Wide plate margin deformation, southern Central America andnorthwestern South America, CASA GPS observationsRobert Trenkampa,*, James N. Kellogga, Jeffrey T. Freymuellerb, Hector P. MoracaDepartment of Geological Sciences, University of South Carolina, Columbia, SC 29208, USAbGeophysical Institute, University of Alaska, Fairbanks, AK 99775, USAcVolcanological and Seismological Observatory, INGEOMINAS, Avenida 12 de Octubre 15-47, Manizales, ColombiaReceived 1 November 2000; accepted 1 January 2002AbstractGlobal positioning system data from southern Central America and northwestern South America were collected during 1991, 1994, 1996,and 1998 in Costa Rica, Panama, Ecuador, Colombia, and Venezuela. These data reveal wide plate boundary deformation and escapetectonics occurring along an approximately 1400 km length of the North Andes, locking of the subducting Nazca plate and strain accumula-tion in the Ecuador±Colombia forearc, ongoing collision of the Panama arc and Colombia, and convergence of the Caribbean plate withPanama and South America. Elastic modeling of observed horizontal displacements in the Ecuador forearc is consistent with partial locking(50%) in the subduction zone and partial transfer of motion to the overriding South American plate. The deformation is hypothesized tore¯ect elastic recoverable strain accumulation associated with the historic seismicity of the area and active faulting associated withpermanent shortening of 6 mm/a. Deformation associated with the Panama±Colombia collision is consistent with elastic strain accumulationon a fully locked Atrato±Uraba Fault Zone suture. q 2002 Elsevier Science Ltd. All rights reserved.Keywords: GPS; North Andes; oblique subduction; arc continent collision; earthquake strain; escape; caribbean1. IntroductionMid-ocean ridge and transform fault azimuths, spreadingrates, and earthquake slip vectors at plate boundaries havesuccessfully explained large-scale features of plate kine-matics and demonstrated that plate interiors generallybehave rigidly over geologic time scales (Le Pichon,1968; Minster et al., 1974; Minster and Jordan, 1978;Chase, 1978; DeMets et al., 1990). However, convergentplate boundaries were recognized early in the developmentof plate tectonics as wide zones of deformation, and thoseboundaries involving a continental plate boundary are muchwider than are those consisting only of oceanic plates(Isaacs et al., 1968; Dewey and Bird, 1970; Freymueller etal., 1993; Kellogg and Vega, 1995; Gutscher et al., 1999).Global positioning system (GPS) measurements providecost-effective and precise constraints on models of platetectonic processes at convergent boundaries, such as conti-nuum deformation versus microplate or block rotation(Thatcher, 1995).The Central and South America (CASA) GPS project wasinaugurated in 1988 to study plate motions and crustaldeformation in a tectonically active area of complex inter-action among the Nazca, Cocos, Caribbean, and SouthAmerican plates (Fig. 1). The tectonic development of theCASA area has been the object of multiple geologic andgeophysical studies (for a comprehensive listing, seeKellogg and Vega, 1995; Ego et al., 1996; Gutscher et al.,1999 and references therein). Previous studies have shownthe CASA region to be a complex area of plate convergenceand deformation, but the location of plate boundariesremains uncertain. At least two microplates, Panama andNorth Andes, have been hypothesized by Kellogg et al.(1985) and Kellogg and Vega (1995) (Figs. 1 and 2). Thispaper presents the results of CASA geodetic measurementsspanning the years 1991±1998 in Costa Rica, Panama,Colombia, Ecuador, and Venezuela and focuses on GPSmeasurements related to the following:1. Oblique subduction at the Ecuador trench and the `esca-pe'of the North Andes;2. Earthquake strain accumulation at the Ecuador trench;3. Island arc±continent collision, Panama±Colombia; and4. Caribbean plate subduction.Journal of South American Earth Sciences 15 (2002) 157±1710895-9811/02/$ - see front matter q 2002 Elsevier Science Ltd. All rights reserved.PII: S0895-9811(02)00018-4www.elsevier.com/locate/jsames* Corresponding author. Tel.: 11-803-777-4501; fax: 11-803-777-9133.E-mail address: [email protected] (R. Trenkamp).2. Data analysisAll data presented in this paper have been analyzed usingGIPSY OASIS or GYPSY OASIS II software developed atthe Jet Propulsion Laboratory (JPL), California Institute ofTechnology (Lichten and Border, 1987; Blewitt, 1989,1990). Details of the 1991 data analysis are given in Frey-mueller (1991) and Freymueller et al. (1993). Each day of1991 data was analyzed independently using GIPSY OASIS,the data from all stations in the CASA region, and data fromtracking sites distributed over almost half the globe. The1994, 1996, and 1998 data were analyzed using GIPSYOASIS II. All days were analyzed independently using allthe stations in the CASA region and a tracking network simi-lar in the number of stations to the 1991 campaign but with amore favorable regional geographic distribution for CASAdue to the increased number of permanent sites, especially inthe southern hemisphere. All the raw data were passedthrough automatic editors: Turboedit (Blewitt, 1990) forRogue and Turborogue receivers and Phasedit (Freymueller,unpublished algorithm) for all other receivers. These editors®nd and correct cycle-slips (phase breaks in the data stream)and remove outliers. The data then are decimated to a 6-minute interval for data collected before 1996 and a 5-minuteinterval for data collected after 1996 to enable the use of theJPL's precise clocks and orbits. The GIPSY OASIS II solu-tion suite of software was then run following a strategy simi-lar to that described by He¯in et al. (1992) to obtain solutionsbased on all data available for a given day. All daily solutions,except those for 1991, were derived with weak constraints oninitial positions and then transformed into ITRF96 usingJPL-produced daily frame ®les. The 1991 data were trans-formed to ITRF96 using a seven-parameter transformation.3. Velocity ®eld descriptionsCASA GPS measurements used in this study were madeduring 1991, 1994, 1996, and 1998. Forty-four usablevectors for tectonic interpretation (Table 2, Fig. 3) weredetermined and are distributed over the major tectonicfeatures of the study area. Two of our sites are located onR. Trenkamp et al. / Journal of South American Earth Sciences 15 (2002) 157±171158Fig. 1. Tectonic map for