whole-rock geochemistry and heavy mineral analysis as petroleum exploration … · 2019-07-17 ·...

320

1 Now with Diamondex Resources Ltd., P.O. Box 11584, 1410 - 650 West Georgia Street, Vancouver, BC V6B 4N8 Canada

ABSTRACT

Principles of chemostratigraphic characterization and correlation employing whole-rock inorganic chemical data andheavy mineral grain counts are applied to the frontier Bowser and Sustut basins. Methodologies commonly used with well samples in mature petroleum provinces can be applied to field samples, providing a vital and practical link between theearliest frontier investigations and more advanced hydrocarbon exploration.

The major stratigraphic divisions of the basins, the Bowser Lake and Sustut groups, have markedly different indicationsof sedimentary provenance from heavy mineral analysis, and are readily differentiated geochemically. Variations in keyelements are related directly to the provenance indications identified by heavy minerals.

Lithofacies assemblages within the Bowser Lake Group also display contrasting provenance signatures. Sandstones andconglomerates from marine facies (Ritchie-Alger, Todagin, and Muskaboo Creek assemblages) are differentiated fromthose of the deltaic and nonmarine units (Eaglenest, Skelhorne, Groundhog-Gunanoot and Jenkins Creek assemblages) by higher Fe2O3 and MgO contents, which may relate to increased glauconite contents in marine units. Sandstones and conglomerates from the deltaic to nonmarine units are separated with less certainty by heavy mineral contents and elementconcentrations or ratios.

RÉSUMÉ

Des principes de caractérisation chimiostratigraphique et de corrélation, utilisant des bases de données compositionnellesde roche totale et des comptes des grains de minéraux lourds, sont appliqués dans les régions pionnières des bassins deBowser et de Sustut. Les méthodologies communément employées sur des échantillons de puits dans les provinces maturesen pétrole, peuvent s’appliquer sur des échantillons prélevés sur le terrain, fournissant un lien vital et pratique entre lesrecherches pionnières les plus anciennes et les explorations d’hydrocarbures les plus avancées.

Les divisions stratigraphiques majeures des bassins, ceux des groupes de Bowser Lake et de Sustut, présentent des indications nettement différentes de provenance sédimentaire à partir d’analyses de minéraux lourds, et se différencientaisément sur le plan géochimique. Des variations d’éléments clés sont reliées directement aux indications de provenance,identifiées par les minéraux lourds.

Les assemblages de lithofaciès, à l’intérieur du groupe de Bowser Lake , montrent des signatures de provenance contrastée. Les grès et les conglomérats, provenant de faciès marins (assemblages de Ritchie-Alger, Todagin, et deMuskaboo Creek), se différencient de ceux des unités deltaïques et non marines (assemblages de Eaglenest, Skelhorne,Groundhog-Gunanoot et de Jenkins Creek) par de hautes teneurs en Fe2O3 et en MgO, qui peuvent être liées à des augmentations de teneur en glauconite dans les unités marines. Les grès et les conglomérats, provenant des unitésdeltaïques à non marines, sont séparés avec moins de certitude par des teneurs en minéraux lourds et des concentrationsou des fractions d’éléments.

Traduction de Gabrielle Drivet

BULLETIN OF CANADIAN PETROLEUM GEOLOGYVOL. 55, NO. 4 (DECEMBER, 2007), P. 320–336

Whole-rock geochemistry and heavy mineral analysis as petroleum exploration tools in the Bowser and Sustut basins, British Columbia, Canada

K.T. RATCLIFFE

Chemostrat Inc. 5850 San Felipe

Suite 500Houston, TX 77057

USA

A.C. MORTON

HM Research AssociatesWoodhouse Eaves, Leics United

Kingdom and Department ofGeology and Petroleum Geology

University of AberdeenUnited Kingdom

D.H. RITCEY1

Geological Survey of Canada625 Robson Street

Vancouver, BC V6B 5J3 Canada

C.A. EVENCHICK

Geological Survey of Canada625 Robson Street

Vancouver, BC V6B 5J3Canada

provenance, and may permit geochemical modeling of aspectssuch as facies variations and clastic mineralogy. Chemical andmineralogical compositions may also provide informationabout spatial or stratigraphic variation in potential reservoirquality, which will aid in prioritizing regions or units in a fron-tier setting.

Chemostratigraphy, as with many stratigraphic techniques,is largely a subjective and interpretative exercise. Within anysedimentary succession, the large number of variables thataffect element concentrations means that not all samples willfall within the “typical” range of compositions defined for thatunit. A successful chemostratigraphic study therefore requires adataset sufficiently large to avoid undue influence of outliers.Outliers may need to be disregarded for the overall characteri-zation, but they cannot be ignored entirely and should be takenas an indication of the limitations of the dataset. Alternatively,they may be viewed as potential indicators of stratigraphicissues not otherwise realized.

A definitive geochemical study of the Bowser and Sustutbasins is beyond the scope of this paper. Rather, the aims of thiswork are to demonstrate the applicability of chemostratigraphyto a frontier setting, and to investigate some implications of thechemical and mineralogical compositions.

GEOLOGICAL SETTING

The area covered in this paper is the northern two-thirds of theBowser and Sustut basins, in north-central British Columbia,Canada (Fig. 1). These are frontier petroleum explorationbasins where much new information pertinent to petroleumsystems has recently been developed and released (e.g.Evenchick et al., 2005a, and references therein).

The Bowser and Sustut basins together occupy an area ofmore than 62,000 square kilometres in the intermontane mor-phogeological belt of the Canadian Cordillera (Fig. 1). TheBowser Basin is the depocentre of the Bowser Lake Group, alate Middle Jurassic to early Cretaceous clastic overlap assem-blage with a total thickness probably in excess of 5000 m(Evenchick and Thorkelson, 2005). Clastic deposition was ini-tiated by accretion of the Stikine Terrane (Stikinia) to NorthAmerica (Gabrielse, 1991; Ricketts et al., 1992). Strata weredeposited upon volcanic arc units of Stikinia, and clasts weresubstantially derived from the accretionary complex andoceanic units of the Cache Creek terrane to the east. (Eisbacher,1974a, 1981; Evenchick and Thorkelson, 2005, and referencestherein).

Within the study area, the Bowser Lake Group has been sub-divided into seven lithofacies assemblages and one defined for-mation (Evenchick and Thorkelson, 2005; Evenchick et al.,2005b). Subdivisions are the Ritchie-Alger assemblage (sub-marine fan), Todagin assemblage (marine slope), MuskabooCreek assemblage (marine shelf to shoreface), Eaglenest,Skelhorne, and Groundhog-Gunanoot assemblages (3 distinctdeltaic successions), Jenkins Creek assemblage (nonmarine),and the fluvial-alluvial Devils Claw Formation. Across thenorthern two-thirds of the basin, the Bowser Lake Group is an

GEOCHEMISTRY AND HEAVY MINERAL ANALYSIS AS EXPLORATION TOOLS 321

INTRODUCTION

CHEMOSTRATIGRAPHIC RATIONALE

The application of whole-rock geochemical data to solvingcorrelation problems is commonly referred to as chemostrati-graphy, or chemical stratigraphy. Strictly, however, chemo-stratigraphic characterization is the zonation of a sequence interms of its chemical characteristics, whereas chemostrati-graphic correlation is the extension of this zonation from onegeographic location to another. Chemostratigraphy in thispaper involves the characterization of strata using variations intheir major and trace element concentrations. The technique isextensively used in the oil industry to define chemostrati-graphic correlation frameworks between well-bore sections(Ehrenberg and Siring, 1992; Racey et al., 1995; Preston et al.,1998; Pearce et al., 1999; Wray, 1999; Pearce et al., 2005a;2005b; Ratcliffe et al., 2004; Ratcliffe et al., 2006). Althoughthe technique is becoming increasingly accepted as a strati-graphic tool in the petroleum industry, most petroleum-relatedchemostratigraphic studies are in mature hydrocarbonprovinces with numerous well penetrations, rather than frontierscenarios where well data are limited or non-existent.

This paper applies chemostratigraphic methodologies to asuite of field samples from frontier basins. The approach inmature petroleum settings is generally to use well cuttings that,by their depth-related nature, carry an inherent stratigraphiccomponent (unless disrupted by faults and folds), that is absentfor the geographically widespread samples in this study.Additionally, chemostratigraphic correlation involves theextension of a geochemical characterization from one geo-graphic area to another (e.g. one well to another), which is notthe case with field samples. Rather, in the situation of field-based samples, the stratigraphic assignment of each field sam-ple is used to build a chemostratigraphic characterization foreach map unit. Using the methods and general approachemployed by authors such as Pearce et al. (2005b) and Ratcliffeet al. (2006), the elemental data from field samples can be usedto develop a chemical “fingerprint” for each lithostratigraphicunit. Once this characterization is accomplished, it can be usedto assign any sample to one of the pre-defined units.Classification of cuttings samples which, by their nature, lackmany of the sedimentary features used to assign field samplesto a unit, provides an important link between frontier investi-gations and more advanced petroleum basin exploration. Lackof definitive features in cuttings samples is exacerbated whenfixed cutter (or PDC) bits are used and the entire fabric of therock is destroyed. Use of these drill bits has no deleteriouseffect on the whole rock geochemistry. In general, the chemical“fingerprint” of a unit may be made up of the absolute concen-tration of particular elements, certain element ratios, or a dis-tinctive trend through the stratigraphic thickness of the unit.

In addition to characterization and classification of units, thelarge dataset routinely obtained for a chemostratigraphic appli-cation (47 element concentrations for each sample in thisstudy) can provide important information about sediment

overall regressive sequence with shallow marine and deltaicfacies belts migrating to the south and southwest through time(Evenchick et al., 2001). The assemblages interfinger laterally,and are substantially overlapping in age (Fig. 2). In addition tothese depositional patterns, the present outcrop distribution ofthese assemblages is further controlled by fold and thrust struc-tures in the Skeena Fold Belt (e.g. Evenchick, 1991; Evenchickand Thorkelson, 2005).

The mid- to Late Cretaceous Sustut Basin extends alongmuch of the northeast side of the Bowser Basin. In part, this isa foreland basin, developed in response to fold and thrustdeformation within the Bowser Basin (Evenchick, 1991).Fluvial, alluvial, and lacustrine units of the Sustut Group havea total thickness of at least 2000 m, and are divided into thelower Tango Creek Formation and the overlying, more con-glomeratic Brothers Peak Formation. On the southwest side ofthe Sustut Basin, strata unconformably overlie deformedBowser Lake Group and Stikinia strata, and on the northeastside they directly overlie Stikinia. Early in its depositional his-tory, the Sustut Group had eastern sources, including metamor-phic and plutonic rocks of the Omineca Belt (Fig. 1), followedin time by a western source largely derived from the BowserLake Group (Eisbacher, 1974b). Distinct sources for theBowser Lake and Sustut groups have been recognized fromclast compositions and paleocurrents (Eisbacher, 1974a, b,1981), and detrital zircon ages (McNicoll et al., 2005).

STUDY METHODS AND DATA SET

This study provides an initial application of chemostratigraphictools to a large, relatively little-studied sedimentary basin.Outcrop samples were selected from all stratigraphic units ofthe Bowser Lake and Sustut groups, with broad geographiccoverage and representation from different grain sizes. Thesample suite comprises 36 coarse siltstones to shales (consid-ered together as siltstones/claystones), 71 sandstones (mainlyfine to medium arenites), and 51 conglomerates (chert granuleto chert pebble conglomerates). Analysed conglomerates gen-erally have a sandy matrix, relatively few clasts greater than5 mm in diameter, and mean grain size less than 1 mm(Reichenbach et al., 2006). Sample locations are shown onFigure 3, and the number and type of samples associated witheach stratigraphic unit is summarized in Table 1. Few samplesare tightly constrained in age, but available age informationsuggests the sample set covers much of the known age range ofthe units. However, Upper Jurassic strata cover a much largerportion of the Bowser Basin than Middle Jurassic strata(Evenchick et al., 2001). Therefore, sampling is biased towardthe younger portions of units in the Bowser Lake Group.

Whole-rock compositional data for ten major elements(expressed as oxides) and thirty-seven trace elements (includingREEs) was acquired by ICP-OES (inductively-coupled plasma-optical emission spectrometry) and ICP-MS (inductively-coupled plasma - mass spectrometry). Sample preparation andanalytical procedures are as outlined in Jarvis and Jarvis (1995)and Pearce et al. (1999). Samples were prepared by crushing

322 K.T. RATCLIFFE, A.C. MORTON, D.H. RITCEY and C.A. EVENCHICK

Fig. 1. Location of the Bowser and Sustut basins within theCanadian Cordillera, showing morphogeological belts and the general-ized distribution of Cache Creek and Stikine terranes. Modified fromWheeler and McFeely, 1991.

Fig. 2. Approximate ages and relationships of units in the BowserLake and Sustut groups in the northern Bowser and Sustut basins.

approximately 250 g of rock, followed by alkali fusion of ahomogenized 0.25 g powdered subsample. Precision for majorelement data is generally below 2%, whereas precision is belowapproximately 3% for the high abundance trace element datadetermined by ICP-OES, and below approximately 5% for elemental data acquired by ICP-MS. Standard rock referencematerials (SRM’s) were analyzed along with the field samples,and used to apply drift corrections.

Heavy mineral analyses were carried out on a subset ofthirty sandstone and sandy conglomerates from this suite.Samples were disaggregated without grinding, sieved,cleaned, and dried. Heavy minerals were separated from the63–125 micron fraction in bromoform with a measured specificgravity of 2.8, and mounted under Canada Balsam for optical

identification and counting using a polarizing microscope.Mineral proportions were estimated by counting up to 200 non-opaque detrital grains.

Geochemical and heavy mineral datasets have been pub-lished in Ritcey et al. (2005), and petrographic data anddescriptions for the same sample suite are in Reichenbach et al.(2006).

SEDIMENT PROVENANCE

HEAVY MINERAL PROVENANCE INDICATORS

Heavy mineral analyses were carried out primarily to identifycontrasts or variations in heavy mineral suites that can berelated to sediment provenance. Additionally, since many


Fig. 3. Location map for samples in this study.

heavy minerals have distinctive concentrations of certain traceelements, the heavy mineral data can be used to better under-stand the factors controlling geochemical variations used forchemostratigraphic characterization and correlation.

Heavy mineral abundances for all 30 samples in Figure 4show markedly different mineral assemblages, both between

and within the Sustut and Bowser Lake groups. Although manyof these variations suggest variations in sediment provenance,heavy mineral suites are not entirely controlled by source rockmineralogy. Other processes, principally weathering, hydrody-namics, and diagenesis, may overprint the original provenancesignal (Morton and Hallsworth, 1999). Such effects can be


Table 1. Distribution of samples by stratigraphic unit and lithology.

Fig. 4. Heavy mineral grain counts, normalized to 100% total abundance.

counteracted by examining ratios or indices of stable mineralswith similar densities, because these indices are not affected bychanges in hydraulic conditions during sedimentation or bydiagenesis (Morton and Hallsworth, 1994). Table 2 containscalculated indices for provenance-sensitive heavy minerals.The index ATi is defined as % apatite in total apatite plus tour-maline, GZi as % garnet in total garnet plus zircon, RZi as %TiO2 minerals in total TiO2 minerals plus zircon, RuZi as %rutile in total rutile plus zircon, CZi as % chrome spinel in total

chrome spinel plus zircon, and CGi as % chrome spinel in totalgarnet plus chrome spinel (after Morton and Hallsworth, 1994).

Sustut Group samples commonly contain abundant epidoteand almost ubiquitously lack chrome spinel (Fig. 4). Althoughepidote is abundant in the Sustut Group overall, it is relativelyunstable during burial diagenesis (Morton and Hallsworth,1999), and is therefore not used as a provenance indicator.Sustut Group samples show substantial variation in garnet:zir-con (GZi) and rutile:zircon (RuZi) (Table 2) resulting from the


Table 2. Provenance-sensitive heavy mineral indices, determined on the 63–125 micron fraction of sandstone samples. ATi = % apatite in total apatite plus tourmaline, GZi = % garnet in total garnet plus zircon, RZi = % TiO2 minerals in total TiO2 minerals plus zircon, RuZi = % rutile in total rutile plus zircon, CZi = % chrome spinel in total chrome spinel plus zircon, CGi = % chrome spinel in

total garnet plus chrome spinel (after Morton and Hallsworth, 1994). Blanks signify lack of accurate data due to poor recovery of relevant heavy minerals.

interplay of two end-member provenances, one characterizedby high GZi and high RuZi, the other by low GZi and lowRuZi. The high GZi, high RuZi component (Table 2) is proba-bly indicative of a metasedimentary source, which could alsohave supplied the abundant epidote. The low GZi, low RuZiend member is probably an acid igneous component, likelyincluding allanite-bearing granitoids in part. The variations inATi may indicate variable involvement of a tourmaline-richsource, probably within the Omineca Belt.

Most samples from the Skelhorne, Eaglenest, MuskabooCreek, Todagin, and Ritchie-Alger assemblages are substan-tially dominated by chrome spinel, suggesting a strong mafic-ultramafic component. The high chrome spinel:zircon ratios(CZi) and chrome spinel:garnet ratios (CGi) of these samplesplot as a tight cluster on Figure 5. Pyroxenes are expected froman ultramafic source, but are probably lacking due to their lowstability during burial diagenesis (Morton and Hallsworth,1999). Samples from the Skelhorne assemblage exhibit a rangeof heavy mineral indices similar to most other BLG sandstonesand conglomerates, but they have notably lower total counts ofheavy minerals (data in Reichenbach et al., 2006).

The Jenkins Creek assemblage is dominated by zircon andapatite, with some chrome spinel, indicative of a predominantly

acid igneous provenance. On the CZi vs. CGi plot (Fig. 5),Jenkins Creek samples plot with high CGi, but low to moder-ate CZi indices supporting the suggestion of an acid igneoussediment source, with minor supply from mafic-ultramaficrocks. Samples from the Groundhog-Gunanoot assemblagecontain a mixture of chrome spinel, zircon and apatite, sug-gesting mixed provenance from mafic-ultramafic and acidigneous sources. This is supported by their position relative toother BLG samples on Figure 5.

PROVENANCE INDICATIONS FROM

WHOLE ROCK GEOCHEMISTRY

Concentration of each element in the full analytical suite isinfluenced by the distribution of minerals, which in turn reflectschanges in a variety of geological conditions, such as sedimentprovenance, facies, syndepositional weathering, basin redoxconditions, and diagenesis. A complete chemostratigraphicframework identifies geological controls on the key elementsthat are employed. Establishing the links between sediment geo-chemistry and mineralogy can only be fully achieved when X-ray diffraction, thin section petrography, and scanning electron microscopy studies are carried out on each samplegeochemically analyzed. In practice, this type of detailed mineralogical analysis is seldom carried out in an oil field setting, particularly when dealing with cuttings samples, whichmeans that the mineralogical affinities of elements can only begeneralized. This approach is adopted herein.

There are marked variations in the heavy mineral speciespresent in sandstones and conglomerates of the various strati-graphic units, and many of these heavy minerals have distinctivemajor and trace element compositions that exert strong influ-ences on the whole rock geochemistry. Profiles of selected traceelement concentrations and ratios in Figure 6 (akin to profilescommonly presented for wellbore studies) display geochemicalvariations between and within the Sustut and Bowser Lakegroups. Zr and Ce are markedly enriched in the Sustut Group rel-ative to the Bowser Lake Group and Cr and Ni are both relativelydepleted. Comparison of element concentrations to heavy min-eral distribution shows that broadly Zr values are high where zir-con (ZrSiO4) is abundant. However, discrepancies exist betweenthe detailed distribution of Zr and zircon. For example, theRitchie-Alger Assemblage has higher Zr contents than theTodagin and Muskaboo Creek assemblages, yet has the lowestzircon contents. This discrepancy probably reflects the size of thezircons. Heavy mineral analysis for this study was carried out onthe 63–125 micron size fraction, yet as discussed below, it isknown that airfall zircons are present in parts of the Bowser LakeGroup. These airfall zircons are likely to be accounted for in thewhole rock geochemistry, but not the heavy mineral analysis,resulting in apparent discrepancies between the two data sets.The mineral association of high Ce in the Sustut Group is moreenigmatic, but the heavy mineral allanite has a general formulaof (Ca, REE, Th)2(Fe2+, Fe3+Al)3Si3O12OH and is a probablesource of the high Ce values within the Sustut Group. Cr values


Fig. 5. Key heavy mineral indices (CGI vs. CZi) plotted as a binarydiagram to demonstrate provenance variations in Bowser Lake andSustut groups. Double-headed arrows indicate possible mixing linesbetween provenance types.

are high where Cr-spinel [(Mg Fe2+)O(CrAl Fe3+)2O3] isabundant, implying this mineral exerts a strong influence on theconcentrations of this element. Ni is a common substitution inCr-spinel, potentially accounting for its high concentrations inthe Bowser Lake Group sediments. The high Ni may also be res-ident in chlorite, the product of pyroxene alteration. In eitherinterpretation, the high Ni values indicate a mafic-ultramaficsediment provenance. For the sandstones and sandy conglomer-ates, simple comparisons of geochemical composition and heavymineral distribution identify the Zr/Cr and Ce/Cr ratios displayedon Figure 6 as provenance indicators, such that high values of theratios indicate acid igneous provenance and low values a strongmafic influence on the sediment provenance.

Heavy mineral analysis is expected to be a more sensitiveindicator of sediment provenance than whole rock geochem-istry, due to the multiple and complex mineralogical controlson major trace element distributions, but it is ordinarily carriedout only on coarse grained samples. Once the geochemicalmodeling of provenance has been calibrated with heavy min-eral data, it may be possible to extend the interpretations tofiner-grained lithologies. This can be vitally important sincefine-grained rocks commonly constitute a high proportion ofunits encountered in many basins. Profiles of provenance-sensitive elements (as determined by calibrating sandstonewhole rock data with heavy mineral data) are shown in Figure 7for siltstone/claystone samples. There is a strong influence of


Fig. 6. Comparisons of selected trace element distributions and heavy mineral abundances for sandstones. Vertical position on profilesdoes not imply relative age of samples. Pie charts for the heavy mineral data are averages calculated for each assemblage.

provenance, probably in the form of silt-sized heavy mineralgrains, specifically the relative significance of acid igneous andmafic igneous sources. Such an approach must be tempered bythe realization that certain provenance indicators for sand-stones, e.g. Cr and Ni, are chemical constituents of some clayminerals, and cannot be used to recognize provenance signa-tures for true claystones. However, in practice, true claystoneswith zero silt content are uncommon, and the profiles inFigure 7 closely reflect those of Figure 6. A similar close rela-tionship between heavy mineral-related elements in sandstonesand claystones was also noted by Ratcliffe et al. (2006), furthersupporting the generalization that provenance indicators can bemodeled from typical fine grained lithologies.

CHEMOSTRATIGRAPHIC CHARACTERIZATION

A strict definition of a chemostratigraphic characterization is“the definition of stratigraphic units based on changes inchemical composition”. However, in a practical sense, it isimportant to develop characterizations that can be used forstratigraphic correlation. A stratigraphic characterizationmade, for instance, on the definition of a unit with high CaO

values due to calcite cements within an otherwise uncementedsandstone is a valid stratigraphic characterization, but offerslittle potential for stratigraphic correlation. The choice ofwhich elements to use for chemostratigraphic characterizationis initially an entirely pragmatic one. Elements that displaysystematic variations between stratigraphic units are selectedas potential key chemostratigraphic elements and ratiosbetween these elements calculated to enhance the often subtlevariations in absolute concentrations. Statistical techniquessuch as multivariate analysis and principal component analy-sis are commonly employed to identify elements that canserve as discriminators.

In well-bore sections, systematic trends in element concen-trations are generally readily identified, since in structurallysimple sequences, changes in element concentrations withdepth are directly related to stratigraphy. For field samplesfrom widespread geographic locations, it is imperative that astratigraphic framework is in place prior to attemptingchemostratigraphic characterization. Once the key chemostrati-graphic elements are identified, element ratios are devised thatemphasize more subtle variations between the stratigraphicunits, enabling chemostratigraphic characterizations that have


Fig. 7. Provenance-sensitive elements in siltstone/claystone samples. Vertical position on profiles does not imply relativeage of samples.

potential as correlation tools. Only then is consideration givento the likely controls on these key elements and ratios.

The primary control on element distributions in anysequence is lithology. Each of the major lithological groups(siltstones/claystones, sandstones, and conglomerates) isexpected to have its own geochemical characteristics, such thatcompositional contrasts between claystones and sandstoneswithin a single unit may be far greater than the variationsbetween units for one lithology (Ratcliffe et al., 2006).

Siltstones/claystones, sandstones and conglomerates of theSustut Group are all geochemically distinct from their counter-parts in the Bowser Lake Group. This is evident from the chem-ical logs displayed on Figures 6 and 7 and can also beexpressed with binary and ternary diagrams (Fig. 8). Once dis-criminatory diagrams have been erected, they can be used asclassification tools to characterize samples of unknown strati-graphic affinity, including well cuttings samples. Variablesused here to differentiate the Sustut and Bowser Lake groupsare Zr, Cr and Ce, elements whose distribution has been shownto be related to changes in sediment provenance in the discus-sions above.

Key geochemical features and potential geological interpre-tations are summarized in Table 3. Sandstones and conglomer-ates from the Bowser Lake Group have geochemicalcharacteristics that can be related to the different lithostrati-graphic assemblages. There are insufficient siltstone/claystonesamples to make meaningful chemostratigraphic characteriza-tions for these finer grained lithologies. The primary geochem-ical characterization of the Bowser Lake Group samples isrelated to changes in the depositional environments of theassemblages. Sandstones and conglomerates from the marineunits (Ritchie-Alger, Todagin and Muskaboo Creek assem-blages) can be readily differentiated from the deltaic to nonma-rine units (Eaglenest, Skelhorne, Groundhog-Gunanoot,Jenkins Creek assemblages and Devils Claw Formation) usingbinary and ternary diagrams (Fig. 9). Marine assemblage sand-stones and conglomerates have generally higher values ofFe2O3 and MgO than the nonmarine samples. Although themineralogical significance of this can only be surmised withoutdetailed X-ray diffraction or petrographic data, it most proba-bly relates to the amount of Fe-Mg minerals, which could becarbonates (e.g. siderite or ferroan dolomite) or Al silicates(e.g. chlorite, mica or glauconite). The increase in Fe2O3 andMgO corresponds to a broad change from nonmarine to marinefacies, which could see an increase in glauconite, pointingtoward the high element values being related to high glau-conite. However, influences of other minerals cannot be ruledout without extensive mineralogical data. Existing XRD andpoint-count data for these samples (Reichenbach et al., 2006)show the highest average chlorite and total clay contents are inthe Ritchie-Alger assemblage, but the sparse dataset, whichonly sporadically reports glauconite, does not indicate a cleardistinction between marine and nonmarine units in terms of anyone Fe-Mg mineral.

Sandstones and conglomerates from the deltaic and non-marine assemblages are characterized and differentiated in


Fig. 8. Graphical plots differentiating the Bowser Lake Group fromthe Sustut Group. A) and B) sandstone and conglomerate data; C) silt-stone-claystone data.

Figures 10 and 11. Data for one sandstone and one conglom-erate from the fluvial-alluvial Devils Claw Formation areincluded in this series of plots: in most cases the samplepoints lie close to those for the fully nonmarine Jenkins Creekassemblage. These two units are distinguished from the strati-graphically lower assemblages by their high Zr and low Crcontents. The Groundhog-Gunanoot and Skelhorne deltaicassemblages are most clearly separated by the Zr vs. Crbinary plots, whereas the Eaglenest assemblage is chemicallydistinguished from other units by low Na2O values for bothsandstones and conglomerates (Fig. 10). Variations in the Zrand Cr values likely reflect changes in sediment provenance,albeit more subtle signatures than those which result in thestrong geochemical contrasts between Bowser Lake andSustut groups. Controlling factors for Na2O values cannot beunambiguously determined, however, Na2O is most com-monly related to clay minerals, evaporites or feldspar (pla-gioclase). On the binary diagrams of Figure 10, there is no

overall linear relationship between Na2O and Al2O3, stronglysuggesting that Na2O is controlled by more than one mineral.There are no evaporite sequences in these units, suggestingthat the Na2O/Al2O3 ratio is related to clay minerals and toplagioclase. A higher value of this ratio indicates higher rela-tive plagioclase content, thereby implying that the Eaglenestassemblage has distinctly lower plagioclase contents than theother deltaic assemblages.

IMPLICATIONS FOR RESERVOIR QUALITY

While the primary aim of chemostratigraphic studies in petro-leum provinces is to provide stratigraphic correlation schemes,the whole rock geochemical data can also be used to makeinterpretations regarding reservoir quality. For this study, deal-ing with limited data in a frontier basin, only very broad gen-eralizations about potential reservoir quality can be made.Although these are generalizations, they do provide insights as


Table 3. Summary of geochemical and provenance indicators for sandstones and conglomerates from subunits of the Bowser Lake and Sustut groups.


Fig. 9. Binary and ternary diagrams used to show characterization and differentiation of marine from nonmarine units of the Bowser LakeGroup. Upper two graphs are constructed from sandstone data, lower two graphs are constructed from conglomerate data.

to where, within a large, poorly explored area, the probabilityof good reservoir development is greatest.

The SiO2/Al2O3 ratio in a siliciclastic sediment is a measureof the amount of quartz vs. Al-silicates. Generally, sandstoneswith high quartz contents are likely to have greater potentialreservoir quality. Most sandstones of the Eaglenest, Skelhorneand Groundhog-Gunanoot assemblages have generally higherSiO2 and lower Al2O3 values than those from the marineMuskaboo Creek, Todagin and Ritchie-Alger assemblages(Fig. 12). Sandstones from the nonmarine Jenkins Creekassemblage display a wide range of SiO2 : Al2O3 ratios. For theentire Bowser Lake Group sandstone and conglomeratedataset, SiO2 and Al2O3 have a correlation coefficient ofapproximately minus 0.60. The most probable interpretation ofthis overall trend is that the fluvial and deltaic sandstones have

less clay content than those deposited in marine environments.As such, the gross change in SiO2 : Al2O3 ratio can be used asa proxy for quartz vs. clay contents and therefore reservoirquality. Furthermore, high Zr concentrations in sandstones, ifdirectly related to detrital zircon and not airfall zircon, suggestdeposition in relatively high energy environments with miner-alogically and texturally mature sediments, i.e. more likely tobe better reservoirs. Therefore, it can be tentatively suggestedthat of the Eaglenest, Skelhorne and Groundhog-Gunanootassemblages, the Eaglenest assemblage, with its high Zr values,may have the best reservoir properties.

Sandstones and conglomerates from the Sustut Group haveSiO2 / Al2O3 values that are generally higher than those fromthe deltaic units of the Bowser Lake Group (Fig. 12).Additionally, using the Zr/Cr ratio values as a proxy for acid


Fig. 10. Binary diagrams constructed to characterize Eaglenest, Skelhorne Groundhog-Gunanoot, and Jenkins Creek assemblages, and DevilsClaw Formation: A) using sandstone data; B) using conglomerate data.

igneous vs. mafic sedimentary provenance, the Bowser LakeGroup is indicated to have higher mafic mineral content rela-tive to the Sustut Group. Generally, mafic minerals are delete-rious to reservoir quality as they are relatively unstable,degrading into clay minerals. Although these parameters do nottake into account important variations in diagenetic history thatobviously affect reservoir quality, they do suggest that in agross sense and on a basin-wide scale, the Sustut Group couldbe expected to have better overall reservoir potential than theBowser Lake Group and better reservoir quality may beexpected in the nonmarine lithostratigraphic assemblages ofthe Bowser Lake Group than in the marine units of that group.

DISCUSSION

The Bowser and Sustut basins offer a frontier petroleum explo-ration scenario, for which energy resource studies have identifiedsignificant petroleum potential (e.g. Evenchick et al., 2005a;Stasiuk et al., 2005; Osadetz et al., 2004), yet as of late 2006,only two conventional hydrocarbon exploration wells have beendrilled [Dome et al. a-3-J/104-A-6 British Columbia Ministry ofEnergy Mines and Petroleum (BCMEM) well File WA#2529,and c-62-G/104-A-6 BCMEM well File WA#3215]. Chemo-stratigraphy has been applied to this basin scenario to testwhether whole rock geochemical data can be used to character-ize the main lithostratigraphic units, and to investigate prove-nance indications and potential influences on reservoir quality.

Contrasts in whole rock composition and provenance-sensi-tive heavy mineral abundances and ratios identified in thisstudy corroborate the findings of previous work that indicateddistinct sediment sources for the Bowser Lake and Sustutgroups. The two groups are readily differentiated geochemi-cally, providing a first order chemostratigraphic characteriza-tion. In addition to the geochemical and mineralogicalindicators presented here and the clast studies of previousworkers (Eisbacher, 1974a, b; Evenchick and Thorkelson,2005, and references therein), contrasting sources for theBowser Lake and Sustut groups are also shown by detrital zir-con geochronology, which identifies North AmericanPrecambrian contributions in the Sustut Group that are absentin Bowser Lake Group samples (McNicoll et al., 2005). Basedlargely on the predominance of chert grains in the Bowser LakeGroup, Eisbacher (1974a; 1981) inferred sources in the CacheCreek terrane, and this source terrane was confirmed by radio-larian fossil ages from chert clasts in the Bowser Lake Group(Currie, 1984; Evenchick and Thorkelson, 2005, and referencestherein). In addition to abundant chert, the Cache CreekTerrane contains alpine ultramafic bodies, and Cookenboo(1993) determined an alpine peridotite source for detritalchrome spinel in the Bowser Lake Group from microprobeanalyses of detrital chrome spinel. Eisbacher (1981), Green(1992) and Cookenboo (1993) inferred Mesozoic arc terranesas contributors of volcanic material noted in Bowser LakeGroup clasts. The predominance of accreted oceanic and arcterranes as sediment sources for the Bowser Lake Group is alsosupported by neodymium isotopic compositions of sedimen-tary rocks from the northwestern Bowser Basin (Samson et al.,1989). In contrast to older units of the Bowser Lake Group,Cookenboo (1993) identified a metamorphic clastic componentin the Devils Claw Formation.

Lithostratigraphic assemblages within the Bowser LakeGroup are diachronous interfingering units representing faciesbelts that occupied different positions in the basin over time. Assuch, the assemblages are expected to have similar sedimentsources, yet some provenance contrasts are identified for thelargely coeval Ritchie-Alger, Todagin, and Muskaboo Creekassemblages. The observed distinctions suggest contributionsfrom sources other than those identified by a simple model ofdetrital transport and deposition from hinterland sources washed


Fig. 11. Ternary diagrams constructed to characterize Eaglenest,Skelhorne, Groundhog-Gunanoot, and Jenkins Creek assemblages,and the Devils Claw Formation: A) using sandstone data; B) using con-glomerate data. Symbols are as in Figure 10.

progressively through proximal (shelf and shoreface) to distal(deeper marine) environments. Some of the contrast may beexplained by the likelihood that most Todagin assemblage sam-ples are probably somewhat older than the majority of Ritchie-Alger and Muskaboo Creek samples. McNicoll et al. (2005)proposed airfall deposition of volcanic material as a significantcontributor to sedimentary sequences within the Bowser Basin,based on U-Pb geochronology of detrital zircons from siltstonesand sandstones having well-constrained paleontological ages.The U-Pb age of the dominant detrital zircon population isequivalent, within uncertainty, to the depositional age of therocks (McNicoll et al., 2005), although volcanic flows and vol-caniclastic rocks in the Bowser Basin are restricted to EarlyOxfordian, and are present only in the south and southeast partof the basin. Contemporaneous volcanism, south, and possiblywest, of the basin depocentre, may have been a significant sed-iment source that contributed an acid igneous provenance sig-nature to portions of the Bowser Lake Group.

In addition to the first-order differentiation of the BowserLake and Sustut groups, geochemical characterization ofmarine vs. nonmarine assemblages of the Bower Lake Group ispossible using relatively simple variables such as Fe2O3, MgOand SiO2/Al2O3. It is also possible to chemostratigraphicallydifferentiate the deltaic lithostratigraphic assemblages(Eaglenest, Skelhorne, Groundhog-Gunanoot) and the nonmarine Jenkins Creek assemblage and Devils Claw Formation.The differentiation of these deltaic and nonmarine units can beconsidered as a third order characterization and it is this differ-entiation that has most potential application in future explo-ration. Differentiation of the lithostratigraphic assemblages is

based on lithologies, small-scale sedimentary structures, con-tained fossils, and macroscopic features such as nature andscale of cyclicity. Most of these features will not be recogniza-ble in cuttings samples, particularly if future explorationemploys fixed cutter (or PDC) bits as is most often the case inmodern hydrocarbon wells. However, the whole rock geo-chemistry will be preserved in the cuttings samples and there-fore, by applying the criteria discussed above, the nonmarineassemblages of the Bowser Lake Group should be geochemi-cally identified from cuttings samples.

CONCLUSIONS

Whole rock geochemistry and heavy mineral analysis are poten-tially very effective tools in frontier basin exploration. An impor-tant general conclusion from this introductory study is thatmineralogical and geochemical variations do exist in the Bowserand Sustut basins, allowing for the application of a chemostrati-graphic approach. The first-order characterization identified hereis the predominantly mafic provenance of the Bowser LakeGroup contrasted with the acid igneous and metasedimentaryprovenance of the Sustut Group. This contrast is evident fromheavy minerals and several element ratios. Chemostratigraphiccharacterization separates sandstones and conglomerates of theBowser Lake Group into marine and deltaic-nonmarine deposi-tional settings, and further characterzation allows geochemicaldifferentiation of each deltaic and nonmarine assemblage. Withthis type of characterization developed from field samples thathave confidently been assigned to a lithostratigraphic unit, it is possible to assign samples of unknown affinity, including


Fig. 12. SiO2 vs. Al2O3 binary diagram for all sandstone and conglomerate samples. Double-headed arrow indicates a silt- or clay-dilution trend.

cuttings samples from hydrocarbon wells, to a chemostrati-graphic, and therefore lithostratigraphic, unit.

Ratios of SiO2 to Al2O3 and provenance indications forsandstones and conglomerates indicate the Sustut Group isexpected to have generally better hydrocarbon reservoir poten-tial than the Bowser Lake Group. Limitations of the currentdataset make this a very broad generalization, and geographicand stratigraphic variations in quality are extremely likely.

The provenance implications of this study are in generalagreement with previous work. Heavy minerals and whole rockgeochemistry identify multiple sources for the Bowser LakeGroup, and this information will be of considerable importanceto models and concepts of basin development. Therefore,acquiring geochemical and heavy mineral data at an early stageof basin investigation is highly desirable not only for the prag-matic issues of unit identification and correlation, but also astool for understanding large-scale basin processes.

ACKNOWLEDGMENTS

This study was initiated by a request from EnCana Corporationto the Geological Survey of Canada to provide a suite of fieldsamples for study. The support of EnCana Corporation in pro-viding geochemical and mineralogical data for public release isgratefully acknowledged by the authors.

Dr. Brian Zaitlin, formerly of EnCana Corporation, was amajor driving force behind the project, and provided manyinsightful comments to early discussions leading to the prepa-ration of this paper. Dr. Tim Pearce of Chemostrat Ltd also sup-plied helpful input into the chemostratigraphic discussionsthroughout preparation of the manuscript. The acquisition andselection of rock samples from storage archives was made pos-sible by the efforts of Jamel Joseph at the Geological Survey ofCanada in Vancouver.

The authors also thank Pierre Cousineau for comments andquestions in his review of this manuscript. This is GSCContribution Number 20070171.

REFERENCES

Cookenboo, H.O. 1993. Lithofacies, provenance, and diagenesis of Jura-Cretaceous strata of the northern Bowser Basin, British Columbia. Ph.D.thesis, The University of British Columbia, Vancouver, British Columbia,216 p.

Currie, L. 1984. The provenance of chert clasts in the Ashman conglomeratesof the northeastern Bowser Basin. B.Sc. thesis, Queen’s University,Kingston, Ontario, 59 p.

Ehrenberg, S. N and Siring, E. 1992. Use of bulk chemical analysis in strati-graphic correlation of sandstones: an example from the Statfjord field,Norwegian continental shelf. Journal of Sedimentary Petrology, v. 62, p. 318–330.

Eisbacher, G.H. 1974a. Deltaic sedimentation in the northeastern BowserBasin, British Columbia. Geological Survey of Canada, Paper 73-33, 13 p.

________ 1974b. Sedimentary history and tectonic evolution of the Sustut andSifton basins, north-central British Columbia. Geological Survey ofCanada, Paper 73-31, 57 p.

________ 1981. Late Mesozoic–Paleogene Bowser Basin Molasse andCordilleran Tectonics, western Canada. In: Sedimentation and Tectonics inAlluvial Basins. A.D. Miall (ed.). Geological Association of CanadaSpecial Paper 23, p. 125–152.

Evenchick, C.A. 1991. Geometry, evolution, and tectonic framework of the Skeena Fold Belt, north-central British Columbia. Tectonics, v. 10, p. 527–546.

________ and Thorkelson, D.J. 2005. Geology of the Spatsizi River map area,north-central British Columbia. Geological Survey of Canada, Bulletin577, 276 p.

________, Ferri, F., Mustard, P.S., McMechan, M.E., Ritcey, D., McNicoll,V.J., Osadetz, K.G., O’Sullivan, P.B., Stasiuk, L.D., Wilson, N.S.F.,Poulton, T.P., Lowe, C., Enkin, R.J., Waldron, J., Snyder, D.B., Turner,R.J.W., Nowlan, G. and Boddy, M. 2005a. Highlights of recent research inthe Bowser and Sustut basins project, British Columbia. Geological Surveyof Canada, Current Research 2005-A1, 11 p.

________, Mustard, P.S., Woodsworth, G.J. and Ferri, F. 2005b. Compilationof geology of Bowser and Sustut basins draped on shaded relief map, north-central British Columbia. Geological Survey of Canada, Open File 4638,scale 1:500 000.

________, Poulton, T.P., Tipper, H.W. and Braidek, I. 2001. Fossils and faciesof the northern two-thirds of the Bowser Basin, northern British Columbia;Geological Survey of Canada, Open File 3956.

Gabrielse, H. 1991. Late Paleozoic and Mesozoic terrane interactions in north-central British Columbia. Canadian Journal of Earth Sciences, v. 28,p. 947–957.

Green, G.M. 1992. Detailed sedimentology of the Bowser Lake Group, north-ern Bowser basin, north-central British Columbia. MSc. thesis, CarletonUniversity, Ottawa, Ontario, 197 p.

Jarvis, I. and Jarvis, K. E. 1995. Plasma spectrometry in earth sciences: tech-niques, applications and future trends. In: Plasma Spectrometry in EarthSciences. I. Jarvis. and K.E. Jarvis (eds.). Chemical Geology, v. 95, p. 1–33.

McNicoll, V.J., Evenchick, C.A. and Mustard, P.S. 2005. Provenance studieson the depositional histories of the Bowser and Sustut basins, and theirimplications for tectonic evolution of the northern Canadian Cordillera.GAC-MAC-CSPG-CSSS Joint Meeting, Halifax, Nova Scotia, AbstractsVolume 30, p. 133.

Morton, A. C. and Hallsworth, C.R. 1994. Identifying provenance-specific fea-tures of detrital heavy mineral assemblages in sandstones. SedimentaryGeology, v. 90, p. 241–256.

________ and ________ 1999. Processes controlling the composition of heavymineral assemblages in sandstones: Sedimentary Geology, v. 124, p. 3–29.

Osadetz, K.G., Jiang, C., Evenchick, C.A., Ferri, F., Stasiuk, L.D., WilsonN.S.F. and Hayes, M. 2004. Sterane compositional traits of Bowser andSustut basin crude oils: indications for three effective petroleum systems.In BCMEM Resource Development and Geoscience Branch, Summary ofActivities 2004, p. 99–112.

Pearce, T. J., Besly, B. M., Wray, D.S. and Wright, D. K. 1999.Chemostratigraphy: a method to improve interwell correlation in barrensequences - a case study using onshore Duckmantian/Stephanian sequences(West Midlands, U.K.). Sedimentary Geology, v. 124, p. 197–220.

________, McLean, D., Wright, D. K., Jeans C. J. and Mearns, E. W. 2005a.Stratigraphy of the Upper Carboniferous Schooner Formation, southernNorth Sea: chemostratigraphy, mineralogy, palynology and Sm-Nd isotopeanalysis. In: Carboniferous hydrocarbon geology: the southern North Seaand surrounding onshore areas. J.D. Collinson, D.J. Evans, D.W. Hollidayand N.S. Jones (eds.). Yorkshire Geological Society, OccasionalPublications series, v. 7, p. 165–182.

________, Wray, D. S., Ratcliffe, K. T., Wright, D. K. and Moscariello, A.2005b. Chemostratigraphy of the Upper Carboniferous SchoonerFormation, southern North Sea. In: Carboniferous Hydrocarbon Geology:The Southern North Sea and Surrounding Onshore Areas. J.D. Collinson,D.J. Evans, D.W. Holliday and N.S. Jones (eds.). Yorkshire GeologicalSociety, Occasional Publications series, v. 7, p. 147–164.

Preston, J., Hartley, A., Hole, M., Buck, S., Bond, J., Mange, M. and Still, J.1998. Integrated whole-rock trace element geochemistry and heavy mineralchemistry studies: aids to correlation of continental red-bed reservoirs inthe Beryl Field, UK North Sea. Petroleum Geoscience, v. 4, p. 7–16.

Racey, A., Love, M. A., Bobolecki, R. M. and Walsh, J. N. 1995. The use ofchemical element analysis in the study of biostratigraphically barrensequences: an example from the Triassic of the central North Sea (UKCS).In: Non-biostratigraphical Methods of Dating and Correlation. R.E.


Dunnay and E.A. Hailwood, E. A. (eds.). Geological Society, LondonSpecial Publication v. 89, p. 69–105.

Ratcliffe. K. T., Wright, A. M., Hallsworth, C., Morton, A., Zaitlin, B. A.,Potocki, D. and Wray, D. S. 2004. An example of alternative correlationtechniques in a low accommodation setting, nonmarine hydrocarbon sys-tem: the (Lower Cretaceous) Mannville Basal Quartz succession of south-ern Alberta. American Association of Petroleum Geologists Bulletin, v. 88,p. 1419–1432.

________, Martin, J., Pearce, T.J., Hughes, A.D., Lawton, D.E., Wray, D.S. andBessa, F. 2006. A regional chemostratigraphically-defined correlationframework for the late Triassic TAG-I Formation in Blocks 402 and 405a,Algeria. Petroleum Geoscience, v. 12, p. 3–12.

Reichenbach, M.E., Gordon, A.A., Evenchick, C.A. and Ritcey, D.H. 2006.Petrographic data and descriptions of the Bowser Lake and Sustut groups,north-central British Columbia, Canada. Geological Survey of CanadaOpen File. 5300, 1 DVD-ROM.

Ricketts, B.D., Evenchick, C.A., Anderson, R.G. and Murphy, D.C. 1992.Bowser Basin, northern British Columbia: constraints on the timing of ini-tial subsidence and Stikinia-North America terrane interactions. Geology, v. 20, p. 1119–1122.

Ritcey, D.H., Evenchick, C.A., and Ratcliffe, K.T. 2005. Geochemical andheavy mineral analyses of the Bowser Lake and Sustut groups, north-central British Columbia, Canada. Geological Survey of Canada Open File5072, 1 CD-ROM.

Samson, S.D., McClelland, W.C., Patchett, P.J., Gehrels, G.E. and Anderson,R.G., 1989. Evidence from neodymium isotopes for mantle contributionsto Phanerozoic crustal genesis in the Canadian Cordillera. Nature, v. 337,p. 705–709.

Stasiuk, L.D., Evenchick, C.A., Osadetz, K.G., Ferri, F., Ritcey, D., Mustard,P.S. McMechan, M. 2005. Regional thermal maturation and petroleumstage assessment using vitrinite reflectance, Bowser and Sustut basins,north-central British Columbia, Geological Survey of Canada, Open File4945; BC Ministry of Energy and Mines, Petroleum Geology Open File2005-2; 1 CD-ROM.

Wheeler, J.O. and McFeely, P. (compilers) 1991. Tectonic Assemblage Map ofthe Canadian Cordillera and adjacent parts of the United States of America;Geological Survey of Canada, Map 1712A, scale 1:2 000 000.

Wray, D. S. 1999. Identification and long-range correlation of bentonites inTuronian–Coniacian (Upper Cretaceous) chalks of northwest Europe.Geological Magazine, v. 136(4), p. 361–371.

Manuscript received: September 19, 2006

Date accepted: July 20, 2007

Associate Editor: Martine Savard


whole-rock geochemistry and heavy mineral analysis as petroleum exploration … · 2019-07-17 ·...

Documents