With support in the Dominican Republic from the Instituto Cartografico Militar, INDRHI, and the Direccion General de Mineria.

Funding from the National Science Foundation (USA), the Centre National de la Recherche Scientifique (France).

1. Tectonic setting

Relative motion of the Caribbean plate with respect to the North American plate is accomodated across a 200 km wide deformation zone, to which belong the island of Hispaniola (Figure 1). The relative motion of the two plates is poorly defined in global plate motion models because of a lack of constraining data such as transform azimuths, ocean ridge spreading rates. The distribution of strain within the plate boundary zone remains also poorly constrained in spite of clear evidences for active deformation. Indeed, most of the major active faults in Hispaniola have been identified from field observations, aerial photos, and satellite images. Late Quaternary tectonic activity is clearly recognized for both the Enriquillo and Septentrional fault zones (e.g. Mann et al, 1984, 1995). Trenching across the Septentrional fault indicates that it last ruptured 730 years ago, slipping about 5 m horizontally and 2 m vertically (Prentice et al., 1993). Holocene terraces offset by the Septentrional fault indicate 13-23 mm/yr of slip over the past 3000 years (Mann et al., 1998). Historical and recorded seismicity show several earthquakes of magnitude greater than 6.5 associated with the Septentrional and Enriquillo fault zones (Robson, 1964 ; Sykes et al., 1982). Significant seismicity and active deformation have also been identified offshore along the North Hispaniola Trench (Dillon et al., 1992, Dolan et al, 1998) and the Muertos Trench to the south (Ladd et al., 1977).

In spite of a wealth of structural and seismotectonic information on these active structures, the strain distribution among them is not yet quantified and the slip rates of the major active faults in and around the island of Hispaniola are not known. This however information is critical for a better understanding of the dynamics of the northern Caribbean plate boundary zone and of the seismic threat posed by these active faults.

We chose to address these issues using Global Positioning System (GPS) measurements (Figure 2). The GPS is a space geodetic technique based on a constellation of 24 satellites that permanently broadcast a radio signal on two frequencies in the L-band. Using antennas and receivers set up on geodetic benchmarks that collect data on those two frequencies, it is possible to compute distances between the different sites of a regional network with an accuracy of about 2 to 3 millimeters. Repeated measurements every year of a network that covers a region of active deformation therefore allows the direct estimation of site displacements and network strain for typical rates of crustal deformation (a few millimeters to a few centimeters per year).
 
 

Figure 2: Field GPS measurements during the 1999 campaign in the Dominican Republic. Left: Ing. Luis Pena after installing a GPS antenna on a spike mount. Right: Ing. Luis Pena leveling a tripod.

In addition to these scientific objectives, precise GPS determination of geodetic site coordinates is of interest for surveying and mapping projects in the Dominican Republic.

2. GPS measurements

2.1. Earlier GPS measurements in the Dominican Republic

In 1986, a sparse network of geodetic sites was occupied by GPS in the northeastern Caribbean, as part of a NASA program to test and validate this new technology in a humid tropical environment (Dixon et al., 1991). Three of the sites were installed in the Dominican Republic (Cabo Frances Viejo, Capotillo, and Cabo Rojo, Figure 3).

In 1994, this network was reoccupied in the framework of the CANAPE (Caribbean North America Plate boundary Experiment) project and number of new sites were added, in particular in the Dominican Republic. Some of these sites have been reoccupied in 1995, 1996, 1998, and 1999.

The CANAPE project is a collaboration between the University of Madison, Wisconsin (C. DeMets), the University of Texas at Austin (P. Mann), the University of Miami (T. Dixon), the University of Puerto Rico at Mayaguez (P. Jansma and G. Mattioli), and the University of Nice, France (E. Calais and B. de Lépinay). The field work has been carried out in collaboration with and thanks to the logistical support of the following Dominican agencies : the Instituto Cartografico Militar, INDRHI, and the Direccion General de Mineria.

2.2. The 1999 CANAPE GPS campaign

In September 1998, as part of the CANAPE project, 27 new geodetic benchmarks have been installed in the Dominican Republic (Figure 3). The primary aim of the CANAPE 99 GPS campaign was to perform a first-epoch occupation of these sites. The field campaign took place during two weeks from January 24 to February 7, 1999. The participants were Ing. Luis Pena (Univ. Santiago), Ramon Mejia (INDRHI), sergents Jose R. Hernandez Almonte, E.N., and Jose F. Jimenez Torres, F.A.D. from the Instituto Cartografico Militar, Alberto Manuel Lopez Venegas, Hector Rodriguez Cesani, Daniel Lao, Manuel Canabal (Univ. Puerto Rico, Mayaguez), B. Mercier de Lépinay, E. Calais (CNRS Univ. Nice, France). The participants were divided into four teams with two GPS receivers per team. Many of the benchmarks installed in 98 are located in secure places where the equipment could be left unattended, which allowed the field work to gain in flexibility (Figure 2).

The measurements were made using 8 dual-frequency Trimble SSI GPS receivers and choke-ring antennas belonging to the University of Puerto Rico, Mayaguez. Each site was occupied during two to three consecutive 24 hour sessions. We were able to occupy all the new 1998 sites but two (Punta Palenque and San Pedro de Macoris), but we also occupied Puerto Escondido and La Colonia de Duverge (1994 sites). The measurement schedule is given in Table 1 (numbers indicate start and end time at each site, yellow boxes indicate sessions less than 12 hour long).

2.3. The 2001 GPS campaign

As we had planned, we are now ready for a second occupation of the sites that were installed in 1998 and measured for the first time in 1999. The campaign will start on February 15 and end on March 4, 2001. We plan to deploy 4 teams of 2/3 persons each, each team carrying 2 GPS receivers and observing two sites simultaneously. We plan to use 8 Trimble 400 SSIs GPS receivers with choke-ring antennas. The site occupation schedule for the 3001 campaign is given in Table 2 (numbers indicate start and end time at each site, yellow boxes indicate sessions less than 12 hour long). Each site was generally measured for 2 consecutive 24 hours sessions.

3. GPS data processing, first results

3.1. Data processing

We processed pseudorange and phase GPS data from the 1994 to 1999 campaigns in single-day solutions using the GAMIT software. We solve for regional station coordinates, satellite state vectors, 13 tropospheric zenith delay parameters per site and day, and phase ambiguities using doubly-differenced GPS phase measurements. We use IGS final orbits, IERS (International Earth Rotation Service) earth orientation parameters, and apply azimuth and elevation dependant antenna phase center models, following the tables recommended by the IGS. We include 11 global IGS stations with position and velocities well determined in the International Terrestrial Reference Frame (ITRF97) to serve as ties with the global reference frame (stations GOLD, BRMU, ALGO, CRO1, KOUR, MAS1, RSM5, RCM6, AREQ, SANT, KOKB).

The least squares adjustment vector and its corresponding variance-covariance matrix for station positions and orbital elements estimated for each independent daily solution are then merged together using a Kalman filter (GLOBK). We also add daily solutions from global tracking sites obtained from the IGS data processing center at Scripps Institution of Oceanography. Site positions and velocities, earth orientation parameters, and orbits are loosely constrained at this stage.
We then impose the reference frame using the resulting combined solution by minimizing the position and velocity deviations of 14 IGS stations with respect to the ITRF97, while estimating an orientation, translation and scale transformation. The height coordinate and velocity are downweigthed by a factor of 10 relative to the horizontal components.

3.2. Velocity field

Figure 4 displays the GPS-derived velocities relative to the north American plate, for the sites that have been observed at least twice. We find an insignificant residual velocity of 3.3±3.3 mm/yr at site TURK on the Bahamas platform, consistent with the previous interpretation of Dixon et al. (1998) that this site is located on the stable North America plate.
For the sites located in Puerto Rico (ISAB, PARA), St. Croix (CRO1), Guadeloupe (GUAD), Martinique (MART), Aves Island, (AVES), and Barbados (BARB), we find GPS velocities in very good agreement with DeMets et al.'s (2000) kinematic model predictions for the Caribbean plate. Residual velocities at these sites range from 0.6 mm/yr to 3.3 mm/yr (1.4 mm/yr on average) and are insignificant at the 95% confidence level. The same is true for ROJO, located at the southernmost tip of Hispaniola. Its velocity is consistent with the prediction of a rigid Caribbean model, with a residual velocity of 1.9 mm/yr.

GPS-derived velocities at five sites in Hispaniola (CAPO, FRAN, CONS, MOCA, SDOM) show significant discrepancies with the prediction of a rigid Caribbean plate model. These sites have consistently more easterly strikes and slower rates than the sites in Puerto Rico, the Virgin Islands and the Lesser Antilles that are moving with the stable Caribbean plate. In addition, one can observe a decrease in velocities from south to north across Hispaniola. This spatial gradient, perpendicular to the trend of the plate boundary zone, could be caused by either elastic strain accumulation on individual locked faults in Hispaniola or continuous and/or anelastic deformation across the entire Hispaniola plate boundary zone.

A detailed geophysical interpretation of these results in currently being preformed. In particular, we would like to interpret the GPS-derived velocities in terms of seismic hazard. In that respect, the densification started in 1999 should provide a better description of the strain distribution and accumulation on active faults across the Dominican Republic, which should, in turn, improve our understanding of seismic hazard in the country.

3.3. Coordinates of the sites installed in 1999

We have also derived precise coordinates in the International Terrestrial Reference Frame 1997 (ITRF97) for the sites that were measured for the first time in 1999. These coordinates, and their transformation into ellipsoidal coordinates on WGS84, are given in Table 3.