Purdue University Purdue e-Pubs

Open Access Dissertations Theses and Dissertations

4-2016 Present day plate boundary deformation in the Caribbean and crustal deformation on southern Haiti Steeve Symithe Purdue University

Follow this and additional works at: https://docs.lib.purdue.edu/open_access_dissertations Part of the Caribbean Languages and Societies Commons, Geology Commons, and the Geophysics and Seismology Commons

Recommended Citation Symithe, Steeve, "Present day plate boundary deformation in the Caribbean and crustal deformation on southern Haiti" (2016). Open Access Dissertations. 715. https://docs.lib.purdue.edu/open_access_dissertations/715

This document has been made available through Purdue e-Pubs, a service of the Purdue University Libraries. Please contact [email protected] for additional information. Graduate School Form

30 Updated ¡ ¢¡£ ¢¡¤ ¥

PURDUE UNIVERSITY GRADUATE SCHOOL Thesis/Dissertation Acceptance

This is to certify that the thesis/dissertation prepared

By Steeve Symithe

Entitled Present Day Plate Boundary Deformation in The Caribbean and Crustal Deformation On Southern Haiti.

For the degree of Doctor of Philosophy

Is approved by the final examining committee:

Christopher L. Andronicos

Chair Andrew M. Freed

Julie L. Elliott

Ayhan Irfanoglu

To the best of my knowledge and as understood by the student in the Thesis/Dissertation Agreement, Publication Delay, and Certification Disclaimer (Graduate School Form 32), this thesis/dissertation adheres to the provisions of Purdue University’s “Policy of Integrity in Research” and the use of copyright material.

Andrew M. Freed Approved by Major Professor(s):

Indrajeet Chaubey 04/21/2016 Approved by: Head of the Departmental Graduate Program Date

PRESENT DAY PLATE BOUNDARY DEFORMATION IN THE CARIBBEAN

AND CRUSTAL DEFORMATION ON SOUTHERN HAITI

A Dissertation

Submitted to the Faculty

of

Purdue University

by

Steeve J. Symithe

In Partial Fulfillment of the

Requirements for the Degree

of

Doctor of Philosophy

May 2016

Purdue University

West Lafayette, Indiana ii

This work is especially dedicated to all the lives lost during the January 12, 2010 Haiti earthquake and thoses whose lives have been also drastically affected by this tragic event. This thesis is my contribution to the ongoing effort of the Haitian community to understand and to mitigate the seismic hazards posed by existing geological structures in Haiti. I want to express also a feeling of profound gratitude to my family, especially my dear regreted mother Carmita Brignolle, my brother Evens Dossous for all his supports prior and also during this process, my aunt Anne Marie Brignolle for all the love that she had given me. A special thanks goes to my dear friends: Dieudonn´e Belto, Ronald Cad´emus, Natacha Dor´elien, Blanc Pierre–William, Chrisla Joseph, Roby Douilly, Cassandre Joseph, Kathiana Raymond and Brainly Eugene. You guys have been a great source of motivation during these last 5 years. I want to thank the people from the Earth, Atmospheric and Planetary Sciences, especially my labmates (Joshua Boscheli, Chen Chen) and all the people from the Business Office. A lot of thanks go to Dave Blair and the Grillot’s for the great memories that I have shared with them during my stay in the US. I also want to thank the colleagues from the Department of Geology in Ecole Normale Superieure de Paris (ENS) for sharing with me their work space during my visit in the ENS. My gratitude goes to the Dieudonn´e, Bien-aim´e and Desrameau families to have welcomed me in their homes during my long trips in Paris. iii

ACKNOWLEDGMENTS

This research is supported by a COCONet fellowship from UNAVCO to Steeve Symithe and the National Science Foundation awards EAR-0409487, EAR-RAPID- 1024990, and EAR-1045809 to professor Eric Calais. GPS data used in these projects were provided by the UNAVCO Facility with support from the U.S. National Science Foundation (NSF) and National Aeronautics and Space Administration (NASA) un- der NSF Cooperative Agreement EAR-0735156. I thank the IGS and its centers for providing open GNSS data and data products to the community. Some of the data were also collected from campaign measurements by agencies in the Dominican Republic (Direccion General de Mineria the Instituto Cartografico Militar, Colegio Dominicano de Ingenieros, Arquitectos y Agrimensores (CoDIA), Holasa S.A.) and in Haiti by the Bureau of Mines and Energy (BME). I want to thank all the individuals who participated in the data collection process, especially: Claude Preptit, Macly Jeannite, Frantz Saint-Preux and Daniel from the BME in Haiti. This work would never have been possible without the guidance and support of my advisor Andrew Freed. Thank you, not only for your contribution to this work but also for your moral support all along this process. I can never be grateful enough to Professor Eric Calais for his contributions in my formation but also for the guidance and all the supports that he have provided me during these 5 long years. I am also thankful to the members of my study committee for insight on the documents: Julie Elliot, Christopher Adronicos and Ayhan Irfanoglu. Finally, my recognitions also go to my former professor from the Facult´e Des Sciences (FDS) of the Universit´e d’Etat d’Haiti (UEH): Roberte Momplaisir, Raould Momplaisir, Edgard Etienne, Janin Jadotte, Dominique Boisson for their contribution in my solid education. iv

TABLE OF CONTENTS

Page LIST OF TABLES ...... vi LIST OF FIGURES ...... vii ABSTRACT ...... ix 1 Introduction ...... 1 1.1 Plate Boundary Deformation and Earthquakes ...... 1 1.1.1 Brief Tectonic Setting ...... 1 1.1.2 Large Earthquakes in the Caribbean ...... 2 1.2 Caribbean Plate Boundary Deformation: State-of-the-art ...... 3 1.3 The January 12, 2010 Haiti Earthquake ...... 5 2 Research Questions ...... 7 3 Current Block Motion and Strain Accumulation on Active Faults in the Caribbean ...... 8 3.1 Background ...... 9 3.1.1 Tectonic Setting ...... 9 3.1.2 Previous GPS-Based Models ...... 13 3.2 Data and Models ...... 16 3.2.1 GPS Data Analysis ...... 16 3.2.2 Model Setup ...... 18 3.2.3 Model Results ...... 20 3.3 Discussion ...... 25 3.3.1 Best Fit Model ...... 25 3.3.2 Low Coupling on the Lesser Antilles Subduction ...... 29 4 Present-day Shortening in Southern Haiti from GPS Measurements ... 34 4.1 Tectonic Setting ...... 35 4.2 GPS Data ...... 37 4.3 Elastic Model ...... 38 4.4 Discussion ...... 41 4.4.1 Present-day Tectonic Model ...... 41 4.4.2 Implication for Seismic Hazard ...... 42 5 Conclusion ...... 45 REFERENCES ...... 75 v

VITA ...... 89 vi

LIST OF TABLES

Table Page 1 χ2 variations amongst tested model ...... 71 2 Angular velocity estimates ...... 72 3 Slip deficit rates for major faults ...... 73 4 Estimated model parameters and associated uncertainties...... 74 vii

LIST OF FIGURES

Figure Page 1 Seismotectonic setting of the Caribbean region ...... 47 2 Current tectonic setting of the northeastern Caribbean ...... 48 3 GPS velocities ...... 49 4 Earthquake focal mechanisms and locations ...... 50 5 Model reduced χ2 ...... 51 6 Block geometry used in the models tested ...... 52 7 Total model χ2 as a function of model tested ...... 53 8 Best-fit model geometry ...... 54 9 observed, modeled and residuals velocities ...... 55 10 observed, modeled and residuals velocities in the Lesser Antilles .... 56 11 Euler poles for block pairs with their 95% confidence ellipse...... 57 12 Test of the consistency of Lesser Antilles GPS velocities ...... 58 13 Coupling ratio estimated along the Greater-Lesser Antilles subduction in- terface ...... 59 14 Predicted motion of the North American plate ...... 60 15 Sections across the Lesser and Greater Antilles Antilles subduction .. 61 16 Resolution tests ...... 62 17 More resolution tests ...... 63 18 Comparison of tectonic interpretations of the eastern termination of the Enriquillo fault system ...... 64 19 GPS velocities shown with respect to the Caribbean plate and to the North American plate ...... 65 20 comparison between the best-fit model and GPS observations ..... 66 21 Parameter estimates for strike-slip rate, dip-slip rate, locking depth, and locking depth ...... 67 viii

Figure Page 22 Ground motion scenarios illustrating the two end-member models dis- cussed in the text...... 68 23 SUPPLEMENTAL: Site selection...... 69 24 SUPPLEMENTAL: Comparison between GPS observations and model re- sults for a range of surface fault trace location and dip angles...... 70 ix

ABSTRACT

Symithe, Steeve J. PhD, Purdue University, May 2016. Present Day Plate Boundary Deformation in the Caribbean And Crustal Deformation on Southern Haiti . Major Professor: Andrew W. Freed.

The Caribbean plate and its boundaries with North and South America, marked by subduction and large intra-arc strike-slip faults, are a natural laboratory for the study of strain partitioning and interseismic plate coupling in relation to large earth- quakes. In this work, I use the available campaign and continuous GPS measurements in the Caribbean to derive a regional velocity field expressed in a consistent reference frame. I use this velocity field as input to a kinematic model where surface velocities result from the rotation of rigid blocks bounded by locked faults accumulating inter- seismic strain, while allowing for partial locking along the Lesser Antilles, Puerto Rico, and Hispaniola subduction. This improved GPS velocity field in the Lesser Antilles excludes more than 3 mm/yr of strain accumulation on the Lesser Antilles-Puerto Rico subduction plate interface, which appears essentially uncoupled. The transition from a coupled to an uncoupled subduction in the northeastern Caribbean coincides with a transition in the long-term geological behavior of the Caribbean plate margin from compressional (Hispaniola) to extensional (Puerto Rico and Lesser Antilles). Also in Haiti, the ∼3 M inhabitant capital region that was severely affected by the devastating M7.0, 2010 earthquake continues to expand at a fast rate. Accurate characterization of regional earthquake sources is key to inform urban development and construction practices through improved regional seismic hazard estimates. I also use this improved GPS data set and show that seismogenic strain accumulation in southern Haiti involves an overlooked component of shortening on a south-dipping reverse fault along the southern edge of the Cul-de-Sac basin in addition to the well- known component of left-lateral strike-slip motion. This tectonic model implies that x ground shaking may be twice that expected if the major fault was purely strike-slip, as assumed in the current seismic hazard map for the region. 1

1 INTRODUCTION

Subduction zones are known to be responsible for the largest earthquakes on Earth. Besides large events, they are also responsible for tsunamis with important destructive capabilities. Even very developed countries, such as: Japan, suffer from the hazards associated with subduction zones. Therefore, the northeastern Caribbean subduction poses a large threat to the less developed countries located in that region. While historically, there are few large events that can be associated with the Puerto-Rico and Lesser Antilles subduction interface, intraplate faulting in the over- riding plate are known to produce several devastating earthquakes in the past. The recent Mw7.0 January 12, 2010 earthquake in Haiti is an example of threat associated with regional faults in the Caribbean. The level of damage recorded during that event is also the proof that these countries are vulnerable to the existing regional seismic hazard. This implies that the understanding and quantification of the potential for regional geological structures to produce large earthquakes or tsunamis is crucial for sustainable development in the region. Furthermore, the availability of numerous geodetic measurements in the Caribbean in the proximity of active faults is also a good opportunity to investigate at a large scale the relation between the observed surface deformation and the strain accumu- lation along the faults.

1.1 Plate Boundary Deformation and Earthquakes

1.1.1 Brief Tectonic Setting

The Caribbean domain and Central America form a small lithospheric plate in- serted between North and South America (Figure 1). The Caribbean plate moves eastward relative to them at 18-20 mm/yr [DeMets et al., 2000]. This motion is ac- 2 commodated by two major quasi-east-west strike-slip fault zones along its northern boundary on either side of the Cayman trough and at its southern boundary along the Oca-El Pilar fault system. The relative plate motion implies oblique convergence and subduction of the Atlantic oceanic lithosphere under the Greater Antilles (His- paniola and Puerto Rico), transitioning to frontal subduction in the Lesser Antilles, then to pure strike-slip motion along the southern boundary of the Caribbean plate in South America. Plate boundary deformation at the Caribbean plate margin is accommodated by slip on a number of relatively well-identified major faults.

1.1.2 Large Earthquakes in the Caribbean

The Caribbean region has been the locus of many large earthquakes in the past. While most of these large events occurred at the western boundary of the Caribbean plate, a few large earthquakes have also been observed on the northeastern and south- ern parts of the plate. The two major strike-slip faults, Septentrional and Enriquillo Plantain Garden (EPGF), which cross the northern part of the region had been rup- tured at several occasions. For example, the Septentrional fault was responsible for the destructive 1842, M8.0, Cap Haitian earthquake in Haiti and the 1562, M7.7, Santiago earthquake in the Dominican Republic [Scherer, 1912, McCann, 2006, ten Brink et al., 2011]. The Mw7.1, 2010, Haiti earthquake ruptured one of the subsidiary faults of the EPGF system close to Port-au-Prince [Calais et al., 2010, Hayes et al., 2010b]. Also a number of historical earthquakes, possibly located on Enriquillo, had struck southern Hispaniola between 1701 and 1770 [Bakun et al., 2012b]. A series of M7.2–8.1 thrust earthquakes associated with the North Hispaniola fault had struck the northeastern Dominican Republic in 1943–1953 [Dolan and Wald, 1998]. More recently, a Mw 6.4, thrust earthquake in 2003 offshore the northern Do- minican Republic [Dolan and Bowman, 2004] also occurred on that portion of the northeastern plate boundary, which marks the oblique subduction of the Atlantic oceanic lithosphere under Puerto Rico [Sykes et al., 1982, Calais et al., 1992]. At the 3

northeastern extent of the Caribbean plate, a significant number of small to moder- ate earthquakes define the interface between the subducting slab and the overiding island arc along the Puerto Rico trench and most of the Lesser Antilles. Seismicity is however less prominent south of 15oN, coincident with the development of a thick acretionary prism fed by sediments shed from the South American continent [Le Pi- chon et al., 1990]. Although it currently appears aseismic, paleoseismic investigations have shown that the Central Range fault in Trinidad produced several large earth- quakes between 2710 and 550 yr B.P. [Prentice et al., 2010b]. The paleo-earthquake record of the El Pilar fault is also well established by trench excavations and his- torical records of destructive earthquakes [Audemard et al., 2000, Mendoza, 2000]. Earthquake focal mechanisms and other kinematic indicators show pure right-lateral strike-slip on vertical east-west trending planes [Audemard et al., 2005]. Focal mechanisms of these large earthquakes near Puerto Rico and the Lesser Antilles show that none of them can be related to the plate subduction interface (see Figure 1). While a few of the large events can be attributed to the plate interface of the North Hispaniola portion of the subduction, they seem to be located mostly in the intra-arc of the upper plate and/or in the deep portion of the slab.

1.2 Caribbean Plate Boundary Deformation: State-of-the-art

Several attempts have been made to quantify the kinematics of plate boundary deformation in the Caribbean, but only for specific plate boundary segments so far. Early on Dixon et al. [1998] used 6 early GPS stations in the Dominican Republic to show eastward motion of the Caribbean plate with respect to North America at a rate twice faster than predicted by the Nuvel-1A geologic model [DeMets et al., 1994]. This result was confirmed later by DeMets et al. [2000].Additional GPS measurements in the northeastern Caribbean allowed Jansma et al. [2000] to show the existence of a Puerto Rico–Virgin Islands block independent from the Caribbean plate, while Calais et al. [2002b] provided the first estimates of strain accumulation rates on 4 regional faults in Hispaniola and showed strain partitioning in an oblique collisional context. In the northern Lesser Antilles, Lopez et al. [2006] observed a systematic misfit between their GPS-derived Caribbean/North America relative plate motion and slip vectors of thrust earthquakes. They interpreted this observation as the result of strain partioning, and predicted 5 to 10 mm/yr of along-arc motion of a possible North Antilles arc sliver, consistent with the values later proposed by Feuillet et al. [2002] on the basis of the mapping of offshore active faults in the northern Lesser Antilles. Manaker et al. [2008b] used a combined GPS solution covering Hispaniola and Puerto Rico, with a few sites in the northernmost part of the Lesser Antilles to show that the data were consistent with a simple block model with strain accumulation on the major plate boundary faults but with largely uncoupled Puerto Rico and Lesser Antilles subduction interfaces. Following the 2010 Haiti earthquake, Calais et al. [2010] show, thanks to a new GPS data set in Haiti, evidence for shortening across Hispaniola, consistent with the transpressional nature of the 2010 Haiti event. Then, the combination of GPS velocities in Jamaica and Hispaniola allowed Benford et al. [2012c] to refine a regional kinematic model for the northern Caribbean by geodetically defining the boundary between the Gonave microplate [Mann et al., 1995b] and the Hispaniola block of Manaker et al. [2008b] through western Hispaniola. No kinematic model is yet available for the southern boundary of the Caribbean plate but several regional studies have already placed constraints on fault slip rates and locking depth, as noted above. P´erez et al. [2001] and Weber et al. [2001] showed that the Caribbean plate is currently moving due east with respect to South America at 20–22 mm/yr, with pure right-lateral strike-slip concentrated in a narrow region along the El Pilar fault in Venezuela, possibly experiencing aseismic slip [Jouanne et al., 2011]. In Colombia and northern Ecuador GPS measurements show that plate boundary deformation involve the Maracaibo and North Andes blocks [Trenkamp et al., 2002, White et al., 2003]. The North Andes block is currently moving north- ward with respect to the Caribbean plate, consistent with recent evidence for large 5 continental slivers along the south American margin from Peru to Ecuador [Nocquet et al., 2014]. Though much is known about the first order active tectonic features of the region thanks to seismotectonic mapping, paleoseismology, and geodetic mea- surements, we are still lacking a geodetically consistent GPS velocity field covering the entire Caribbean plate and its boundaries. The complexity of plate boundary deformation in the region and the paucity of GPS data to define a stable Caribbean plate require a large-scale approach in order to simultaneously estimate plate/block motions in a kinematically consistent manner.

1.3 The January 12, 2010 Haiti Earthquake

The Haiti January 12, 2010 Haiti earthquake occurred on a north dipping blind fault called the L´eogˆane fault [Calais et al., 2010, Symithe et al., 2013b]. This result was very surprising in the fact that previously published results have expected such possible earthquakes to occur either on the left–lateral Enriquillo Plantain Garden (See Figure 2), which is located near by the epicenter of this event or on the other major strike–slip fault (Septentrional) which crosses the northern part of the country [Manaker et al., 2008b]. Several slip distribution models have been published after this event with most of them supporting the existence of this blind fault responsible for the Mw 7.0 January 12 2010 catastrophic earthquake. While one of these models proposed by Hayes et al. [2010b] does not oppose the hypothesis that a structure as the L´eogˆane fault might be a key element during the 2010 rupture, they also proposed that part of the total energy released also comes from a segment of the Enriquillo fault as well as a south dipping structure located to the east of the L´eogˆane fault. Using temporary local seismic stations installed in Haiti after the earthquake by different research groups, Douilly et al. [2013] was able to relocate a series of aftershocks recorded during the first months following the earthquake. The spatial distribution of these aftershocks was used to illuminate the L´eogˆane fault geometry [Douilly et al., 2013] and Symithe et al. [2013b] used 6 that new geometry to improve the slip distribution model for that earthquake and to determine the associated stress change on surrounding faults. Symithe et al. [2013b] found a significant Coulomb stress increase along the Enriquilo fault on the western and eastern extents of the rupture area. While they also observed important decrease of Coulomb stress (> 0.08 MPa) at the bottom of the Enriquillo section adjacent to the L´eogˆane fault, they also found significant positive Coulomb stress change (> 0.08 MPa) at the top of this section of the fault. More importantly, they found that the production of aftershocks on the eastern segment of the Trois-Baies fault near the Gonave Island is associated with important Coulomb stress increase (∼ 0.06 MPa) on this section of the fault [Symithe et al., 2013b]. These results proved that the 2010 Haiti earthquake although a moderate earthquake had significantly influenced the seismic hazards in southern Haiti. Therefore, the vulnerability of poorly engineered constructions in very populated cities near the Enriquillo fault line along the southern Peninsula makes it of great importance to study in more detail the crustal deformation in southern Haiti. 7

2 RESEARCH QUESTIONS

Though much is known about the first order active tectonic features of the re- gion thanks to seismotectonic mapping, paleoseismology, and geodetic measurements, we are still lacking a geodetically consistent GPS velocity field covering the entire Caribbean plate and its boundaries. The complexity of plate boundary deformation in the region and the paucity of GPS data to define a stable Caribbean plate require a large-scale approach in order to simultaneously estimate plate/block motions in a kinematically consistent manner. The quantification of the threat posed by the long northeastern Caribbean subduction to the region is crucial for the less developed countries located in that region. There is some important questions that need to be addressed in this region. Is the Lesser Antilles subduction capable of large events such as: Mw8.0 earthquake? What is the recurrence time for those large events? Is the existing historical earthquake pattern consistent with the interseismic deforma- tion? The availability of a more dense GPS network in southern Port–au–Prince, Haiti with measurements periodically performed on these sites now also offers the possibility to study with greater detail the deformation pattern observed in this part of the Peninsula. Therefore, in the following I propose several projects that would address different aspects of these important geophysical problems. First, I process all available GPS data for the Caribbean to constrain a block model for the plate to determine the kine- matic motions of the different blocks while we solve for strain accumulation on the regional faults and for plate coupling along the northeastern subduction. Then, I se- lect a sub-sample of the kinematic solution in Haiti to investigate crustal deformation that is occurring along the Cul-de-Sac plain in southern Port–au–Prince. 8

3 CURRENT BLOCK MOTION AND STRAIN ACCUMULATION ON ACTIVE FAULTS IN THE CARIBBEAN

Most of the seismic energy of our planet is released at subduction zones by earthquakes that occur either at the plate interface or on active faults in the overriding plate. It has long been thought that interplate coupling in such contexts – hence their seismogenic potential – depended for a large part on the age of the subducting plate [Ruff and Kanamori, 1980]. Recent large subduction earthquakes have shaken this paradigm, to the point that some now claim that all subduction have the capacity of generating mega-earthquakes regardless of the age of the subducting crust [McCaffrey et al., 2008]. The issue has therefore refocused on spatial and temporal variations of interplate coupling and how these may relate to asperities and barriers on the plate interface and to the deformation of the overriding plate [e.g., Chlieh et al., 2011, Wallace et al., 2012a]. For instance, recent results along the Chilean subduction indicate that lateral variations of interseismic coupling are correlated with the rupture areas of the Maule (2011, Mw8.8) and Valdivia (1960, Mw9.5) earthquakes, and with the frictional prop- erties of the plate contact derived from fore-arc morphology [Cubas et al., 2013]. Similarly, the Colombian subduction shows lateral variations of interseismic coupling correlated with large historical earthquakes and with the segmentation of the upper plate into continental slivers translating with respect to both the Nazca and South American plates [Nocquet et al., 2014]. The subduction of the north American plate under the Caribbean plate (Fig. 1) also shows lateral variations of interseismic coupling correlated with a segmentation of the tectonic regime along the arc [Manaker et al., 2008b]. Over a short distance, the plate boundary evolves from a frontal subduction (Lesser Antilles) with arc- parallel extension, to a very oblique subduction (Puerto Rico) with little deformation 9 of the arc, to a subduction-collision (Hispaniola) with large strike-slip faults in the overriding plate and an active mountain range culminating at 3300 m (Pico Duarte, Dominican Republic). Interestingly, the large historical earthquakes that struck the Lesser Antilles subduction may not be interplate events [Stein et al., 1982], while those appear to concentrate in the north at the Hispaniola–Bahamas collision. The hypothesis has been put forward that this geological segmentation was corre- lated with interseismic coupling, reflecting the ability of the plate contact to transfer shear stress to the overriding plate [Mann et al., 2002]. The segmentation would then also reflect the seismogenic capacity of the plate interface and possibly of the intra- arc faults. Here we test this hypothesis by jointly estimating interplate coupling and block kinematics in the whole Caribbean region east of 85◦W. We also aim at provid- ing a first-order kinematic description of surface deformation across the study area that can inform seismc hazard assessments and be compared with paleoseismological information. We do so using a new combined velocity field derived from campaign and continuous measurements at 300 sites, a 5 times increase since Manaker et al. [2008b]. In particular, we include a new data set that significantly improves resolution in Puerto Rico and the Lesser Antilles and address the southern part of the Lesser Antilles subduction and the Caribbean plate boundary in South America.

3.1 Background

3.1.1 Tectonic Setting

The Caribbean domain and Central America form a small lithospheric plate in- serted between North and South America (Figure 1). While the North and South American plates show little relative motion [Patriat et al., 2011], the Caribbean plate moves eastward relative to them at 18-20 mm/yr [DeMets et al., 2000]. This dis- placement is accommodated by two major east-north-east strike-slip fault systems along its northern boundary on either sides of the Cayman trough and at its south- ern boundary along the Oca-El Pilar fault system. The relative plate motion implies 10 oblique convergence and subduction of the Atlantic oceanic lithosphere under the Greater Antilles (Hispaniola and Puerto Rico), transitioning to frontal subduction in the Lesser Antilles, then to pure strike-slip motion along the southern boundary of the Caribbean plate in South America. Plate boundary deformation at the Caribbean plate margins is accommodated by slip on a number of relatively well-identified major faults. We briefly describe these structures thereafter as they will serve to define the geometry of our kinematic model. At the northeastern edge of the studied area (Figure 1), the Oriente Fault bounds the Cayman trough, a 45-50 Ma oceanic pull-apart basin [Rosencrantz et al., 1988], to the north. Earthquake focal mechanisms show pure left-lateral strike-slip motion from the mid-Cayman spreading center to the southern Cuban margin [Perrot et al., 1997] where the fault trace slightly changes direction to become transpressional [Calais and de Lepinay, 1991], with earthquake focal mechanisms showing a combination of thrust and strike-slip faulting [Van Dusen and Doser, 2000]. To the east, the Oriente fault pursues its course through the Windward Passage, along the northern Haitian coast as the “Septentrional fault” [Calais and de L´epinay, 1990], then further east on land through the Cibao Valley of the Dominican Republic Calais and Mercier de L´epinay [1992] where paleoseimological studies indicate a Holocene slip rate of 9±3mm/yr [Prentice et al., 2003], in agreement with GPS estimates [Calais et al., 2002b]. The Septentrional fault extends offshore to the east as far as the Mona rift, one of the active structures marking the extensional boundary between Hispaniola and Puerto Rico through the Mona Passage [Grindlay et al., 1997, Gestel et al., 1998]. The con- tinuation of the Septentrional fault east of the Mona Passage is less clear (Figure 3), it may gradually merge with the Puerto Rico trench through a series of small faults including the Bunce and Bowin faults [Brink et al., 2004, Grindlay et al., 2005a]. The Septentrional fault was the locus of significant historical earthquakes, among which the destructive 1842, M8.0, Cap Haitian earthquake in Haiti and the 1562, M7.7, Santiago earthquake in the Dominican Republic [Scherer, 1912, McCann, 2006, ten Brink et al., 2011]. 11

The Cayman trough is bounded to the south by a second series of left-lateral strike-slip faults, starting at the mid-Cayman spreading center with the purely strike- slip Walton fault [Rosencrantz and Mann, 1991]. The Walton fault connects through Jamaica with the Enriquillo-Plantain Garden Fault (EPGF) through a series of re- lays whose geometry remains debated [Benford et al., 2012a]. The EPGF continues offshore as a purely strike-slip fault east of Jamaica then on land in southern Haiti where it is marked by a well-defined narrow valley continuing eastward just north of Port-au-Prince and into the Enriquillo Valley in the Dominican Republic [Mann et al., 1995b]. Further east, the EPGF appears to merge with thrust faults at the western termination of the Muertos Trough [Mauffret and Leroy, 1999]. A number of historical earthquakes, possibly located on that fault, struck southern Hispaniola between 1701 and 1770 [Bakun et al., 2012b]. The Mw7.1, 2010, Haiti earthquake ruptured one of the subsidiary faults of the EPGF system close to Port-au-Prince in a transpressional context [Calais et al., 2010, Hayes et al., 2010b]. No geological estimate of slip rate is yet available for that fault. The Muertos Trough marks the front of a large accretionary prism that has devel- oped along the southern margin of the Dominican Republic and Puerto Rico [Byrne et al., 1985, Granja Bru˜na et al., 2009]. It is associated with some seismicity, in particular a Ms6.7 thrust faulting event in 1984, that may mark the contact be- tween a downgoing slab of Caribbean lithosphere under the Greater Antilles island arc [McCann and Sykes, 1984]. Evidence for sediment deformation and active fault- ing become more tenuous eastward toward the Anegada Passage, a system of basins apparently bounded by active strike-slip and normal faults that separate Puerto Rico and the Virgin Islands from the Lesser Antilles to the south [Masson and Scanlon, 1991, Jany et al., 1990a]. The connection between the Anegada faults and the Puerto Rico/Lesser Antilles subduction is unclear. We will highly simplify its geometry in our model. At its southern termination, the lesser Antilles subduction bends northern margin of the Greater Antilles island arc marks the contact between the Caribbean plate 12

and the obliquely subducting oceanic lithosphere of the North American plate. It is marked, in the west, by a narrow basin between the Bahamas carbonate platform and the island of Hispaniola, bounded along its southern edge by active compressional features revealed by side-scan sonar and seismic reflection data [Dillon et al., 1996, Dolan and Wald, 1998] that define the North Hispaniola Fault. Active shortening perpendicular to that fault is shown by the series of M7.2–8.1 thrust earthquakes that struck the northeastern Dominican Republic in 1943–1953 [Dolan and Wald, 1998] and by the more recent Mw6.4, 2003 thrust earthquake of 2003 offshore the northern Dominican Republic [Dolan and Bowman, 2004]. The North Hispaniola basin and fault are continuous to the east with the Puerto Rico Trench, deepest point of the Atlantic ocean (>8 km) and largest negative free-air gravity anomaly on Earth (−400 mGals), which marks the oblique subduction of the Atlantic oceanic lithosphere under Puerto Rico [Sykes et al., 1982, Calais et al., 1992, Grindlay et al., 2005a]. To the east, the Puerto Rico Trench curves around the Virgin islands but remains continuous with the Lesser Antilles trench further south. Contrary to Puerto Rico where plate motion is highly oblique to the trench direction, in the Lesser Antilles plate motion becomes perpendicular to the trench. A significant number of small to moderate earthquakes define the interface between the subducting slab and the over- riding island arc along most of the Lesser Antilles subduction. Seismicity is however less prominent south of 15◦N, coincident with the development of a thick accretionary prism fed by sediments shed from the South American continent [Le Pichon et al., 1990]. North of that latitude, the arc is cross-cut by a series of normal faults [Feuillet et al., 2002], the Anegada fault zone possibly representing the northernmost of that extensional system. At its southern termination, the lesser Antilles subduction bends around towards South America and connects with a mostly strike-slip plate boundary. GPS stud- ieshaveshownthat∼65% of the relative motion between the Caribbean and the South American plates was accommodated by strike-slip faulting on Trinidads’ Cen- 13 tral Range fault, a major structure connecting with the southern extent of the Lesser Antilles near Tobago and crossing the central part of Trinidad. Although it currently appears aseismic, paleoseismic investigations have shown that the Central Range fault in Trinidad produced several large earthquakes between 2710 and 550 yr B.P. [Pren- tice et al., 2010b]. Additional motion most likely occurs on the Los Bajos Fault of southern Trinidad and other offshore faults south of the island [Weber et al., 2010, Soto et al., 2007]. To the east, the main plate boundary fault continues offshore with a transtensional relay in the Gulf of Paria [Babb and Mann, 1999], then connects with the El Pilar–San Sebastian fault system which marks the Caribbean-South America in northern Venezuela [Mann et al., 1990]. An unknown quantity of north-south shortening is taken up by compressional, accretionary prism-like structures offshore Venezuela and Colombia that form the “South Caribbean Deformed Belt” [Kroehler et al., 2011].

3.1.2 Previous GPS-Based Models

Several attempts have been made to quantify the kinematics of plate boundary deformation in the Caribbean, but only for specific plate boundary segments so far. Early on Dixon et al. [1998] used 6 early GPS stations in the Dominican Republic to show eastward motion of the Caribbean plate with respect to North America at a rate twice faster than predicted by the Nuvel-1A geologic model [DeMets et al., 1994], with slip accommodated on the Septentrional, Enriquillo, and North Hispan- iola faults at rates that were then quite uncertain. The present-day kinematics of the Caribbean plate was later quantified in more detail by DeMets et al. [2000], who used 4 GPS stations in the plate interior to compute its first geodetically-derived angular velocity. They confirmed that the Caribbean plate motion was significantly faster than predicted by Nuvel-1A, which was later shown to result from a global bias introduced by earthquake slip vectors at obliquely convergent plate margins [DeMets and Dixon, 1999]. Additional GPS measurements in the northeastern Caribbean al- 14 lowed Jansma et al. [2000] to show the existence of a Puerto Rico–Virgin Islands block independent from the Caribbean plate, while Calais et al. [2002b] provided the first estimates of strain accumulation rates on regional faults in Hispaniola and showed strain partitioning in an oblique collisional context. In the northern Lesser Antilles, Lopez et al. [2006] observed a systematic misfit between their GPS-derived Caribbean/North America relative plate motion and slip vectors of thrust earth- quakes. They interpreted this observation – dependent on the definition chosen for the Caribbean frame – as indicative of a Northern Antilles block distinct from the Caribbean plate and moving with respect to it at rates up to 5 mm/yr. This would be consistent with the slip partitioning model proposed by Feuillet et al. [2002, 2010] on the basis of fault mapping and shallow focal mechanisms, which predicts 5 to 10 mm/yr of left-lateral shear along en ´echelon normal faults west of the islands and distributed trench parallel extension along the northern half of the Lesser Antilles. The first regional-scale kinematic model for the Caribbean was proposed by [Man- aker et al., 2008b] using a combined GPS solution covering Hispaniola and Puerto Rico, with a few sites in the northernmost part of the Lesser Antilles. They showed that the data was consistent with a simple block model with strain accumulation on the major plate boundary faults but with largely uncoupled Porto-Rico and Lesser Antilles subduction interfaces. They observed that the transition from a coupled to uncoupled plate interface coincided with the onset of oblique collision between the Greater Antilles island arc and the Bahamas platform [Grindlay et al., 2005b]. The number of reliable GPS sites available in the Lesser Antilles was however small and the predicted strain accumulation highly uncertain, leading for instance to a wide range of possible earthquake and tsunami scenari [Hayes et al., 2014]. Manaker et al. [2008b] results were updated locally following the 2010 Haiti earth- quake [Calais et al., 2010]. A surprising result then, thanks to a new GPS data set in Haiti, was evidence for shortening across Hispaniola, consistent with the trans- pressional nature of the 2010 Haiti event. Benford et al. [2012a] then used a dense campaign GPS network in Jamaica and argued for ∼7 mm/yr of strike-slip motion 15 across the island accommodated by the central Jamaican fault system on land (Rio Minho-Crawle River fault zone) and about 2 mm/yr of convergence offshore to the south of the island on unmapped faults on the northern Nicaragua rise. The combi- nation of GPS velocities in Jamaica and Hispaniola allowed Benford et al. [2012c] to refine a regional kinematic model for the northern Caribbean by geodetically defining the boundary between the Gonave microplate [Mann et al., 1995b] and the Hispan- iola block of Manaker et al. [2008b] through western Hispaniola. They provided new estimates for the angular velocities of the Gonave, Hispaniola, north Hispaniola and Puerto Rico microplates. No kinematic model is yet available for the southern boundary of the Caribbean plate but several regional studies have already placed constraints on fault slip rates and locking depth, as noted above. P´erez et al. [2001] and Weber et al. [2001] showed that the Caribbean plate is currently moving due east with respect to South America at 20–22 mm/yr, with pure right-lateral strike-slip concentrated in a narrow region along the El Pilar fault in Venezuela, possibly experiencing aseismic slip [Jouanne et al., 2011]. In Colombia and northern Ecuador GPS measurements show that plate boundary deformation involve the Maracaibo and North Andes blocks [Trenkamp et al., 2002, White et al., 2003]. The North Andes block is currently moving north- ward with respect to the Caribbean plate, consistent with recent evidence for large continental slivers along the south American margin from Peru to Ecuador [Nocquet et al., 2014]. Though some is known – at least locally – on the first order active tectonic features of the region thanks to seismotectonic mapping, paleoseismology, and geodetic mea- surements, we are still lacking a geodetically consistent GPS velocity field covering the entire Caribbean plate and its boundaries. The complexity of plate boundary de- formation in the region, the limited land areas, and the small number of GPS sites on the stable Caribbean plate require a large-scale approach in order to simultaneously estimate plate/block motions and plate boundary deformation in a kinematically con- sistent manner. The proximity of many GPS sites to active faults imposes that strain 16 accumulation be accounted for, with the possibility of laterally variable coupling on subduction interfaces. In the following, we present a large-scale GPS velocity field covering the entire study, which we interpret in terms of rigid block motions and strain accumulation on locked or partially locked faults.

3.2 Data and Models

3.2.1 GPS Data Analysis

We processed data from 342 episodic and continuously recording GPS sites in the study area from 1994 onward using the GAMIT–GLOBK software package [Herring et al., 2010b]. The data is usually openly available on public databases or upon request from private ftp sites. We process double-difference phase measurements to solve for station coordinates, satellite state vectors, seven daily tropospheric delay parameters per site, two parameters for horizontal tropospheric gradients, and phase ambiguities using final satellite orbits from the International GNSS Service (IGS), Earth orientation parameters from the International Earth Rotation Service (IERS), applying corrections for solid-earth tides, polar tides, time-variable ocean loading following the IERS conventions 2010 [Petit and Luzum, 2010] and antenna phase- centre variations using the latest IGS tables [Schmid et al., 2007], and subsequent updates. For the later years when the number of stations increases significantly, we process the data in subnetworks of up to 40 sites, including twelve reference sites from the International GNSS Service (IGS) common to all subnetworks (BDOS, BOGT, BRAZ, BRMU, CRO1, FORT, KOUR, MAS1, MDO1, NLIB, SCUB, WES2), and two additional common sites between subnetworks. We identify discontinuities or offsets caused by changes in the instrumentation or to earthquakes by visually inspecting daily position time series. We account for these discontinuities in the velocity solution by allowing for a three-component offset while equating velocities before and after the offset. We remove some sections of time series with non-linear deviation from the background trend due to postseismic deformation. 17

This is for instance the case of 2009, M7.3, Swan Islands earthquake [Graham et al., 2012], which caused significant co- and post-seismic displacement at several GPS sites used here. After this cleaning step, we combine the regional loosely-constrained daily solu- tions with global daily solutions for the whole IGS network available from the Mas- sachusetts Institute of Technology IGS Data Analysis Center into weekly position solutions. This helps improve signal resolution over the noise level and allows us to optimally tie our solution to the International Terrestrial Reference Frame (ITRF) [Altamimi et al., 2011]. We finally combine these weekly solutions into a single posi- tion/velocity solution using GLOBK [Herring et al., 2010a], which we tie to the ITRF by minimizing position and velocity deviations from a set of globally-defined IGS ref- erence sites common to our solution via a 12-parameter Helmert transform (scale and scale change are not estimated). We downweigh the variance of the height coordi- nates by a factor of 10 because of the reduced precision of the vertical component in standard GPS solutions. We estimate time-correlated noise at continuous sites using the First-Order Gauss-Markov Extrapolation (FOGMEx) algorithm of Herring [2003, see also Reilinger et al. 2006a] in order to obtain realistic velocity uncertainties. For √ episodic sites, we include a 2 mm/ yr random walk component to account for colored noise in velocity uncertainties. The resulting solution is a set of coordinates and velocities expressed in ITRF2008, which can then be used to model the regional kinematics. Velocity uncertainties vary as a function of observation time span and reach 0.5 mm/yr for the oldest stations that have close to 20 years of continuous data (e.g., CRO1, BRMU, BOGT, SCUB). A number of important observations are readily apparent on the velocity field shown on Figure 3 with respect to the North American (top) and Caribbean (bottom) plates. The Caribbean plate moves in an ENE direction with respect to North America at about 19 mm/yr, while North and South America are converging toward each other at a very slow rate (<2 mm/yr) increasing from east to west. This implies frontal convergence along most of the Lesser Antilles arc, transitioning to oblique subduction 18 to the north in the Greater Antilles and pure strike-slip along the boundary with South America. In a Caribbean frame, GPS sites on the Nicaragua Rise, the Venezuelan basin, and the Lesser Antilles show velocities that are within error of zero, with a weighted RMS of residuals of 0.5 mm/yr and a maximum residual velocity of 1.5 mm/yr. We also find velocities close to zero at GPS sites in Puerto Rico and the Virgin Islands, except for the recently installed CN03 site on Virgin Gorda, with however a slight but systematic pattern of counterclockwise rotation. We observe about 3 mm/yr of extension across the Mona Passage, consistent with findings from earlier GPS measurements [Jansma et al., 2000] and evidence for extensional faulting between Puerto Rico and Hispaniola [Gestel et al., 1998]. West of the Mona Passage in Hispaniola velocities show a N- S gradient consistent with strain accumulation on ∼E-W directed, strike-slip plate boundary faults. The velocity difference is about 10 mm/yr across the Septentrional fault in the northern Dominican Republic and 6 mm/yr across the Enriquillo fault in southern Haiti. We also observe a N-S gradient in velocities across the central and western parts of Hispaniola with up to 4 mm/yr of shortening. The velocity gradient across Jamaica is consistent with strain accumulation on east-west directed faults across the island [Benford et al., 2012c]. Along the southern boundary of the Caribbean plate, most of the Caribbean-South America relative motion is taken up by the El Pilar fault, then by the Bocomo fault system further to the west, while the Oca fault appears to accommodate slow relative motion between the Maracaibo block and the Caribbean plate.

3.2.2 Model Setup

We now model the observed GPS velocities by assuming that they represent the sum of rigid block rotations and strain accumulation on faults locked to a given depth in an elastic half-space. This approach, commonly called “kinematic block modeling”, is widely used to analyse regional GPS velocity fields and quantify plate 19 motions and slip deficit on locked or partially-locked active faults [e.g., McCaffrey, 2002, Meade et al., 2002, Reilinger et al., 2006b, Saria et al., 2014]. Here we use the modeling approach and associated code “blocks” described in Meade and Loveless [2009] where active faults are treated as freely slipping at the full relative plate motion below a given seismogenic depth above which there are either locked or partially slipping. In both cases the fault elastic contributions are calculated using the classical back-slip approach [Savage, 1983b] with Okada Green’s functions in an elastic half- space [Okada, 1985]. Partially locked faults (in this case the subduction interface) are discretized using triangular elements. Continuity between fault elements and regularization of the solution are ensured by imposing a smoothing constraint that minimizes the Laplacian of the slip estimated along the fault plane. This method leads to a set of linear equations which allow for a well-defined solution. It also allows the plate interface to slip not only in the direction of the rigid differential block motion but also in the opposite direction to account for coseismic slip that could be associated with documented earthquakes, slow slip events, or post-seismic deformation. A coupling coefficient can be calculated a posteriori by dividing the full relative plate motion by the estimated slip rate on each fault element. Here we use the major known active faults as plate and block boundaries, following previous authors [Manaker et al., 2008b, Benford et al., 2012c,a]. We treat all faults as locked to a given depth, except for the Hispaniola-Puerto Rico-Lesser Antilles subduction which we discretize with triangular elements on which we solve for slip. We determined the optimal fault locking depth by running a series of models with locking depth ranging from 5 to 25 km. We find the lowest χ2 for a locking depth of 14 km (Figure 5), which we then impose in all model runs. This locking depth is consistent with the seismogenic depth on intra-arc faults along the Caribbean margin and other similar tectonic settings [Sanchez-Rojas and Palma, 2014, Miller et al., 2009]. We did not allow for the shallower locking depths proposed for the El Pilar fault [Jouanne et al., 2011, Weber et al., 2010] because our data set does not include sufficiently dense measurements in these regions to be sensitive to aseismic slip. 20

We derive the geometry of the subduction interface using its surface trace along the Lesser Antilles, Puerto Rico, and North Hispaniola trenches and the depth- distribution of thrust earthquakes derived from the Engdahl et al. [1998] database (Figure 4). We limit the downdip end of the partially locked subduction interface using theoretical isotherms of forearc thermal structure between Guadeloupe and Barbados [Gutscher et al., 2010, Manga et al., 2012] which place the 350◦C isotherm, considered as the downdip limit of purely seismogenic behavior, at a depth of about 40 km [Hyndman, 2007]. We find that shallow seismicity generally defines a plate interface of fairly constant and low-angle dip down to 40-60 km, below which the dip angle increases as the subducting slab bends [Laigle et al., 2013], as apparent on Figure 4. We therefore use a single dip angle for each subduction segment. From the depth distribution of thrust earthquakes we determine a plate interface dip of 20◦ north of Hispaniola, 25◦ along the Puerto Rico trench, and 16◦ for the Lesser Antilles. This latter value is consistent with seismic refraction data [Kopp et al., 2011, Laigle et al., 2013] at close distance to the trench. Therefore, we model the subduction as a curved fault with 20 segments of constant dip and a constant locking depth of 40 km (Figure 4). Assuming a Moho depth of 25-30 km under the Lesser Antilles as imaged by Kopp et al. [2011] between Guadeloupe and Dominica, the 40 km down-dip end of the plate interface in the models is located just below the tip of the mantle wedge, and allows for “deep flat-thrust” earthquakes as identified by Laigle et al. [2013].

3.2.3 Model Results

Our objective is to determine the best-fit angular velocity for rigid block motions along the boundaries of the Caribbean plate to estimate the rate of interseismic strain accumulation on the major plate boundary faults. To do so, we ran a series of models with various block geometries starting with a single Caribbean plate and progressively add more complexity to the model geometry by fragmenting the plate boundaries into blocks (Figure 6). We assess the improvement obtained by increasing 21

the model complexity (more blocks, i.e., increased degree of freedom) by testing the significance of the decrease in χ2 from a model with fewer blocks to a model with more blocks using the F-ratio statistics (e.g., Stein and Gordon, 1984) given by: 2 − 2 − (χp1 χp2 )/(p1 p2) F = 2 (3.1) χp2 /p2

2 2 where χp1 and χp2 are the chi-square statistics of two models with p1 and p2 degrees of freedom, respectively. We compare this experimental F-ratio to the expected value

of a F (p1 − p2,p1) distribution for a given risk level α% (or a 100 − α% confidence level) that the null hypothesis (the decrease in χ2 is not significant) can be rejected. We set the acceptable significance level to 99%, i.e., a probability of rejection less than 1%. Figure 7 shows the variation in χ2 and its significance level from one model to the next (see also Table 1). Model 1, with a single unfragmented Caribbean plate, naturally results in the largest χ2 (Table 1) and residuals, in particular in Colombia, Hispaniola, and Puerto Rico. Model 2 adds the northern Caribbean blocks defined by Manaker et al. [2008b]. As expected, the improvement is significant well above the 99% confidence level, with much smaller residuals in Puerto Rico and Hispaniola, though still significant ones (3-4 mm/yr) in Jamaica. Model 3 further fragments the southern margin of the Caribbean plate with three blocks, Maracaibo, North Andes, and North Venezuela (Figure 6). The improvement is again significant at the 99% confidence level, as expected given the known regional tectonics, with residual velocities for these blocks less than 1.5 mm/yr. Model 4 splits the Hispaniola block into two microplates, with a Gonave microplate extending from the mid-Cayman spreading center to the Neiba-Matheux thrust front (NMF on Figure6) across central Hispaniola following Benford et al. [2012c]. Velocity residuals decrease significantly in northern Jamaica and Hispaniola, in particular in the northern part of Haiti where they reach 2-3 mm/yr. However, this model pre- dicts a large (up to 10 mm/yr) slip rate on the Neiba-Matheux faults, inconsistent with the lack of significant historical earthquakes on this structure and the subdued 22 geomorphic expression of active deformation compared to the Enriquillo and Septen- trional faults [Mann et al., 1995b, Pubellier et al., 2000b]. As shown by Benford et al. [2012c] with a more sparser GPS data set, the location of the eastern boundary of the Gonave microplate is uncertain and could also be a broad zone of deformation across Hispaniola. We tested several locations for that boundary and obtained the lowest reduced χ2 (2.047) for a trace that is coincidental with the Plateau Central-San Juan Valley area across Haiti and the Dominican Republic (perhaps the Peralta – Rio Ocoa belt of Heubeck et al. [1991]). In this configuration (model 5), residual velocities are < 1 mm/yr for all GPS sites in Hispaniola. This location of the block boundary, with predicted slip rates < 4 mm/yr, is consistent with historical events in 1761, which caused severe damage in the Neiba-San Juan area, and 1911 (M6.9?), which destroyed the cathedral of San Juan and was strongly felt in Hinche and Grande Rivi`ere in Haiti [Scherer, 1912]. We then test a scenario (model 6) that further splits the northeastern Caribbean plate margin with an additional North Lesser Antilles block, following [Lopez et al., 2006]. We bound the block to the north by the Anegada passage fault, to the west by the strike-slip back-arc fault system described by [Feuillet et al., 2002]. We set its southern boundary to the latitude of Dominica which coincides with the intersec- tion of the Lesser Antilles subduction with the North America–South America plate boundary [Pichot et al., 2012]. The reduced χ2 improves slightly but with a confidence level less than 99% compared to a model that does not include that block (model 5, Table 1). Residual velocities are essentially unchanged compared to model 5. They are within observation errors for all GPS sites in the Lesser Antilles, and do not show any systematic pattern. This therefore indicates that the existence of a North Lesser Antilles block separate from the Caribbean plate is not required by the GPS data. We further test this result by computing a series of angular velocities for the Caribbean plate starting with the four sites used by DeMets et al. [2000] (SANA, ROJO, CRO1, AVES), then adding one by one the best-determined GPS sites in the Lesser Antilles (velocity uncertainty <2.5 mm/yr). We use the F-test described 23

above to determine whether the null hypothesis – a model with or without a given site are similar – can be confidently rejected. We choose a conservative 99% confidence here. If the null hypothesis can be confidently rejected, then the site velocity is not consistent with the rigid plate rotation. We find that the null hypothesis can be confidently rejected for only 3 of the 42 reliable GPS sites in the Lesser Antilles (Figure 12). Two of them (sites SOUX and PSA1), less than a kilometer from each other, are located near the top of the Soufri`ere volcano of Guadeloupe in an area affected by non-tectonic deformation related to local hydrothermal processes and slope instabilities. The third one, site MPCH, was measured in survey mode only and has a short and discontinuous time series. We conclude that the GPS data is therefore consistent with a Lesser Antilles arc that moves coherently with the rest of the Caribbean plate, at the uncertainty level of the GPS errors (0.6 mm/yr WRMS). Model 5 predicts <1.2 mm/yr motion on the Anegada Passage fault system (Virgin Island basin, Anegada Passage s.s, and Sombrero basin [Jany et al., 1990b]) and only up to 1.5 mm/yr of shortening across the eastern Muertos trough south of Puerto Rico. We therefore tested a model that merges the Puerto Rico and North Lesser Antilles blocks into a single microplate distinct from the Caribbean plate (model 7). We find a similar reduced χ2 as in model 5, with insignificant slip (<0.1 mm/yr) on the Caribbean-Northern Lesser Antilles boundary, consistent with the lack of relative motion resolvable by the data between the Lesser Antilles and the Caribbean plate. We further test a model where the Puerto Rico and the North Lesser Antilles blocks are both part of the Caribbean plate (model 8). We find that the decrease in χ2 from that model to one where the Puerto Rico block is separate from the Caribbean plate (model 5) is significant at the 82% confidence level only. We therefore conclude that the GPS data do not require a North Lesser Antilles block and are only marginally able to distinguish a Puerto Rico block separate from the Caribbean plate. This is already visible on Figure 3 and consistent with the very low slip rates predicted by model 5 along the eastern Muertos and Anegada Passage faults, close the GPS velocity uncertainties. However, because model 5 is 24 still statistically superior and because active deformation is documented – though at an unknown rate – on the eastern Muertos and Anegada Passage faults, we will keep it as our best-fit model thus far. We then turn our attention to Jamaica, starting with model 5 which includes a single fault through central Jamaica, shows residuals within measurement errors north of that fault but large westward residuals south of it. Residual velocities at continuous GPS sites CN10 and CN11 (or their corresponding campaign sites MCAY and PEDR) offshore Jamaica on the Nicaragua Rise are however close to zero, consistent with the Caribbean plate motion. This indicates that the active fault responsible for residual velocities in southern Jamaica must lie somewhere between these sites and the island. We therefore replaced the central Jamaica fault by a southern Jamaica fault following the South Coastal and Aeolus Valley faults also tested by Benford et al. [2012c] as the boundary between the Gonave microplate and the Caribbean plate (model 9). This results in a much improved fit across Jamaica while residual velocities at sites CN10 and CN11 remains consistent with the Caribbean plate. We also tested the preferred block configuration of Benford et al. [2012c] (model 10) with a Nicaragua Rise block carrying sites CN10/CN11 (MCAY/PEDR) and bounded to the north by a central Jamaica fault. This model predicts small slip rates (< 1 mm/yr) on the central Jamaica fault and significant ones (up to 6 mm/yr) along the west, east, and south boundaries of that block. This is difficult to justify in the absence of significant offshore geological features capable of accommodating this large amount of displacement. The best-fit model therefore emphasizes fault activity in the southern part of the island, consistent with the location of the largest historical earthquakes to strike Jamaica in 1692 and 1907 [Wiggins-Grandison, 2004]. However, given the uncertainties in the GPS velocities, this result does not preclude the existence of other seismogenic faults through and around Jamaica accommodating part of the Gonave/Caribbean relative motion. Finally, we evaluate with model 11 whether the data require splitting the Caribbean plate into two subplates, for instance along the eastern edge of the Beata Ridge as 25

proposed by Leroy and Mauffret [1996]. We obtain a fit very similar to model 9. A F-test shows that the null hypothesis that a two-plate versus one-plate model are similar cannot be rejected at a significance level greater than 99%. We conclude that the current regional data set does not require splitting the Caribbean plate. Longer observational time series at continuous sites on the Nicaragua Rise in particular will be essential to confirm or refine this statement.

3.3 Discussion

3.3.1 Best Fit Model

In the end, we find that model 9 (Figure 8) provides the best fit to the GPS data used here with the simplest block geometry required by the data. Velocity residu- als are within measurement uncertainties at all sites and are randomly distributed, without systematic pattern (Figure 9 and Figure 10). Estimated plate and block an- gular velocities (Table 2 and Figure 11) are well determined thanks to the significant number of GPS sites available, except for the North Hispaniola block whose angular velocity estimate relies on a small number of sites over a small geographic footprint. Contrary to Manaker et al. [2008b] we did not constrain the angular rotation of the Caribbean plate to the DeMets et al. [2000] value but estimated it. Our Caribbean- North America angular velocity agrees with the 3.2 Ma MORVEL geological estimate [DeMets et al., 2000] within errors. It is close to, but significantly different from the Benford et al. [2012a] or the older DeMets et al. [2000] estimates (Figure 11). We also find a Puerto Rico-North America angular velocity similar to that of Benford et al. [2012a]. In the northern Caribbean, the best-fit model predicts pure strike-slip motion along the EPGFZ (∼9 mm/yr), Septentrional (∼10 mm/yr), and Oriente (∼10 mm/yr) faults (Figure 8). These results are consistent with previous estimates [e.g., Manaker et al., 2008b, Calais et al., 2010]. In addition, the model predicts significant short- ening along the Enriquillo fault system through southern Haiti, with 5 to 7 mm/yr 26

(west to east) of plate boundary-perpendicular motion. This is simlar to – though slightly larger than – the results of the latest kinematic block model for the north- ern Caribbean by Benford et al. [2012a]. This new finding likely results from the significant improvement in GPS site distribution in Haiti and is consistent with the transpressional nature of the 2010 Haiti earthquake whose moment release was 2/3 strike-slip and 1/3 reverse motion [Calais et al., 2010]. The best-fit model predicts slow N-S convergence (2 to 4 mm/yr from east to west) across the north-Hispaniola fault, consistent with the source mechanisms of the 1946 [Dolan and Wald, 1998] and 2003 [Dolan and Bowman, 2004] earthquakes off the northern coast of the Dominican Republic. A similar conclusion was reached by Manaker et al. [2008b], who however imposed earthquake slip vector directions as a constrain in the kinematic inversion. This result confirms the partitioning of the highly oblique convergence between the Caribbean and North America plate be- tween intra-arc strike-slip faults and convergence at the plate interface [Calais et al., 2002b]. The model also predicts a moderate amount of convergence (∼1–3 mm/yr) across central Hispaniola along the boundary between the Gonave and Hispaniola microplate, possibly coincident with the Hinche-San Juan valley fault zone [Pubel- lier et al., 2000b] or distributed over a broader region. This result should prompt additional geological work to try to identify paleo-earthquakes in this supposedly aseismic region. Finally, the model indicates significant oblique shortening across the western part of the Muertos trough (∼5–6 mm/yr), consistent with active compres- sional structures well imaged by offshore seismic reflection data [Ladd et al., 1977, Granja Bru˜na et al., 2014]. We find no evidence for internal deformation of the Puerto Rico-Virgin Island block at the ∼0.5 mm/yr level (RMS of residual velocities), consistent with the ab- sence of significant active faulting [Frankel et al., 1980]. Holocene faulting within Puerto Rico reported, in particular, on the Cerro Goden-Great Southern fault zone [Grindlay et al., 2005b] therefore has to be occurring on very slow slipping faults. The Caribbean-North America plate motion along that segment of the plate bound- 27

ary is accommodated by oblique slip on the plate interface, consistent with slip vec- tor directions of instrumental earthquakes (Figure 1), though a portion of the plate boundary-parallel slip could be accommodated by the shallow Bowin and Bunce faults on the inner wall of the Puerto Rico trench [Grindlay et al., 2005a, ten Brink and Lin, 2004]. The partitioning of the oblique convergence between the Caribbean and North American plates into plate interface convergence and intra-arc strike-slip fault- ing therefore ceases as the plate boundary transitions from the oblique collision of the Bahamas platform with Hispaniola to the oblique subduction of old (> 100 Ma) lithosphere under Puerto Rico [Dolan et al., 1998, Mann et al., 2002, Grindlay et al., 2005b, Mondziel et al., 2010]. The model predicts little motion (<1.4 mm/yr) across the Anegada Passage fault system while the relative motion between the arc and the North American plate remains constant in direction and magnitude. This holds all the way down the Lesser Antilles arc, where velocity residuals with respect to the Caribbean plate are within measurement errors at all reliable GPS sites (see discussion above). This is also true for site BDOS on Barbados island, which has a 9–years continuous observation time span and is located on the emerged part of the old Antilles accretionary prism [Bangs et al., 2003]. The GPS data are not consistent with the 5 mm/yr motion of a northern Lesser Antilles rigid sliver proposed by Lopez et al. [2006]. They also exclude the slip partitioning model proposed by Feuillet et al. [2002, 2010] with 5 to 10 mm/yr of distributed deformation throughout the northern Lesser Antilles. The observed active faults within the arc and forearc must therefore be accumulating strain at a rate of at most 1–2 mm/yr, the average residual of the best-fit model in the Lesser Antilles. Along the southern boundary of the Caribbean plate, the best-fit model predicts pure strike-slip motion on the El Pilar fault in northern Venezuela at ∼18 mm/yr. Further west the Caribbean/South America relative motion splits into 1.5 mm/yr of pure strike-slip motion on the Oca fault and 2-2.5 mm/yr on the left-lateral Santa Marta-Bucaramanga fault, while the bulk of the plate motion is accommodated by 28

∼12 mm/yr of right-lateral strike-slip on the Bocono fault. Consistent with previous results, the Bocono fault also accomodates up to 3 mm/yr of shortening. The left- lateral strike-slip motion predicted by the model along the Santa Marta-Bucaramanga faults is at the low end of slip rate estimates from maximum ages of Quaternary offset features (0.2 mm/yr) [Paris et al., 2000], paleoseismological studies at its northern termination (5-15 mm/yr) [Diederix et al., 2012, Idarraga-Garia and Romero, 2010], or earlier GPS results (6+−2 mm/yr) [Trenkamp et al., 2002]. The model slip rate for the Bocono fault is consistent with estimates of right-lateral strike-slip motion from paleoseismological investigations [Audemard et al., 1999, 2005]. The model predicts ∼N-S convergence across the South Caribbean Deformed Belt at ∼3 mm/yr offshore northeastern Venezuela, transitioning to ∼5 mm/yr offshore central Venezuela and ∼8 mm/yr offshore northern Colombia. These values are however not very robust because of the limited number of GPS sites available on the north Venezuela and Andes blocks. Finally, we find a rotation pole for the relative motion between North and South America located in the central part of the Fifteen-Twenty Fracture Zone (Figure 14), close to the stage pole found by [M¨uller and Smith, 1993] for the 0-10.4 Ma pe- riod (17.2o latitude, -53.5o longitude), with a similar rotation rate (0.2o/Ma). This indicates that the South America-North America relative motion has not varied sig- nificantly over the past∼10 Ma. Our model predicts about 1 mm/yr of present-day N-S shortening across the Barracuda and Tiburon ridges to the west, consistent with offshore geological data showing thrusting and thrust-related folding affecting Qua- ternary sediments along both ridges indicative of north-south compression [Patriat et al., 2011]. This prediction is in closer agreement with geological observations there than other recent estimates [DeMets et al., 2010, Argus et al., 2010] that predict a significant amount of strike-slip motion that does not appear in the offshore geological data [Pichot et al., 2012]. Our South America-North America rotation parameters also predicts up to 1 mm/yr of N-S plate divergence to the east, consistent with extension at the Royal Trough [Roest and Collette, 1986] (see also one CMT focal 29 mechanism at the Royal Trough on Figure 14) and with a cluster of anomalous focal mechanisms off the mid-Atlantic ridge to the south and east that show N-S directed T-axes [Escartin et al., 2003].

3.3.2 Low Coupling on the Lesser Antilles Subduction

A robust and important output of the block models is the very low coupling estimated along the Lesser Antilles and Puerto Rico subduction interface (Figure 13). Taken at face value, this would indicate that the 19 mm/yr of plate convergence builds little to no slip deficit on the plate interface. As for many subduction, GPS measurements on the island arc provide limited coverage of the plate interface, in particular at close distance to the trench. Trench to Island Arc distances, ranging from 175 to 250 km with a depth of the subduction interface beneath the arc ranging from 50 to 100 km, are however similar to Japan or South America, where locked sections of the plate interface do show strain accumulation at coastal GPS stations [e.g., Mazzotti et al., 2000, Loveless and Meade, 2010, Nocquet et al., 2014]. As discussed above, the fact that the velocities of Lesser Antilles GPS sites are con- sistent with the stable Caribbean plate suggests a plate interface that is at least only partially coupled. This is shown on cross-sections across the central (Guadeloupe), northern (St Martin), and southern (Barbados) parts of the subduction (Figure 15) which compare the arc deformation predicted by the best-fit model to that predicted using a plate interface fully locked to 40 km depth. It is apparent on Figure 15 that a fully locked plate interface would cause measurable deformation of the arc, in particular at the Desirade (Guadeloupe, site ADE0), St Martin (site SMRT), and Barbados (site BDOS). These sites, as well as the ones more distal to the trench, show no evidence of elastic strain accumulation. We ran resolution tests in order to determine the spatial resolution of the cou- pling ratio along the plate interface allowable by the data (Figure 16). To do so we calculated slip rates on the subduction patches using the best-fit block model and 30 prescribed either full locking (i.e., slip rate = full relative plate motion) or no locking (i.e., slip rate = 0) in a forward model where we calculate predicted velocities at all GPS sites. We kept the same geometry as the best-fit model for all synthetic models unless specified. We perturbed the predicted velocity using Gaussian white noise with a standard deviation of 0.5 mm/yr, consistent with the RMS scatter of the data to the block model. We then used this predicted velocity field, together with the observed velocity uncertainties, to solve for both rigid block motion and slip on the subduction patches. We first seek to determine whether the data can resolve the lateral variation in the slip rate estimate in the best-fit model, from 0 mm/yr north of Hispaniola (i.e., fully coupled plate interface) to close to the full relative plate motion rate along the Puerto Rico and Lesser Antilles subduction (i.e., uncoupled plate interface). We run three successive models where we prescribe uniform slip on the North Hispaniola fault (3 mm/yr), Puerto Rico subduction (19 mm/yr), and Lesser Antilles subduction (19 mm/yr) (Figure 16a to c). As shown on Figure 16d to f, the data allows for the recovery of this spatial variability with some smearing at the segment edges. This shows that the transition from locked to unlocked from Hispaniola to Puerto Rico is resolvable by the data. It also shows that the data are sufficient to detect fully locked plate interfaces along each of the Hispaniola, Puerto Rico, and Lesser Antilles subduction segments. A similar test with alternating locked/unlocked segments of shorter length (∼200 km, Figure 16g and 16j) shows that the data can resolve small scale variability on the North Hispaniola thrust, but with significant smearing and poor resolution along the Puerto Rico and Lesser Antilles subduction segments. We then seek to determine how well the available data can resolve along-dip variations in interplate coupling. To do so, we impose full locking or unlocking along the upper or lower half of the subduction interface. As shown on Figure 16h–k and 16i–l, the down-dip resolution of the models is poor, a consequence of the distribution of sites on the island arc at more than 200 km from the subduction trench. We then ask whether offshore geodetic sites could improve the along-dip resolution of the 31 estimated coupling and how many would be needed. We placed fictitious offshore sites at various distances to the trench and ran a series of tests where we impose full locking on the lower half of the subduction interface (Figure 17–a). As expected, the resulting resolution varies greatly with the number of sites and their distribution. With 20 GPS sites uniformly distributed on top of the subduction interface we obtain a good along-dip resolution with some smearing in regions where the space between the GPS sites become large. The resolution is still acceptable with 10 GPS sites (Figure 17–d) and becomes poor with 5 only sites. We conclude that low coupling (0-10%) along the Lesser Antilles/Puerto Rico sec- tion of the Caribbean-North America plate interface is a robust feature of the model, although one cannot determine the depth distribution of coupling with confidence. Stein et al. [1982] first pointed out the lack of significant thrust faulting earthquakes (M>7) in the Lesser Antilles in the historical and instrumental record, which they interpreted as indicative of a uncoupled and aseismic subduction where “only a small fraction of the slip occurs as interface thrust earthquakes”. Although there are a few moderate magnitude (4accretionary wedge. The largest historical earthquakes in the Lesser Antilles in 1690 and 1843 [Robson, 1964] are both reported to have reached an intensity of IX [Dorel, 1981]. The magni- tude of the better documented 1843 event has been estimated 7.5 to 8.0 [Bernard and Lambert, 1988, ten Brink et al., 2011], 8.0 to 8.5 [Mccann et al., 1982], 8.5 [Feuillet 32

et al., 2001], or greater than 8.5 [Hough, 2013]. Because of its magnitude and location at a subduction plate boundary, this earthquake, which did not cause a tsunami, is often interpreted as a thrust event on the plate interface [Mccann and Sykes, 1984, Bernard and Lambert, 1988] though no direct evidence for this is available. If this event is indeed a thrust, then the very low slip deficit on the subduction interface inferred from GPS data implies that similar events have a long recurrence time. Us- ing the empirical relations between average slip and moment magnitude [Wells and Coppersmith, 1994], we find a recurrence time of 2000 years for Mw8.0 events and 3 mm/yr of average slip on the subduction interface. Alternatively, the 1843 event may have had a source mechanism similar to its large normal faulting neighboring events of 1969 (Ms7.5), 1974 (Ms7.4), and 2007 (Mw7.4). A similar depth range of 40-140 km would explain the lack of a tsunami and the lack of substantial coseismic uplift reported [Bernard and Lambert, 1988]. The low coupling inferred here for the Lesser Antilles subduction bears similari- ties with a similar result at the Hellenic subduction, where Vernant et al. [2014] find that strain accumulation on the plate interface accommodates <20% of the Nubia- Aegean convergence rate. The context is interestingly similar, with a ∼900 km long arcuate subduction, a slow convergence rate (∼3 cm/yr), and the subduction of old (>100 Myr) and dense oceanic crust. Slab-pull and roll-back of the subduction have been invoked as a mechanism that reduces the normal stress acting on the plate interface and facilitate aseismic slip [Scholz and Campos, 1995]. This mechanism would be consistent with the pervasive normal faulting documented throughout the northern half of the Lesser Antilles arc [Feuillet et al., 2002]. Alternatively, Wallace et al. [2012b] proposes that fluid overpressure, because it reduces frictional strength of faults, lowers the depth of the transition from frictional to viscous behavior. As a consequence, subduction zones experiencing significant fluid overpressure should be locked to greater depths than those where high permeability prevents overpressure to develop. These authors propose that the former (e.g., highly overpressured, deep locking) would occur in regions of upper plate contractional tectonics, while the latter 33

(e.g., more permeable upper plate, less overpressure, and shallower locking depths) would correspond to an extensional stress regime in the forearc. A correlation is indeed observed between along-strike transition from locked to creeping subduction and a coincidental transition from compressional to extensional stress regime in the upper plate in New Zealand, Japan, and Vanuatu. The same observation holds in the Caribbean, where the transition from upper plate contraction to extension between Hispaniola and in Puerto Rico also corresponds to the transition from deep interseis- mic locking to aseismic creep on the plate interface. Mechanical models, beyond the scope of the present study, are now needed to test this hypothesis. The Lesser Antilles subduction has been compared with the Tohoku segment of the Japan subduction, which ruptured on March 11, 2011 in a Mw9.3 earthquake. Both share seismicity in the mantle wedge (mostly normal faulting) and thrust events at the plate interface just below the mantle wedge [Laigle et al., 2013]. Recent studies show that the Tohoku rupture may have extended under the mantle wedge, which somewhat broadens the width of the seismogenic zone [Koketsu et al., 2011]. However, full and deep interseismic coupling was documented by GPS measurements along the Tohoku segment before the March 2011 earthquake [Mazzotti et al., 2000, Loveless and Meade, 2010], a characteristic that is opposite to the findings reported here for the Lesser Antilles subduction. 34

4 PRESENT-DAY SHORTENING IN SOUTHERN HAITI FROM GPS MEASUREMENTS

On January 12, 2010, Haiti was struck by a devastating – though not unexpected – earthquake [Bilham, 2010] (Figure 1). Because its epicenter was located in the near vicinity of the Enriquillo Plaintain Garden Fault (EPGF), a major active fault part of the left-lateral strike-slip boundary between the Caribbean and North American plates (Figure 1), it was first thought that the event had ruptured a portion of the strike-slip EPGF. Independent studies based on geodetic [Calais et al., 2010, Hayes et al., 2010a, Hashimoto et al., 2011, Mercier de L´epinay et al., 2011, Symithe et al., 2013a], geological [Prentice et al., 2010a], and seismological data [Douilly et al., 2013] showed that the earthquake had actually ruptured a previously unmapped fault with a source mechanism combining strike-slip and reverse faulting, in a setting resembling the 1989 Loma Prieta earthquake in California [Dietz and Ellsworth, 1990, Beroza, 1991]. This composite source mechanism was a first indication that the conventional interpretation of the EPGF in southern Haiti as a purely strike-slip fault system may need revisiting. GPS measurements in the northeastern Caribbean have established that the 19 mm/yr relative motion between the Caribbean and North American plates, slightly oblique to the plate boundary direction in Hispaniola, is accommodated by shortening on the North Hispaniola fault to the north and strike-slip motion on the ∼E–W striking Septentrional and Enriquillo faults throughout the island (Calais et al. [2002a], Man- aker et al. [2008a], Figure 1). Further geodetic studies making use a larger number of geodetic sites identified an additional component of boundary–normal shortening in southern Haiti [Calais et al., 2010, Benford et al., 2012b, Symithe et al., 2015], consistent with the composite source mechanism of the 2010 earthquake but, again, 35

questioning the purely strike-slip nature of the EPGF system in the region [Mann et al., 1995a]. Here, we revisit the present-day tectonic setting of southern Haiti by testing pub- lished tectonic models in light of geodetic data acquired since the 2010 earthquake. We show that the well-known strike-slip motion affecting the area is accompanied by an even larger component of shortening on a south-dipping reverse fault along the southern edge of the Cul-de-Sac basin, which holds the ∼3 M–people capital city of Port-au-Prince. This tectonic scenario implies that ground shaking may be twice that expected if the major fault was purely strike-slip, as asssumed in the current seismic hazard map for the region.

4.1 Tectonic Setting

The EPGF has long been recognized as a major tectonic feature of southern Haiti, where it was first named “Lin´eament Tiburon-P´etionville” from the eponymous end- points of its surface trace mapped using airphoto interpretation [Duplan, 1975], then “D´ecrochement S´enestre Sud Ha¨ıtien” [Calmus, 1983]. Subsequent studies however tended to minimize its importance as an active plate boundary fault. The EPGF does not appear as a major feature on the 250,000 scale Haiti geological maps [Momplaisir and Boisson, 1988]. Pubellier et al. [1991] proposed that slip on the EPGF stopped in the Late Miocene when deformation in southern Hispaniola became compressional as the Haiti fold-and-thrust belt (Figure 1) was propagating southward to its present position with its active front currently thrusting onto the Cul-de-Sac basin along the Neiba–Matheux range [Pubellier et al., 2000a]. In their view, the EPGZ was reactivated as a normal fault in the Quaternary as a result of crustal loading imparted by the fold-and-thrust belt to the north. Mann et al. [1995a] proposed a drastically different interpretation where the EPGF is a quasi-vertical, left-lateral, active strike-slip fault running across southern His- paniola from the westernmost tip of Haiti to the Enriquillo basin in the Dominican 36

Republic. The fault is indeed well-marked in the morphology as a quasi-linear feature with push-ups and pull-aparts throughout the southern Peninsula of Haiti [Calmus, 1983, Momplaisir, 1986]. This holds until about 72.27°W near the city of P´etionville where the EPGF intersects the Cul-de-Sac plain (Figure 18). East of that point, its imprint on the morphology becomes more subtle. Mann et al. [1995a] describe a series of gentle folds affecting plio-quaternary sediments along the southern edge of the Cul-de-Sac basin and into the Enriquillo basin in the Dominican Republic that they interpret as en ´echelon drag folds marking the trace of EPGF (Figure 18A). The Cabritos island in the Enriquillo lake is the easternmost of these folds and, in that interpretation, is the surface expression of a vertical EPGF at depth. Hence, most maps show the EPGF as a single – though segmented – left-lateral strike-slip fault extending from Jamaica to the west all the way through southern Haiti and into the Enriquillo basin in the Dominican Republic to the east. This interpretation contrasts with early geologic work in southern Haiti where the EPGF, well expressed along the Southern Peninsula, is mapped as abutting against a north-verging reverse fault system that mark the southern edge of the Cul-de-Sac basin (Figure 18B, Bourgueil et al. [1988]). Recent geological surveys of the Port-au- Prince area support Mann et al. [1995a] interpretation and show that these N110°E reverse-sinistral faults affect Quaternary alluvial sediments throughout the city and to the east [Terrier et al., 2014]. At a broader regional scale, Saint Fleur et al. [2015] used high-resolution air photos and lidar topography to revisit the Plio-Quaternary folds identified by Mann et al. [1995a] along the southern edge of the Cul-de-Sac basin. They propose that they are fault-propagation folds on top of shallow-dipping decollements emerging well into the Cul-de-Sac basin and rooted on a south-dipping low-angle reverse fault underneath the high relief Massif de la Selle (Figure 18B). In their interpretation, the Cul-de-Sac basin is actively overthrusted on both sides by the Matheux-Neiba range in the north and the Massif de la Selle in the south while the EPGF is a young (<2 Ma) fault propagating eastward throughout southern Haiti. 37

4.2 GPS Data

We use GPS data acquired in Haiti and the Dominican Republic since 2003 to test the tectonic models described above, determine the geometry of the major active faults in southern Haiti, and quantify the related elastic strain accumulation to inform regional hazard assessment. GPS data acquisition and processing procedures are provided in Symithe et al. [2013a]. The solution used here is an improved sub-sample of their data set that includes additional GPS measurements collected in Haiti and the Dominican Republic from June 2014 to April 2015. As a result, we now have a dense distribution of GPS sites covering the whole island, in particular the southern part of Haiti. The regional velocity field (Figure 19) shows left-lateral motion between the Caribbean and North American plates at 17–19 mm/yr, slightly oblique to the plate boundary direction. Velocities in Hispaniola show a north–south gradient with up to 15 mm/yr of integrated left-lateral shear strain across the island, consistent with previous findings [Symithe et al., 2015]. A new finding, however, is a component of boundary–normal shortening readily visible on Figure 19. This shortening affects the central and western parts of Hispaniola, except the Southern Peninsula of Haiti west of 72.5°W where velocities in a Caribbean frame are parallel to the E–W trending EPGF [Calais et al., 2016]. This shortening component that adds to the well-known regional left-lateral shear is readily visible on the profile displayed on Figure 19 where we project the GPS velocities onto directions parallel and normal to the EPGF strike. We observe a well-defined velocity gradient, both in the shortening and strike-slip components, coincident with the contact between the Cul-de-Sac basin and the Massif de la Selle. Although strike-slip motion at about 6–7 mm/yr was expected [Manaker et al., 2008a, Symithe et al., 2015], the significant component of shortening visible on the profiles had not been documented before. In the following, our objective is to determine 38

the geometry of the fault system that accommodates the observed strike-slip and shortening localized along the southern edge of the Cul-de-Sac basin.

4.3 Elastic Model

We implement a simple model where faults are simulated as rectangular dislo- cations in an elastic half-space. We follow the classically-used backslip approach of Savage [1983a] and calculate surface deformation due to faults locked in the upper, seismogenic, crust as the difference between the deformation caused by an infinitely wide fault plane with uniform slip from the surface downward and the deformation caused by a fault plane with uniform slip across a finite width. We relate fault slip to surface motion using the Green’s functions of Okada [1992] for rectangular dislo- cations assuming uniform slip and dip angle. In the “one-fault” models, surface deformation is caused by uniform slip on a fault whose dip is varied from 10°to 90°, surface trace location from 18.38°to 18.82°latitude, and strike from 70°E to 110°E. In the “two-fault” models, we test whether strike- and dip-slip motion may be partitioned on two different, nearby faults. These fault geometries are obviously simplifications of the actual geology, as GPS measurements are rarely able to capture the details of fault plane geometries, even in regions where dense networks and ample geological data are available [e.g., Argus, 2005]. Because the density of GPS sites is not sufficient to capture the short wavelengths of the strain rate field, our models are therefore meant to reproduce the average behavior of a fault system whose geometry is likely to be more complex in actuality. In order to score the models, we only consider GPS sites that are not affected by fault systems other than the Enriquillo–Cul-de-Sac basin. We model a N–S profile that originates just south of Haiti at longitude 72°W and latitude 17.8°N. We run and score a series of models with profile lengths extending northward from 100 km – i.e., to the southern edge of the Neiba-Matheux mountains – to 200 km – i.e., encompass- ing additional sites in northern Haiti close to the Septentrional fault (Supplemental 39

Figure 23A). As expected, model misfit increases with profile length as more data is included, but with a significant jump at 175 km profile length, wherer sites start being significantly affected by elastic strain accumulation on the Septentional and North Hispaniola faults. We also run and score a series of models where we vary the profile width to determine reasonable eastern and western bounds for the set of sites to include. We seek to encompass a region where lateral variations in fault geometry and tectonic regime are negligible so that our two-dimensional models hold, while still including as many GPS sites as possible in the analysis. We find that model misfit increases significantly for profile widths of 58 km and larger (Supplemental Figure 23B). The therefore select, for the sake of this analysis, GPS sites located inside the 175 km–long by 58 km–wide box shown on Figure 19. In order to optimally search the parameter space and determine reliable uncer- tainties associated with the estimated parameters, we make use of Bayes’ rule [Segall, 2002, Johnson, 2004, Hilley et al., 2005]:

P (x|mi) ∗ P (mi) P (mi|x)= (4.1) ΣP (x|mj) ∗ P (mj)

where mi is a vector of four unknown parameters (slip, dip, locking depth, and fault

location), x is a set of observed GPS velocities, P (mi|x) is the probability of parameter

set mi to explain the observed data set x,andP (mi) is some prior probability that this model parameter set actually occurs derived from independent information. P (x|mi) is the probability of observing x as interseismic surface deformation when the model

parameter is mi. Since we have minimum a priori information on the estimated

parameters, we use a uniform distribution as prior for all parameters (P (mi)) over intervals 0-20 mm/an for slip rates, 0-90 °for dip angle, and 2-15 km for locking depth. In order to optimally evaluate equation 4.1, we use a Metropolis-Hastings sampling method which uses a Markov Chain with stationary distribution [Metropolis et al., 1953]. The best-fit model in the one-fault configuration, shown on Figure 20, is obtained for a fault–surface intersection at 18.55°latitude, with a dip of 37°±13 to the south 40

underneath the Massif de la Selle, and a locking depth of 5±3.5 km (Table 4). It is remarkable that the inferred location of the fault–surface intersection coincides precisely with the southern toe of the fault-propagating folds identified by Saint Fleur et al. [2015], for instance the Ganthier fold (Figure 20, bottom panel). The model scores a χ2 of 69.92 with WRMS of 1.3 mm/yr and 0.8 mm/yr for the E-W and N-S components of the GPS velocities, respectively. Parameters inferred from the Bayesian analysis, their uncertainties, and their covariance are shown on Figure 21. We observe that inferred parameters are well determined from the data which favor a south-dipping fault with oblique slip. Supplemental Figure 24 shows model/data comparisons for a range of fault dip angles and surface trace locations. It is readily apparent that the GPS data favor a shallow, south–dipping fault with a surface trace located well within the Cul-de-Sac basin. We also tested a two-fault model where deformation is partitioned onto (1) pure dip-slip motion on the best-fit shallow-dip reverse fault determined above, and (2) pure strike-slip motion on a vertical fault located at the contact between the Massif de la Selle and Cul-de-Sac basin, i.e., the usually assumed EPGF fault geometry. The best-fit slip rates are 6 mm/yr and 9 mm/yr for the strike-slip and reverse faults, similar to the one-fault model. This model scores a χ2 of 78.62 with WRMS of 1.3 mm/yr and 0.8 mm/yr for the E-W and N-S components of the GPS velocities, respectively. Decreasing the dip of the strike-slip fault systematically increases the misfit (WRMS=1.35 mm/yr and χ2=85.88 for a 70°dip). Therefore, although parti- tioning slip on two faults slightly increases the model misfit compared to the one-fault model, the overall fit remains reasonable. However, the data still requires a significant amount of dip-slip motion on a shallow, south–dipping reverse fault with a surface trace located well within the Cul-de-Sac basin. 41

4.4 Discussion

4.4.1 Present-day Tectonic Model

Plate boundary deformation in southern Hispaniola is currently accommodated by a combination of plate boundary–parallel strike slip motion that appears to decrease from west to east and plate boundary–normal shortening that decreases from west to east (Figure 19). West of Port-au-Prince, the GPS velocities imply that deformation is partitioned between strike-slip on the EPGF along the Southern Peninsula of Haiti and shortening across the Gulf of Gonave. This is consistent with seismic surveys that have long shown a series of active folds and reverse faults along the northern coast of the Southern Peninsula and across the Gonave Gulf, the Gonave Island itself being one of these anticlinal folds [Goreau, 1981, Bruneton et al., 1988, Momplaisir, 1986]. One of these reverse faults, the Trois Baies fault, was the locus of triggered seismicity with purely reverse focal mechanisms in the wake of the 2010 Haiti earthquake Douilly et al. [2013]. The reverse faulting required by the GPS data along the southern edge of the Cul-de-Sac basin is consistent with geological observations of active folding [Mann et al., 1995a, Terrier et al., 2014] and with the geological interpretation that the Massif de la Selle is actively thrusting over the Cul-de-Sac basin [Bourgueil et al., 1988, Saint Fleur et al., 2015]. It remains possible that the strike-slip component is accommodated by an additional, EPGF-type vertical fault, different but close to the reverse fault required by the data. However, the configuration where a strike-slip and a dip-slip fault would intersect at depth does not appear to be mechanically stable – it is difficult to imagine how the two faults would coexist over the long term unless one of them has a very low slip rate, which is not the case with dip-slip and strike-slip components of 9 and 6 mm/yr, respectively. In addition, we note that no direct geological evidence has yet been provided of the existence of a vertical strike-slip fault along the southern edge of the Cul-de-Sac basin. There is however direct field evidence for reverse faulting in Quaternary alluvial 42

deposits throughout Port-au-Prince [Terrier et al., 2014] and just to the east of the city at the Dumay locale [Saint Fleur et al., 2015], consistent with an early geologic map of Haiti where the southern edge of the Cul-de-Sac basin is a north-verging reverse fault system intersecting the EPGF just east of Port-au-Prince [Bourgueil et al., 1988]. The interpretation of a vertical EPGF continuous throughout the Cul-de-Sac basin owes to Mann et al. [1995a]’s reproduction of a seismic line across the Enriquillo Lake (MOBIL LINE 1127) in the Dominican Republic where recent faulting just north of the Cabritos Island anticline is interpreted as the trace of the EPGF. Interestingly, seismic reflectors below about 1.5 sec on this profile are well-marked and continuous across it, indicating that surface faulting does not continue vertically at depth. This points to an alternative interpretation where the Cabritos anticline may also be a fault-propagation fold developing above a shallow decollement rooted to the south on a south-dipping reverse fault underneath the Selle-Bahoruco Massif, similar to the interpretation of the Cul-de-Sac folds by Saint Fleur et al. [2015]. We therefore conclude that the geodetic and geological information currently avail- able are consistent with significant present-day shortening coincident with the south- ern edge of the Cul-de-Sac basin, in addition to the strike-slip motion component additionally required to fit the data. The data support a model where this deforma- tion results from slip on a locked, shallow-angle fault dipping southward underneath the Massif de la Selle and accommodating a combination of 9±3 mm/yr of reverse motion and 6±2 mm/yr of strike-slip motion. It is readily apparent on Figure 20 that the GPS data do not indicate significant deformation along the Mattheux-Neiba front at the precision level of the measurements, about 1 mm/yr.

4.4.2 Implication for Seismic Hazard

The current Haiti seismic hazard maps, produced shortly after the 2010 earthquake [Frankel et al., 2011], follow the Mann et al. [1995a] interpretation of a vertical, purely 43

strike-slip EPGF throughout southern Haiti and include active reverse faulting along the Matheux-Neiba front. The tectonic scenario advocated for here on the basis of GPS data is different. We therefore seek to determine the impact of either scenario on earthquake ground motion. To do so, we follow the ShakeMap procedure [Wald et al., 1999] using the openSHA framework [Field et al., 2005]. In the alternate tectonic scenario advocated for in this paper, we assume a fi- nite source rupture on a fault dipping 37°to the south, parallel to the reverse faults mapped by Saint Fleur et al. [2015] along the southern edge of the Cul-de-Sac basin (Figure 18B). We arbitrarily simulate a rupture of magnitude 7.0, consistent with historical records in the region [Ali et al., 2008, Bakun et al., 2012a]. We model a rupture plane extending laterally from 72.35°W to 71.88°W and vertically from 0 to 5 km, combining dip-slip and strike-slip motion (rake=56°). In the classic EPGF sce- nario, we assume a finite source rupture on a vertical fault following the trace shown in Mann et al. [1995a] and assumed in Frankel et al. [2011] (Figure 18A). We use the same lateral extent as in the alternate scenario with vertical dip from 0 to 8 km and pure strike-slip motion (rake=0). All other simulation parameters are identical for both scenarios. We use the same attenuation relations as in the 2010 Haiti national seismic hazard maps [Frankel et al., 2011] which apply with equal weight three of the Next Generation of Attenuation (NGA) relations [Boore and Atkinson, 2008, Campbell and Bozorgnia, 2008, Chiou and Youngs, 2008], derived from a global compilation of strong-motion records from tectonically active areas. The additional details provided by the microzonation study of Port-au-Prince Gilles et al. [2013] are unimportant for our discussion. We make use of Vs30 estimations from topographic slope derived from SRTM30 Version 2 [Allen and Wald, 2009]. Figure 22 shows the resulting maps of the median peak ground acceleration (PGA) for the two scenarios described above. They are meant to qualitatively illustrate these scenarios – exploring other attenuation relations or producing an actual seismic hazard map is beyond the scope of this paper. 44

We observe contrasted results between the two scenarios, with PGA reaching 0.4 g for the vertical strike-slip rupture but up to 0.75 g for the oblique-slip rupture on a south-dipping fault with g =9.82ms−2. The new interpretation of the Cul-de-Sac basin tectonic regime required by the geodetic data is therefore much more detrimen- tal for southern Haiti than the strike-slip source assumed in the current earthquake hazard maps. This holds in particular for the Port-au-Prince area where ground ac- celeration was likely limited to 0.1–0.4 g during the 2010 earthquake according to models [Mavroeidis and Scotti, 2013, Douilly et al., 2015] and geotechnical analysis of ground performance [Olson et al., 2011]. 45

5 CONCLUSION

We have assembled an up-to-date GPS velocity field for the Caribbean plate and its boundaries with North and South America which we used to quantify the kinematics of active deformation in the region. Our results confirm several earlier findings, in particular for the Caribbean-North America plate boundary, and extends them to the Lesser Antilles and the Caribbean-South America boundary. A new key finding is the low coupling required by the GPS data along the Lesser Antilles subduction interface, which was previously poorly resolved. As a consequence, seismic hazard associated with strike-slip plate boundary faults along the northern and southern margins of the Caribbean plate is at least as impor- tant as the threat posed by the subduction plate interface in the Lesser Antilles. This seems to be reflected in the distribution of large historical earthquakes in the region [Stein et al., 1982]. Under the paradigm that the magnitude of large earthquakes depends on the slip deficit accumulated on a potential rupture of sufficient area, the very low seismogenic coupling found here on the Lesser Antilles subduction allows at most one M8 earthquake every 2000 years, or one Tohoku-size event every 3500 years. The limited coverage of GPS measurements in the region calls for additional sites where still possible, though significant progress has been made in the framework of the COCONet initiative [Braun et al., 2012]. Given the oceanic nature of most of the Caribbean plate boundaries, offshore geodesy appears to be a viable way forward, though the implementation cost may be large. In addition, systematic efforts to obtain a reliable catalog of paleoearthquakes in the region is key to our understanding of the regional seismic and tsunami hazard. This improved GPS data set across Haiti shows that active strain accumulation in the southern part of the island involves a significant component of north-south shortening in addition to the well-known component of left-lateral strike-slip motion. 46

Horizontal gradients in GPS velocities together with simple elastic models require that this deformation is accommodated by a combination of reverse (9±3 mm/yr) and strike-slip (6±2 mm/yr) faulting on active faults along the southern edge of the Cul-de-Sac basin. The data require that reverse – and possibly strike-slip – motion is accommodated by a fault dipping at a shallow angle of 37°±13 under the Massif de la Selle with the surface trace located well within the Cul-de-Sac basin, consistent with the location of fault-propagation folds described in the basin morphology [Saint Fleur et al., 2015]. This new interpretation has important implications for seismic hazard as rupture on the reverse or oblique fault required by the geodetic data implies ground motion that may be up to twice larger than rupture on the usually assumed vertical, purely strike-slip, EPGF. There is much at stake in this ∼3 M inhabitant region that is still recovering from the 2010 earthquake and where urbanization continues at a fast rate. Additional geodetic and geological work in this region is key to further investigate this interpretation and, eventually, lead to revised and improved seismic hazard estimates for the region.

FIGURES

51

2 £

©

¨

¥

¤

¢

1

§

¦

¥

¤

R

1 ¡

£ £ 2£ 2 £

5 1 15 5 3

L  

Figure 5. Model reduced χ2 as a function of fault locking depth illus- trating the minimum found for 14 km.

52

!" #!" $!"

N ¡¢£¤¥¦¡§¨©



¸¹º¹»¹¼

¹ »

½¾ ¿ÀÁÂ

«

¬

­

®

¯

a

b

c

d

e

f

b

g

h

e

‹

Œ



Ž

r

7 

q

p 

2!"

8 o

2!"

9

:

n

m

8

l

k

;

j

i

:

<

=

>

4

6

5 ;

4

3

0 ?

O

@





©





ª







©

¨

§

H P

¦

G 

¥

H

z

¤

{

¶·

³¶

µ

£

³´

²

±

°

|

¢

|

{

¡

J

I

F

D

}

B

W

ÃÄÅ

R

Q

L

  K

M

%

&

'

(

)

*

 

~



€



‚

‚

ƒ

„



C

C

†

‡

ˆ

‰

s

—

t

’

Š

u –

v

•

w •

”

x “

’

t ‘

y



œ

ž

›

Ÿ

P 

š

™ ˜

40

30

^

_ ` 20

10

E+ ,-+./

!"

1

!"

1

Y



Z

[

U

X

\

U

V

[

U

T ]

Z

[

A

S ¢£¤¥¦¡§¨©

!" #!" $!"

Figure 6. Block geometry used in the models tested. Solid black lines show the block boundaries for the best-fit model, thick dashed lines show other tested block boundaries. NHIS = North Hispaniola, PRVI = Puerto Rico and Virgin Islands, GONA = Gonave, HISP = Hispan- iola, NLAB = North Lesser Antilles Block, SJAM = South Jamaica. CARW = Caribbean West, CARE = Caribbean East, NVEN = North Venezuela, MARA = Maracaibo, ANDE = Andes, HFBT = Hispan- iola Fault and Thrust Belt, NMF = Neiba-Matheux Thrust, SJF = South Jamaica Fault. Thin dashed lines are depth contours of the subduction interface used in the model, derived from the earthquake hypocenters cross-sections shown on Figure 4.

53

¡

3

3



a







i

¡¤



a



*



)





t

,



¡

£



a





(



i

a

o





i

a

t

o



¡

¢



2



t

u





'





u

&

t '



 $

¡¡

u



&



a

$

)













a

i





i

t



i

t

t

#





a

a













¡

i

a





m

(

t









a i









"

n



a

v



a

e

a







m













a



%



v

n

%



1¤

-

a

i



! $

t





$



u

a

%

)

+ +

+

*



!

$

t

"

i



+

-

e

t

(

?

i

1

£



-

#

t



i

-

t

5



i

#



#

5

a

5



1

¢

i





m





1¡

+

1

1 ¡ ¥ 7 ¤ ¦ 1 11 1¡

3 ¢ £

M§ ¨©

Figure 7. Total model χ2 as a function of model tested. The line joining two models is green if the null hypothesis that the two models are similar can be rejected at a confidence level greater than 99%. It is red otherwise.

54

¥¦§ ¨ ¥ ¨ ¥ ¨ ¥ ¦¨ ¥ ¨ ¥ §¨ ¥ ¨

© ©



N  

%$9

%$

)

&$

'

9$

(

&$

1

9$9

§¨

$

1* *

9$9

§¨

$

1* 3

$

1 (

$ 6$

1* ) 1 )

$ $% $9

1* ' 1( 1'

$

1* 3

&$

'

$

11 '

$

1 )

$&

13

HP"

G !



%$9

%$&

$

1 1

$9

3

%$&

9$

*

9$%

9$

3

 H

$

1* 1

$

( (

$&

11

9$

(

9$

'

9$ &$

1 )

$

13 )

6$9

$&

*

$

) )

$%

1

 ¨

$

13 (  ¨

C  

  

9$%

1

$&

*

%$

3

%$

1

#

" 

%$

3

9$9

1

 ¨

 ¨

&$

3

$%

'

6$& &$

(

& $

* )

&$

'

9$9

1

& $&

*

%$

'

$

1) *

& $%

*

9$%

! !

M P

&$

3

%$%

1

&$

'

&$6

1

¦¨

¦¨

! A#



S     2 ¡¡¢£¤

¥ ¨ ¥ ¦¨ ¥ ¨ ¥ §¨ ¥ ¨ ¥¦§ ¨

©

Figure 8. Best-fit model geometry with block boundaries as solid black lines and predicted relative block motions as arrows with velocity indicated in mm/yr with their 95% confidence ellipse according to the parameters listed in Table 2. Red = strike-slip (i.e., slip direction with ±30o from fault strike), blue = reverse or transpressional, gree = normal of transtensional. Residual velocities are show with grey arrows. We omitted their error ellipses for a sake of readability, see Figures 9 and 10 for a close up on Hispaniola and the Lesser Antilles. The thin dashed line indicates the boundary of the Bahamas Platform.

55

      

£¥ ¥

d

1§ ¨¨© r

¨ £ ¦

m







e $

N% &' (i )

r ¡ ¢ £¤¥¦

1§ ¨¨© r

LA" 

H ,-$# o.$

N% i i





C

 

C 

L

M

 

C 

A 

F

A

C 

Go#$ve

D 





C

J!A

A  

 

M 



L"



H ,-$# o.$

i i

J



*$ ++e $#

(i

     

Figure 9. Top: observed and modeled velocities in Hispaniola shown with respect to the North American plate. Bottom: residual veloci- ties. Dashed blue lines show the block boundaries, with block names labeled in blue. Error ellipses are 95% confidence.

56

© ©  © © 

P0 23

m¢ d£¤

1¥ mm¦ §¨

A V

A V B

B

¨£ d ¤

d ¡ r 

1¥ mm¦ §¨



C



C

 

S T 

 C

S T



 

C

CO

N  !i"#

C O

© ©

AV

AV

'

(

)

*

$#!i%% #&

 

+

,

-

.

/



B 



B 

G 



G 



 

Figure 10. Left: observed and modeled velocities in the Lesser Antilles shown with respect to the North American plate. Right: residual velocities. Dashed blue lines show the block boundaries, with block names labeled in blue. Error ellipses are 95% confidence. 57

carb-noam B12 MORVEL THIS WORK

DM00

ptrc-noam THIS WORK ptrc-noam B12

nhis-noam gona-noam

gona-noam B12 mara-noam ande-noam

Figure 11. Euler poles for block pairs with their 95% confidence el- lipse. The four stars in northernmost part of the figure show the Caribbean-North America Euler pole for this work in red, Demotes et al., 2000 DeMets et al. [2000] in black, Benford et al., 2012-bBenford et al. [2012a] in blue, and DeMets et al. [2010] in green. Note the general agreement between this work and Benford et al., 2012-b Ben- ford et al. [2012a], although the two 95% confidence error ellipses do not overlap. The same holds for the Gonave-North America relative motion. The agreement between the two studies is however excellent for the Puerto Rico-North America angular velocity estimate.

58

$ S (

D & A &CB

100



A 

u./379:;.9< =>>?<7@ # 





¤

 



*+ 5 ,+* ,+ 5 -+* -+ 5

©

$

&

¦

$ !



80





¡



CB%

§

$ 

B D

!

C

' 

 

¥

¥

 

& 





 C

# 



#

©

"

A

"

A B

60

"



$' !

O

¦

¨

 

D

" !"



¡

¡

" !

)

¤



&CB

¦

©

$



40

##

 %

©



A

"

CA

¨



A 

'



¡

A

§

$



#

O 

¦

' 

A

¤

¥

¤

¢

#

O )

A

£

20

¢

! 

 A

# 



¡

!"#

A

$



P

 

B

0

10 12 14 16 18

L

Figure 12. Test of the consistency of Lesser Antilles GPS velocities with the rigid rotation of the Caribbean plate. The y-axis shows the probability that the null hypothesis – a model with or without a given site are similar – can be rejected. We find that only 3 sites (SOUX, PSA1, MPCH) can be confidently rejected as not consistent with a rigid Caribbean plate motion. Colors show the site velocity uncertainties.

59

¥¦§¨ ¥© ¨ ¥© ¨ ¥ ¦¨ ¥ ¨ ¥ §¨ ¥ ¨

 !" #$%

N coupling ratio

B 

Plto

0.0 0.2 0.4 0.6 0.8 1

T

r

e

3

4



5



i



§¨





 §¨



u



G)* +

./0,

H,-.

 ¨

 ¨

C '' (

%# "%

01

* *

 ¨

 ¨

M /

+ +

¦¨

¦¨

A1

+*

S& !" #$%

2 ¡¡¢£¤

¥© ¨ ¥ ¦¨ ¥ ¨ ¥ §¨ ¥ ¨ ¥ ¦§¨

Figure 13. Coupling ratio estimated along the Greater-Lesser Antilles subduction interface estimated on the discretized plate interface also shown on Figure 4. Residual velocities are show with grey arrows. We omitted their error ellipses for a sake of readability. The thin dashed line indicates the boundary of the Bahamas Platform. Note the coincidence between the transition from coupled to uncoupled plate interface with the transition from Bahamas Platform collision to oceanic subduction at the Puerto Rico Trench.

61

¨

¨

1 %

& %

 ' (e*'

1

&

#

#

$

$

/

MN T

 '(e*'

§

§

§

§

¥

¥

8

8

l

7

#

#

c

9

£

£

:

 ' *

U W X

;

"

"

!

!

l

7

c

9

:

S' ()© '*+

;

¡

¡

V

V

¨

¨

§

¦

§

¥

¦

¥

¤

£

1

¤

¢

£

¡

1

¢

¡

D

D

1

6

2

C C0

Q Q0

1 2

5 4

1 2

¨

¨

%

1

8

%

2

#

#

$

$

 '(e*'

§

O-P

§

§

¥ §

R

¥

1

#

#

£

"

£

"

!

!

<

=

Y©*'*

Z[

\]^_

>

?

G©  

@

¡

A

¡

V

V

¨

 '(e* '

¨

§

¦

§

¥

¦

¥

¤

£

¤

1

¢

£

¡

¢

¡

6

D

D

  

2

0

3 3

` `0

1 2 1 2

5 4 5 4 6

¨

sp

abdfgh abgfg h aiifg h aijfgh akmfg h

%

2

#

$

 '(e*'

,-./

q

p

§

jgfgh

§

¥

1

r

p

#

£

"

F

H

I

!

J

K

L

nmfgh

B© © 

¡

s

o

p

r

V

q

¨

§

¦

nifgh

¥

o

¤

£

¢

uvwxyz{x

¡

ndfgh

D

1

t

p

t

njfgh

0

E E

1 2

5 4 6 &

   

Figure 15. Sections across the Lesser and Greater Antilles Antilles subduction showing topography (grey line), earthquake hypocenter Engdahl et al. [1998], velocity magnitude at the GPS sites (red cir- cles with 95% confidence error bar), velocity predicted by the best-fit model (solid red line), and velocity predicted by a forward model where we impose full-coupling on the subduction interface (dashed blue line). The misfit of the data to a fully locked plate interface is apparent on the three Lesser Antilles cross-sections. 62

20 mm/yr 20 mm/yr 20 mm/yr

s

l

e

d

¦©¨

o

M

c

i t

e Slip rate (mm/yr) Slip rate (mm/yr) Slip rate (mm/yr)

h

t

£ ¤ ¥

0 ¡ ¡0 £ ¤ ¥

0 ¡ ¡0

0 ¡ ¢

¦§¨

n

y

S

a b

20 mm/yr

20 mm/yr 20 mm/yr

¦©¨

t

u p

t Slip rate (mm/yr)

Slip rate (mm/yr) Slip rate (mm/yr)

u

£ ¤ ¥

0 ¡ ¡0

O

£ ¤ ¥

0 ¡ ¡0

0 ¡ ¢

¦§¨

f



20 mm/yr 20 mm/yr 20 mm/yr

s

l

e

d

¦©¨

o

M

c

i t

e Slip rate (mm/yr) Slip rate (mm/yr) Slip rate (mm/yr)

h

t

0 ¡ ¢ 0 ¡ ¢

0 ¡ ¢

¦§¨

n

y

S

g

20 mm/yr 20 mm/yr 20 mm/yr

¦©¨

t

u

p

t

¦§¨ u Slip rate (mm/yr) Slip rate (mm/yr)

O Slip rate (mm/yr)

0 ¡ ¢

0 ¡ ¢

0 ¡ ¢

j



k

©¨ §¨ ¨ ¨ ¨ ©¨

©¨ §¨ ¨ ¨ ¨ ©¨

©¨ §¨ ¨ ¨ ¨ ©¨

Figure 16. Tests of the ability of the data to resolve lateral and depth-dependent variations in coupling along the Greater-Lesser An- tilles subduction plate interface. Input synthetic forward models are shown on panels a-b-c and g-h-i, corresponding outputs from an inver- sion using the velocities predicted by the forward model are shown on panels d-e-f and j-k-l. It is readily apparent that the data distribution allows us to determine lateral variations in plate coupling with con- fidence, but that depth-dependent variations are poorly constrained by the data.

63 a

20 mm/yr b 20 mm/yr

§¥£

¦£ © Strain accumulation rate (mm/yr) Strain accumulation rate (mm/yr)

0 2 4 6 8 10 12 14 16 18 20 22 0 2 4 6 8 10 12 14 16 18 20 22

§£

© c

20 mm/yr d 20 mm/yr

§¥£

¦£ © Strain accumulation rate (mm/yr) Strain accumulation rate (mm/yr)

0 2 4 6 8 10121416182022 0 2 4 6 8 10 12 14 16 18 20 22

§£

©

¡¢£ ¡¤£ ¡¥£ ¦¦£ ¦§£ ¨¢£

¡¢£ ¡ ¤£ ¡ ¥£ ¦¦£ ¦§£ ¨¢£

Figure 17. Synthetic forward model (a) and recovered slip distribution adding 100 (b), 20 (c), or 10 (c) fictitious offshore GPS sites to the existing GPS sites used in this study.

65



A

N   

    

'>

¨©

£¥

'@

s    s!"

ML

¤§

2

K

'

3

MQ

I

J

U

O

0

0

.

I

1

¤

,

H

L

OP

UQ

G

>

F

E

*

V

s   

#$ % %&

B C    

?

¨©

£¥

)

TS

RS

¤§

)

2

()

0

/

.

-

s  s86<

,

+

1

¤

75!! = s"5  !59

% $

*

'))

D

:

;

¡ ¢ £ ¤ ¥ ¦ §

'()

) () ')) '() )) ()

3 3

4s5 6 7 89 %

Figure 19. GPS velocities shown with respect to the Caribbean plate (A) and to the North American plate (B). Error ellipses are 95% confidence. North-south profile including GPS sites shown with the dashed box on plate A. Velocities are projected onto directions parallel (blue) and normal (red) to the EPGF direction. MS = Massif de la Selle, CdS = Cul-de-Sac basin, MN = Matheux-Neiba range, PC = Plateau Central, PN = Plaine du Nord, EF = Enriquillo fault, SF = Septentrional fault.

66

6

©

¨

4

§

§

¦

2

¥

i

¤

£

¢

e

0

¡

i

t

S

-2

57)(%9%:;37)(%9%

;'3%)(%9%: )(%9%

<

8

$r%&'(%)*, /','3&,

©

¨

6

§

§

¦

4

¥

i

¤

£

¢

2

¥

i

D

0

Ma a

!"

Cua

#

Mau

G

0

§

¡

¦



t

.

¥

1

0

e

+

D

2 6

0 0 40 0

 m

Figure 20. Top and middle: comparison between the best-fit model (solid lines) and GPS observations for the strike-slip (blue) and dip- slip (red) components. Bottom: interpretative geological cross-section using information from Saint Fleur et al., 2015 Saint Fleur et al. [2015]. Red lines indicate model faults, the locked portion is shown as solid. The data density is not sufficient to distinguish between a one fault– and a two fault–model but requires a south-dipping fault with 9±3 mm/yr of reverse motion in addition 6±2 mm/yr of strike-slip motion. The surface trace of the fault in the best-fit model coincides with the northern limb of the Ganthier fold, indicated by the letter G. The gradient of GPS velocities coincides with the southern edge of the Cul-de-Sac basin, while the Matheux range appears devoid from present-day strain accumulation. D = Dumay locale where Terrier et al., 2014 Terrier et al. [2014] report reverse faulting affecting Quater- nary sediments. G = Ganthier fold Mann et al. [1995a].

67

1 1

1

1











0 0

0

 0









 

0 0



0



0







0 0



0

0











02 02

02

02



N

9 3 7

1 ¡

¡ 1 1¡ 2 2¡

  

12 1 2   1 12 1 1

2

D¢£¤ ¥ ¦¢£ § ¨© ¥ § S©§¢# ¤ ¥ ¦¢£ § ¨© ¥ § D¢£ ¨ !¦ " ! §  L$%#¢ !" £©& # 

+ ')

(

+

(

P

')

P

O

P P

J

O O

M

[

K

M M

I

('

K K

(R

J +

(

a

('

J J

`

J

E

I

J J

I I

_

E

8

('

^

C

()

E E

H

C C

G

H H

]

F

8

G G

\

F F

_

C

[

E E

[

8

C

C C

*

O

Z

R

`

B B

Y

g

X

*

*

+

' -' () (* Q) () c) R) b) () c) R) b) Q) () c) R) b) Q)

,./ 4 56 :;4< =>>?@/A dUe;.f4< T4< UVW;6 T4< UVW;6 -

Figure 21. Top row: parameter estimates for strike-slip rate, dip-slip rate, locking depth, and locking depth. Grey area shows the standard deviation. Bottom row: covariations of parameter estimates strike- slip rate / dip-slip rate, and locking depth / fault dip angle.

69

Which GPS sites contribute to the strain accumulation along the Ganthier Fault?

240 250

220

200 200

180 150

 160

100 140

120 50

100

80 0 100 110 120 130 140 150 160 170 180 190 200 40 45 50 55 60 65 70 75 80 Distance along profile (km) Width of Profile (km)

Figure 23. SUPPLEMENTAL: Site selection.

70

CFGH IJKLMNO HPQR L TN UJR GPOLWMJXFXNYPZPLMNO H PQR L TN UJR

8

S    S   

6

¦

©

¨

¨

§

¦

¥

¤

£

0

¢

¡

V

-

-

0

S !"  

S ! "     

¦

©

¨

¨

§

0

¦

¥

¤

£

¢

¡

-

V

- 0

B000

0

0

o

45

¨

o

45

o

3

7

5

o

§

7

5

o

9

5

o

2

9

5

¨

o

¥

§

:5

o

:5

1

o

;

5

A

o

;

5

o

¢

<5

o

¤

/

<5

¥

o

=5

o

- 0

@

=5

o

>

? 5

o

>

5

¡

E

00 0 00 0 0

0 0 00 0

D#$%&' ()*+,.

D#$%&'()*+,.

Figure 24. SUPPLEMENTAL: Comparison between GPS observa- tions and model results for a range of surface fault trace location and dip angles.

TABLES 71

Table 1. χ2 variations amongst tested model and associated F-ratio test results. See also Figure 7. DOF = degrees of freedom, P = probability of rejection of the null hypothesis.

2 ModelA ModelB DOF-A DOF-B Δ(χ ) Fratio P (%) Model1 Model2 588 579 1976.145 47.023 0.01 Model2 Model3 579 570 1320.590 60.476 0.01 Model3 Model4 570 567 222.050 36.149 0.01 Model4 Model5 567 567 -91.539 - - Model5 Model6 567 564 15.424 2.5313 5.67 Model5 Model7 567 567 9.608 - - Model8 Model5 570 567 10.128 1.6488 17.7 Model5 Model9 567 567 153.782 - - Model9 Model10 567 564 85.456 17.430 0.01 Model9 Model11 567 564 11.796 2.228 8.39 72 zz C yz C 2 yy /Myr 2 C radian xz 10 C − xy C xx C z Ω y Ω /Myr 10 x Table 2. ◦◦ ◦ ◦ /Myr ◦ uncert. smaj smin azim Ω ω /Myr ◦ E ◦ Angular velocity Rotation Error ellipse Rotation vector Covariance elements N ◦ Angular velocity estimates withfor respect to the the best North fitvalues. American model plate smaj, described smin, inlength, and the semi-minor azim text axis are, and length,clockwise respectively, previously from and the reported azimuth north semi-major of forthe axis angular the the velocity semi-major 95% estimate. axis confidence ellipse associated with Block/Plate lat. lon. This work: CaribbeanSouth AmericaPuerto -71.60 Rico 16.61 47.77Maracaibo -53.58 0.179 0.136 -34.76Andes 107.47 0.004North 0.006 Venezuela 0.24 0.518 -0.26Gonave 8.0 2.6 -65.42 0.138 -64.37North 0.635 0.8 Hispaniola 1.7 2.40 11.6 0.687 17.37Dom. 15.2 83.8 0.180 Republic -69.31 0.6 -65.73 0.194 0.0380 0.0774 10.24Benford 1.003 et -24.34 70.6 8.4 0.234 al., -0.1049 0.0418 2012: -74.11 9.7 109.98 -0.1278Caribbean 0.0389 -0.1698 0.054 1.0 0.533 0.282 0.4059 1.115 1.5Puerto 126.7 71 Rico 42 -0.2954 0.9 15.9 0.047 118.0 0.168 0.2642 9085Gonave -73.80 0.2972 -136 0.3 -0.5774 3.5 -50 -20935 1.9 -0.6194 2.1 21.00DeMets 0.0026 34.00 141.4 154.7 et 7575 al., -0.0032 44 2010 0.5 0.3541 8971 0.0918 -73.20 (MORVEL): 0.192 0.4 -1 12712Caribbean 48338 -0.9375 -0.2036 -28470 103.2 -36097 0.605 -17482 87.6 276 -4.10 0.004DeMets 0.0420 0.0698 0.1436 5132 7945 et al., 6336 -0.3471 83 2000: 0.135 23501 106.50 -0.5045 104016 657 90980Caribbean 0.9548 9.6 -67220 73.90 -88 0.0948 -22854 0.473 -16387 -0.4595 25268 -2321 9.4 -147.40 2.1 5176 1 373 3013 193030 9265 0.065 0.190 32 268 0.7 -72514 -25135 4.2 -1452 64.9 27361 112.5 0.005 5.0 10 9081 8352 0.0500 495 0.1450 250.5 68419 1.1 0.0192 8.2 -0.4802 -962 0.214 -24730 5866 -0.1844 0.3383 80.4 1.5 8957 126 -1994 -0.1340 8642 48 0.030 169.6 0.4524 -19831 -0.0444 688 29.2 -108 -0.0338 7143 -0.0284 3.0 0.1825 769 45684 30 -35.0 -16448 -2880 46 -0.0303 5935 513 -0.0647 987 -109 -0.2017 11494 -125 29 – -3934 50 1358 356 – -94 – 56 – – – 73

Table 3. Slip deficit rates in mm/yr for major faults (minim and maximum strike-slip and dip-slip values are provided) for the best-fit model and for a model with the same geometry where full locking is enforced at the Greater–Lesser Antilles subduction.

Best-fit model Locked Subduction Uncertainties

Faults strike-slip strike-slip strike-slip strike-slip σSS σDS max min max min max min max min Oriente Fault 10.3 8.9 0 0 10.3 8.9 0 0 0.4 0.0 Septentrional Haiti 10 8.8 0 0 10.0 8.8 0 0 0.7 0 Septentrional D.R. 10.1 9.7 0 0 10.1 9.7 0 0 0.7 0 South Jamaica 9.4 5.7 0 0 9.4 8.7 0 0 0.3 0 EPGF Haiti 10.3 8.1 0 0 10.3 8.1 0 0 0.3 0.0 Muertos West 4.9 1.5 13.7 5.1 4.9 1.5 13.7 5.1 0.5 0.5 Muertos East 0.4 0 1.5 0 0.4 0 1.5 0 0.4 0.5 Anegada Passage 0.5 0.3 0 0 0.5 0.3 0 0 0.5 0.0 Mona Passage 2.3 0.3 0 0 2.3 0.3 0 0 0.8 0.0 North Hispaniola 3.1 0 4.2 2.5 1.2 0.4 4.6 1.5 0.8 1.2 Puerto Rico Trench 0.3 0 1.0 0.2 15.5 12.6 12.8 3.3 0.5 0.7 Lesser Antilles N 0.3 0 2.2 1.2 12.8 2.6 19.2 13.6 0.2 0.4 Lesser Antilles S 0.3 0 3.1 2.3 18.5 0.6 20.2 8.5 0.3 0.3 El Pilar Fault 20.3 15.9 0 0 20.3 15.9 0 0 0.5 0.0 Oca Fault 1.5 1.2 0 0 1.5 1.2 0 0 1.0 0.0 Bocono Fault 12.7 10.9 0 0 12.7 10.9 0 0 0.8 0.0 Santa Marta Fault 2.3 2 0 0 2.3 2 0 0 0.7 0.0 Venezuela thrust 3.2 0.2 5.3 0.3 3.2 0.2 5.3 0.3 1.6 1.6 Colombia thrust 5.6 0.8 9.8 5.4 5.6 0.8 9.8 5.4 0.9 0.9 74

Table 4. Estimated model parameters and associated uncertainties.

Estimated Parameters Average Optimum Std. deviation Strike–slip rate (mm/yr) 5.7 6.0 2.4 Dip–slip rate (mm/yr) 8.9 9.0 2.8 Dip angle (◦) 37.0 36.0 13.0 Locking depth (km) 5.0 7.0 3.5 REFERENCES 75

REFERENCES

Syed Tabrez Ali, Andrew M Freed, Eric Calais, David M Manaker, and William R McCann. Coulomb stress evolution in Northeastern Caribbean over the past 250 years due to coseismic, postseismic and interseismic deformation. Geophysical Jour- nal International, 174(3):904–918, September 2008. T I Allen and D J Wald. On the Use of High-Resolution Topographic Data as a Proxy for Seismic Site Conditions (VS30). Bulletin of the Seismological Society of America, 99(2A):935–943, April 2009. Zuheir Altamimi, Xavier Collilieux, and Laurent M´etivier. Itrf2008: an improved solution of the international terrestrial reference frame. Journal of Geodesy, 85(8): 457–473, 2011. Donald F Argus. Interseismic strain accumulation and anthropogenic motion in metropolitan Los Angeles. Journal of Geophysical Research, 110:B04401, 2005. Donald F Argus, Richard G Gordon, Michael B Heflin, Chopo Ma, Richard J Eanes, Pascal Willis, W Richard Peltier, and Susan E Owen. The angular velocities of the plates and the velocity of Earth’s centre from space geodesy. Geophysical Journal International, 180(3):913–960, March 2010. FA Audemard, M Machette, J Cox, R Dart, and K Haller. Map and database of quaternary faults and folds in venezuela and its offshore regions. USGS Open-File report 00-0018, 2000. Franck Audemard, Daniela Pantosti, Michael Machette, Carlos Costa, Koji Oku- mura, Hugh Cowan, Hans Diederix, and Carlos Ferrer. Trench investigation along the m´erida section of the bocon´o fault (central venezuelan andes), venezuela. Tectono- physics, 308(1):1–21, 1999. Franck A Audemard, Gloria Romero, Herbert Rendon, and Victor Cano. Quaternary fault kinematics and stress tensors along the southern Caribbean from fault-slip data and focal mechanism solutions. Earth Science Reviews, 69(3-4):181–233, 2005. Stephen Babb and Paul Mann. Structural and sedimentary development of a neogene transpressional plate boundary between the caribbean and south america plates in trinidad and the gulf of paria. Sedimentary Basins of the World, 4:495–557, 1999. W H Bakun, C H Flores, and U S ten Brink. Significant Earthquakes on the En- riquillo Fault System, Hispaniola, 1500-2010: Implications for Seismic Hazard. Bul- letin of the Seismological Society of America, 102(1):18–30, February 2012a. William H Bakun, Claudia H Flores, and S Uri. Significant earthquakes on the enriquillo fault system, hispaniola, 1500–2010: Implications for seismic hazard. Bul- letin of the Seismological Society of America, 102(1):18–30, 2012b. 76

Nathan L Bangs, Gail L Christeson, and Thomas H Shipley. Structure of the lesser antilles subduction zone backstop and its role in a large accretionary system. Journal of Geophysical Research: Solid Earth (1978–2012), 108(B7), 2003. B Benford, C DeMets, and E Calais. GPS estimates of microplate motions, northern Caribbean: evidence for a Hispaniola microplate and implications for earthquake hazard. Geophysical Journal International, 191(2):481–490, September 2012a. B Benford, C DeMets, and Eric Calais. GPS estimates of microplate motions, north- ern Caribbean: evidence for a Hispaniola microplate and implications for earthquake hazard. Geophysical Journal International, 191(2):481–490, September 2012b. B Benford, C DeMets, B Tikoff, P Williams, L Brown, and M Wiggins-Grandison. Seismic hazard along the southern boundary of the gˆonave microplate: block mod- elling of gps velocities from jamaica and nearby islands, northern caribbean. Geo- physical Journal International, 190(1):59–74, 2012c. Pascal Bernard and J´erˆome Lambert. Subduction and seismic hazard in the northern Lesser Antilles: revision of the historical seismicity. Bulletin of the Seismological Society of America, 78(6):1965–1983, 1988. Gregory C Beroza. Near-source modeling of the Loma Prieta earthquake: Evidence for heterogeneous slip and implications for earthquake hazard. Bulletin of the Seis- mological Society of America, 81(5):1603–1621, October 1991. Roger Bilham. Structural geology: Invisible faults under shaky ground. Nature Geoscience, 3(11):743–745, November 2010. David M Boore and Gail M Atkinson. Ground-Motion Prediction Equations for the Average Horizontal Component of PGA, PGV, and 5sand 10.0 s. Earthquake Spectra, 24(1):99–138, February 2008. B Bourgueil, P Andreieff, J Lasnier, R Gonnard, J Le Metour, and Jean-Philippe Rancon. Synth`ese G´eologique de la R´epublique d’Haiti. Technical report, Bureau des Mines et de l’Energie, Port-au-Prince, Haiti, October 1988. John J Braun, Glen S Mattioli, Eric Calais, David Carlson, Timothy H Dixon, Michael E Jackson, E Robert Kursinski, Hector Mora-Paez, M Meghan Miller, Rajul Pandya, Richard Robertson, and Guoquan Wang. Focused study of interweaving hazards across the Caribbean. Eos, 93(9):89–90, February 2012. Uri ten Brink, William Danforth, Christopher Polloni, Brian Andrews, Pilar Llanes, Shepard Smith, Eugene Parker, and Toshihiko Uozumi. New seafloor map of the puerto rico trench helps assess earthquake and tsunami hazards. Eos, Transactions American Geophysical Union, 85(37):349–354, 2004. A Bruneton, R Gonnard, Y Gou, J Lasnier, and P Negroni. Evaluation du potentiel p´etrolier, Synth`ese g´eologiquedelaR´epublique d’Haiti. Technical Report BME/BID ATN-SA 2506-HA, Bureau des Mines et de l’Energie, Port-au-Prince, Haiti, 1988. Byrne, G DB, WR Suarez, and McCann. Muertos trough subduction–microplate tectonics in the northern caribbean? Nature Geoscience, 317(6036):420–421, 10 1985. 77

E Calais and B Mercier de L´epinay. A natural model of active transpressional tectonics the en ´echelon structures of the oriente deep, along the northern caribbean transcurrent plate boundary (southern cuban margin). Oil & Gas Science and Technology, 45(2):147–160, 1990. E Calais and B Mercier de L´epinay. The northern Caribbean transcurrent plate boundary in Hispaniola: paleogeographic and structural evolution during the Ceno- zoic. Bull.Soc.G´eol. France, 163:309–324, 1992. E Calais, JY Han, C DeMets, and JM Nocquet. Deformation of the north american plate interior from a decade of continuous gps measurements. Journal of Geophysical Research: Solid Earth (1978–2012), 111(B6), 2006. Eric Calais and Bernard Mercier de Lepinay. From transtension to transpression along the northern Caribbean plate boundary off Cuba: implications for the Recent motion of the Caribbean plate. Tectonophysics, 186(3-4):329–350, February 1991. Eric Calais, Nicole B´ethoux, and Bernard Mercier L´epinay. From transcurrent fault- ing to frontal subduction: A seismotectonic study of the northern caribbean plate boundary from cuba to puerto rico. Tectonics, 11(1):114–123, 1992. Eric Calais, Y Mazabraud, B Mercier de Lepinay, Paul Mann, Glen S Mattioli, and P Jansma. Strain partitioning and fault slip rates in the northeastern Caribbean from GPS measurements. Geophysical Research Letters, 29(18):1856, 2002a. Eric Calais, Yves Mazabraud, Bernard Mercier de L´epinay, Paul Mann, Glen Matti- oli, and Pamela Jansma. Strain partitioning and fault slip rates in the northeastern caribbean from gps measurements. Geophysical Research Letters, 29(18):3–1, 2002b. Eric Calais, Andrew Freed, Glen Mattioli, Falk Amelung, Sigurj´on J´onsson, Pamela Jansma, Sang-Hoon Hong, Timothy Dixon, Claude Pr´epetit, and Roberte Mom- plaisir. Transpressional rupture of an unmapped fault during the 2010 haiti earth- quake. Nature Geoscience, 3(11):794–799, 2010. Eric Calais, Steeve Symithe, Bernard de L´epinay, and Claude Pr´epetit. Plate bound- ary segmentation in the northeastern Caribbean from geodetic measurements and Neogene geological observations. Comptes rendus - Geoscience, 348(1):42–51, Jan- uary 2016. TCalmus.Contribution `al’´etude g´eologique du massif de Macaya (sud-ouest d’Haiti, Grandes Antilles), sa place dans l’orog`ene nord caraibe. PhD thesis, Paris, France, 1983. Kenneth W Campbell and Yousef Bozorgnia. NGA Ground Motion Model for the Geometric Mean Horizontal Component of PGA, PGV, PGD and 5Ranging from 0.01to10 s. Earthquake Spectra, 24(1):139–171, February 2008. BrianS-J Chiou and Robert R Youngs. An NGA Model for the Average Horizontal Component of Peak Ground Motion and Response Spectra. Earthquake Spectra,24 (1):173–215, July 2008. Mohamed Chlieh, Hugo Perfettini, Hernando Tavera, Jean-Philippe Avouac, Do- minique Remy, Jean-Mathieu Nocquet, Fr´ed´erique Rolandone, Francis Bondoux, Germinal Gabalda, and Sylvain Bonvalot. Interseismic coupling and seismic poten- tial along the Central Andes subduction zone. Journal of Geophysical Research- Atmospheres, 116(B12):B12405, December 2011. 78

Nadaya Cubas, Jean-Philippe Avouac, Pauline Souloumiac, and Yves Leroy. Megathrust friction determined from mechanical analysis of the forearc in the maule earthquake area. Earth and Planetary Science Letters, 381:92–103, 2013. Charles DeMets and Timothy H Dixon. New kinematic models for pacific-north america motion from 3 ma to present, i: Evidence for steady motion and biases in the nuvel-1a model. Geophysical Research Letters, 26(13):1921–1924, 1999. Charles DeMets, Richard G Gordon, Donald F Argus, and Seth Stein. Effect of recent revisions to the geomagnetic reversal time scale on estimates of current plate motions. Geophysical research letters, 21(20):2191–2194, 1994. Charles DeMets, Pamela E Jansma, Glen S Mattioli, Timothy H Dixon, Fred Farina, Roger Bilham, Eric Calais, and Paul Mann. Gps geodetic constraints on caribbean- north america plate motion. Geophysical Research Letters, 27(3):437–440, 2000. Charles DeMets, Richard G Gordon, and Donald F Argus. Geologically current plate motions. Geophysical Journal International, 181(1):1–80, 2010. Hans Diederix, Catalina Hern´andez, Eliana Torres, Jairo Alonso Osorio, and Paola Botero. Resultados preliminares del primer estudio paleosismol´ogico a lo largo de la falla de bucaramanga, colombia. Ingenier´ıa Investigaci´on y Desarrollo, 9(2), 2012. Lynn D Dietz and William L Ellsworth. The October 17, 1989, Loma Prieta, Califor- nia, Earthquake and its aftershocks: Geometry of the sequence from high-resolution locations. Geophysical Research Letters, 17(9):1417–1420, August 1990. William P Dillon, N Terence Edgar, Kathryn M Scanlon, and Dwight F Coleman. A review of the tectonic problems of the strike-slip northern boundary of the caribbean plate and examination by gloria. Geology of the United States’ Seafloor: The View From GLORIA, pages 135–164, 1996. Timothy H Dixon, Frederic Farina, Charles DeMets, Pamela Jansma, Paul Mann, and Eric Calais. Relative motion between the caribbean and north american plates and related boundary zone deformation from a decade of gps observations. Journal of Geophysical Research: Solid Earth (1978–2012), 103(B7):15157–15182, 1998. James F Dolan and David D Bowman. Tectonic and seismologic setting of the 22 september 2003, puerto plata, dominican republic earthquake: implications for earthquake hazard in northern hispaniola. Seismological Research Letters, 75(5): 587–597, 2004. James F Dolan, Henry T Mullins, and David J Wald. Active tectonics of the north- central caribbean: Oblique collision, strain partitioning, and opposing subducted slabs. SPECIAL PAPERS-GEOLOGICAL SOCIETY OF AMERICA, pages 1–62, 1998. JE Dolan and David J Wald. The 1943-1953 north-central caribbean earthquakes: Active tectonic setting, seismic hazards, and implications for caribbean-north amer- ica plate motions. SPECIAL PAPERS-GEOLOGICAL SOCIETY OF AMERICA, pages 143–169, 1998. J Dorel. Seismicity and seismic gap in the lesser antilles arc and earthquake hazard in guadeloupe. Geophysical Journal International, 67(3):679–695, 1981. 79

R Douilly, J S Haase, W L Ellsworth, M P Bouin, Eric Calais, S J Symithe, J G Armbruster, B M de Lepinay, A Deschamps, S L Mildor, M E Meremonte, and S E Hough. Crustal Structure and Fault Geometry of the 2010 Haiti Earthquake from Temporary Seismometer Deployments. Bulletin of the Seismological Society of America, 103(4):2305–2325, July 2013. R Douilly, H Aochi, Eric Calais, and A M Freed. Three-dimensional dynamic rupture simulations across interacting faults: The Mw7.0, 2010, Haiti earthquake. Journal of Geophysical Research: Solid Earth, 120(2):1108–1128, February 2015. L Duplan. Carte Structurale de la R´epublique d’Haiti, R´egion Sud, April 1975. G Ekstrom, M Nettles, and A M Dziewo´nski. Physics of the Earth and Planetary Interiors. Physics of the earth and planetary interiors, 200-201(C):1–9, June 2012. E Robert Engdahl, Rob van der Hilst, and Raymond Buland. Global teleseismic earthquake relocation with improved travel times and procedures for depth deter- mination. Bulletin of the Seismological Society of America, 88(3):722–743, 1998. J Escartin, DK Smith, and M Cannat. Parallel bands of seismicity at the MidAtlantic ridge, 12–14n. Geophysical Research Letters, 30(12), 2003. doi: 10.1029/2003GL017226. URL http://dx.doi.org/10.1029/2003GL017226. N Feuillet, I Manighetti, and P Tapponnier. Active extension perpendicular to subduction in the lesser antilles island arc; guadeloupe, french antilles. CR Acad. Sci., Ser. II, 333(9):583–590, 2001. N Feuillet, I Manighetti, P Tapponnier, and E Jacques. Arc parallel extension and localization of volcanic complexes in guadeloupe, lesser antilles. Journal of Geophysical Research: Solid Earth (1978–2012), 107(B12):ETG–3, 2002. Nathalie Feuillet, Fr´ed´erique Leclerc, Paul Tapponnier, Fran¸cois Beauducel, Georges Boudon, Anne Le Friant, Christine Deplus, Jean-Fr´ed´eric Lebrun, Alexandre Ner- cessian, Jean-Marie Saurel, and Valentin Cl´ement. Active faulting induced by slip partitioning in Montserrat and link with volcanic activity: New insights from the 2009 GWADASEIS marine cruise data. Geophysical Research Letters, 37(19), April 2010. Edward H Field, Hope A Seligson, Nitin Gupta, Vipin Gupta, Thomas H Jordan, and Kenneth W Campbell. Loss Estimates for a Puente Hills Blind-Thrust Earthquake in Los Angeles, California. Earthquake Spectra, 21(2):329–338, May 2005. Arthur Frankel, William R McCann, and Andrew J Murphy. Observations from a seismic network in the virgin islands region: Tectonic structures and earthquake swarms. Journal of Geophysical Research: Solid Earth (1978–2012), 85(B5):2669– 2678, 1980. Arthur Frankel, Stephen Harmsen, Charles Mueller, Eric Calais, and Jennifer Haase. Seismic Hazard Maps for Haiti. Earthquake Spectra, 27(S1):S23–S41, October 2011. Jean-Paul Gestel, Paul Mann, James F Dolan, and Nancy R Grindlay. Structure and tectonics of the upper cenozoic puerto rico-virgin islands carbonate platform as determined from seismic reflection studies. Journal of Geophysical Research: Solid Earth (1978–2012), 103(B12):30505–30530, 1998. 80

R Gilles, D Bertil, C Pr´epetit, M Belvaux, A Roull´e, J Jean-Philippe, and G Noury. Seismic microzoning in the metropolitan area of Port-au-Prince: Complexity of the subsoil. In American Geophysical Union Fall Meeting, pages Abstract S51A–2312, San Francisco, 2013. P Goreau. The Tectonic Evolution of the North Central Caribbean Plate Margin. Master’s thesis, MIT, January 1981. Shannon E Graham, Charles Demets, Heather R DeShon, Robert Rogers, Manuel Rodriguez Maradiaga, Wilfried Strauch, Klaus Wiese, and Douglas Her- nandez. GPS and seismic constraints on the M = 7.3 2009 Swan Islands earthquake: implications for stress changes along the Motagua fault and other nearby faults. Geophysical Journal International, 190(3):1625–1639, June 2012. JL Granja Bru˜na, A Carb´o-Gorosabel, P Llanes Estrada, A Mu˜noz-Mart´ın, US ten Brink, M G´omez Ballesteros, M Druet, and A Pazos. Morphostructure at the junc- tion between the beata ridge and the greater antilles island arc (offshore hispaniola southern slope). Tectonophysics, 618:138–163, 2014. Jos´e Luis Granja Bru˜na, Uri S Ten Brink, Andr´es Carb´o-Gorosabel, Alfonso Mu˜noz- Mart´ın, and Mar´ıa G´omez Ballesteros. Morphotectonics of the central muertos thrust belt and muertos trough (northeastern caribbean). Marine geology, 263(1):7–33, 2009. N Grindlay, Ps Mann, and J Dolan. Researchers investigate submarine faults north of puerto rico. Eos, Transactions American Geophysical Union, 78(38):404–404, 1997. Nancy R Grindlay, Meghan Hearne, and Paul Mann. High risk of tsunami in the northern caribbean. Eos, Transactions American Geophysical Union, 86(12):121– 126, 2005a. NR Grindlay, LJ Abrams, L Del Greco, and P Mann. Toward an integrated un- derstanding of holocene fault activity in western puerto rico: constraints from high- resolution seismic and sidescan sonar data, in active tectonics and seismic hazards of puerto rico, the virgin islands, and offshore areas. Geol. Soc. Am. Spec. Paper, 385:139–160, 2005b. Marc-Andre Gutscher, Graham K Westbrook, Boris Marcaillou, David Graindorge, Audrey Gailler, Thibaud Pichot, and Ren´e C Maury. Along strike variations in the width of the seismogenic zone of the lesser antilles subduction predicted by thermal modeling. In EGU General Assembly Conference Abstracts, volume 12, page 10600, 2010. Manabu Hashimoto, Yo Fukushima, and Yukitoshi Fukahata. Fan-delta uplift and mountain subsidence during the Haiti 2010 earthquake. Nature Geoscience, 4(4): 1–5, March 2011. G P Hayes, R W Briggs, A Sladen, E J Fielding, C Prentice, K Hudnut, P Mann, F W Taylor, A J Crone, R Gold, T Ito, and M Simons. Complex rupture during the 12 January 2010 Haiti earthquake. Nature Geoscience, 3(11):800–805, October 2010a. 81

Gavin P Hayes, Daniel E McNamara, Lily Seidman, and Jean Roger. Quantifying potential earthquake and tsunami hazard in the lesser antilles subduction zone of the caribbean region. Geophysical Journal International, 196(1):510–521, 2014. GP Hayes, RW Briggs, A Sladen, EJ Fielding, C Prentice, K Hudnut, Paul Mann, FW Taylor, AJ Crone, R Gold, et al. Complex rupture during the 12 january 2010 haiti earthquake. Nature Geoscience, 3(11):800–805, 2010b. T. Herring. MATLAB Tools for viewing GPS velocities and time series. GPS Solu- tions, 7:194–199, 2003. doi: 10.1007/s10291-003-0068-0. T Herring, B King, and S McClusky. Introduction to gamit/globk reference manual global kalman filter vlbi and gps analysis program. release 10.3. eaps, 2010a. T Herring, R King, and S McClusky. Gamit and globk reference manuals, release 10.3. Mass. Inst. of Technol., Cambridge, 2010b. C Heubeck, P Mann, J Dolan, and S Monechi. Diachronous uplift and recycling of sedimentary basins during cenozoic tectonic transpression, northeastern caribbean plate margin. Sedimentary Geology, 70:1–32, 1991. G E Hilley, R B¨urgmann, and P Z Zhang. Bayesian inference of plastosphere vis- cosities near the Kunlun Fault, northern Tibet. Geophysical Research Letters, 32: L01302, 2005. Susan E Hough. Missing great earthquakes. Journal of Geophysical Research-Solid Earth, 118(3):1098–1108, March 2013. R D Hyndman. The seismogenic zone of subduction thrust faults. The seismogenic zone of subduction thrust faults, pages 15–40, 2007. J Idarraga-Garia and J Romero. Journal of South American Earth Sciences. Journal of South American Earth Sciences, 29(4):849–860, October 2010. Pamela E Jansma, Glen S Mattioli, Alberto Lopez, Charles DeMets, Timothy H Dixon, Paul Mann, and Eric Calais. Neotectonics of puerto rico and the virgin islands, northeastern caribbean, from gps geodesy. Tectonics, 19(6):1021–1037, 2000. I Jany, KM Scanlon, and A Mauffret. Geological interpretation of combined seabeam, gloria and seismic data from anegada passage (virgin islands, north caribbean). Marine Geophysical Researches, 12(3):173–196, 1990a. I Jany, KM Scanlon, and A Mauffret. Geological interpretation of combined seabeam, gloria and seismic data from anegada passage (virgin islands, north caribbean). Marine Geophysical Researches, 12:173–196, 1990b. K M Johnson. Viscoelastic earthquake cycle models with deep stress-driven creep along the San Andreas fault system. Journal of Geophysical Research, 109:B10403, 2004. Fran¸cois Jouanne, Franck A Audemard, Christian Beck, Aurelien Van Welden, Reinaldo Ollarves, and Carlos Reinoza. Present-day deformation along the el pi- lar fault in eastern venezuela: Evidence of creep along a major transform boundary. Journal of Geodynamics, 51(5):398–410, 2011. 82

Kazuki Koketsu, Yusuke Yokota, Naoki Nishimura, Yuji Yagi, Shin’ichi Miyazaki, Kenji Satake, Yushiro Fujii, Hiroe Miyake, Shin’ichi Sakai, Yoshiko Yamanaka, and Tomomi Okada. Earth and Planetary Science Letters. Earth and Planetary Science Letters, 310(3-4):480–487, October 2011. H Kopp, W Weinzierl, A Becel, P Charvis, M Evain, E R Flueh, A Gailler, A Galve, A Hirn, A Kandilarov, D Klaeschen, M Laigle, C Papenberg, L Planert, and E Roux. Deep structure of the central Lesser Antilles Island Arc: Relevance for the formation of continental crust. Earth and Planetary Science Letters, 304(1-2):121–134, April 2011. Margaret E Kroehler, Paul Mann, Alejandro Escalona, Gail Christeson, et al. Late cretaceous-miocene diachronous onset of back thrusting along the south caribbean deformed belt and its importance for understanding processes of arc collision and crustal growth. Tectonics, 30(6), 2011. John W Ladd, J Lamar Worzel, and Joel S Watkins. Multifold seismic reflection records from the northern venezuela basin and the north slope of the muertos trench. Island Arcs, Deep Sea Trenches and Back-Arc Basins, pages 41–56, 1977. M Laigle, A Hirn, M Sapin, A Becel, P Charvis, E Flueh, J Diaz, J F Lebrun, A Gesret, R Raffaele, A Galve, M Evain, M Ruiz, H Kopp, G Bayrakci, W Weinzierl, Y Hello, J C L´epine, J P Viod´e, M Sachpazi, J Gallart, E Kissling, and R Nicolich. Tectonophysics. Tectonophysics, 603(C):1–20, September 2013. Xavier Le Pichon, Pierre Henry, and Siegfried Lallemant. Water flow in the Barbados Accretionary Complex. Journal of Geophysical Research-Solid Earth, 95(B):8945– 8967, June 1990. S Leroy and A Mauffret. Intraplate deformation in the caribbean region. Journal of Geodynamics, 21(1):113–122, 1996. AM Lopez, S Stein, T Dixon, G Sella, E Calais, P Jansma, J Weber, and P LaFem- ina. Is there a northern lesser antilles forearc block? Geophysical research letters, 33(7), 2006. John P Loveless and Brendan J Meade. Geodetic imaging of plate motions, slip rates, and partitioning of deformation in japan. Journal of Geophysical Research: Solid Earth (1978–2012), 115(B2), 2010. D M Manaker, Eric Calais, A M Freed, S T Ali, P Przybylski, G Mattioli, P Jansma, CPr´epetit, and J B de Chabalier. Interseismic Plate coupling and strain partitioning in the Northeastern Caribbean. Geophysical Journal International, 174(3):889–903, September 2008a. David M Manaker, Eric Calais, Andrew M Freed, S T Ali, P Przybylski, Glen S Mattioli, PE Jansma, C Pr´epetit, and J B De Chabalier. Interseismic Plate cou- pling and strain partitioning in the Northeastern Caribbean. Geophysical Journal International, 174(3):889–903, September 2008b. Michael Manga, Matthew J Hornbach, Anne Le Friant, Osamu Ishizuka, Nicole Stroncik, Tatsuya Adachi, Mohammed Aljahdali, Georges Boudon, Christoph Bre- itkreuz, Andrew Fraass, et al. Heat flow in the lesser antilles island arc and adjacent back arc grenada basin. Geochemistry, Geophysics, Geosystems, 13(8), 2012. 83

P Mann, F W Taylor, R L Edwards, and T L Ku. Actively evolving microplate formation by oblique collision and sideways motion along strike-slip faults: An ex- ample from the northeastern Caribbean plate margin. Tectonophysics, 246(1-3): 1–69, 1995a. Paul Mann, Carlos Schubert, and Kevin Burke. Review of caribbean neotectonics. In IN: The geology of North America. Vol. H-The Caribbean region. Boulder, CO, Geological Society of America, 1990, p. 307-338. Research supported by University of Texas, CONICIT, and Universidad de Los Andes., volume 1990, pages 307–338, 1990. Paul Mann, FW Taylor, R Lawrence Edwards, and Teh-Lung Ku. Actively evolv- ing microplate formation by oblique collision and sideways motion along strike-slip faults: An example from the northeastern caribbean plate margin. Tectonophysics, 246(1):1–69, 1995b. Paul Mann, Eric Calais, Jean-Claude Ruegg, Charles DeMets, Pamela E Jansma, and Glen S Mattioli. Oblique collision in the northeastern caribbean from gps mea- surements and geological observations. Tectonics, 21(6):7–1, 2002. DG Masson and Kathryn M Scanlon. The neotectonic setting of puerto rico. Geo- logical Society of America Bulletin, 103(1):144–154, 1991. Alain Mauffret and Sylvie Leroy. Neogene intraplate deformation of the caribbean plate at the beata ridge. Sedimentary Basins of the World, 4:627–669, 1999. G P Mavroeidis and C M Scotti. Finite-Fault Simulation of Broadband Strong Ground Motion from the 2010 Mw 7.0 Haiti Earthquake. Bulletin of the Seismolog- ical Society of America, 103(5):2557–2576, September 2013. St´ephane Mazzotti, Xavier Le Pichon, Pierre Henry, and Shin-Ichi Miyazaki. Full interseismic locking of the nankai and japan-west kurile subduction zones: An anal- ysis of uniform elastic strain accumulation in japan constrained by permanent gps. Journal of Geophysical Research: Solid Earth (1978–2012), 105(B6):13159–13177, 2000. Robert McCaffrey. Crustal block rotations and plate coupling. Plate boundary zones, pages 101–122, 2002. Robert McCaffrey, Laura M Wallace, and John Beavan. Slow slip and frictional transition at low temperature at the hikurangi subduction zone. Nature Geoscience, 1(5):316–320, 2008. William R McCann. Estimating the threat of tsunamogenic earthquakes and earth- quake induced-landslide tsunami in the Caribbean. World Scientific Publishing, Sin- gapore, 2006. William R Mccann and Lynn R Sykes. Subduction of aseismic ridges beneath the Caribbean plate: implications for the tectonics and seismic potential of the north- eastern Caribbean. Journal of Geophysical Research: Solid Earth (1978–2012),89 (B6):4493–4519, 1984. William R McCann and Lynn R Sykes. Subduction of aseismic ridges beneath the caribbean plate: implications for the tectonics and seismic potential of the north- eastern caribbean. Journal of Geophysical Research: Solid Earth (1978–2012),89 (B6):4493–4519, 1984. 84

William R Mccann, James W Dewey, Andrew J Murphy, and Samuel T Harding. A large normal-fault earthquake in the overriding wedge of the Lesser Antilles subduc- tion zone: The earthquake of 8 October 1974. Bulletin of the Seismological Society of America, 72(6A):2267–2283, 1982. Brendan J Meade and John P Loveless. Block modeling with connected fault-network geometries and a linear elastic coupling estimator in spherical coordinates. Bulletin of the Seismological Society of America, 99(6):3124–3139, 2009. Brendan J Meade, Bradford H Hager, Simon C McClusky, Robert E Reilinger, Semih Ergintav, Onur Lenk, Aykut Barka, and Haluk Ozener.¨ Estimates of seismic poten- tial in the marmara sea region from block models of secular deformation constrained by global positioning system measurements. Bulletin of the Seismological Society of America, 92(1):208–215, 2002. C Mendoza. Rupture history of the 1997 cariaco, venezuela, earthquake from tele- seismic p waves. Geophysical research letters, 27(10):1555–1558, 2000. Bernard Mercier de L´epinay, Anne Deschamps, Frauke Klingelhoefer, Yves Maz- abraud, Bertrand Delouis, Val´erie Clouard, Yann Hello, Jacques Crozon, Boris Mar- caillou, David Graindorge, Martin Vall´ee, Julie Perrot, Marie-Paule Bouin, Jean- Marie Saurel, Philippe Charvis, and Mildor St-Louis. The 2010 Haiti earthquake: A complex fault pattern constrained by seismologic and tectonic observations. Geo- physical Research Letters, 38:L22305, November 2011. Nicholas Metropolis, Arianna W Rosenbluth, Marshall N Rosenbluth, Augusta H Teller, and Edward Teller. Equation of State Calculations by Fast Computing Ma- chines. The Journal of Chemical Physics, 21(6):1087–7, 1953. Meghan S Miller, Alan Levander, Fenglin Niu, and Aibing Li. Upper mantle structure beneath the caribbean-south american plate boundary from surface wave tomogra- phy. Journal of Geophysical Research: Solid Earth (1978–2012), 114(B1), 2009. Roberte Momplaisir. Contribution `al’´etude g´eologique de la partie orientale du massif de la Hotte (Presqu’ile du Sud d’Haiti): synth`ese structurale des marges de la Presqu’ile. PhD thesis, Paris, 1986. Roberte Momplaisir and D Boisson. Carte G´eologique de la R´epublique d’Haiti, South-East sheet (Port-au-Prince). BME, 1988. S Mondziel, N Grindlay, P Mann, A Escalona, and L Abrams. Morphology, structure, and tectonic evolution of the mona canyon (northern mona passage) from multibeam bathymetry, sidescan sonar, and seismic reflection profiles. Tectonics, 29(2), 2010. doi: 10.1029/2008TC002441. URL http://dx.doi.org/10.1029/2008TC002441. DM¨uller and WF Smith. Deformation of the oceanic crust between the north american and south american plates. Journal of Geophysical Research: Solid Earth (19782012), 98(B5):8275–8291, 1993. doi: 10.1029/92JB02863. URL http://dx.doi.org/10.1029/92JB02863. JM Nocquet, JC Villegas-Lanza, Mohamed Chlieh, PA Mothes, F Rolandone, P Jar- rin, D Cisneros, A Alvarado, Laurence Audin, Francis Bondoux, et al. Motion of continental slivers and creeping subduction in the northern andes. Nature Geo- science, 7(4):287–291, 2014. 85

Yoshimitsu Okada. Surface deformation due to shear and tensile faults in a half- space. Bulletin of the seismological society of America, 75(4):1135–1154, 1985. Yoshimitsu Okada. Internal deformation due to shear and tensile faults in a half- space. Bulletin of the Seismological Society of America, 82(2):1018–1040, April 1992. Scott M Olson, Russell A Green, Samuel Lasley, Nathaniel Martin, Brady R Cox, Ellen Rathje, Jeff Bachhuber, and James French. Documenting Liquefaction and Lateral Spreading Triggered by the 12 January 2010 Haiti Earthquake. Earthquake Spectra, 27(S1):S93–S116, October 2011. G Paris, MN Machette, RL Dart, and KM Haller. Map and database of quaternary faults and folds in colombia and its offshore regions. USGS Open-File Report 00- 0284, 2000. M Patriat, T Pichot, G K Westbrook, M Umber, E Deville, F Benard, W R Roest, B Loubrieu, and the ANTIPLAC Cruise Party. Evidence for Quaternary convergence across the North America-South America plate boundary zone, east of the Lesser Antilles. Geology, 39(10):979–982, September 2011. Omar J P´erez, Roger Bilham, Rebecca Bendick, Jos´e R Velandia, Napole´on Hern´andez, Carlos Moncayo, Melvin Hoyer, and Mike Kozuch. Velocity field across the southern caribbean plate boundary and estimates of caribbean/south-american plate motion using gps geodesy 1994–2000. Geophysical Research Letters, 28(15): 2987–2990, 2001. J Perrot, E Calais, and B Mercier de L´epinay. Tectonic and kinematic regime along the northern caribbean plate boundary: New insights from broad-band modeling of the may 25, 1992, ms= 6.9 cabo cruz, cuba, earthquake. pure and applied geophysics, 149(3):475–487, 1997. G´erard Petit and Brian Luzum. Iers conventions (2010). Technical report, DTIC Document, 2010. T Pichot, Martin Patriat, GK Westbrook, Thierry Nalpas, Marc-Andr´e Gutscher, WR Roest, E Deville, Maryline Moulin, D Aslanian, and Marina Rabineau. The cenozoic tectonostratigraphic evolution of the barracuda ridge and tiburon rise, at the western end of the north america–south america plate boundary zone. Marine Geology, 303:154–171, 2012. C S Prentice, P Mann, A J Crone, R D Gold, K W Hudnut, R W Briggs, R D Koehler, and P Jean. Seismic hazard of the Enriquillo–Plantain Garden fault in Haiti inferred from palaeoseismology. Nature Geoscience, 3(11):1–5, October 2010a. Carol S Prentice, Paul Mann, Luis R Pe˜na, and G Burr. Slip rate and earthquake recurrence along the central septentrional fault, north american–caribbean plate boundary, dominican republic. Journal of Geophysical Research: Solid Earth (1978– 2012), 108(B3), 2003. Carol S Prentice, John C Weber, Christopher J Crosby, and Daniel Ragona. Pre- historic earthquakes on the caribbean–south american plate boundary, central range fault, trinidad. Geology, 38(8):675–678, 2010b. 86

M Pubellier, J M Vila, and D Boisson. North Caribbean neotectonic events: The Trans-Haitian fault system. Tertiary record of an oblique transcurrent shear zone uplifted in Hispaniola. Tectonophysics, 194(3):217–236, 1991. Manuel Pubellier, Alain Mauffret, Sylvie Leroy, Jean Marie Vila, and Helliot Amil- car. Plate boundary readjustment in oblique convergence: Example of the Neogene of Hispaniola, Greater Antilles. Tectonics, 19(4):630–648, August 2000a. Manuel Pubellier, Alain Mauffret, Sylvie Leroy, Jean Marie Vila, and Helliot Amil- car. Plate boundary readjustment in oblique convergence: Example of the neogene of hispaniola, greater antilles. Tectonics, 19(4):630–648, 2000b. R. Reilinger, S. McClusky, P. Vernant, S. Lawrence, S. Ergintav, R. Cakmak, H. Ozener, F. Kadirov, I. Guliev, R. Stepanyan, M. Nadariya, G. Hahubia, S. Mah- moud, K. Sakr, A. ArRajehi, D. Paradissis, A. Al-Aydrus, M. Prilepin, T. Guseva, E. Evren, A. Dmitrotsa, S. V. Filikov, F. Gomex, R. Al-Ghazzi, and G. Karam. GPS constraints on continental deformation in the Africa-Arabia-Eurasia continen- tal collision zone and implications for the dynamics of plate interactions. Journal of Geophysical Research, 111, 2006a. doi: 10.1029/2005JB004051. Robert Reilinger, Simon McClusky, Philippe Vernant, Shawn Lawrence, Semih Ergintav, Rahsan Cakmak, Haluk Ozener, Fakhraddin Kadirov, Ibrahim Guliev, Ruben Stepanyan, et al. Gps constraints on continental deformation in the africa- arabia-eurasia continental collision zone and implications for the dynamics of plate interactions. Journal of Geophysical Research: Solid Earth (1978–2012), 111(B5), 2006b. GR Robson. An earthquake catalogue for the eastern caribbean 1530-1960. Bulletin of the Seismological Society of America, 54(2):785–832, 1964. WR Roest and BJ Collette. The fifteen twenty fracture zone and the north american– south american plate boundary. Journal of the Geological Society, 143(5):833–843, 1986. Eric Rosencrantz and Paul Mann. Seamarc ii mapping of transform faults in the cayman trough, caribbean sea. Geology, 19(7):690–693, 1991. Eric Rosencrantz, Malcolm I Ross, and John G Sclater. Age and spreading history of the cayman trough as determined from depth, heat flow, and magnetic anomalies. Journal of Geophysical Research: Solid Earth (1978–2012), 93(B3):2141–2157, 1988. Larry Ruff and Hiroo Kanamori. Seismicity and the subduction process. Physics of the Earth and Planetary Interiors, 23(3):240–252, 1980. N Saint Fleur, N Feuillet, and R Grandin. Seismotectonics of southern Haiti: A new faulting model for the 12 January 2010 M7 earthquake. Geophysical Research Letters, 42, 2015. J Sanchez-Rojas and M Palma. Crustal density structure in northwestern south america derived from analysis and 3-d modeling of gravity and seismicity data. Tectonophysics, 2014. David T Sandwell and Walter HF Smith. Global marine gravity from retracked geosat and ers-1 altimetry: Ridge segmentation versus spreading rate. Journal of Geophysical Research: Solid Earth (1978–2012), 114(B1), 2009. 87

E Saria, E Calais, DS Stamps, D Delvaux, and CJH Hartnady. Present-day kine- matics of the east african rift. Journal of Geophysical Research: Solid Earth, 119 (4):3584–3600, 2014. J C Savage. A dislocation model of strain accumulation and release at a subduction zone. Journal of Geophysical Research, 88(B6):4984–4996, 1983a. JC Savage. A dislocation model of strain accumulation and release at a subduction zone. Journal of Geophysical Research: Solid Earth (1978–2012), 88(B6):4984–4996, 1983b. J Scherer. Great earthquakes in the island of haiti. Bulletin of the Seismological Society of America, 2(3):161–180, 1912. Ralf Schmid, Peter Steigenberger, Gerd Gendt, Maorong Ge, and Markus Rothacher. Generation of a consistent absolute phase-center correction model for gps receiver and satellite antennas. Journal of Geodesy, 81(12):781–798, 2007. C H Scholz and J Campos. On the Mechanism of Seismic Decoupling and Back-Arc Spreading at Subduction Zones. Journal of Geophysical Research-Solid Earth, 100 (B11):22103–22115, 1995. P Segall. Integrating geologic and geodetic estimates of slip rate on the San Andreas fault system. International Geology Review, 44(1):62–82, 2002. MD Soto, P Mann, A Escalona, and LJ Wood. Late holocene strike-slip offset of a subsurface channel interpreted from three-dimensional seismic data, eastern offshore trinidad. Geology, 35:859–862, 2007. S Stein, J F ENGELN, D A WIENS, K FUJITA, and R C SPEED. Subduction Seismicity and Tectonics in the Lesser Antilles Arc. Journal of Geophysical Research- Atmospheres, 87(NB10):8642–8664, 1982. Lynn R Sykes, William R McCann, and Alan L Kafka. Motion of caribbean plate during last 7 million years and implications for earlier cenozoic movements. Journal of Geophysical Research: Solid Earth (1978–2012), 87(B13):10656–10676, 1982. S Symithe, Eric Calais, and J B Chabalier. Current block motions and strain accu- mulation on active faults in the Caribbean. Journal of Geophysical Research Solid Earth, 120(5):3748–3774, 2015. S J Symithe, Eric Calais, J S Haase, A M Freed, and R Douilly. Coseismic Slip Distribution of the 2010 M 7.0 Haiti Earthquake and Resulting Stress Changes on Regional Faults. Bulletin of the Seismological Society of America, 103(4):2326–2343, July 2013a. Steeve J Symithe, Eric Calais, Jennifer S Haase, Andrew M Freed, and Roby Douilly. Coseismic slip distribution of the 2010 m 7.0 haiti earthquake and resulting stress changes on regional faults. Bulletin of the Seismological Society of America, 103(4): 2326–2343, 2013b. U ten Brink and J Lin. Stress interaction between subduction earth- quakes and forearc strikeslip faults: Modeling and application to the north- ern caribbean plate boundary. Journal of Geophysical Research: Solid Earth (19782012), 109(B12), 2004. doi: 10.1029/2004JB003031. URL http://dx.doi.org/10.1029/2004JB003031. 88

Uri S ten Brink, William H Bakun, and Claudia H Flores. Historical perspective on seismic hazard to hispaniola and the northeast caribbean region. Journal of Geophysical Research: Solid Earth (1978–2012), 116(B12), 2011. M Terrier, A Bialkowski, A Nachbaur, C Pr´epetit,andYFJoseph.Revisionofthe geological context of the Port-au-Prince metropolitan area, Haiti: implications for slope failures and seismic hazard assessment. Natural Hazards and Earth System Science, 14(9):2577–2587, 2014. Robert Trenkamp, James N Kellogg, Jeffrey T Freymueller, and Hector P Mora. Wide plate margin deformation, southern central america and northwestern south america, casa gps observations. Journal of South American Earth Sciences, 15(2): 157–171, 2002. SHELLEY R Van Dusen and DIANE I Doser. Faulting processes of historic (1917– 1962) m 6.0 earthquakes along the north-central caribbean margin. pure and applied geophysics, 157(5):719–736, 2000. Philippe Vernant, Robert Reilinger, and Simon McClusky. Earth and Planetary Science Letters. Earth and Planetary Science Letters, 385(C):122–129, January 2014. David J Wald, V Quitoriano, T H Heaton, and H Kanamori. Relationships between peak ground acceleration, peak ground velocity, and modified Mercalli intensity in California. Earthquake Spectra, 15(3):557–564, 1999. L M Wallace, P Barnes, J Beavan, R Van Dissen, N Litchfield, J Mountjoy, R Lan- gridge, G Lamarche, and N Pondard. The kinematics of a transition from subduction to strike-slip: An example from the central New Zealand plate boundary. Journal of Geophysical Research, 117(B2), February 2012a. L M Wallace, A˚ Fagereng, and S Ellis. Upper plate tectonic stress state may in- fluence interseismic coupling on subduction megathrusts. Geology, 40(10):895–898, September 2012b. J C Weber, J Saleh, S Balkaransingh, T Dixon, W Ambeh, T Leong, A Rodriguez, and K Miller. Triangulation-to-GPS and GPS-to-GPS geodesy in Trinidad, West Indies: Neotectonics, seismic risk, and geologic implications. Marine and Petroleum Geology, 28(1):200–211, December 2010. John C Weber, Timothy H Dixon, Charles DeMets, William B Ambeh, Pamela Jansma, Glen Mattioli, Jarir Saleh, Giovanni Sella, Roger Bilham, and Omar P´erez. Gps estimate of relative motion between the caribbean and south american plates, and geologic implications for trinidad and venezuela. Geology, 29(1):75–78, 2001. Donald L Wells and Kevin J Coppersmith. New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement. Bulletin of the Seismological Society of America, 84(4):974–1002, 1994. Scott M White, Robert Trenkamp, and James N Kellogg. Recent crustal deformation and the earthquake cycle along the ecuador–colombia subduction zone. Earth and Planetary Science Letters, 216(3):231–242, 2003. Margaret D Wiggins-Grandison. Simultaneous inversion for local earthquake hypocentres, station corrections and 1-d velocity model of the jamaican crust. Earth and Planetary Science Letters, 224(1):229–240, 2004. VITA 89

VITA

My name is Steeve Julien Symithe. I was born in P´etion-ville, a little city in the South of Port-au-Prince, the capital of Haiti. I studied in the Facult´e des Sciences of the University dEtat dHaiti (UEH) since 2004 to 2009, where I have obtained a diploma in civil engineering. After the earthquake that struck Port-au-Prince in January 12 2010, I was selected based on academic record between two students to pursue a master in geophysics co-financed by Purdue University and the Voila foundation (a foundation of a telecommunication company in Haiti). I have spent 2 years working on the coseismic aspect of the 2010 Haiti earthquake during the Master program. I successfully defended my master thesis during summer 2012. And later, I was able to summarize the results of this study into a scientific article that was accepted and published in the Bulletin of Seismological Society of America (BSSA). Then, I decided to continue with my research and pursue a PhD degree at Purdue in the same field of geophysics. My application to the Purdue PhD program was accepted and I have continued to work on several projects during 3 more years. I have then published one more article to summarize my results from a kinematic model for the whole Caribbean plate and another paper where I studied the crustal deformation in Southern Haiti is accepted for publication. I worked under the guidance of professors Andrew Freed from the Earth, Atmospheric and Planetary Sciences (EAPS) and Eric Calais from Ecole Normale Superieure in Paris on using GPS observations and other geodetic measurements to model crustal rheology and also to understand the different period of the earthquake cycle. I have also collaborated form other scientists here at Purdue and in the Institut de Physique du Globe in Paris.