JOURNAL OF GEOPHYSICAL RESEARCH,VOL. 95, NO. Bll, PAGES 17,459-17,474,
OCTOBER10, 1990
The 1967CaracasEarthquake:FaultGeometry,Directionof
RupturePropagation
andSeismotectonic
Implications
GERARDOSU,•EZ
Institutode Geofisica,
Universidad
NacionalAut6norna
deMgxico,MexicoCity
JOHNN,Xf3h.•-:K
Collegeof Oceanography,
OregonStateUniversity,Corvallis
The faultplaneorientation
of the July30, 1967,Caracas
earthquake
(Mw=6.6)hasbeena sourceof
controversy
for severalyears.This strike-slipeventwasoriginallythoughtto haveoccurredon an east-
westoriented
faultplane,reflecting
therelativemotionbetween
theCaribbean
andSouthAmerican
plates.
Morerecently,however,thecomplexseismicradiationfromthiseventwasinterpreted
asbeingindicative
of a north-south
strikingfaultthatrupturedalongthreeen echelonsegments.
In thisstudywe synthesize
evidencebasedon the intensityand damagereports,the distributionof aftershocks,
and the resultsof a
jointformalinversion
of theP andSH wavesandshowthatthesedataclearlyindicate
thattherupture
of the
1967 earthquake
occurredon an east-west
trendingfault system.Usinga mastereventtechnique,
the
largestaftershock,
whichoccurred
40 minafterthemainevent,is shownto lie 50 km eastof theepicenter
of the mainshock.The epicentraldistances
of smallaftershocks
registeredin Caracas,basedon the S-P
arrival time differencesand the polarizationsof the P waves, are also consistentwith these events
occurringon an east-westorientedfault systemnorthof Caracas.A joint inversionof the teleseismic
P and
SH waves,recorded
on long-period
seismographs
of theWorld-WideStandardized
Seismographic
Network,
showsthat in a time frame of 65 s, four distinctburstsof seismicmomentrelease(subevents)occurred,
witha totalseismic
moment
of 8.6x 10•8N m.Thefirstthreesubevents
triggered
sequentially
fromwest
to east, in a directionthat is almost identicalto the east-westtrendingnodal planesof the source
mechanisms.The average depth of these three subeventsis 14 km. The fourth, and last identifiable,
subeventof the sequence
showsa reversefaultingmechanism
with the nodalplanesorientedroughlyeast-
west.It occurredat a 21-kmdepth,about50 km to the northof the fault zonedefinedby the strike-slip
subevents.This fourth subeventappearsto reflect compressional
deformationof the southernCaribbean,
possiblyrelatedto underthrusting
alongthe proposedCuraqaotrench.The complexityof the fault system
causingthe 1967 earthquakesuggeststhat the relativemotionalongthe Caribbean-South
Americaplate
boundaryin central Venezuela is taken up over a broad, highly faulted, and highly stressedzone of
deformationand not by a simple,major throughgoingfault.
INTRODUCTION
Although the tectonic fabric in north central Venezuela is
On July 30, 1967, parts of the city of Caracasand several
neighboring
coastaltownswere severelydamagedby a moderate-sizedearthquake (Mw=6.6). Although the so-called
Caracas
earthquake
was not a greatplate boundaryearthquake,
it is one of the largest event to have occurred along the
Caribbean-South
Americanplateboundaryin centralVenezuela
sincethe magnitude8.4 (Ms) earthquakeof 1900 [Richter,
1958].Thus the Caracasearthquakerepresentsan important
pieceof evidencefor understanding
the tectonicregimeand
seismic
hazardof thiscomplexplateboundary.
Thefaultplanesolutionof theCaracasearthquake
basedon P
wavefirst motions[Molnar and Sykes,1969] showsa nearly
purestrike-slipmechanismwith nodalplanesorientedN10øW
andN80øE,respectively.Molnar and Sykes[1969] originally
interpreted
this earthquakeas occurringon an east-west
trending,right-lateralstrike-slipfault, reflectingthe relative
motionbetweenthe Caribbeanand South Americanplates.
Rial [1978],however,suggested,
on thebasisof bodywave
modeling,
thatthe Caracasearthquake
occurredon a seriesof
en echelonfaults strikingnorth-southand with a left-lateral
motion. Rial [1978] reached his conclusionbased on the
interpretation
of theasymmetry
in theobserved
waveforms.
Copyright
1990by theAmerican
Geophysical
Union.
Papernumber89JB03008.
0148-0227/90/89JB-03008
$05.00
dominatedby a seriesof right-lateralstrike-slipfaults oriented
roughly east to west [Rod, 1956; Schubert, 1981; Soulas,
1986] (Figure 1), the region near Caracasexhibits a complex
faulting pattern including faults oriented in a NNW-SSE
direction (Figure 2). This complexity may be related to the
bend in the plate boundary that takes place in the vicinity
(Figure 1). The motivation for reanalizing the Caracas
earthquakecomes from the observationthat althoughrupture
on faults oriented north-southis quite feasible in this tectonically complex area, the solutionproposedby Rial [1978]
does not explain adequately the intensity distribution and
damage,nor the location of aftershocks.
Furthermore,asidefrom beingan importantdatumin an area
of low backgroundseismicityand of relatively short seismic
history, the Caracas earthquake is also a peculiar and
particularly interestingearthquakefrom the point of view of
seismicrupture propagation.As Rial [1978] pointedout, the
ruptureprocessof this earthquakeis very complex. The body
wave trains last up to 100 s, almost5 times as long as would
be expectedfor a shallow earthquakeof that magnitude,and
show severaldistinctsubevents.The studyby Rial [1978] is of
particularinterestbecauseit represents
the first attemptto use
body wave modelingin orderto investigatethe ruptureprocess
of a complex event.
The purposeof thiswork is to furtherthe studyof the 1967
earthquake
initiatedby Rial [1978].In orderto characterize
the
time history and geometry of the rupture process, we
17,459
17,460
SU,•,REZ
ANDN,•,B•LEK:CARACAS
EARTHQUAKE
13øN
/
•
/
-''- '•'- - - •' "''-'--'•,-....
/ ,e
•
..,,m._
'"•,,.
•>BAR
BA
DOS
0
•"
BONAIRE
&
SEA
GRANADA
EBASTiANo...
• • LAR,&
/k
I0 •
2'x,///'
COLOMBIA
... i
75 ø
,,'/
\
/
/I
...
,oo _
VZUELA
•
•
/'"'. !
/
•uS
•'•_•
',_
•
•
GUYANA
SHI ELD
70 ø
I
o
J
HISTORICAL
Io
,
•,o
I Km
'
65 ø
6.4•
M• 7.0
I
60 ø
Fig. 1. Regionalseismotectonic
settingof the southernCaribbean.
The main fault zonesaccommodating
the motionof
the SouthAmericanplaterelativeto the Caribbean
are shown.The dateson the faultsareof largehistoricalearthquakes.
The
subructionzoneto the northof BonaireandLos Roquesis thatinferredby Talwaniet al. [1977] (modifiedafter Y. P.
Aggarwa! (personalcommunication, 1987)).
synthesizeevidence from the damageand intensityreports,the
epicentral location of the mainshock, the distribution of
aftershocks, and the geologic information. Furthermore, we
presenta detailed study of the body waves using a formal least
squares inversion algorithm [Ndb•lek,
1984]; this
deterministic approach eliminates many of the resolution
problemsencounteredin a trial-and-error,visual matchingof
the observedand syntheticwaveforms.The main conclusionof
our study is that, contraryto Rial's [1978] initial suggestion,
the 1967 earthquakeoccurredon a set of east-westtrending
fault planes.
On average, the background seismicity along this plate
boundaryis low, probably due to the low relative velocity
between the two plates. Nonetheless, large historical
earthquakes
have occurredalong varioussegmentsof this fault
system [Centeno-Grau, 1940]. Great earthquakesprobably
occurredon the E1Pilar fault in 1530 and 1766 [Centeno-Grau,
1940; Grases, 1980], and the great earthquakeof 1812 caused
destructionalong the Bocono and Moron faults, from Merida
to Caracas(Fiedler [1961] mappedintensityIX in both cities).
Active Faults in North Central
Venezuela
The Moron fault zone that presumably defines the plate
boundarybetweenthe Caribbeanand SouthAmericanplatesin
centralnorthernVenezuela, near Caracas,is not a single,wellVENEZUELA
definedstrike-slipfault but is a systemof en echelon,rightRegional Plate Tectonic Setting
lateralfaults, orientedroughly east-west[Rod, 1956;Schubert
The tectonic structure of northern Venezuela reflects a
and Krause, 1984; Soulas, 1986] (Figure 2). Moreover,there
complexgeodynamicregime in which the major geologic are somefaultswith a NW-SE trend showingboth strike-slip
featuresare a consequence
of the relativeplatemotionbetween andnormalfaulting[Schubertand Laredo,1979;Singer,1977]
the South American and Caribbean plates [Barazangi and (Figure 2). The complexity and segmentationof the plate
in thisregion,comparedto the morelinearandwellDorman, 1969; Burke et al., 1984;Dewey, 1972;Molnar and boundary
Sykes,1969;Schubert,1981, 1984; Schubertand Krause, definedfaults to the east (El Pilaf fault) and to the SW (Bocono
TECTONICSAND SEISMICITY OF NORTHERN
the sharpchangein strike
1984;Sykesand Ewing,1965](Figure1). The Caribbean
plate fault), probablyis a consequence
(about45ø) betweenthesetwo major fault zones(Figure1).
to be about100km
1975; Minster and Jordan, 1978; Steinet al., 1988]. East of Althoughthe apexof thebendis presumed
it is expectedthatthe deformation
produced
the island of Trinidad, the plate boundarychangesfrom a westof Caracas,
subduction
regimein theLesserAntilles,to a systemof right- by thebendwouldcovera large area.
to haveoccurred
in north
lateralstrike-slipfaults.Pdrezand Aggarwal[1981]suggest Thetwo largest,recentearthquakes
thatthe Los Bajos-E1
Soldadofaultsystem(Figure1) defines centralVenezuelaare the 1967 Caracasearthquakeand an event
moves to the east relative to the SouthAmerican plate [Jordan,
the plateboundary
betweenthe LesserAntillessubductionon October29, 1900, for which Richter [1958] estimateda
zone and the Paria Peninsulain Venezuela. West of the Paria magnitude
of 8.4. As Kelleheret al. [1973]pointedout,it is
Peninsula,the right-lateralstrike-slipsystemextendsover difficultto establishwhich segmentof the fault systemis
1200 km and is subdividedinto three major fault zones: E1 responsiblefor the 1900 earthquake.The instrumental
Pilar, Moron,and Bocono(Figure1). The widthof major epicenterdeterminedby Richter [1958] lies 100 km to the east
reported
by
tectonicdeformation
alongthisfault systemis of theorderof of theareaof maximumintensity(aroundCaracas)
Fiedler
[1961].
Furthermore,
the
magnitude
appears
to
be
100km [Garc[aandAggarwal,
1982;Soulas,
1986].
SU,•.REZ
ANDN,•BI•LEK:
CARACAS
EARTHQUAKE
II ø
N
17,461
SOUTH
AMERICA
7/30/1967
IO ø
-._/
ACTIVE
FAULTS
IN
NORTH-CENTRAL
VENEZUELA ( AFTER T. SUBIETA)
(•) MORON
0
25
5o
I
•1
I
km
(•)
(•)
(•
":':':'/•PLIOCENEQUATERNARY(•)
68ow
67 ø
MACUTO
LA VICTORIA
CHARALLAVE
SAN DIEGO
(•) SANANTONIO
(•)
(•)
(•)
(1•
ARAGUA'T'A
TACAGUA
CARAYACA
CHARACAITO
66 ø
Fig. 2. Geologicfaults presumedto be active in north-centralVenezuela(modifiedafter Subieta [1983]). Solid lines
indicatemappedfaultsanddashedlinesinferredfaults.Stippledregionsindicatepresence
of Plioceneand Quaternary
sediments.
The starmarkstheepicenter
of the 1967Caracasearthquake.
largelyoverestimatedconsideringthat the zone of intensityIX
to X is only about 50 km long [Cisternas and Gaulon, 1984;
Fiedler, 1961]. Based on the intensity reports, it is possible
that the rupture of the 1900 earthquakeincluded some of the
samefault segmentsas the 1967 event. The only other major
historicalearthquakethat occurredin this region besidesthe
1812and 1900 earthquakes
is an earthquakein 1641 [CentenoGrau,1940].The fact thatno greatearthquakes
haveoccurred
alongthis plate boundaryduring the last 87 years highlights
theimportance
of the 1967Caracasearthquake.
placesthis fault systemso that it passesthroughCaracasand
would, at the first glance,adequatelyexplain the damageand
intensitiesreportedby Espinosa and Algerrnissen[1972] and
thosewe obtainedby reviewing accountsof the earthquakein
various Venezuelan journals and magazines (Figure 4). To
place this fault systemgeographically,Rial usesa preliminary
epicentral location determinedby Fiedler [1968]. The location
proposed by Fiedler [1968] lies 70 km offshore NNW of
Caracas. It was obtained by measuring the arrival time
differencesof P n wavesat variouspairs of seismicstationsin
the Caribbeanbasin.The epicenterwas determinedgraphically
as the intersectionof the epicentral distance hyperbolaefor
DISTRIBUTION: IMPLICATIONS FOR
the individual station pairs. In spite of drastically different
THE FAULT ORIENTATION
propagationpaths,Fietiler [1968] assumedthat the velocity of
the P n phase is the same for all stations. Given the
EpicentralLocationof the Mainshock
assumptionsused, this graphical method gives only a rough
The epicenterof an earthquake,being determinedfrom estimateof the epicenter.
arrivaltimesof short-periodbody waves,reflectsthe location
The epicenterobtainedby Fiedler [1968] is locatedabout40
of thepointon the faultplanewhereruptureinitiates.Usually, km to the NE from that determined by the International
whenan eventwith a complexfault ruptureprocess
is studied SeismologicalCenter (ISC) (Figure 3). The epicentral locausingthe shapeand durationof the body waves,the locations tions reportedby the ISC, the U.S. Coast and GeodeticSurvey,
of the subevents
that composethe earthquake
are estimated and Moscow show differencesof lessthan 10 km amongthem.
relative to the point of nucleation of the rupture (the The teleseismicstation coverageof the southernCaribbeanis
epicenter).
A goodestimateof the epicenteris necessary
for good, and the errors in epicentrallocationsare generallyless
geographicalplacement of the subevents.As we discuss than about 15 km for earthquakesgreaterthan magnitude5.0
EPICENTER,
INTENSITYREPORTS,
ANDAFTERSHOCK
below,in the caseof theCaracasearthquake,
thechoiceof the that are well recorded worldwide. This observation is confirmed
epicenter
is alsocrucialfor the interpretation
of the intensity by comparingthe epicentrallocationsobtainedby the ISC and
data and the distribution of aftershocks.
by a local network in Venezuela (installed after the 1967
Themodelderivedby Rial [1978]fromwaveformmodeling earthquake),for 15 earthquakesselectedat random (Figure 5).
is composed
of threeen echelonfaultsorientedNNW-SSEthat For these earthquakes,all smaller than the 1967 event, the
rupturedsequentiallyfrom north to south (Figure 3). Rial
differences in epicentral locations between those obtained
17,462
SU•REZ ANDN,•B•LEK:CARACAS
EARTHQUAKE
Teresa,amongothers(Figure 4). Particularlylarge
accelerations
wouldbe expectedin this region,becauseit lies
within 10-15 km awayfrom the third segmentin a north-south
oriented
fault system,whichin Rial's modelhadthehighest
II ø
N
momentrelease.On the otherhand, faulting along the nodal
plane strikingalmosteast-westto the north of Caracas(dashed
thin line in Figure 4) would explain the high intensities
observedin and to the north of Caracas(>VI) and the smaller
values
(< V) observed
in townsto thewestandsouthof thecity
(Figure 4).
Althoughintensityinformationis missingfrom theregions
to the west and southwest of Caracas because these areasare
largely uninhabited,there is sufficient information available
indicatingthat the largest intensitiesoccurrednorth of Caracas
along the Caribbean coast, with intensity decreasing
southwardfrom the coast. Local ground amplification due to
unconsolidatedsediments was possibly responsible for the
IO ø
-
0
2_5
q
50
-
COASTAL
but the distribution of Quaternary deposits(Figure 2) shows
that this factor alone cannotreconcilethe observeddamage
and intensitydistributionwith the fault systemproposedby
Rial [1978]. The constructionof low-rise housingis fairly
km
.....
localizeddamageto high-risebuildingsin Caracas[Espinosa
and Algermissen,1972; Seed et al., 1970; Whitman, 1969],
RANGE
I
67øW
Fig. 3. Location of the subevents(solid circles) and the north-south
trending fault system (solid lines) proposed by Rial [1978]. The
location of the first subeventis made to coincide with the preliminary
epicenterdeterminedby Fiedler [1968] (markedF). The opencirclesand
dashedlines indicate the location relative to the epicenterdetermined
by the International Seismological Center (ISC). Note that when the
fault systemis locatedrelative to the more preciseISC location (error
bars shown), it lies well to the south and west of Caracas.
uniform throughoutthe area and provides a good measureof
variations in intensity (and acceleration) in the macroseismic
zone. The damage sustainedby this type of constructionsis
higher along the coastnorth of Caracasthan it is elsewhere.
Spatial Distribution of Aftershocks
The most direct way for determining the orientationof an
earthquakerupture,other than actually mappingits surfaceexpression,is from the distribution of aftershocks.The Caracas
earthquake,however, had no large enough aftershocksthat
using local instrumentsand the worldwide networkare usually were well recordedby the worldwide seismic networks, and the
local seismographsin Caracas capable of recording smaller
less than 15 km; only three small events (mr,<4.0)show
differences
of 30-35
km. We believe that 40-km distance
eventswere damagedfor a few days after the earthquake.It is
between the location proposedby Fiedler [1968] and that of remarkablethat an earthquakeof the magnitude of the Caracas
the ISC reflectsprimarily the inaccuracyof the methodusedby event, and with suchcomplexhistory and geometry of rupture,
the former.
did not producelarger aftershocks.
The only suggestionfor a moderately large aftershocks
Intensity Distribution and the Orientation
comes from newspaper accounts describing other strong
of the Fault System
quakes shaking the city of Caracas after the mainshock.The
The choiceof the epicenteris crucial for the model proposed largest of these aftershockscaused panic among the populaby Rial [1978], becausewhen the fault systemis movedto the tion and occurred about 40 min after the main event. The ISC
more precise ISC epicentrallocation, his proposedfaults lie reportsthis event as occurringoffshoreat a depth of 114 km.
about 60 km to the west and south of Caracas. With the faults
The location is poor becausethe magnitudeof this earthquake
in
at this location (Figure 3), it is difficult to explain the (mr,=4.4) is near the thresholdof detectionfor earthquakes
observed intensity data (Figure 4). The regions that this region by the worldwide networks and the local stations
experienceddamage and high intensitiesare located near were not operating.It was possible,however,to identify clear
Caracas and along the coast immediately to the north of the first arrivals of the aftershock on several stations in the
capital. If isoseismalcurveswere drawn on Figure 4, they Caribbean (Table 1). We relocated the aftershock relative to
would show a semicirculardistributioncenteredroughly north
of Caracas,with intensitiesgraduallydecreasingaway from the
city. It is difficult to explain how cities like Maracay,
the mainshockusing the master event technique of Dewey
[19711.
The relocatedepicenterof the aftershocklies 50 km to the
Turmero, and Valencia, for example, that lie at a similar eastof the mainshock'sepicenter,at a focal depthof 10 km,
distance from the fault system defined by Rial [1978] as andabout10 km offshore(Table2 andFigure6). The location
Caracasand are located also on loosely consolidatedsediments near Caracaaexplainswhy this relatively small aftershock
(Figure2), consistently
showintensities
threeto four degrees (mr,=4.4)was so stronglyfelt and "heard"in Caracasandthe
coastaltowns.The locationof the largestaftershockto the
lower thanthosereportedat the northcoastof Caracas.
In fact, no matterwhich epicenteris usedto geographically east of the mainshockepicenterstronglysuggestsa fault
placeRial's [1978] north-south
trendingfault system,it is
difficult to explainthe absenceof high intensitiessouthof
extending east-west,north of Caracas.
The short-periodinstrumentsat station CAR (Caracas),
Caracas, in the towns of Cua, Ocumare del Tuy, and Santa member of the World-Wide Standardized Seismographic
SU•REZ
ANDN•B•LEK:CARACAS
EARTHQUAKE
MMI
INTENSITIES
17,463
REPORTED
oF
II ø
N
0
25
t
5o
i
i
km
..;....::.:.: ß
TACACAS ß ß
:.
'
30/1967
--...
"-'•i'•'.
-"•*
-,
:........
•
7/
•
.................
;...[.•':""
•:•
•
•
..... -':'•"
'9
;
*
RAY ß
'•q• ..___.
ß
c•T,^
•• •^•../:••c.•- •i'•0x•...........
'......
'-•• '•••••
t
'X
--•
•-
i
- ' ß
;'-
HIGUER•
SANTATERESA
•
• GUACARA
• •• -TURMERO
--
•
'-•
-t
VALENCIA
•-_•'-•-
ANTIMANO
e:•' r'' PETARE
G•ARENAS
'
-- •
_MARACAY
* ....
........:½):
__
,'.•.
ß
eCUA
.
,,.•,O_4.._..•,
•
OCUMARE
DEL
TUY
GUIGUE
IO o
',
e S•N JUAN
68øW
•
',
67 ø
66 ø
Fig.4. ModifiedMercalli intensities(MMI) observedin centralVenezueladuringthe 1967earthquake.
The datacome
fromthestudyof Espinosa
and Algermissen
[ 1972]andfromour ownreadings
of Venezuelan
journals,magazines,
and
governmentreports.F marksthe locationof the preliminaryepicenterdeterminedby Fiedler [1968]. The short-dashed
lines
indicatethe fault systemproposed
by Rial [1978], locatedrelativeto the epicenterdetermined
by the ISC (star).The longdashedline showsthe strikeof the east-westnodalplaneof the focal mechanism.
Network(WWSSN), becameagainoperationala fews daysafter
distance from the recording station may be estimated and
plotted on a map (Figure 7).
recordedat CAR in the following month. Although epicentral
Assumingthe rupturenucleatednear the ISC epicenter,most
locations of these small aftershocks cannot be obtained with
of the 22 aftershocksanalyzedwould fall well to the east of the
only one station, it was possibleto identify in the original proposed north-south fault system (Figure 7). Clearly, the
recordsclear compressional(P) and shear wave (S) phases. epicentral distancesof the aftershocksto station CAR would
Measuring the $-P travel time differences, the epicentral agree better with an east-westoriented fault north of Caracas,
as most of the epicentral distance radii intersect the line
representingthe east-westnodal plane of the focal mechanism
(Figure 7). Furthermore, based on the polarization of the P
the earthquake occurred. Dozens of small aftershocks were
waves from the aftershocks
that were well recorded both on the
vertical and horizontal instruments at station CAR, it is clear
that all of them have a location
.c:
2
E
Z
1
2.5
7.5
12.5
17.5
22.5
27.5
32.5
37.5
42.5
Distance Difference (km)
Fig.5. Histogramshowingdifferencesin epicentrallocationfor 15
earthquakes
locatedby a local seismicnetworkin Venezuelaand by the
ISC.
of
The results presented above strongly favor the initial
assumption of Molnar and Sykes [1969] that the Caracas
earthquake occurred on one of the various east-west running
faults that define this complex plate boundary.This prompted
us to reexaminequantitatively the radiation of P and SH waves
using a formal inversion scheme. The following section
describesthe results of this analysis.
w
o
to the north and northwest
CAR.
3
INVERSION OF BODY WAVES
Data Preparation and InversionMethod
The source parametersand rupture processof the Caracas
earthquake are investigated through an inversion of longperiod P wavesand SH wavesrecordedby the WWSSN. Only
17,464
SU•P•Z ANDN•B•LEK: CARACAS
EARTHQUAKE
seismograms
of stations
withinan epicentral
distance
range
of
TABLE 1. StationsUsed in the Relocationof the Aftershock
30o-90ø were includedin the analysis;within this distance
range,theGreenfunctions
for a sourceembedded
in a layered
StationName
Arrival Time of P Wave,
earth are simple to computeand can be accuratelyestimated
[e.g.,Langstonand Helmberger,1975].The crustalstructure
in
the source region and beneath recording stations was
approximatedby a half-spacewith a compressional
wave
GMT
TRN
SVT
SJG
BOG
HUA
PNS
LPB
WMO
TFO
LF2
BMO
ILG
0041:02.5
0041:08.4
0041:27.7
0041:56.5
0044:58.0
0045:29.2
0045:27.3
0046:57.8
0048:09.4
0048:27.8
0049:18.4
0050:50.8
velocityof 6 km/s, a Poissonratio of 0.25, and densityof
2.75 g/cm3. Anelasticattenuation
alongthepropagation
path
wasparameterized
usinga t* of 1 s for? wavesanda t* of4 sfor
SH waves.
The inversionprocedureis discussedin detail by Ndb•lek
[1984] and has been applied succesfullyto other large
earthquakes
[Ndb•lek,1985;Ndb•leket al., 1987].Depending
on the size and complexityof the earthquake,the sourcecanbe
parameterized
either as a singlepoint sourceor as an event
consistingof severalpoint sources(subevents)separated
in
TABLE 2. Focal Parameters of the Mainshock and the Relocated Aftershock
OriginTime,
Date
Latitude,
GMT
July 30, 1967
July 30, 1967
0000:02.7
0039:49.9
II o
N
Longitude,
Depth,
øN
øW
km
10.68
10.70
67.40
66.95
26
10
Source
ISC
this study
FI
MAIN
EVENT
AFTERSHOCK
07/30/67
.....,........._.
_....,----'"-"'"•--'
07/30/67
0 hrO'2.7"
0 hr39' 49.9"
•.[
I
CATIA
CARBALLEDA
LA
LA MAR
A.•I•AcuTo
NAIGUATA
MAIQUETIA
10.5"
LOS
TEQUES
•
RACAYLAVICTORIA 0I
I
25
I
kin
AFTERSHOCK
67'.5ow
I
50
RELOCATION
67 •
Fig.6. Master
eventrelocation
of thelargest
aftershock
(solid
triangle)
feltin theCaracas
region
about
40 rainafterthe
mainshock
(solidcircle).Themainshock
wasusedasthemaster
eventin therelocation.
Thelinesthrough
thesesymbols
are errorestimates
of the location.The dashedline showstheorientation
of thenodalplaneorientedeast-west.
SU•REZANDN,•B•LEK:CARACAS
EARTHQUAKE
17,465
stationsto ascertaindirectly if substantialdifferencesexist in
the shapeof the body wavesthat would suggestdirectivityof
the sourceruptureprocess.For this purpose,we comparethe
body wavesrecordedat threeeast-westandnorth-south
pairsof
seismic stationslocated symmetricallyrelative to the nodal
planes of the fault plane solutionof the Caracasearthquake
(Figure 8). For a pure strike-slip event, this symmetric
AFTERSHOCK LOCATION
S-P TIMES
( mb= 4.4)
LA GUAIRA
location
•_ARACAS
relative
waveforms
to the source mechanism
have a similar
location
relative
insures that the
to the radiation
patternof direct and reflectedphases.The polaritieshave been
made the samefor eachpair of stationsin orderto facilitatethe
direct visual comparison.Thus for each pair of stationsthe
waveforms may be directly compared, and any drastic
differencesin waveformsshouldreflect the effects of rupture
directivity and sourcefiniteness.
The P wavesrecordedat stationMAL (Spain) are more compressedin time than thoseobservedat station UNM (Mexico).
67, 5 øW
67 ø
The first two peaks are almost in phase, but the third and
Fig.7. Circles of possibleepicentrallocationsof small aftershocks
relativeto stationCAR. The epicentraldistancesare basedon readings largest peak is clearly delayed at UNM relative to its
counterpart at MAL. Subsequentpeaks and troughs in the
of S-P travel time differences.Other symbolsare as in Figure 6.
signals show a much longer, low-frequency signal at UNM
than
at MAL (Figure 8). This observation is even more
time and space.For each subevent,the model parametersthat
apparent
comparingSH waveformsrecordedat TOL (Spain)and
we invert for from the observed seismograms include the
source mechanism (strike, dip, and rake), depth, seismic KIP (Hawaii). At TOL the observedSH waves are compressed,
moment, source time function, and the relative location and exhibiting a much shorter duration than at KIP (Figure 8).
time delay of each subeventwith respectto the first one. The These two setsof stations,locatedroughly to the east and west
sourcetime function is allowed to take an arbitrary shape, of the epicenter,suggesta rupture propagatingfrom west to
making the parameterization suitable for studies of large, east; the recedingrupturefront produceslong-durationsignals
complicatedearthquakes.The data used in the inversion are at stationsto the west of the epicenter,while thoseto the east,
hand-digitizedtime series (two samples per second) of the lying along a similar azimuthas the directionof rupture,are
observedP (vertical component) and SH seismograms,in a strongly compressed.
The S H waves recorded at CMC (Canada) and LPA
time window that includes the direct body wave phases and
extendsin time to include all surfacereflected waves (i.e., pP, (Argentina),seismicstationslocatedto the northand southof
sP, and sS). The seismic phases and stations used in the the epicenter,respectively,are very similar to one another.
inversionare shown in Table 3. Theoretical seismogramsare The prominentpeaksand troughsof both time seriesare in
iteratively matched to the observedwaveforms using a least phase, and the durationof the signal is almost identical in
both cases. We interpret this as evidence that no major
squarescriterion.
CAR
directivity occurredin the north-southdirection.No northsouth,P wave pair is presentedbecauseof the lack of long-
Visual Comparisonof Waveforms
Beforeproceedingto the formal inversionof the data, it is
instructive to visually compare the waveforms at several
periodP wavedatato thesouthof theepicenter
in the300-90ø
epicentraldistancerange.
TABLE 3. SeismicStationsUsedin the Inversionof Body Waves
Station
Azimuth,
deg
EpicentralDistance,
deg
WaveType
GDH
KEV
5.7
20.5
35.1
54.1
289.5
311.3
323.2
334.4
352.3
0.5
59.2
81.6
75.5
62.1
31.8
38.9
31.1
76.3
65.9
44.1
P
P
P
P
P, $H
P
P
P
P
SH
34.1
50.6
169.2
184.0
291.1
317.3
342.0
66.7
62.8
46.3
43.8
86.7
44.4
65.1
SH
SH
SH
SH
$H
COP
MAL
UNM
LUB
OXF
COL
RES
SCH
ESK
TOL
LPA
PEL
KIP
GOL
CMC
SH
SH
Network
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
CSN
CSN
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
WWSSN
CSN
17,466
SU•,REZANDN•d•LEK: CARACAS
EARTHQUAKE
EAST
P
MAL
EAST
SH
TOL
NORTH
SH
CMC
i
•
I
i'
I
I
I
I
I
I
I
I
SOUTH
SH
'
1 min
i
WEST
P UNM
WEST
SH
KIP
P
LPA
SH
Fig. 8. Comparisonof P andSH waveslocatedsymmetrically
with respectto the focal mechanism
of the Caracas
earthquake.
Noticehowstations
to theeast(TOLandMAL) showa morecompressed
wavetrain
thanthoseto thewest(UNM
andKIP). SH wavesto the northand southof the epicenter(CMC andLPA) do not showmajordifferencessuggesting
the
ruptureprocess
hadan east-west
propagation.
The ray pathsto the stations
andthenodalplanesareprojected
of thelower
part of the focal sphere.
1.1
Inversion Strategy
The waveforms indicate that the Caracas earthquake is a
1.0
complex,multiple event. Rial [1978] showedthat at least
three subeventswere necessaryto match the observedwave
trains. The number of free source parameters necessary to
0.9
describe these waveforms is quite large. To deal with the
complexityof the sourcemechanism,
whiletryingto keepthe
numberof free parameters
to a minimum,a stepwise
inversion
strategywasadopted.As a first step,the Caracasearthquake
was assumedto be a single-pointsourcewith a long source
time function. The starting source mechanism for the
inversion was that of MoInar
N
-,._•.
0.8
E
o
z
0.7
and Sykes [1969]. This
inversiongave the average(centroidal)mechanism
with a
strikeof 262ø, dip of 74ø, anda rakeof 183ø andshowed
that
0.6
0
four subeventsare neededto model the rupture.Next, in order
4
8
120
160
200
240
280
320
360
Azimuth (deg)
to determine
thepredominant
directionof rupturepropagation,
we introducedthe effect of rupturedire•tivity into the model
[Ndb•lek, 1984, 1985] and variedthe rupturedirectionas a Fig.9. Varianceof the residualsas a functionof azimuthof the
functionof azimuth;here we assumedthe rupturevelocityof rupturepropagationdirectionfor an unilaterallypropagatingsource
with a velocityof 2.4 km/s.The clearvarianceminimumindicates
that
2.4 km/s andusedthe abovesourcemechanism.
The resulting theruptureof the Caracasearthquake
propagated
predominantly
along
of 80ø.Theaverage
mechanism
anddepthgivenin Table4
plotof residual
variance
versus
azimuth
is displayed
in Figure anazimuth
9, showinga variancereduction
of about30% alongthe
are used.
SUXREZ
ANDN•B•LEK:CARACAS
EARTHQUAKE
17,467
17,468
SU•REZANDN•B •LEK: CARACAS
EARTHQUAKE
TABLE 4. SourceParameters
of the 1967Caracas
Earthquake
Determinedby Inversionof
Body Wave Seismograms
Strike, Dip, Rake, Depth, Moment, Delay, Distance, Azimuth, Duration,
deg
deg deg
km
1018
Nm
s
km
deg
s
Subevent
1
2
3
4
Average*
261
265
230
276
85
69
99
59
180
194
191
128
14.4
14.1
7.7
21.0
3.1
4.5
1.3
1.6
14.8
36.5
50.6
42
89
91
76
69
34
6
9
4
6
262
74
183
14.3
8.6
-
-
-
57
* Bestdouble-couple
of themomenttensorsumof thefoursubevents.
azimuth of 80 ø. This result substantiates formally
The final model was obtained in the following manner.
the
conclusionbased on the visual inspectionof waveformsthat
to the first order the predominantdirection of rupture was
eastward, approximatelyalong the strike indicated by the
average mechanism. The simple unilaterally propagating
source model, however, did not produce completely
satisfactory fit between the observed and synthetic seismogramsat all stations(Figure 10e), indicatingthat a more
complicatedmodelinvolvingprecisepositionsandchangesin
mechanisms of subevents is necessary.
Using the roughsubdivisionof the earthquakeinto subevents
based on the first
set of inversions,
each subevent was
analyzedindependentlyby freeing the parameterspertaining
to that subevent.For example,Figure 10a showsthe resultsof
invertingthe time serieswithin a time frame includingonly
the first subevent. After the analysis of the first event was
completed,the inversionschemewas appliedto the nexttime
window corresponding
to the secondsubevent(Figure 10b).A
similar strippingapproachwas followed sequentiallyfor the
Event 4
h = 21.0 km
Mo = 1.6 x 1018 N rn
Event 3
Fo
h = 7.7 km
Mo = 1.3 x 1018 N m
Event 2
h-- 14.1 km
Mo = 4.5 x 1018 N m
Event
1
h = 14.4 km
& Aftershock
CA
RA
CA
S
' "•";;
•
0
25
•
I
50 km
Fig.l l. Themechanism
andlocation
(solid
circles)
of thefoursubevents
thatformtheCaracas
earthquake.
The
mechanisms
areplotted
using
a lowerhemisphere
stereographic
projection,
thedarkquadrants
representing
compressional
firstmotion.Solidtriangleis thelocationof therelocated
aftershock.
SU•REZ
ANDN•BI•LEK:
CARACAS
EARTHQUAKE
17,469
thirdandfourthsubevents
(Figures10cand10d).Theresulting distinctsubevents
we!! separated
in time andbecause
our main
theoreticalseismogramsshow now a better fit to the observed
concern was to fit the major long-period arrivals in the
wavetrains,most significantlyfor the later parts of the SH
observed waveforms.
waves(Figure 10d). The inversion was not continuedbecause
at several stations the arrival of PP and ScS contaminates the
records.In total, a timeframe of 65 s at each station was used in
the inversion.
A West-to-East
RupturePropagation
The final inversionyields for each subeventthe seismic
moment,focal mechanism,depth,sourcetime function,and
Once this piecewise inversion of the waveforms was
finished, all parameters were freed, and the inversion was
allowedto iterate to the best fitting solution. What was done
the location relative to the first subevent. We find that the
bythestepwiseinversionschemedescribedabovewas to find
west to east, with distances of about 50 km from one another.
mainseismicradiationis dominated
by four subevents
(Table4
andFigure11).The firstthreeeventstriggersequentially
from
a stableand realistic starting model for the final inversion in
which all parameters were allowed to vary at once. This
These three eventsshow almost pure strike-slipmotion,
althoughthe mechanism
of the thirdonebeginsto showsome
minimized wild iterations and insured that the inversion
dip-slipcomponent
(Figure11). The averagedepthof these
algorithmwould iterateto the true minimum in the residual threeshocksis 14 km. The fourtheventin the sequence
is
space.The reason we were able to take this simplified deeper,havinga depthof 21 km, andit hasa reverse-faulting
approachis that the Caracasearthquakewas composedof mechanismwith a small strike-slip component.The total
Caracas,
1967
P waves
60 s
GDH
COL
RES
OXF
KEV
LUB
UNM
MAL
mm
60
I
sec
Fig. 12a. Comparisonbetweenthe observedP waves(solid lines) and the theoreticalseismograms
(dashedlines)for the
best fit model obtainedby the simultaneousinversionof P andSH waves.The focal mechanismof the first subeventis
shownin a lower hemisphereprojectionwhere the solid circlesindicatecompressions
and the open circlesdilatations.
Amplitudes of the waveforms are normalized to an instrumentalmagnificationof 1500 and a geometrical spreading
corresponding
to an epicentraldistanceof 40ø.
17,470
SU•REZ
AND
N•B•,LEK:
CARACAS
EARTHQUAKE
Caracas, 1967
SH waves
CMC
GOL
T
KIP
I2060 PEL
,• UNM
mm
I
ß
sec
,0
Fig. ]2b. Comparison
betweenthe observed
SH wavesandtheth½orefica!
seismograms
for thesameinversion
as in
Figure] 2a. 'Thesymbolsa•d amplitudenormaJization
arethesameasfor P waves.
seismicmoment(tensorsum)of the Caracasearthquake
is 8.6
(Figures 12a and 12b). Subevents1 and 2 are the best resolved
x 10Is N m (Mw=6.6).
of thesequence.
Theyarewell separated
in timeandproduce
The orientationof the fault systemdefinedby the locations strong contributionson all seismograms(Figure 10).
of the subevents
formingthe Caracasearthquake
is subparallel Subevent4 is also well resolved.Due to its dip-slip
to the east-west oriented nodal plane of the fault plane mechanism,
it hasa strongereffecton the ? wavesthanonthe
solutions,indicatingthat the rupture occurredon an east-west SH waves(Figure10).Thethirdsubevent
is partlybuffedinthe
orientedfault system.The syntheticseismograms
computed coda of the first and second subevents and has the smallest
using the sourcemodel derived from the joint inversionof P
effecton reducingthe residuals;
consequently,
it is the least
andSH wavesagreewell with theobserved
waveforms
(Figures resolvedsubeventof the sequence.
12a and 12b). For all the stations analyzed, there are no
In orderto visualizethe effectof the fault geometryon the
significantmismatches
in the arrivaltimesof the majorpulses body waves and to evaluate the resolution of the data, it is
between the observed and the synthetic waveforms. As instructiveto comparethe solution obtainedhere with that
expected, the matches are not as good for stations near the
nodes of the radiation patterns, where small errors in the
assumedray takeoff anglesproducelarge errorsin shapeand
determinedby Rial [1978]. In the case of the P waves, the
largest differencesbetween the two sourcemodels occur in the
westernstations.For example,at stationLUB (United States),
amplitudes of the arrivals. Also, due to the underpa- thenorth-south
trendingfaultmodelshowslargephaseoffsets
rameterization of the source, the matches are better for the
lower frequencies than for the higher frequencies. This is
reflected in somewhatunderestimatedamplitudesof P wavesin
the direction of rupture. Overall, within a time frame of about
65 s from the onset of the body waves, the rupture processof
the 1967 earthquakeis well representedby the four shocks
relativeto the observations
(Figure13). SH wavesare more
sensitiveto rupturepropagationthan P waves,and herethe
orientationof the fault systemis even more crucial in fitting
the observedseismograms.
At stationsUNM and TOL, neither
the amplitudenor the phaseof the syntheticseismograms
produced with a north-south oriented fault match the
SU,•REZ
AND
N,J•B•.LEK:
CARACAS
EARTHQUAKE
observations
(Figure 13). This is particularly true for TOL,
wherethe compressednature of the observedSH waves cannot
bematchedby a fault systempropagatingnorth to south.The
17,471
west trendingfaults that characterizethe plate boundaryin
northernSouthAmerica.The subevents
composing
therupture
are separatedin spaceby distancesmuch greaterthan the
opposite
is trueat stationUNM, wherethe synthetic
seismo- length of the individual faults, as inferred from the duration of
gramfor Rial's [1978]modelis too shortcompared
to the theirrespectivesourcetime functions.The averagedurationof
recorded
SH wave. It shouldbe pointedout that we were unable
toreproduce
the syntheticwaveforms
presented
in Rial's paper
usingthe parametersshown in his Table 1. Many of the
synthetic
waveformspresentedby Rial [1978] appearto be
missingthe third subeventwhich he interpretedto be the
largest
oneof all.
the source time function of each subevent indicates fault
lengths shorterthan about 12 km, whereasthe average
distance between the subeventsis about 50 km. Thus the four
shocksaccountingfor the main part of the radiatedseismic
energy rupture sequentiallyfour different fault segments,
insteadof a single, long, or en echelonfault. No seismic
As we discussed before, stations located to the north and energyappearsto havebeenradiatedin the segments
between
southof the fault zone do not exhibit large differencesin the faults. This is indicativeof a complexand severelyfaulted
waveshapes,indicatingthey do not "see" a directivity effect platemarginwherethe faultsarehighlystressed.
The distances
in this direction. Our model matches both sets of stations well
separatingeach subeventand the time delay amongthem
(Figure13). RiM [1978], however,wasnot ableto fit the later suggestit was not stressconcentrationat the end of one fault
partof theSH waveat LPA, becausethe assumed
directivity rupturethat triggeredthe next earthquake.Perhapsit was the
towardthe southdrasticallycompresses
the seismogram.Rial passageof the surfacewavesthat triggereddaischainreaction
suggested
that the last part of the seismogram,which couldnot of faulting.
bematchedby a north-south
fault, wasperhapsdueto a coupled
It is interestingto point out that the great earthquakeof
PL wave arriving "off-azimuth" after the SH. The sourcemodel 1812 that affectedan elongatedarea along the Boconofault
derivedhere matches that part of the waveform at LPA (and from Caracasto Merida hasbeeninterpreted
as consisting
of
PEL) without the need of invoking additional nonstandard two or more distincteventsalongthe plate margin[Kelleheret
waves (Figure 13). Many of the changes in waveform al., 1973]. This is suggested
by the fact that severalareasof
interpreted
by Rial [1978] as being due to the directivityof the
sourceare caused by the changesin focal mechanism of the
very high intensitynear the cities of Merida and Caracas,are
third and fourth shocks. Notice that stations UNM,
fault [Centeno-Grau,
KIP and
GOL, for example,lie very closeto the nodal plane of the SH
wavein the focal sphere(Figure 12b). Thus small changesin
thefocal mechanismshow up as large variationsin the amplitudeandpolarity of the radiatedSH waves.
separatedby others showingonly slight damagealong the
1940; Fiedler,
1961; Kelleher et al.,
1973]. Complex earthquakescomposedof several shocks
breakingdifferent fault segmentsmay thereforebe character-
isticof'thishighlyfaultedplateboundary.
It is not possibleto determineon which fault the individual
shocksof the 1967eventoccurred;
offshoregeologyhasnot
been mappedin this region of Venezuela.It appearsthat the
sequenceof eventsformingthe Caracasearthquaketook place
Strike-SlipFaulting Along the Plate Margin
on faults that are part of the Moron-La Victoria fault system
The results of the inversion presentedhere show that the (Figure 2), indicatingthat the relative plate motion between
Caracasearthquakeoccurredon the complex systemof east- the Caribbeanandthe SouthAmericanplatesis accommodated
DISCUSSION OF RESULTS
SH
waves
CMC
P
waves
LPA
MAL
R
. .[
SN
UNM
TeL
LUB
•o
R'•,ff•
'."/"
R '••
R ••
SN
.'"':.•
--
SN
SN
..
.
.:
Fig. 13. Comparison
betweenthe observed
P andSH waveseismograms
(solidlines)andtheoretical
seismograms
(dashed
lines) producedby the fault model suggested
by Rial [1978] (R) and by the modelproposed
here (SN). Only several
representative
stations
areshown.Because
ourmomentestimate
is differentfromRial's,we compare
onlythe waveshapes
andnottheabsolute
amplitudes;
theamplitudes
of thetheoretical
seismograms
arescaledto matchthepowerin thewindow
of the observed waveform.
17,472
SU•,REZANDN•d3•LEK:CARACAS
EARTHQUAKE
over a broadareaof deformation,insteadof beingconcentrated
on a main throughgoing fault.
The observationthat in this segmentof the plate boundary
relative motionis accommodated
over a large zoneof deformation is not surprising,consideringthat the right-lateral strike
slip faults defining the plate margin to the east and to the
southwestundergo a sharp change in strike (Figure 1). The
location of the 1967 earthquakeis near the apex of the bend.
The complex stressfield due to the bend probably causesthe
complicated pattern of faulting observedin the north central
coast of Venezuela, where faults of different type and
orientation coexist with the dominant right-lateral strike-slip
faults (Figure 2).
The importantrole played by fault bendsin the initiation
and and terminationof faulting has been stressedbefore [e.g.,
Aki, 1979; King and Ndb•lek, 1985; Barka and KadinskyCade, 1988]. The rupturesin individual earthquakesappearto
be frequently limited to regions between bends in the fault
trace, with bends acting as barriers to rupture propagation.
This suggeststhat faulting initiated on the Boconoor on the
Moron fault systems would not propagate around this
curvature. Furthermore, the intense fault segmentationmay
also limit the largest possiblesize of an earthquakein the
region near the bend.
Althoughthe evidencepresentedhere stronglysuggests
that
the 1967 Caracasearthquaketook place on a systemof faults
oriented east-west, the other conjugate faults mapped in the
area are also probablyseismogenic.
In fact, the two relatively
relativeto the Caribbean
alongan azimuthof aboutN75ow.
The compression
producedby this relativeplatemotioncould
accountfor this underthrusting.
The reverse-faulting
mechanism
shownby the fault plane solutionof the fourth
subevent
suggests
compressional
deformation
induced
by the
thrustzone to the north. The depthobtained(21 kin) would
place the earthquakewithin the underthrustbasementof the
VenezuelanBasin, southof the CuracaoTrough.
Geometryof Faulting and Damageand IntensityReports
Theepicenter
of the second
subevent
liesnorthof Caracas,
about 12 km offshore.This subeventwas the largestof thesequence(Table 4) with a magnitudeof 6.4 (Mw). The relocated
epicenter
of the largestaftershock
is lessthan10 km away
fromthissecond
subevent,
suggesting
thattheaftershock
may
haveoccurredon the fault planeof the latter. The epicenter
of
thesecond
subevent
is verycloseto theareashowing
the
highest intensities. It appears that the damage observedin
Caracas
andin neighboring
coastal
townswasproduced
mainly
by thiseventandnot by the threeothereventsin the sequence
whichareover50 km awayfromthe areaof largestdamage.
More recently,RiaI [ 1984] has arguedthat the CaracasBasin
acts as a lens focusing seismic energy in the Palos Grandes
area of Caracas,where several buildings collapsed[e.g.,
Whitman, 1969]. His resultsimply that this caustic is formed
when the seismic waves approachthe basin from the north.
The secondsubeventin the seriesis located almost exactlyto
the north of Caracas and would be capable of causing the
focusingeffect suggested
by Rial [1984] to partly explainthe
localized damage in Caracas.Whatever the case, it is clear
have occurred on NNW-SSE striking faults (anonymous Caracassits on a highly faulted plate boundary,and the 1967
reviewer,personalcommunication,
1989). This is a common earthquakeshows that even earthquakeswith moderate
mayproduceintenselocalizeddamage.
mode of deformationin a complextectonicenvironment.
The magnitudes
recent swarms in northcentral Venezuela, near towns of Los
Teques(August1986)andAyagua(November1980),appearto
recentSuperstition
Hills earthquake
in southern
California,
for
example,occurred
on two conjugate
faults[e.g.,Buddingand
Sharp, 1988;Johnson,1988;Hudnutet al., 1989].Moreover,
the 1968 Borrego Mountain earthquake,also in southern
California,producedsurfacerupturealongthe main strand
CoyoteCreekfault, but the high seismicactivityprior and
after the earthquake
occurredalmostentirelyon crossfaults
(M. Petersenet al., The interaction
betweensecondary
and
SUMMARY
The Caracasearthquakeof July 30, 1967, is one of the
largesteventsto haveoccurredalongthe SouthAmericanand
Caribbeanplate margin since the earthquakeof 1900. The
ruptureprocess
of the 1967earthquake
is complex.The formal
inversion of the P and SH waveforms shows the source is
composedof at least four subeventsseparatedin time and
space.The focal mechanismof the first three subevents
manuscript
in preparation,
1990).The NW to SE trending indicatestrike-slipfaulting at an averagedepth of 14 kin,
master faults within the southern San Jacinto fault zone,
faultsmapped
in northcentralVenezuela
(Figure2) should
be,
consistent with the relative
motion between the South
therefore,considered
whenassessing
the seismichazardof this American and Caribbean plates. These first three eventsare
triggeredsequentially
fromwestto east,in a directionsimilar
highly populatedregion.
to the strike of the E-W trending nodal plane of the source
mechanism.
This fact stronglysuggests
thatrupturetookplace
Compressional
Deformation
in theSouthern
Caribbean
The epicenter
of the fourthsubevent
lies about100 km on a systemof faults orientedeast-west.
offshore of Caracas. The mechanismshows mainly reverse
The fourtheventin the sequence
showsreversefaultingat a
faultingat a depthof 21 km (Figure11).Thisfourthshock
is depth of 21 km and appearsto reflect compressional
alongthe southern
marginof the Caribbean
plate.
perhaps
related
to thecompressional
deformation
observed
in deformation
tectonicregimeto the northis
the VenezuelaBasinbasementbeneaththe CuracaoRidge in Evidencefor a compressive
of the
the southern Caribbean [Talwani et al., 1977]. Seismic foundin seismicreflectiondatashowingunderthrusting
Caribbean
plate
beneath
South
America
near
the
Curacao
Ridge
reflection recordsshow evidenceof folding in Quaternary
and
the
associated
deformation
of
Quaternary
sediments
in the
sediments
alongthe CuracaoRidgeandin the LosRoques
Basin offshoreVenezuela[Kellogand Bonini,1982;Ladd et Los RoquesBasin.
The epicentraldistances
of aftershocks
recordedin Caracas
al., 1984;Silveret al., 1975;Talwaniet al., 1977].Caseet al.
dataobserved
aftertheearthquake
support
the
[1984]havenamed
thisregion,
containing
asmuchas 10km andtheintensity
east-west
orientation
of
the
fault
system
derived
from
the
of deformed
pelagicandturbidite
sediments
of Neogene
age,
inversion
of bodywaves.Moreover,
therelocated
epicenter
of
the SouthernCaribbeanDeformed Belt.
(m•,=4.4)thatoccurred
40 minafterthe
Thepolesof rotation
proposed
byJordan[1975]andStein thelargestaftershock
of the second
et al. [1988]predictmotionof the SouthAmerican
plate maineventlieswithin10 km of the epicenter
SUXREZ
ANDN,•B•_,LEK:
CARACAS
EARTHQUAKE
17,473
de la regi6nde Caracascon relaci6nal
subeventof the mainshock.It appears that this large Fiedler, G., Estudiosismo16gico
terremoto del 29 de julio de 1967, Bol. Inst. Materiales Modelos
aftershock
occurredon the samefault segmentresponsible
for
Estracturales (IMME), VI (23-24), 127-222, 1968.
thesecond
subevent.
The secondshockof the rupturesequence Garcfa, D., and Y. P. Aggarwal, Seismotectonics of southern
is about12 km offshore,near the city of Caracasand the
Venezuelan Andes, Eos Trans. AGU, 63, 1126, 1982.
coastaltowns that experienced the highest intensities. Gras6s, J., Investigaci6n Sobre los Sismos Destructores que Hah
Afectado el Centro y Occidente de Venezuela, 3 vol., Informe
Thereforeit appearsthat it was the secondsubeventof the
Caracasearthquakewhich was mainly responsiblefor the
damage.
The average separation of about 50 km between the
subevents
of the mainshockis much larger than the fault
INTEVEP, Caracas, Venezuela, 1980.
Hudnut, K. W., L. Seeber,and J. Pacheco,Cross-faulttriggeringin the
November 1987 SuperstitionHills earthquakesequence,southern
California, Geophys.Res. Lett., 16, 199-202, 1989.
Johnson, C. E., Tectonic implications of the November 24, 1978
SuperstitionHills earthquakes,Imperial Valley, CA, Seismol.Res.
lengths
inferred
fromtheduration
of thesource
timefunctions.
Lett., 59, 48, I988.
This indicates that the Caracas earthquake ruptured several Jordan, T. H, The present-daymotions of the Caribbeanplate, J.
faultsseparatedby a large distanceand not adjacentor en
Geophys. Res., 80, 4433-4439, 1975.
echelonsectionsof a single fault. Each of these individual Kelleher, J., L. R. Sykes,and J. Oliver, Possiblecriteria for predicting
faultsegments
brokesequentially,
perhapstriggeredby the
passage
of thesurface
wavesof theprevious
subevents.
Thestyleof faultingduringthe earthquake
suggests
thatthe
plateboundary
nearCaracas
is complex,
highlystressed,
and
verysegmented.
Thisgeometry
of faultingmaybe explained
earthquakelocationsand their applicationto major plate boundaries
of the Pacific and the Caribbean,J. Geophys.Res., 78, 2547-2585,
1973.
Kellog,J. N., andW. E. Bonini,Subduction
of the Caribbeanplateand
basementupliftsin the overridingSouthAmericanplate,Tectonics,
1, 251-276,
1982.
Role of fault bendsin the initiation and
bythepresence
of a bendin theplateboundary,
whichchanges King, G., andJ. N,4.b•,lek,
froman east-weststrikealongthe E1 Pilar fault systemto a NE-
SW strikealong the Boconofault zone. The intensefault
terminationof rupture,Science,228, 984-987, 1985.
Ladd, J. W., M. Truchan, M. Talwani, P. L. Stoffa, P. Buhl, R. Houtz,
A. Mauffret, and G. Westbrook,Seismic reflection profiles across
segmentation
may limit the largestpossiblesize of an
the southernmarginof the Caribbean,Mere. Geol. Soc. Am., 162,
earthquake
in the regionof the bend.Althoughthe Caracas 153-159, 1984.
earthquake
of 1967tookplacealongfaultsoriented
east-west, Langston,C. A., and D. V. Helmberger,A procedurefor modeling
shallow dislocation sources, Geophys.J. R. Astron. Soc., 42,
the possibilityof moderatelysized eventsoccurringon
117-130,
1975.
conjugate
faultsoriented
NW-SEshould
notbe discounted
in Minster, J. B., and T. H. Jordan, Present-day plate motions, J.
assessing
the seismichazardof the region.
Geophys.Res., 83, 5331-5334, 1978.
Molnar, P., and L. R. Sykes,Tectonicsof the Caribbeanand Middle
Acknowledgments.
The authors
thankY.P. AggarwalandO. P6rez
Americanregionsfromfocalmechanisms
andseismicity,
Geol.Soc.
fordiscussions
andJ. Dewey,C. Schubert,
andJ.P. Soulasfor detailed
Am. Bull., 80, 1639-1689, 1969.
reviewsof the manuscript.O. P6rez made available copies of Ntil•lek, J. L., Determinationof earthquakesourceparametersfrom
seismograms
recorded
in CAR.L. Jones
helpedto familiarize
uswith
inversionof body waves, Ph.D. thesis, 361 pp., Mass. Inst. of
themaster
eventrelocation
programs.
K. NagaoandS. Campos
drafted
Technol., Cambridge, 1984.
mostof thefigures.This workwasinitiatedwhileoneof the authors Na'b•lek,J., Geometryand mechanismof faulting of the 1980 E1
(G.S.)wasrecipient
of the Lamont-Doherty
Postdoctoral
Fellowship. Asnam,Algeria,earthquakefrom inversionof teleseismicbody
Thisworkwasalsofundedby theNSF grantEAR 88-96187.
waves and comparisonwith field observations,
J. Geophys.Res.,
90, 12,713-12,728,
1985.
Ntib•lek, J., W.-P. Chen, and H. Ye, The Tangshanearthquakesequence
REFERENCES
Aki, K., Characterizationof barriers on an earthquakefault, J.
Geophys.
Res., 84, 6140- 6148, 1979.
Barazangi,
M., andJ. Dorman,Worldseismicity
mapsof ESSA,Coast
andGeodeticSurveyepicenter
datafor 1961-1967,Bull. Seismol.
Soc. Am., 59, 369, 1969.
Barka,A. A., and K. Kadinsky-Cade,Strike-slipfault geometryin
Turkeyand its influenceon earthquakeactivity, Tectonics,7,
and its implications
for the evolutionof the NorthChinaBasin,J.
Geophys.Res.,92, 12,615-I2,628, 1987.
P6rez, O. J., and Y. P. Aggarwal,Present-daytectonicsof the southeastern Caribbean and northeasternVenezuela, J. Geophys.Res.,
86, 10,791-10,804,
1981.
Rial, J. A., The Caracas, Venezuela earthquakeof July 1967: A
multiplesourceevent,J. Geophys.
Res.,83, 5405-5414, 1978.
Rial, J. A., Causticsand focusingproducedby sedimentarybasins:
Applicationsof catastrophe
theory to earthquakeseismology,
633-684, 1988.
Geophys.J. R. Astron.Soc.,79, 923-938, 1984.
Budding,
K. E., andR. V. Sharp,Surfacefaultingassociated
withthe
768 pp., W. H. Freeman,New
ElmoreDesertRanchandSuperstition
Hills, California,earthquakes Richter,C. F., ElementarySeisinology,
of 24 November1987, Seismol.Res. Lett., 59, 49, 1988.
York, 1958.
Burke,K., C. Cooper,J. F. Dewey,P. Mann, and J. L. Pindell, Rod, E., Strike slip faults of northernVenezuela,Am. Assoc.Petr.
Caribbean tectonics and relative motions, Mere. Geol. Soc. Am.,
162, 31-63,
1984.
Geol. Bull., 40, 457-476,
1956.
Schubert,C., Are the Venezuelanfault systemspart of the southern
Caribbean
plateboundary?,
Geol.Rundsch.,
70, 542-551, 1981.
Case,J. E., T. L. Holcombe,and R. G. Martin, Map of geologic
Pilar fault
provincesin the Caribbeanregion,Mere. Geol. Soc. Am., 162, Schubert,C., Basin formationalongthe Bocono-Moron-E1
system,Venezuela,
J. Geophys.
Res.,89, 5711-5718, 1984.
1-30, 1984.
Centeno-Grau,
M., EstudiosSismolSgicos,
210-300 pp.,Litograffadel Schubert, C. and F. F. Krause, Moron fault zone, north-central
Comercio,Caracas,Venezuela, 1940.
Venezuelan borderland: Identification, definition, and neotectonic
character,Mar. Geophys.Res.,6, 257-273, 1984.
Schubert,C., and C. Laredo,Late Pleistoceneand Holocenefaultingin
Dewey,
J. W., Seismicity
studies
withthemethodof jointhypocentral Lake Valencia basin, north-central Venezuela, Geology, 7,
289-292, 1979.
determination,
Ph.D. thesis,Univ. of Calif., Berkeley,1971.
betweensoil
Dewey,J. W., Seismicity
andtectonics
of westernVenezuela,
Bull. Seed,H. B., I. M Idriss,andH. Dezfulian,Relationships
conditionsandbuildingdamagein the Caracasearthquake
of July
$eismol. Soc. Am., 62, 1711-1751, 1972.
Espinosa,
A. F., andS. T. Algermissen,
A studyof soilamplification 29, 1967, Univ. of Calif., Rep. EERC70-2, 201 pp., Berkeley,
Cisternas,
A., andR. Gaulon,Sintesissismotect6nica
del nordestede
Venezuela,Rev. Geoffs.,40, 3-10, 1984.
factors
inearthquake
damage
areas,
Caracas,
Venezuela,
NOAATech.
Rep. ERL, 280-ESL 31, 1972.
Fiedler,
G.,Areasafectadas
porterremotos
enVenezuela,
Bol.Geol.,3,
1791, 1961.
1970.
Silver,E. A., J. E. Case,andH. J. MacGillavry,Geophysical
studyof
the Venezuelan borderland, Geol. Soc. Ant. Bull., 86, 213-226,
1975.