Shape of the Subducted Rivera and Cocos Plates in Southern Mexico

Home , 1932 Jalisco earthquakes, Cocos Plate, Nazca Plate, North American Plate, Rivera Plate, Tehuantepec Ridge

JOURNALOF GEOPHYSICAL RESEARCH, VOL. 100, NO. B7, PAGES 12,357-12,373, JULY 10, 1995

Shapeof the subductedRivera and Cocosplates in southern Mexico: Seismic and tectonicimplications Mario Pardo and Germdo Sufirez Insfitutode Geoffsica,Universidad Nacional Aut6noma de M6xico

Abstract.The geometry of thesubducted Rivera and Cocos plates beneath the North American platein southernMexico was determined based on the accurately located hypocenters oflocal and te!eseismicearthquakes. The hypocenters ofthe teleseisms were relocated, and the focal depths of 21 eventswere constrainedusing a bodywave inversion scheme. The suductionin southern Mexicomay be approximated asa subhorizontalslabbounded atthe edges by the steep subduction geometryof theCocos plate beneath the Caribbean plate to the east and of theRivera plate beneath NorthAmerica to thewest. The dip of theinterplate contact geometry is constantto a depthof 30 kin,and lateral changes in thedip of thesubducted plate are only observed once it isdecoupled fromthe overriding plate. On thebasis of theseismicity, the focal mechanisms, and the geometry ofthe downgoing slab, southern Mexico may be segmented into four regions ß(1) theJalisco regionto thewest, where the Rivera plate subducts at a steepangle that resembles the geometry of theCocos plate beneath the Caribbean plate in CentralAmerica; (2) theMichoacan region, where thedip angleof theCocos plate decreases gradually toward the southeast, (3) theGuerrero-Oaxac.a region,bounded approximately by theonshore projection of theOrozco and O'Gorman fracture zones,where the subducted slab is almostsubhorizontal and underplates the upper continental plate for about250 kin, and (4) the southernOaxaca and Chiapas region, in southeasternMexico, wherethe dip of the subductiongradually increases to a steepersubduction in CentralAmerica. Thesedrastic changes in dip do not appearto takeplace on tearfaults, suggesting that smooth contortionsaccommodate these changes in geometry.The inferred80 and100 km depthcontours of the subducted slab lie beneath the southernfront of the Trans-Mexican Volcanic Belt, suggestingit is directlyrelated to the subduction.Thus the observed nonparallelism with the MiddleAmerican Trench is apparentlydue to thechanging geometry of theRivera and Cocos platesbeneath the North American plate in southernMexico, and not to zones of weakness in the crustof the North Americanplate as someauthors have suggested.

Introduction takes place on a fold. They also suggestedthat to the southeast,the changein dip may be related to differencesin The seismicity and tectonics of southern Mexico are the age of oceanic lithosphere on either side of the characterizedby the subductionof the oceanicCocos and Tehuantepecridge. Riveraplates beneath the North Americanlithosphere along Until recently, the acceptedview was that the Cocos plate the Middle American Trench. Several studies have shown that subductsbeneath central Mexico at a constantand relatively this subductionzone exhibits lateral variations in the dip of shallowangle of ~12 to 15ø [Stoiberand Carr, 1973;Hanus the subductedoceanic lithosphere[e.g., Molnar and Sykes, and Vanek, 1978; Havskov et al., 1982; Burbach et al., 19.84; 1969; Stoiber and Carr, 1973; Dean and Drake, 1978; Nixon, Bevis and lsacks, !984; Singh and Mortera, 1991]. Using 1982; Beyis and lsacks, 1984; Burbach et aL, 1984; Ponce et data from a permanentlocal network, Sutirez et al. [1990] al., 1992]. A steep subductiongeometry in southeastern definedthe geometryof the subductedplate in more detail and Mexico [Ponce et al., 1992] and in the Jaliscoregion of showed that in central Mexico the slab does not subduct at a westernMexico [Pardoand Sudrez,1993] bound a slabdipping steadyshallow angle. Instead,after an initial dip of-15 ø, the apparentlyat a very shallowangle in centralMexico. slab becomesalmost subhorizontal and underplatesthe North On the basis of teleseismic data, Burbach et aI. [19134] American upper lithosphere. This plate geometry is concludedthat the subductedCocos plate can be dividedinto reminiscent of that found in central Peru and central Chile, thesethree major regions dipping at differentangles. Beyis where the Nazca plate also subductshorizontally beneath the andlsacks [1984], using a hypocentraltrend surface analysis, SouthAmerican upper lithosphere. findno evidenceof tearfaulting separating these segments of Perhapsone of the more striking tectonic features in thesubducted lithosphere •[nd" suggested that the transition southern Mexico is the location of the Trans-Mexican Volcanic Belt. The Trans-Mexican Volcanic Belt crosses southern Mexico from east to west at about I9øN. The distributionof the volcanicarc is oblique(-16 ø) relativeto the 1Nowat Departamento deGeoffsica, Universidad deChile, Santiago. Middle American Trench (Figure 1), in an unusal geometry which is not parallel to the subductionzone. The Trans- Copyright1995 by theAmerican Geophysical Union. MexicanVolcanic Belt is ~100 km wide andis cut by a Paperhumber 95JB00919. sequenceof grabensthat are obliquerelative to the general 0148-0227/95/95JB.,00919505.00 trend of the arc [Nixon, 1982]. '

12,357 12,358 PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES

-115" - 110 ø - 105" - 1 O0 ø -95 ø -90 ø -85 ø -80 ø 25 25 ø

NORTH AMERICA

CARIBE 20" 20 ø

PACIFIC

15' 15"

-'OCOS.'

10' 10 ø

-115' -110 ø -105 ø -100 ø -95 ø -90 ø -85 ø -80 ø

Figure 1. General tectonic setting of the study area. Five lithosphericplates arc shown: Pacific, Rivera, Coco.,,, North A•nerica, and Caribbean. Relative convergence rates (cm/yr) between the {•ceanic and continental plates are indicated by open arrows, and absolute motions of the upper plates relative to a hot spot rra•ne of reference ,ire shown by shaded arrows. The abrcviations are EPR, East Pacific Rise; TFZ, Tamayo Fracture Zone; RFZ, Rivera FractureZone; OFZ, Orozco Fracture Zone; OGFZ, O'Gorman Fracture Zone; TR, TchuantcpecRidge; MAT, Middle AmericanTrench; JB, JaliscoBlock; CG, Colima Graben;EG, E1 Gordo Graben. Solid triangles indicate Quaternaryvolcanism, and the crossedarea indicates the Trans- Mexican Volcanic Belt (TMVB).

The volcanic arc erupted since the Miocene and mostly Volcanic Belt is presentedhere. The availability o1'data from during the Plioccne and Quaternary. A diversity of structures permanentand temporary local networksand the occurrenceof such as large strato-volcanoes, monogenetic cineritic cones, recentearthquakes recorded at tcleseismicdistances which had shield volcanoes, and several calderas are found on the not been studied previously otter tin excellent opportunityto volcanic belt. The chemical and petrologicai characteristics understand this region ol' complex subduction geometry in .of the Trans-Mexican Volcanic Belt indicate that, in general, greater detail. the volcanic sequences that form this belt are c,lic-alkaline, The purposeof this study is to determinethe geometryof although some localized zones of alealine volcanism exist the subducted Cocos and Rivera plate, based on all the along the Trans-Mexican Volcanic Belt, especially in western available and reliable hypocenter locations of earthquakes Mexico [e.g., Luhr et al., 1985; Allan et al., 1991]. The recorded with local and teleseismic networks, and to studythe radiometric dating of volcanic rocks and the broad nature of regional distribution of stresseswith reported and newly the Trans-Mexican Volcanic Belt suggest the migration of determined focal mechanisms. Together, these data are volcanism toward the trench sihce 5.3 Ma [Delgado et al., analyzedto determinean integralmodel of the subductionin 1993]. southernMexico and its implicationson the seismicityand A number of contrasting hypotheseshave been proposedto tectonics of' the area. account for the nonparallelism of the Trans-Mexican Volcanic Belt relative to the Middle American Trench and its relation to Tectonic Setting the subductedoceanic plate. These hypothesescan be divided This studyincludes the subductionof the Rivera andCocos into two groups: (1) those which favor a direct association plates in southernMexico, west of the intersectionof the between subduction and volcanism [e.g., Molnar and Sykes, 1969; Demant and Robin, 1975; Nixon, 1982; Burbach et al., Tehuantepecridge with the Middle American Trench (Figure 1). Accordingto Mammerickxand Klitgord [1982], the tectonic 1984; Suc•rezand Singh, 1986], and (2) those which suggest that the Trans-Mexican Volcanic Belt bears no direct tectonic historyof this regionfor the last 25 m.y. hassuffered at least three major plate reorganizations.The old Farallonplate relation with the subduction along the Middle American evolved first to the Guadalupe plate, which was later Trench and explain its origin as resulting from zones of weakness within the crust of southern M6xico, inherited from segmentedinto the actual Rivera and Cocosplates. The positionof thespreading centers moved east, from about 12.5 earlier episodes of deformation [e.g., Mooser, 1972; Gastil to 11 Ma, to the Mathematician seamounts. Later, between and Jensky, 1973; Shurbet and Cebull, 1973; Johnson and 6.3 to 3.5 Ma, it shifted to its actual position of the East Harrison, 1989]. Pacific Rise (EPR). An integral study that takes into account the geometry of the subducted Rivera and Cocos plates beneath the North Rivera Plate American lithosphere (NW of 94øW), the stressdistribution in the subducted slab, and the relation between the subduction Theexistence of thesmall Rivera plate was first suggested process and the anomalous location of the Trans-Mexican by Atwater [1970]. Since then, severalauthors have shown PARDOAND SUAREZ: SUBDUCTION OF TIlE RIVERAAND COCOS PLATES 12,359 that the Rivera plate is kinematicallydistinct from the North Guerreroregion. Hypocentrallocations were considered Americanand Cocosplates [Bandy and Yah,1989; Eissler and reliableand includedin the final datasetwhen they met the McNaIly, 1984;DeMets and Stein, 1990]. The junction followingcriteria: (1) a minimumof sevenphase readings betweenthe East Pacific rise and Rivera fracturezone is located withat leastone S wavearrival time; (2) a rootmean square 165km to thewest of theMiddle American Trench [Bourgois errorof lessthan 0.25 s; (3) estimatedhypocentral errors of and Michaud, 1991]. The preciselocation of the Rivera-Cocos lessthan 10 km; (4) a hypocentralvariation of lessthan 10 boundaryis still controversial,and no clear bathymetric km whenthe initialtrial depthwas varied from !0 to 150 km; featurescan be associatedto a distinctand clear plate boundary and(5) a maximumangle gap between the epicenterand the [Eisslerand McNally, 1984;Mamrnerickx, 1984; Bourgois local stations of less than 230 ø. From a total of over 3000 and Michaud, 1991]. Bandy [1992] suggestedthat the E1 events,a final subsetof only 1101 hypocenterssatisfied these Gordograben may be partof theRivera-Cocos plate boundary criteria (Figure 2). (Figure1). The Rivera lithosphere consumedat the trench is of late Relocated Earthquakes Mioceneage (~9 Ma)[Klitgord and Mammerickx,1982]. Althoughthe seismicityrelated to the subructionof the Rivera Earthquakeswith magnitudern b > 4.5 were relocatedin southernMexico, usingthe JHD method[Dewey, 1971] and platebeneath the Jaliscoblock is low (Figure2), at leastsix the phase readings reported by the International largeeathquakes (M s > 7.0) havebeen documented since ! 837, SeismologicalCentre (ISC) between1964 andJanuary 1990. includingthe great 1932 Jaliscoearthquake (Ms=8.2) [Eissler and McNally, 1984; Singh et al., 1985]. This contradictsthe Severalcalibration events were used along the region:(1) for the zonebetween 103 ø and 105ø, we usedthe largestaftershock suggestionthat the Riveraplate subductsaseismically [Nixon, of the 1973 Colima earthquake(Ms=7.5). This aftershock 1982]. On the basisof locally recordedmicroearthquakes and (Ms=6.2) was located with a temporary seismic network relocatedhypocenters of teleseismicevents, Pardo and Sudrez deployedafter the mainshock[Reyes et al., 1979]; (2) for the [1993]concluded that the subductionof the Riveraplate takes zone between 100ø and 103øW, the 1979 Petatlanearthquake placewith a steepangle (-50 ø) belowdepths of 40 km. (Ms=7.8)recorded with portablestations [Va!dds et aI., 1982] was used;and (3) the 1981 Ometepecearthquake (Ms=7.2) Cocos Plate [Nava, 1984] for the easternmostregion. The Cocos plate subductsat a rate that increasesto the The JHD method [Dewey, 1971] determines relocated southeastfrom 4.8 cm/yr at 104.5øWto 7.5 cm/yr at 94øW hypocenters relative to the hypocenter of the calibration [DeMetset al., 1990]. The age of the Cocosplate alsovaries event, applying time adjustmentsto the phasereadings of P, alongthe Middle American Trench, with jumps associatedto p P, and S. These time adjustments of the reporting fracturezones (Figure 1). To the east,the Tehuantepecridge seismologicalstations were obtained from the variance of the intersectsthe Middle AmericanTrench near--.95øW. This ridge different phasearrival times of the calibrationevents and from is the largestlinear bathymetricfeature on the easternflank of the earthquakesreported in this studyfor which the focal depth the East Pacific Rise. Apparently, it is the morphological was constrainedusing a long-period body wave inversion. expressionof a fracture zone that presumablyseparates crust These calibration station-phaseswere then used in a single- of significantlydifferent ages. A youngerand shallowercrust event location method [Dewey, 1971], in order to determine to the northwest differs in age to that southeast of the the relocatedhypocenters within the zonesof each calibration Tehuantepecridge by approximately10 to 25 m.y. [Couchand event. A sphericalEarth model was usedin the JHD methodto Woodcock, 1981]. relocate the hypocenters. For the eastern part of the region, we used the relocated hypocentersobtained by M. Cruz andG. Su•rez (manuscriptin Data Used in the Analysis preparation)who applieda similarprocedure. The precisionof The data used to infer the morphologyof the subducted each relocatedhypocenter relative to the calibrationevent is Rivera and Cocos plates beneath southernMexico are (1) estimatedby computingerror ellipsoidsfor the hypocentral accurately determined hypocenters obtained from local coordinatesat a 90% confidencelevel. A hypocentrallocation microearthquakedata recordedby permanentand temporary was consideredreliable when the ellipsoidsemiaxes were less seismologicalstations, and (2) eventsrecorded at teleseismic than 20 km; all events which did not meet these criteria were distancesand relocatedusing the methodof JointHypocenter discarded. A final subset of 207 reliable hypocenterswas Determination(JHD) [Dewey, 1971]. For someof the larger obtained from an initial set of over 300 events. The events(m/• >_ 5.0), the focal depth was constrainedfrom the hypocentraland focal parameters of all theearthquakes studied inversionof long-periodbody waves [Ndbelek, 1984]. are shown on Figure 2 and provided as an electronic supplementfor thereader. Local Microearthquake Data Long-Period Body Wave Inversion All the available Iocal data from the permanent and temporaryseismic networks installed in southernMexico were The formal inversion of P, S V, and S H waveforms used in order to accurately locate the hypocentersof constrainsthe focal mechanismand the centroidaIfocal depth microearthquakeswithin the zone [Nava, 1984; UNAM [NdbeIek, 1984]. Long-period seismogramsof teleseismic SeisrnoIogyGroup, 1986; Sansores, 1990; Pardo and Suc•rez, earthquakes(25 ø > A > 90ø ) from the Global Digital 1993;Ponce et al., 1992;Zu•iga et al., 1993]. SeismographNetwork (GDSN) were usedfor the eventsthat The hypocenterswere locatedusing HYPO7! [Lee and occurred between 1980 and 1987 and from the World-Wide Valdes,1985] with the velocitymodel reported by Sudrezet al. StandarizedSeismograph Network (WWSSN) for the events [1992],and a velocityratio (Vp/V s = 1.76)determined forthe that occurredprior to 1980. The analogseismograms were 12,360 PARDO AND SUAREZ: SUBDUCTION OFTHE RIVERA AND COCOS PLATES

ß ß

ß +-i--I- " .•

++.

.m PARDOAND SUAREZ: SUBDUCTION OFTHE RIVERA AND COCOS PLATES 12,361 hand-digitizedand filtered with a high-passfilter with cutoff the southeast,the maximumdepth observed for earthquakesin frequencyof 0.017 Hz (60 s) to eliminatelong-period noise. the Cocosplate, between the E1 Gordograben and the point All of theevents modeled are in themagnitude range 5.0 >_ rnt, where the Orozco Fracture Zone apparentlyintersects the _>6.1, and therefore a point sourcewas assumedin all cases. trench, is 100 km. Farther east, in the region bounded Keyexamples of themodeled events are shown on Figure 3. approximatelyby the onshoreprojection of the Orozco and the O'Gormanfracture zones, where a goodcoverage of local Analysisof Hypocentral Data stationsexists, the maximum focal depths observedare consistentlyless than 70 km. Between the O'Gorman fracture In plan view, the seismicityof southernMexico showsa zone and the TehuantepecRidge the maximum depth of distributionin bands parallel to the trench [Sudrez et al., earthquakesincreases continuosly down to depthsof 180 km. 1990](Figure 2). The first band alongthe coastis relatedto The depthextent of the various segmentsof the subduction the interplateseismogenic contact. North of it lies an almost zonedirectly correlates with the observeddip of the slab. aseismic zone, which becomes broader between 96øW and Smooth contours of the Wadati-Benioff zone were obtained 10!øW. Fartherinland, a secondband of deeperintraplate using a two-dimensionalspline interpolation [Smith and events(depths > 50 km) in the oceanicdowngoing slab lies at Wessel,1990] (Figure6). In centralMexico, however,to the a distanceof 150-250 km from the trench(Dt). Beyondthis northof 18øNand at distancesbeyond 300 km from the trench, last band, which lies immediately to the south of the Trans- there is a remarkabledearth of seismicevents (Figures 2 and MexicanVolcanic Belt, the seismicactivity disappears. 6). Thusthe 80- and 100-km-deepcontours cannot be mapped In order to map the geometryof the subductedslab, 12 cross directly in central Mexico. sectionsof the seismicitywere drawn (Figures2 and 4). The These two contourswere drawn following the trend of the originof all the crosssections is the trench,and they are all slab at shallowerdepths and extrapolatingtheir locationfrom orientedin the directionof convergencepredicted between the the regionswhere they are well defined by the seismicity:to Rivera-North America [DeMets and Stein, 1990] and Cocos- the westof 103øWand east of 96øW (Figure 6). The resulting NorthAmerica [DeMets et al., 1990] platepairs (Figure2). contoursshow that the Cocos plate lies at depthsof 80 to 100 The shallow part of the subductionzone (h < 30 km) km beneath the southern front of the Trans-Mexican Volcanic indicatesa constantinterplate geometrythat initially dips at Belt (Figure 6). This depth of the Wadati-Benioff zone of ~10ø and gradually increasesto --25ø at a depth of 30 km approximately100 km beneaththe volcanicarc is observedin (Figure 4). This geometry is observed throughout the most subructionzones of the world. This suggeststhat the subruction zone in southern Mexico and appears to be Trans-Mexican Volcanic Belt is directly related with the independentof the age and the relative convergencevelocity subruction of the oceanic slab beneath the North American of the subducted oceanic plate; no lateral changesof the plate and that the observednonparallelism of the volcanicbelt interplate geometry are observed where major bathymetric with the trenchis mainly due to the geometryof the subducted featuresare being subducted. plate.

Geometry of the Subducted Plate Focal Mechanisms and State of Stress The subductedRivera and Cocos plates show rapid lateral In this study, 21 new focal mechanismsof earthquakes variationsin dip at depthsgreater than -30 km (Figure4). In within the subductingCocos plate are reported(see electronic the Isthmusof Tehuantepec(cross section AA' on Figures2 supplements).The lower hemisphericprojection of the focal and 4a), the seismicitydefines a Wadati-Benioff zone dipping mechanismsobtained from a waveform inversion reportedin at an angle of-30 ø. Immediatly toward the west (Figure4b), this studyand in previousworks are shownon Figure 7, and the dip of the downgoingslab is shallowerand decreasesto an severalkey examplesof the earthquakesmodeled in this study angle of-25 ø. In central Mexico, the geometry of the are shownon Figure 3. These focal mechanismsare also downgoingslab becomesalmost subhorizontal at distancesof shownon lateralprojection on the crosssections (Figure 4). between110 km to 275 km from the trench and at depthsof Owingto the absenceof seismicstations to the southand west about50 km (Figures 4c to 4g). A similar geometrywas of southern Mexico, the focal mechanisms of small inferredfor theGuerrero region (Figures 4f and4g) bySudrez et earthquakesreported from the first motionof P wavesare al. [1990] andSingh and Pardo [ 1993] usinglocal data. poorlyconstrained. Therefore only onemechanism from first In the westernmostregion of the Cocosplate (Figures4h motion data alone was used; its small magnitudeprecluded the and4i), the dip of the subductedslab changes again rapidly use of a formal inversion scheme (event 67 on the electronic from the subhorizontalsubduction observed in the region of supp!ementl). Guerreroand Oaxaca (Figures 4c to 4g) to a steeperdip of-30 ø. In a manneranalogous to the seismicity,the stresspattern This dip angie is similarto that observedon Figure3a, near observedalong the subductionzone may be dividedin zones the Isthmusof Tehuantepec.Toward the northwest,the slab continuesto steepenprogressively, and beneaththe Colima 1An electronic supplement of this material may be obtained on a volcano,the Wadati-Benioffis celarlyfollowed dipping at an disketteor AnonymousFTP from KOSMOS.AGU.ORG. (LOGIN to angleof ~50ø downto a maximumdepth of approximately130 AGU's FTP account using ANONYMOUS as the usernameand km (Figure4j). GUEST as the password. Go to the right directoryby typing CD APEND. TypeLS to seewhat files are available.Type GET andthe A crosssection parallel to thetrench, from (19øN, 106øW) name of the file to get it. Finally, type EXIT to leave the system.) to (15øN, 95øW), showsthe rapid and drasticvariations in (Paper95JB00919, Shape of the subductedRivera and Cocos Plates in southernMexico: Seismicand tectonic implications,by Mario Pardo maximumdepth extent of the hypocenterswithin southern and Gerardo Su.4rez). Diskette may be ordered from American Mexico(Figure 5). In thecase of theRivera plate beneath the GeophysicalUnion, 2000 FloridaAvenue, N.W., Washington,DC Jaliscoregion, the maximumfocal depth is 130 kin. Toward 20009;$15.00. Payment must accompany order. 12,362 PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES

, IdLUe ts Id

• •, ...s.... :.•.-. .

idtue PARDO AND SUAREZ: SUBDUCTION OFTHE RIVERA AND COCOS PLATES 12,363

• 8ø • 12,364 PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES

< m 0 z (D (D Z

. ?

o o Iduae IduJe

o PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES 12,365

• o idt•e Idu•e 12,366 PARDO AND SUAREZ: SUBDU•ON OFTHE RIVERA AND COCOS PLATES

/ / / /

o o .• o o o • o o o • o PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES 12,367

RIVERA • N COCOS JALISCOMICHOACAN o GUERRERO kmo ,- ', ,, '

(•ou, •w) to (•su, 1o 2• • t : •'" ' I ' ' I ' ] ' I L K d I H • F E D C B A 40

20 cm/yr m.y. l__w.a----I1-.• *ß AgeRate [m.y.b.p.][cm/yr] I ' I ' I ' I ' I ''' I ' I • I ' ß I ' I ' I " ' I . 0 200 400 600 800 1000 1200 km

Figure5. Relocatedhypocenters projected onto a crosssection parallel to thetrench, from (19øN, 106øW) to (15øN,95øW); symbols as on Figure1. SouthernMexico can be separatedinto four major segments: Jalisco,Michoacan, Guerrero, and Oaxaca. The bottom part shows the variations in ageof thesubducted slabat thetrench [Klitgord and Mammerickx, 1982] and in therate of relativemotion of theoceanic plates at thetrench [DeMets and Stein,1990, DeMets et al., 1990]. Crosssections A to L arethose shown on Figures2 and 4. parallelto thetrench (Figures 4 and7). A shallowdepth zone subduction zone are the relative convergence rate in the closeto the trench showsinterplate thrust faulting followed subductionzone, the age of the subductedslab, the absolute bydowndip, normal faulting events. The thrustearthquakes motion of theoverriding plate, and the subductionof aseismic areobserved down to a maximumdepth of-25 km alongthe bathymetricfeatures, like ridgesor intraplateseamounts [e.g., Cocosplate (Figure 4). Crossand Pilger, 1982; Jarrard, 1986]. Exceptfor the age of Downdipfrom the interplatecontact, a zoneof lowseismic the slab,all of theseparameters correlate negatively with the activityis observedwhich is followedfarther inland by anarea dip of the downgoingplate. Thusyoung lithosphere is of intraplateevents showing normal faulting at depthsgreater apparentlymore buoyant than old platesand subduers at a than50 km. The normal faulting eventsin the downgoing lower dip angle. slab(Figure 7) show,in general,T axesoriented in a direction Therefore, accordingto these empirical relations, the parallelto the gradientof the subductedoceanic plate (Figure shallowdip observedin the subductionzone of southern 6). Only wherethe slab showsa rapidchange in dip, near Mexicocould be explainedby the relativelyyoung age of the 96øW and 101øW, the stress distribution become more subductingslab, its moderateconvergence rate, and the rapid complex(Figure 8). absolutemotion of theupper North American plate relative to Beneaththe normalfaulting events near the coastline,there the trench(Figure 1). To the eastof theTehuantepec ridge, is a sheetof compressionalevents, with depths ranging from wherethe Cocosplate subducts beneath the Caribbeanplate 40 km to 50 km (events47, 70, 158, 28, 152 and67 on the witha steeperdip angle,the subducting slab is 10 to 25 m.y. electronicsupplement; Figures 4b, 4d, 4e,4f, 4h and4j). This olderthan to the westof TR [Couchand Woodcock,1981], and doublezone in stressis probablyassociated to flexural the relativemotion of the upperCaribbean plate towardthe stressesassociated to the slab bending upward to become trench is lower than to the northwest[Gripp and Gordon, subhorizontal[SutSrea et al., 1990]. 1990] (Figure 1). Practicallyno crustal seismicityis observedin the SouthwesternMexico is the region of the subductionzone overridingplate over the subducted Cocos plate (Figure 4). In wherethe Cocos plate is youngest.In thissame manner, the contrast,there is a relativelyhigh crustalseismicity within ageof the subductingRivera plate is alsorelatively young. theJalisco region in westernMexico (Figures 2 and4j-41), However,the dip of the downgoingslab in this areais steep. suggestingthat the deformationof theJalisco block is greater Thusthe simple correlation between the age of theslab and its thanthat observed in the NorthAmerican plate over the Cocos dip doesnot holdin thiscase. The only majordifference betweenthe Riveraand the Cocossubduction zones is the slow slab[Pardo and SutSrez,1993]. convergencerate betweenthe Riveraand NorthAmerican plates(---2 cm/yr) [DeMets and Stein,1990], relative to a Discussion greaterCocos-North America convergence rate (-6 cm/yr) Althoughthe conditionscontrolling the geometryof [DeMetset al., 1990](Figures 1 and4). Perhapsthis low rate subductionzones are still controversial,several authors have of relativemotion becomes an importantparameter affecting suggestedthatthe principal factors affecting the geometry of a thegeoinetry of thesubducted Rivera plate. 12,368 PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES

II I

t I o

• +'++ • ,• / , •-+++ '•++++++++•-I-+ -t- -I- • ... * -' r.+ , ++ .1.+ ++++ +++ .1.+ +%.1. 4-++•+++ -•''+ +_++ %+ +++++++++++++•-• + + + + + .•.,+ f + 4- + .4-+ i-+++-.1.+.1.4 ' ++++' ! I ß k+++++++, M-+++++'+ •-, + + + + r "-t- • .1.++.1.++ + + + ++ +%+ +- +-t-+ + +++++_, cq.x +++++++ + + + + +++++-+ + + -I!+•,+ .ß •+++++++++•-,-+++++ ++++I +. ß + + +-L,•- + + + +•.+ + • . + • •.+ + + ++l•p++l ß

+ + + + + + + + :•1•+ -t• 0 + + + + + + + + F++++++++. + + + + + + +'11111•+•+ Y + + + 4- + + + -t--lq•- + + •+ %++ + +++++•"+ + + •++ ++++p-+++-• + + ++ ++++:•'*++++dl•+.+ T'• + + ++' ..•• •- ++ ++ ++ + + + + + +++,•++++I+ + • + ++•'++ +'• 0 0<: 0

'•'+ + + ß + +++'• . ß -+++• +•l•_% + + +-• ' •,11 o

o . U')

ß ß \

ß1. ,. ++.'+'

/ o '• Idd3

v- I PARDO AND SUAREZ: SUBDUCYION OF THE RIVERA AND COCOS PLATES 12,369 12,370 PARDO AND SUAREZ: SUBDUCTION OF THE RIVERA AND COCOS PLATES

O oZ..c:

O or) -,-•

I I

'._..•

! ! I

• I z tic: I % v- ILl

I--

o I

/ i I

i PARDO AND SUAREZ: SUBDUCTION OFTHE RIVERA AND COCOS PLATES 12,371

earthquakesalong the coupled interplate contactin southern Mexico [Singh and Mortera, 1991] is probably related to variationsin the strengthof couplingalong the plate interface 50 and not to the geometry of the subductionor to the depth extent of the coupledinterplate zone. kill The depthextent of the seismogenicplate contactcan be estimated from the maximum depth of the shallow thrust 150 earthquakes[e.g., Sudrezand Comte,1993]. In the caseof the Cocosplate, this depth is consistentlyshallower than .--25 km. This resultsyield an unusuallynarrow seismogeniczone 0 which is only .-.60 km wide. This shallow depth of seismogeniccoupling is abouthalf as deepas that observedin 50 most subductionzones of the world [e.g., Pacheco et al., -g 1993] km , In the case of the Rivera plate, the maximum depth of seismogeniccoupling appears to be slightlydeeper and reach 2 (GUERRERO) ß-.40 km. This observation is suggested by a coastal 150 .'" ,' . , . , . , . ', . , '.', . - earthquakeshowing a nodal plane dipping to the northeastat an angleof 34ø (event92 on Figure4j and on the electronic . I ',,,, i supplement). If this earthquakestill reflectedthe interplate contact,the width of the seismogeniczone of the Rivera plate 5O would be wider (.-.75 km) than that of the Cocos plate, and, potentially, earthquakesof greater magnitude could be h generatedon its plate interface. Interestingly,the largest earthquakethis centuryon the Middle AmericanTrench, the 1932, Jaliscoearthquake (Ms=8.2), occurredon the Rivera- 150km Iv 3 . (MICHOACAN) , . , , , . , . i I l'"' I ' North America plate interface[Eissler and McNally, 1984; Singh et al., 1985]. _ .• . I • I ,• I ß 1 The lateralvariations in the dip angleof the subductedplate at depthsgreater than 30 km (Figure4, 5, and6) suggestthat southern Mexico can be divided into four regions bounded 5O approximatelyby major bathymetric features identified offshore(Figures 6 and9). (1) The Jaliscoregion, where the kill steep-dippingRivera plate subducts beneath North Americais boundedto the east by the subductionof the El Gordo graben. 150 4 (JALISCO)...... IxX j_, In fact, the significantdifferences in the dip of the slab on both sides of the El Gordo graben may indicate that this 0 I i ,,,I ., 1 tectonicfeature decouples the downgoingRivera and Cocos plates. (2) The Michoacanregion, is a transitionzone between the steep subductionof the Rivera plate and the 50 almost subhorizontal subduction to the east. As mentioned before,it is boundedby the E1 Gordograben to the westand by krn the onshoreprojection of the Orozco fracturezone to the east. (3) The Guerrero-Oaxaca region shows an almost 150 , , , , , subhorizontal geometry of the subducted Cocos plate. Geographically,this region can be definedas boundedby the 0 I00 200 300 km 400 Orozco and the O'Gorman fracture zones. (4) The southern Oaxacaregion in southeasternMexico is a transitionzone Figure 9. Comparisonof the crosssections of the four betweenthe flat subductiongeometry beneath Guerrero and the regionsinto whichsouthern Mexico was divided based on the steepsubduction geometry of the Cocosplate beneaththe seismicityand shape of the subductionzone. The seismicity Caribbeanplate towards the east. andthe plate geometryare a superpositionof the cross The remarkable absence of seismic events with depths sectionson Figure4, andare labeledwith the sameletters. greaterthan 70 km, northof---18øN and of theTrans-Mexican Thebottom figure shows a comparisonof the geometryof Volcanic Belt suggeststhat the downgoingplate lossesits subductionin eachof theseregions in southernMexico. brittle behavior and sinks aseismically into the mantle to the north of the volcanic belt. Perhaps low-magnitude Seismic and Tectonic Implications earthquakesoccur which are too smallto be detectedby the The depthextent and dip of the interplatemegathrust is insufficient seismic network coverage that exists today. The constantand doesnot exhibitany clear correlationwith the inferred80- and 100-km-deepcontours of the downgoingslab age of the subductedplate, with the relative convergence appearto lie beneaththe southern front of theTrans-Mexican velocity,or with the subductionof bathymetricfeatures along VolcanicBelt (Figure 6). This observationstrongly suggests thetrench (Figures 4 and6). Thusthe different rupture modes the associationof this volcanic belt with the subductionof the andcomplexity of the seismic sources reported for large thrust oceanicplate beneath the North Americanplate. Thusthe 12,372 PARDO AND SUAREZ:SUBDUCTION OFTHE RIVERA AND COCOSPLATES observednonparallelism of the Trans-Mexican Volcanic Belt Bandy,W.L., and C. Y. Yan, Present-dayRivera-Pacific and Rivera- with the trench is mainly due to the geometryof the subducted Cocosrelative plate motions (abstract) Eos Trans. AGU, 70, 1342, 1989. Rivera and Cocos platesin southernMexico. Beyis,M., and B.L. Isacks, Hypocentral trendsurface analysis: Probing' thegeometry of Benioffzones, J. Geophys.Res., 89, 6153-6170, Summary and Conclusions 1984. Bourgois,J., and F. Michaud, Active fragmentationof theNorth The shapeof the subductedRivera and Cocosplates beneath Americanplate at theMexican triple-junction area off Manzanillo, the North American plate in southernMexico was determined Geo Mar. Lett., 11, 59-65, 1991. using accuratelylocated hypocentersof earthquakes. A total Burbach,G.,C. Frolich, W. Pennington, andT. Matumoto,Seismicity andtectonics of the subduetedCocos plate, J. Geophys.Res., 89, of 1101 microearthquakes hypocenters recorded by local 7719-7735, 1984. networks and 207 relocated hypocenters of earthquakes Chael,E. P., andG.S. Stewart,Recent large earthquakes along the recorded at teleseismic distances were used as the database. For middle Americantrench and their implicationsfor the subduction 2! teleseismicevents (rob>5.0), the source parameterswere process,J. Geophys.Res., 87, 329-338, 1982. determinedusing a formal long-period,body wave inversion Couch, R., and S. Woodcock, Gravity and structureof the continental marginsof southwesternMexico and northwesternGuatemala, j. [Ndbelek, 1984]. In the relocationprocedure [Dewey, 1971], Geophys.Res., 86, 1829-1840,1981. the focal depth of these latter events was constrainedby the Cross,R.H., andT.A. Pilger, Controlsof subductiongeometry, location inversion and used within the initial subset of events to of magmaticarcs, and tectonicsof are and back-arcregions, Geol. generatethe variance of the residualsat each teleseismic Soc. Am. Bull., 93, 545-562, 1982. Dean, B.W., and C.L. Drake, Focal mechanism solutionsand tectonics station, which were then applied to all the events in the of the Middle America arc, J. Geol., 86, 111-128, 1978. relocalization scheme. Delgado,H., P. Cervantes,and R. Molinero, Origende la faja The subructionin southernMexico may be approximatedas vo!e/inica trans-mexicanahate 8.3 May sus migracioneshacia el a subhorizontalslab subductingbeneath the North American W, SW y SSW desde5.3 Ma (abstract)GEOS Uni6n Geofs.Mex., plate,which is boundedby a steepsubruction geometry of the 13, 31-32, 1993. Cocosplate beneaththe Caribbeanplate to the east and the Demant, A., and C. Robin, Las rasesdel volcanismoen M6xico;una sfntesisen relaci6ncon !a evoluci6ngeodin•mica desde el Cret•cico, Riveraplate beneaththe North Americanplate to the west. No Rev. Inst. Geol., UNAM, 75, 70-82, !975. lateral changesin the dip of the coupled interplate zone are DeMets, C., and S. Stein, Present-daykinematics of the Riveraplate observedat depthsof lessthan 30 km. The changesin dip of and implicationsfor tectonicsin southwesternMexico, J. Geophys. the downgoing slab geometry are observed once it is Res., 95, 21,931-21,948, 1990. DeMets, C., R.G. Gordon, D.F. Argus, and S. Stein, Currentplate decoupled from the overriding plate. The width of the motions,Geophys. Res. J. [nt.,101,425-478, 1990. seismogeniczone of the Rivera plate is apparentlywider (--,75 Dewey,J.W., Seismicitystudies with the methodof joint hypocenter km) than that of the Cocos plate (~60 kin). determination,Ph.D. thesis,Univ. of Calif., Berkeley, 1971. The downgoingslab shows drastic changes in the dip angle, Eissler,H., and K.C. MeNally, Seismicityand tectonics of theRivera ranging from subhorizontal subduction beneath central plate and implicationsfor the 1932 Jalisco,Mexico, earthquake, J. Geophys.Res., 89, 4520-4530,1984. Mexico to a relativelysteep dip wherethe Tehuantepecridge is Gastil, R.G., and W. Jensky, Evidencefor strike-slipdisplacement being subducted. These changesin the dip of the downgoing beneaththe Trans-Mexicanvolcanic belt, StanfordUniv. Publ. Geol. slabdo not appearto take place on tear faults, suggestingthat Sci., 13, 171-180, 1973. smoothcontortions accomodate these dip changes. The 80- Gonz•lez-Ruiz,J., Earthquakesource mechanics and tectonophysics of the middle America subruction zone in Mexico, Ph.D. thesis,Univ. and 100-km-depthcontours of the subductedplate lie beneath of Calif., Santa Cruz, 1986. the southern front of the Trans-Mexican Volcanic Belt, Gripp,A., andR. Gordon,Current velocities relative to thehotspots suggestingthere is a direct associationof this abnormally incorporatingthe NUVEL-1global plate motion model, Geophys. aligned volcanic belt with the complex subruction geometry Res.Lett., 17, I ! 09-1112, 1990. of the subductedRivera and Cocosplates in southernMexico. Hanus,V., andJ. Vanek,Subduction of theCocos plate and deep active fracturezones of Mexico, Geofis.Int., 17, 14-53, 1978. Havskov,J., S.K.Singh, and D. Novelo,Geometry of theBenioff zone Acknowledgments. We thank the Servicio Sismo16gicoNacional, in theTehuantepec area in southernMexico, Geofis. !nt., 21, 325- the !nstitutode Geoffsica,and the Institutode Ingenieria,all of them at 330, 1982. UNAM, and the Universidadde Colima for facilitatingdata from their Jarrard,R.D., Relationsamong subduction parameters, Rev. Geophys., local and temporarynetworks. We are gratefulto S.K. Singh and J. 24, 2 ! 7-284, 1986. Paehecofor helpful discussionsand a review of the manuscript.The Johnson,C.A., and C.G.A. Harrison, Tectonics and volcanism incentral manuscriptwas greatlyimpoved by commentsmade by D. E. Jamesand Mexico:A Landsatthematic mapper perspective, Remote Sens. an anonymous reviewer. Thanks are due to W.H.F. Smith and P. Environ., 28, 273-286, 1989. Wessel for the use of the program GMT to draft the figures. This Klitgord,K., and J. Mammerickx, Northern East Pacific Rise: Magnetic research was partially supported by CONACYT and by project anomalyand bathymetric framework, J. Geophys. Res., 87, 6725- UACPyP-UNAM 30309. One of us (M.P.) acknowledgesa fellowship 6750, 1982. from the Secretaria de Relaciones Exteriores, M6xico. Lee,W.H.K., and C. Valdes,HYPO71PC: A personal computer version of theHYPO71 earthquake location program, U.S. Geol. Surv. Open References File Rep. 85-749, 1985. LeFevre,L.V., andK.C. McNally, Stress distribution and subduction on Allan, J.F., S.A. Nelson, J.F. Luhr, I.S.E. Charmichael,M. Wopat, and aseismicridges in themiddle America subduction zone, J. Geophys. P.J. Wallace, Pliocene-Holocenerifting and associatedvolcanism in Res., 90, 4495-4510, 1985. southwestMexico: An exotic terrane in the making, in The Gulf and Luhr,J., S. Nelson, J. Allan,and I.S.E Charmichael, Active rifting in PeninsularProvinces of the Califomias,edited by J.P. Dauphinand southwesternMexico: Manifestations of an incipienteast-ward R.R.T. Simoneit, AAPG Mere., 47, 425-445, 1991. spreading-ridgejump, Geology, 13, 54-57,1985. Atwater, T., Implicationsof plate tectonicsfor the Cenozoictectonic Mammerickx,J., Themorphology of propagatingspreading centers: evolution of western North America, Geol. Soc. Am. Bull., 81, 3513- Newand old, J. Geophys.Res., 89, 1817-1828,1984. 3536, 1970. Mammerickx,J.,and K. Klitgord, Northern East Pacific Rise: Evolution Bandy,W.L., Geologicaland geophysicalinvestigation of the Rivera- from25 m.y.B.P. tothe present, J.Geophys. Res., 87, 6751-6759, Cocosplate boundary:Implications of plate fragmentation,Ph.D. !982. thesis,Tex. A&M Univ., CollegeStation, 1992. Molnar,P., Faultplane solutions ofearthquakes anddirection ofmotion PARDOAND SUARF. Z:SUBDUCI•ON OFTHE RIVERA AND COCOS PLATES 12,373

inthe Gulf of Californiaand on the Rivera Fracture zone, Geol. Soc. Singh,S.K., L. Ponce,and S.P. Nishenko, The great Jalisco, Mexico, Am. Bulk, 84, 1651-1658, 1973. earthquakeof 1932: Subductionof the Rivera Plate, Bull. SeisrnoI. Molnar,P., andL. R. Sykes, Tectonicsof the Caribbeanand Middle Soc.Am., 75, 1301-1313,1985. Americanregion from focal mechanisms andseismicity, Geol. Soc. Smith,W.H.F., andP. Wessel,Gridding with continuouscurvature Am. Bull., 80, 1639-1684, 1969. splinesin tension,Geophysics, 55, 293-305,1990. Mooset,F., TheMexican volcanic belt: Structure andtectonics, GeofiT StoiberR.E., and M.J. Carr, Quaternaryvolcanic and tectonic Int., 12, 55-70, 1972. segmentationof CentralAmerica, Bull. Volcanol.,37, 304-325, 1973. N/ibelek,J.L., Determinationof earthquake source parameters from Su/trez,G., andD. Comte,Comment on "Seismiccoupling along the inversionof bodywaves, Ph.D. thesis, 346 pp.,Mass. Inst. of Chilean subductionzone" by B.W. Tichelaar and L.R Ruff, J. Technol.,Cambridge, 1984. Geophys.Res., 98, 15,825-15,828,1993. Nava,E., Estudiode los temblores de Ometepee del 7 deJunio de 1982 Su,Srez,G., and S.K. Singh, Tectonicinterpretation of the Trans- y susreplicas, thesis, Univ. Nac. Aut6noma de M6xico, M6xieo D.F., MexicanVolcanic Belt-Discussion, Tectonophysics, I27, 155-160, 1984. 1986. Nixon,G.T., The relationshipbetween Quaternary volcanism in Sugrez,G., T. Mortfret,G. Wittlinger, and C. David, Geometryof centralM6xico and the seismicityand structure of subduetedoceanic subductionand depthof the seismogeniczone in the Guererogap, lithosphere,Geol. Soc. Am. Bull., 93, 514-523,1982. Mexico, Nature, 345, 336-338, 1990. Pathcoo,J.F., L.R. Sykes, and C. H. Scholz,Nature of seismic coupling Su,Srez,G., J.P.Ligorda, and L. Ponce,Preliminary crustal structure of alongsimple plate boundaries of the subduction type, J. Geophys. the coastof Guerrero,Mexico, using the minimumapparent velocity Res.,98, 14,133-14,159, 1993. of refractedwaves, Geofis. Int., 31, 247-252, 1992. Pardo,M., andG. Su•ez, Steepsubduction geometry of theRivera UNAM SeismologyGroup, The September1985 Michoacan platebeneath the Jalisco block in westernMexico, Geophys. Res. earthquakes:Aftershock distribution and history of rupture,Geophys. Lett., 20, 2391-2394, 1993. Res. Lett., 13, 573-576, 1986. Ponce,L., Gaulon,G. Sufu'ez,and E. Lomas, Geometryand state of Vald6s,C., R. Meyer,R. Zufiiga,J. Havskov,and S.K. Singh, Analysis stressof thedowngoing Cocos plate in theIsthmus of Tehuantepec, of the Petarianaftershocks: Numbers, energy release and asperities, Mexico, Geophys.Res. Lett., 19, 773-776, 1992. J. Geophys.Res., 87, 8519-8529, 1982. Reyes,A., J. Brune, and C. Lomnitz, Sourcemechanism and aftershock Zufiiga,F.R., C. Gutierrez,E. Nava, J. Lermo, M. Rodriguezand R. studyof the Colima,Mexico earthquake of January30,!973, Bull. Coyoli, Aftershocksof the sanMarcos earthquake of April 25, 1989 SeismokSoc. Am., 69, 1819-1840, 1979. (Ms=6.9) and some implicationsfor the Acapulco-SanMarcos Sansores,L., Sismicidad,esfuerzos y atenuaci6n previos y posteriores seismicpotential, Pure AppI. Geophys., 1;10, 287-300, 1993. al terremorodel 29-Xi-1978,Oaxaca (Ms=7.8), thesis, Univ. Nac. Aut6nomade M6xico, Mexico D.F., 1990. M. Pardo,Departamento de Geoffsica,Universidad de Chile, Casilla Shurbet,D.H. and S.E. Cebull, Tectonicinterpretation of the Trans- 2777, Santiago,Chile. (e-mail:mpardo@dgf. uchile.cl) Mexicanvolcanic belt, Tectonophysics,101, 159-165,1973. G. Su,Srez,Instituto de Geoffsica, Universidad Nacional Aut6noma Singh,S.K., and F. Mortera, Sourcetime functionsof largeMexican subductionearthquakes, morphology of theBenioff zone, age of the de M6xico, Apartado Postal 70-172, Mexico D.F., Mexico. (e-mail: plate,and their tectonicimplications, J. Geophys.Res., 96, 21,487- [email protected]) 21,502, 1991. Singh,S.K., and M. Pardo, Geometryof the Benioffzone and state of stressin the overridingplate in centralMexico, Geophys. Res. Lett., (ReceivedJuly 7, 1994;revised March 1, 1995; 20, 1483-1486, 1993. acceptedMarch 16, 1995.)