Geometry and state of stress of the Wadati-Benioff zone

in the Gulf of Tehuantepec, Mexico

Hugo BravoGerencia de Estudios de Ingenieria Civil, Departamento de Sismotectonica, Mexico D. F., Mexico

Cecilio J. RebollarCentro de Investigacion Cientifica y Educacion Superior de Ensenada (CICESE), Ensenada, Baja California, Mexico

Antonio Uribe and Oscar JimenezGerencia de Estudios de Ingenieria Civil, Departamento de Sismotectonica, Mexico D. F., Mexico

Received 15 October 2003; revised 19 January 2004; accepted 12 February 2004; published 16 April 2004.

[1] Data recorded in the Gulf of Tehuantepec and published moment tensor solutionswere used to infer the geometry of the Wadati-Benioff zone (W-B zone) and the stressdistribution of the subducted Cocos Plate. From northwest to southeast the subductedslab gently increases its dip, bends up at depths of the order of 100 km and about240 km inland from the trench, then unbends and increases its dip toward thesoutheast below Chiapas. Contour lines of equal depth at 50 and 100 km show asignificant contortion of the W-B zone west of the Isthmus of Tehuantepec and coincidewith significant changes in the topographic features of the crust. We calculated a Vp/Vs

ratio of 1.75 for the whole area and a Moho depth in the Gulf of Tehuantepec of 28.5 ±3.5 km. Near the trench, along the coupling zone, T axes of interplate and intraplateearthquakes are oblique to the interface and parallel to the dip direction of the subductedplate, respectively. At depths greater than 50 km the T axes are roughly horizontal. Bestfit stresses using the Gephart and Forsyth [1984] code resulted in the followingvariation of the minimum stresses (T axes) from northwest to southeast as (plunge,azimuth) s3(40�, 37�) to s3(28�, 72�), then to s3(11�, 54�) and finally to s3(27�, 35�).We also analyzed recordings from the 30 September 1999, Oaxaca earthquake Mw = 7.5and its aftershocks. We inferred an approximate fault area of 80 by 30 km inside thesubducted Cocos Plate. INDEX TERMS: 7203 Seismology: Body wave propagation; 7205

Seismology: Continental crust; 7230 Seismology: Seismicity and seismotectonics; KEYWORDS: seismicity,

Wadati-Benioff zone, stress distribution

Citation: Bravo, H., C. J. Rebollar, A. Uribe, and O. Jimenez (2004), Geometry and state of stress of the Wadati-Benioff zone in the

Gulf of Tehuantepec, Mexico, J. Geophys. Res., 109, B04307, doi:10.1029/2003JB002854.

1. Introduction

[2] The convergence of the Tehuantepec Ridge (TR)between Oaxaca and Chiapas has produced a complexpattern of deformation in the continental crust as well asin the subducted lithosphere, and consequently, it hasgenerated a high rate of seismic activity in the whole area.West of the TR the dip of the subducted plate is �20�, andhypocenters of the earthquakes are no deeper than 100 km[Pardo and Suarez, 1995]. On the other hand, east of theTR the subducted slab deepens at an angle of 40�, and thehypocenters can reach depths of the order of 270 km[LeFevre and McNally, 1985; Rebollar et al., 1999a,1999b]. This zone has been proposed as a seismic gap[Kelleher et al., 1973; McCann et al., 1979; Singh et al.,1981; McNally and Minster, 1981; Astiz et al., 1988].

LeFevre and McNally [1985] suggested an aseismic sub-duction of the TR. Figure 1 shows the location of the TRand the controversial location of the Tehuantepec gap.Ponce et al. [1992] carried out a seismic study in this area,and they found that the dip of the subducted slap occurredgradually over a length of 150 km parallel to the trenchfrom west to east; however, they reached that conclusionusing a limited data set.[3] In this study, we used data from a network of 13

analog stations deployed in Oaxaca and Chiapas that span alarge period of recording. This network has been operatedsince January 1999 up to now by the Departamento deSismotectonica de la Comision Federal de Electricidad(CFE). In this study, we report the analysis of the seismicactivity recorded from January 1999 to September 2002,and we infer the geometry of the Wadati-Benioff (W-B)zone, the depth of the crust using the converted S-to-Pphases (Sp), and from published moment tensor solutions(MTS) the stress orientations on the subducted Cocos Plate

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109, B04307, doi:10.1029/2003JB002854, 2004

Copyright 2004 by the American Geophysical Union.0148-0227/04/2003JB002854$09.00

B04307 1 of 14

in the Tehuantepec Ridge. We also report the aftershocks ofthe intraplate Oaxaca normal earthquake of 30 September1999 Mw = 7.5 magnitude event.

2. Tectonic Setting

[4] In the area under investigation the Cocos (CO),Caribbean (CA), and North American (NA) Plates(Figure 1) converge. The oceanic Cocos Plate subductsbeneath North American and the Caribbean Plates whichgive origin to the Middle American Trench (MAT). Themotion between the Caribbean and the North AmericanPlates takes place along the Motagua-Polochic Fault Zone(MOFZ), a large sinister strike-slip fault. The CO Platemoves at a rate of �7.1 cm/yr in the northeast direction, theNA Plate moves in a west-southwest direction at a rate of�3.0 cm/yr, and the CA Plate moves with an approximatelynortheast direction with a relative velocity of �1.9 cm/yrfrom the MAT [DeMets et al., 1990].[5] The Cocos Plate varies in age roughly from west

to east at the same time that it subducts the MAT inincreasing rates from northwest (�4.8 cm/yr) to southeast(�7.5 cm/yr) [DeMets et al., 1990]. The Cocos Plate hasfractures zones, and the Tehuantepec Ridge (TR) convergestoward the MAT. As the TR converges to the MAT, it has agreat influence in the way the Cocos Plate subducts inthis region (Figure 1). Younger age crust subducts at ashallow dip angle north-northwest of the TR that contrastswith subducted crust with a steep angle and older crust ofabout 10 to 25 Myr southeast of the TR [Couch andWoodcock, 1981; Rebollar et al., 1999a].[6] West of the TR a stronger coupled plate interface

has been interpreted on the basis of the occurrence of large

thrust interplate earthquakes [Singh and Mortera, 1991;Pardo and Suarez, 1995]. The region east of the TR hasbeen interpreted as a seismic gap [Kelleher et al., 1973;McCann et al., 1979; Singh et al., 1981; McNally andMinster, 1981; Astiz et al., 1988]; however, the existence ofthis gap is controversial because in this area have occurredmedium sized earthquakes (R. Madariaga, personal com-munication, December 2002). Pardo and Suarez [1995]have observed a maximum downdip depth of �25 km thatis subhorizontal, west of the TR, while Ponce et al. [1992]inferred the Wadati-Benioff (W-B) zone with a dip in therange of 45� to 50� in the TR. On the other hand, Rebollaret al. [1999a] estimated a W-B zone �39 km thick dipping�40� southeast of the TR.[7] The continental crust in this area is characterized by

different ‘‘tectonostratigraphic’’ terrenes, which ideally cor-respond to geologic blocks with a coherent stratigraphicsequence and bounded by major tectonic subvertical dis-continuities [Campa and Coney, 1983; Sedlock et al., 1993].However, their location, boundaries, kinematics, and struc-ture at depth are still under controversy. Nevertheless, in theOaxaca region, gravity modeling shows that the crust is�40 km thick with a density consistent with continentalaffinity [Arzate et al., 1993].[8] Some paleostress analysis in conjunction with geo-

logical observations has been carried out in order to definethe continental stress regime in the continental portion of theIsthmus of Tehuantepec during the Mesozoic and Tertiary[Barrier et al., 1998; Guzman-Speziale and Meneses-Rocha,2000]. There it has been proposed a tensional regime insouthern Mexico for the last �6 Myr as a consequence ofthe deformation produced by the subduction of the CocosPlate and the eastward displacement of the western portion

Figure 1. Map of the tectonic features of the Gulf of Tehuantepec. Open triangles are the seismicstation, and dots are the location of the earthquakes bigger than 7. Numbers and their identification aredescribed in Table 1. MAT is the Middle America Trench, CA is the Caribbean Plate, MOFZ is theMotagua Polochic Fault Zone, NA is the North American Plate, and TR is the Tehuantepec Ridge. Heavyshaded area is the Tehuantepec Ridge, and light shaded area is the controversial area of the Tehuantepecseismic gap. Large arrow is the relative motion of the Cocos Plate.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

2 of 14

B04307

of the Caribbean Plate. This extensional regime has beenassociated with the development of both the Motagua-Polochic Fault Zone and the evolution of the passive marginof the Gulf of Mexico.

3. Historical Seismicity

[9] Table 1 shows coordinates of large events (greaterthan 7) that occurred during the last century between96�W and 92�W and reported by different authors.Figure 1 shows the location of the events. The proposedseismic gap in the Gulf of Tehuantepec between 95.3�Wand 94�W is also depicted. Indeed, the last earthquakegreater than 7.0 that occurred in this area was in 1941(see Table 1).

4. Data Recording

[10] We deployed a network of thirteen portable analogseismic stations in Oaxaca and Chiapas since 1999 as wellas a strong motion digital station. Figure 1 shows thelocation of the seismic stations. The deployment andoperation of this network has been logistically difficultbecause of the analog recording and the topographicfeatures of this region. However, we were able to maintainthe network in a good health status. The stations consist ofMQ800 Sprengnether analog smoke paper data recordersconnected to 1 s Mark velocity seismometers. An operatorsynchronized the internal clock of the MQ800 every day.However, sometimes the internal clock was not synchro-nized. In that case we calculated the correction of theinternal clock by doing a linear interpolation betweenthe last two times that the station was synchronized withthe WWVB ratio time signal. The velocity of the recordingdrum was 120 cm/mm. This kind of continuous analogrecording, even cumbersome, is ideal to record all arrivalsof P and SV waves of small to medium size magnitudeearthquakes. In order to have a good control in the locationof the activity in the Gulf of Tehuantepec, four stations

were deployed in a small aperture array. Maximum analogreading errors of P and S waves are of the order of 0.1 and0.2 s, respectively.

5. Event Locations and Vp//Vs Estimate

[11] Earthquakes were located using HYPO71 computercode [Lee and Lahr, 1972]. The crustal one-dimensional(1-D) P wave velocity model estimated by Valdez et al.[1986] was used to locate the earthquakes. We calculatedS wave velocities using a Vp/Vs ratio of 1.75 (laterestimated). We located earthquakes recorded at least in fourstations with 6 or more phases that included at least twoS waves picks. We started to locate earthquakes with trialdepths of 25, 50, 75, and 100 km. We compared HYPO71outputs and we selected events with root mean square errorof time residuals (RMS) of less than 1 s and with standarderror of the epicenter (ERH) and standard error of the focaldepth (ERZ) of less than 15 km. From the thousands ofearthquake recorded, 1897 fulfilled the aforementionedcriteria. Figure 2 shows the earthquake locations. It is wellknown that focal depths are well constrained when we havestations close to the epicenter. Therefore we selected eventswith nearly vertical source-station paths (80% of the eventsfulfill this criterion). We found a cluster of seismic activitynorth of the intersection of the TR and the MAT (square 1of Figure 2a). There are 103 events located in the square 1.On the other hand, in the adjacent square 2 in Figure 2aonly 21 events were located. Therefore we think that thereis a vacancy of activity in this area.[12] In order to calculate Vp/Vs ratios we proceeded in the

following manner. First, we plotted S-P times versus Pwaves arrival times and calculated origin times by a linearleast squares algorithm. We considered events with acorrelation coefficient of 0.95. Then, we plotted Ts-Tpversus Tp-T0 times, where Tp and Ts are the P and S traveltimes and T0 is the origin time. A line was fitted through theorigin up to 40 s that included 95% of our P and S wave

Table 1. Earthquakes Greater Than 7.0 That Occurred During the Twentieth Century Obtained From Different Authors Between

Parallels 96�W and 92�Wa

Event DateLatitudeNorth

LongitudeWest

Depth,km Magnitude Reference

1 23 Sept. 1902 16.58 92.58� 100 7.8 Figueroa [1976]2 14 Jan. 1903 16.70� 92.68� --- ---- Nunez-Cornu and Ponce [1989]2 14 Jan. 1903 15.00� 98.00� --- 8.3 Duda [1965]3 30 March 1914 16.76 92.15 100 8.1 Figueroa [1976]4 29 Dec. 1917 15.56� 94.91� --- ---- Nunez-Cornu and Ponce [1989]4 29 Dec. 1917 15.00� 97.00� --- 7.7 Duda [1965]5 10 Dec. 1925 15.50� 92.50� ---- 7.0 Miyamura [1976]6 22 March 1928 16.23� 95.45� --- 7.7 Singh et al. [1984]7 14 Dec. 1935 16.72� 93.08� --- 7.3 Figueroa [1976]8 11 Feb. 1941 15.20� 94.40� --- 7.0 Miyamura [1976]8 27 June 1941 16.25� 93.52� 200 6.5 Figueroa [1976]9 16 June 1943 14.60� 93.00� -- 7.0 Miyamura [1976]10 22 Dec. 1949 15.90� 93.00� 32 7.0 Miyamura [1976]10 22 Dec. 1949 16.40� 93.08� 100 6.5 Figueroa [1976]11 23 Aug. 1965 16.30� 95.80� 16 7.8 Singh et al. [1984]12 29 April 1970 14.45� 92.71� 44 7.3 Yamamoto and Mitchell [1988]13 10 Sept. 1993 14.41� 92.99� 29 7.3 CMT Harvard14 21 Oct. 1995 16.79� 93.65� 165 7.2 Rebollar et al. [1999b]aEarthquakes 3 and 7 are located in the seismic gap of the subduction zone of the Gulf of Tehuantepec. Earthquake 8 dates do not agree; however, we

included both the Miyamura and Figueroa dates.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

3 of 14

B04307

picks. An average Vp/Vs ratio of 1.75 was obtained (seeFigure 3). Following Riznichenko [1958] and Nicholson andSimpson [1985], we calculated average P wave velocities asa function of depth, and with the already calculated Vp/Vs

we estimated the S wave velocities as a function of depth.Figure 3 shows P and S wave velocities as a function ofdepth, where we can see a monotonically increase ofvelocities with depth (Figure 3).

Figure 2. Map of the well-located events. Triangles are the location of the seismic stations, and circlesare the events. White circles are shallow events, and black circles are deeper events. Lines show thecross-section profiles. Squares 1 and 2, described in the text, are areas with different rates of seismicactivity. (b) Plots of the cumulative number of events versus time and number of events versus magnitudeof earthquakes that occurred in squares 1 and 2. Top and middle panels indicate different rates of seismicactivity on those two zones. (c) Cross section of the aftershock activity of the Oaxaca earthquake. Star isthe hypocenter of the event, and it also shows the preferred fault plane.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

4 of 14

B04307

[13] Two large earthquakes were recorded by our net-work, the Puebla earthquake on 15 June 1999 Mw = 7.0[Yamamoto et al., 2002] and the Oaxaca earthquake on30 September 1999 Mw = 7.5. The Puebla earthquake is outof the area of influence of our network; however, theOaxaca earthquake lies inside our network, and we wereable to locate its aftershocks. A star on Figure 2 depicts thelocation of the main event, and a profile perpendicular to thetrench is also shown in Figure 2.

6. Moho Depth Estimate Using S-to-P (Sp)Converted Phases

[14] Inspection of the seismograms recorded in the seis-mic stations, and displayed in the square of Figure 4,showed clear converted Sp phases at the Moho. Figure 5ashows examples of the events with converted Sp phases.Sixteen events were selected. Figure 5b shows an exampleof its trajectories in a cross section from the source to theHUA station.[15] We calculated the Moho depth by minimizing

observed and calculated travel times in an iterative process.Figure 5b shows a vertical profile with the hypocenter ofthe earthquakes with Sp phases. Therefore, in order tocalculate the P and S travel times we considered a layer

Figure 3. Vp/Vs and average P and S wave velocities as a function of depth calculated with the Wadatiand Riznichenko techniques. Light grey lines are the P and S wave velocities as a function of depth.

Figure 4. Map showing the location of the events with Spconverted phase at the Moho discontinuity. Triangles are theseismic stations.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

5 of 14

B04307

Figure 5. (a) Examples of seismograms of events that occurred in the subducted plate that show clearconverted Sp phase at the Moho. Sixteen events were selected. (b) Cross section of the selected eventswith Sp phases recorded at HUA station as well as the parameters used in the estimate of the Moho depth.Number is their identification.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

6 of 14

B04307

(with thickness H1) and a half-space. First, we appliedSnell’s law at the crust-mantle boundary of the incidentand transmitted rays, so we have

V2D1

D21 þ H2

1

� �1=2 � D� D1ð ÞV1

D� D1ð Þ2þ H � H1ð Þ2h i1=2 ¼ 0; ð1Þ

where V1 is the average velocity of the crust (P or S), V2 isthe average velocity of the upper mantle (P or S), H1 isthe thickness of the crust, H is the focal depth, D1 is thehorizontal distance between the seismic station and thevertical projection to the surface of the transmitted P-to-Sphase in the Moho, and D is the epicentral distance (seeFigure 5b). Numerically solving equation (1), we obtainedD1. Then we calculated P and S wave travel time’s tp and tsin each of the layers (crust and mantle). From the Wadatiand Riznichenko diagram (Figure 3) we calculated a Vp1 =6.2 km/s and a Vp2 = 8.3 km/s, and from the Vp1/Vs1 = 1.76and Vp2/Vs2 = 1.73 we calculated S wave velocities. Traveltimes were calculated by varying the Moho depths from 5 to60 km. Once travel times were calculated, we computed

RMS

¼ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffitp � ts� �

o� tp � ts� �

c

h i2þ tp � tsp

� �o� tp � tsp� �

c

h i2þ tsp � ts

� �o� tsp � ts� �

c

h i2r

residuals for different Moho depths. Where (tp-ts)o is the Pminus S observed travel time and (tp-ts)c is the P minus Scalculated travel time and so on. Figure 6 shows thevariations of the RMS residuals for different Moho depths.From the variations of the RMS residuals we estimated aMoho depth of the order of 28.5 ± 3.5 km.

7. Aftershocks of the 30 September 1999 Event

[16] We recorded the 30 September 1999 Oaxaca eventand its aftershocks up to June 2000, when we recorded amagnitude 4.6 event that occurred at a depth of 50 km.

Source mechanisms of the main event and five aftershocksindicate that the intraplate events were normal fault with thestrike parallel to the trench and dipping to the north-northeast [Singh et al., 2000; Hernandez et al., 2001].Hernandez et al. [2001] found that the slip was mainlyreleased at a depth of 45 km inside the subducted CocosPlate. The depths of the hypocenters range from 12 to50 km. Figure 2c shows a cross section perpendicular tothe Middle America Trench where we show the location ofthe main event. From Figure 2c it can be seen that theaftershocks occurred updip of the main event. The preferredfault plane, taken from Hernandez et al. [2001], is depictedin Figure 2c as well as the Cocos-North America interface.

8. Shape of the Wadati-Benioff Zone and StressDistribution of the Subducted Lithosphere

[17] In order to investigate the shape of the Wadati-Benioff zone we constructed a mesh of 12 by 13 circleswith a diameter of 30 km that cover the whole area of study(Figure 7). Within each circle we choose events from ourcatalog with standard error of the focal depth (ERZ) of lessthan 10 km. We calculated the mode [Mood and Graybill,1963, p. 107] of the focal depth with the selected events oneach circle. With this criterion we were able to find themode of 83 points (black circles in Figure 7). White circlesin Figure 7 are areas were it was not possible to define themode. Figure 8 shows vertical cross sections perpendicular(A–L) and parallel (1–13) to the MAT. Open circles are theevents, and open squares are the mode focal depths thatdefine the W-B zone. From the cross sections perpendicularto the trench we can see that the dip of the W-B zone gentlyincreases from east to west. Cross sections E-H are the bestdefined and suggest that the subducted lithosphere increasesits dip and bends up at about a depth of 100 km. Farthersoutheast the subducted slab unbends and increases its dip.A 2-D contour surface of the W-B zone was calculated

Figure 6. Plot of the RMS time residuals versus Mohodepth. The estimated average thickness of the Moho is28.5 ± 3.5 km.

Figure 7. Map showing the mesh used to project theseismicity in cross sections perpendicular and parallel to theMiddle America Trench. Each circle has a diameter of30 km. Black circles contain seismic activity, and whitecircles do not contain seismic activity.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

7 of 14

B04307

using Surfer 8 [Golden Software, Inc., 2002] code that usesthe Kriging gridding method. Figure 9 shows contour linesof equal depth of the subducted Cocos Plate at 10 kmintervals. Contour labels were placed on contour lines of 50,100, and 150 km. The contour lines of 50 and 100 kmconverge west of the Isthmus of Tehuantepec and corre-spond to the E, F, and G cross sections. It can be observedthat cross section F has shallow and low-magnitude (M < 4)seismic activity in the continental crust. Ponce et al. [1992]

also reported this pattern of shallow seismic activity. Wethink that this nest of seismic activity could be linked to thebend of the MAT to the east. From Figure 9 it can be seenthat there are significant changes in the topographic featuresin the crust in the contortion of the W-B zone. The MAT liesto the west of this area, and a wide continental shelf lies tothe east. The hachured area is the coupled zone between thesubducted Cocos Plate and the overriding North AmericanPlate defined up to a depth of 40 km. However, from the

Figure 8. Vertical cross section of seismicity projected perpendicular (A–L) and parallel (1–13) fromthe Middle America trench. Numbers in the top of the plots are centered in the circles of Figure 7. Opencircles are the events, and open squares are the mode. Description of the seismicity is given in the text.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

8 of 14

B04307

analysis of the seismicity it can be seen that the coupledzone is not uniform in this region.[18] Cross sections (30 km wide) parallel and oblique to

the trench, following the coast, were plotted in order to seeroughly the coupled zone of the Cocos Plate and theoverriding North American Plate from east to west alongthe Gulf of Tehuantepec (see Figure 10). Profile ABCC’shows the coupled zone and depth extent of the seismicactivity. It also shows a pocket of activity in the neighbor-hood of point C. This area coincides with topographicfeatures and the closeness of the lines of 50 and 100 kmcontour lines of the W-B zone. Roughly, half the waybetween C and C’, there is a small group of events. In thiszone, Nunez-Cornu and Ponce [1989] proposed the locationof the 1917 magnitude 7.7 event (Table 1). Pardo andSuarez [1995] also reported thrust events in this area thatcoincides with the projection of the Tehuantepec Ridgeunder the North American Plate.[19] Profile ABDD’ shows features similar to the former

profile; however, in the neighborhood of point D theactivity migrates to depths of the order of 50 km. In thiszone a normal event of magnitude 6.7 occurred at a depth of60 km [Gonzalez-Ruiz, 1986]. Figure 10 shows the locationof the event as well as the strike and dip of one fault planeof the reported fault plane solution. The gently increase ofthe activity with depth suggests that the Cocos Plate startedto bend down dip because we moved away from the MAT.Profile ABEE0E00 follows the coastline of the Isthmus ofTehuantepec. Near point E we can see a shallow pocket of

activity with events with magnitudes smaller than 4. Ponceet al. [1992] reported this zone of shallow seismic activity.Also near this point the W-B zone starts to bend down dip,and the activity along the coast follows the W-B contourline of 100 km.[20] Now, it is possible to have access to moment tensor

solutions (MTS) from the U.S. Geological Survey (USGS)Web page (http://eqint.cr.usgs.gov/neic/), which includesHarvard and preliminary determination of epicenters(PDE) moment tensor solutions. We went though thatcatalog and searched for MTS of events located in theIsthmus of Tehuantepec. We found 123 MTS with magni-tudes greater than 5.0 since 1977. In our analysis weincluded the 30 September 1999 (Mw = 7.5) Oaxacaearthquake and five of its aftershocks [Singh et al., 2000].Fault plane solutions reported by Rebollar et al. [1999a] inChiapas and the 30 January 2002 (Mw = 5.9) Veracruzearthquake was also included. We applied the Gephart andForsyth [1984] and Gephart [1990] inversion code in orderto invert for the best fit P and T axes along the downgoingsubducted plate.[21] As a mean of visualizing variations of P and T stress

vectors along the subducted lithosphere, we plotted crosssections A, C, E, and G described in Figure 7. All crosssections started at the trench. Figure 11 shows cross sectionsthat included T axes projections along the strike of theprofiles, and Figure 12 shows plots of P and T axes on anequal-area lower projection. Profile A (Figure 11a) showsprojections of T axes before the MAT started to bend to theeast. Near the trench along the coupling zone, T axes ofinterplate earthquakes are oblique to the interface (P axesnormal to the interface of the overriding and subductedslab), while T axes of intraplate events lie along the dipdirection of the subducted plate, which means that belowthe coupling zone, the slab is mainly under tension (slab-pull events). At depths greater than 50 km, T axes aremainly horizontal, indicating that the slab is rupturing alongthe dip direction or with steeply dipping angles. Best fitstresses (s1 > s2 > s3) obtained from the inversion of P andT axes along this profile are (plunge, azimuth) s1(50�, 220�)and s3(40�, 37�) (Figure 12a). Profile C (Figure 11b)included events from the area where the trench starts tobend from northwest to east. Again we observed the samefeatures as before for interplate events; however, there arefew intraplate slab-pull events with T axes along the dipdirection. There are less T axes at depths greater than100 km; however, they indicate that the slab is undertension. Best fit stresses are s1(44�, 193�) and s3(28�, 72�)(see Figure 12b). Along E profile (Figure 11c) the MATstarted to follow its original strike toward the southeastwhere the continental shelf starts to widen. T axes alongthe coupling zone are no longer oblique, and T and P axesorientations are complex along the coupling zone; however,intraplate events indicate that the slab is under tension. Atdepths greater than 100 km the subducted plate is dominatedby slab-pull events. Best fit stresses are s1(62�, 165�) ands3(11�, 54�) (Figure 12c). Finally, G profile (Figure 11d) hadmore MTS that the previous profiles, and the dip of the W-Bincreases to �40�. The strikes of the T axes of interplateevents are oblique to the coupling zone with a few excep-tions. There are intraplate events that indicate that thesubducted plate is under tension. T axes of deeper events

Figure 9. Map of a 2-D shape of the Wadati-Benioff zonein the Gulf of Tehuantepec. Continuous lines are contourlines of equal depth. Contour depth lines of 50, 100, and150 km are shown as heavy black solid lines. It can be seenthat contour depth lines of 50 and 100 km converge west ofthe Isthmus of Tehuantepec. Light gray shaded area is thecoupling zone between the subducted Cocos and overridingNorth American Plates. White lines are the politicalboundaries of Mexico. White lines marked with 50, 100,and 150 km are the contour depth lines of the W-B zonesuggested by Ponce et al. [1992].

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

9 of 14

B04307

Figure 10. (top) Map showing lines parallel and oblique to the trench that we chose to project in crosssections the seismic activity (light white lines). Dashed line shows the location of the MAT. (bottom)Projection of the seismic activity in cross sections parallel and oblique to the MAT. White circles are theshallow events, and black circles are deeper events. Star shows the location of the event of magnitude 6.7reported by Gonzalez-Ruiz [1986] and its possible fault plane.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

10 of 14

B04307

Figure 11.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

11 of 14

B04307

indicate that the subducted slab in under tension; however,there is one event that indicate that the subducted slab inunder compression (slab-push event). The best fit stresses ofthis profile are s1(63�, 200�) and s3(27�, 35�) (Figure 12d).

9. Summary and Conclusions

[22] We have analyzed the seismic activity recorded in anetwork of analog seismic stations deployed in Oaxaca andChiapas that encompass the Isthmus of Tehuantepec. Weused the USGS catalog of moment tensor solution to have

an insight of the stress orientations within of the subductedplate in this area, and we used 1897 events to infer thegeometry of the W-B zone. To obtain principal stressorientations (P and T axes), we used MTS compiled bythe USGS and fault plane solution reported by Rebollar etal. [1999a] and Singh et al. [2000]. Duration magnitudes ofthe events recorded in our network varied from 1.8 to 4.2.We calculated a Vp/Vs of 1.75 that within the errorsrepresents a Poisson solid. Within the time window ofobservation we recorded the Oaxaca normal fault earth-quake of 30 September 1999 (Mw = 7.5) and its aftershocks.

Figure 11. Projections of T axes in cross sections (depth versus distance) from northeast to southeast that were obtainedfrom the USGS catalog, Rebollar et al. [1999a], and Singh et al. [2000]. (a) T axes of the cross section of lines A and B ofFigure 7. (b) T axes of the cross section of lines C and D of Figure 7. (c) Cross section of lines E and F of Figure 7.(d) Cross section of lines F and G of Figure 7. Arrows are the projection of the T axes along the cross section. Circledarrows at the right are the best fit T axes of stresses obtained with the Gephart and Forsyth [1984] and Gephart [1990]inversion code. Upper left and right coordinates are the points where cross sections were projected. Plotted with the Louvariand Kiratzi [1997] program.

Figure 12. A lower hemisphere Schmidt stereonet showing P (solid circle) and T (open circle) axesfrom the cross sections of Figure 11. Large circles are the best fit stress orientation obtained with theinversion process. Plotted with the Louvari and Kiratzi [1997] program.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

12 of 14

B04307

The aftershocks were located at depths that varied from 20to 50 km, and from its locations we inferred a fault area witha length of 80 km along the strike of the trench and 30 kmwide along the dip direction of the subducted plate.[23] We plotted seismic cross sections from northwest to

southeast perpendicular from the MAT (profiles A–L) andparallel (profiles 1–13) from the MAT in order to see thegeometry of the W-B zone. Profiles A and B (Figure 8)show that there is a vacancy of events at depth at about50 km that extends between about 60 and 120 km from thetrench. There is also a vacancy of seismic activity, observedin profiles C and D, in the depth range from 100 km up to180 km and about 300 km from the trench. LeFevre andMcNally [1985] also observed this vacancy of seismicity.Profiles E, F, G, and H do not show any gap of events alongthe downgoing slab. However, these profiles suggest thatthe subducted slab becomes buoyant and flattens up. Pro-files I and J show that the subducted slab loses its buoyancyand the dip of the W-B zone started to increase. From thesecross sections it is evident that the lateral variations in thedip angle of the subducted slab are from northwest tosoutheast.[24] Figure 10 shows profiles 30 km wide parallel and

oblique to the MAT. Those profiles clearly show theseismogenic zone of the subducted slab along the strike ofthe MAT. The activity is mainly concentrated up to depthsof 50 km along the strike; however, depths of the eventsstarted to increase between points C and D (see Figure 10)since the lithosphere started to bend down and because weare not able to locate events in this coupling zone of thesubducted and overriding plate. In this area of the subductedplate, Nunez-Cornu and Ponce [1989] proposed the locationof the 1917 magnitude 7.7 earthquake and Gonzalez-Ruiz[1986] located an earthquake of magnitude 6.7. As we gofarther southeast, the depths of the events seems to increaseto about 100 km because we can not locate events below thecontinental shelf with enough precision. The inversion ofthe P and T axes, using the Gephart and Forsyth [1984] andGephart [1990] technique, indicate that the best fit T axes(s3(plunge, azimuth)) varied from northwest to southeastlike s3(40�, 37�) to s3(28�, 72�), to s3(11�, 54�) and finallyto s3(27�, 35�).[25] The observed stress pattern in this study has been

inferred in Oaxaca with the intraplate normal fault 1931,magnitude 7.8 earthquake, the thrust fault interplate 1978,magnitude 7.6 earthquake and again the normal faultintraplate 1999, magnitude 7.5 earthquake [Singh et al.,1985, 2000]. Farther northwest in the geographic limits ofMichoacan and Guerrero, the 1985 interplate thrust fault,magnitude 8.1, earthquake and the 1997 intraplate normalfault earthquake follows the same pattern. Mikumo et al.[1999] suggested that the 1985 thrust fault earthquaketransferred enough stresses inside the subducted plate sothat it triggered the normal 1997 intraplate earthquake. Weobserved the same pattern in our area of study; compres-sional events (thrust faulting) in the interplate coupled zoneof the subducted and overriding plates and intraplate slab-pull (extensional) events below coupled zone. LeFevre andMcNally [1985] found out the same stress distribution onthe subducted slab in the Gulf of Tehuantepec. Lemoineet al. [2002] studying intermediate depth slab-pull andslab-push earthquakes in Mexico, Peru, and north central

Chile suggested that the slab-push interplate events andslab-pull intraplate events can be explained as due toflexure of the downgoing slab.[26] In conclusion, we were able to infer the geometry of

the W-B zone. We also clearly observed lateral variations inthe dip of the W-B zone from northwest to southeast. It wasobserved a shallow dip below Oaxaca that smoothlyincreases toward Chiapas to an angle of 40�. However,during this transition we observed a contortion of thesubducted slab, clearly seen in the depth contour lines of50 and 100 km northwest of the intersection of the Tehuan-tepec Ridge and the MAT. This zone coincides with theregion where Sierra Madre Occidental ends and starts theChiapas continental shelf. Using travel times of converted Sto P (Sp) phases at the Moho, we were able to calculate thethickness of the continental crust. We estimated a thicknessof 28.5 ± 3.5 km in the Gulf of Tehuantepec. T and P axesshow that the interplate interface is under compression andthat the plate below the coupled zone is under tension. AsLemoine et al. [2002] pointed out, slab-push intraplateevents are rare; however, one slab-pull intraplate event isobserved in cross section of Figure 11b, and Rebollar et al.[1999a] reported one slab-push intraplate event at the end ofthe subducted slab of Figure 11d.

[27] Acknowledgments. The authors thank Maria E. Perez andPrimitivo Lopez, who participated in the picks of the arrival times. AntonioMendoza helped us with the figures. Raul Madariaga, Denis Hatzfeld, RaulCastro, and the Associate Editor provided helpful suggestions to improvethe manuscript. We greatly appreciate the Comision Federal de Electricidad(CFE) for letting us use the data recorded in the Oaxaca and Chiapasnetwork. One of us, C.J.R., was partially supported by CONACYT (Project37038-T).

ReferencesArzate, J. A., M. Mareschal, and J. Urrutia-Fucugauchi (1993), A prelimin-ary crustal model of the Oaxaca continental margin and subduction zonefrom magnetotelluric and gravity measurements, Geofis. Int., 32, 441–452.

Astiz, L., T. Lay, and H. Kanamori (1988), Large intermediate-depth earth-quakes and the subduction process, Phys. Earth Planet. Inter., 53, 80–166.

Barrier, E., L. Velasquillo, M. Chavez, and R. Gaulon (1998), Neotectonicevolution of the Isthmus of Tehuantepec (southern Mexico), Tectono-physics, 287, 77–96.

Campa, M. F., and P. J. Coney (1983), Tectonostratigraphic terranes andmineral resources distribution in Mexico, Can. J. Earth Sci., 20, 1040–1051.

Couch, R., and S. Woodcock (1981), Gravity and structure of the con-tinental margins of southwestern Mexico and northwestern Guatemala,J. Geophys. Res., 86, 1829–1840.

DeMets, C., R. G. Gordon, D. F. Argus, and S. Stein (1990), Current platemotions, Geophys. J. Int., 101, 425–478.

Duda, S. J. (1965), Secular seismic energy release in circum-pacific belt,Tectonophysics, 2, 409–452.

Figueroa, J. (1976), Sismicidad de Chiapas, technical report, 21 pp., Fac. deIng., Univ. Nac. Auto. Mex., Mexico City.

Gephart, J. W. (1990), FMSI: A FORTRAN program for inverting fault/slick-inside and earthquake focal mechanism data to obtain the regionalstress tensor, Comput. Geosci., 16, 953–989.

Gephart, J. W., and D. W. Forsyth (1984), An improved method for deter-mining the regional stress tensor using earthquake focal mechanism data:Applications to the San Fernando earthquake sequence, J. Geophys. Res.,89, 9305–9320.

Golden Software, Inc. (2002), Surfer 8, Contouring and 3D surface map-ping for scientists and engineers.

Gonzalez-Ruiz, J. R. (1986), Earthquake source mechanics and tectono-physics of the Middle-America subduction zone in Mexico, Ph.D. thesis,Univ. of Calif., Santa Cruz.

Guzman-Speziale, M., and J. J. Meneses-Rocha (2000), The North Amer-ica-Caribbean plate boundary west of the Motagua-Polochic fault system:A fault jog in southeastern Mexico, J. Am. Earth Sci., 13, 459–468.

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

13 of 14

B04307

Hernandez, B., N. M. Shapiro, S. K. Singh, F. J. Pacheco, F. Cotton,M. Campillo, A. Iglesias, V. Cruz, J. M. Gomez, and L. Alcantara(2001), Rupture history of September 30, 1999 intraplate earthquakeof Oaxaca, Mexico (Mw = 7.5) from inversion of strong-motion data,Geophys. Res. Lett., 28, 363–366.

Kelleher, J., L. Sykes, and J. Oliver (1973), Possible criteria for predictingearthquake locations and their application to major plate boundaries inthe Pacific and the Caribbean, J. Geophys. Res., 78, 2547–2585.

Lee, W. H. K., and J. C. Lahr (1972), HYPO71: A computer program fordetermining hypocenter, magnitude, and first motion pattern of localearthquakes, U.S. Geol. Surv. Open File Rep. 73-311.

LeFevre, L. V., and K. C. McNally (1985), Stress distribution and subduc-tion of aseismic ridges in the Middle America subduction zone, J. Geo-phys. Res., 90, 4495–4510.

Lemoine, A., R. Madariaga, and J. Campos (2002), Slab-pull and slab-pushearthquakes in the Mexican, Chilean and Peruvian subduction zones,Phys. Earth Planet. Inter., 132, 157–175.

Louvari, E., and A. Kiratzi (1997), Rake: A windows program to plotearthquake focal mechanisms and the orientation of principal stresses,Comput. Geosci., 23, 851–857.

McCann, W., S. Nishenko, L. Sykes, and J. Krause (1979), Seismic gap andtectonics: Seismic potential for major boundaries, Pure Appl. Geophys.,117, 1082–1147.

McNally, K. C., and B. Minster (1981), Nonuniform seismic slip ratesalong the Middle America trench, J. Geophys. Res., 86, 4949–4959.

Mikumo, T., S. K. Singh, and M. A. Santoyo (1999), A possible stressinteraction between large and normal-faulting earthquakes in the Mexicansubduction zone, Bull. Sesimol. Soc. Am., 89, 1418–1427.

Miyamura, S. (1976), Provisional magnitudes of Middle American earth-quakes not listed in the magnitude cataloge of Gutenberg-Richter, Bull.Int. Inst. Seismol. Earthquake Eng. Tokyo Univ., 14, 41–46.

Mood, A. M., and F. A. Graybill (1963), Introduction to the Theory ofStatistics, McGraw-Hill, New York.

Nicholson, C., and D. W. Simpson (1985), Changes in Vp/Vs with depth:Implications for appropriate velocity models, improved earthquake loca-tions, and material properties of the upper crust, Bull. Seismol. Soc. Am.,75, 1105–1123.

Nunez-Cornu, F., and L. Ponce (1989), Zonas sismicas de Oaxaca Mexico:Sismos maximos y tiempos de recurrencia para el periodo 1542–1988,Geofis. Int., 28, 587–641.

Pardo, M., and G. Suarez (1995), Shape of the subducted Rivera and Cocosplates in southern Mexico: Seismic and tectonic implications, J. Geophys.Res., 100, 12,357–12,373.

Ponce, P., R. Gaulon, G. Suarez, and E. Lomas (1992), Geometry and stateof stress of the downgoing Cocos plate in the Isthmus of Tehuantepec,Mexico, Geophys. Res. Lett., 19, 773–776.

Rebollar, C. J., V. H. Espindola, A. Uribe, A. Mendoza, and A. Perez-Vertti(1999a), Distribution of stresses and geometry of the Wadati-Benioffzone under Chiapas, Mexico, Geofis. Int., 38, 95–106.

Rebollar, C. J., L. Quintanar, J. Yamamoto, and A. Uribe (1999b), Sourceprocess of the Chiapas, Mexico, intermediate-depth earthquake (Mw = 7.2)of 21 October 1995, Bull. Seismol. Soc. Am., 89, 348–358.

Riznichenko, Y. V. (1958), Methods for large-scale determination of focuscoordinates of nearby earthquakes and velocities of seismic waves in thefocal region, Tr. Inst. Fiz. Zemli. Akad. Nauk SSSR, 4.

Sedlock, R. L., F. Ortega, and R. C. Speed (1993), Tectonostratigraphicterranes and tectonic evolution of Mexico, Spec. Pap. Geol. Soc. Am.,278, 153 pp.

Singh, S. K., and F. Mortera (1991), Source time functions of largeMexican subduction earthquakes, Morphology of the Benioff zone,age of the plate and their tectonic implications, J. Geophys. Res.,96, 21,487–21,502.

Singh, S. K., L. Astiz, and J. Havzkov (1981), Seismic gaps and recurrenceperiods of large earthquakes along the Mexican subduction zone: Areexamination, Bull. Seismol. Soc. Am., 71, 827–843.

Singh, S. K., M. Rodrighez, and J. M. Espindola (1984), A catalog ofshallow earthquakes of Mexico from 1990–1981, Bull. Seismol. Soc.Am., 74, 267–279.

Singh, S. K., G. Suarez, and T. Dominguez (1985), The Oaxaca, Mexico,earthquake of 1931: Lithospheric normal faulting in the subducted Cocosplate, Nature, 317, 56–58.

Singh, S. K., et al. (2000), The Oaxaca earthquake of 30 September 1999(Mw = 7.5), A normal-faulting event in the subducted Cocos plate,Seismol. Res. Lett., 71, 67–78.

Valdez, C. M., W. D. Money, S. K. Singh, R. P. Meyer, C. Lomnitz, J. H.Luetgert, C. E. Helsley, B. T. R. Lewis, and M. Mena (1986), Crustalstructure of Oaxaca, Mexico, from seismic refraction measurements, Bull.Seismol. Soc. Am., 76, 547–563.

Yamamoto, J., and B. J. Mitchell (1988), Rupture mechanics of complexearthquakes in southern Mexico, Tectonophysics, 154, 25–40.

Yamamoto, J., L. Quintanar, C. J. Rebollar, and Z. Jimenez (2002), Sourcecharacteristics and propagation effects of the Puebla, Mexico earthquakeof June 15, 1999, Bull. Seismol. Soc. Am., 92, 2126–2138.

�����������������������H. Bravo, O. Jimenez, and A. Uribe, Gerencia de Estudios de Ingenieria

Civil, Departamento de Sismotectonica, Augusto Rodin 263, Col. NocheBuena Mexico D. F. 03720, Mexico.C. J. Rebollar, Centro de Investigacion Cientifica y Educacion Superior

de Ensenada (CICESE), km 107 carretera Tijuana-Ensenada. Ensenada,Baja California 22860, Mexico. (rebollar@cicese.mx)

B04307 BRAVO ET AL.: WADATI-BENIOFF ZONE IN THE GULF OF TEHUANTEPEC

14 of 14

B04307