Plate Motions in the North Andean Region Jeffrey T

Home , Caribbean Plate, South American Plate

University of South Carolina Scholar Commons

Faculty Publications Earth, Ocean and Environment, School of the

12-10-1993 Plate Motions in the North Andean Region Jeffrey T. Freymueller

James N. Kellogg University of South Carolina - Columbia, [email protected]

Victor Vega

Follow this and additional works at: https://scholarcommons.sc.edu/geol_facpub Part of the Earth Sciences Commons

Publication Info Published in Journal of Geophysical Research, Volume 98, Issue B12, 1993, pages 21,853-21,863. Freymueller, J. T., Kellogg, J. N., & Vega, V. (1993). Plate motions in the North Andean region. Journal of Geophysical Research, 98 (B12), 21,853-21,863. ©Journal of Geophysical Research 1993, American Geophysical Union

This Article is brought to you by the Earth, Ocean and Environment, School of the at Scholar Commons. It has been accepted for inclusion in Faculty Publications by an authorized administrator of Scholar Commons. For more information, please contact [email protected]. JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 98, NO. B12, PAGES 21,853-21,863, DECEMBER 10, 1993

Plate Motionsin the North AndeanRegion

JEFFREYT. FRE•J•-:• ,1 JAMESN. Km•Oc_,G,At, O VICTOR VEGA

Departmentof GeologicalSciences, University of SouthCarolina, Columbia

Repeated geodetic measurementswith the Global Positioning System (GPS) provide direct measurementsof displacementsdue to plate motionsand active crustal deformation in Central Americaand northern South America, an area of complex interaction of the Nazca, Cocos, Caribbean and South Americanplates. The displacementrates for the period 1988-1991, obtainedfrom the resultsof the first threeCentral AndSouth America (CASA) GPScampaigns, are in generalagreement with the predictionsof the NUVEL-1 plate motion model, but there are differencesin detail betweenthe observationsand the model. The Nazca-NorthAndes convergence rate vector measuredby GPS is differentfrom the NUVF.•I vector at 95% confidence.The differenceimplies that the North Andesare moving northwardrelative to SouthAmerica. The measured convergence between the Caribbeanplate andthe North Andessuggests that the southernmargin of the Caribbeanplate is locatedin the SouthCaribbean deformed belt. The April 1991 CostaRica earthquakeand the Cocos-Caribbeanconvergence rate determinedby GPS suggestthe possibility of significant ongoing deformationbetween Central Americaand the stableinterior of the Caribbeanplate. Our GPS resultsare consistent with deformationof the overriding plates at the convergentmargins of Central andSouth America and confirmthat activeconvergence is occurringaround muchof the southernmargin of the Caribbeanplate, from Colombia west to Costa Rica. Costa Rica and Panam• are not part of the stable Caribbeanplate. Instead,the SouthCaribbean deformed belt andthe North Panam•fold belt probablyrepresent the southernmargin of the Caribbeanplate.

INTRODUCTION The rate of change of baseline vectors, three-dimensional vectors between two sites, measurethe rates of motion of the The CASA (Central and South America)Global Positioning major plates, as well aslocal or regional deformationat each System(GPS)project was inauguratedin 1988 to studyplate site. The baseline rate of change vectors are decomposedinto motions and crustal deformation in Central America and components which represent the motion rates in the local northern SouthAmerica, a tectonically active areaof complex north, east and vertical directions. Details of the GPS data interaction between the Nazca, Cocos, Caribbean and South analysis and the precision and accuracyof the GPS solutions American plates. There is general agreementthat the three are given in the appendix. A single baseline is insufficient to plates are converging, but the locations of certain plate computea pole of relative motion, so the observedbaseline boundarieshave long been a subject of controversy [e.g., ratesof changeare compared to modelrates computed from the Dewey, 1972; Shagam, 1975; Aggarwal, 1983]. Diffuse NUVEL-1 plate motion model [DeMets et al., 1990; Argusand seismicity is spreadout over a wide area from the Cocos Gordon, 1991]. Further studies will compute displacement transform fault south of Panamfi to the East Andean and rates for all the sites of the CASA network, to study the Bocon6 fault systems at the eastern edge of the Andes internal deformationof the plates. [Pennington, 1981]. Kellogg et al. [1985] have interpretedthe tectonicsof the region in termsof two additionalmicroplates, REGIONAL •NICS Panatariand the North Andes, both of which are subject to internal deformation(Figure 1). Colombia-EcuadorSubduction System Repeated geodetic measurementswith GPS provide direct Subduction(70 mm/yr) of the oceanic Nazca plate occursat measurementsof the displacementsof the individualGPS sites, the western margin of South America, and the trench is which can be used to compute rates of plate motion and seismicallyactive (Table 1). Four large (Ms > 7) shallow intraplate deformation.The presentCASAnetwork consistsof subduction-relatedearthquakes have occurredwithin this area 30 stations in five countries (Figure 1), most of which have in the last century.The 1906 event rupturedmore than 400 km been occupied in at least two of the three CASA GPS of the trench, from the equatorto the north [Kelleher, 1972] campaigns. Five sites (Isla Baltra, Isla del Coco, Liberia, (Figure 2). Three smaller events, in 1942, 1958, and 1979 Lim6n and Jerusalen)positioned to measurethe velocities of have together rupturedapproximately the same length of the the major plates in the region have been occupiedin all three trench as did the 1906 event (Figure 2), althoughthe seismic campaigns. This paper concentrateson the plate motion momentof the 1906 event is approximately5 times as large as determinationsmade using thesefive sites, and two sites (Isla the sum of the moments of the smaller events [Kanamori and San Andrts and Cartagena) which measurethe convergence McNally, 1982]. Based on the seismic record of the last between the Caribbean plate and the North Andes. The century, Nishenko [1989] estimated the probability of a measurementscover a 3-year period, 1988-1991. magnitude7+ earthquakein these segmentsof the subduction zone within the next 10 years to be greater than 70%. A 1Nowat Departmentof Geophysics,Stanford University, magnitude M ~ 6.6 earthquakeoccurred on November 1991, Stanford, California. roughly 100 km to the north of the northern extent of the Copyright1993 by the AmericanGeophysical Union. 1906 rupture(Figure 2). The preinstrumentalrecord of seismicity in this region is Paper number 93JB00520. incomplete, due to the historically small population on the 0148-0227/93/93 JB-005 20505.00 Pacific coast.Only one eventprior to 1906 can be attributedto

21,853 21,854 FREYMUF•LERET AL.: PLATEMOTIONS IN NORTHANDEAN REGION

l CARIBBEAN Isla San Andr•s 10mm/yr •

10 ø N cocos

I mm/yr

¾ I / • •s•ade• Coco I I

NAZCA

o • IslaBaltra 69 90øW 80øW 70øW

Fig. 1. ObservedGPS (solid lines) andmodel NUVEL-1 [DeMets et al., 1990] (dashedlines) baselinerates of change,with their 95% confidenceellipses. Baselinerates of changemeasure the relative motions of two GPS sites. The Caribbean-NorthAndes platemotion is basedon only two epochsof GPSdata(1988and 1991), andthe uncertaintygiven is conservative.GPS sites used in this studyare indicatedby squares;other GPS sitesare indicatedby circles. this segment of the trench, an earthquakenear Tumaco late Cenozoic [Bowin, 1976; Mann and Burke, 1984; Pindell (Colombia) in 1798, which was large enough to produce et al., 1988]. notable damage300 km away in Quito, Ecuador[Comit6 del On the east side of the North Andes, right-lateral Afio Geoffsico Internacional del Ecuador,1959]. The damage transpressivemovement is taking place on the Bocon6 and causedin Quito was similar to the damagecaused by the 1979 East Andean fault zones. The Bocon6 fault in Venezuela is Tumaco earthquake,but no other information is available to principally a right-lateral strike slip fault [Schubert,1982] determinethe magnitudeof this event. whichruns throughthe M6rida Andes.Subparallel thrust faults flank the Bocon6 fault, and small-scalegeodetic networks in Motion and Deformation of the North Andes the M6rida Andesindicate significant compressionnormal to the Bocon6 fault in additionto strike-slip motion [Henneberg, The North Andes are boundedby the Colombia-Ecuador 1983]. The East Andean frontal fault system consists of trench and Panamfi on the west, the South Caribbean deformed subparallelwestward-dipping faults. Based on a studyof focal belt to the north, and the Bocon6 and East Andean fault zones mechanisms in the East Andean frontal fault system, to the east(Figures 1 and2) [Bowin, 1976; Pennington,1981; Pennington [1981] proposedthat transpressiveright-lateral Kellogg et al., 1985; Adamek et al., 1988]. Within the slip is occurring on these faults, and that the North Andes northern part of the North Andes, significant Cenozoic block is movingNNE relative to the SouthAmerican plate. displacementshave occurredon the Oca (right-lateral, east- To the south the Ecuadorian Andes include a system of west trending) and Santa Marta-Bucaramanga(left-lateral, subparallel northeast trending right lateral strike slip faults northwest-southeast trending) faults (Figure 2) [e.g., and north trending thrust faults. The relationships between Campbell, 1968; Kellogg, 1984; Mann and Burke, 1984; thesefaults arenot well understood;perhaps the best studiedof Mann et al., 1990]. There areno estimatesof the slip ratesof these faults is the right-lateral strike-slip Pallatanga fault either of these two faults, but both may be active. The Santa (Figure2) [Winter andLaVenu,1988]. The focal mechanismof Marta-Bucaramangafault plays a major role in the vectorblock the magnitudeMS ~ 6.9 earthquakein March 1987 east of model of Dewey andPindell [1985, 1986]. The Bocon6 fault, Quito indicatespure thrust faulting on a north-southtrending SantaMarta-Bucaramanga fault, andSouth Caribbeandeformed fault, while focal mechanisms of two smaller events about 100 belt define the wedge-shaped"Maracaibo block" which has km directly to the south show oblique (transpressional) moved northward relative to stable South America during the displacement(Table 1 andFigure 2) [Lyon-Caenet al., 1990]. FREYMUFI..LERET AL.: PLATEMOTIONS IN NORTHANDEAN REGION 21,855

TABLE 1. Summaryof EarthquakesShown in Figures2 and 3

M o, M s Area, Slip, Date Latitude Longitude Depth 102oNt-m 1000 km2 m Source

North AndeanEarthquakes 1 1906 1.0 -81.0 -- 200.0 8.7 114.0 5.2 1 2 1942 0.0 -80.0 -- 3.2 7.9 7.1 1.3 1 3 1958 1.0 -79.0 -- 5.2 7.8 6.6 2.3 1 4 1979 1.6 -79.4 33 29.0 7.7 28.0 2.7 1 5 March 1987 0.06 -77.72 8 0.4 7.0 .... 2 6 Sept. 1987 -0.90 -77.96 3 0.05 ...... 2 7 Sept. 1987 -0.98 -78.04 10 0.005 ...... 2 8 Nov. 1991 4.7 -77.5 25 0.7 ...... 2 9 Oct. 1992 6.9 -76.7 15 1.1 6.6 .... 3 10 Oct. 1992 7.1 -76.5 44 7.8 7.2 .... 3

CostaRican Earthquakes 1 1 March 1990 9.6 -84.9 22 1.0 7.0 .... 3,4 12 April 1991 9.6 -83.1 17 2.0 7.5 4.5 2.1 4,5 13 Sept. 1992 10.7 -87.7 15 1.6 7.0 1.6 -- 3

Prior to 1987, only great earthquakes(M > 7) areshown. After 1987, significant or well-studiedevents are listed. Not all information is available for all earthquakes.Sources are (1)Kanamori andMcNally [1982], (2) Ines Cifuentes, personal communication, 1992, (3) Harvard CMT solution, 1990, (4) J. Marino Protti-Quesada,personal communication,1991, and (5) Plafkerand Ward [ 1992].

From the short-termseismicity on the Bocon6 Fault zone in Nicaraguasouth to Panama.The first and largest event (focal the VenezuelanAndes, Aggarwal [1983] deduceda seismic-slip mechanismplotted in Figure3)occurred just southof the tip of rate of 10-!-_2mm/yr, and from historical seismicity, slip rates the Nicoya peninsula,approximately on a line betweenLiberia between8 and 11 mm/yr for the East Andeanand Bocon6 fault and Isla del Coco. Based on the seismic moment and the systems.Schubert [1980] estimateda similar averagerate of approximatearea of the aftershockzone, the principal event strike slip motion on the Bocon6 fault, 8 mm/yr for the last hadclose to 2 m of slip. Elastic dislocation modelingpredicts 12,000 years, basedon 100-m right-lateral offsets of glacial that this earthquakecaused only a small displacement(roughly moraines.Henneberg [1983] measured4.5 mm/yr of slip along 2 cm) at Liberia, located ~100 km from the epicenter. Model the Bocon6 fault using 6 yearsof geodeticdata. The slip rate displacementsat Lim6n and Isla del Coco were insignificant. on the Pallatangafault in Ecuadorhas been estimatedas 4-5 On the Caribbean margin of Costa Rica and Panama, mm/yr basedon the offset of a glacial moraine [Winter and convergence between the stable Caribbean plate and the LaVenu, 1988]. The seismic activity of the Iliniza fault Central American arc occurs on the North Panam• fold belt. (subparallel to and 70 km to the north of the Pallatanga fault) The North Panam• fold belt comes on land north of Puerto is similar to that of the Pallatanga fault, and it probably has a Lim6n. The subaqueouspart of the fold belt, off the northern similar slip rate (Hugo Yepes, EscuelaPolitecnica Nacional, coast of Panama,consists of young sedimentsbeing deformed Ecuador,personal communication,1991). Based on observed into anticlinal folds. Multichannel seismic and SeaMARC II slip rates, Kellogg et al. [1985] predictedthat relative to the surveys show undeformedoceanic crust dipping landward South American plate, the North Andes were moving beneath folded sediments of the deformed belt [Lu and approximately 10 mm/year toward the northeast. McMillen, 1983; Silver et al., 1990]. A two-dimensional density model constructedfrom refraction interval velocities Costa Rica subductionsystem predicts landward-dippingoceanic crust beneath the North The Cocos plate is being subductedbeneath the Caribbean Panam• fold belt [Kellogg et al., 1991]. Earthquakefocal plate at the southern Middle America trench. The NUVEL-1 mechanismsand seismicity suggest a poorly definedWadati- modelpredicts a subductionrate of 90 mm/yr nearly normalt o Benioff zone dipping about 50 degreesto the southwest the trench. The submarineCocos ridge is being subducted beneath the Panam• fold belt [Wolters, 1986; Adamek et al., beneath southwesternCosta Rica (Figure 3), and this feature 1988]. representsthe northern boundaryof the distinct Costa Rica The April 1991 Costa Rica earthquake(MS ~ 7.4) occurred segment of the trench [Burbach et al., 1984]. The subducted on the Caribbeancoast of Costa Rica shortly after the 1991 slab in this segment is shallower than in the remainderof the measurementswere made (Figure 4). The earthquakewas a trenchto the north, and the slab appearsto have been torn due thrust event on a plane striking between 102ø and 120ø and to the effectsof the attemptedsubduction of the buoyantCocos dipping between17 ø and 30øto the southwest[Giiendelet al., Ridge. 1991; Plafker and Ward, 1992; Goeset al., 1993]. The fault is In March and April of 1990, two months after the second one of a systemof faults that makeup the northwestextension measurementcampaign, a seriesof five shallow earthquakes, of the North Panam• fold belt. The earthquakeproduced up to the largestbeing MS ~ 6.9, andthree others being largerthan 20-30 cm of uplift along the coast, measuredby uplift of coral MS ~ 6.0, rupturedportions of the trench from southern reefsand other shoreline features[Plafker andWard, 1992] and 21,856 FREYMUELLERET AL.: PLATEMOTIONS PN NORTH ANDEAN REGION

10 ø N

80øW 70øW

Fig. 2. Tectonicfeatures and boundaries of the NorthAndes block. All plate velocityvectors are relative to theSouth American plate, exceptthe greyvector which represents the Caribbean-North Andes vector of Kellogget al. [ 1985]. Focalmechanisms or aftershockzones are shown for 10 significantearthquakes in the region.The numbers next to eachfocal mechanism or aftershock zonerefer to Table1. Earthquakes1-7 occurredprior to the 1988GPS campaign, and 8-10 afterthe 1991campaign. by repeatedgeodetic leveling [De Obaldia et al., 1991]. (Figure2). The southernmargin of the Caribbeanplate is now Verticalcoseismic displacements of a similar magnitudehave generally accepted to be a distributed zone of deformation, beenmeasured using GPS [Lundgrenet al., 1991], andalmost 2 although there remains disagreementover how muchof the m of horizontal coseismic displacementwere observedat North Andesshould be considered"Caribbean." The principal Puerto Lim6n [Lundgren et al., 1991]. The horizontal part of the plate boundaryzone consistsof a seriesof deformed displacementat Lim6n observedwith GPS is not in accordwith belts along the Caribbeancoast of Southand Central America, a dislocationmodel basedon the vertical displacementdata, including the South Caribbean deformed belt, the North andmay be duein part to local seawardslumping in response Panam•fold belt andthe LeewardAntilles [e.g., Pindell and to the earthquake[Plafker and Ward, 1992]. Barrett, 1990; Mann et al., 1990]. In Colombia, the South Caribbean deformedbelt includes the SouthernCaribbean Plate Boundary'Zone Sinfiand San Jacinto belts (Figure2), consistingof up to 12 km of Tertiaryturbidites, carbonates, and fluvial andlacustrine The locationof the southernmargin of the Caribbeanplate sediments[Duque-Caro, 1979; Duque-Caro,1984]. The wedge has beenthe subjectof considerablecontroversy. It has been of sedimentshas a very low topographicslope (<2 ø) [Totoand placed as far south as the Bocon6 and East Andean fault Kellogg, 1992]. Toto and Kellogg [1992] interpretedthe systems,and as far north as the South Caribbeandeformed belt SinfiandSan Jacinto belts as an accretionarywedge similar to FREYMUELLERET AL.: PLATE MOTIONS IN NORTH ANDEAN REGION 21,857

fairly consistentwith the given uncertainties.Notice that in Figure4 all estimatesfrom a given campaignfrequently lie either above or below the best fit line. This is an indication CARIBBEAN that all estimatesfrom a given campaignare slightly biased

(12) due to systematic errors, but the reducedchi-square statistics suggestthat the uncertaintiesused are reasonable,and that our Limon 10 ø uncertaintieshave accountedsufficiently for these systematic COCOS N errors. Since only two epochs are available for the Isla San Andr6s/Cartagena (Caribbean-North Andes) baseline, the displacementrate is the differencebetween the two solutions divided by the time elapsed between the experiments (3.05 years). The uncertaintyin the rate for this baselineis basedon the mean formal precision of eachexperiment's solution, with /•'sladel Coco the uncertainty due to the possible systematic error sources ,," / I I . I I describedabove addedin quadrature.This is a conservative / /' o 4ookm approach that reducesthe chance of interpreting systematic errors as plate motions. We feel it is appropriate to observe 90ow 80øW caution in the assessment of uncertainties for baselines with Fig. 3. The Costa Rica subductionsystem. Focal mechanisms only two occupations. are shown for the 1990 Gulf of Nicoya earthquake(M ~ 6.9), The observed baseline displacement rates are in general the 1991 Valle de Estrella.earthquake (M ~ 7.4), andthe 1992 agreementwith the predictions of the NUVEL-lplate motion Nicaraguanearthquake (M ~ 7.0). The numbersnext to each model, but there are significant differenceson most baselines; focal mechanism refer to Table 1. only for the Nazca-Cocos spreading rate do the NUVEL-1 predictionslie within the 95% confidenceellipses of the GPS baseline rates (Figure 1, Table 2). The plate motion model predictions and their uncertainties are shown with dashed the Barbados, Makran and Cascadia wedges, with pore lines. Notice that for the Cocos-Caribbeanconvergence rate, pressuresapproaching lithostatic. The high pore pressuresand the uncertainty in the GPS estimate is smaller than the low slip rate explainthe low rate of seismicityin the wedge. uncertainty in the NUVEL-1 model. No significant vertical Multichannel seismic profiles across the South Caribbean motions wereobserved at any sites. A summaryof the GPS and deformedbelt showevidence for activefolding of the youngest sediments and show that Caribbean acoustic basement has NUVEL-1 displacementrates is given in Table 2. underthrustthe deformedbelt [Luand McMillen, 1983; Lehner DISCUSSION et al., 1983; Laddet al., 1984]. Kellogg and Bonini [1982] have interpreteda zone of earthquakesdipping 20ø to the southeastbeneath the continent [Dewey, 1972; Pennington, Differences between our GPS baseline rates and the 1981] as a weakly definedWadati-Benioff zone resulting from predictions of the NUVEL-1 model for those baselinesmay be the slow underthrustingof Caribbean crust. Kellogg et al. explained in four possible ways: (1) systematicbiases in the [ 1985] useda block vector modelto estimatethe convergence GPS solutions which change systematically with time, (2) rate betweenthe Caribbeanand North Andesto be 17 mm/yr, to deficiencies in the NUVEL-1 model, (3) motion of the GPS the southeast.Dewey and Pindell [1985, 1986] find a similar sites with respectto the stableplate interiors, or (4)temporal convergencedirection between the Colombian basin and the 'variations in motions of certainstations dueto the effects of Maracaiboblock. Mann and Burke[1984] arguefor northerly the earthquake cycle. We consider the third and fourth directedconvergence between the Caribbeanand the Maracaibo possibilitiesto be the most likely. block. Systematic biases in the GPS solutions which bias the baselinerates are possible, but unlikely for two reasons.First, we have been very conservative in assigning uncertainties which considerthe magnitudesof potentialsystematic errors. Displacement rates were computed for the Isla Second,when baselines have been observedmultiple times Baltra/Jerusalen(Nazca-North Andes plates), Isla Baltra/Isla over severalyears, the systematicerrors would have to change del Coco (Nazca-Cocos)andthe Isla del Coco/Liberia (Cocos- linearly with time to bias the baselinerates. This is a primary Caribbean or Cocos-Panama) baselines based on the GPS reason for assigning very conservative uncertainties to the baseline estimates from 1988, 1990 and 1991. Each baseline Isla San Andr6s/Cartagenabaseline which has been measured was consideredindependently. The computeddisplacement only in 1988 and 1991. rates and their uncertaintiesare based on a weightedlinear fit Deficiencies in the NUVEL-1 model are also possible; for of the GPS solutions as a function of time for each baseline example, the motion of the Caribbean plate is poorly (Figure 4). The weights usedwere determinedfrom the formal constrainedrelative to other plates by the types of datausedin uncertaintiesof each estimateand an estimate of magnitudeof the NUVEL-1 model. Three years of GPSdata have determined the potential systematic errors (Appendix). The uncertainties the relative motion of Liberia andIsla del Cocomore precisely due to the potential systematicerrors make up a significant than does the NUVEL-1 model. Also, the NUVEL-1 model is part of the error budget(Appendix, Table A2). The reducedchi- probably less reliable at convergentplate boundariesthan at squarestatistics for the linear fits rangedfrom 0.44 to 1.29, extensional or transform boundaries, due to the lack of reliable indicating that the scatter of the data aboutthe best fit line is data which directly measureconvergence rate and direction. 21,858 FREYMUEL•RET AL.: PLATEMOTIONS IN Nom'• Am•EnN REOION

Cocos / Liberia (Cocos-Caribbean)

East North 50 IO0

o o

-50 -lOO -100

-150 -200

-200 . , - , - , -300 1988.0 1989.0 1990.0 1991.0 1988.0 1989.0 1990.0 1991.0

Baltra/Jerusalen (Nazca-North Andes)

East North 50 150

0 100 -50

50 -150

-200

-250

-,300 -50 1988.0 1989.0 1990.0 1991.0 1988.0 1989.0 1990.0 1991.0

Baltra / Cocos (Nazca-Cocos)

East North lOO 200 150

50 100

o 0 -50 • • • • • • -50 100

150

-lOO 200 1988.0 1989.0 1990.0 1991.0 1988.0 1989.0 1990.0 1991.0

Fig. 4. Baseline solutions as a function of time for three baselineswhich measurerelative motion of the major plates. The horizontalseparation of solutionsfor each campaignhas been exaggeratedto showthe individualsolutions.

Nevertheless,the NUVEL-1 model is in good agreementwith South America. We expect that baseline rates of change geodetic measurementsin many parts of the world, and it involving sites in the North Andes(Jerusalen and Cartagena in representswell the motions of the stable plate interiors on a this study) will differ from the predictions of the NUVEL-1 globalbasis [DeMets et al., 1990;Gordon and Stein, 1992]. modelby the amount that those sites are in motion relative to Applicationof the NUVEL-1model to specificsites is the South American plate. possibleonly wheneach site can be placedunambiguously on Temporal effects of the earthquakecycle may have affected a stable plate. In plate boundaryzones or other regions of the GPS baseline rates, but the magnitudeof these effectsis active deformation, we would expect that the motion of GPS small for the baselines considered.The two large earthquakes sites would differ from the motion predictedby NUVEL-1. A in Costa Rica both had substantial coseismic offsets but caused considerablebody of evidence,summarized in the first section, only small displacementsat the sites used in this study. An suggeststhat the North Andes region is moving relative to elastic dislocationmodel for the March 19.90 magnitude6.9 FREYMUELLERET AL.: PLATE MOTIONS IN NORTH ANDEAN REGION 21,859

TABLE 2. ObservedDisplacement Rates ComparedWith Predictionsof NUVEL-1 PlateMotion Model.

NUVEL-1 GPS

Baseline East North East North Zv2

Isla Baltra / Jerusalen -70+4 -1+2 -68 +7 15 +5 0.44 Isla Balt•a / Isla del Coco -4+3 65+5 -15 +6 66 +4 1.11 Isla del Coco / Liberia -54+4 -73+6 -40 +3 -59 +2 1.29 Isla San Andr•s / Cartagena -14+3 3+4 -11 +10 7 +6 --

The motion of the secondstation relative to the first station is given. Since thesebaselines are mostly oriented in the direction of relative plate motion, negative displacementrates indicatec'onvergence and positive rates divergence. The Zv 2 columngives the reducedchi square statistic of the linear fit to the GPS measurements. earthquakewas computedbased on fault parametersderived Caribbeancrust underthrusting Central America[Plaficer and from the Harvard Centroid Moment Tensor solution and a Ward,1992], further supportingthis hypothesis. solution of the University of California, Santa Cruz basedon The March 1990 magnitude6.9 earthquakeprobably did not local and teleseismic data (J. M. Protti-Quesada,personal producesignificant deformationat either of the sites that make communication, 1991). The model predicts 20 mm of up this baseline. Accordingto an elasticdeformation model, southward motion at Liberia (located ~100 km from the 20 mm of southwardmotion is predictedat the Liberiastation. epicenter), and insignificant displacementsat Limtn andIsla Removing the model coseismic displacementfrom the GPS del Coco. The 1991 earthquakeoccurred 2 months after the estimates for 1991 would improve the fit of the baseline 1991 measurements. The Limtn site, located close to the 1991 estimatesabout a line (chi-squareof 1.05 rather than 1.29) and epicenter,was not used in this study. produceslightly better agreementwith the NUVEL-lmodel's Nazca-NorthAndes convergence is measuredby the baseline predicted direction of convergence(0.1 ø away from the from Baltra to Jerusalen (1323 km). Jerusalen(on the North NUVEL-1 direction rather than 2.4ø). The results shown in Andes microplate) has a northward component of motion Figure 1 do not remove the model displacementof Liberia. relative to South America, so Baltra appears to be moving Additional GPS datain Costa Rica, taken in 1990 and 1991 but southward relative to the NUVEL-1 Nazca-South America not usedin this study, will allow a more reliable estimate of prediction. If the entire differencebetween the GPS estimate the coseismicdisplacement of Liberia [Niemeieret al., 1993]. and NUVEL-1 were due to motion of the North Andes, the North Isla SanAndrts on the Caribbeanplate moved southward at a Andesmicroplate would be moving relative to SouthAmerica rate of 7+6 mm/yr and eastwardat a rate of 11+10 mm/yr at a rate of 16 +5 mm/yr toward353 +25 ø. Kellogg et al. relative to Cartagenaon the northern coast of Colombia. As [1985] predictedthe motion of the North Andes microplate mentioned above, the uncertainties for this baseline are relative to SouthAmericato be roughly 10 mm[yr toward55ø. conservativebecause the baselinehas been occupiedonly If the predictednortheastward motion of the North Andes is twice.Given the large uncertainties, the observed convergence addedto the GPS displacementrate, the NUVEL-1 modelwould is consistent with both the NUVEL-1 Caribbean-South lie within the 95% confidence ellipses of the GPS estimate. America relative motion prediction and the Kellogg et al. Although the inferred motion of Jerusalenis directedmore [1985] Caribbean-NorthAndes prediction. Most of the northward than predicted, the differenceis not significant at convergencein the South Caribbeanplate boundaryzone must the 95% confidence level. be takenup offshore,north of Cartagena. Convergencebetween the Cocos and Caribbean plates is The rate of seafloor spreadingat the Gal•pagosRise is measuredby thebaseline between Isla delCoco on the Cocos measuredby the baseline between Isla Baltra on the Nazca plate andLiberia (595 km) or Limtn (636 km) in Costa Rica. plate andIsla del Coco on the Cocosplate (754 km). Relative The geodeticmonument used at the Limtn site was destroyed to Isla Baltra, Isla del Coco has a largerwestward component andreset between the 1990 and1991 experiments,so only the of motion than the NUVEL-1 modelpredicts, but the difference baseline to Liberia is usedhere. No significant motion was is not significant at the 95% confidencelevel. The scatterof observedbetween Limtn and Liberia during the period 1988- the estimatesfor the eastcomponent of this baselineis much 1990. The observedconvergence rate of 71 mm/yr is only 78% larger than for any other baseline. The north componentof of the 91 mm/yr predictedby the NUVEL-1, but the directionof this baselineis muchbetter determined, and the rate agrees motion is almost identical to that predicted by the plate with theNUVEL-1 predictionto within 1 mm/yr. motion model. If the NUVEL-1 Cocos-Caribbean relative Additional occupationsof the network will significantly motion vector is correct, then the observedconvergence rate improve the uncertaintiesin the GPS-derivedplate motion wouldbe consistent with the hypothesis that the stations in rates. With the first three occupations,the uncertaintyin the Costa Rica are moving northeastwardrelative to the stable Cocos-Caribbean(Panama)vector from GPS is comparableto Caribbean plate. Relative to Isla San Andrts, which is the uncertaintyfor the samevector in the NUVEL-1 model, and presumablylocated on the stable Caribbean plate, Liberia with future occupationsGPS can be expectedto determine moved northward at a rate of 14+6 mm/yr during the period relative plate motions in the region with a precision 1988-1991. The 1991 Costa Rica earthquake occurredon a comparableto or betterthan that of the NUVEL-1plate motion shallow southwestward-dipping thrust consistent with model. Initial GPS resultspresented here are consistentwith 21,860 FREYMUEL•RET AL.: PLATE MOTIONS IN NORTHANDEAN REGION deformation in the overriding plates at the Colombia-Ecuador (VLBI). In addition,data were u.sed from asmany other tracking and Middle America trenches.We expect that in the futureGPS sites aspossible. The positionsof these additionalsites were data from the full CASAnetwork will resolve details of active not constrained, but their use significantly improves the slip partitioning within the complex North Andean and estimates of the orbit corrections [e.g., Freymueller and Central Americanplate boundaries. Kellogg, 1990; Kornreich Wolf et al., 1990]. This approach makes the solutions sensitive to errors in the coordinates of APPENDIX ANALYSIS OFTHEGPS MEAS• the constrained stations; any such errors will bias the estimates for the baselines of interest in the CASA region. The GPS data were analyzed using the GIPSY software Also, the distribution of tracking sites aroundthe world can [Lichten andBorder, 1987; Blewitt, 1989; Sovers and Border, changethe estimatesfor the baselinesof interestby an amount 1990; Blewitt, 1990]. Details of the analysis are given in comparableto their uncertainties[Freymueller and Kellogg, Freymueller [1991]. Each day of data was analyzed 1990]. We attempt to estimatethe magnitudeof theseeffects, independently,using data from all stationsin the CASA region and add it to the formal uncertainties. By doing so, the and from a tracking network of sites distributedover almost uncertaintieson our plate motion estimateswill better reflect half the globe (FigureA1). Carrierphase data were used, along the true uncertainties in the GPS solutions. with pseudorangedata when they were available, to estimate station coordinates, corrections to the satellite orbits, time- Considerationof Potential SystematicErrors varying tropospheric path delays at each station, and the carrierphase ambiguities and receiverand satellite clock errors The tracking sitesused in eachexperiment were different, as (independently for each epoch). When possible, the carrier a different set of sites were operating at the times of each phase ambiguities were resolved to integer values using the CASAexperiment (Figure A1). In particulax;the 1990 tracking method of Blewitt [1989], which produces a dramatic network spannedless of the globe than the others, so improvement,of a factor of 2-4 or more, in the precision and additionaluncertainty is includedin the baselineestimates accuracyof GPS solutions [Dong and Bock, 1989; Blewitt, from 1990 [Freymueller,1991]. All of the available tracking 1989]. Dueto achangeof receivertypes betweenthe 1988 and sites were used in each solution, and three of the following 1990 experiments, most ambiguities in the 1990 and 1991 sites were constrained:Owens Valley (California), Fort Davis solutions could not be resolved (none were resolved for the (Texas),Richmond (Florida) andWestford (Massachusetts). The stationsconsidered here) [Freymueller, 1992]. coordinates used for these stations are taken from the Goddard Estimating correctionsto the GPS satellite orbits using data SpaceFlight CenterVLBImodel GLB223 (C. Ma, publishedas from a tracking network is necessaryto obtain a sufficient Table 1 by Larson et al. [ 1991]), in which North Americais level of precisionfor tectonicstudies. For a networkthe size of constrainedto move accordingto the AM0-2 absoluteplate the CASA network, orbit improvementresults in a factor of 5- motion model of Minster and Jordan[1978]. A small geocenter 10 increase in precision. Using a standardapproach, the offset to the GLB223 coordinatesis applied to provide a coordinates of three tracking sites are constrained to the geocentric reference frame. Further detail is given by coordinatesdetermined from very long baselineinterferometry Freymueller[ 1991].

i 70 ø I I I

I :::::::::::::::::::::::::::::::::::::::::::::::::::::::::::::: I I I all I

40 ø

-70 ø 90 ø 150 ø 210 ø 270 ø 330 ø 30 ø 90 ø

Fig. A1. Trackingnetworks used in the threecampaigns. The yearsin whicheach site wasused are indicated by numbersnext to each site. FREYMUFJ.LF.RET AL.: PLATEMOTIONS IN NORTHANDEAN REGION 21,861

TABLE A1. Error Budgetfor ConsideredSystematic ObservedPrecision of the GPS BaselineEstimates Errorsfor Two SampleBaselines in the CASA Region In general, the precision and accuracyof GPS baseline Error Source East North Vertical solutionsdegrade with baselinelength becausethe effectsof satellite orbital errors become more important as the station Ida BaltraIJerusalen (1323 kin) separationincreases. There is also a componentof precision 1988 Fiducial error 24 20 10 which is independent of baseline length. In the 1988 1990 Fiducial error 31 17 18 experiment, the precision of our GPS solutions was 1990 Trackingnet difference 20 2 0 2 0 approximately(5-6 mm)+ (1-3) partsin 108for thehorizontal 1990 Local net difference 10 10 10 baselinecomponents for baselinesup to 750 km in length. 1991 Fiducial error 31 17 18 Ambiguityresolution was less successful in the 1990and 1991 solutions, but there weremore satellites available (7 in 1988, Ida del Coco/Liberia (585 km) 12 in 1990, 16 in 1991). As a result, it is not simple to 1988 Fiducial error 14 8 10 comparebaseline precision in the threeexperiments, as there 1990 Fiducial error 10 10 18 are different constant and length-dependentterms for each 1990 Trackingnet difference 10 10 10 experiment.In general,for the baselinesused here, the highest 1990 Local net difference 5 5 5 precisionwas obtained in 1988, followedin orderby1991 and 1991 Fiducial error 10 10 18 1990 (Table A2). The precision of the solutions for the baselinevertical component varies from 10 to 60 mm in 1988, Thesenumbers are an estimateof the potential magnitude and 20 to 40 mm in 1990 and 1991. While it is not of the systematicerror in the baselinedue to each systematic independentof baselinelength, there is no simple relation error source.These errors are addedin quadratureto the formal with length. uncertaintiesto give a truer estimateof the baselineaccuracy. TABLE A2. GPS BaselinePrecision for the BaselinesUsed in the Study

Baseline Length Formal Repeatability Consider

1988

Isla B altra / Jerusalen 1323 20/14 36/14 24/20 Isla Baltra / Isla del Coco 754 14/11 54/16 16/10 Isla del Coco / Liberia 585 8/7 11/8 14/8 Isla San Andr•s / Cartagena 710 9/8 11/8 14/8

1990

Isla B altra / Jerusalen 1323 24/12 24/20 38/28 Isla Baltra / Isla del Coco 754 20/7 23/7 25/21 Isla del Coco / Liberia 585 14/5 23/4 15/15

1991

Isla B altra / Jerusalen 1323 19/12 28/17 31/17 Isla Baltra / Isla del Coco 754 17/7 17/22 20/15 Isla del Coco / Liberia 585 12/4 15/12 10/10 Isla SanAndr6s / Cartagena 710 14/6 14/4 20/15 Given are the baseline length in kilometers, the mean formal uncertainty("formal") for the baseline and the observedrms repeatability ("repeatability") for each campaign, and the added uncertainty to accountfor possible systematiceffects ("consider").All uncertaintiesare given in millimeters for East/Northcomponents.

An error budgetfor two samplebaselines used is given in Acknowledgments.The work presentedhere wasfunded principally Table A1, including estimates of the magnitudesof the by National Science Foundationgrants EAR-8617485 and EAR- systematicchanges in the baselineestimates due to errorsin 8904657 to JamesN. Kellogg. Many other institutions contributed significant additional funding directly to the CASA Project or the coordinates of the constrained sites, and differences in the indirectlythrough the supportof global trackingnetworks (see Table trackingnetworks and the local CASAnetwork in 1990. The 1 and the acknowledgmentsgiven by Kellogg et al. [1989]). The latter effect is included because the 1990 local network was contribution of NASA through the Jet PropulsionLaboratory is also very sparse, consisting of a small numberof stations especiallynotable. In additionto thoseacknowledged by Kellogget widely separated.A test using 1988 dataindicated that there al. [ 1989],we gratefully acknowledge the collaborationand assistance of WolfgangNiemeier (Technische Universitiit Braunschweig, Institut weresmall changesto the estimatesof the longestbaselines fiir Vermessungskunde),Andreas Bagge (Universifiit Hannover), Heinz when sucha sparsenetwork wasused. It is not clear whether Saler (Universitiit Stuttgart), field engineersMichael Jackson, the changesare systematic,but to be conservativewe have DouglasRoberts and JamesNormandeau (University NAVSTAR assumedthat they are. In effect, the considerationof these Consortium),and the many field operatorswho worked on oneor more 'potential systematicerrors deweightsthe 1990 solutions projects. This manuscriptwas improved with carefulreviews by RobertSpeed and James Pindell. relative to the others. 21,862 FREYMUELLERET AL.: PLATEMOTIONS IN NORTHANDEAN REGION

303-316, 1984. Freymueller, J. T., CASA- Central and South America GPS geodesy:Crustal motions determinedfrom 1988 and 1990 Adamek,S., C. Frohlich, andW. D. Pennington,Seismicity v f the Caribbean-Nazca boundary:Constraints on block epoch measurementsin Colombia, Costa Rica and Ecuador, tectonicsof the Panam•region, J. Geophys.Res., 93, 2053- Ph.D. thesis,Univ. of S.C., Columbia, 1991, 172pp. 2075, 1988. Freymueller,J. T., Comparisonof baseline resultsfor the TI- Aggarwal, Y., Seismic slip rates andearthquakes rupture zones 4100 andTrimble 4000SDT geodetic GPS receivers, Bull. Geod., 66, 272-280, 1992. in the southernCaribbean:implications for plate motions Freymueller,J. T., and J. N. Kellogg, The extendedtracking and earthquakehazard in this region (abstract)Abs. loth Caribb. Geol. Conf., 16, 1983. network and indications of baseline precision and accuracy Argus, D. E, and R. G. Gordon, No-net-rotation model of in the North Andes, Geophys. Res. Lett., 17, 207-210, 1990. current plate velocities incorporating plate motion model NUVEL-1, Geophys.Res. Lett., 18, 2039-2042, 1991. Goes, S. D. B., A. A. Velasco, S. Y.. Schwartz,and T Lay, The Blewitt, G., Carrier phaseambiguity resolutionfor the Global April 22, 1991, Valledela Estrella,Costa Rica (Mw = 7.7) Positioning System applied to geodetic baselines up to earthquake and its tectonic implications: A broadband 2000 km, J. Geophys.Res., 94, 10,187-10,203, 1989. seismicstudy, J. Geophys.Res., 98, 8127-8142, 1993. B lewitt, G., GPS techniquesfor monitoring geodynamicsat Gordon, R. G., and S. Stein, Global tectonics and space regional scales, in Proceedingsof the SecondInternational geodesy,Science, 256, 333-342, 1992. Symposium on Precise Positioning with the Global Giiendel, F., et al., Mainshock-aftershock sequenceassociated Positioning System, pp. 494-507, The CanadianInstitute of with the Costa Rica earthquakeofApril 22, 1991 (abstract), Surveying and Mapping, Ottawa,Canada, 1990. Eos Trans.AGU, 72 (44), Fall Meeting suppl., 301, 1991. Bowin, C., Caribbean gravity field and plate tectonics, Spec. Henneberg, H. E., Geodetic control of neotectonics in Pap. Geol. Soc. Am., 169, 79pp, Geological Society of Venezuela,Tectonophysics, 97, 1-15, 1983. America, Boulder, Colorado, 1976. Kanamori, H., and K. McNally, Variable rupturemode of the Burbach, G., C. Frohlich, W. Pennington, and T Matumoto, subductionzone along the Ecuador-Colombiacoast, Bull. Seismicity and tectonics of the subductedCocos plate, J. Seismol. Soc. Am., 72, 1241-1253, 1982. Geophys. Res., 89, 7719-7735, 1984. Kelleher, J. A., Rupture zones of large South American Campbell, C. J., The Santa Marta wrench fault of Colombia earthquakesand some predictions,J. Geophys. Res., 77, and its regional setting, in Trans. IV Caribb. Geol. Conf., 2087-2103, 1972. Trinidad, 247-260, 1968. Kellogg, J. N., Cenozoic tectonic history of the Sierra de Comit6 del Afio Geoffsico Internacional del Ecuador, Breve Perij• and adjacent basins, in Caribbean-SouthAmerican Historia de los PrincipalesTerremotos en la Repdblica del Plate Boundary and Regional Tectonics, edited by W. E. Ecuador, Ministerio de Educaci6n Publica, Quito, Ecuador, Bonini, R. B. Hargravesand R. Shagam, Geol. Soc. Am. 1959. Mem., 162, 239-261, 1984. DeMets, C., R. G. Gordon, D. F. Argus and S. Stein, Current Kellogg, J. N., andW. E. Bonini, Subductionof the Caribbean plate motions,Geophys. J. Intl., 101, 425-478, 1990. plate and basementuplifts in the overriding SouthAmerican De Obaldia, F., T. Marino, R. Van der Laat, F. Hernandez, E. plate, Tectonics,1, 251-276, 1982. Malavassi, and K. C. McNally, Coseismicuplift associated Kellogg, J. N., I. J. Ogujiofor, and D. R. Kansakar,Cenozoic with the CostaRica earthquake(M s 7.5)of the April 22, tectonics of the Panam• and North Andes blocks, Mern. 1991 (abstract), Eos Trans. AGU, 72 (44), Fall Meeting Sixth Lat. Am. Geol. Congr., Bogota, Colombia, 1, 40-59, suppl., 301, 1991. 1985. Dewey, J.E, Seismicity and tectonics of WesternVenezuela, Kellogg, J. N., T H. Dixon, and R. E. Neilan, CASA: Central Bull. Seism. Soc. Am., 62, 1711-1751, 1972. and South America GPS geodesy, Eos Trans. AGU, 70, Dewey, J. F., and J. L. Pindell, Neogene block tectonics of 649,651,655-656, 1989. eastern Turkey and northern South America: Continental Kellogg, J. N., V. M. Godley,C. Ropain U., A. Bermudez,and applications of the finite differencemethod, Tectonics,4, C. L.V. Aiken, Gravityfield of Colombia,eastern PanarnJ, 71-83, 1985. and adjacentmaxine areas,Geol. Soc. Am. Map Chart Ser., Dewey, J. E, and J. L. Pindell, Reply to comment by Amos MCH070, 1991. Salvadoron "Neogeneblock tectonics of easternTurkey and KornreichWolf, S., T Dixon, andJ. Freymueller,The effectof northern South America: Continental applications of the tracking network configuration on GPS baseline estimates finite difference method;' Tectonics, 5, 703-705, 1986. for the CASA UNO experiment, Geophys. Res. Lett., 17, Dong, D.-N., and Y. Bock, Global Positioning System 647-650, 1990. network analysiswith phaseambiguity resolution applied to Ladd, J. W., M. Truchan, M. Talwani, P. L. Stoffa, P. Buhl, R. crustaldeformation studies in California, J. Geophys.Res., Houtz, A. Mauffret, and G. Westbrook, Seismic reflection 94, 3949-3966, 1989. profiles acrossthe southernmargin of the Caribbean,in Duque-Caro, H., Major structural elements of northern Caribbean-South American Plate Boundary and Regional Colombia, Mem. Am. Assoc. Pet. Geol., 29, 329-351, Tectonics,edited by W. E. Bonini, R. B. Hargravesand R. 1979. Shagam,Geol. Soc.Am. Mem., 162, 153-159, 1984. Duque-Caro,H., Slxucturalstyle, diapirism, and accretionary Larson, K. M., E H. Webb, and D.C. Agnew, Application of episodes of the Sinfi-San Jacinto terrane, southwestern the Global Positioning System to crustal deformation Caribbean borderland, in Caribbean-South American Plate measurement, 2, The influence of errors in orbit Boundaryand Regional Tectonics,edited by W. E. Bonini, R. determinationnetworks, J. Geophys. Res., 96, 16,567- B. Hargraves,and R. Shagam, Mem. Geol. Soc. Am., 162, 16,584, 1991. FREYMLrELLERET AL.: PLATEMOTIONSIN NORTHANDEAN REGION 21,863

Lehner, P., H. Doust, G. Bakker, P. Allenbach, and J. Gueneau, GeologicalSociety of America, Boulder,Colorado, 1990. Active margins- Caribbean margin of South America, Pindell, J. L., S.C. Cande,W. C. Pittman III, D. B. Rowley, J. profiles C-1422, C-1412, and C-1413, in Seismic F. Dewey, J. LaBrecque,and W. Haxby, A plate-kinematic Expression of Structural Styles 3, edited by A. W. Bally, framework for models of Caribbean evolution, Stud. Geol., 15, 3.4.2-111-3.4.2-128, 1983. Tectonophysics,155, 121-138, 1988. Lichten, S.M., andJ. S. Border,Strategies for high precision Platker, G., and S. N. Ward, Backarc thrustfaulting and tectonic GPS Orbit Determination, J. Geophys. Res., 92, 12,751- uplift along the Caribbean sea coast during the April 22, 12,762, 1987. 1991 Costa Rica earthquake,Tectonics, 11, 709-718, 1992. Lu, R. S., andK. J. McMillen, Multichannel seismic surveyof Schubert, C., Late-Cenozoicpull-apart basins, Bocon6 fault the Colombia Basin and adjacent margins, in Studies in zone, Venezuelan Andes, J. Struct. Geol., 2, 463-468, 1980. Continental Margin Geology, editedby J. S. Watkins andC. Schubert, C., Neotectonics of the Bocon6 fault, western L. Drake, AAPG Men,., 34, 395-410, 1983. Venezuela,Tectonophysics, 52, 205-220, 1982. Lundgren,P., et al., GPS measurementsof crustaldeformation Shagam, R., The northern terminationof the Andes,in The following the 22 April 1991 Costa Rica earthquake OceanBasins andMargins, 3, editedby A. E.M. Nairn and (abstract), Eos Trans. AGU, 72 (44), Fall Meeting suppl., E.G. Stehli, 325-420, Plenum, New York, 1975. 301, 1991. Silver, E. A., D. L. Reed, J. E. Tagudin, and D. J. Heil, Lyon-Caen, H., R. Armijo, and I. Cifuentes, Recent Implications of the North annd SouthPanama deformed belts deformation of the Ecuadorcan sub-Andes (abstract), Eos for the origin of the Panamaorocline, Tectonics,9, 261- Trans. AC•, 71, 1560, 1990. 282, 1990. Mann, P., and K. Burke, Neotectonics of the Caribbean, Rev. Sovers, O. J., and J. S. Border, Observation model and Geophys.,22, 309-362, 1984. parameter partials for the JPL geodetic GPS modeling Mann, P., C. Schubert, and K. Burke, Review of Caribbean software "GPSOMC;' JPL Publ. 87-21, Rev. 2, Jet Propul. neotectonics, in The Caribbean Region, vol. H, The Lab., Pasadena, Calif., 1990. Geologyof North America,edited by Dengo, G., andCase, J. Toro, E. A., and J. N. Kellogg, Structureof the Sinfi-San E., pp. 307-338, Geological Society of America, Boulder,, Jacinto fold belt - An active accretionaryprism in northern Colorado, 1990. Colombia, J. South Am. Earth Sci., 5, 211-222, 1992. Minster; J. B., andT H. Jordan,Present-day plate motions, J. Winter, T., and A. LaVenu, Evidencias morfologicas y Geophys. Res., 83, 5331-5354, 1978. microtectonicas de una falla de rumbo actira en la parte Niemeier, W., J. Roldan, L. Aguilar, H. Pelzer, A. Bagge, W. central del Ecuador,paper presented at Quinto Congreso Augath, and W. Seifert, The project CORBAS- Ecuatoriano de Geologfa, Minas, Petrolcos y Ciencias Determination of recent crustal movements in Costa Rica, in Afines, Loja, Ecuador,November, 1988. Recent Geodetic and Gravimetric Research in Latin America, Wolters, B., Seismicity and tectonics of southern Central editedby W. Torge, A. Gonzalez Fletcher,and J. 13.Tanner, America and adjacentregions with special attention to the 146-156, Springer-Verlag,Berlin, 1993. surroundingsof Panam•, Tectonophysics, 128, 21-46, Nishenko, S., Circum-Pacific Seismic Potential 1989-1999, 1986. U.S. Geol. Surv. Open File Rep., 89-85, 1989. J. T. Freymueller, Department of Geophysics,Stanford Pennington, W. D., Subductionof the easternPanam• basin University, Stanford,CA 94305. and the seismotectonics of northwestern South America, J. J. N. Kellogg and V. Vega, Department of Geological Geophys. Res.,86, 10,753-10,770, 1981. Sciences,University of SouthCarolina, Columbia,SC 29208. Pindell, J. L., and S. F. Barrett, Geological evolution of the Caribbean region; A plate tectonic perspective, in The (ReceivedMay 15, 1992; CaribbeanRegion, vol. H, The Geology of North America, revised December 7, 1992; edited by Dengo, G., and Case, J. E., pp. 405-432, acceptedMarch 3, 1993.)