming, U.S.A (abs.): AAPG Annual MeetingProgram, v. 12, p. A81.

Jol, H. M., and C. S. Bristow, 2003, GPR in sediments: Adviceon data collection, basic processing and interpretation, a goodpractice guide, in C. S. Bristow and H. M. Jol, eds., Groundpenetrating radar in sediments: Geological Society (London)Special Publication 211, p. 9–27.

Jol, H. M., and D. G. Smith, 1991, Ground penetrating radar ofnorthern lacustrine deltas: Canadian Journal of Earth Sciences,v. 28, p. 1939–1947.

Klein, J. S., P. Mozley, A. Campbell, and R. Cole, 1999, Spatialdistribution of carbon and oxygen isotopes in a laterally exten-sive carbonate-cemented layers: Implications for mode of growthand subsurface identification: Journal of Sedimentary Research,v. 69, p. 184–191.

Lee, K., X. Zeng, G. A. McMechan, C. D. Howell, Jr., J. P.Bhattacharya, F. Marcy, and C. Olariu, 2005, A GPR survey ofa delta-front reservoir analog in the Wall Creek Member,Frontier Formation, Wyoming: AAPG Bulletin, v. 89, p. 1139–1155.

Lemke, S. R., and L. C. Mankowski, 2000, GPR attribute analysis:Proceedings of the Symposium on the Application of Geo-physics to Engineering and Environmental Problems, Environ-mental and Engineering Geophysical Society, Wheat Ridge,Colorado, Arlington, Virginia, February 20–24, 2000, p. 263–272.

McBride, E. F., 1996, How predictable are patterns of calcite ce-ment in sandstones? (abs.): AAPG Annual Meeting Program,v. 5, p. A79.

McBride, E. F., M. D. Picard, and K. L. Milliken, 2003, Calcite-cemented concretions in Cretaceous sandstone, Wyoming andUtah, U.S.A: Journal of Sedimentary Research, v. 73, p. 462–483.

McMechan, A. G., G. C. Gaynor, and R. B. Szerbiak, 1997, Use ofground-penetrating radar for 3-D sedimentological character-ization of clastic reservoir analogs: Geophysics, v. 62, p. 786–796.

Pepola, C. P., and E. J. Hickin, 2003, Long-term bed load transportrate based on aerial-photo and ground penetrating radar sur-veys of fan-delta growth, Coast Mountains, British Columbia:Geomorphology, v. 57, p. 169–181.

Roberts, M. C., H. P. Niller, and N. Helmstetter, 2003, Sedimentaryarchitecture and radar facies of a fan delta, Cypress Creek,WestVancouver, British Columbia, in C. S. Bristow and H. M. Jol,

eds., Ground penetrating radar in sediments: Geological Society(London) Special Publication 211, p. 111–126.

Siegenthaler, C., 1982, Tidal cross-strata and the sediment transportrate problem: A geologist’s approach, Marine Geology, v. 45,p. 227–240.

Snelgrove, S. H., C. B. Forster, A. Mattson, and M. A. Chan, 1996,Impact of lithofacies architecture and distribution on fluid flow:Examples from the Cretaceous Ferron Sandstone, east-centralUtah (abs.): AAPG Annual Meeting Program, v. 5, p. A132.

Snelgrove, S. H., A. Mattson, C. B. Forster, M. A. Chan, and P. B.Anderson, 1998, From outcrop to simulation in the Ferron Sand-stone; integrated characterization of a fluvial-dominated delta-front reservoir analog-integrated characterization of a fluvial-dominated delta-front reservoir analog (abs.): AAPG AnnualConvention Program, v. 7, p. A616.

Tang, H., 2005, Geostatistical integration of geophysical, well bore,and outcrop data for flow modeling of a deltaic reservoir analog:Ph.D. dissertation, Louisiana State University, Baton Rouge,Louisiana, 131 p.

van Heerden, I. L., and H. H. Roberts, 1988, Facies development ofAtchafalaya delta, Louisiana: A modern bayhead delta: AAPGBulletin, v. 72, no. 4, p. 439–453.

Walderhaug, O., and P. A. Bjorkum, 1992, Effect of meteoric waterflow on calcite cementation in the Middle Jurassic OsebergFormation, well 30/3-2, Veslefrikk field, Norwegian North Sea:Marine and Petroleum Geology, v. 9, p. 308–318.

White, C. D., D. Novakovic, S. P. Dutton, and B. J. Willis, 2003,A geostatistical model for calcite concretions in sandstone: Math-ematical Geology, v. 35, p. 549–575.

Willis, B. J., 2005, Deposits of tide-influenced river deltas, in L.Giosan and J. P. Bhattacharya, eds., Deltas concepts, models,and examples: SEPM Special Publication 83, p. 87–129.

Willis, B. J., J. P. Bhattacharya, S. L. Gabel, and C. D. White, 1999,Architecture of a tide-influenced river delta in the FrontierFormation of central Wyoming, U.S.A.: Sedimentology, v. 46,p. 667–688.

Winn, R. D., 1991, Storm deposition in marine sand sheets: WallCreekMember, Frontier Formation, Powder River Basin, Wyo-ming: Journal of Sedimentary Petrology, v. 61, p. 86–101.

WyGISC (Wyoming Geographic Information Center), 1990, Oiland gas potential: http://www.sdvc.uwyo.edu/clearinghouse/oilshare.html (accessed March, 2004).

Zeng, X., G. A. McMechan, J. P. Bhattacharya, C. L. V. Aiken,X. Xu, W. S. Hammon, III, and R. M. Corbeanu, 2004, 3-Dimaging of a reservoir analogue inpoint bar deposits in the FerronSandstone, Utah, using ground-penetrating radar: GeophysicalProspecting, v. 52, p. 151–163.

214 Geohorizons