Characterization of a coalesced, collapsed paleocave ...

School of Natural Sciences and MathematicsCenter for Lithospheric Studies

Characterization of a Coalesced, Collapsed PaleocaveReservoir Analog Using GPR and Well-Core Data

UT Dallas Author(s):

George A. McMechanXiaoxian Zeng

Rights:

©2016

Citation:

McMechan, G. A., R. G. Loucks, P. Mescher, and X. Zeng. 2017."Characterization of a coalesced, collapsed paleocave reservoir analog usingGPR and well-core data." Geophysics 67(4): 1148-1158.

This document is being made freely available by the Eugene McDermott Libraryof the University of Texas at Dallas with permission of the copyright owner. Allrights are reserved under United States copyright law unless specified otherwise.

GEOPHYSICS, VOL. 67, NO. 4 (JULY-AUGUST 2002); P. 1148–1158, 11 FIGS.10.1190/1.1500376

Characterization of a coalesced, collapsed paleocave reservoiranalog using GPR and well-core data

George A. McMechan∗, Robert G. Loucks‡, Paul Mescher∗∗, and Xiaoxian Zeng∗

ABSTRACT

The three-dimensional architecture, spatial complex-ity, and pore-type distribution are mapped in a near-surface analog of a coalesced, collapsed paleocavesystem in the Lower Ordovician Ellenburger Group nearthe city of Marble Falls in central Texas. The surface areaof the site has dimensions of about 350× 1000 m. Thedata collected include about 12 km of 50-MHz ground-penetrating radar (GPR) data arranged in a grid oforthogonal lines, 29 cores of about 15-m length, anddetailed facies maps of an adjacent quarry face. Elec-trical property measurements along with detailed coredescriptions were the basis of integrated interpretationof the GPR data. Three main GPR facies are defined onthe basis of degree of brecciation in the correspondingcores: undisturbed host rock, disturbed host rock, andpaleocave breccia. This GPR facies division defined themajor paleocave trends and the distribution of porositytypes, which correlate with reservoir quality. Highly brec-ciated zones are separated by disturbed and undisturbedhost rock. The breccia bodies that outline the trend ofcollapsed cave passages are up to 300 m wide; the in-tervening intact areas between breccias are up to 200 mwide. Understanding the breccia distribution in a reser-voir analog will help in defining strategies for efficient de-velopment of coalesced, collapsed paleocave reservoirs.

INTRODUCTION

Ground-penetrating radar (GPR) data have previously beenused for characterizing both ancient and modern clastic rocks inthree dimensions (Baker, 1991; Jol and Smith, 1991; Gawthorpeet al., 1993; Beres et al., 1995; Bridge et al., 1995; McMechan

Manuscript received by the Editor May 26, 2000; revised manuscript received October 29, 2001.∗University of Texas at Dallas, Center for Lithospheric Studies, Post Office Box 830688, Richardson, Texas 75083-0688. E-mail: [email protected];[email protected].‡University of Texas at Austin, Bureau of Economic Geology, Austin, Texas 78713-8924. E-mail: [email protected].∗∗Geological Resources Company, 3131 Custer Road, Suite 175, Post Office Box 172, Plano, Texas 75075.c© 2002 Society of Exploration Geophysicists. All rights reserved.

et al., 1997; Corbeanu et al., 2001; Szerbiak et al., 2001).Similar work in carbonates is rare, but there are examples ofsuccessful GPR data acquisition in carbonate environments(Beck and Wilson, 1988; Cai and McMechan, 1995; Casaset al., 1996; Zolotarev et al., 1996; McMechan et al., 1998;Sigurdsson and Overgaard, 1998; Dagallier et al., 2000; Pipanet al., 2000). Thus, it makes sense to determine the feasibilityof using GPR data for characterization of carbonate reservoiranalogs at the scale of interwell spacing. This paper reports asuccessful example.

Paleocave systems form an important class of carbonatereservoirs that are products of near-surface karst processes andlater burial compaction and diagenesis (Loucks and Anderson,1985; Kerans, 1988; Candelaria and Reed, 1992; Holtz andKerans, 1992; Loucks and Handford, 1992; Stoudt, 1996;Loucks, 1999). Paleocave reservoirs tend to have complex his-tories of formation and pronounced lateral and vertical hetero-geneity, especially when multiple phases of karsting and burialare involved (Loucks and Mescher, 1996, 1997; McMechanet al., 1998; Loucks, 1999).

The field site is at the Dean Word Quarry, east of U.S. High-way 281 between Burnet and Marble Falls in central Texas(Figure 1). The outcrop at the site is Lower Ordoviciandolomite of the Ellenberger Group, which is a major oil reser-voir in west Texas. This site was chosen because it has a rela-tively planar topographic surface with little soil or vegetation,has a fresh quarry wall exposed at its southern edge (Figure 2),and previously yielded good quality GPR data in a pilot study ofportions of the same collapsed paleocave system (McMechanet al., 1998). At the study site (Figure 1), two main phasesof burial and cave collapse are interpreted to have occurredduring the Mississippian-Pennsylvanian and Early Cretaceous(Loucks and Mescher, 1996), with uplift, erosion, and karstingin between. This burial history is consistent with that presentedby Kupecz and Land (1991), and with observations from bothcores and quarry faces at the site.

1148

Paleocave Reservoir Analog Architecture 1149

Most paleocave reservoirs are not the result of isolated col-lapsed cave passages (each only a few meters in diameter, andhundreds of meters long); rather, they contain coalesced, col-lapsed cave systems (that can be many hundreds of metersacross and thousands of meters long). Details of this modelare presented by Loucks (1999) and Loucks et al. (1999).The study reported here is an implicit test of Loucks’ model,and reveals geometries that are consistent with the modelpredictions.

The study of the internal 3-D distribution of in-situ reservoirproperties is fundamentally limited by the lack of availabilityof 3-D data. Even analogs consisting of reservoir environmentsthat crop out at the surface provide only 2-D exposures of 3-Dstructures (Miall and Tyler, 1991; Tyler et al., 1992; Flint andBryant, 1993). Thus, any method that can view the third dimen-sion is of great practical importance. In the multidisciplinaryproject described below, we obtained a large, high-quality database from a carbonate reservoir analog. The data are compre-hensive and consist of about 12 km of GPR data acquired onan orthogonal grid of 2-D lines, 29 cores (each of about 15-mlength), electrical and petrophysical property measurementson 44 samples extracted from the cores, petrographic and min-eralogic analyses of thin sections, palynology and conodontbiostratigraphy, whole-core porosity, permeability and grain-density measurements, cave-facies logs from the cores, and de-tailed visual logging of the adjacent quarry face. The presentpaper focuses on the acquisition, processing, and analysis ofthe GPR and electrical property data. The data are interpreteddata in terms of the 3-D distribution of three reservoir facies.Other aspects of the project are presented by Loucks et al.(1999, 2001) and Loucks and Mescher (2001).

The main constraints on the interpreted 3-D facies distri-butions are the 3-D GPR data and the core descriptions. TheGPR data were acquired first and gave an initial look at the 3-D

FIG. 1. Study location in central Texas. The star is the quarrysite.

distribution of undisturbed host rock, disturbed host rock, andpaleocave breccia, which guided selection of the core locations.The cores provided explicit ties between the GPR facies andthe carbonate facies at the core locations. Finally, the GPR dataprovided the mechanism for interpolation between the holes,resulting in a refined analysis of the facies distributions. Themain constraints on the petrophysical properties and reservoirquality within the facies are the measurements from the coresamples.

ACQUISITION OF GPR DATA

In recent years, GPR technology has matured to the pointwhere its applicability and limitations are fairly well known.Good introductions to and summaries of the method are givenby Daniels et al. (1988), Davis and Annan (1989), Daniels(1996), and Reynolds (1998, 681–744). GPR is a high-frequency(0.1–1.5 GHz) electromagnetic method that produces high-resolution images of shallow earth structures in environmentsin which the electrical conductivity is sufficiently low. The dataare acquired in the form of a digital trace at each of a seriesof closely spaced points along a survey line. A GPR sectionappears similar (except for scale) to a seismic section, andcontains reflections and diffractions from spatial variations inelectromagnetic properties that correlate primarily with lithol-ogy, fabric, and structure. Other factors that can influence thequality and character of GPR data include water content andnongeologic reflectors such as metal objects (above as well asbelow ground); examples are shown below in the section enti-tled “Pitfalls and nongeologic signals.” In this study, the GPRcharacteristics are interpreted by detailed comparisons of theGPR facies with the paleocave facies seen in the cores andquarry faces.

Grid control positions were surveyed every 75 m on bothnorth-south and east-west lines (Figure 2). To ensure accuracyand repeatability, surveying was performed using Global Posi-tioning System equipment (a Lieca SR-399 real-time kinematicsystem). The measurements were differential, relative to a basestation; repeatability was approximately 1 cm in the horizontalcoordinates.

Constant-offset GPR data were acquired along each of thenorth-south and west-east grid lines, except for short segmentsof lines B, 7, and 8, which went over a large pile of quarryrock debris. The GPR equipment used was two PulseEKKOmodel IV systems with 1000-V transmitters, manufactured bySensors & Software, Inc. The transmitter-to-receiver antennaseparation was 3 m, the station spacing along the grid lineswas 0.2 m, the dominant frequency was 50 MHz, and the to-tal recording time window was 500 ns at a sample interval of1600 ps. These parameters were determined by a series of testsprior to the main acquisition. Summing over all the lines givesa total of about 12.5 km of line and about 62,400 traces. Linelocations and GPR station spacing were controlled by stretch-ing survey tapes between stakes at the survey control points.Each 75-m line segment was acquired digitally and saved asa separate file. Two crews operated simultaneously to reducethe total survey time by half. Minimizing the total field timealso reduces variations in data acquisition conditions and sopromotes internal consistency. Quality control was maintainedby real-time display of the data as it was acquired.

1150 McMechan et al.

Data processing included subtraction of the exponentialbaseline signal [caused by diffusion in the near-surface rocks(see Ward and Hohmann, 1987)] from each trace, time zero cor-rections, subtraction of direct air and ground waves, trace edit-ing, filtering, concatenation of data for line segments to formcomplete survey lines, and plotting with a modified automaticgain control (AGC). Unlike a traditional AGC, this one in-cludes a maximum gain to be applied to any sample, so itreduces, but does not eliminate, the differences between thehigh- and low-amplitude regions. Migration was not done asthe structures are clearly 3-D, but we have only 2-D data. Moreimportantly, migration was not required to define the facies norto map their spatial distribution (which was the main goal ofthe project). Time-to-depth conversion used a single averagevelocity of 0.102 m/ns, which was estimated using direct waveslopes and reflection moveouts in common midpoint gathers,lab measurements (see the section entitled “Core and lab mea-surements”), and ties with depths of features observed in thecores and of the level of the water table in the floor of thequarry (at about 12–15 m below the top of the quarry face).This velocity appeared to be fairly consistent throughout the

FIG. 2. The GPR grid lines superimposed on an aerial photo of the quarry. North-south grid lines are identifiedby letters from A through L; east-west lines are identified by numbers from 1 through 13. The grid is boundedon its south edge by the quarry face, and on its northeast edge by a fence line.

survey area. The maximum depth of usable signals is limitedby the high attenuation caused by the high conductivity thatoccurs beneath the water table.

GPR DATA INTERPRETATION

Facies definitions

The GPR facies interpretation relies on the continuity, regu-larity, and strength of the GPR reflections. These, in turn, corre-late with cave facies of various scales and degrees of brecciation(Loucks and Mescher, 1996, 1997). The geologic significance ofeach of the GPR facies reflection patterns is defined by corre-lation with the core and quarry-face data (McMechan et al.,1998). The 50-MHz GPR data theoretically have vertical reso-lution of about 0.5 m (one-quarter wavelength); breccia blocksless than this size are not detected as distinct features. Thus,while six paleocave-related facies were defined in the cores andquarry faces (Loucks et al., 1999), only three GPR facies aredefined: undisturbed host rock, disturbed host rock, and brec-cia. The transition between facies is sometimes gradational and


sometimes abrupt. This three-part facies division was found tobe adequate to define the major paleocave trends and porositytypes via detection and mapping of the corresponding GPRfacies.

The undisturbed host rock of GPR facies 1 has reflectionsthat are continuous and predominantly parallel (Figure 3a).In the cores (Figure 4a), the undisturbed host rock rangesfrom a non-to-slightly fractured, burrowed peloidal/skeletallime packstone or mudstone with stylolites, to a thin laminatedor bedded intraclast lime grainstone. The GPR data show re-flections from the larger scale bedding.

The disturbed host rock of GPR facies 2 has relatively con-tinuous reflections of spatially variable separation (Figure 3b).The reflections appear to show folds that are broken by smallfaults. The faults are interpreted as being caused by sagging andcollapse into underlying chambers and passages. The lithologyof facies 2 is essentially the same as that of facies 1, but gener-ally contains crackle brecciation caused by collapse processsesand that increases permeability.

The breccia of GPR facies 3 corresponds generally to datawith scales of continuity that vary with the size and shapeof the breccia clasts as well as the amount of matrix mate-rial present between the clasts (Figure 3c). Clasts range from

FIG. 3. Type patterns of GPR reflections used to define thethree GPR facies. (a) Corresponds to undisturbed host rock(GPR facies 1). The data are from a portion of line E, centeredon corehole E-5. (b) Corresponds to disturbed host rock (GPRfacies 2). The data are from a portion of line H, between core-holes H-9 and H-10. Folding and small-scale faulting is visiblein otherwise fairly continuous reflections. (c) Corresponds tobreccia (GPR facies 3). The data are from a portion of line 10,between coreholes A-10 and A-11. The maximum horizontalextent of reflection continuity is only a few meters. The relativehorizontal scale of panel (a) also applies to panels (b) and (c).

tens of meters to subcentimeter in size (Figure 4b). The cor-responding GPR responses range from discontinuous dippingreflections to incoherent fine-scale chaotic scattering. Largerslabs tend to have lesser amounts of matrix-rich material be-tween them. The high permeability of the breccia relative tothe undisturbed host rock resulted in multistage dolomitizationduring burial.

Transitional relations

Assuming that the GPR reflection interpretations in the pre-vious sections are correct, we can use the spatial distribution

FIG. 4. Representative core segments. (a) Undisturbed hostrock from a depth of 9.6–10.0 m from the core at grid loca-tion C-8. (b) A multiphase breccia from a depth of 9.1–9.4 mfrom the core at grid location D-11. Core diameter is 2.5 inches(6.35 cm).


of reflection patterns to gain insight into collapsed paleocavefeatures and the relations between the paleocave facies. Inte-gration of core data with the GPR reflection patterns guidesand substantiates the interpretations.

Figure 5a shows disturbed host rock dipping into a collapsedpaleocave passage (the chaotic reflection zone). A number offaults cut the disturbed host rock. The lack of reflections in thepaleocave brecciated zone may indicate fine clast size, and/orabundant matrix and cave sediment fill. Figure 5b shows a sharptransition from cave breccias into undisturbed host rock. Theparallel reflections of the host rock roll over into the collapsedzone. Within the collapsed area, the variation in scales of con-tinuity is probably a response to varying scales of brecciation.Figure 5c contains a transition from intact host rock to largeblocks to breccia. Figure 5d is interpreted as a pillar of dis-turbed host rock surrounded by chaotic reflections associatedwith the paleocave breccia facies. The reflections in the brec-cia show some continuity, which may indicate the presence of

FIG. 5. Boundaries between GPR facies: (a) between faulteddisturbed host rock and breccia, (b) between intact host rockand breccia, (c) between intact host rock, disturbed host rock,and breccia, and (d) an intact pillar of host rock surroundedby breccia. The location and orientation of each of the GPRlines are labeled on each panel. The relative horizontal scaleof panel (a) also applies to panels (b), (c), and (d).

large slabs. In all cases, the collapse and brecciation is impliedto result from burial of a cave network that was formed nearthe surface (Loucks, 1999).

Structural features

A wide range of structural features are associated with pa-leocaves. Small-scale folds (100 m or less in horizontal extent)and faults (with offsets of up to several meters) are apparenton many GPR lines. The lower part of Figure 6a has relativelygood continuity and is interpreted as disturbed host rock; it con-tains several faults. The upper part of the section has relativelydiscontinuous reflections and is interpreted to be composed oflarge brecciated slabs. Similar features are seen in the quarryface (Figure 6b). The sags are related to folding into collapsedintervals deeper in the section, not to regional compressionalforces. Similar collapse features have been previously docu-mented in the Ellenburger in seismic data (Hardage et al.,1996).

Several domal or anticlinal features and associated faultingare also evident. In Figures 7a and 7c, the faults show bothnormal and reverse movement, which is characteristic of sags.The structure in Figure 7a is interpreted as a composite pileof large breakdown slabs where the roof of a large cavity col-lapsed. Faults also occur in relatively flat-lying strata over col-lapsed passages. The complex graben in the disturbed host rockin Figure 7b appears to overlie a collapsed chamber in whicha large dislocated ceiling block is suggested. Figure 7c showsfaults in the disturbed host rock in the quarry face.

Pitfalls and nongeologic signals

As with seismic data, there are a number of factors that canproduce false structures or poor quality, and hence ambiguousinterpretations, in GPR data. Two of the main factors at this siteare metal objects and water. Metal objects in the area (aboveor beneath the ground surface) can produce strong reflections.A windmill between GPR grid points E-3 and F-4 produced alarge diffraction (Figure 8a) that could be misinterpreted as ananticlinal reflection.

Groundwater or moisture in the rocks also affects GPRdata. The GPR signal is diminished where the rocks are sat-urated with electrically conductive groundwater. In Figure 8b,damp soil in a surface depression locally attenuates the signalstrength. The signal reflection amplitudes on all lines decreasesharply at about 12–13 m depth, which corresponds to the localwater table. The greater depth of penetration in Figures 5b and5c is consistent with a reduced volumetric water content belowthe water table in the tight intact host rock.

Another interesting observation is shown in Figure 8c. Here,there are anomalous continuous signals in a zone (at >10-mdepth) that is known, from the cores, to be highly brecciated.This apparent bedding may be caused by bedded cave fill withinthe breccia (especially if contrasts are enhanced by moisture),subhorizontal pressure-release fractures produced during up-lift, or offline reflections. Alternatively, it may be a recording ar-tifact. Horizontal lines caused by signals reverberating betweenthe antennas are often present in GPR data if the ground sur-face is moderately conductive. These reverberations are usu-ally of low amplitude, but may be enhanced by the gain used in


plotting and may be the only visible signal if the real reflectionsare weak (as expected in a fine-scale breccia), or if the signalattenuation is high at depth.

Features at the regional (interwell) scale are best seenin GPR lines that cross large portions of the survey in thenorth-south or east-west direction. These are produced by con-catenating the 75-m long segments. The three examples inFigure 9 show both large-scale features and spatial variationsin facies.

Cores and lab measurements

The distribution of GPR facies resulting from an initial in-terpretation of the GPR lines guided the selection of core loca-tions to provide core samples of all the facies and all scales ofbreccias. Continuous 2.5-inch (6.35-cm) diameter cores about15 m in length were extracted from the 29 (white dot) grid loca-tions marked on Figure 2. Figure 4 shows examples of two smallcore segments; complete sets of core descriptions and repre-

FIG. 6. GPR data (a), and a quarry face photomosaic (b) showing sags caused by collapse of passages deeper in the section. Thescale of (b) is indicated by the person in the lower right corner of the photo. The photo (b) is of the quarry face that is along theextension of line I (Figure 2); (a) is the portion of north-south line I between east-west lines 7 and 8. A few representative faultsare marked in (a). Many others are also present.

sentative core photos are provided in Loucks et al. (1999). Thecore data, in turn, allowed refined interpretation of the GPRdata.

As part of the analysis of the cores, 44 samples were extractedfor the purpose of making frequency-dependent measurementsof the electrical properties. Sample sites were chosen to containa variety of compositions (from limestone to dolomite) and fab-rics (from homogeneous to layered to brecciated), with a rangeof porosity, permeability, and degrees of fracturing. Measure-ments of complex dielectric permittivity were made with anHP-8752A network analyzer with an HP-8750A coaxial termi-nation probe. Eight measurements were made (four on thetop and four on the bottom) of each sample and averagedto produce representative measurements. Measurements weremade at each of 101 frequencies from 20 MHz to 1.3 GHz atboth ambient saturation and after oven drying, as describedby Klein and Santamarina (1997). Results for 50 MHz alongwith the corresponding whole-core porosity and permeabilitydata at the same locations are shown in Figure 10. The GPR


propagation velocity in the samples ranges from 0.097 to0.108 m/ns, and the effective electrical conductivity ranges from0.37 to 2.48 mS/m. Porosity ranges from 1.7 to 9.7%, and per-meability ranges from 0.02 to about 1400 md. [It should benoted that the data in Figure 10 are only a subset (those forwhich the electrical properties were measured). Much morepetrophysical data were obtained and are analyzed in the con-text of reservoir quality by Loucks et al., (1999) and Loucksand Mescher (2001).]

The lab measurements are subject to scaling uncertaintieswhen using them to infer larger scale properties under in situconditions. The superposition of multiphase diagenetic andstructural effects upon the original fabric produces a highlyvariable set of properties, both petrophysical and electrical.Because the number of measurements is relatively small andtheir variance is high, we did not attenpt to fit curves or todo any statistical analysis. Some general observations may bemade, however:

1) There are trends of increasing GPR velocity with de-creasing electrical conductivity (Figure 10e), with de-creasing grain density (i.e., with decreasing dolomiti-

FIG. 7. Faults in disturbed host rock. GPR lines (a) and (b)show large-scale brecciation produced by collapse of cham-bers, and the resulting faults in the disturbed host rock above.(c) Similar faults exposed in the north quarry face. Both normaland reverse offsets of (white) chert beds within the dolomiteare clearly visible. The scale of (c) is indicated by the person inthe lower center of the photo. The location and orientation ofeach of the GPR lines are labeled on each panel. The relativehorizontal scale of panel (a) also applies to panels (b) and (c).

zation) (Figure 10b), and with decreasing permeability(Figure 10a).

2) The limestone data cluster at higher velocity (Figures 10aand 10b), lower electrical conductivity (Figures 10c and10d), lower permeability (Figure 10c), and lower poros-ity (Figure 10d) than the dolomite. The separation is notperfect because the limestone/dolomite composition is acontinuum. Thus, the limestone/dolomite ratio may beapproximately estimated from GPR data.

3) Differences in measurements between the ambient satu-ration and oven-dried samples are small because the am-bient saturation is small (about 0.1% average by weight).However, the dry samples have consistently higher ve-locity and lower conductivity than the ambient samples,so relative variations in saturation may potentially be es-timated from velocity and conductivity measurements.

4) These lab results reveal a potential for achieving alonger term goal of predicting lithology and petro-physical properties from GPR data. The new data pre-sented here contribute to the data base. However thelarge variance of the data distribution suggests that muchadditional work needs to be done.

FIG.8. Interpretation pitfalls. The large, downward curving sig-nal (at the white dashes) in (a) is a diffraction traveling in theair from a metal windmill on the ground surface, not a geo-logic feature. The low signal amplitudes in (b) correspond to ashallow sinkhole at the ground surface. Retention of water bymud in the sinkhole increases the electrical conductivity whichattenuates the signal. The anomalous (horizontal banding) sig-nals at times >200 ns in (c) may be real or may be recordingartifacts as discussed in the text. The location and orientationof each of the GPR lines are labeled on each panel. The relativehorizontal scale of panel (a) also applies to panels (b) and (c).


Integration of core and GPR data: 3-D interpretation

The integration of data from the cores and the GPR surveyallows a 3-D analysis of the study area. The cores are the bestdata for interpreting facies, but they provide only limited refer-ence points. Large breccia features and associated structures,such as faults and sags, are difficult to define from cores. TheGPR data are good for correlating reflection patterns that canbe used to delineate structures and the main facies distribu-tions, and for recognizing facies changes, but only generaliza-tions can be made about rock types and reservoir potential.Thus, integration of the cores and GPR data provides a morerobust interpretation of facies types and their distribution; theirstrengths are complementary.

Overall, there is a good agreement between the paleocavefacies seen in the core and the three paleocave facies inter-preted from the GPR data (see also Loucks et al., 1999). Thebest agreement is within the undisturbed host rock where GPRreflections are strong, parallel, and continuous. The agreementin the disturbed host rock is also generally good. Where thedisturbed host rock grades into collapsed ceiling and wall rockbreccia, it gradually becomes more difficult to distinguish thescale of brecciation from the GPR data (which is why all brec-cias were treated as a single GPR facies). Individual brecciablocks cannot be differentiated below the resolution of the50-MHz GPR waves (one-quarter wavelength is about 0.5 m),but they can still contribute to the overall scattered wave pat-tern in the breccia. Thus, although the internal details of the

FIG. 9. Three 2-D GPR lines crossing large portions of the survey area. (a) Facies transitions in north-south line F. (b) Sags innorth-south line I. (c) A complicated structure/facies interaction in west-east line 3. The relative horizontal scale of panel (a) alsoapplies to panels (b) and (c).

finer breccias are not resolvable from the GPR data, the over-all geometry of the distributions of the key facies for reservoirdescription is well defined. Adding the detailed informationfrom the cores provides a continuum of properties from kilo-meter to millimeter scales. [Analyses at the finer scales arepresented by Loucks et al. (1999) and by Loucks and Mescher(2001)].

By correlating the interpreted facies along the GPR lines,constrained by the core and quarry face data, and then in-terpolating between the GPR lines, a pseudo 3-D image of the3-D geometry and spatial distribution of the three facies is pro-duced. Figure 11 shows the interpreted horizontal distributionof the dominant facies in three depth windows. Each of thesemaps shows a similar series of paleocave breccias trending tothe northeast. The breccia zones are separated by disturbedand undisturbed host rock. The orientation of the paleobrecciabodies does not correspond to the present-day fracture orien-tations (which are north-northwest and west-northwest). Thus,if a fracture system controlled Ellenburger cave development,then that (northeast-striking) fracture system had a differentorientation than the present-day (north-northwest- and west-northwest–striking) systems.

DISCUSSION AND SUMMARY

This paper demonstrates that GPR is a viable componentof characterization of a reservoir analog in a collapsed paleo-cave environment. GPR does not have the capability to resolve


fine-scale details, but distinct GPR facies can be defined thatcorrelate with the undisturbed host rock and the larger scales ofbrecciation observed in outcrop and cores. It should be notedthat the present study involves a depth increment of only about15 m. Similar paleocave facies also exist at greater depths, asseen in deeper (100-m) cores obtained by the quarry operatorsand by evidence of the existence of deeper collapsed zones inthe present data.

The combination of facies description and definition fromcores, with ties to the GPR lines (Loucks et al., 1999) to ex-trapolate interpretations away from the coreholes, providesthe basis of 3-D interpolation and interpretation at interwelldimensions. GPR data interpretation must include awarenessof the effects of nongeologic factors and the distribution of wa-ter that potentially produce locally erroneous interpretations.

The spatial complexity of paleocave systems is revealed inthe cores and imaged in the GPR lines. This study is the first toprovide a 3-D picture at the interwell scale and has implicationsfor both exploration and production strategies. For example, anexploration well may penetrate a poor reservoir section, and asidetrack or stepout well can encounter a high-quality reservoir

FIG. 10. Cross-plots of lab measurements of effective conductivity, permeability, porosity, grain density, and GPRvelocity. In general, the correlations are low, but trends can be observed. There are only small differences betweenthe oven dried and ambient saturation measurements because the average ambient saturation was only about0.1% by weight. The logarithms on the permeability axes are base 10.

section. Passage-related thief zones and fractures may causesevere production problems. Understanding of the geometryand facies distributions of collapsed paleocave reservoirs willbe crucial in optimizing production.

The composite GPR/core methodology employed appearsto be successful and cost effective. The extension to true 3-Dacquisition and processing, as previously applied to clastic sys-tems (McMechan et al., 1997; Corbeanu et al., 2001; Szerbiaket al., 2001) now also seems warranted for carbonate strata.GPR surveys of reservoir analogs can provide internal detailsof depositional surfaces and sedimentary features, and imagethe spatial variation of facies that correlate with porosity andpermeability, at a scale that is of direct value for reservoir mod-eling and flow-unit zonation. The analog model is a more re-alistic and detailed picture than can be obtained from any realreservoir at depth, even though the conditions at depth in areal reservoir are not the same as those in an analog at thesurface.

This study supports the model of paleocave reservoirs as avolume (with dimensions of hundreds or thousands of meters)of coalesced, collapsed cave systems (Loucks, 1999) rather than


FIG. 11. Pseudo 3-D paleocave facies distributions. The three horizontal slices show the dominant facies present in the depthintervals 3–6 m (10–20 ft), 6–9 m (20–30 ft), and 9–12 m (30–40 ft).

single collapsed cave passages a few meters in diameter. Thebreccia bodies that outline the trend of former passages are upto 300 m wide; the intervening areas between breccias are up to200 m wide. This model, and the accompanying sag and faultstructures, should also be a useful guide to interpretation ofseismic data from similar environments, including other occur-rences of the Ellenburger strata, such as in west Texas, whereit contains hydrocarbons at depth.

ACKNOWLEDGMENTS

The research leading to this paper was supported by a con-sortium of companies including Altura Energy, ARCO Per-mian, ARCO Technology and Operational Services, Burling-ton Resources, Conoco, Enron International, Kerr McGee Oiland Gas, Oryx Energy, Texaco, TotalFinaElf, and Unocal. Theproject was administered through the Gas Research Instituteunder Contract 5097-210-3889. The authors appreciate the kindcooperation of Tom Cook and Rick Hohman of the Dean WordQuarry in allowing access to the field site. The grid layoutwas performed using GPS by Carlos Aiken, Mamadou Balde,and Greg Lyman. The whole-core porosity and permeability

measurements were done by Reservoir Inc., and the electricalproperty measurements were done by Carlos Santamarina atthe Georgia Institute of Technology. Special thanks to the 12UTD graduate students who survived fire ants and 100◦ F tem-peratures to complete the GPR data acquisition. This paper isContribution No. 953 from the Geosciences Department at theUniversity of Texas at Dallas.

REFERENCES

Baker, P. L., 1991, Fluid, lithology, geometry, and permeability infor-mation from ground penetrating radar for some petroleum industryapplications: SPE 22976.

Beck, B. F., and Wilson, W. L., 1988, Interpreting ground penetratingradar profiles in karst terrane: National Water Well Association Sec-ond Conference on Environmental Problems in Karst Terranes andtheir Solution, 347–367.

Beres, M., Green, A., Hugenberger, and P., Horstmeyer, H., 1995,Mapping the architecture of glaciofluvial sediments with three di-mensional georadar: Geology, 23, 1097–1090.

Bridge, J. S., Alexander, J., Collier, R. E. Ll., Gawthorpe, R. L., andJarvis, J., 1995, Ground penetrating radar and coring used to studylarge-scale structure of point bar deposits in three dimensions: Sed-imentology, 42, 839–852.

Cai, J., and McMechan, G. A., 1995, Ray-based synthesis of bistaticground penetrating radar profiles: Geophysics, 60, 87–96.


Candelaria, M. P., and Reed, C. L., Eds., 1992, Paleokarst related di-agenesis and reservoir development: Examples from Ordovician-Devonian age strata of west Texas and the mid-continent: PermianBasin Section SEPM 1992 Annual Field Trip.

Casas, A., Lazaro, R., Vilas, M., and Busquet, E., 1996, Detectingkarstic cavities with ground penetrating radar at different geologicalenvironments in Spain: Proceedings, 6th International Conferenceon Ground Penetrating Radar, 455–460.

Corbeanu, R. M., Soegaard, K., Szerbiak, R. B., Thurmond, J. B.,McMechan, G. A., Wang, D., Snelgrove, S., Forster, C. B., andMenitove, A., 2001, Detailed internal structure of a fluvial chan-nel sandstone determined from outcrop, cores, and 3-D ground-penetrating radar: Example from the mid-Cretaceous FerronSandstone, east-central Utah: AAPG Bull., 85, 1583–1608.

Dagallier, G., Laitinen, A. I., Malartre, F., Van Campenhout, I. P. A. M.,and Veeken, P. C. H., 2000, Ground penetrating radar applicationin a shallow marine Oxfordian limestone sequence located on theeastern flank of the Paris Basin, NE France: Sedimen. Geol., 130,149–165.

Daniels, D. J., 1996, Surface-penetrating radar: Inst. Electr. Eng.Daniels, D. J., Gunton, D. J., and Scott, H. F., 1988, Introduction to

subsurface radar: IEE Proceedings, 135F, 278–320.Davis, J. L., and Annan, A. P., 1989, Ground penetrating radar for high

resolution mapping of soil and rock stratigraphy: Geophys. Prosp.,37, 531–551.

Flint, S. S., and Bryant, D., Eds., 1993, The geological modeling ofhydrocarbon reservoirs and outcrop analogues: Blackwell Sci. Publ.

Gawthorpe, R. L., Collier, R. E. Ll., Alexander, J., Bridge, J. S.,and Leeder, M. R., 1993, Ground-penetrating radar: application tosandbody geometry and heterogeneity studies, in North, C. P., andProsser, D. J., Eds., Characterization of fluvial and aeolian reservoirs:The Geological Society Spec. Pub. 73, 421–432.

Hardage, B. A., Carr, D. L., Lancaster, D. E., Simmons, Jr., J. L.,Elphick, R. Y., Pendleton, V. M., and Johns, R. A., 1996, 3-D seismicevidence of the effects of carbonate karst collapse on overlying clas-tic stratigraphy and reservoir compartmentalization: Geophysics, 61,1336–1350.

Holtz, M. H., and Kerans, C., 1992, Characterization and classificationof west Texas Ellenburger reservoirs, in Candelaria, M. P., and Reed,C. L., Eds., Paleokarst related diagenesis and reservoir development:Examples from Ordovician-Devonian age strata of west Texas andthe mid-continent: Permian Basin Section SEPM 1992 Annual FieldTrip, 45–54.

Jol, H. M., and Smith, D. G., 1991, Ground-penetrating radar of north-ern lacustrine deltas: Can. J. Earth Sci., 28, 1939–1947.

Kerans, C., 1988, Karst-controlled reservoir heterogeneity in Ellen-burger Group carbonates of west Texas: AAPG Bull., 72, 1160–1183.

Klein, K., and Santamarina, J. C., 1997, Methods for broadband di-electric permittivity measurements (soil-water mixtures, 5 Hz to1.3 GHz): Geotech. Test. J., 20, 168–178.

Kupecz, J. A., and Land, L. S., 1991, Late-stage dolomitization of theLower Ordovician Ellenburger Group, west Texas: J. Sed. Petrol.,61, 551–574.

Loucks, R. G., 1999, Paleocave carbonate reservoirs: Origins, burial-depth modifications, spatial complexity, and reservoir implications:AAPG Bull., 11, 1795–1834.

Loucks, R. G., and Anderson, J. H., 1985, Depositional facies, dia-genetic terranes, and porosity development in Lower OrdovicianEllenburger Dolomite, Puckettt field, west Texas, in Roehl, P. O., andChoquette, P. W., Eds., Carbonate petroleum reservoirs: Springer-Verlag: 19–38.

Loucks, R. G., and Handford, C. R., 1992, Origin and recognitionof fractures, breccias and sediment fills in paleocave-reservoir net-works, in Candelaria, M. P., and Reed, C. L., Eds., PaleokarstRelated Diagenesis and Reservoir Development: Examples fromOrdovician-Devonian Age Strata of West Texas and the Mid-Continent: Permian Basin Section SEPM 1992 Annual Field Trip,31–44.

Loucks, R. G., and Mescher, P. K., 1996, Architecture of collapsed-paleocave reservoirs (abst.): 1996 AAPG Ann. Conv. OfficialProgram, Am. Assn. Petr. Geol., A87.

——— 1997, Interwell scale architecture, heterogeneity, and pore net-work development in paleocave reservoirs: SEPM Fieldtrip Guide-book for the 1997 AAPG Annual Meeting.

——— 2001, Paleocave facies classification and associated pore types:AAPG Southwest Sec. Ann. Mtg. Papers and Abstracts, Am. Assn.Petr. Geol., CD-ROM.

Loucks, R. G., Mescher, P. K., and McMechan, G. A., 1999, Archi-tecture of a coalesced, collapsed-paleocave system in the LowerOrdovician Ellenburger Group, Dean Word Quarry, Marble Falls,Texas: Gas Res. Inst. Rep. 5097-210-3889.

——— 2001, Characterizing the three-dimensional architecture of acoalesced, collapsed-paleocave system in the Lower Ordovician El-lenburger Group by integrating ground-penetrating radar, shallowcore, and outcrop data (abst.): 2001 AAPG Ann. Conv. Official Pro-gram, Am. Assn. Petr. Geol., A121.

McMechan, G. A., Gaynor, G. C., and Szerbiak, R. B., 1997, Use ofground-penetrating radar for 3-D sedimentological characterizationof clastic reservoirs: Geophysics, 62, 786–796.

McMechan, G. A., Loucks, R. G., Zeng, X., and Mescher, P., 1998,Ground penetrating radar imaging of a collapsed paleocave systemin the Ellenburger dolomite, central Texas: Appl. Geophys., 39, 1–10.

Miall, A. D., and Tyler, N., Eds., 1991, The three-dimensional faciesarchitecture of terrigenous clastic sediments and its implications forhydrocarbon discovery and recovery: Soc. Econ. Paleont. Mineral.

Pipan, M., Baradello, L., Forte, E., and Prizzon, A., 2000. GPR studyof bedding planes, fractures and cavities in limestone; in Noon,D. A., Stickey, G. F., and Longstaff, D., Eds., Proc., 8th Internat.Conf. on GPR, Expanded Abstracts, SPIE 4084, 682–687.

Reynolds, J. M., 1998, An introduction to applied and environmentalgeophysics: John Wiley & Sons.

Sigurdsson, T., and Overgaard, T., 1998, Application of GPR for 3-Dvisualization of geological and structureal variation in a limestoneformation: J. Appl. Geophys, 40, 29–36.

Stoudt, E. L., 1996, Precambrian-Devonian geology of the FranklinMountains, west Texas—Analogs for exploration and production inOrdovician and Silurian reservoirs in the Permian Basin: West TexasGeological Society 1996 Annual Field Trip Guidebook.

Szerbiak, R. B., McMechan, G. A., and Corbeanu, R. M., 2001, 3-Dcharacterization of a clastic reservoir analog: From 3-D GPR datato a 3-D fluid permeability model: Geophysics, 66, 1026–1037.

Tyler, N., Barton, M. D., Bebout, D. G., Fisher, R. S., Grigsby, J. D.,Guevara, E., Holtz, M., Kerans, C., Nance, H. S., and Levey, R. A.,1992, Characterization of oil and gas reservoir heterogeneity: U.S.Dept. Energy Rep. 14403-3.

Ward, S. H., and Hohmann, G. W., 1987, Electromagnetic theory forgeophysical applications, in Nabighian, M. N., ed., Electromagneticmethods in applied geophysics: Soc. Expl. Geophys., 131–311.

Zolotarev, V. P., Grigorieff, K. N., Glotov, V. P., and Stanzyte, B., 1996,Detection of cavities in dolomite and profiling of alluvial depositsusing GPR in Lithuania: Proceedings, 6th International Conferenceon Ground Penetrating Radar, 445–448.

Characterization of a coalesced, collapsed paleocave ...

Documents

Transcript of Characterization of a coalesced, collapsed paleocave ...