Pre Seismic Magnetic Field Measurements

8/7/2019 Pre Seismic Magnetic Field Measurements

http://slidepdf.com/reader/full/pre-seismic-magnetic-field-measurements 1/6

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 86, NO. B2, PAGES 921-926, FEBRUARY 10, 1981

Preseismic nd CoseismicMagnetic Field MeasurementsNear the

Coyote Lake, California, Earthquake of August 6, 1979

M. J. S.-JOHNSTON, R. J. MUELLER, AND V. KELLER

U. $. GeologicalSurvey,Menlo Park, California 94025

The epicenter f the CoyoteLake earthquake M L -- 5.9 + 0.2) of August6, 1979, s locatedwithin anarray of recordingmagnetometers hich hasbeen n operationsince1974.The nearest nstrument,COY,was within 5 km of the epicenter. t was installed n October 1978 and is locatedon sedimentary ock,althoughvolcanicand ultramafic ockswith magnetizations f up to I A/m outcrop2 km to the west.Asecond ecordingmagnetometerwas operated or 18 days,beginning4 daysafter the main event, to rec-ord the latter stages f the aftershock ctivity. Although onger-termmagnetic ield variationswere re-cordedat stationCOY early in 1979relative to other sites n the area, no anomalouschangeswithin thetwo monthsprior to the earthquakewere observedoutside he presentmeasurement ncertaintyof 0.8nT for hourlyaverage ifferences. uring the ate aftershock tage,no magnetic ield changegreater han0.25 nT occurred or more than a day. We conclude hat in contrast o the 2-nT changeobserved efore apreviousM = 5.2 earthquakenear Hollister, California, no demonstrable reseismic, oseismic, r post-seismic ectonomagneticffectwas detected.A reasonable eismomagnetic odel of the earthquake n-dicates hat stationCOY was poorly located o detectstress-generated agneticperturbations rom thisearthquake.Using a magnetizationdistribution ndicatedby modeling he aeromagnetic ata over thearea, we have calculated hat homogeneous hearstress hanges f about 5 MPa or greaterwould havebeen necessaryo produceany observable ffectat COY. This change n stresss precludedby geodeticdata from over the area. However, COY is ideally situated or detectionof electrokinetically eneratedmagneticanomalies.This initial null observation ndicates hat the assumptions sed n the calculationof electrokinetic effects have, in this case, not been satisfied.

INTRODUCTION

A moderate earthquake (M,• = 5.9 _+ 0.2) occurred near

Coyote Lake, California, on the Calaveras ault on August 6,

1979, at a depth of about 10 km. The earthquake generated

minor surface upture (-.•5 mm) along about 15 km of the fault

trace to the southeastof Coyote Lake [Herd et al., 1979]. Af-

tershocks ccurredat depthsof from 4 to 12 km along a 21-

km fault segment,also to the southeast f the main shock [Lee

et al., 1979]. The radiation pattern of the main shock ndicatesa predominantly strike-slip ocal mechanismwith a rupture

direction rom northwesto southeastLee et al., 1979;Arch-uleta, 1979]. The seismicmoment, seismic stressdrop, and

length of the aftershockzone were about 0.5 x 10 8 N m, 1

MPa, and 21 km, respectively Lee et al., 1979].

The earthquake occurred within an array of continuously

recording magnetometers nstalled since 1974 as part of a

search or earthquakeprecursors nd indicationsof active ec-

tonic behavior [Johnston t al., 1976]. For the only previous

earthquake of magnitude greater than 5 in this area, Smith

and Johnston 1976] reported anomalousmagnetic ield varia-

tions of about 2 nT within a 2-month period prior to this

earthquake in November 1974. The August 6, 1979, Coyote

Lake earthquake provided an opportunity to detect similar

magneticanomaliesbefore or coincidentwith the earthquake

and to test the different modelsof magnetic field generationnear active faults [Stacey, 1964; Talwani and Kovach, 1974;

Mizutani et al., 1976;Johnston,1979;Fitterman, 1979].

OBSERVATIONS

The locations of recording magnetometers n the central

Californiarea re hownn he/insetfFigure .Becauseheinstrumentseparation n the region of the Coyote earthquake

is quite large, only one instrument,COY, was recordingwithin 10 km of the epicenter.The area around the earth-

This paper s not subject o U.S. copyright.Publishedn 1981 bythe American GeophysicalUnion.

quakeepicenter nd the locationof the instrument OY isshown n Figure 1. The aftershocks ll occurredwithin thedashed reaapproximately 1 km longand4 km wide.All theinstruments hownoperate t 0.25-nT sensitivity, nd the digi-tal data are telemetered o Menlo Park at one sampleevery 10

min. The instrumentCOY was installed n September 1978.

The portable nstrumentCIY was nstalled daysafter themain shock t a pointwherewe calculated n expectedmaxi-

mum coseismicield perturbationdue to changesn stress e-suiting from failure.

The simplest method of isolating local magnetic field

changesand reducing he effectsof ionospheric nd magneto-

spheric isturbances,nd,most mportantly,he generaldiur-nal variation, is to difference he magnetic ield observations

between adjacent stationsno more than a few tens of kilome-

ters apart. In particular, we wish to search for and isolate

changes hat might have occurredat COY, the magnetometer

nearest o the Coyote Lake main shock.Figure 2a shows he

magnetic ield differencesn 10-min samplesbetweenstationCOY and the surroundingstationsMTH, EUC, and QSB for

the period around the time of the earthquake. For com-

parison, the differences excluding COY are also shown.

Clearly, the local magnetic ield did not changesignificantlycoincidentwith or immediately before the earthquake.

We have applied several methods for noise reduction to

thesedata to improve discriminationof signals hat may be of

tectonic origin. The simplestof these methods s a variation

on the weighted-difference echnique [Rikitake, 1966]. In our

application his technique ocuses rimarily on the short-termnoise in the difference data that arises rom incomplete can-

cellation of the diurnal variation. By plotting the data from

other sitesagainst hose rom COY during the daily interval

of significantdiurnal variation, a mean-response ifferenceor

'weight factor' was obtained by a least squares ine fit. Figure

2b shows he result of applying this technique o data fromCOY and MTH.

Paper number 80B1098. 921



922 JOHNSTON T AL.: MAGNETIC MEASUREMENTS•COYOTEAKE EARTHQUAKE

/•11111112,.x x / • N• N • STATION

, .•//I/A' AUG , 197& • • )•, .• • '•MTH

• • • L • I SJ•HOLL STER

•/b coyI 1X/•X• • • I

-37o00 '

GILROY \ \ \

\ '.• \\ \

VOLCANICND \ • \77-/] LTRAMAFICOCK \ \ \

k 9 AFTERSHOCKONE • .,••SAN I

I, I, •. •, I, I • • •FEL/PE0 5 km N •LmKE

121•5 121•0 • • / 121•25i i I

Fig. 1. Map of centralCalifornia howingocationsf continuouslyecording agnetometers.he ocation f the Au-gust earthquakesshownsanasteriskn theexpandedectionogether ith heaftershockone, eneralizedeology,and he ocations f the permanenttationCOY and he emporary tationCIY. The profileAA' is used n the text formodeling the magnetic structure.

The upper plot in Figure 2b shows, or comparison, he rawmagnetic field differences rom Figure 2a. The standard de-

viation is 0.8 nT. The middle plot shows he weighteddiffer-

encesaccording o the expression

b(COY) - (MTH) + c

whereb and c, obtained y leastsquaresine fittingon 30 daysof data, are 1.038and -94.4 nT, respectively.he standard e-

viation for thesedata is 0.65 nT. This represents 19% m-

provement in the standard deviation using weighted differ-

ences.The bottomplot shows six-point i.e., hourly) averageof the weighted-differencedata. The standard deviation here

is 0.44 nT. We note that the magnetic field variation at the

time of the August 6 earthquake s not significantat the 95%

confidenceevel if we use a data segmentonger han 4 days(Figure 2c).

The aftershockactivity of the Coyote Lake earthquakede-

creased apidly after the main event [Lee et al., 1979].By thetime the instrument CIY was installed 4 days after the main

than0.25nT overperiodsonger han day,eitherof system-atic form or coseismic ith any aftershocksuring he latterstageof the aftershock ctivity.

DISCUSSION

It hasbeenargued hat magneticieldchanges ightbe ex-pectedpreseismically,oseismically,r postseismicallyn thebasis of two primary physical mechanisms.These are the

seismomagneticffect Stacey, 964; hamsi ndStacey, 969;Staceyand Johnston, 972; Talwaniand Kovach, 1972;John-

ston,1978]and the electrokinetic ffect Mizutaniet al., 1976;Dmowska,1977;Fitterman,1979].The seismomagneticffectis derived rom the stress ependencef the magnetic roper-tiesof rocksand the electrokineticffect rom streaming o-

tentialsset up by pore pressure ariationsnear active faults.Regarding seismomagnetic ffects, we can calculate the

form and amplitudeof magnetic hanges xpected t COY asa resultof finiteslipon the Calaverasault usingmodels imi-lar to thoseof Stacey 1964],Talwani ndKovach 1972],and

event, his activitywas ess han 25%of the original ate of Johnston1978]. he surface nomaly t a pointon theearth'sabout 150 earthquakes er day, and no earthquakes ccurredsubsequentlywith magnitudes greater than 3.6. Figure 3shows 9days'ofmagneticielddifferencesetween IY andCOY, togetherwith the aftershockactivity. The standard de-viation of these differences is 0.50 nT.

Because hese stationsare much closer ogether, he mea-

surement esolution s much higher. However, we are still un-

able to determineany significant hangesn thesedata greater

surfaces a functionprimarilyof the fault geometry,he dis-tributionof magnetization, nd the change n stress tate nthe region. ollowing ohnston1978],we calculatedn upperestimate f the seismomagneticffectat COY usinga modelof a finite slip patch rom I to 11 km deepand 21 km in ex-tent. The fault slip was estimatedas 10 cm from the seismic

momentof from 0.5 to 0.6 x 1018 m reported or thisearth-quake by Lee et al. [1979].We assumedhat the magnet-



JOHNSTONET AL.: MAGNETIC MEASUREMENTS---COYOTELAKE EARTHQUAKE 923

AUGUST 6 1979EARTHD'UAKE

COY-UC

COY-SB

.Oz

I•1

,MTH -QSB

EUC -QSB

I I I

4 5 6 7 8

AUGUST, 1979

Fig. 2a. Magnetic ielddifferencesetween ata ecorded t COYand variousother stationswithin a 2$-km radiusof the epicenter or 2

daysbefore nd dayafter he 1979 arthquake.tandard eviationsof theplotsare 0.8, 0.8, 0.7, 0.8, 1.3,and 1.25nT, respectively.

ization was 1 A/m and that the remanent and induced com-

ponents ere n thesame irection. hissolutions llustratedin Figure4. Clearly, f thissolutions correct,he locationofstation COY for this earthquakewas unfortunate. On theother hand, if further stress elease occurred over the same

area, then C1Y would be ideally situated o detect t if the as-

AUG. 6EARTHQUAKE

COY-MTH10IN.)

COY-MTHWTD.0 IN)

COY-MTH WTD.1HR.)

I I I I I

4 5 6 7 8

AUGUST, 1979

Fig. 2b. Noise eductionn weighted ifferencesetween10-minrecordsobtained betweenCOY and MTH. The weightswere calcu-lated from a leastsquares it of COY data to MTH data. The lowerplot shows ourly averages f weighted ifferences.

AUGUST 6,1979COY- MTH EARTHQUAKE

_ COY- QSB

MTH-EU

MTHQSB

I

1978 [ 1979

Fig. 2c. Five-point running averageon daily averagevalues ofdifferentialmagneticdata for the stationsaround the Coyote earth-quake.

sumptionsof uniform magnetizationand geometryare cor-rect. Taking the geometryof the slip patch to be from only 4to 11 km deep,as ndicatedby the aftershock istribution Leeet al., 1979], he calculatedanomaly n Figure 4 is attenuated

slightly and the contour pattern is extended. ncreasing he

I

i0 15 20 25 $0AUGUST 1979

oz<[z

LLI

4:2 I I I I I I I I I I I I I I I J I I I I"' I0 15 20 25 ;.'50AUGUST 1979

Fig. 3. Ten-minute (top) and daily-average middle) magneticfield differences etweenCIY and COY for the period August 10 toAugust 8, 1979.The magnitudesndoccurrenceimesof aftershocksduring hisperiodwhosehypocenters erewithin 10 km of COY areplottedat the bottomof the figure.



924 JOHNSTON ET AL.: MAGNETIC MEASUREMENTS•COYOTE LAKE EARTHQUAKE

MAGNETIZATION = I A/m(UNIFORM)

L = 21.8 krnD= Ilkrn

d -- Ikrn

u = IOcm

\ 0 I0 km

• I ,,,,•,,,,I\

\\ x•x•x

..0\\ +37O'Ni

I

//

COY

GILROYt

/ \I \

I

\ Ix I

• /+ '. •./ +36o50'

iZio40, '"'_ • 121o20,

FiS. 4. MaSactic anomaly •mours (in nanotcslas)orslip on a fault •twcca I and 11 • deepby 2ncousmaterial with a normal masnctization f I A/re. The Ausust6

ca•bquake epicmet is marked with an asterisk.

fault slip from 10 cm to 33 cm, as indicatedby the geodeticdata to be the total amount of slip on the fault from May to

September 979 King et al., 1979],correspondinglymplifiesthe calculatedanomaly at COY by about 5 to 0.3 nT. We con-clude, therefore, that reasonablevariations in geometry and

fault slip still appear nsufficiento generate measurable o-seismic ignalat station COY.

As regards he locationof COY for the present ault geome-try, other parameters hat might controlwhethera piezomag-netic anomaly could be generated re the distributionof mag-

netization or a more complex and heterogeneous tressstatethan has been assumed.The surfaceanomaly field could be

greatly perturbed f the distributionof magnetizationwerenonuniform. Some indication of this actual distribution in this

area, albeit an underestimate,can be obtained by surfacesam-

pling. Observed alues anged rom 0.1 to 0.001 A/m on theeast side of the fault near COY, from 0.5 to 2 A/m on the west

side of the fault across he volcanicand ultramafic rocks Fig-

ure 1), and from 0.1 to 0.001 A/m further o the west.Aero-magnetic urveys t a heightof 914 m [U.S. Geological urvey,1974] indicate regional magnetic anomaliesabove the vol-canic and ultramafic rocks of about 100 nT. A better in-

dicationof the regionalmagneticstructure an be obtainedby

determining he simplestdistributionof magnetization hat

generates n anomalywhich best fits the observed egionalanomaly.The observed ast-west rofile hroughstationCOY

(A-A', Figure 1) is plotted in Figure 5 togetherwith a calcu-

lated profile at the same elevation obtained by assuming he

magnetizationdistributionof the model magneticstructure l-

lustrated n the lower part of the figure. Although other simi-

lar distributionscould be chosen, he necessary eature ofthesedistributions s that the dominant near-surfacemagnetic

material occurson the west side of the fault, as indicated by

the surfacegeology.The simplest approximation to this distribution for a

seismomagnetic alculation is a magnetizationof 1 A/m onthe west side of the fault and 0.1 A/m on the east side. The

solution or this case,with all other parameters he sameas in

the previous ase Figure 4), is illustrated n Figure 6. The co-seismicchange now expectedat COY is still too small to be

detectedunambiguously.A further difficulty, hat C1Y is also

poorly locatedwith respect o the anomaly, s now apparent.

A better fit to the observedaeromagnetic nomaly requires

the inclusionof more complexity n the magnetizationpatternto the west of the fault. This does not seem warranted at this

point, since t is unlikely to changesignificantly he expectedanomaly at COY. We note that the changes n mean shear

stress n the fault necessaryo generatean observable oseis-mic anomaly greater than 1.5 nT at COY for the two cases

considered re about 21 and 5.2 MPa, respectively. ollowing

Chinnery 1963], hesechanges ould be producedby uniform

slip in the two casesof 150 and 37 cm, respectively.The first

value is certainly precluded by geodeticobservations ver the

area [King et al., 1979]. The secondcase s possible f com-

parable preseismic r coseismic lip occurredand may be in-

dicatedperhapsby the longer-termchangeearlier n the year.

Mizutani et al. [1976] suggested hat electrokinetic effects

may be associatedwith active faulting. Fitterman [1979] re-

cently presented he most complete ormulation and modeling

of this effect.According o Fitterman's 1979]model, the maxi-

mum magnetic ield perturbation should be recordedaround

the center of the rupture zone with an amplitude of up to 10nT dependingon the electrical conductivities, treamingpo-

tential coefficients, nd pore pressure hanges hosen.Thus if

electrokinetic ffectsoccur,site COY shouldbe ideally located

to detect hem. However, sinceno changewas apparentlyob-

served n thesedata, the assumptions sedby Fitterman [1979]

100._•

• •oo

• 40

• o

Profilelong A

.. --•,P ..... Calculated(914m)

.:.//•i -- Observed914m)

.,,..../eee/''....•.ß tationOYM' • , , , , , I , m -- I ... I''-•.m

!I

,,I

J:0.01im'•J-1A/m•=0.01A/m=lkm• -

J 0.]A/m•,,,/ J=0.]/rn D=krnJ=0.] A/m

D=Skrn

Fig. 5. Observedaeromagnetic nomaly (solid curve) at 914 malong the profile A-A' (Figure 1). The calculated nomaly dottedcurve) is for the model magnetic structure shown in the lower dia-gram. ,



JOHNSTON ET AL.: MAGNETIC MEASUREMENTS•COYOTE LAKE EARTHQUAKE 925

\ \

\

\

\

MAGNETIZATION '- I A/m

( WEST SIDE ONLY

L = 21.8 km

D= Ilkmd = Ikmu = IOcm

0 I0 km

$

x2

37 O'N

CO YOT•' I A t(•'

t GLROY// \

/ \/ \

/ \I \

\0

\ I

.0 x //\ /\ /

\ /

IzI 40' /

7ø00'N

$6ø50'N121ø20'W

Fig. 6. Magneticanomalycontours in nanoteslas)or the model n Figure 4 but for which the magnetization n the eastside of the fault is 0.01 A/m.

have, n someway, not beensatisfiedor this earthquake.Theleast well known parametersare the sourcegeometry, the

streaming otentialcoefficients,nd the changesn pore pres-sure.This argumentcan be pursued n a differentway if we

questionwhether he longer-termmagnetic ield changese-

cordedearlier n 1979were generatedby electrokinetic ffects.If so, the electrokineticmodels would indicate that electric

fieldsexceeding 00 mV/km shouldhave accompaniedhesemagnetic ield changes. nfortunately, lectric ield measure-ments are not made in the area. However, about 25 km to the

south,changesn electric ield greater han a few millivoltsper kilometerapparentlydid not occur T. Madden, personalcommunication, 1980).

It is not easy o quantify the relation, f any, between he

longer-termmagnetic ieldvariations ecorded t COY earlierin 1979 Figure2c) and those round he time of the August6earthquake. y referring ll data to a singledistantstation, tis easy o show hat these onger-term ariations ccuronly atCOY. However, he variationsdo appear o reflectpart of the

annual cycle, and until several more years of data are ob-tained, a conservative pproach o their interpretation seems

appropriate.

CONCLUSIONS

After examining he data from the single ecordingmagne-tometersituatednear the August6, 1979,Coyote Lake earth-

quake n somedetail,we concludehat, n contrasto the No-vember 1974 Hollister, California, earthquake, no anomaly

precedinghis earthquake an be identifiedwith any signiti-canceas a precursor.Although some onger-termmagneticfield changes id occurearlier n 1979, t is unclearwhetherorhow these changes elate to longer-term fault activity. We

note the following implications regarding either piezomag-

netic or electrokineticmechanisms or tectonomagnetic f-fects:

1. I n termsof a piezomagnetic xplanation, he absence f

any clear observation f magnetic ield changecould be dueto poor ocationof COY with respect o the subsequent arth-



926 JOHNSTONET AL.: MAGNETIC MEASUREMENTS•COYOTELAKE EARTHQUAKE

quake slip plane, nsufficientmagneticmaterial,or a marginalchangen meanstress tate.For the present eometry ndlikely distribution of magnetization, a mean shear stress

change n excess f 5 MPa probablywould be required o ob-tain a significantobservation. he absence lso of any post-seismic changes on a second recording magnetometer in-

stalled4 days after the main event at the point of maximum

expectedsignal ndicates hat postseismic tress ariation, freflected in these data, was minimal after that time. We note

that the frequency and magnitudesof aftershocks ecayed

more rapidly than expected or an earthquakeof this magni-tude [Lee et aL, 1979].During the operatingperiod, no after-shocks with Mr > 2.6 occurred within 10 km of the station,

and the daily number of aftershockswas less han 25% of the

original rate.2. It appears hat the stationCOY was deally situated or

the detection of electrokinetic effects. The absenceof any

clear observation n thesedata indicates hat the assumptionsused in the calculations of these effects have not been satisfied

for this earthquake.

Acknowledgment. We thank D. Woodward for use of his mag-netic modelingprogram.

REFERENCES

Archuleta, R. J., Rupture propagationeffects n the Coyote Lakeearthquake abstract),Eos Trans.AGU, 60, 890, 1979.

Chinnery, M. A., The stress hangeshat accompany trike-slip ault-ing, Bull. Seismol.Soc.Am., 53, 921-932, 1963.

Dmowska, R., Electromechanical henomenaassociated ith earth-quakes,Geophys. urv.,3, 157-174, 1977.

Fitterman, D. V., Theory of electrokinetic-magnetic nomalies n afaulted half-space,J. Geophys. es., 84, 6031-6041, 1979.

Herd, D. G., R. J. McLaughlin, A.M. Sarna-Wojcicki,W. H. K. Lee,R. V. Sharp, D. H. Sorg, W. D. Stuart, and P. W. Harsh, Surfacefaulting accompanying he August 6, 1979, Coyote Lake earth-

quake:Secondaryault movement?, os Trans.AGU, 60, 890, 1979.

Johnston,M. J. S., Local magnetic ield variations nd stress hangesnear a slip discontinuity n the San Andreas ault, J. Geomagn.Geoelectr., 30, 511-522, 1978.

Johnston,M. J. S., B. E. Smith, and R. J. Mueller, Tectonicexperi-mentsand observationsn westernU.S.A., J. Geomagn.Geoelectr.,28, 85-97, 1976.

King, N. E., J. C. Savage,W. H. Prescott, nd M. Lisowski,Geodeticmeasurementsear the epicenter f the CoyoteLake earthquakeof

August 6, 1979, Eos Trans.AGU, 60, 890, 1979.Lee, W. H. K., D. G. Herd, V. Cagnetti,W. H., Bakun,and A. Rap-

port, A preliminarystudyof the CoyoteLake earthquake f August6, 1979,and its major aftershocks, .S. Geol.Surv.OpenFile Rep.,79-162, 43 pp., 1979.

Mizutani, H., T. Ishido, T. Yokokura, and S. Ohnishi, Electrokineticphenomenaassociatedwith earthquakes,Geophys.Res. Lett., 3,365-368, 1976.

Rildtake,-T., Elimination of non-local changes rom total intensityvaluesof the geomagneticield, Bull. EarthquakeRes. nst. TokyoUniv., 44, 1041, 1966.

Shamsi, ., and F. D. Stacey,Dislocationmodels nd seismomagneficcalculationsor California 1906and Alaska 1964earthquakes, ull.Seismol. Soc. Am., 59, 1435-1448, 1969.

Smith, B. E., and M. J. S. Johnston,A tectonomagnetic ffect ob-servedbefore a magnitude 5.2 earthquakenear Hollister, Califor-nia, J. Geophys.Res., 81, 3556-3560, 1976.

Stacey,F. D., The seismomagneficffect,PureAppl. Geophys.,8, 5-22, 1964.

Stacey, . D., and M. J. S. Johnston, heoryof the piezo-magneticf-fects in titanomagnetite-bearingocks, Pure Appl. Geophys., 7,146-155, 1972.

Talwani, P., and R. L. Kovach, Geomagnetic bservations nd faultcreep n California, Tectonophysics,4, 245-256, 1972.

U.S. Geological Survey, Aeromagneticmap of parts of San Jose,Santa Clara and San Francisco1ø by 2ø quadrangles,Open FileRep., 74-79, 1974.

(Received March 24, 1980;accepted uly 11, 1980.)

Documents

Pre Seismic Magnetic Field Measurements