GEOPHYSICAL

RESEARCH
LETTERS,VOL. 15, NO. 9, PAGES969-972, AUGUSTSUPPLEMENT
1988

ELEMENTAL AND MINERALOGICAL ANALYSES USING GEOCHEMICAL LOGS FROM THE

CAJON PASS SCIENTIFIC DRILLHOLE, CALIFORNIA, AND THEIR PRELIMINARY
COMPARISON WITH CORE ANALYSES

Roger
N. Anderson
1,RoyE.Dove
2,Cristina
Broglia
1,LeonT. Silver3,EricW. James
3 andB.W.Chappell
4

Abstract. Estimates of elemental and mineralogical abundances of Si, Ca, Fe, S, K, Ti, H, C1 and Gd. The
abundancesfrom geochemicallogs are comparedto additionaluseof thephotoelectric capturecross-section of the
preliminary
chemicalandmodalanalyses fromcoresin the formation allows for the calculation of the residual abundance
CajonPassScientific of log-computed of Mg + Na (Hertzoget al., 1987).
Drillhole.Accuracies
weightpercentoxideandmineralogydeterminations range The majorelementcomposition in dry weightpercentof the
from 10 to 30%. oxides is estimated by assuming that in all rocks,
approximately50 % is oxygenby weight (Hertzog et al.,
Introduction
1987). The assumptions of the distributionof oxygensare
somewhat different in igneous rocks than in sediments,
The properinterpretationof the heatflow, stress,andother particularly since the Hertzog et al.'s (1987) model is
geophysicaland geochemicalexperimentsplannedfor the carbonate-based. Therefore,we have modified the oxygen
CajonPassdrillholedependscriticallyuponan understanding distribution coefficients from carbonate to silicate abundances.
of the structure,stratigraphyandlithologyof the countryrock. The gamma-ray spectroscopymeasurementsmade in the
Conventionalgeophysical logsdo not displaygreatcontrast well detecta yield for eachactivatedelementthat is linearly
betweenunitsthat haveporositiesof < 2 % anddensitiesof > proportionalto the concentration
of thatelementin thevolume
2.7gm/cm 3,asisthecase ofthecrystalline
rocks of theCajon of the measurement.Thus, the relative yields from each
Pass basement. We have, therefore, made extensive use of
new geochemical toolsthatmakecontinuousmeasurements of
theconcentrations of majorandtraceelementsin thewell. The
geochemicallogs provideexcellentreconnaissance analyses /r'rO / / ,'COO/ rS ! U
for quick and relatively accurate interpretation of !0 I 510 5010 a,o,
jo . 51;)5 ppm 0
lithostratigraphy(Anderson et al., this issue), structure ....I I...... ......
(Pezardet al., thisissue),andthermalconductivity(Williams
et al., this issue). i rK•0/ ,S,Oa I rF,O I ,A,aO,/
The GeochemicalLogging Tool (mark of Schlumberger) [,o , ,011øø,. Ol_IO__ø15ø 0,,,{
was developed for use in sedimentarybasins by the oil IOOO

industryand is now beginningto be used in igneousand

metamorphic rocksduringscientific
drilling.(e.g.,in theKTB 1500
drillhole in West Germany,in the Siilgen Ring well in
Sweden,andin theOceanDrillingProgram).The dataquality 2000
in crystalline
rocks,however,
hasnotyetbeenverified.
700
The Geochemical LoggingTool carriesthreemeasurement
2500
devicesinto the borehole:a) a naturalgammaray spectral 800
analyzerthatdetectstheenergylevelsof naturallyemitted
gammaraysfromtherock.Theweightpercent of K, aswell 30O0

as the abundances
of Th and U are determinedby this sensor, I000
b) a Californium
neutronsource/NaIscintillation
detector
that 3500
IlO0
activatesAluminum atomsin the formation,and c) a pulsed
neutron,capturegammaray spectrometer
thatmeasures
the 4000 --' . 12(X)

Borehole
Research
Group,Lamont-Doherty
Geological 1400
Observatoryand Departmentof Geological Sciences,
50(03 .• -• 15cx3
ColumbiaUniversity
Schlumberger
WellServices 1600
5500
Department
ofGeological
andPlanetary
Sciences,
California 170(3
Instituteof Technology
Bureauof Mineral Resources,Australia
Fig.1.Log-computedoxidesof themajorelements expressed
Copyright1988by theAmericanGeophysical
Union. in dryweightpercent
andtraceelements (Th,U'andGd,) in
ppm.Reconstructions(labeledr) of theoxidesgivea good
Papernumber8L7340. measureof thegoodness-of-fit
of themineralogical
inversion
0094- 8276/88/008L-7340 $03.00 to theoriginaldata.

969
970 Anderson
et al.: Geochemical
AnalysesfromWell Logs

Table 1. CajonPassOxideandTraceElementAnalyses:
Corevs.Log

Core # C-5 C-10 C-15 C-17 C-18 C-19 C-20

Depth(m) 525.5-525.7 744.4-744.9 1139.1-1140.2 1353-1353.3 1500.6-1502 1655.3-1657 1740-1741.4
Unit # 1-GD 11-GR 12-GR 6-GD 7-GD 14-INT. gabbro

SiO2 (core) 66.4 73.3 71.1 62.9 61.1 67.2 44.7

Oog) 66.8+1.0 73.3+0.2 71.4_-_-t-0.6 60.9-!-_0.5 65.8+0.5 66.4+0.7 50.2+1.5

TiO2 (core) 0.63 0.32 0.46 0.88 1.07 0.56 1.3

0og) 0.9+0.1 0.22_-_-t-0.1 0.7+0.3 1.1+0.5 1.0+0.2 0.8+0.5 1.2+0.1

A1203 (core) 16.0 14.0 14.0 17.3 15.2 15.8 19.0

(log) 12.7+1.5 15.9'•_0.3 13.5+0.3 20.3+2.0 15.0-•_0.2 16.8+0.8 18.2+0.4

FeO (core) 3.3 1.7 3.1 4.3 6.7 3.0 11.5

(log) 4.1+0.5 1.4+0.1 2.6+0.1 3.6+0.1 6.0+0.2 2.6+0.1 9.3+0.6

CaO (core) 3.1 1.8 1.9 4.7 3.5 3.2 11.5

(log) 6.8+1.0 3.4+0.2 1.1+0.6 5.3+0.2 4.7+0.3 4.5+0.3 9.5+0.4

K20 (core) 3.2 4.5 4.8 2.4 3.4 4.3 0.9

(log) 3.3+0.1 3.7+0.1 4.0+0.2 2.4+0.2 3.0•.1 3.6+0.1 1.5+0.1

Na20+MgO(core) 5.1 3.9 3.6 6.0 6.8 4.9 8.9

(log) 4.0+0.6 3.6+0.1 3.5+0.1 5.9+0.1 3.1+0.1 4.1+0.2 10.5+2.0

U (core) 1.91 1.01 - 1.25 1.67 3.82 0.32

(log) 3.8+0.5 1.7+0.2 - 2.0+0.1 2.9__+0.2 4.9+0.2 0.4+0.3

Th (core) 13.3 13.5 - 7.1 14.4 11.0 0.3

(log) 11.0&0.2 12.0-k__0.1 - 7.0+0.1 14.0-!__0.4 13.1+0.1 2.5+1.1

elementmeasuredthroughthermalneutroncapturereactions (Hertzoget al., 1987).Aspartof anextensive geochemical

are renormalized analysis
to elemental oxides which must sum to 100 programonthecorefromtheCajonPass well,Silver
% at eachsamplingdepthinterval(0.15m). Sincetheelements et al., (thisissue)sought
to minimizethesecross-correlation
measured by eithercaptureactivationor naturalspectroscopy difficulties.Great carewas takento insuremaximumrecovery,
make up almost all of the significant oxides of a rock
(particularly when Na and Mg are also estimated), the 8

calculationsshouldbe reliable in any geologicalformation

(Hertzog et al., 1987). We will test the accuracyof these
calculations over the seven cored intervals for which
preliminary laboratory analyses from Cajon Pass are
completed. 40 50 60 70 80 0 3 6 9 12

ElementalAnalysesfromLogversusCore •z5/
•20
AI203
ß
//I
The variations in major and trace element chemistry o1õ ß
determinedfrom the elemental yields measuredby the
geochemical
log areshownin Figure1 for thefirst 1,829m of 5 I0 15 20 25

theCajonPasswell.Thesevalueswerecalculated
withno
calibrationor otherreferenceto coreanalyses. I2 ß 12 12 ß
Laboratory
coreanalysesareoftendifficulttolocaterelative
to logdepthsbecause
of incompletecorerecovery. If a 10 m
intervalis coredbut only 1 m of rockis recovered, thatcore 0 04 08 I 2 I 6 16

couldhavecomefrom anywherewithin the 10 m interval. LOG (wt %)

Alsothe sampling intervalof coresamples is differentfrom Fig.2. Cross

correlation
between
average
loganalyses
and
logs.Thecoreisfroma specificcylinder coreanalyses
thatof geochemical forseven
intervals
ofCajonPass
well(squares).
of rockwhereasthe Geochemical
LoggingTool irradiatesan Bad hole conditionsmake data from core C-5 less accurate
approximately
onecubicmetervolumesurrounding
thehole (soliddots;seeTable 1).
 Anderson
et al.: Geochemical
Analyses
fromWell Logs 971

and > 90% of each interval was indeed recovered. Also, core
sampleswerepreparedfor chemicalanalysisin a specialway
to insurethatthe depthintervalmostcloselyrepresented that
corresponding to the sampleintervalof the geochemical
log. 20 ß

Vertical wedgeswere slicedfrom top to bottomof the entire

coreinterval.Thesewedgeswere thenpowdered,thoroughly • O'
•. =,
0 I0 20 50 40
mixed, and sampledseveraltimes. The elementalanalyses 0 I0 20:50 40

from corearethusaveragesfor theintervalof eachcore(Table

1). The log analyseswerethenaveragedoverthe sameinterval 60
tPLAGIOCLASE,./
HORNBLENDe'
(Table 1); the comparison is shownin Figure2.
The geochemicallogs reproducethe core analyseswith
ß1- I
I--... VOLUME% i--.. Ok' ,
20 30 40 50 60 0 I0 20 3o 40 50 60 70

n..- 0 %0 IOOt•E L06 (%)

I000 .... '-" '

_ .•_ . ._,-3oo Fig. 4. Comparisonbetweenpreliminarymodal analysesof
thin sectionsfrom sevencore intervals (Table 1) and modal
--4OO predictionsfrom themineralinversionof geochemical logging
data.Squaresare from the mostreliable dataintervals,circles
1500--
representdatafrom bad holeintervals.
--5OO

variable
degrees
ofaccuracy
(Table1).SiO2,FeO,K20andU
--600
20O0 providethemostlinearof thecorrelations,althoughU seems
tobesomewhat outof calibration
(slopenot1:1).Badholeis a
--70O significantcauseof errorin any loggingdataset,andthe
geochemical dataappearto be affectedaswell. Discarding
5oo-t- --800
geochemicallogging data from hole that is out-of-size
(determined
fromthefourarmcalipermeasurements: dotsin
Figure2) increasesthe linearityof theA1203 andCaO
-900 correlations.
TiO2 appears to bethemostpoorlydetermined
....
major elementbecauseof its low abundance,but even this
analysis is accurate
to within30%.Th isreasonably accurateat
-IOOO

:3500
concentrations greaterthan5 ppm( 10-15%).
Thoughtheleastaccurate of themajorelemental correlations,
-ILOO Na20 + MgO from theresidualis within20 % of thatfromthe
coreanalyses, exceptfor an inexplicableerrorof 50 % in the
analyses fromcoreC-18 (Table1 andFigure 2).
'- -• ' ' -•------ --1200 Thesesevencoreintervalsareonly the firstof morethan50
..,.

that will be eventually cross-correlatedwith Geochemical

--1300 LoggingTool results.We will await the more extensivedata
setbeforedetermining moreaccuratecalibrationproceedures
4500 --1400
for geochemical loggingin crystallinelithologies.Thesedata
are presentedin their preliminary form to verify that the
geochemical logging data are indeed reliable as a
reconnaissance tool for therapiddetermination
of geochemical
--1500
variationswithintheCajonPassscientificdrillhole.

--1600 MineralogyInversion

5500• 1700
Elementalconcentrations from the geochemicallogscan be
invertedusinga correlationmatrixto determinethevolumesof
ideal mineral compositionsthat might be presentin the well
1800 (Hertzoget al., 1987). We havechosenaveragecompositions
of quartz,orthoclase,
plagioclase(An 45), biotite,hornblende,
0 50 I00
pyrite, sphene,and calcitefor the mineralinversion.When
VOLUME %
microprobeanalysesof theelementalcompositions of minerals
Fig. 3. Mineralogyof the CajonPassscientificdrillhole in the CajonPasscoresarecompleted,mineralinversionwill
determined from an inversion of the elemental abundances be better constrainedthan in this preliminary study. The
assumingideal mineralcompositions.
From left to right: mineralcompositions were thenenteredasend-members in an
yellow=quartz; purple=k-feldspar; red=plagioclase; inversion
matrixalongwiththeweightpercentoxidevaluesof
blue=biotite;
white=hornblende; green=pyrite. Si, A1, Fe, Ca, K, Ti, Na+Mg, and S from the geochemical
yellow=sphene;
972 Andersonet al.: Geochemical
AnalysesfromWell Logs

logs.The resultingbest-fitmineralogyof the well is shownin data.So far, however,the resultsindicatethat the log-derived
Figure 3. chemicalanalysesarean excellentcomplement to geophysical
Preliminarymodal analyseswere done on representative logging, experimentalmeasurements,and core analysesin
thin-sections from the cored intervals for which we have studiesof the CajonPassscientificdrillhole.
chemical analyses.The cross-correlationof core-derived
modes and the predicted mineral volumes from the Acknowledgements.Wewish to thank Honor O'Malley,
geochemicallogs is shown in Figure 4. The log-derived ColinWilliams, andPhilippePezardfor helpfuladviceduring
mineralogiesdisplay somewhatlower quartz and mafic thepreparationof thispaper.ElizabethPratsondevelopedthe
contents(perhaps5 % low) and slightly high plagioclase silicateelementalweight percenttransform.We especially
modes.However,we mustawait the full laboratoryanalyses thank Mark Zoback for conceivingof and carryingout the
of core mineral compositionsbefore adjustingthe input drilling of this magnificent scientific laboratory into the
standards for themineralogyinversion. workingsof the San AndreasFault, and DOSECC and the
A good test of the reliability and self-consistencyof the NationalScienceFoundation for supportingtheeffort.
mineral inversion comes from the reconstruction of elemental
compositions from thebest-fitmineralogymodel.Becausethe References
matrix inversionseeksthe most statisticallysignificantfit of
elementsto minerals,the exactreconstructionof that model-fit Anderson,R. N., C. Broglia, P.Pezard,and C. Williams,
can show where elemental concentrations from the Lithostratigraphy determined
fromdiscriminant analysis
of
geochemicallogs were not fitted well by the ideal mineral geochemicalwell logs from the Cajon Pass Scientific
compositions of the model. Whereas most elemental Drillhole, California,Geophys.Res.Lett., thisissue.
reconstructsare indistinguishablefrom the input values, Hertzog, R., L., Colson,B. Seeman,M. O'Brien, H. Scott,
reconstructed
MgO actuallyagreeswith coreanalysesbetter D. McKeon, P. Wraight, J. Grau, D. Ellis, J. Schweitzer,
thantheinputresidualcurve(Figure1). and M. Herron,GeochemicalLoggingwith Spectrometry
Tools,Soc.Pet. Eng., paper 16792, 1987.
Summary Pezard,P.A., R.J. Weldon, G.R. Ollier, and C. Wilkinson, A
structuralmodel for the sedimentarysectionwithin the
Preliminary evaluation of the comparisonbetween the vicinity of the Cajon PassScientificDrillhole, California,
geochemical log-derived elemental and mineralogical Geophys.Res. Lett., this issue,1988.
abundancesand core analysesshowsthat the log-derived Silver,L. T., E.W. James,andB.W. Chappell,Petrological
chemicalcompositionsare accurateenoughto be extremely and geochemicalinvestigationsat the Cajon Passdeep
usefulgeologicalreconnaissance datain the CajonPasswell. drillhole, Geophys.Res.Lett., this issue,1988.
The elemental weight percent oxide analyses compare Williams, C. F., R.N. Anderson, C. Broglia, A.H.
reasonablywith core analyses;K, U, Fe, and Si being Lachenbruch, and J.H. Sass,In Situ investigationof
determinedmostaccurately. A1,Ca, Th andTi weresomewhat thermalconductivity,heat production,and hydrothermal
less well-determined,and the Na+Mg residualis the least circulationin the CajonPassScientificDrillhole,California,
well-constrainedof the log-derived chemical results. All Geophys.Res. Lett., this issue,1988.
appearto be accurateto within10 to 30 %.
The subsequent mineralogicaldeterminationprovesto be
RogerN. AndersonandCristinaBroglia,BoreholeResearch
extremelyusefulfor severalof themajorstudiesin progress at
Group,LamontDohertyGeologicalObservatory,Palisades,
Cajon Pass. Thermal conductivity determined from the NY 10964.
mineralogyprovidesa continuous descriptionof theheatflow
Roy Dove, Schlumberger
Well Services,365 Canal Street,
variationwith depthwhen combinedwith thermalgradient New Orleans, LA 70130.
measurements (Williams et. al., this issue).Lithologicunits
detem'dned from elemental variations can be used to define the Leon T. Silver and Eric W. James, Department of
Geologicaland PlanetarySciences,California Institute of
lithostratigraphyof the well (Andersonet el., this issue).
Technology,Pasadena,CA 91125.
Furthermore,the structuralinterpretationof the crystalline
B.W. Chappell, Bureau of Mineral Resources,Canberra,
basementcan be constrainedby the mineralogicalchanges Australia.
recordedby geochemical logs(Pezardet al., thisissue).
We mustwait for thefull-scaleinterlaboratory
comparison
to (ReceivedMarch 11, 1988;
determinethe quantitativeaccuracyof the geochemicallog accepted:
May 4, 1988)