FAULTING AND STRAIN PARTITIONING IN JAMAICA FROM
GPS AND STRUCTURAL DATA: IMPLICATIONS FOR GÔNAVE
AND HISPANIOLA MICROPLATE KINEMATICS, NORTHERN
CARIBBEAN
by Bryn Benford
A dissertation submitted in partial fulfillment of the requirements for the degree of
Doctor of Philosophy (Geoscience)
at the University of Wisconsin – Madison 2012
Date of final oral examination: 3/22/2012
The dissertation is approved by the following members of the Final Oral Committee: D. Charles DeMets, Albert and Alice Weeks Professor, Geoscience Basil Tikoff, Professor, Geoscience Laurel Goodwin, Professor, Geoscience Clifford Thurber, Professor, Geoscience Harold Tobin, Professor, Geoscience ! "! FAULTING AND STRAIN PARTITIONING IN JAMAICA FROM GPS AND
STRUCTURAL DATA: IMPLICATIONS FOR GÔNAVE AND HISPANIOLA
MICROPLATE KINEMATICS, NORTHERN CARIBBEAN
Bryn Benford
Under the supervision of Professors Chuck DeMets and Basil Tikoff
At the University of Wisconsin-Madison
A series of small microplates separate the Caribbean and North America plates in the northern Caribbean. My dissertation focuses on understanding the structural evolution and neotectonic deformation of Jamaica, and how it relates to the overall microplates and tectonics of the northern Caribbean. Jamaica, which lies along the same seismically active plate boundary as Haiti, has had twelve earthquakes with Modified Mercalli intensities of VII to X since 1667. However, remarkably little is known about which faults presently constitute the most significant seismic hazards. This research provides insight into tectonic processes and facilitates mitigation of geological hazards in the region.
Two chapters focus on characterizing deformation in Jamaica through modeling GPS velocities and through field mapping. The best-fitting models based on GPS velocities place most strike-slip motion on faults in central Jamaica and suggest that faults in northern
Jamaica have minimal motion. I estimate 4-5 mm yr-1 of slip for faults near the capital city of
Kingston of southeastern Jamaica, implying significant seismic hazard. Field mapping combined with present-day topography, focal mechanisms, geology, gravity, and well and borehole data indicate that east-west contraction is accommodated by reactivated, NNW- striking reverse faults, which are bound by E-striking strike-slip faults in southern Jamaica. ! ""! The other two chapters of my thesis focuses on understanding the behavior of the microplates along the Caribbean-North America plate boundary: I model GPS velocities and use shear-wave splitting to understand the crustal and mantle behaviors, respectively of the microplates. The GPS data require an independently moving Hispaniola microplate between the Mona Passage and a likely diffuse boundary just west of or within western Hispaniola.
The new microplate angular velocities predict 6.8±1.0 mm yr-1 of left-lateral slip and 5.7±1 mm yr-1!of convergent motion surrounding the seismically hazardous Enriquillo fault of western Hispaniola, suggesting that one to two M=7 earthquakes are expected for Haiti each century. Using shear-wave splitting analyses, I document fast axis of polarization parallel to the Gônave microplate boundaries along its northern, southern, and eastern boundaries. In the interior of the microplates, weak/no fabric is documented suggesting that the microplate boundaries continue into the upper mantle.
! """! ACKNOWLEDGMENTS
! #"$%&'!(!)*+,-!,"./!&*!&012.!31%",!4".*556!!40$*+70!&0"%!/2&"$/!8$*9/%%'!0/!01%!
9*2%&12&,:!;//2!%+88*$&"
&012.!>0+9.!?/@/&%6!!(!188$/9"1&/!&0/!=12:!0*+$%!0/!%8/2&!)"&0!=/!=1."27!=/!1!;/&&/$!
%9"/2&"%&!12-!1!;/&&/$!)$"&/$6!!(!1,%*!)*+,-!,"./!&*!&012.!>0+9.A%!51=",:!5*$!1,,!*5!&0/!&"=/!
&01&!&0/:!%19$"5"9/-!)0",/!>0+9.!12-!(!B.:8/-6!C1%&,:'!&012.!:*+!C1+$/,!D**-)"2!5*$!*+$!
=12:!$12-*=!&1,.%!"2!&0/!"%*,1&"*2!)1$-6!!4012.!:*+!1,%*!&*!>,"55!40+$;/$!12-!E1$*,-!
4*;"2'!)0*!01
G+/%&"*2%6!
! @12:!&012.%!7*!&*!H,5$/-!@9?/$=*&&'!)0*!-$*
J1=1"912!$*1-%6!(!)*+,-!1,%*!,"./!&*!&012.!?$6!@1$71$/&!K"77"2%LD$12-"%*2'!)0*!0/,8/-!
=/!-+$"27!=:!5"$%&!5"/,-!%/1%*2!"2!J1=1"916!?$6!C:2-*2!3$*)2!12-!C1+$/,!>0*:'!12-!
/%8/9"1,,:!M1+,!K",,"1=%!1&!&0/!J1=1"91!N1$&0G+1./!O2"&!01
19G+"$"27!12-!%/2-"27!=/!-1&16!
(!)*+,-!1,%*!,"./!&*!&012.!P/1,!C*$-!5*$!0/,8"27!=/!12-!&/190"27!=/!"2!&0/!5"/,-!
12-!;19.!"2!&0/!,1;6!!(!1,%*!&012.!C//!M*)/,,!5*$!0"%!81&"/29/!12-!1%%"%&129/!)"&0!
8$*9/%%"27!3$*1-;12-!%/"%=*=/&/$!-1&1!Q!(!-"-2A&!=1./!"&!/1%:!5*$!0"=6!(!2//-!&*!&012.!
J/$/=:!M/%"9/.!12-!P"251!3/22"27&*2!5*$!0/,8"27!=/!;/9*=/!1!8%/+-*L7/*80:%"9"%&!12-!
,"%&/2"27!&*!1,,!*5!=:!$1=;,"27%6!
! (!)*+,-!,"./!&*!&012.!&0/!%&$+9&+$/!7$*+8!81%&!12-!8$/%/2&6!J/22"/!>**.!%8*",/-!
=/!1%!51$!1%!*55"9/!%*+,!=1&/%!7*6!(!)*+,-!2*&!01
-/5"2"&/,:!&0/!;/%&!5"/,-!1%%"%&12&!*+&!&0/$/6!
! #"21,,:'!(!01
01
%90**,!;:!2*)!"5!(!)/$/!&0/=6!!4012.%!&*!T1&/!5*$!=*<"27!&*!K"%9*2%"2!12-!1,)1:%!
;/"27!&0/$/!5*$!=/6! !
! TABLE OF CONTENTS ! ! Abstract "! ! Acknowledgments """! ! Table of Contents
The M=7 Haiti earthquake in 2010 is a tragic reminder that the strike-slip faults that define much of the northern boundary of the Caribbean plate constitute major seismic hazards (Prentice et al., 2010). After the earthquake, an international effort began to better understand the seismic risks associated with the numerous faults that define the boundaries between a series of small microplates that separate the Caribbean and North America plates in the northern Caribbean (e.g., Calais et al., 2010). Jamaica, which lies along the same seismically active plate boundary as Haiti, has had twelve earthquakes with Modified
Mercalli intensities of VII to X since 1667, including the MMI X earthquake in 1692 that destroyed much of the city of Port Royal (near the present capital Kingston) and the M=6-6.5 earthquake in 1907 that damaged or destroyed 85% of the buildings in Kingston (Taber,
1920; Versey et al., 1958; Pereira, 1977; Tomblin and Robson, 1977; Clark, 1995; Wiggins-
Grandison, 1996; Natural Disaster Research, 1999; Wiggins-Grandison, 2001). Despite
Jamaica’s long history of damaging earthquakes, remarkably little is known about which fault(s) were responsible for previous large earthquakes and which faults presently constitute the most significant seismic hazards, including the locations of the 1692 and 1907 ruptures.
My dissertation focuses on understanding the structural evolution and neotectonic deformation of Jamaica, and how it relates to the overall tectonics of the northern Caribbean.
The first chapter of my dissertation focuses on building on work of DeMets and
Wiggins-Grandison (2007) by modeling GPS velocities estimated from 13 years of continuous and campaign GPS measurements at 30 Jamaican sites and 96 other sites from the northern Caribbean. I determine a lower bound of 6.0±0.5 mm yr-1 of WSW motion of the
Gônave microplate in Jamaica along the Gônave-Caribbean plate boundary. Additionally, 2 2.6±0.6 mm yr-1 of primarily southward movement of all GPS sites in the Jamaica archipelago suggests that oblique convergence across the Gônave microplate’s southern boundary in Jamaica may be partitioned, such that ~N-S shortening is accommodated on a fault system south of Jamaica. The best-fitting models place most strike-slip motion on faults in central Jamaica and suggest that faults in northern Jamaica have minimal motion. I estimate 4-5 mm yr-1 of slip for the Plantain Garden fault and Blue Mountain restraining
bend of southeastern Jamaica, implying significant seismic hazard for the nearby capital of
Kingston.
The second chapter concentrates on determining the location of the Gônave
microplate’s eastern boundary in Hispaniola and estimating angular velocities for the
Gônave, Hispaniola, Puerto Rico-Virgin Islands, and two smaller microplates. Similar to the
first chapter, I use elastic block modeling of the same 126 GPS velocities. A model in which
the Gônave microplate extends from the Mona Passage to the Cayman spreading center is
rejected at a high confidence level. The data instead require an independently moving
Hispaniola microplate between the Mona Passage and a likely diffuse boundary just west of
or within western Hispaniola. The new microplate angular velocities predict 6.5±0.5 mm yr-1
of left-lateral slip along the seismically hazardous Enriquillo-Plantain Garden fault zone of
western Hispaniola, 10.9±2.0 mm yr-1 of slip along the Septentrional fault of northern
Hispaniola, and ~14-15 mm yr-1 of left-lateral slip along the Oriente fault south of Cuba.
In the third chapter, I focus on the neotectonics of southern Jamaica and how the two
dominant fault sets interact with one another. I use present-day topography, focal
mechanisms, geology, gravity, and well and borehole data to show that contraction is
accommodated by reactivated, NNW-striking reverse faults. Reactivated, EW-striking 3 strike-slip faults accommodate some of the plate motion but also act to separate the zone of
contraction from essentially undeformed areas. Both fault sets were originally part of an
extensional transfer zone in the Paleogene (NNW-striking normal faults and east-west strike- slip faults), which has since been reactivated to accommodate contraction as part of the
Jamaica restraining bend.
The final chapter of my thesis tests whether the boundaries of the microplates, documented in the first two chapters, extend into the upper mantle. Using shear-wave splitting analysis for eight broadband seismometers across the northern Caribbean, I document shear-wave splitting delay times of ~1.5 s on the northern, southern, and eastern boundaries of the Gonave microplate. In all cases, the fast axis of polarization is parallel to the microplate boundaries, similar to results along the southern margin of the Caribbean plate
(Russo et al., 1996). Since shear-wave splitting is commonly attributed to deformation of the lithospheric mantle, I conclude that deformation occurs in the upper mantle below the microplate boundaries.
This dissertation provides a better understanding of the neotectonic deformation of the northern boundary of the Caribbean plate, at a variety of scales and depths. The geodesy and elastic modeling provide a “snapshot” of current deformation, the geological studies allow insights into the cumulative effects of deformation, and the seismic anisotropy studies constrain deformation of the mantle along the plate margin. The integration of these approaches provide insight into tectonic processes, facilitating mitigation of geological hazards in the region.
4 REFERENCES
Calais, E., Freed, A., Mattioli, G., Amelung, F., Jónsson, S., Jansma, P., Hong, S.-H., Dixon,
T., Prépetit, C., & Momplaisir, R., 2010, Transpressional rupture of an unmapped fault
during the 2010 Haiti earthquake, Nature Geoscience, doi: 10.1038/NGEO992.
Clark, G.R., II, 1995, Swallowed up, Earth, 4, 34-41.
DeMets, C., & Wiggins-Grandison, M., 2007. Deformation of Jamaica and motion of the
Gônave microplate from GPS and seismic data, Geophys. J. Int., 168, 362-378, doi:
10.1111/j.1365-246X.2006.03236.x.
Natural Disaster Research, Inc., Earthquake Unit, Mines & Geology Division, 1999.
Kingston Metropolitan Area Seismic Hazard Assessment Final Report, Caribbean
Disaster Mitigation Project, Organization of American States,
www.oas.org/cdmp/document/kma/seismic/kma1.htm, 82 pp.
Pereira, J., 1977. An engineering seismology study of Jamaica, MSc thesis, Imperial College
London University, London, pp. 147.
Prentice, C.S., Mann, P., Crone, A.J., Gold, R.D., Hudnut, K.W., Briggs, R.W., Koehler,
R.D., and Jean, P., 2010. Seismic hazard of the Enriquillo-Plantain Garden fault in Haiti
inferred from palaeoseismology, Nature Geoscience, doi: 10.1038/NGEO991.
Russo, R. M., P.G. Silver, M. Franke, W. B. Ambeh, and D.E. James (1996), Shear-wave
splitting in northeast Venezuela, Trinidad, and the eastern Caribbean, Phys. Earth
Planet. Inter., 95, 251-275.
Taber, S., 1920. Jamaica earthquakes and the Bartlett Trough, Bull. Seism. Soc. Am., 10, 55-
86. 5 Tomblin, J.M. & Robson, G.R., 1977. A catalogue of felt earthquakes for Jamaica, with
references to other islands in the Greater Antilles, 1564-1971, Ministry of Mining and
Natural Resources (Jamaica), Mines & Geology Division Special Publication No. 2, 243
pp.
Versey, H.R., Williams, J.B., & Robinson, E., 1958. The earthquake of March 1, 1957,
Geonotes; quarterly newsletter of the Jamaica Group of the Geologists Association, 1,
54-65.
Wiggins-Grandison, M.D., 1996. Seismology of the January 1993 earthquake, J. Geol. Soc.
Jam., 30, 1-14.
Wiggins-Grandison, M.D., 2001. Preliminary results from the new Jamaica seismograph
network, Seism. Res. Lett., 72, 525-537. 6
CHAPTER 1
Seismic hazard along the southern boundary of the Gônave microplate: Block modeling
of GPS velocities from Jamaica and nearby islands, northern Caribbean
B. Benford1, C. DeMets1, B. Tikoff1, P. Williams2, L. Brown2, and M. Wiggins-Grandison3
1Department of Geoscience
University of Wisconsin – Madison
1215 W. Dayton St.
Madison, Wisconsin 53706, USA
2Earthquake Unit
University of the West Indies – Mona Campus
Kingston 7, Jamaica
3Preparatory Commission for the Comprehensive Nuclear-Test Ban Treaty Organization
Vienna International Centre
P.O. Box 1250
1400 Vienna, Austria
In press, Geophysical Journal International. 7
Abstract
We use block modeling of GPS site velocities from Jamaica and nearby islands, including
Hispaniola, to test alternative plate boundary geometries for deformation in Jamaica and estimate slip rates along the island’s major fault zones. Relative to the Caribbean plate, GPS sites in northern Jamaica move 6.0±0.5 mm yr-1 to the WSW, constituting a lower bound on
the motion of the Gônave microplate across its southern boundary in Jamaica. Obliquely
convergent motion of all 30 GPS sites on and near Jamaica relative to the island’s ~E-W- trending strike-slip faults may be partitioned into 2.6±0.6 mm yr-1 of ~N-S shortening across submarine faults south of Jamaica and 5-6 mm yr-1 of E-W motion. Guided by geologic and
seismic information about the strikes and locations of faults in Jamaica, inverse block
modeling of the regional GPS velocities rejects plate boundary configurations that presume
either a narrow plate boundary in Jamaica or deformation concentrated across a restraining
bend defined by the topographically high Blue Mountains of eastern Jamaica. The best-fitting
models instead place most deformation on faults in central Jamaica. The 4-5 mm yr-1 slip rate we estimate for the Plantain Garden fault and Blue Mountain restraining bend of southeastern
Jamaica implies significant seismic hazard for the nearby capital of Kingston.
Key Words: Neotectonics; Jamaica; Caribbean; Gônave microplate; Plate motions
1. Introduction
The January 12, 2010 M=7 Haiti earthquake (Calais et al. 2010; Hayes et al. 2010),
which ruptured an onland segment of the southern boundary of the Gônave microplate (Fig.
8 1), was a tragic reminder that the strike-slip faults that define much of the northern boundary of the Caribbean plate constitute major seismic hazards where those faults come on land
(Prentice et al. 2010). The island of Jamaica (Fig. 1), which lies along the same seismically active plate boundary, has had twelve earthquakes with Modified Mercalli intensities (MMI) of VII to X since 1667, including the MMI X earthquake in 1692 that destroyed much of the city of Port Royal (near the present capital of Kingston) and the M=6-6.5 earthquake in 1907 that damaged or destroyed 85% of the buildings in Kingston (Taber 1920; Versey et al. 1958;
Pereira 1977; Tomblin and Robson 1977; Clark 1995; Wiggins-Grandison 1996; Natural
Disaster Research 1999; Wiggins-Grandison 2001). Despite Jamaica’s long history of damaging earthquakes, remarkably little is known about which faults were responsible for previous large earthquakes and which faults presently constitute the most significant seismic hazards, including the locations of the 1692 and 1907 ruptures.
In an effort to better understand seismic hazard and deformation rates in Jamaica, GPS measurements began on the island in 1998. DeMets and Wiggins-Grandison (2007)
(hereafter abbreviated DWG07) report the initial results from these measurements along with focal mechanisms determined from teleseismic and local seismograms. DWG07 demonstrate that GPS sites move dominantly to the WSW relative to a stationary Caribbean plate at rates that increase from 3 mm yr-1 in southern Jamaica to 7-8 mm yr-1 in northern Jamaica and further demonstrate that the mean P- and T-axes for earthquakes on and near the island have nearly horizontal plunges and trend 45° from the island’s E-W-oriented strike-slip faults (Fig.
2c). From these observations, they conclude that deformation in Jamaica is dominated by a combination of left-lateral shear along E-W striking strike-slip faults and convergence across the island’s NNW-striking mountain ranges, both consistent with a general model of the
9 island as a major restraining bend in the left-slipping Gônave-Caribbean plate boundary
(Mann et al. 1985).
Herein, we build on the DWG07 study by modeling GPS velocities estimated from 13
years of continuous and campaign GPS measurements at 30 Jamaican sites and 96 sites in
Hispaniola (Manaker et al. 2008; Calais et al. 2010) and other islands that span the
Caribbean-Gônave-North America plate boundary. Using the locations of major faults and earthquakes in Jamaica and the northern Caribbean as a guide, we test a series of progressively more complex models for the geometry of the Gônave-Caribbean plate boundary in Jamaica, ranging from simple discrete/narrow boundaries to boundaries with an independently moving block in Jamaica. Based on our preferred plate boundary geometries, we estimate present-day fault slip rates and the locus of present deformation in Jamaica.
Modeling of the GPS velocity field is accomplished using the Blocks software of Meade &
Loveless (2009), which treats the crust as an elastic homogeneous half-space consisting of
rotating blocks bound by frictionally locked plate-boundary faults.
2. Tectonic Setting
!"#" $%&'()*$+,-./01(2*$
Jamaica straddles the boundary between the Gônave microplate and Caribbean plate (Fig.
1) and is located at the northern, emergent end of the Nicaragua Rise (Fig. 1), where the rise
collides obliquely with the southern edge of the Gônave microplate. Westward motion of 8-
13 mm yr-1 of the Gônave microplate relative to the Caribbean plate (DWG07), likely driven by the oblique collision of the Bahama platform with Hispaniola (Mann et al. 1995, 2002),
10 gives rise to left-lateral slip on the Enriquillo, Plantain Garden, and Walton faults along the
southern edge of the Gônave microplate (Fig. 1).
The Gônave microplate boundaries consist of the Cayman spreading center in the west,
the Oriente transform fault in the north, multiple faults in the south including the Enriquillo
fault of Hispaniola, the Plantain Garden fault, faults in Jamaica, and the Walton fault west of
Jamaica (Fig.1; Rosencrantz & Mann 1991, Tyburski 1992). The eastern boundary of the
microplate may lie west of Hispaniola, within central Hispaniola, or may be diffuse
(Manaker et al. 2008; Calais et al. 2010; Chapter 2).
!"!" 3(4125$,'$6(+(,-($
The ~50-km right step between the Plantain Garden fault (PG) of eastern Jamaica and
Walton fault (WF) west of Jamaica define the Jamaica restraining bend (Fig. 2a). The right-
stepping Jamaica restraining bend has given rise to widespread faulting and reactivation of
faults in and near Jamaica consisting of east-west striking, left-lateral strike-slip faults and
NNW-striking faults dominated by reverse dip-slip motion (Figs 1 and 2a; Horsfield 1974;
Wadge & Dixon 1984; Mann et al. 1985; Leroy et al. 1996). The latter faults are typically
steeply east-dipping, in many cases are blind, and are Paleogene extensional structures
reactivated in contraction (Horsfield 1974; Draper 2008).
Previous studies (Horsfield 1974; Wadge & Dixon 1984; Mann et al. 1985; Leroy et al.
1996; DeMets & Wiggins-Grandison 2007) of the structures and topography of Jamaica
define four major, east-west striking, left-lateral strike-slip fault systems on the island (Fig.
2a), the Duanvale fault (DF) of northern Jamaica, the South Coast fault zone (SCFZ) of
southern Jamaica, the Plantain Garden and Aeolus Valley (AV) faults of southeastern
11 Jamaica, and the central Jamaica fault system, consisting of the Cavaliers fault (CF), Rio
Minho-Crawle River fault (RMCR), and Siloah fault system (SFS). None of these east-west faults have the typical geomorphic expression of a large-offset, strike-slip fault (e.g., fault scarps, sag ponds) and none can be traced continuously across the island. However, they show up as prominent lineaments in aerial photographs and in the 90-m Shuttle Radar
Topography Mission (SRTM) elevations (Fig. 2b). Below, we describe each briefly.
The Plantain Garden fault extends ~150 km from the Morant Trough (Fig. 1), an active pull-apart basin east of Jamaica (Mann et al. 1990) that defines the western
termination of the Enriquillo fault of Hispaniola, to the Wagwater deformed belt (WW in Fig.
2b) at the western edge of the Blue Mountains (Mann et al. 1985). Estimates of the fault
offset east of the island range from 30-45 km based on the offset of the eastern Jamaica shelf
to ~60 km based on the width of the Morant Trough. These are consistent with 3-7 mm yr-1 of slip on the Plantain Garden fault assuming the offset has occurred since ~9 Ma (Natural
Disaster Research 1999). Farther west, where the Plantain Garden fault of eastern Jamaica separates the Blue Mountains to the north from topographically lower and younger rocks to the south, rocks of similar ages and lithologies flanking the fault are offset by only 10-12 km
(Mann et al. 1985).
The difference in the offsets estimated for the eastern and western halves of the
Plantain Garden fault may be due to partitioning of slip between the western segment of the
Plantain Garden fault and a previously undescribed, unnamed WSW-striking fault that is visible in both aerial photographs and 90-m SRTM elevations (Figs 2a and b; AV). Hereafter, we refer to this fault as the Aeolus Valley fault for the valley in eastern Jamaica where the fault is located (Figs 2a and b). The fault, which is mapped but not named on the 1992
12 Jamaica Mining and Commerce structure maps, intersects the eastern Plantain Garden fault and may carry motion offshore to the South Coast fault zone (Fig. 2a; SCFZ).
In northern Jamaica, the Duanvale fault (DF) has a locally prominent topographic signature, where it is lower than the surrounding areas and creates a lineament at the macroscale (Wadge & Dixon 1984). The small offset estimated across the Duanvale fault, less than 10 km and as little as 3 km (Wadge & Dixon 1984), based on offset of Cretaceous units (Grippi 1978), either argues against this fault as a primary plate boundary structure or indicates that it became active too recently to accumulate significant offset. We test these possibilities below.
The left-lateral Rio Minho-Crawle River fault (RMCR) of central Jamaica offsets
Cretaceous features in the Central Inlier by 8 km, is reactivated from the Cretaceous
(Mitchell 2003), and is the most seismically active strike-slip fault on the island (Fig. 2;
DeMets & Wiggins-Grandison 2007). Field mapping, structure maps, and topographic maps define the Siloah fault system (SFS) west of the Cavaliers and Rio Minho-Crawle River faults; these three faults define a continuous central Jamaica fault system (Chapter 3; Fig. 2).
The Siloah fault system is a reactivated fault from the Eocene (Wright 1975). No reverse faults or mountain ranges are continuous across the central Jamaica fault system at the 90-m
SRTM-scale or based on fieldwork.
The South Coast fault zone (SCFZ) cuts across the alluvium-covered Vere Plain of south-central Jamaica and closely parallels the southern coast of southwestern Jamaica, where it creates prominent cliffs as it crosscuts the NNW-oriented ranges that occur to the north. The cliffs associated with the South Coast fault zone extend along the western third of the island. Offset is not constrained for the South Coast fault zone.
13 North-northwest-striking faults are interpreted as dip-slip features (Fig. 2a; Horsfield
1974). In eastern Jamaica, the Blue Mountain and Yallahs faults accommodate some or all
contraction across the Blue Mountains, which are Jamaica’s highest (>2000 m) and most
seismically active region (Fig. 2). Farther west, uplift of the Don Figuerero and Santa Cruz
Mountains, which reach ~800 m and ~660 m, respectively, occurs along NNW-striking
reverse faults (Fig. 2). Recent mapping (Chapter 3) shows that these NNW-striking ranges are cored by east-dipping reverse faults that terminate at and are bounded by the South Coast fault zone in the south and the central Jamaica fault system in the north.
3. Geodetic Data
7"#" $6(+(,-($%89$'*2:/.;$<*5-.,02,/'$
The Jamaica GPS network consists of 30 geodetic benchmarks (Table 2; Fig. 3), 28 on
the main island of Jamaica and one each on Morant Cay and Pedro Cay. GPS data described
and used by DWG07 include data from 20 sites spanning the period 1998-2005; the present
velocity field includes 18 of the 20 sites used by DWG07 and twelve new sites, notably
including two sites on limestone cays, 50 km southeast and 80 km south of the main island
(MCAY and PEDR in Fig. 3a). Data for the present study span a 13-yr-long period from
1998 to August of 2011 (Table 2). The velocities for GPS stations BAMB and COFE used by DWG07 are omitted from this study due to concerns about the stability of these two sites.
Continuous or quasi-continuous measurements constrain the velocities at six of the 30 sites
(Table 2); the remaining station velocities are derived from campaign measurements,
14 typically lasting five or more days per site occupation (Table 2). All data from the Jamaica
GPS sites are available through UNAVCO or the National Geodetic Survey CORS archive.
7"!" %89$<(2($('(1=5,5$$
Except for the data for stations from Hispaniola, whose velocities are taken from Calais
et al. (2010), all the GPS data used for this study were processed with Release 6.1 of the
GIPSY software suite from the Jet Propulsion Laboratory (JPL). Non-fiducial daily GPS
station coordinates were estimated using a precise point-positioning strategy (Zumberge et al.
1997), including constraints on a priori tropospheric, hydrostatic, and wet delays from
Vienna Mapping Function (VMF1) parameters (http://ggosatm.hg.tuwien.ac.at), elevation- and azimuthally-dependent GPS and satellite antenna phase center corrections from IGS08
ANTEX files (available via ftp from sideshow.jpl.nasa.gov), and corrections for ocean tidal loading from the TPX0.7.2 ocean tide model (http://froste.oso.chalmers.se). Wide- and narrow-lane phase ambiguities were resolved for all the data using GIPSY's single-station ambiguity resolution feature.
All daily non-fiducial station location estimates were transformed to ITRF2008
(Altamimi et al. 2011) using daily seven-parameter Helmert transformations from JPL. The resulting station coordinate time series have day-to-day scatter of 2.8 and 3.1 mm yr-1 in their latitudes and longitudes relative to simple linear-fit models. We further estimated and removed common-mode noise for all the stations using noise common to the coordinate time series of 20-70 well-behaved continuous GPS stations within 2000 km of Jamaica
(representing the maximum interstation distance over which GPS noise remains strongly correlated, i.e., Marquez-Azua & DeMets 2003). For the 30 stations in Jamaica, the common-
15 mode noise corrections reduces the magnitudes of the random and longer-period noise in the
station coordinate time series by ~20 percent in both horizontal components relative to the
unadjusted coordinate time series. The corresponding changes in the GPS station velocities
average only 0.05 and 0.15 mm yr-1 in the north and east velocity components, but range up to 0.7 and 1.1 mm yr-1 in the north and east velocity components at two campaign sites.
Each GPS site velocity was transformed from ITRF2008 to a Caribbean plate reference
frame (Fig. 3) by subtracting from each site velocity a velocity that is predicted at the site by
the angular velocity for the Caribbean plate relative to ITRF2008 (Table 2). We determined
the angular velocity for the Caribbean plate relative to ITRF2008 from the motions of 12
GPS sites on the Caribbean plate, five located in the eastern Caribbean, one in southern
Hispaniola, and six in the western Caribbean and Central America. Both the angular velocity
and its weighted root-mean-square misfit, 0.93 mm yr-1 in both the north and east velocity components, are close to those reported by DeMets et al. (2007) from an inversion of 15 velocities for Caribbean plate sites. The ITRF2008 velocities of all sites used for the analysis were corrected assuming the motion of ITRF2008 relative to Earth’s center of mass is the same as that for ITRF2005, 0.3, 0.0, and 1.2 mm yr-1 in the X, Y, and Z directions,
respectively (Argus 2007). Given the small geographic extent of our study area, none of our
results are sensitive to the geocentral translation correction that is applied to all of the GPS
velocities in the analysis.
7"7" %89$)*1/-,2,*5$.*1(2,)*$2/$2>*$?(.,@@*('$01(2*$
Jamaica: Relative to the Caribbean plate, all GPS sites in Jamaica move generally to the
SW at rates that decrease from 7.3±1.0 mm yr-1 at locations in northern Jamaica to 3.9±0.8
16 mm yr-1 at locations in southern Jamaica (Fig. 3). Highly oblique convergent motion across
the east-west-trending Gônave-Caribbean plate boundary thus occurs, consistent with the
velocities described previously by DWG07. The average southward component of motion of
2.6±0.6 mm yr-1 is remarkably consistent everywhere on the island (Fig. 4b), in contrast to the decrease from 6.5±1.0 mm yr-1 along the island’s north coast to 2.3±0.6 mm yr-1 along its
southern coast in the magnitude of the west-directed component of motion (Fig. 4a).
Consequently, GPS site directions rotate progressively CCW between the north and south
coasts of the island (Fig. 3).
The velocities of GPS sites of MCAY and PEDR ~50 and 80 km south of the main island
(Figs 3 and 4) reinforce the patterns described above. The southward velocity component at
both sites agrees with the average 2.6±0.6 mm yr-1 southward component of motion measured
at all the Jamaican stations (Fig. 4b), possibly suggesting that south-directed convergence is
accommodated within the northern Nicaragua Rise south of Jamaica. The west-directed
velocity component at both sites agrees with the velocity gradient defined by the GPS sites
on the main island (Fig. 4a) and differs insignificantly from zero at site PEDR ~80 km south
of the island. This suggests that all shear-related deformation occurs on plate boundary faults
north of site PEDR.
The velocity gradients described above are even better displayed when site velocities are
referenced to a station centrally located on the island (Fig. 3b). Relative to the continuous
GPS station PIKE (Table 2), sites located north of the Plantain Garden fault and the central
Jamaica fault system are either stationary or move no faster than ~1-2 mm yr-1 to the SE (Fig.
3b). In contrast, sites located south of the central Jamaica fault system move to the ESE at rates that increase from 1 to 6 mm yr-1 southward from the central Jamaica fault system.
17 Other locations: Figure 5a shows velocities for the other 96 stations used in our analysis.
The velocities of sites in Hispaniola are described and modeled by Calais et al. (2010); we
reserve an in depth analysis and discussion of those velocities for Chapter 2. Velocities from
Puerto Rico and vicinity are also described and used by Chapter 2. We refer readers to
Jansma & Mattioli (2005) for modeling and interpretation of earlier velocity fields for the
Puerto Rico-Virgin Island region.
4. Plate-boundary geometry and fault slip rates from block modeling
We next use the GPS velocities described above and shown in Figure 5a to evaluate a series of models with differing assumptions about how faults in Jamaica transfer slip from the Plantain Garden fault of southeastern Jamaica to the Walton fault offshore of western
Jamaica. Previous authors have variously proposed that slip is transferred by (1) left-lateral shear across a broad, east-west striking zone that crosses the island (Burke et al. 1980;
Wadge & Dixon 1984), (2) two right-stepping restraining bends that connect the Plantain
Garden and South Coast fault zone to the Duanvale fault (Mann et al. 1985), and (3) a series of CCW-rotating blocks bounded by the island’s major E-W strike-slip faults (Draper 2008).
Following their lead, we first test different possible geometries and locations for an assumed discrete Gônave-Caribbean plate boundary passing through the island (Fig. 6). We then test more complex models in which deformation on the island is characterized using one fault-bounded block (Fig. 7). Throughout the analysis, we use geologic and seismic information to constrain the proposed fault and block geometries and use the model fits for rigorous comparisons of the alternative models. Although our GPS velocity field is too
18 widely spaced to test Draper’s (2008) concept of multiple rotating blocks on the island, we show below that the velocity field is well fit by a simpler, one-block model.
A"#" B/<*1$+*2>/<5$('<$(554+02,/'5$
Our models were produced using Blocks software described by Meade & Loveless
(2009), who apply linear spherical block theory to decompose interseismic GPS velocities ( v v ˜ I ) into the rotation of fault-bounded blocks (˜ B ), internal homogenous strain within the v v blocks (˜" ), and elastic strain accumulation on the faults (˜ E ). This relationship can be expressed by:
v v v v ˜ I = ˜ B + ˜" + ˜ E , [1]
(from Meade & Loveless 2009, equation 1). The rates and directions of fault slip predicted by a given block model are determined solely from the relative rotations of the adjacent blocks and fault geometry. These rotations, however, are estimated from an inversion of the observed interseismic velocities that simultaneously estimates the effects of the block rotations, internal homogeneous block strain, and elastic strain from frictional locking of all the faults in the model. We do not estimate internal homogeneous strain of any of the blocks in our models, primarily because too few GPS sites are located on any likely block in
Jamaica to estimate both its rotation and internal strain.
Three assumptions are required about faults for our modeling, as follows: their locking depths, their dips, and the nature of frictional coupling across the faults. Wiggins-Grandison
(2004) finds that relocated earthquakes in Jamaica occur primarily at depths between 10 and
22 km. Here, we assign a uniform fault-locking depth of 15 km. We also evaluated model results for locking depths as shallow as 10 km and as deep as 20 km, but determined that the
19 estimated fault slip rates changed by only 10-15 percent, within their estimated uncertainties.
Strike-slip faults are assigned vertical dips; reverse faults are assigned dips of 60°, based on the known fault geometries of Jamaica, earthquake focal mechanisms, and gravity modeling
(e.g. Horsfield 1974; DeMets and Wiggins-Grandison 2007; Chapter 3).
For our modeling, we assume complete and uniform interseismic coupling across all block boundary faults, thereby maximizing the elastic deformation component in equation
(1). Consequently, the angular velocities that describe each block’s motion are the only parameters estimated during each velocity field inversion. As described below, this simple model fits GPS velocities in Jamaica within their estimated uncertainties.
We impose a full interseismic fault locking assumption for two reasons. First, only one
GPS transect of Jamaica’s E-W-striking faults has enough stations (77.4°-77.3°W in Fig. 3) to reliably estimate the local magnitude of interseismic coupling. Second, an inverse problem in which block rotations and fault coupling are estimated simultaneously may be poorly posed at deformation rates as slow as those in Jamaica because the sub-mm yr-1 differences
between the velocity field gradients associated with fully- or partially-locked faults are
smaller than the underlying GPS velocity uncertainties. At these slow rates, estimates of fault
coupling and block rotations will trade off strongly, guaranteeing that neither will be well
determined. The assumption of full interseismic coupling is a limiting factor of our analysis
and merits further investigation when more closely spaced, better-determined GPS site
velocities become available.
An inversion of the GPS site velocities described above using Blocks gives as output an
estimate of the angular velocity for each block in the model and the angular velocity
uncertainties. Each GPS site is affiliated with a block in the model depending on the
20 geometry of the faults that define the blocks; the estimated angular velocities therefore depend implicitly on the block geometry. The angular velocity of a given block also depends, to varying degrees, on the velocities of GPS sites exterior to its boundaries due to elastic deformation associated with locked faults along all the block boundaries.
A"!" 81(2*$C*/+*2.,*5$('<$;,'*+(2,-$-/'52.(,'25$
The boundaries of the North America plate, the largest block in the model, are well known and are not varied in the models tested below. Similarly, the boundaries of the
Caribbean plate are well defined and not varied except along its boundary with the Gônave microplate in Jamaica. We use the MORVEL Caribbean-North America angular velocity
[73.9° S, 32.6° E, 0.190°/Myr; DeMets et al. 2010] to tie the North America plate to the
Blocks model.
Although the northern, western, and most of the southern boundaries of the Gônave microplate are well defined by the Oriente fault, Cayman spreading center, and Walton,
Enriquillo, and Plantain Garden faults (Fig. 1), the location of the eastern boundary of the
Gônave microplate in Hispaniola is poorly defined. Specifically, the boundary could be located as far east as the Mona Passage between eastern Hispaniola and Puerto Rico (Fig. 5), as assumed by Manaker et al. (2008), or might be located in central or western Hispaniola
(Mann et al. 1995, 2002). In a related paper, we use the same GPS velocities from the northern Caribbean to demonstrate that a Hispaniola block moves independently from the
Gônave microplate and has a shared, likely diffuse boundary in western Hispaniola (Chapter
2). Hereafter, we use the Hispaniola block geometry preferred in Chapter 2 (Fig. 5).
21 All our models include a North Hispaniola block bordered by the Septentrional fault to
the south and the northern Hispaniola fault and Puerto Rico trench to the north (Fig. 5),
consistent with the regional block configuration used by Manaker et al. (2008) and Calais et al. (2010).
A"7" ?/+0(.,5/'5$/D$<,5-.*2*$@/4'<(.=$('<$@1/-;$C*/+*2.,*5$
A"7"#" B*2>/<5$
The goodness-of-fit for each assumed geometry for the Gônave-Caribbean plate
2 boundary in Jamaica is quantified using reduced chi-squared (! ") from the Blocks inversion
2 2 (Table 3), where ! " is the weighted, summed least-squares misfit ! divided by the degrees
2 of freedom in the model. Values of ! " that are smaller than 1 indicate that a model geometry
2 fits the GPS velocities within their estimated uncertainties. Conversely, values of ! " greater than 1 indicate that the misfits exceed the estimated uncertainties. In order to better compare
2 models and their fits to the subset of GPS site velocities of Jamaica, we calculate ! " for just
the 30 sites in the Jamaica archipelago (Table 3). The degrees of freedom are defined by the
number of GPS velocity components that are inverted (126 velocities and 252 velocity
components) to estimate all of our trial models reduced by the number of parameters that are
adjusted to fit those velocities.
All the models we tested have either five plates (i.e. Caribbean, Gônave, Hispaniola,
North Hispaniola, and Puerto Rico-Virgin Islands) or six plates (including Jamaica) whose
angular velocities are estimated. We use the Stein & Gordon (1984) F-ratio test to evaluate
the improvements in fit of the more complex six-plate models relative to the five-plate
22 models; the F-ratio test is well suited for this analysis given its inherent insensitivity to
incompletely known data uncertainties. Best-fitting angular velocities and their uncertainties
are given in Table 4 for plates directly relevant to this analysis (Jamaica, Gônave, Caribbean,
and North America). Angular velocities for other microplates included in this analysis
(Hispaniola and Puerto Rico-Virgin Islands) are documented in Chapter 2.
A"7"!" E*525$/D$<,5-.*2*$01(2*$@/4'<(.=$C*/+*2.,*5$
We first tested geometries for the Gônave microplate boundary through Jamaica,
assuming in each case that the plate boundary is defined by one or more faults that connect to
form a continuous, discrete boundary. Each boundary follows the traces of known faults and
is guided to some degree by earthquake epicenter locations (Figs 5 and 6). The geometry
referred to as the northern discrete model, in which the Blue Mountains of eastern Jamaica
form a large restraining bend that links the Plantain Garden fault of southeast Jamaica to the
Duanvale fault of northern Jamaica (Fig. 6a) fits the data more poorly than any discrete-
boundary geometry we tested (including geometries not shown in Fig. 6). The combined
weighted root-mean-square (WRMS) east and north velocity misfits of 1.4 mm yr-1 for the 30
Jamaica GPS sites (Table 3) for this geometry exceeds their estimated velocity uncertainties
2 by a factor of 1.9 (i.e., #! " = #3.5).
The central discrete geometry in which slip on the Plantain Garden fault is transferred to the central Jamaica fault system across a short segment of the Blue Mountain restraining bend (Fig. 6b) better fits the GPS velocities than the other discrete boundary geometries we tested (Fig. 6d), mainly due to improved fits to GPS velocities in central and northern
Jamaica. The WRMS misfit for this geometry is 1.0 mm yr-1, about 30% larger than the
23 average velocity uncertainty. The velocities of sites located north of the assumed boundary are well fit for this geometry (Fig. 6b). South of the assumed boundary, eight of the ten stations that are located 20 km or more south of the assumed boundary have southwest- directed residual motions of 0.5 to 2.2 mm yr-1, including the two GPS sites at Pedro and
Morant Cays south of the main island.
Finally, we tested a southern discrete geometry in which slip on the Plantain Garden fault is transferred by the Aeolus Valley fault to the South Coast fault zone and is then transferred northward to the offshore Walton fault across an assumed restraining bend along the southwest coast of Jamaica (Fig. 6c). Although the WRMS misfit for this geometry is only about 15% larger than for an assumed discrete boundary in central Jamaica (Fig. 6b), site velocities in the northern half of the island are fit more poorly for an assumed boundary in southern Jamaica.
We also tested other more complex discrete boundary geometries assuming different locations for the restraining bend in Jamaica. None, however, improved the fit relative to the central discrete geometry in which Gônave-Caribbean plate motion is assumed to follow a narrow boundary in central Jamaica (Fig. 6b). The best discrete boundary geometry thus misfits the GPS velocities at a level ~30% larger than the estimated velocity uncertainties, and moreover leads to a systematic southwest-directed residual velocity field in southern
Jamaica and south of the island. We thus reject the hypothesis that the plate boundary is narrow and test more complex geometries in the following section.
24 A"7"7" E*525$/D$+/.*$-/+01*F$01(2*$@/4'<(.=$C*/+*2.,*5$
Each of the more complex plate boundary geometries we tested includes a single fault-
bounded block that is sandwiched between the Gônave microplate and Caribbean plate (Fig.
7). We tested eight geologically plausible block models, each with three additional degrees of
freedom relative to the discrete boundary models described above. Six of the eight block
models improved the fit relative to the best discrete boundary model at the 99% or better
confidence level (Table 3).
Two models failed to significantly improve the fit, one that includes a block defined by
the major faults that bound the topographically high and seismically active Blue Mountains
of eastern Jamaica (Fig. 7b), and the other includes a block bounded by the Duanvale fault
and the central Jamaica fault system (Fig. 7d). The former model fits the data worse than any
2 other block model (! " = 2.9), consistent with the poor fit of the analogous discrete-boundary geometry (Fig. 6a). The motion of the block is constrained solely by stations along its edges, making this a weak test for a Blue Mountain block. The latter model, shown in Fig. 7d, reinforces the poor fit we found for the northern discrete boundary model (Fig. 6a). These argue against significant slip along the Duanvale fault of central Jamaica, where our GPS network crosses the fault as part of a N-S transect of the island (Figs 2b and 3).
One of the two best-fitting block fault models defines an elongate southern block bounded by the central Jamaica fault system and South Coast fault zone (Fig. 7e). This model fits the GPS velocities better than all three discrete boundary models (Fig. 6b) with a WRMS misfit (0.9 mm yr-1) that is only 15% larger than the estimated velocity uncertainties.
Reflecting this good fit, the southern block model fits the GPS velocities everywhere on the
25 island. Significant misfits of this model to the GPS velocities at Morant and Pedro Cays ~50-
80 km south and southeast of the main island (Fig. 6e) are similar to those for all but one of the block models and are treated below.
Given the success of the southern block geometry, we examined whether two plausible variations on the assumed southern block geometry would further improve the fit (Figs 7f and g). A geometry in which the eastern restraining bend is assumed to coincide with the
Porus fault of south-central Jamaica, which is mapped but not named on the 1992 Jamaica
Mining and Commerce structure maps, instead of the Blue Mountain restraining bend in eastern Jamaica (Fig. 7f) degrades the fit and is thus rejected. A model in which motion along the central Jamaica fault system is assumed to step northward to the Walton fault along the
Montpelier-Newmarket fault zone of northwestern Jamaica (Fig. 7g) improves the fit and has
an average velocity misfit equal to the average estimated uncertainty. This model predicts
4.6±1.0 mm yr-1 of left-lateral motion on the Montpelier-Newmarket zone of western
Jamaica. Given the absence of major throughgoing strike-slip faults in this area, we are skeptical of this model. Nonetheless, given the sparse distribution of GPS stations in northwestern Jamaica, more stations are needed for a stronger test of this and other geometries in which faults in northwestern areas of the island transfer slip northward off the central Jamaica fault system.
Finally, given that 2-3 mm yr-1 of southward motion is measured everywhere in Jamaica, including its southern cays (Fig. 4b), and that none of the discrete or block models described above fit the velocities at the cays (Figs 6 and 7a-g), we constructed a model in which Pedro and Morant cays are assumed to lie on the same block as much of southern Jamaica (inset to
26 Fig. 7h). In effect, this model tests whether deformation can be partitioned into largely E-W
shear on Jamaica and southward convergence south of the main island and cays.
The model described above (Fig. 7h) fits the velocities better than any of the previous
2 -1 discrete or block models, with ! " of 0.84 and combined WRMS of 0.7 mm yr (Fig. 7i). The
average velocity misfit is thus only 90% of the estimated velocity uncertainties, making this
model unique amongst the many models we tested. Although deformation south of the island
is almost surely accommodated by multiple faults over a wide area within the Nicaragua
Rise, the good fit of our simplified geometry is encouraging. Given its superior fit, we adopt
this hereafter as our preferred model and in Table 4 give the angular velocities that fully
specify this model. Further discussion of the block motions and fault slip predicted by this
model is found in the next section.
To determine whether the more complex single-block model is warranted by the improvement in fit, we compared the fits of the best discrete boundary model (Fig. 6d) and both the southern block geometry (Fig. 7e) and Nicaragua Rise block geometry (Fig. 7h).
Using the Stein & Gordon (1984) F-ratio test for an additional plate and least-squares misfits
!2 = 375.5 for the best discrete boundary model and !2 = 300.1 for the Nicaragua Rise block
geometry (Fig. 7h), the value for F is 19.6. For comparison, the 99% threshold for a
significant improvement in fit for 3 versus 234 degrees of freedom is F = 3.87. The
probability that random errors in the GPS velocities could result in this F-value is only 2
parts in 1011. The additional block is thus warranted at much greater than the 99% confidence
level. Repeating the calculations for the southern block geometry (Fig. 7e) gives F = 9.7 (!2 =
333.9 for this geometry; Table 3). The improvement in fit is also significant at a high
27 confidence level, with a probability of only 5 parts in 106 that random errors in the GPS
velocities could result in this F-value.
5. Discussion
G"#" %89$)*1/-,2=$D,*1<$,'2*.0.*2(2,/'$
Independent of any modeling, the gradient in the Jamaica GPS velocity field along a N-S transect of the island (Fig. 3b) strongly indicates that one or more active plate boundary faults are located on the island. In particular, relative to a fixed GPS site (PIKE) near the center of the island, all sites in Jamaica north of the central Jamaica fault system move only 1
mm yr-1 or slower, consistent with little or possibly no deformation in much of the northern
half of the island. In particular, four stations that comprise a N-S transect of the Duanvale
fault near 77.4°W show no evidence for active slip across that fault. Similarly, GPS site
NGLF in western Jamaica and north of the central Jamaica fault system moves with the same
velocity (within uncertainties) as GPS site CASL on the east coast of Jamaica (Fig. 3 and
Table 2). By implication, little or no east-to-west contraction appears to occur across the
northern half of the island.
All of the GPS sites on the southern half of the island move significantly relative to
PIKE, with velocities increasing to the south to rates as fast as 6 mm yr-1 to the east (Figs 3 and 4). To first order, the velocities are consistent with strike-slip motion on the central
Jamaica fault system, on the South Coast fault zone, or on both.
All Jamaica GPS sites move 2.6±0.6 mm yr-1 southward toward the Caribbean plate
interior (Fig. 4b), including the two sites located 50 and 80 km south of the island. The
28 consistent southward motion may indicate that ~N-S contraction may occur in the northern
Nicaragua Rise south of Jamaica. Some evidence supports this explanation – one of the two
largest earthquakes in or near Jamaica in the past 70 years was the Mo = 6.9 April 7, 1941 earthquake ~80 km southwest of the island within the Nicaragua Rise (Event #1 in Fig. 2c;
Van Dusen & Doser 2000). The P-axis for this strike-slip earthquake is oriented NNE-SSW, consistent with approximate N-S contraction across the Nicaragua Rise. Alternatively, the slow southward motions of all the network sites could be an artifact of a ~2 mm yr-1
systematic bias in the plate velocities predicted by the Caribbean plate angular velocity near
Jamaica. We consider this less likely because the Caribbean plate angular velocity estimate is
relatively robust with respect to the subset of site velocities that are used to estimate it
(DeMets et al. 2007).
G"!" B/<*1$,+01,-(2,/'5$('<$1,+,2(2,/'5$
G"!"#" H,5-.*2*$)*.545$<,52.,@42*<$01(2*$@/4'<(.=$,'$6(+(,-($
A key goal of our work was to test whether GPS velocities from Jamaica are more
consistent with a geometry in which a discrete plate boundary passes through the island or
with plate boundary geometries that incorporate an independently moving block between the
Gônave and Caribbean plates. The highly significant improvements in the fits of both the
southern block and Nicaragua Rise block geometries (Figs 8a and b) relative to the fit for the
best-fitting discrete boundary geometry (Fig. 6b) argue against the simpler discrete boundary
assumption. Instead, 7±1 mm yr-1 of WSW-ENE motion between the Gônave and Caribbean
plates (Figs 8a and b) is accommodated across the boundaries of a block whose northern and
southern boundaries are defined respectively by the central Jamaica fault system and
29 unknown faults within or near southern Jamaica (Fig. 8a) or in the northern Nicaragua Rise
(Fig. 8b).
Seismic and geologic observations reinforce the kinematic evidence against the simpler discrete-boundary model. Seismicity in Jamaica is widespread (Figs 2c and 8), contrary to the presumption of concentrated deformation inherent in a discrete-boundary model.
Numerous earthquakes in the Blue Mountains of eastern Jamaica (Figs 2c and 8) and active anticlines (i.e. Long and Dallas Mountains in northeast Kingston) at the western edge of the
Blue Mountains (Draper 2008) argue strongly for the existence of such a restraining bend.
However, neither of the discrete-boundary models that include restraining bends solely in eastern Jamaica (Figs 6a and 6b) fits the GPS velocities well.
G"!"!" 8.*D*..*<$('<$(12*.'(2,)*$@1/-;$C*/+*2.,*5$
Two block geometries, shown in Figures 8ab, fit the GPS velocities close to or within their estimated uncertainties. For several reasons, we prefer the geometry that assumes deformation extends south of the island on as-yet unidentified faults within the Nicaragua
Rise (Fig. 8b). Foremost, this geometry fits the velocities ~20% better than the alternative geometry. Moreover, the Blocks velocity inversion that employs this geometry cleanly partitions deformation in the study area into left-lateral strike-slip motion along the WNW- trending central Jamaica fault system and N-S convergence across submarine structures south of Jamaica, in accord with the geological and seismic evidence for shear-dominated deformation on the island. Finally, the elastic component (red arrows in Fig. 8c) of the GPS velocities predicted by Blocks for the preferred geometry parallels Jamaica’s strike-slip faults
30 everywhere on the island, in accord with elastic deformation observed in many strike-slip settings worldwide (e.g., Genrich et al. 2000; Pearson et al. 2000; Spinler et al. 2010).
Using the alternative geometry (Fig. 8a), the Blocks model predicts 4-5 mm yr-1 of N-S convergence across the South Coast fault zone of southern Jamaica, inconsistent with the geology and topography of this apparent strike-slip fault zone.
For the preferred geometry, the Caribbean-Gônave plate angular velocity (Table 4) predicts 6-7 mm yr-1 of left-lateral slip between the two plates east and west of Jamaica (Fig.
8b). This motion is partitioned into ~4-5 mm yr-1 of left-lateral, fault-parallel slip along the
Plantain Garden fault and central Jamaica fault system (Fig. 8b) and 2.4±1 mm yr-1 of N-S convergence across the hypothetical block boundary south of Jamaica (Fig. 8b).
One weakness of the preferred model is the absence of compelling seismic or marine geophysical evidence for active deformation within the Nicaragua Rise south of Jamaica. We are unsurprised that deformation in this area has gone unrecognized given the slow predicted deformation rate, the absence of high-resolution marine surveys in this area, and the possibility that multiple faults within a diffuse deformation zone accommodate the convergence. A second shortcoming of the new model is that the current resolution of the
GPS and seismic data put most of the seismic hazard on the Plantain Garden fault, active faults in the Blue Mountain restraining bend, and the central Jamaica fault system. We assess the possible seismic hazard for the Kingston area in Section 5.2.4 based on this model. We consider it unlikely that all motion occurs on this series of faults based on recent work
(Chapter 3); however, stronger tests for active deformation on faults other than those listed above will require more widely distributed GPS sites and better determined site velocities.
31 G"!"7" I+01,-(2,/'5$D/.$50*-,D,-$D(412$J/'*5$
We next describe briefly the implications of the preferred (Fig. 8b) and other models
(Figs 6 and 7) for several prominent fault zones in Jamaica, namely, the Blue Mountains restraining bend of eastern Jamaica, the central Jamaica fault system, the South Coast fault zone and its postulated eastern continuation the Aeolus Valley fault, and the faults in northern and eastern Jamaica, where little deformation appears to occur. Because the formal velocity uncertainties estimated by the Blocks software are based on the assumption that the fault locking depths and block geometries are perfectly known, neither of which is true for our analysis, the formal uncertainties are surely too small. Numerical experiments suggest that the true uncertainties may be a factor of 2-3 times larger than the formal uncertainties.
Uncertainties stated below represent our best estimate for a given rate.
Blue Mountains and a restraining bend in eastern Jamaica: Although we approximate the Blue Mountain restraining bend of eastern Jamaica as a single fault at the western edge of the Blue Mountains, the widespread seismicity associated with this restraining bend (shown in Figs 8a and b) instead suggests that 2.6-4.0 mm yr-1 of predicted contraction (Figs 8a and b) is accommodated by multiple structures, possibly including the Blue Mountain fault, the
Yallahs faults (Mann & Burke 1990), and actively growing anticlines in the capital Kingston
(Draper 2008). Although the topographically prominent John Crow Mountains of far eastern
Jamaica (JC in Figs 2a and 5c) may accommodate some restraining bend deformation, none of the discrete-boundary or block geometries that include the John Crow Mountains as a block boundary adequately fit the GPS velocity field (e.g., Figs 7a, b, and d). Discrete boundary and block geometries that extend the Blue Mountain restraining bend farther north
32 along the Wagwater Belt (fault WW in Figs 2a and 5c and geometries shown in Figs 6a and
7b) also fit the velocities poorly, consistent with the northward decrease in both the seismicity and topography associated with the Blue Mountain restraining bend (Fig. 8b).
Central Jamaica fault system and South Coast fault zone: Estimates of present slip rates
along the central Jamaica fault system range from 2.4 to 4.2 mm yr-1 of left-lateral strike-slip in central Jamaica (Figs 8a and b), depending on which block geometry is employed. Slip of
4.2±1 mm yr-1 is predicted in central Jamaica by our preferred model (Fig. 8b), whereas the
South Coast fault zone is assumed to accommodate negligible or no slip. In contrast, the
southern block model (Fig. 8a) partitions Gônave-Caribbean plate slip nearly evenly between
the South Coast fault zone (2.2-2.8 mm yr-1), and the central Jamaica fault system (2.4-3.0
mm yr-1), with negligible deformation assumed within the Nicaragua Rise south of Jamaica.
Reasonable bounds for slip rates along both fault systems are 2-4 mm yr-1, depending on the assumed block geometry (Figs 8a and b).
Aeolus Valley fault: The southern block geometry requires that some motion on the
Plantain Garden fault be transferred southwestward to the South Coast fault zone (Fig. 8a) by one or more faults onshore and south of Jamaica. The Aeolus Valley fault, which intersects the eastern Plantain Garden fault (Fig. 2a), may be the best candidate for transferring this slip. The southern block model predicts that Caribbean-Gônave plate motion is partitioned nearly evenly between the Aeolus Valley fault and western Plantain Garden fault, such that each fault accommodates ~3-5 mm yr-1 of oblique slip motion (Fig. 8b).
Duanvale fault and northern Jamaica: Although previous authors have hypothesized
that the Duanvale fault plays an important role in Jamaica’s tectonics (Mann et al. 1985;
Mann et al. 2007; Draper 2008), none of the geometries that include the Duanvale fault give
33 an acceptable fit to the GPS velocity field (Figs 6a, 7a and b). Combined with an absence of seismicity along much of the fault, we infer that at least the eastern segment of the Duanvale fault is inactive or slips too slowly to detect with GPS. Significant slip, however, may still occur along the Duanvale fault in northwestern Jamaica, where only one GPS station is located (site PYRA in Fig. 3). In this region, several small earthquakes along the Duanvale
fault just east of Montego Bay (Fig. 8b) indicate that the fault is still active. A Mw = 4.5 left- lateral oblique slip earthquake that occurred in 2005 in central Jamaica north of the central
Jamaica fault system (Fig. 2c, event #48) may indicate that some slip on the central Jamaica fault system is transferred north to the western portion of the Duanvale fault. Additional GPS sites in northwest Jamaica are needed to determine whether the fault is active in this region.
Walton fault: SeaMARC II side-scan mapping off the coast of western Jamaica identifies multiple active strands of the Walton fault, one at the same latitude (Fault E) as the central
Jamaica fault system (Tyburski 1992). Modeling of our GPS velocities suggests that significantly more slip occurs along the central Jamaica fault system, east of Fault E, rather than along the Duanvale fault of northern Jamaica. Some slip along the central Jamaica fault system may thus continue westward to Fault E instead of stepping north to the Duanvale fault. If correct, this implies a smaller right-stepping restraining bend than postulated by previous authors, who instead envision much of the island as a restraining bend (Burke et al.
1980; Wadge & Dixon 1984; Mann et al. 1985; Draper et al. 2008).
G"!"A" I+01,-(2,/'5$D/.$5*,5+,-$>(J(.<$
Elastic strain accumulation appears to be highest in eastern Jamaica, where deformation is localized along the Plantain Garden fault and Aeolus Valley fault. Consistent with the
34 conclusions of DWG07, our results imply high seismic risk for the capital city Kingston and
its surrounding suburbs, which are the most densely populated areas on the island (Fig. 2a).
Using plausible assumptions for the maximum rupture length of a future earthquake in
eastern Jamaica (~150 km), our modeling results indicate that enough elastic strain has
accumulated since the destructive 1907 earthquake in eastern Jamaica to generate a Mw $ 7-
7.5 earthquake.
G"!"G" B/<*1$1,+,2(2,/'5$('<$4'-*.2(,'2,*5$
The slow deformation rates on the island (<7 mm yr-1) and heterogeneous distribution of the GPS sites limit our ability to resolve deformation in more detail, even given the small uncertainties in our site velocities. For example, although east-west contraction likely occurs across the Santa Cruz and Don Figuerero Mountains in the interior of the southern block, we are unable to resolve any E-W contraction within this block given the slow deformation rates.
Most of what we know about where fault slip extends offshore in western Jamaica is based on the velocities of just three GPS sites (PYRA, NGLF, and FONT in Fig. 3a). Although these sites suggest that most and possibly all motion exits the island south of site NGLF, presumably on the central Jamaica fault system, better site coverage is needed to determine whether some fault slip may be transferred northward from the central Jamaica fault system to the Duanvale fault near the densely populated tourist area of Montego Bay (Figs 2a and
8a).
Because of the limited number of GPS sites, we largely limited our exploration of alternative block models to those with a single block. Given that the best-fitting block geometry (Fig. 8b) reduces the data variance by 98.3% and has a WRMS misfit of only 0.7
35 mm yr-1, our approach seems warranted. As a test, we subdivided the Jamaica block in our preferred geometry (Fig. 8b) into two blocks, one consisting of the elongate block in southern
Jamaica (Fig. 8a) and the other extending southward from the South Coast fault zone into the
Nicaragua Rise. Inverting the GPS velocities with this more complex geometry failed to improve the fit significantly and moreover predicted insignificant slip along the South Coast fault. A more complex block model awaits better-determined velocities at the present sites and new measurements in other areas of the island.
6. Conclusion
Relative to the Caribbean plate, all GPS sites in Jamaica move generally to the SW at rates that decrease from 7.3±1.0 mm yr-1 at locations in northern Jamaica to 3.9±0.8 mm yr-1 at locations in southern Jamaica. GPS sites located north of the central Jamaica fault system have differential movements that are typically 1 mm yr-1 or less, consistent with little or possibly no deformation in much of the northern half of the island. In contrast, a well-defined
GPS velocity gradient in central and southern Jamaica indicates that elastic deformation is concentrated across one or more faults in these areas of the island or south of the island.
Modeling of GPS velocities from Jamaica and adjacent islands for a series of candidate geometries for the Gônave-Caribbean plate boundary yields two best-fitting geometries, one that assigns most of southern Jamaica to an independently moving block and another that partitions strike-slip motion along the Plantain Garden fault and central Jamaica fault system and southward convergence in the Nicaragua Rise. Both models improve the fit to the velocity field at better than the 99% confidence level relative to a best-fitting discrete- boundary geometry and one fits the velocities within the estimated uncertainties. Both
36 models partition 7±1 mm yr-1 of Gônave-Caribbean plate boundary motion between the
central Jamaica fault system and faults farther south. Both moreover predict 3-4±1 mm yr-1 of convergence across the seismically active Blue Mountain restraining bend of eastern
Jamaica, adjacent to the densely populated capital city of Kingston.
None of the models that postulate significant slip along the prominent Duanvale fault of northern Jamaica adequately fit the GPS velocity field. In particular, slip faster than 1 mm yr-
1 seems unlikely along the eastern half of the Duanvale fault. Almost no GPS velocities constrain our results along the western half of the Duanvale fault, where the fault passes near the densely populated tourist area of Montego Bay. Seismic hazard in northwestern Jamaica is thus still more poorly defined than elsewhere on the island.
7. Acknowledgments
None of this work would have been possible without more than a decade of dedicated support from the Jamaica Earthquake Unit of the University of West Indies. We thank
Gregory Peake of Spatial Innovision in Kingston and the Jamaican National Land Agency for continuous data used in our analysis, and Glendon Newsome and Dr. Raymond Wright for their support and data. We thank the Jamaica Coast Guard, for sea transport to Morant and
Pedro Cays and Dr. Byron Wilson for sea transport to the Hellshire Hills GPS site. We thank
Jack Loveless for patient advice on the use of his Blocks software and constructive comments as a reviewer, and further thank Glen Mattioli for his in-depth, constructive review. This project was funded by National Science Foundation grant EAR-0609578. Some of this material is based on data provided by the UNAVCO Facility with support from the
National Science Foundation (NSF) and National Aeronautics and Space Administration
37 (NASA) under NSF Cooperative Agreement No. EAR-0735156. Figures were produced using
Generic Mapping Tools software (Wessel & Smith 1991).
8. References
Altamimi, Z., Collilieux, X., & Métivier, L., 2011. ITRF2008: An improved solution of the
International Terrestrial Reference Frame, J. Geod., 85, 457-473, doi: 10.1007/s00190-
011-0444-49.
Argus, D.F., 2007. Defining the translational velocity of the reference frame of Earth,
Geophys. J. Int., 169, 830-838, doi: 10.1111j.1365-246X.2007.03344.x.
Burke, K., Grippi, J. & Sengor, A.M.C., 1980. Neogene structures in Jamaica and the
tectonic style of the northern Caribbean plate boundary zone, J. Geol., 88, 375-386.
Calais, E., Freed, A., Mattioli, G., Amelung, F., Jónsson, S., Jansma, P., Hong, S.-H., Dixon,
T., Prépetit, C., & Momplaisir, R., 2010, Transpressional rupture of an unmapped fault
during the 2010 Haiti earthquake, Nature Geoscience, doi: 10.1038/NGEO992.
Clark, G.R., II, 1995, Swallowed up, Earth, 4, 34-41.
DeMets, C., Gordon, R.G., & Argus, D.F., 2010. Geologically current plate motions,
Geophys. J. Int., 181, 1-80, doi: 10.1111/j.1365256X.2009.04491.x.
DeMets, C., Mattioli, G., Jansma, P., Rogers, R.D., Tenorio, C., & Turner, H.L., 2007.
Present motion and deformation of the Caribbean plate: Constraints from new GPS
geodetic measurements from Honduras and Nicaragua, in Geologic and Tectonic
Development of the Caribbean Plate in Northern-Central America, Geol. Soc. Am.
Spec. Paper, ed. Mann, P., 428, 21-36, doi: 10.1130/2007.2428(02), The Geological
Society of America, Boulder.
38 DeMets, C., & Wiggins-Grandison, M., 2007. Deformation of Jamaica and motion of the
Gônave microplate from GPS and seismic data, Geophys. J. Int., 168, 362-378, doi:
10.1111/j.1365-246X.2006.03236.x.
Draper, G., 2008. Some speculations on the Paleogene and Neogene tectonics of Jamaica,
Geol. J., 43, 563-572.
Genrich, J.F., Bock, Y., McCaffrey, R., Prawirodirdjo, L., Stevens, C.W., Puntodewo,
S.S.O., Subarya, C., and Wdowinski, S., 2000. Distribution of slip at the northern
Sumatran fault system, Geophys. Res. Lett., 105, 28,327-28,341, doi:
10.1029/2000JB900158.
Grippi, J., 1978. Geology of the Lucea Inlier, western Jamaica, M.S. thesis, State University
of New York at Albany, Albany, New York.
Hayes, G.P., Briggs, R.W., Sladen, A., Fielding, E.J., Prentice, C., Hudnut, K., Mann, P.,
Taylor, F.W., Crone, A.J., Gold, R., Ito, T., and Simons, M., 2010. Complex rupture
during the 12 January 2010 Haiti earthquake, Nature Geoscience, doi:
10.1038/NGEO977.
Horsfield, W.T., 1974. Major faults in Jamaica, J. Geol. Soc. Jamaica, 14, 1-15.
Jansma, P.E., & Mattioli, G.S., 2005. GPS results from Puerto Rico and the Virgin Islands:
Constraints on tectonic setting and rates of active faulting, in Active tectonics and
seismic hazards of Puerto Rico, the Virgin Islands, and offshore areas, Geol. Soc. Am.
Spec. Paper, ed. Mann, P., 385, 13-30 doi: 10.1130/2007.2428(02), The Geological
Society of America, Boulder.
39 Leroy, S., Mercier de Lepinay, B., Mauffret, A., & Pubellier, M., 1996. Structural and
tectonic evolution of the eastern Cayman Trough (Caribbean Sea) from seismic
reflection data, AAPG Bull., 80, 222-247.
Manaker, D.M, Calais, E., Freed, A.M., Ali, S.T., Przybylski, P., Mattioli, G., Jansma, P.,
Prepetit, C., & de Chabalier, J.B., 2008. Interseismic plate coupling and strain
partitioning in the northeastern Caribbean, Geophys. J. Int., 174, 889-903, doi:
10.1111/j.1365-246X.2008.03819.x.
Mann, P. & Burke, K., 1990. Transverse intra-arc rifting: Palaeogene Wagwater Belt,
Jamaica, Mar. Pet. Geol., 7, 410-427.
Mann, P., Draper, G., & Burke, K., 1985. Neotectonics of a strike-slip restraining bend
system, Jamaica, in Strike-slip deformation and sedimentation, Special Publication,
SEPM 37, pp. 211-226, eds Biddle, K. & Christie-Blick, N., SEPM, Tulsa.
Mann, P., Taylor, F.W., Edwards, R.L., & Ku, T.-L., 1995. Actively evolving microplate
formation by oblique collision and sideways motion along strike-slip faults; An example
from the northeastern Caribbean plate margin, Tectonophysics, 246, 1-69.
Mann, P., Calais, E., Ruegg, J.-C., DeMets, C., Jansma, P., & Mattioli, G.S., 2002. Oblique
collision in the northeastern Caribbean from GPS measurements and geological
observations, Tectonics, 37, doi: 0.1029/2001TC001304.
Mann, P., DeMets, C., & Wiggins-Grandison, M., 2007. Toward a better understanding of
the Late Neogene strike-slip restraining bend in Jamaica: geodetic, geological, and
seismic constraints, in Tectonics of Strike-Slip Restraining and Releasing Bends, Special
Publications, 290, pp. 239-253, eds Cunningham, W.D. & Mann, P., Geological Society,
London.
40 Mann, P., Schubert, C., & Burke, K., 1990. Review of Caribbean neotectonics, in The
Geology of North America, The Caribbean region, eds Dengo, G., and Case, J.E., H,
307-338, Geological Society of America, Boulder, Colorado.
Mann, P., Taylor, F.W., Edwards, R.L, & Ku, T., 1995. Actively evolving microplate
formation by oblique collision and sideways motion along strike-slip faults: an example
from the northeastern Caribbean plate margin, Tectonophysics, 246, 1-69.
Marquez-Azua, B., & DeMets, C., 2003. Crustal velocity field of Mexico from continuous
GPS measurements, 1993 to June 2001: Implications for the neotectonics of Mexico, J.
Geophys. Res., 108, 2450, doi: 10.1029/2002JB002241.
Meade, B.J., & Loveless, J.P., 2009. Block modeling with connected fault-network
geometries and a linear elastic coupling estimator in spherical coordinates, Bull. seism.
Soc. Am., 99, 3124-3139, doi: 10.17850120090088.
Mitchell, S.F., 2003. Sedimentology and tectonic evolution of the Cretaceous rocks of central
Jamaica: Relationships to the plate tectonic evolution of the Caribbean, in The Circum-
Gulf of Mexico and the Caribbean: Hydrocarbon habitats, basin formation, and plate
tectonics, AAPG Mem., eds Bartolini, C., Buffler, R.T., & Blickwede, J., 79, pp. 605-
623.
Natural Disaster Research, Inc., Earthquake Unit, Mines & Geology Division, 1999.
Kingston Metropolitan Area Seismic Hazard Assessment Final Report, Caribbean
Disaster Mitigation Project, Organization of American States,
www.oas.org/cdmp/document/kma/seismic/kma1.htm, 82 pp.
Pearson, C., Denys, P., and Hodgkinson, K., 2000. Geodetic constraints on the kinematics of
the Alpine Fault in the southern South Island of New Zealand, using results from
41 Hawea-Haast GPS transect, Geophys. Res. Lett., 27, 1319-1322, doi:
10.1029/1999GL008412.
Pereira, J., 1977. An engineering seismology study of Jamaica, MSc thesis, Imperial College
London University, London, pp. 147.
Prentice, C.S., Mann, P., Crone, A.J., Gold, R.D., Hudnut, K.W., Briggs, R.W., Koehler,
R.D., and Jean, P., 2010. Seismic hazard of the Enriquillo-Plantain Garden fault in Haiti
inferred from palaeoseismology, Nature Geoscience, doi: 10.1038/NGEO991.
Rosencrantz, E. & Mann, P., 1991. SeaMARC II mapping of transform faults in the Cayman
Trough, Caribbean Sea, Geology, 19, 690-693.
Sandwell, D.T. & Smith, W.H.F., 1997. Marine gravity anomaly from Geosat and ERS 1
altimetry, J. Geophys. Res., 102, 10039-10054.
Spinler, J.C., Bennett, R.A., Anderson, M.L., McGill, S.F., Hreinsdottir, S., and McCallister,
A., 2010. Present-day strain accumulation and slip rates associated with southern San
Andreas and eastern California shear zone faults, J. Geophys. Res., 115, B11407, doi:
10.1029/2010JB007424
Stein, S. & Gordon, R.G., 1984. Statistical tests of additional plate boundaries from plate
motion inversions, Earth planet. Sci. Lett., 69, 401-412.
Taber, S., 1920. Jamaica earthquakes and the Bartlett Trough, Bull. Seism. Soc. Am., 10, 55-
86.
Tomblin, J.M. & Robson, G.R., 1977. A catalogue of felt earthquakes for Jamaica, with
references to other islands in the Greater Antilles, 1564-1971, Ministry of Mining and
Natural Resources (Jamaica), Mines & Geology Division Special Publication No. 2, 243
pp.
42 Tyburski, S.A., 1992, Deformational mechanisms along active strike-slip faults: SeaMARC
II and seismic data from the North America-Caribbean plate boundary, M.S. thesis,
University of Texas – Austin, 194 pp.
Van Dusen, S.R. & Doser, D.I., 2000. Faulting processes of historic (1917-1962) M ! 6.0
earthquakes along the north-central Caribbean margin, Pure appl. Geophys., 157, 719-
736.
Versey, H.R., Williams, J.B., & Robinson, E., 1958. The earthquake of March 1, 1957,
Geonotes; quarterly newsletter of the Jamaica Group of the Geologists Association, 1,
54-65.
Wadge, G. & Dixon, T.H., 1984. A geological interpretation of SEASAT-SAR imagery of
Jamaica, J. Geology, 92, 561-581.
Wessel, P. & Smith, W.H.F., 1991. Free software helps map and display data, EOS, Trans.
Am. Geophys. Un., 72, 441-446.
Wiggins-Grandison, M.D., 1996. Seismology of the January 1993 earthquake, J. Geol. Soc.
Jam., 30, 1-14.
Wiggins-Grandison, M.D., 2001. Preliminary results from the new Jamaica seismograph
network, Seism. Res. Lett., 72, 525-537.
Wiggins-Grandison, M.D., 2004. Simultaneous inversion for local earthquake hypocenters,
station corrections, and 1-D velocity model of the Jamaican crust, Earth planet. Sci.
Lett., 224, 229-240.
Wright, R.M., 1975. Aspects of the geology of Tertiary limestones in west-central Jamaica,
Ph.D. thesis, Stanford University, Stanford, CA, United States, 320 p.
43 Zumberge, J.F., Heflin, M.B., Jefferson, D.C., Watkins, M.M., & Webb, F.H., 1997. Precise
point positioning for the efficient and robust analysis of GPS data from large networks,
J. geophys. Res., 102, 5005-5017.
44
9. Figure captions
Figure 1. Tectonic setting of Jamaica and vicinity. Jamaica is located in a restraining bend on the left-lateral strike-slip boundary between the Caribbean plate and the Gônave microplate. Arrow shows MORVEL estimate of North America plate motion in mm yr-1 relative to the Caribbean plate (DeMets et al. 2010). CSC – Cayman spreading center. 2-min seafloor bathymetry and land topography are from Sandwell & Smith (1997).
Figure 2. A - Major faults of Jamaica modified from Wiggins-Grandison & Atakan (2005) and Chapter 3. Long-dashed line shows assumed offshore continuation of the Aeolus Valley fault. NNW-striking faults are BM - Blue Mountain fault, JC – John Crow fault, PO – Porus fault, ST – Spur Tree fault, SCr – Santa Cruz fault, WMN and EMN – Western and Eastern
Montpelier-Newmarket faults, respectively, and WW – Wagwater Belt. Strike-slip faults are:
AV – Aeolus Valley fault, CF – Cavaliers fault, DF – Duanvale fault, PG – Plantain Garden fault, RMCR – Rio Minho-Crawle River fault, SCFZ – South Coast fault zone, SFS – Siloah fault system, WF – Walton fault, YA – Yallahs fault, and approximate location of Fault E of the Walton fault (Tyburksi 1992). The CF, RMCR, and SFS constitute the central Jamaica fault system. Faults are overlain on a population density map as of 2007. Faults are also shown in panels B and C. B - Locations of major cities, including the capital city of
Kingston, and major geographic features. GPS station locations (red circles) on 90-m Space
Shuttle Topographic Radar Mission topography illuminated from the southwest. C -
Earthquakes from the International Seismological Centre for the period 2000-2010 (June); magnitudes range from 2.0 to 5.9. Earthquake focal mechanisms #1-2 are from Van Dusen &
45 Doser (2000), #3-4 from Wiggins-Grandison (2001), #5-48 from DeMets & Wiggins-
Grandison (2007), and #49-54 (shown in red) from the ISC for the period 2005-2008. Focal
mechanism parameters for events #49 to 54 are given in Table 1. Focal mechanisms are
scaled to magnitude. Black or red dots in the focal mechanisms indicate pressure axes.
Figure 3. A - Jamaica GPS site velocities relative to the Caribbean plate, with one sigma, 2-
D error ellipses. Continuous and campaign site velocities are shown in red and black, respectively. Star shows origin of N-S velocity transect in Figure 4. Velocity scale is in lower right corner of the map. B - Jamaica GPS site velocities relative to the campaign station
PIKE. Velocities shown in black are north of the Plantain Garden fault and central Jamaica fault system, whereas velocities shown in red are south of these faults. Velocity scale is in upper right corner of the map. Both figures show 90-m Space Shuttle Topographic Radar
Mission topography illuminated from the southwest.
Figure 4. A - West velocity component of GPS velocities from Fig. 3 versus distance along a
N-S transect whose origin is shown by the star in the map beneath. B - South velocity component of GPS velocities projected along the same transect. Gray shaded area is the 95% confidence interval in the weighted mean velocity component. All rates are in a Caribbean plate reference frame and are shown with their standard errors. GPS velocities from stations south of the Plantain Garden fault and central Jamaica fault system are shown in red. PEDR and MCAY indicate GPS velocities for stations on cays south of the Jamaica (locations shown in the map beneath). Fault locations are shown in Figure 2a.
46 Figure 5. A – Regional GPS velocities relative to Caribbean plate. Velocities from
Hispaniola are taken from Calais et al. (2010). Uncertainty ellipses are 2-D, one-sigma.
Velocity scale is in lower right corner of the map. Red lines mark plate boundaries and
major faults used for the analysis. Red arrows show North America plate motion predicted
by MORVEL. All plates and blocks included in the analysis are labeled.
B – Faults for northern Caribbean plate boundary used in the analysis. GPS site locations are
shown by red circles. Fault name abbreviations are: CSC – Cayman spreading center, EF –
Enriquillo fault, MP – Mona Passage, MT – Muertos trench, NHF – North Hispaniola fault,
OF - Oriente fault, PG – Plantain Garden fault, SF – Septentrional fault, and WF - Walton
fault. C - Simplified fault geometry of Jamaica used in the analysis. OF1, OF2, and OF3
refer to offshore faults 1, 2, and 3, respectively. Other fault name abbreviations are given in
the caption to Fig. 2a.
Figure 6. A-C – Discrete-boundary models and their residual velocity fields, as described in
text. Each assumed discrete-boundary is shown by the thick black line and is comprised of
some subset of the faults shown in Fig. 5c. Red arrows show the observed GPS site velocities
from Fig. 3a reduced by the velocity predicted by its corresponding discrete-boundary model.
Residual velocity scale is shown in upper right. Stations PEDR and MCAY from cays south
2 of the main island (Fig. 3) are shown as insets. D - Reduced chi-squared (! ") for all three
2 discrete-boundary models – lower values of ! " correspond to improved fits.
Figure 7. A-H. Fits of candidate block models for deformation in Jamaica, as described in text. Block boundaries are shown by black lines and are comprised of subsets of the faults
47 shown in Fig. 5c. Red arrows show the observed GPS site velocities from Fig. 3a reduced by
the velocity predicted by its corresponding block model. Residual velocity scale is shown in
upper right. Stations PEDR and MCAY from cays south of the main island of Jamaica (Fig.
3) are shown as insets. Inset in H shows block geometry of Nicaragua Rise model. I -
2 2 Reduced chi-squared (! ") for all eight block models – lower values of ! " correspond to improved fits.
Figure 8. Slip rates predicted by the southern block model (Panel A) and the Nicaragua Rise block model (Panel B). All slip rates are in mm yr-1 and are specified as boundary-parallel
and boundary-normal components for each fault segment. Red circles show ISC seismicity
from Figure 2. Slip rate estimates along the Blue Mountains fault are 2.0 mm yr-1 of left- lateral motion and 2.6 mm yr-1 of convergence for Panel A and 2.5 mm yr-1 of left-lateral
motion and 4.0 mm yr-1 of convergence The dashed line in Panel A marks both the Duanvale fault and a zone where motion may be transferred to it from the central Jamaica fault system.
C - Elastic component of the modeled GPS velocities (blue arrows) for the southern block model shown in Panel A and the Nicaragua Rise model shown in Panel B (red arrows). Red- blue dashed line shows the northern block boundary assumed for both geometries. Velocity scale is in the lower left corner.
48
49
50
51
52
53
54
55
56
57
58
59
60
CHAPTER 2
GPS estimates of microplate motions, northern Caribbean: Evidence for a Hispaniola
microplate and implications for earthquake hazard
B. Benford1, C. DeMets1, and E. Calais2
1Department of Geoscience
University of Wisconsin – Madison
1215 W. Dayton St.
Madison, Wisconsin 53706
2Department of Earth and Atmospheric Sciences
Purdue University
550 Stadium Mall Dr.
West Lafayette, Indiana 47906
In review, Geophysical Research Letters.
61 Abstract
We use elastic block modeling of 126 GPS site velocities from Jamaica, Hispaniola,
Puerto Rico, and other islands in the northern Caribbean to test for the existence of a
Hispaniola microplate and estimate angular velocities for the Gônave, Hispaniola, Puerto
Rico-Virgin Islands, and two smaller microplates relative to each other and the Caribbean
and North America plates. A model in which the Gônave microplate spans the whole plate
boundary between the Cayman spreading center and Mona Passage west of Puerto Rico is
rejected at high confidence level. The data instead require an independently moving
Hispaniola microplate between the Mona Passage and a likely diffuse boundary within or
offshore from western Hispaniola. Our updated angular velocities predict 6.8±1.0 mm yr-1 of
left-lateral slip along the seismically hazardous Enriquillo fault zone of southwest
Hispaniola, 9.8±2.0 mm yr-1 of slip along the Septentrional fault of northern Hispaniola, and
~14-15 mm yr-1 of left-lateral slip along the Oriente fault south of Cuba. Additionally, they
predict 5.7±1 mm yr-1 of fault-normal motion in the vicinity of the Enriquillo fault, faster
than previously estimated and possibly accommodated by folds and faults in the Enriquillo
fault borderlands. Along with previous estimates, our new estimate for Gônave-Caribbean
plate motion suggests that enough elastic strain accumulates to generate one to two Mw~7 earthquakes per century along the Enriquillo and nearby faults of southwest Hispaniola.
Key Words: Neotectonics; Hispaniola; Caribbean; Gônave microplate; Plate motions
62
1. Introduction
Following the January 12, 2010 Mw=7.0 Haiti earthquake (Calais et al. 2010; Hayes et al.
2010), along the southern boundary of the Gônave microplate, an international effort began to better understand the slip rates and hence seismic hazards of the numerous faults in the northern Caribbean region (e.g., Frankel et al. 2010; Prentice et al. 2010). These efforts depend critically on estimates of the present motions of microplates between the Caribbean and North America plates (Fig. 1), which in turn depend on our still-evolving understanding of the configuration and number of these microplates.
To date, the Gônave, Puerto Rico-Virgin Islands, and south Jamaica microplates (Fig. 1) have been defined from geologic, seismic, and geodetic observations (Mann et al. 1995,
2002; Jansma & Mattioli 2005; Benford et al. 2012). In addition, previous authors (Byrne et al. 1985; Jansma et al. 2000; Mann et al. 2002) invoke a Hispaniola microplate in their discussions of the regional tectonics, but do not treat the questions of whether the Hispaniola microplate is distinct from the larger Gônave microplate and if so, where its western boundary with the Gônave microplate is located. The answers to these questions are important because accurate estimates of fault slip rates on the seismically hazardous island of
Hispaniola depend critically on whether the island is part of an independent microplate or instead moves with the larger Gônave microplate.
Here, we apply elastic block modeling to an updated GPS velocity field spanning all of the Caribbean-North America plate boundary east of the Cayman spreading center (Fig. 1b) to re-examine the configuration of microplates used by previous authors and estimate angular velocities for all the microplates along the plate boundary (Jansma et al. 2000; Jansma &
63 Mattioli 2005; Manaker et al. 2008; Calais et al. 2010). In particular, we test rigorously for the existence of a distinct Hispaniola microplate and examine whether the eastern limit of the
Gônave microplate coincides with the Mona Passage between Hispaniola and Puerto Rico
(Figs 1 and 3; Manaker et al. 2008; Calais et al. 2010), with the topographically high and
seismically active Haiti fold-and-thrust belt of central and western Hispaniola (Mann et al.
1995; Pubellier et al. 2000; Mann et al. 2002) (Figs 1c and 2), or is located even farther west
in the Jamaica Passage (Fig. 3). The new set of angular velocities defined by our analysis
provides a useful basis for determining long-term fault slip rates and hence seismic hazard in
this earthquake-prone region.
2. Tectonic setting and microplate configuration
Along the Greater Antilles islands of Puerto Rico, Hispaniola, and Jamaica (Fig. 1), motion between the Caribbean and North America plates is dominated by obliquely convergent, left-lateral slip at rates of 19-20 mm yr-1 (Dixon et al. 1998; DeMets et al. 2000;
2007; Lopez et al. 2006; DeMets et al. 2010). From east to west, the Puerto Rico trench and
north Hispaniola and Oriente faults define the southern boundary of the North America plate
(Fig. 1), and the Muertos Trough and Enriquillo, Plantain Garden, and Walton faults define
the northern boundary of the Caribbean plate (Fig. 1). Between these two boundaries, four
distinct microplates have been identified. The Puerto Rico-Virgin Islands microplate
(hereafter abbreviated PRVI) defined by Byrne et al. (1985) and Masson & Scanlon (1991)
moves roughly westward at 2.6±2.0 mm yr-1 relative to the Caribbean plate interior (Jansma
et al. 2000; Jansma & Mattioli 2005). The Gônave microplate, first proposed by Rosencrantz
and Mann (1991), moves westward at a more rapid 6-8 mm yr-1 (DeMets & Wiggins-
64 Grandison 2007; Chapter 1), also relative to the Caribbean plate. The Hispaniola block defined by Byrne et al. (1985) and Manaker et al. (2008) occupies the eastern end of the
Gônave microplate and as described below is a key subject of this paper. Finally, two smaller microplates, the south Jamaica (Chapter 1) and north Hispaniola microplates
(Manaker et al. 2008), also appear to partition deformation along the plate boundary (Fig. 1) and are used for this analysis.
The forces responsible for moving these microplates are incompletely understood, but almost surely change along-strike given the irregular trace of the plate boundary and its well- described transition from oblique, steep-angle subduction along the Puerto Rico Trench to shear-dominated deformation in Hispaniola and locations farther west (Calais et al. 1992).
The prominent restraining bend where the Bahama platform collides obliquely with the north edge of Hispaniola (Mann et al. 1995, 2002) and a smaller restraining bend where the
Nicaragua Rise collides obliquely with southern Jamaica (Chapter 1) are both associated with distributed seismicity and deformation within the Gônave microplate and are likely to be responsible for some of the slip partitioning and microplate fragmentation characteristic of the plate boundary.
3. Data and Methodology
The 126 GPS velocities used for this analysis (Fig. 1b and Supplementary Information) consist of 30 velocities from Jamaica (Chapter 1), 63 from Hispaniola (Calais et al. 2010), 17 from the Puerto Rico-Virgin Islands microplate, 6 from smaller islands scattered along the
North America-Caribbean plate boundary, and 10 from other locations on the Caribbean plate. Processing of the raw GPS data from Hispaniola is described by Calais et al. (2010)
65 and employs ITRF05. Processing of the raw data from other locations is described by
Benford et al. (2012) and employs ITRF08. We used the Caribbean-ITRF05 angular velocity
from DeMets et al. (2010) and a Caribbean-ITRF08 angular velocity (Table 1) to transform
the respective sets of GPS velocities to a Caribbean plate reference frame.
All the GPS velocities we use are interseismic and hence include both a steady long-term component associated with the rotation of a corresponding microplate and an elastic component associated with one or more nearby active faults (Manaker et al. 2008). Modeling that allows for elastic, rotating blocks is thus required to separate and estimate the two effects
(McCaffrey 2002; Meade & Loveless 2009). We model the GPS velocity field using the
Blocks software of Meade & Loveless (2009), which treats the crust as an elastic homogeneous half-space consisting of rotating plates that are fully locked along the faults that define their boundaries. The output of each velocity field inversion includes one angular velocity per plate (unless the plate angular velocity is specified as a model constraint), the goodness-of-fit as measured by the summed, weighted least-squares misfit !2, long-term slip rates parallel and orthogonal to each plate boundary fault, and elastic deformation components at each GPS site.
Three assumptions are required about faults for our modeling: their locking depths, their dips, and their degree of frictional coupling. Here, we assign a uniform fault-locking depth of
15 km, but also evaluate model results for locking depths as shallow as 10 km and as deep as
20 km. For faults common between our own analysis and that of Manaker et al. (2008), we use the same fault dips as adopted by Manaker et al. For other faults, we assign dips of 90° for strike-slip faults, 60° for high-angle reverse faults, and 15° or 30° for thrust faults.
66 We assume complete and uniform interseismic coupling across all block boundary faults,
thereby maximizing the elastic deformation component everywhere in our model. This
approach differs modestly from that adopted by Manaker et al. (2008), who assign full
coupling to most faults in their model and estimate variations in coupling along the Puerto
Rico and Lesser Antilles trenches. Given the remoteness of these features from much of our
study area and uncertainties in those coupling estimates, variations in coupling along those
features are unlikely to significantly influence our model results and are ignored hereafter.
For the analysis below, we vary only the assumed location of the eastern boundary of the
Gônave microplate. The boundaries of the Gônave microplate at most other locations are
well defined, as are the boundaries of the Caribbean, North America, and PRVI plates. The
geometry of the southern boundary of the Gônave microplate in Jamaica is constrained by
GPS and geologic observations (Chapter 1). We use the MORVEL Caribbean-North
America angular velocity (73.9° S, 32.6° E, 0.190°/Myr; DeMets et al. 2010) to tie the North
America plate to the Blocks model. Estimates of fault slip rates and boundary geometries for
Hispaniola are insensitive to alternative assumed geometries for the Gônave microplate boundary in Jamaica.
Finally, a general limitation of our models is the requirement that plate boundaries be narrow, discrete features. For example, deformation may be distributed across a ~100-km- wide zone in central and western Hispaniola, where we and other authors postulate the existence of a plate boundary (Jansma et al. 2000; Jansma & Mattioli 2005; van Benthem &
Govers 2010). Similarly, deformation in southwest Haiti may be partitioned between the strike-slip Enriquillo fault and nearby structures (Calais et al. 2010).
67 4. Test for geometry and existence of the Hispaniola microplate
If the Gônave microplate extends from the Cayman spreading center in the west (Fig. 1) to the Mona Passage between eastern Hispaniola and Puerto Rico, it includes much of
Hispaniola and Jamaica, the two major landmasses on the largely submarine Gônave microplate. If this geometry were correct, then the GPS velocities for sites in areas of
Hispaniola and from the island of Jamaica should be consistent with rotation about a single angular velocity after accounting for the elastic deformation from locked plate boundary faults. Alternatively, motion between independently moving Hispaniola and Gônave microplates would be manifested as an inconsistency between velocities recorded by sites in
Jamaica and Hispaniola relative to the velocities predicted by a single angular velocity.
We test for an independent Hispaniola microplate comparing the fit of a five-microplate model that excludes a Hispaniola microplate to the fits of several six-microplate models in which we vary the assumed locations for the Gônave-Hispaniola plate boundary (locations indicated by the colored lines in Fig. 3a). For each assumed geometry, we invert all 126 GPS velocities (Fig. 3) using Blocks to estimate each microplate angular velocity and the elastic effects of the faults that bound the microplates. The goodness-of-fit for each assumed boundary location is given by chi-squared (!2) from the Blocks inversion (Fig. 3), where !2
2 is the weighted, summed least-squares misfit. Lower values of ! correspond to a better fit of the GPS velocities for a given boundary location. Angular velocities are estimated for the
Caribbean, Gônave, north Hispaniola, Puerto Rico-Virgin Islands, and south Jamaica microplates; models that also include a Hispaniola microplate have three additional adjustable parameters (one angular velocity). We use the Stein & Gordon (1984) F-ratio test to evaluate the improvements in fit of the more complex six-plate models relative to the five-
68 plate model; the F-ratio test is well suited for this analysis given its inherent insensitivity to
incompletely known data uncertainties.
Figure 3 shows the goodness-of-fit to all 252 GPS velocity components for all the models we tested. The worst-fitting model, with !2 = 423.3 (Fig. 3b), corresponds to the five-
plate model, which excludes the Hispaniola microplate. In this model, the Gônave microplate
extends east to the Mona Passage and thus includes all of Hispaniola and much of Jamaica.
This model significantly misfits GPS velocities in both western and eastern Hispaniola (Fig.
4a), indicating that the velocities from Jamaica and Hispaniola are fit poorly if forced onto
the same microplate.
The fits of all of the six-plate models we tested improve on that of the five-plate model
by ~20% or more (Fig. 3b), with the best fits for assumed boundaries that coincide with
either the Haiti fold-and-thrust belt (green boundary in Fig. 3a) or an assumed boundary
offshore from western Hispaniola (blue boundary in Fig. 3a). An assumed boundary in the
Jamaica Passage (purple boundary in Fig. 3a) fits the data worse than the two best-fitting six-
plate models and as described below is rejected at the 99% confidence level.
The respective least-squares misfits, !2 =300.1 and !2 =300.7, for the two best-fitting models, those with assumed boundaries that coincide with the Haiti fold-and-thrust belt or just offshore from western Hispaniola (labeled “Gônave Island” in Fig. 3a), correspond to
WRMS misfits of 0.9 mm yr-1 and 0.8 mm yr-1 in the north and east velocity components.
These misfits are only ~10% larger than the estimated velocity uncertainties. As shown by
the small and randomly oriented residual GPS velocities in Fig. 4b, the best-fitting models
eliminate the systematic misfits to the GPS velocities in western and eastern Hispaniola
observed for a model without a Hispaniola microplate (Fig. 4a).
69 We used the Stein & Gordon (1984) F-ratio test for an additional plate to assess whether
the best overall fit (!2 =300.1) for the assumed Haiti fold-and-thrust belt plate boundary
improves significantly on that for the model that excludes the Hispaniola microplate. For the
three additional degrees of freedom corresponding to the additional Hispaniola plate angular
velocity, F = 32.0. The probability that random errors in the GPS velocities could spuriously
give rise to an F-value this high is only 3 parts in 1017. The GPS velocities thus strongly support the existence of a Hispaniola microplate, a result not previously reported in the literature.
Based on a F-ratio comparison of the squared misfits for all the assumed boundary locations relative to that for the best-fitting location, boundaries in the Jamaica Passage or within or east of central Hispaniola fit the data significantly worse than the two best-fit models (Fig. 3b). We conclude that the regional GPS velocities strongly justify the addition of a Hispaniola microplate, bounded to the east by the Mona Passage and to the west by the
Haiti fold-and-thrust belt or structures in far western Hispaniola (e.g., the Matheux Neiba fault zone shown in Fig. 2), including offshore faults. Although our block modeling requires plate boundaries to be discrete, the seismicity and topography in western Hispaniola are distributed and may define a diffuse plate boundary that spans the faults listed above (shown by the hachured region in Fig. 5).
We tested the degree to which the 30 GPS site velocities from Jamaica influence the above results by repeating all the above inversions while excluding those velocities (thereby eliminating nearly all the GPS sites assumed to be on the Gônave microplate). The improvement in fit for an additional Hispaniola microplate gives F = 6.8, which is significant at the 99% confidence level (or 2 parts in 104). The kinematic evidence for a distinct
70 Hispaniola microplate is thus still significant if we exclude data from Jamaica that help
determine the Gônave microplate angular velocity.
5. Microplate angular velocities and predicted fault slip rates
Table 1 lists the best-fitting angular velocities for the microplates in our study area relative to each other and relative to ITRF08 based on the best-fitting Gônave-Hispaniola microplate boundary that corresponds to the Haiti fold-and-thrust belt (Fig. 3a). The formal angular velocity covariances estimated from Blocks (Table 1) are likely to be too small since they are based on the simplifying assumptions that the fault locking depths, fault dips, and microplate geometries are well known, that full coupling occurs across all faults, and that the microplate boundaries are discrete. To better understand some of the modeling tradeoffs and hence estimate more realistic uncertainties, we derived models using a variety of alternative microplate geometries, different subsets of the GPS velocity field, different geologically plausible constraints for the slip type along particular faults, and fault locking depths as shallow as 10 km and as deep as 20 km. The uncertainties stated below, which are based on these sensitivity tests, are typically larger than the formal uncertainties by a factor of two to three and approximate the 95% confidence limits.
Septentrional fault: The complexly deforming island of Hispaniola includes the seismically hazardous Septentrional and Enriquillo faults (Fig. 5), as well as seismically active faults in offshore areas around the island (Figs 1 and 3a). Left-lateral slip of 9.8±2 mm yr-1 is predicted for the Septentrional fault (Figs. 4b and 5), near the middle of the 6-12 mm yr-1 geologically estimated slip rate for the fault (Prentice et al. 2003) and modestly
slower than but consistent with the 12±2 mm yr-1 rate estimated by Calais et al. (2010) from
71 GPS velocity block modeling with DEFNODE software (McCaffrey 2002). Sensitivity tests
in which we vary the factors described in the preceding paragraph suggest a 13 mm yr-1
maximum slip rate for this fault.
Enriquillo fault: The Gônave-Caribbean angular velocity (Table 1) predicts 8-9.5 mm yr-1
of obliquely convergent slip along the Enriquillo fault of western Haiti (Fig. 5). For
comparison, Calais et al. (2010) estimate motion of 5.1-5.8 mm yr-1 from their inversion of
GPS velocities solely from Hispaniola (Fig. 5). Our revised estimate is thus faster than and more oblique to the fault than that of Calais et al. At a central point along the N85°E-striking fault, our newly predicted Gônave-Caribbean plate velocity resolves into 6.8±1 mm yr-1 of
fault-parallel left-lateral slip and 5.7±1 mm yr-1 of boundary-normal convergence (Fig. 4b).
For comparison, the velocity estimated by Calais et al. (2010) resolves into 5±1 mm yr-1 of
fault-parallel and 2±1 mm yr-1 of fault-normal motion. Our respective fault-parallel rate estimates are therefore consistent within their estimated uncertainties. In contrast, our model predicts fault-normal convergence that is a factor of three faster (5.7 mm yr-1 versus 2 mm yr-
1) than predicted by Calais et al.’s model.
Most of the difference in the predictions of the two models is caused by the different
Gônave microplate geometries used in our two studies and the new GPS velocities from
northern Jamaica that we use to constrain motion of the Gônave microplate (Fig. 1 and
Chapter 1). The Gônave-Caribbean plate angular velocity for our five-plate model, which
simulates the microplate geometries assumed by Calais et al. (2010), predicts a fault-normal
convergent component along the Enriquillo fault of only 3.1±1 mm yr-1, close to that (2 mm
yr-1) predicted by Calais et al. However, as described above, this sub-optimal microplate
72 geometry increases the misfit by more than 30% (Fig. 3b) and is rejected at high confidence level.
The more rapid convergence predicted by our model for the Gônave-Caribbean plate boundary in southwest Hispaniola poses a conundrum given that geologic and seismic evidence indicate that the primary plate boundary structure, the Enriquillo fault, is a strike- slip fault. We thus postulate that the oblique convergence predicted by our model is partitioned onto the Enriquillo fault and structures north and possibly south of the fault.
Mann et al. (1995) describes underwater folds and reverse faults north of Haiti’s southwest peninsula that could play a role in the postulated partitioning. The 2010 Haiti earthquake, which occurred on a buried fault close to, but north of the Enriquillo fault, accommodated 2.6 meters of slip parallel to and 1.8 meters of slip orthogonal to the Enriquillo fault (Calais et al.
2010). The nearly 40% reverse dip-slip component for this earthquake and predominance of thrust-faulting aftershocks following the earthquake (Nettles & Hjörleifsdóttir 2010; Mercier de Lépinay et al. 2011) are both strong evidence for a significant component of boundary- normal convergence, as predicted by our model.
We tested the robustness of our model velocity estimates for the Enriquillo fault by re- inverting the GPS velocity field while imposing the following constraints on the outcomes of our velocity field inversions: (1) pure strike-slip motion was required along the Enriquillo fault, (2) pure strike-slip motion was required along the Oriente fault west of Cuba (Fig. 1),
(3) pure strike-slip motion was enforced across both the Oriente and Walton faults west of
Jamaica (Fig. 1), and (4) various subsets of GPS velocities on the Gônave microplate were excluded. Enforcing the first constraint increases the model misfit ~30% and is rejected at high confidence level. For cases (2) through (4), the estimated slip rate component parallel to
73 the Enriquillo fault differs by no more than 0.6 mm yr-1 from our best-fitting estimate of 6.8
mm yr-1. Cases 2 and 3 reduce the estimated convergent slip rate component orthogonal to
the Enriquillo fault to 4.5-4.8 mm yr-1, slower than the 5.7 mm yr-1 best estimate, but still
faster than the 2±1 mm yr-1 estimated by Calais et al. (2010). Case 4 variously changes the
boundary normal rate component to 5.3-6.2 mm yr-1. We conclude that the rate of strike-slip
motion in western Hispaniola is well determined and that the faster-than-expected convergent
component of Gônave-Caribbean plate motion in western Haiti is a robust feature of our data
and modeling.
Hispaniola-Gônave microplate motion: Slow convergence in western Hispaniola: The
Hispaniola-Gônave rotation pole is located in western Hispaniola and remains nearly fixed
whether we adopt the boundary just west of Hispaniola or a boundary that coincides with the
Haiti fold-and-thrust belt. Surprisingly, the best-fitting angular velocity (Table 1), located in
western Hispaniola at 19.1°N, 72.2°W, predicts that almost no deformation occurs across the
Haiti fold-and-thrust belt (Figs 4bc and 5). For example, at a central location in the fold-and-
thrust belt, it predicts NW-directed dextral shear of only 0.6±1.5 mm yr-1. If we instead
assume the plate boundary is located just west of the island, the resulting Hispaniola-Gônave
angular velocity (19.1°N, 72.4°W, 1.30 ° Myr-1) predicts oblique convergence at rates of 2.3-
3.6 mm yr-1 across the NW- to WNW-oriented reverse faults in western Haiti and offshore
(Fig. 5).
Although the kinematic evidence for significant motion between distinct Gônave and
Hispaniola microplates may appear to be at odds with the prediction of slow or no motion
across their shared boundary, the primary kinematic evidence for these two distinct plates is
the inconsistency of the well-determined GPS site velocities in Jamaica with the GPS site
74 velocities from eastern Hispaniola (as illustrated in Fig. 4a). The apparently slow motion
across their boundary may be caused by strong mechanical coupling across the fault(s) that
separate the two microplates. More and better-determined GPS site velocities from western
Hispaniola are needed to better understand the kinematics of this likely diffuse plate
boundary.
Oriente fault: The new angular velocities predict that Caribbean-North America plate motion is partitioned into 14.2-14.5 (±1) mm yr-1 of left-lateral slip along the Oriente fault
south and west of Cuba and 4.1-7.3 (±1) mm yr-1 of left-lateral slip along faults that define the southern edge of the Gônave microplate (Figs. 1, 4c, and 5). The latter agrees with a 3-7 mm yr-1 geologic estimate based on post-9 Myr fault offsets in eastern Jamaica and the
Jamaica Passage (Natural Disaster Research 1999), and supersedes a somewhat faster 8 mm yr-1 estimate of the minimum slip rate based on an earlier velocity field from Jamaica
(DeMets & Wiggins-Grandison 2007).
Cayman spreading center: The best-fitting Gônave-North America angular velocity
(Table 1) predicts a 12.6±0.6 mm yr-1 seafloor-spreading rate across the Cayman spreading center (Fig. 4c). Long-term opening rates based on Cayman spreading center magnetic anomalies (Macdonald & Holcombe 1978; Leroy et al. 2000) and on seafloor depth and seafloor age profiles (Rosencrantz et al. 1988) range between 12 and 20 mm yr-1. DeMets &
Wiggins-Grandison (2007) estimate a 6-11 mm yr-1 short-term opening rate based on
geodetic results from Jamaica. The modestly faster rate predicted by our new model is more
consistent with the estimated long-term opening rate.
75 6. Discussion: Seismic hazard and driving forces
Seismic hazard: Our work implies greater earthquake hazard along the Enriquillo fault
of southwestern Haiti than previous studies. The 6.8±1 mm yr-1 of fault-parallel left-lateral slip predicted by our best model agrees with prior estimates within uncertainties (Dixon et al.
1998; Calais et al. 2002; Mann et al. 2002; Manaker et al. 2008; Calais et al. 2010).
However, the 5.7±1 mm yr-1 of boundary-normal convergence predicted by our model is a factor of 3 greater than estimated by Calais et al. (2010). Given that the surprisingly rapid estimated fault-normal component of motion is robust with respect to a variety of sensitivity tests and that the 2010 Haiti earthquake included a reverse slip component equal to nearly
40% of the seismic moment (Calais et al. 2010), partitioning of the oblique plate motion onto separate sets of faults may occur.
If the Enriquillo fault accommodates all of the boundary-parallel component of motion, which appears likely given the evidence it has ruptured multiple times over the past few centuries (Prentice et al. 2010), our best model predicts that the fault accrues a 0.68±0.1 meter slip deficit per century. Simple calculations show that it takes ~600 years for a 30-km- long by 10 km-deep fault, approximately the rupture dimensions of the 2010 Haiti earthquake
(Calais et al. 2010), to accumulate enough seismic moment at the predicted slip rate to produce a Mw=7.0 earthquake. For the 300-km-long Enriquillo fault, the accumulated seismic
moment deficit is therefore large enough to cause one Mw~7 strike-slip earthquake every 60
years on a ~30-km-long segment somewhere along the fault. For comparison, the 5-6 mm yr-
1 fault-parallel slip rate predicted by Calais et al. (2010) predicts a modestly longer 75-90 yr
recurrence interval for a Mw~7 strike-slip earthquake somewhere along the fault.
76 Unlike the kinematic model described by Calais et al. (2010), which predicts a relatively small 2±1 mm yr-1 component of motion orthogonal to the Enriquillo fault, our model predicts that a fault-normal slip deficit accrues at 0.57±0.1 meters per century. Assuming 60° dips for assumed reverse-slip faults that extend along the 300-km-long boundary and repeating the calculations above for assumed 30-km-long locked segments, we find that enough sufficient seismic moment accumulates to generate one Mw~7 reverse-slip earthquake every 60-70 years.
Our kinematic model therefore suggests that as many as two Mw~7 earthquakes per century may occur somewhere along the Gônave-Caribbean plate boundary in southern Haiti.
The Calais et al. (2010) kinematic model predicts less frequent earthquakes, possibly only one Mw~7 earthquake per century. The two models may thus bracket estimates of the seismic hazard, the former constituting a worst-case scenario and the latter a best-case for the recurrence interval of assumed Mw~7 earthquakes.
Driving forces: Our kinematic model offers intriguing new clues about the forces that influence microplate motions along the shear-dominated Caribbean-North America plate boundary. Slow CCW rotation of the PRVI microplate about a pole immediately south of the microplate (Table 1), giving rise to 1-1.5 mm yr-1 of dominantly westward PRVI motion relative to the Caribbean plate, may be driven by oblique left-lateral shear across the plate boundary. Our kinematic results strongly support those reported earlier by Jansma et al.
(2000) and Jansma & Mattioli (2005), who report GPS evidence for 2.5±2 mm yr-1 of dominantly westward PRVI motion.
Masson & Scanlon (1991) propose that the PRVI microplate rotates slowly CCW and translates slowly to the west along the Puerto Rico Trench, consistent with their
77 interpretation of long-range sidescan sonar data and seismic reflection profiles near Puerto
Rico. Our kinematic results strongly support their conclusions. An earlier proposal by Jany et
al. (1987) for eastward tectonic “escape” of the PRVI block conflicts with GPS
measurements later reported by Jansma et al. (2000) and Jansma & Mattioli (2005), who find westward rather than eastward motion of the PRVI microplate. Our model results also demonstrate slow westward motion of PRVI, as found by Jansma et al. (2000) and Jansma &
Mattioli (2005).
More rapid CCW rotation of the Hispaniola microplate is presumably driven by oblique collision of the Bahama platform with north-central Hispaniola (Mann et al. 1995, 2002).
The effects of the collision are profound and include rifting in the Mona Passage and a distinct change from westward motion in western Puerto Rico to SW motion in eastern
Hispaniola (Figs 1b and 4c). From forward modeling of the forces and hence torques that act on the Caribbean plate and comparisons to the observed regional stresses, vertical-axis rotations, and earthquake slip directions, van Benthem & Govers (2010) conclude that microplates in the northeast Caribbean should rotate CCW. This agrees with our Caribbean- fixed angular velocities of the Hispaniola and PRVI microplates (Table 1), which predict slow CCW rotations for both microplates.
Our model also includes the first estimate of Gônave plate angular velocities. Contrary to the CCW rotations of the Hispaniola and PRVI microplates, the Gônave microplate rotates slowly CW (Fig. 4c). The estimated CW rotation agrees with the slow CW rotation predicted by van Benthem & Govers (2010) for the Gônave microplate. Its opposite-sense rotation may be caused by two factors, namely, the southward push on the eastern end of the Gônave microplate from rapid CCW rotation of the Hispaniola microplate (Fig. 4c) and oblique
78 convergence of the Nicaragua Rise with the southern boundary of the Gônave microplate,
which may pin and hence create a pivot point for the microplate or push it northward (Fig. 1).
The Gônave microplate is elongate subparallel to the direction of plate motion and is thus
unlikely to rotate rapidly in response to shear imposed along its northern and southern
boundaries (Lamb 1994).
7. Conclusion
GPS site velocities from the two largest islands potentially on the Gônave microplate,
Hispaniola and Jamaica, are fit poorly by a single angular velocity when the site velocities are corrected for interseismic elastic deformation from faults locked along the microplate boundaries. In particular, the velocities of sites in eastern Hispaniola are misfit systematically and significantly when inverted simultaneously with well-defined velocities from sites on parts of Jamaica and western Hispaniola that lie on the Gônave microplate. A discrete or diffuse boundary between the Gônave and Hispaniola microplates thus lies between eastern
Hispaniola and Jamaica. Inversions of the 126 velocities from GPS sites along the Caribbean-
North America plate boundary for a range of assumed Gônave-Hispaniola boundary locations indicate that the boundary is most likely located in central or western Haiti, coinciding with the Haiti fold-and-thrust belt or offshore faults west of Hispaniola.
An inversion of the regional GPS site velocities using the newly defined Hispaniola microplate, as well as the Gônave, north Hispaniola, south Jamaica, and Puerto Rico-Virgin
Islands microplates gives new angular velocities for these microplates relative to each other and relative to the Caribbean and North America plates and ITRF2008. These angular velocities and their uncertainties provide a useful new basis for studies of the regional-scale
79 deformation, as well as seismic hazard for specific plate boundary faults. Our new estimate
and an estimate previously published by Calais et al. (2010) for Gônave-Caribbean plate
motion suggest that enough elastic strain accumulates to generate one to two Mw~7 earthquakes per century along the Enriquillo and nearby faults of southwest Hispaniola.
8. Acknowledgements
This project was funded by National Science Foundation grant EAR-0609578. Some of this material is based on data provided by the UNAVCO Facility with support from the
National Science Foundation (NSF) and National Aeronautics and Space Administration
(NASA) under NSF Cooperative Agreement No. EAR-0735156. Figures were produced using Generic Mapping Tools software (Wessel & Smith 1991).
9. References
Argus, D.F., 2007. Defining the translational velocity of the reference frame of Earth,
Geophys. J. Int., 169, 830-838, doi: 10.1111j.1365-246X.2007.03344.x.
Byrne, D.B., Suarez, G., and McCann, W.R., 1985. Muertos Trough subduction-microplate
tectonics in the northern Caribbean?, Nature, 317, 420-421.
Calais, E., Béthoux, N., & Mercier de Lépinay, B., 1992. From transcurrent faulting to
frontal subduction: a seismotectonic study of the northern Caribbean plate boundary
from Cuba to Puerto Rico: Tectonics, 11, 114-123.
Calais, E., Mazabraud, Y., Mercier de Lépinay, B., Mann, P., Mattioli, G., & Jansma, P.,
2002. Oblique strain partitioning and fault slip rates in the northeastern Caribbean from
GPS measurements, Geophys. Res. Lett., 20, 1856-1859, doi: 10.1029/2002GL015397.
80 Calais, E., Freed, A., Mattioli, G., Amelung, F., Jónsson, S., Jansma, P., Hong, S.-H., Dixon,
T., Prépetit, C., & Momplaisir, R., 2010, Transpressional rupture of an unmapped fault
during the 2010 Haiti earthquake, Nature Geoscience, doi: 10.1038/NGEO992.
DeMets, C., & Wiggins-Grandison, M., 2007. Deformation of Jamaica and motion of the
Gônave microplate from GPS and seismic data, Geophys. J. Int., 168, 362-378, doi:
10.1111/j.1365-246X.2006.03236.x.
DeMets, C., Jansma, P.E., Mattioli, G.S., Dixon, T.H., Farina, F., Bilham, R., Calais, E., and
Mann, P., 2000. GPS geodetic constraints on Caribbean-North America plate motion:
Geophys. Res. Lett., 27, 437-440, doi: 10.1029/1999GL005436.
DeMets, C., Mattioli, G., Jansma, P., Rogers, R.D., Tenorio, C., & Turner, H.L., 2007.
Present motion and deformation of the Caribbean plate: Constraints from new GPS
geodetic measurements from Honduras and Nicaragua, in Geologic and Tectonic
Development of the Caribbean Plate in Northern-Central America, Geol. Soc. Am.
Spec. Paper, ed. Mann, P., 428, 21-36, doi: 10.1130/2007.2428(02), The Geological
Society of America, Boulder.
DeMets, C., Gordon, R.G., & Argus, D.F., 2010. Geologically current plate motions,
Geophys. J. Int., 181, 1-80, doi: 10.1111/j.1365256X.2009.04491.x.
Dixon, T.H., Farina, F., DeMets, C., Jansma, P., Mann, P., and Calais, E., 1998. Relative
motion between the Caribbean and North American plates and related boundary zone
deformation from a decade of GPS observations, J. Geophys. Res., 103, 15,157-15182,
doi: 10.1029/97JB03575.
81 Frankel, A., Harmsen, S., Mueller, C., Calais, E., and Haase, J., 2010. Documentation for
initial seismic hazard maps for Haiti, Open-File Rep. – U. S. Geol. Surv. 2010-1067, 20
p.
Hayes, G.P., Briggs, R.W., Sladen, A., Fielding, E.J., Prentice, C., Hudnut, K., Mann, P.,
Taylor, F.W., Crone, A.J., Gold, R., Ito, T., and Simons, M., 2010. Complex rupture
during the 12 January 2010 Haiti earthquake, Nature Geoscience, doi:
10.1038/NGEO977.
Jansma, P.E., & Mattioli, G.S., 2005. GPS results from Puerto Rico and the Virgin Islands:
Constraints on tectonic setting and rates of active faulting, in Active tectonics and
seismic hazards of Puerto Rico, the Virgin Islands, and offshore areas, Geol. Soc. Am.
Spec. Paper, ed. Mann, P., 385, 13-30 doi: 10.1130/2007.2428(02), The Geological
Society of America, Boulder.
Jansma, P.E., Mattioli, G.S., Lopez, A., DeMets, C., Dixon, T.H., Mann, P. and Calais E.,
2000. Neotectonics of Puerto Rico and the Virgin Islands, northeastern Caribbean, from
GPS geodesy, Tectonics, 19(6), 1021-1037.
Jany, I., Mauffret, A., Bouysse, P., Mascle, A., Mercier de Lépinay, B., Renard, V., &
Stephan, J.F., 1987. Relevé bathymétrique Seabeam et tectonique en décrochement au
sud des Iles Vierges [Nord-Est Caraibes], C.R. Acad. Sci. Paris, 304(Ser. II), 527-532.
Lamb, S. 1994. A model for tectonic rotations about a vertical axis, J. Geophys. Res., 99,
4457-4483, doi: 10.1029/93JB02574.
Leroy, S., Mauffret, A., Patriat, P., & Mercier de Lepinay, B., 2000. An alternative
interpretation of the Cayman trough evolution from a reidentification of magnetic
anomalies, Geophys. J. Int., 141, 539-557.
82 Lopez, A.M., Stein, S., Dixon, T., Sella, G., Calais, E., Jansma, P., Weber, J., and LaFemina,
P., 2006. Is there a northern Lesser Antilles forearc block?, Geophys. Res. Lett., 33,
L07313, doi: 10.1029/2005GL025293.
Macdonald, K.C. & Holcombe, T.L., 1978. Inversion of magnetic anomalies and sea-floor
spreading in the Cayman Trough, Earth planet. Sci. Lett., 40, 407-414.
Manaker, D.M, Calais, E., Freed, A.M., Ali, S.T., Przybylski, P., Mattioli, G., Jansma, P.,
Prepetit, C., & de Chabalier, J.B., 2008. Interseismic plate coupling and strain
partitioning in the northeastern Caribbean, Geophys. J. Int., 174, 889-903, doi:
10.1111/j.1365-246X.2008.03819.x.
Mann, P., Taylor, F.W., Edwards, R.L, & Ku, T., 1995. Actively evolving microplate
formation by oblique collision and sideways motion along strike-slip faults: an example
from the northeastern Caribbean plate margin, Tectonophysics, 246, 1-69.
Mann, P., Calais, E., Ruegg, J.-C., DeMets, C., Jansma, P., and Mattioli, G.S., 2002. Oblique
collision in the northeastern Caribbean from GPS measurements and geological
observations, Tectonics, 37, doi: 0.1029/2001TC001304.
Masson, D.G. and Scanlon, K.M., 1991. The neotectonic setting of Puerto Rico, Geol. Soc.
Am. Bull., 103, 144-154, doi: 10.1130/0016-7606(1991)103<0144:TNSOPR>2.3.CO;2.
McCaffrey, R., 2002. Crustal block rotations and plate coupling, in Plate boundary zones,
Geodynamics Series, eds Stein, S. & Freymueller, J.T., 30, 101-122, American
Geophysical Union.
McCann, W.R., 2006. Estimating the threat of tsunamigenics earthquakes and earthquake-
induced landslide tsunami in the Caribbean, in, Caribbean tsunami hazard, Proceedings
83 of the NSF Caribbean Tsunami Workshop, March 30-31, 2004 eds Mercado-Irizarry, A
& Liu, P., 43-65, World Scientific Publishing Co., Singapore.
Meade, B.J. & Loveless, J.P., 2009. Block modeling with connected fault-network
geometries and a linear elastic coupling estimator in spherical coordinates, Bull. seism.
Soc. Am., 99, 3124-3139, doi: 10.17850120090088.
Mercier de Lépinay, B., Deschamps, A., Klingelhoefer, F., Mazabraud, Y., Delouis, B.,
Clouard, V., Hello, Y., Crozon, J., Marcaillou, B., Graindorge, D., Vallée, M., Perrot, J.,
Bouin, M.-P., Saurel, J.-M., Charvis, P., & St-Louis, M., 2011. The 2010 Haiti
earthquake: A complex fault pattern constrained by seismologic and tectonic
observations, Geophys. Res. Lett., 38, L22305, doi:10.1029/2011GL049799.
Natural Disaster Research, Inc., Earthquake Unit, Mines & Geology Division, 1999.
Kingston Metropolitan Area Seismic Hazard Assessment Final Report, Caribbean
Disaster Mitigation Project, Organization of American States,
www.oas.org/cdmp/document/kma/seismic/kma1.htm, 82 pp.
Nettles, M. & Hjörleifsdóttir, V., 2010. Earthquake source parameters for the 2010 January
Haiti main shock and aftershock sequence, Geophys. J. Int., 183, 375-380, doi:
10.1111/j.1365-246X.2010.04732.x.
Prentice, C.S., Mann, P., Peña, L.R. and Burr, G., 2003. Slip rate and earthquake recurrence
along the central Septentrional fault, North American-Caribbean plate boundary,
Dominican Republic, J. Geophys. Res., 108(B3), 2149, doi:10.1029/2001JB000442.
Prentice, C.S., Mann, P., Crone, A.J., Gold, R.D., Hudnut, K.W., Briggs, R.W., Koehler,
R.D., and Jean, P., 2010. Seismic hazard of the Enriquillo-Plantain Garden fault in Haiti
inferred from palaeoseismology, Nature Geoscience, doi: 10.1038/NGEO991.
84 Pubellier, M., Mauffret, A., Leroy, S., Vila, J.M., and Amilcar, H., 2000. Plate boundary
readjustment in oblique convergence: Example of the Neogene of Hispaniola, Greater
Antilles: Tectonics, 19, 630-648.
Rosencrantz, E. & Mann, P., 1991. SeaMARC II mapping of transform faults in the Cayman
Trough, Caribbean Sea, Geology, 19, 690-693, doi: 10.1130/0091-
7613(1991)019<0690:SIMOTF>2.3.CO;2.
Rosencrantz, E., Ross, M., & Sclater, J., 1988. Age and spreading history of the Cayman
Trough as determined from depth, heat flow, and magnetic anomalies, J. geophys. Res.,
93, 2141-2157.
Sandwell, D.T. & Smith, W.H.F., 1997. Marine gravity anomaly from Geosat and ERS 1
altimetry, J. geophys. Res., 102, 10039-10054.
Stein, S. & Gordon, R.G. 1984. Statistical tests of additional plate boundaries from plate
motion inversions, Earth planet. Sci. Lett., 69, 401-412. van Benthem, S. & Govers, R., 2010. The Caribbean plate: Pulled, pushed, or dragged?, J.
geophys. Res., 115, B10409, doi: 10.1029/2009JB006950.
Wessel, P. & Smith, W.H.F., 1991. Free software helps map and display data, EOS, Trans.
Am. Geophys. Un., 72, 441-446.
85
10. Figure captions
Figure 1. A - Tectonic setting of the northern Caribbean. Bold black arrows in both panels
show MORVEL estimate of North America plate motion in mm yr-1 relative to the Caribbean plate (DeMets et al. 2010). CSC – Cayman spreading center. PR – Puerto Rico. 2-min seafloor bathymetry and land topography from Sandwell & Smith (1997). B – GPS site velocities relative to Caribbean plate, with 1", 2-D error ellipses. Velocities from Hispaniola are taken from Calais et al. (2010). Velocities are color-coded based on plate. Scale is shown in upper right corner. Black lines mark plate boundaries and major faults used for the analysis. All plates included in the analysis are labeled. PRVI microplate – Puerto Rico-
Virgin Islands microplate. C – Red circles show earthquakes from 1964-2010 overlain on faults, bathymetry, and topography from A.
Figure 2. – Major faults of Hispaniola overlain on 2-min seafloor bathymetry and land topography from Sandwell & Smith (1997). Abbreviations: Haiti FTB – Haiti fold-and-thrust belt, Matheux Neiba – Matheux Neiba fault zone, NA plate – North America plate, and North
Hisp. Fault – North Hispaniola fault.
Figure 3. A – Trial locations for the eastern boundary of the Gônave microplate shown in different colors (including the Mona Passage in black). Orange circles show 1964-2010 earthquakes. B – Least-squares misfits (!2) for trial boundaries shown in A. Boundary
locations with misfits above the green line are excluded at the 99% confidence level.
86 Figure 4. A – Velocity misfits and modeled slip rates for a model that excluded an
independent Hispaniola microplate. Slip rates are specified as boundary-parallel and
boundary-normal components. Velocities are in Gônave microplate reference frame and are
color-coded per microplate. Uncertainty ellipses are 2-D, 1". Maximum strike-slip and
convergent-slip uncertainties are 1.8 mm yr-1 and 2.1 mm yr-1, respectively. B – Velocity misfits and modeled slip rates for a model with an independent Hispaniola microplate.
Velocities are in a Hispaniola microplate reference frame. Maximum strike-slip and convergent-slip uncertainties are 1.1 mm yr-1 and 1.8 mm yr-1, respectively. Fault name abbreviations are: EFZ – Enriquillo fault zone, NHF – North Hispaniola fault, and SF –
Septentrional fault zone. C – GPS velocities relative to Caribbean plate and corrected for elastic effects from our best-fitting Blocks model. Slip rates for Hispaniola are shown in B.
Velocities are in a Gônave microplate reference frame. Maximum strike-slip and convergent- slip uncertainties are 1.9 mm yr-1 for both. Scale is shown in upper right corner in all panels.
Figure 5. Comparison of plate velocities estimated herein (green arrows) and by Calais et al.
(2010) (red arrows). Velocities shown in green are predicted from the best-fitting microplate angular velocities in Table 1 and are for an assumed Gônave-Hispaniola boundary that coincides with the Haiti fold-and-thrust belt. Blue arrows west of Hispaniola indicate motion of the Gônave microplate relative to the Hispaniola microplate for an assumed plate boundary just west of Hispaniola (indicated by blue line in Fig. 3a). Plate velocities show motion of microplate located south of the plate boundary with respect to that north of the boundary. Long-term fault slip rates derived from GPS inversions are given in mm yr-1.
Estimated historical rupture areas are derived from archives (McCann 2006). 1701, 1860, and
87 1953 are the dates of smaller magnitude, poorly located events. Vertical strike-slip events are shown as lines; dip-slip events are shown as projected surface areas. Hachured area may approximate the diffuse Gônave-Hispaniola plate boundary. Modified from Fig. 1 of Calais et al. (2010).
Figure S1. Elastic (A) and rotational (B-D) components of GPS velocities based on best- fitting model. Scale is in upper right corner of all panels. A – Best-fitting elastic component of GPS velocities estimated with Blocks. These velocities show the elastic effects of fault locking, independent of the rotational component (B-D). B - GPS velocities relative to
Gônave microplate corrected for elastic effects shown in A. Velocities thus illustrate rotational component of the site motions, independent of the elastic effects associated with locked faults. Scale is in upper right corner. C – Same as B but fixed North America plate. D
– Same as B but fixed Hispaniola plate.
88
89
90
91
92
93
94
95
96
97
98
99
CHAPTER 3
Fault interaction of reactivated faults within a restraining bend: Neotectonic
deformation of southwest Jamaica
B. Benford, B. Tikoff, and C. DeMets
Department of Geoscience
University of Wisconsin – Madison
1215 W. Dayton St.
Madison, Wisconsin 53706, USA
To be submitted, Lithosphere. 100 Abstract
Jamaica is located on a restraining bend that occurs on the E-trending, left-lateral plate boundary between the Gônave microplate and Caribbean plate. Deformation in southern
Jamaica occurs on two reactivated and simultaneously active fault sets: NNW-striking reverse faults and E-striking strike-slip faults. Movement on NNW-striking reverse faults form fault-propagation folds that are expressed topographically as the Don Figuerero, Santa
Cruz, and Brisco Mountains. The NNW-trending ranges (and faults) of southern Jamaica terminate against the E-W oriented strike-slip faults. The two dominant E-striking, left- lateral strike-slip faults are the South Coast fault zone in the south and the central Jamaica fault system (Cavaliers fault, Rio Minho-Crawle River fault, and the Siloah fault system) in the central part of the island.
We propose that the restraining bend is the result of a reactivated transfer zone in southern Jamaica. The two fault systems, inherited from Paleogene deformation, are both reactivated to accommodate current deformation. The NNW-striking reverse faults accommodate E-W shortening. The E-striking strike-slip faults accommodate both the plate motion and the differential motion of the fault blocks bounded by the NNW-striking reverse faults. This geometry results in strike-slip faults that are characterized by topographic highs and lows along strike, as a result of vertical displacement on the NNW-striking reverse faults.
101 1. Introduction
The reactivation of inherited structures is commonly observed in both ancient and neotectonic settings. Pre-existing faults are particularly subject to reactivation during later deformation events, since they are weaker than surrounding intact rock (e.g., Etheridge,
1986). Particular patterns of reactivated faults have been described in the literature (e.g.,
Chandler et al., 1989; Gomez et al., 2000; De Paola et al., 2006; Quintana et al., 2006; Vos et
al., 2006). For example, normal faults are commonly reactivated as reverse faults during
contraction, resulting in inversion tectonics (e.g., Withjack et al., 1995; Beauchamp et al.,
1996, 1999; Gomez et al., 2000; Sagir, 2001; Hill et al., 2004; Konstantinovskaya et al.,
2007). Strike-slip faults may be reactivated as strike-slip, reverse, or normal faults
(Kennedy, 1946; Holgate, 1969; Stewart et al., 1999; Santos et al., 2000; Cobbold et al.,
2001).
Much less work has concentrated on if, and how, different fault sets can be
reactivated during regional deformation. Jamaica provides us the opportunity to observe how
pre-existing fault sets operate to accommodate deformation associated with a plate-margin
restraining bend (Fig. 1). Studies of neotectonic deformation in Jamaica have documented
two major active fault sets: an E-striking set and a NNW-striking set (Horsfield, 1974;
Wadge and Dixon, 1984; Draper, 1998) (Fig. 1b). The E-striking fault set developed during
the Cretaceous (Mitchell, 2003) and the NNW set initiated in the Paleogene (Green, 1977;
Eva and McFarlane, 1985; Mann and Burke, 1990). Both fault sets have been reactivated in
the current tectonic setting (Horsfield, 1974; Burke et al., 1980; Mann et al., 1985; Draper,
1998, 2008). 102 The goal of this paper to is document ongoing deformation in southern Jamaica by
presenting data for the NNW-striking faults and then for the E-striking faults. Recent
geodetic modeling indicates that deformation associated with the Caribbean-Gônave plate
boundary occurs primarily in southern Jamaica (Chapter 1). Our work highlights the interaction of the two reactivated fault sets and how they interact with one another and accommodate deformation associated with the current restraining bend. The geometries of both fault sets are constrained by geological mapping, regional topography, direct observation of faults, and focal mechanisms; the NNW-striking reverse faults are further constrained by gravity transects and borehole data. These data, combined with available geodetic measurements, allow us to show that the E-striking strike-slip faults bound a
“panel” of NNW-striking reverse faults. The reverse faults accommodate contraction associated with the restraining bend, commonly through the development of fault- propagation folds. Consequently, differential vertical and horizontal motion occurs along the strike-slip faults as a result of movement along the NNW-striking reverse faults. Overall, our work indicates that a network of pre-existing faults controls the neotectonic deformation in
Jamaica.
2. Tectonic and geologic overview
Jamaica is the uplifted, northernmost extent of the Northern Nicaragua Rise (Fig. 1a).
The Nicaragua Rise is a Cretaceous submarine volcanic plateau overlain with 5-7 km of
Tertiary carbonates (Arden, 1975; Duncan et al., 1999; Mutti et al., 2005). The Nicaragua
Rise extends for over 700 km NE-SW from Nicaragua to Jamaica (Lewis and Draper, 1990;
Robinson, 1994). Currently, the Cayman Spreading Center and the Hess Escarpment bound 103 the Nicaragua Rise to the north and south, respectively. Jamaica occurs partly on oceanic
crust and partly on the stretched continental Chortis Block, originally part of the North
America plate (Pindell, 1993; Cunningham, 1998; Pindell and Kennan, 2001), both of which
were intruded by a Cretaceous volcanic arc. The Caribbean plate surrounding the Nicaragua
Rise near Jamaica consists of atypically thick (10-15 km) oceanic crust (Officer et al., 1959;
Ewing et al., 1960; Edgar et al., 1971; Houtz and Ludwig, 1977) that lacks identifiable
magnetic anomaly patterns (Duncan and Hargraves, 1984).
Jamaica occurs along the Gônave microplate-Caribbean plate boundary (Fig. 1; Mann et
al., 1985, 2007). The relative motion between the Caribbean plate and the Gônave
microplate is 7±1 mm yr-1 of left-lateral slip (DeMets and Wiggins-Grandison, 2007;
Chapter 1). The Gônave microplate extends 100-150 km from north to south and ~1100 km from east to west. The extent of the Gônave microplate is well constrained in the west by the
Cayman spreading center, in the north by the Oriente transform fault, and in the south along the Enriquillo fault in Haiti and the Plantain Garden fault and the Walton fault, east and west of Jamaica, respectively (Rosencrantz and Mann, 1991).
Detailed documentation of Jamaica’s geology is limited by tropical weathering, abundant vegetation, and the fact that nearly two-thirds of the island is covered by a single lithology
(the Eocene-Middle Miocene White Limestone Group) (Kashfi, 1983). Older rocks are preserved in 27 Cretaceous structural inliers (Robinson, 1994; Fig. 2a). The geology of these areas, however, preserves a record of four major episodes of deformation: 1) Island arc development from the Early Cretaceous to early Cenozoic; 2) Rifting of the arc in the
Eocene; 3) Subsidence in the Oligocene-Miocene; and 4) Faulting and folding from the 104 Middle Miocene to the Holocene (Meyerhoff and Krieg, 1977; Mitchell, 2003; Draper,
2008). Evidence for these events is given in Appendix 1.
3. Neotectonic deformation
Jamaica is interpreted as a restraining bend where slip is transferred from the Plantain
Garden fault in the southeast to the submarine Walton fault in the northwest (Fig. 1)
(Horsfield, 1974; Mann et al., 1985; Leroy et al., 1996). The island contains a series of left- lateral E-striking strike-slip faults. These include, from S to N, the South Coast fault zone, the Plantain Garden fault, the central Jamaica fault system (Siloah fault system, Rio Minho –
Crawle River fault, Cavaliers fault), and the Duanvale fault. There is no clear indication of major strike-slip offset on any of these faults. Mitchell (2003) documents 8-10 km of left- lateral slip on the Rio Minho – Crawle River fault. Additionally, the Plantain Garden fault offsets rocks of similar ages and lithologies flanking the fault by only 10-12 km (Mann et al.,
1985). Estimates of the fault offset east of the island range from 30-45 km, based on the offset of the eastern Jamaica shelf, to ~60 km, based on the width of the Morant Trough
(Natural Disaster Research et al., 1999). Regardless, there is no consensus on a single strike- slip zone that acts as a plate boundary.
In addition to the strike-slip faults, there are NNW-striking faults throughout Jamaica.
The major faults adjacent to the Blue Mountains – the Blue Mountain and Yallahs faults – strike NNW and are associated with the high topography. Other NNW-striking faults are more widely spaced and create isolated and distinctive mountain ranges in SW Jamaica, such as the Don Figuerero and Santa Cruz Mountains (Wright, 1975; Fig. 1b). These faults are interpreted to have formed during Paleogene extension and have been reactivated as reverse 105 faults in the present tectonic regime (Horsfield, 1974; Green, 1977; Burke et al., 1980; Eva
and McFarlane, 1985; Mann et al., 1985; Mann and Burke, 1990; Draper, 1998, 2008). The
best evidence for reactivation is the inverted rifts of the Wagwater Belt (adjacent to Blue
Mountains) and Montpelier-Newmarket Zone (northwestern Jamaica) (Mann et al., 1985).
The Wagwater belt, in particular, contains abundant clastic sedimentary rocks interlayered
with Eocene volcanic rocks, and is interpreted as a Paleogene rift. Documentation of NNW
reverse faults away from the Blue Mountains, however, is very limited.
Seismologic data confirm strike-slip movement on the E-striking faults and reverse
motion on the NNW faults. Left-lateral motion, reverse motion, or oblique-slip (left-lateral,
reverse motion) dominate the focal mechanisms across the island (Fig. 1c; Wiggins-
Grandison and Atakan, 2005; DeMets and Wiggins-Grandison, 2007; Chapter 1). Overall, seismicity is widespread throughout Jamaica, although it is particularly prevalent in the Blue
Mountains. Focal mechanisms indicate dominantly reverse motion on its western side along the Blue Mountain and Yallahs faults. Left-lateral focal mechanisms are documented on the
E-striking Plantain Garden fault in eastern Jamaica and central Jamaica fault system (CJFS;
Wiggins-Grandison and Atakan, 2005; DeMets and Wiggins-Grandison, 2007). Reverse motion focal mechanisms have relatively deep (20-30 km) foci (Wiggins-Grandison, 2004), and occur through the island, including southwest Jamaica.
4. Geodesy
Geodetic results also indicate bulk left-lateral motion across Jamaica (DeMets and
Wiggins-Grandison, 2007; Chapter 1). Relative to the Caribbean plate, GPS sites in northern
Jamaica move 6.0±0.5 mm yr-1 to the WSW, constituting a lower bound on the motion of the 106 Gônave microplate across its southern boundary in Jamaica (Fig. 1a). The westward movement of Jamaica relative to the Caribbean plate – requiring left-lateral motion – is apparent. A 2.6±0.6 mm yr-1 southward component of motion is remarkably consistent everywhere on the island and on two small islands south of the main island of Jamaica.
Deformation within Jamaica is better displayed if the site velocities are referenced to a well-established continuous site centrally located on the island (PIKE; Fig. 3; Chapter 1).
This reference station highlights deformation in Jamaica and elastic strain accumulation of locked faults (e.g., CJFS). Relative to this continuous GPS site, sites located north of the
CJFS and Plantain Garden fault are either stationary or move no faster than ~1-2 mm yr-1 to the SE (Fig. 3). Sites in westernmost Jamaica and north of the CJFS move with the same velocity (within uncertainties) as sites on the east coast of Jamaica (Fig. 3). Little or no east- to-west shortening occurs across the northern half of the island. In contrast, sites located south of the CJFS move to the ESE at rates that increase from 1 to 6 mm yr-1 southward from the CJFS. The gradient in the Jamaica GPS velocity field along a N-S transect of the island strongly indicates that one or more active plate boundary faults are located on the island.
Further, there is an east-to-west velocity gradient, south of the CJFS, with GPS sites in the west moving less rapidly eastward when the elastic effects of the Plantain Garden fault are removed (Chapter 1).
The geodetic results, corroborated by the block modeling of Chapter 1, suggest that:
1) The CJFS and/or the South Coast fault zone may accommodate a large percentage of the transcurrent offset in Jamaica; and 2) Velocity gradients in southwest Jamaica suggest that present-day deformation occurs in this region. 107 5. Southwest Jamaica
We define Southwest Jamaica as the area west of the Vere Plain and south of the
CJFS. The area contains a series of NNW-trending mountains. The two largest are the Don
Figuerero and Santa Cruz Mountains; smaller ones include Brisco Mountain and Hill Top
Hill (Fig. 2b). These ranges are typically topographically asymmetric with the west slope dipping steeper than the east slope (Figs 4 and 5). Further, the ranges end abruptly in the north and south, against the CJFS and the South Coast fault zone (SCFZ), respectively.
Because the SCFZ is spatially coincident with the south coast of the island in the west, there are areas of high relief against the south coast (Lover’s Leap) where the mountain range
(Don Figuerero) intersects the fault. Likewise, there are low relief areas along the south coast (Alligator Pond), where a NNW-oriented valley intersects the coast. The Vere Plain is covered by Quaternary alluvial deposits, with low relief mountains to both the east
(Brazilletto) and the west (Don Figuerero).
SW Jamaica was investigated using a combination of (1) topographic analysis to constrain the geomorphology, (2) geological observation to understand the structure, and (3) gravity transects and subsequent modeling to understand the subsurface geometry of bedding and faults. The geological observations are largely based on the map of Wright (1975), corroborated and added to by our observations. However, Wright’s (1975) material is not available in any geological publication, and thus we include it here. We summarize the results for the Santa Cruz Mountains, Don Figuerero Mountains, and Vere Plain area below. 108 5.1. Santa Cruz Mountains
5.1.1. Topography
One of the most prevalent NNW-oriented ranges is the Santa Cruz Mountains (Figs
2b and 6). These mountains are ~35 km long and end in the south and north where they are
crosscut by the SCFZ and CJFS, respectively (Fig. 5). They are topographically asymmetric
in that their west flank is steeper than their east flank. Detailed topographic maps are shown in Figures 4 and 5, which quantify the slope of the ground surface throughout Jamaica. The steepest slopes occur in the Blue Mountains of eastern Jamaica but also in southwestern
Jamaica, on the western and southern slopes of the ranges, particularly the Santa Cruz
Mountains (Fig. 4a). The steep western slopes occur over short distances in contrast to the gentler eastern slopes (Fig. 4b). This technique utilizes the generic mapping tools (GMT) of
Wessel and Smith (1991), using a method developed by W. Haneberg (pers. comm., 2010).
Just north of the town of Lacovia, the Black River cuts across the Santa Cruz
Mountains, creating a significant valley across the range. North of this valley, the Santa Cruz
Mountains change in character by widening and the eastern slope becoming steeper than the western slope (Fig. 4).
5.1.2. Geology
The geology of the region around the Santa Cruz Mountains is described in Wright’s
(1975) Ph.D. thesis; his map and one of the cross-sections are reproduced with modifications
(e.g., fault dips) as Figure 6. Although numerous faults occur within the Santa Cruz
Mountains, the White Limestone Group Miocene Newport Formation is the only unit exposed. Further, bedding orientation parallels the topography. On the west slope of the 109 Santa Cruz Mountains, the limestone beds dip gently to the west, and on the east slope, the
limestone beds dip more gently to the east (Figs 6 and 7a). Bedding measurements from
Wright (1975) and recent field data throughout the Santa Cruz Mountains indicate a NNW- striking, upright, slightly asymmetric fold with an axial plane oriented 334, 87°E and the fold
hinge plunging 1° ! 328 (Fig. 7a).
Although a single lithology occurs at the surface, a borehole along the ridge of the Santa
Cruz Mountains provides information about the subsurface geology. The Santa Cruz
Mountains borehole (Meyerhoff and Krieg, 1977) occurs at an elevation of 786 m and
17°55.5’ N, 77°41.0’ W. The drillcore contains the Newport Formation (to 1000 m),
Oligocene Brown’s Town Formation of the White Limestone Group (1000 m – 1571 m),
Eocene Somerset and Troy Formations of the White Limestone Group (1571 m – 1768 m),
the Eocene Chapelton Formation of the Yellow Limestone Group (1768 m – 1966 m),
Cretaceous volcanic rocks (1966 m – 2529 m), and greenschists (2529 m – 2662 m; the base
of borehole). The Oligocene Brown’s Town Formation is 120-390 m thick where observed at the surface, but it is 571 m thick in the borehole. This apparent doubling in thickness suggests that the stratigraphy is duplicated by a reverse fault in the well.
5.1.3. Gravity measurement
The subsurface geometry of the faults below the western Santa Cruz Mountains was previously unknown. Wright (1975) mapped two vertical, NNW-striking faults with east- side-up motion. We conducted gravity surveys along two transects with 200-700 m spacing, oriented perpendicular to the mountain ranges (Fig. 8) to constrain the subsurface geometry and orientation of the Santa Cruz faults. These transects were designed to also investigate 110 the subsurface geometry of the Don Figuerero Mountains and the Spur Tree faults; these
results are discussed in the next section. A 9-km ENE-oriented transect with ten stations
crosses the Don Figuerero Mountains, and six kilometers north of this transect, a 27-km long
transect with 34 stations crosses both mountain ranges (Fig. 8).
Gravity measurements were taken using a LaCoste and Romberg G-meter (G-19) in the
winter of 2008. This gravimeter has an accuracy of 0.01 mGal and a long-term drift of less
than 0.5 mGal/mo. Differential real-time kinematic Global Positioning System (GPS),
running in short station survey mode, allowed centimeter-level elevation control for each
station. A local base station was designated for the survey and measured at the start and end
of each day to account for instrument drift. Gravity corrections (earth tides, instrument drift,
latitude, elevation, and terrain) were made using QC Tool software following the methods of
Hays (1976). Topographic and terrain corrections are based on both local topographic
measurements and high-resolution SRTM data.
To constrain our gravity models, we include data from geologic maps, known geologic
structures, and the Santa Cruz Mountains borehole (Meyerhoff and Krieg, 1977). We use a
density of 2.5 g/cm3 for all limestone formations and 2.7 g/cm3 for the volcanic rocks, following Wadge et al. (1983). Additionally, we assign a density of 2.9 g/cm3 for the
greenschist.
5.1.4. Gravity results and interpretation
At a latitude of 17°56’N, the Bouguer gravity anomaly across the Santa Cruz
Mountains is ~6-7 mGal (Fig. 8). The anomaly increases steadily by ~0.5 mGal per km from
west to east except at the western flank of the mountains. On the western flank, there is a
sharper increase of ~2.5 mGal per km. 111 We carried out an iterative 2D forward model of the gravity anomaly, along transect
B-B’ that intersects the Santa Cruz Mountains borehole. We used the software Grav2D
(freeware from the University of Liverpool), which is based on the method of Talwani et al.
(1959). This simple 2D forward model program calculates the gravitational effect of a body along a profile, and assumes that the body extends infinitely in a horizontal direction perpendicular to the profile direction. The results of the model are shown in Figure 8.
The model constrains several aspects of the subsurface geology of the Santa Cruz
Mountains, including: 1) The location of the reverse faults below the Santa Cruz Mountains;
2) The dip of these faults; and 3) The approximate magnitude of offset. Two major faults occur in the Santa Cruz Mountains, one directly below the western flank and the other just west of the ridge. Both faults dip steeply eastward at 77° and 80º. The amount of throw is based on offsets in the limestone-volcanic rock contact and the volcanic rock-schist contact.
The eastern fault dips ~77º E and has ~2000 m of throw and ~500 m of heave. The second, smaller fault occurs just west of the ridge; it is modeled to have ~200 m throw and ~30 m heave.
In summary, the Santa Cruz Mountains are cored by two reverse faults, both of which dip steeply eastward. This interpretation is consistent with the Bouguer gravity anomaly increasing from west to east across the mountains, and with the steep gradient on the western limb of the fault. The surface geometry, however, is one of a slightly asymmetric upright fold, with vergence toward the west. Reverse slip on the fault at depth results in the unfaulted rocks at the surface folding. Based on the orientation of the faults coring this fold and the folding of the units at the surface, the Santa Cruz Mountains have the geometry of a fault- propagation fold. 112 5.2. Don Figuerero Mountains
5.2.1. Topography
The Don Figuerero Mountains, ~18 km east of the Santa Cruz Mountains, are similar
topographically to the Santa Cruz Mountains (Figs 2b and 6). They are ~35 km long, end in the south where the east-west-striking SCFZ crosscuts them (Fig. 5), and are topographically
asymmetric. The west and east faces dip as much as ~18° and ~12°, respectively (Figs 4 and
5). These mountains are about twice as wide as the Santa Cruz Mountains and are a tilted
plateau (dips ~1° E) with the ridge at the western edge of the plateau (Figs 4 and 5). The
eastern slope of the mountains defines the western boundary of the Vere Plain. Where the
SCFZ crosscuts the Don Figuerero Mountains, the southern slopes reach ~12°. The
mountains narrow in width to the NNW and their strike changes to a more east-west strike at
their northernmost extent, where they are crosscut by the CJFS.
5.2.2. Geology
The western Don Figuerero Mountains are a km-scale anticline, herein named the
Don Figuerero anticline. The Don Figuerero anticline mostly exposes the White Limestone
Group of the Eocene Troy Formation in the north and Miocene Newport Formation in the
south. Beds dip gently (<5°) east on the east flank of the mountains, and the gentle dip of the
topography mostly reflects dip slopes in resistant limestone beds. In contrast, the dips on the
western flank are up to 56° to the west. Bedding measurements by Wright (1975) throughout
the western Don Figuerero Mountains indicate an asymmetric, NNW-striking, gently
plunging fold with a more steeply dipping west limb. The axial plane of the fold is 349,
85°E with the fold hinge oriented 5° ! 170 (Fig. 7b). 113 The Don Figuerero Mountains also expose other stratigraphic units. The Oligocene
Brown’s Town and Eocene Somerset formations are exposed on the east flank of the
mountain, in the north. These units are crosscut by the Spur Tree fault system. The Eocene
Chapelton Formation of the Yellow Limestone and Cretaceous volcaniclastics are only
exposed along the western ridge of the mountains (exposed on transect A-A’), along the trace of the fold hinge (Fig. 6).
Mapped faults in the Don Figuerero Mountains include three faults in the west, which are all considered splays of the Spur Tree fault. One fault cuts across the foot of the western slope and the other two faults occur part way up the western slope of the mountains (Fig. 6;
Wright, 1975). Wright (1975) mapped these faults as vertical with east-side-up motion for
the westernmost and easternmost faults (Fig. 6). The middle fault is mapped as having a
minor amount of west-side-up motion.
Other faults occur within the Don Figuerero Mountains. The Porus fault, named after
the nearby town of Porus, and referred to as the Porus graben by Burke et al. (1980), occurs
in the easternmost Don Figuerero Mountains and defines the eastern slope (Figs 1b and 2b).
The Porus fault is mapped as having west-side-up motion. One major unnamed fault, herein
referred to as the Carpenter fault after the Carpenter Mountains along its strike, occurs in the
interior of the Don Figuerero Mountains (Fig. 5). This fault varies in strike between NW and
WNW, cuts across the Don Figuerero Mountains from their southeastern corner, and bends
into parallelism with the Spur Tree fault in the west (Fig. 1b). This fault is one of the few
faults in Jamaica with a WNW strike, has a small isolated range (Carpenter Mountains) of
limestone outcrops, and has NE-side up motion (Figs 4a and 5). 114 5.2.3. Gravity measurement
The same gravity transect used for the Santa Cruz Mountains was continued across the
Don Figuerero Mountains and the Spur Tree faults. The same equipment, forward modeling gravity program, and density values were used for the gravity measurements and modeling of the Don Figuerero Mountains.
5.2.4. Gravity results and interpretation
Across the Don Figuerero Mountains, gravity surveys reveal an ~8 mGal Bouguer gravity anomaly in the southern transect and ~12 mGal anomaly in the northern transect (Fig.
8). Similar to the Santa Cruz Mountains, in both transects the anomaly steadily increases from west to east, except at the western flank of the Don Figuerero Mountains. In this location, the gradient in the anomaly increases from ~1 mGal per km to ~4 mGal per km. A
2D forward model of the gravity anomaly using Grav2D was carried out along transect B-B’, which intersects the borehole in the Santa Cruz Mountains.
The forward gravity model for the Don Figuerero Mountains constrains the location, dip, and offset of the main strand of the Spur Tree fault. Based on the forward model, the
Spur Tree fault dips ~73º E, with ~1400 m of throw and ~400 m of heave (Fig. 8d). The offsets on the other two splays of the Spur Tree fault to the east cannot be constrained, as the gravity signature associated with these faults is overprinted by the large signature of the main strand of the Spur Tree fault and the associated fold.
In summary, the Don Figuerero Mountains share many characteristics with the adjacent
Santa Cruz Mountains. The increase in the Bouguer gravity anomaly from west to east across the ranges and valley in between them suggests that most faults in this region are east- dipping reverse faults. The gravity anomaly is a result of the reverse faults bringing dense 115 volcanic rocks and schists closer to the surface. With progression to the east across the
reverse faults, these denser rocks are closer to the surface. Offset of these dense rocks in the
east where they are closer to the surface results in a larger gravity anomaly than in the west
where they are deeper, thus the Don Figuerero Mountains have smaller offsets on the faults
but a larger gravity anomaly than the Santa Cruz Mountains.
Finally, the bedding and outcrop patterns of the Don Figuerero Mountains define a
westward-verging anticline. The clear correlation between the fault location, the fault
dipping steeply east, and the westward-verging anticline implies that the uplift of the western
Don Figuerero Mountains is also the result of a fault-propagation fold. Reactivation of the
east-dipping fault at depth has resulted in folding of the units exposed at the surface.
5.3. Vere Plain
5.3.1. Topography
The Vere Plain, in south-central Jamaica, is a broad, flat area, which is ~10-15 km
wide east-west. Uplifted mountains surround its western (Don Figuerero Mountains),
northern (Mocho Mountains), and eastern (Brazilletto Mountains) limits. The Vere Plain slopes ~0.1º over ~25 km, from ~50 m in the north to sea level in the south (Fig. 4).
Three major topographic features occur in the Vere Plain along the projected trace of the E-striking SCFZ: the southern edge of the Brazilletto Mountains (130 m elevation in the south), Kemp’s Hill (95 m in elevation), and Round Hill (330 m in elevation; Figs 2 and 5).
Round Hill and the Brazilletto Mountains define the western and eastern extents of the southern Jamaica peninsula. Similar to the Santa Cruz and Don Figuerero Mountains, the southern slope of the Brazilletto Mountains dips ~12º, however the elevation change is only 116 ~130 m. The southern edge of the Vere plain is either the Caribbean ocean or an E-W oriented ridge, known as Portland Ridge (Fig. 4).
5.3.2. Geology
The subsurface geometry of the Vere Plain is constrained by well data. Up to 200 m of
Quaternary clastic sediments overlie the Miocene limestone bedrock (Versey and Prescott,
1958). The deepest area of alluvium is oriented north-south, and is located just west of
Kemp’s Hill with the thickest alluvium in the south. This elongate area has been interpreted as a buried valley (Versey and Precott, 1958; Robinson, 2004). The sediment deposition is a result of the migration of the Rio Minho River alluvial fan across the area (Robinson, 2004).
The Rio Minho River drains a major portion of central Jamaica, where the rocks are primarily Cretaceous volcanic and siliciclastic sedimentary rocks.
Meyerhoff and Krieg (1977) generated island-wide isopach maps of the White and
Yellow Limestone Groups based on seven boreholes and field mapping. In the Vere Plain, the White Limestone is about ten times as thick as the Yellow Limestone. The maps indicate that the White Limestone Group increases in thickness from ~600 m in the north to ~1500 m just north of the trace of the SCFZ (figure 35 of Meyerhoff and Krieg, 1977). South of the trace of the SCFZ, isopach contours are closely spaced and parallel the fault. The thickness of the White Limestone Group increases from 1800 m to 2700 m over ~5 km, moving southward. In the area of high topography in southernmost Jamaica – known as Portland
Ridge - the White Limestone decreases to ~1800 m.
Minor NNW-striking faults have been mapped to the east of Round Hill, and to the west of both Kemp’s Hill and the Brazilletto Mountains (Horsfield, 1974). The northwest side of 117 Kemp’s Hill contains a series of NNW-striking faults with well-developed subhorizontal
slickenfibers in the Miocene carbonates (Fig. 7d).
5.3.3. Gravity methods
To better understand the SCFZ and its relation to the isolated hills of the Vere Plain, a
gravity survey consisting of 328 stations, covering an area of 500 km2 was performed using a
Lacoste and Romberg G-meter (G-19) in the winter of 2008. This survey updated and extended the 54-station survey performed by Wadge and others (1983), which did not include
Kemp’s Hill and Round Hill. We use the same densities as Wadge et al. (1983) - 2.0 g/cm3 for the Quaternary alluvium, 2.5 g/cm3 for the Miocene White Limestone, and 2.7 g/cm3 for the Cretaceous volcanic rocks. We carried out 2D iterative forward gravity models of E-W and N-S transects across the Vere Plain (Fig. 9a) using the Grav2D software.
5.3.4. Gravity results and interpretation
Prominent negative Bouguer anomalies oriented both N-S and E-W occur in the Vere
Plain (Fig. 10). The N-S anomaly has the largest magnitude (~14 mGal) at the latitude of
Kemp’s Hill and Round Hill and decreases in magnitude to the north. The E-W negative gravity anomaly occurs just south of the projected trace of the SCFZ (Fig. 10). The magnitude of this anomaly ranges between 3 and 11 mGal, with the greatest anomaly differences relative to Kemp’s Hill, the Brazilletto Mountains, and Round Hill. The gravity highs associated with Kemp’s Hill, the Brazilletto Mountains, and Round Hill continue to the north, however, they terminate abruptly in the south at the SCFZ.
In order to test the accuracy of the Meyerhoff and Krieg (1977) isopach maps of the
White and Yellow Limestone Groups in the Vere Plain, we generated an east-west transect 118 across the Vere Plain, north of Kemp’s Hill. Since our gravity modeling cannot distinguish between Yellow and White Limestone because of their similar density, we group these units together in our models. In this location, we have sufficient well data for the depth of the
Quaternary alluvium-White Limestone Group contact (Fig. 9b), and so, if the isopach maps of the limestone are accurate, then we should be able to accurately model the Bouguer gravity anomaly. However, the estimated thicknesses of the limestone units based on the isopach maps do not generate a large enough anomaly. For this reason, in all of our gravity models, we use the well data as constraints and the isopach maps as a secondary guide since they are not as precise.
South of the SCFZ, well data are sparse and so we have few constraints on the thickness of the alluvium (Fig. 9b). Since both the alluvium and limestone thicknesses have to be estimated, models we present for the subsurface of the southern Vere Plain are nonunique (Fig. 9b,c). Using 2D transects (Fig. 9a), we extend the structure contour map of
Versey and Prescott (1958) for the alluvium-limestone contact south of the SCFZ (Fig. 9b) and generate a structure contour map of the base of the limestone for the Vere Plain (Fig. 9c).
Transect A-A’ (shown in Figures 9a and 10) crosses four faults. The two westernmost faults dip steeply west. The fault that defines the eastern edge of the Brazilletto
Mountains is shown as vertical because the gravity data were not sufficient to determine dip.
In the north-south transects B-B’ and C-C’ (shown in Figures 9a and 10), the SCZF has an apparent north-side-up motion. Since the SCFZ is primarily a strike-slip fault and the volcanic rock-limestone and limestone-alluvium contacts were not originally planar and horizontal, it is not possibly to quantify vertical displacement across these faults. A second E- 119 striking fault just south of Round Hill appears to have even more vertical displacement than
the SCFZ.
Kemp’s Hill is not a simple isolated hill, but rather the southern, emergent tip of an
uplifted block (Figs 9 and 10). The elongate buried valley to the west of Kemp’s Hill, visible
using the extension of the Versey and Prescott map, parallels the block and is fault bounded.
Based on the Bouguer gravity map, the gravity transects, and the structure contour maps of
the base of the alluvium and of the limestone (Figs 9 and 10), we propose that Kemp’s Hill is
the product of a buried or blind reverse fault that strikes N, dips ~77º E, and terminates at the
SCFZ (Fig. 9). The buried valley to the west of Kemp’s Hill is a down-dropped block
between the Kemp’s Hill fault and the west-dipping fault just to the west.
The combination of the structure contour maps of Meyerhoff and Krieg (1977; their
figures 31 and 35) and the gravity data allow us to put constraints on the bedrock structure of
the southern Vere Plain (Fig. 9). The base of the alluvium is about a hundred meters lower and the base of the limestone is about a thousand meters lower in the south than north of the
SCFZ. This pattern could result from either vertical displacement on the SCFZ or significant strike-slip motion. However, an additional constraint can be obtained by considering the buried valley of Versey and Prescott (1958). This depression apparently continues south of the SCFZ, as noted by the gravity data. The data indicate that there are depressions in both the limestone-alluvium and limestone-volcanics contacts (Fig. 9c), suggesting that this feature had to have formed originally prior to or during limestone deposition. If this depression was originally part of the same river valley located north of the SCFZ, it indicates that strike-slip displacement on the SCFZ is minimal whereas vertical motion is significant. 120 Finally, a positive gravity anomaly exists at Portland Ridge in southernmost Jamaica
(Fig. 10). Wadge et al. (1983) assert that a fault just to the north of Portland Ridge has 1.2 km of vertical displacement. However, it is not clear how Portland Ridge and its associated faults fit into the other fault systems of Jamaica.
5.4. Other NNW-striking faults
The Brompton fault is the one exception to all NNW faults ending against the E- striking faults. The Brompton fault is crosscut by the central Jamaica fault system in the north, but this fault dies out before reaching the SCFZ in the south. The fault and associated topography end near the town of Black River, where the river by the same name empties into the Caribbean. The Black River cuts across the prominent Santa Cruz Mountains lowering the topography to near sea level. The river has deposited alluvium (Fig. 6) along its length.
South of the town of Black River, the Pondside fault extends to the SCFZ (Fig. 1 and
Appendix 2). Therefore, it is possible that the Pondside and Brompton faults are the same fault (Figs 1b and 11a), and the rapid erosion rates and Quaternary deposits of the Black
River may be hiding a portion of the fault. If the Black River has maintained this location since these faults were reactivated, it is possible that it was able to continually erode the growing topography associated with reverse motion on the NNW fault, especially if the fault produces ~1 mm/yr of vertical motion. Alternatively, an E-striking left-lateral fault could occur between the Pondside and Brompton faults. A future gravity survey of the alluvium near the town of Black River could provide evidence for existence of this portion of the fault. 121 5.5. E-striking, left-lateral strike-slip fault systems
There are two dominant E-striking, left-lateral strike-slip fault systems in southwestern Jamaica: the South Coast fault system and the central Jamaica fault system.
Studying the E-striking faults is more challenging than studying the NNW faults. The E- striking faults are commonly expressed as linear topographic lows, and there is poor to no exposure of the fault surfaces themselves. The most apparent evidence for these faults is their expression in the topography: NNW ridges terminate when they encounter the E- striking faults. There is evidence for these features in the subsurface, such as in the case of the Vere Plain, in which the NNW ridges terminate against the SCFZ. Below, we discuss the two E-striking fault systems in southwest Jamaica.
5.5.1. South Coast fault zone (SCFZ)
The SCFZ parallels the coast in southwest Jamaica from Great Pedro Bluff, the southwesternmost tip of Jamaica, to Round Hill. It then extends across the Vere Plain
(Horsfield, 1974) (Figs 2 and 5). The submarine extent of the SCFZ, both west of Great
Pedro Bluff and east of the Vere Plain, is not well constrained. From Great Pedro Buff to
Round Hill, cliffs are up to several hundred meters high (e.g., Lover’s Leap). East of Round
Hill, no cliffs are present and the coastline extends south of the projected trace of the SCFZ
(Fig. 2b). The SCFZ truncates all of the N- and NNW-striking faults along its trace, observable in both the island topography and the subsurface data (e.g., gravity, wells) from the Vere Plain.
Wadge et al. (1983) state that the SCFZ is primarily a strike-slip fault. However, their gravity model across the eastern Vere Plain (corroborated by our study) indicates up to
2.5 km of vertical separation. No focal mechanisms exist for the SCFZ. The only direct 122 evidence for left-lateral motion on the SCFZ is the slickenlines at Kemp’s Hill (Fig. 7d). We
interpret Kemp’s Hills to be on the N side of the SCFZ, and thus the faults on Kemp’s Hill
are not part of the SCFZ. However, their parallelism to the main fault zone is strongly
suggestive that they are subsidiary structures. The faults strike EW with kinematic indicators consistent with left-lateral slip.
5.5.2. Central Jamaica fault system (CJFS)
Central Jamaica consists of a series of E-striking faults including, from east to west, the Cavaliers fault, the Rio Minho-Crawle River fault, and the Siloah fault system. These faults appear to be segments of a continuous feature, which for simplicity, we refer to as the central Jamaica fault system (CJFS; Figs 1b and 5). The CJFS marks the southern margin of the Paleogene carbonate platform (Wright, 1975) where ~1370 m of shallow-water Newport
Formation of the White Limestone Group was deposited during the Miocene (Wright, 1975)
(Fig. 2b). The CJFS is expressed as a topographic depression, although local lows and highs
occur in close proximity along the CJFS (Fig. 5).
The Siloah fault system, the westernmost segment of the CJFS, is characterized by a
series of structures (e.g., faults, folds, foliations, deformation bands). Planar structures strike
between 285° and 305°. The folds plunge gently, vary between open and tight, and generally
strike NW or NNW. Some fold hinges are oriented 11°!140 and 24°!286 (Figs 7e,f).
Several small-scale faults in this area strike between 285° and 305° and dip 45°-90° NE.
Striations on one fault surface indicate either a left-lateral normal fault or a right-lateral
reverse fault. 123 Along the Siloah fault system and north of Brisco Mountain and the Brompton fault, a
pervasive foliation striking ~290° occurs in the Eocene-Oligocene Bonny Gate Formation.
Additionally, numerous fractures and deformation bands parallel a fault oriented 290, 87° S with slickenlines pitching 33° from the west. In the hanging wall, the rocks have a foliation oriented 306, 61° SW that cuts subhorizontal bedding. In contrast, no foliation occurs in the footwall and bedding dips ~30° SE. Based on the fault orientation, foliation, and bedding,
this fault appears to be a left-lateral fault with a reverse component.
The CJFS shows an interesting relation between topography and kinematics from
focal mechanisms. In areas where the CJFS forms a topographic low, focal mechanisms are
dominantly left-lateral. In contrast, in places where a topographic high is associated with the
CJFS (e.g., termination of the Spur Tree, Santa Cruz, and Porus faults), focal mechanisms
have a left-lateral component but primarily indicate top-to-the-south reverse motion (Fig. 1c).
6. Discussion
6.1. Reactivation and interaction of fault sets
The two major fault sets of Jamaica are known to be reactivated fault systems. The
NNW faults have been described as inversion structures from the Paleogene (Horsfield,
1974), whereas the E-striking faults are assumed to have formed in the Cretaceous and
Paleogene. No study, however, documents how the two fault sets interact.
Based on the datasets presented above, the NNW faults are primarily reverse faults with possibly a minor component of left-lateral motion (e.g., focal mechanisms, striations at
Kemp’s Hill). The major NNW-striking reverse faults from east to west are the Blue
Mountain fault, blind faults under Dallas Mountain and Long Mountain anticlines (Draper, 124 2008), and the Kemp’s Hill, Porus, Spur Tree, Santa Cruz fault, Pondside, and Brompton
faults. Aside from the Porus fault, and possibly the northern extent of the Santa Cruz
Mountains, the NNW faults dip east and are either blind or occur at the western base of the
associated ridge.
The Pondside and Brompton faults may mark the western boundary of the restraining
bend. However, GPS velocities from DeMets and Wiggins-Grandison (2007) and Chapter 1 indicate that westernmost Jamaica still has motion relative to the Caribbean plate, indicating that it is likely that at least one other fault occurs offshore. The other boundaries of the restraining bend are likely the Blue Mountain fault in the east and the SCFZ and CJFS in the south and north, respectively (Fig. 11).
We interpret the NNW ridges as fault-propagation folds. In these ranges, Eocene normal faults were reactivated as reverse faults and the previously unfaulted, younger Yellow and White Limestone Groups have been included into the present deformation. Bedding measurements from the ridges indicate slightly asymmetric NNW-striking folds with the west limb dipping more steeply than the east limb (Fig. 7a-c). The axial planes strike between 333-349° and dip 85-88° E. The eastern dips of the axial planes further support
east-dipping reverse faults coring these structures (Figs 6-8; Erslev, 1991). Draper (2008)
documents a similar geometry for Long Mountain and Dallas Mountain anticlines, just east
of the capitol city of Kingston (Fig. 2b). The Don Figuerero Mountains differ from the Santa
Cruz Mountains and Brisco Mountain (Appendix 2) in that it has prominent reverse faults,
the Spur Tree and Porus faults, on its western and eastern edges, respectively. Based on these
two faults and the SCFZ to the south and the CJFS to the north, the Don Figuerero Mountains
have the geometry of a Laramide-style block uplift. 125 The E-striking strike-slip faults are not as straightforward to interpret as the NNW- striking faults. Despite being major topographic features, they demonstrate minimal strike- slip offset (e.g., Mitchell, 2003). Rather, the E-striking faults are strike-slip faults with variable vertical displacement as a result of motion of the hanging walls of the reverse faults.
Consequently, the vertical displacement on these faults varies along strike, depending on whether the fault block that is bounded by NNW-striking faults moves up or down at that location (e.g., the local lows and highs).
The two prominent fault sets in Jamaica are not orthogonal. The E-striking faults are parallel to the plate boundary and the contraction orientation, whereas the NNW faults, which are primarily contractional features, are not orthogonal to the E-striking faults, the plate boundary, or the direction of shortening. Because of this counterclockwise obliquity, the NNW faults have a small component of left-lateral motion that also accommodates contraction across the restraining bend (Fig. 12). This left-lateral component is apparent in the focal mechanism for the Santa Cruz fault (Fig. 1c) and possibly in the topography related to the Spur Tree fault – the Don Figuerero Mountains curve to the west at their northern extent (Figs 2 and 6). We interpret this curving as a result of the small amount of left-lateral motion along the Spur Tree fault that has forced more material to the north.
6.2. Reactivated transfer zone model for the Jamaica stepover
Previous work in Jamaica has focused on synthesizing existing datasets (e.g., Burke et al.,
1980; Wadge and Dixon, 1984; Mann et al., 1985, 2007; DeMets and Wiggins-Grandison,
2007; Draper, 2008), and so we continue their work here, by incorporating new gravity, geologic, and topographic data and recent geodetic results of Chapter 1 into all pre-existing 126 datasets. We use all of the datasets to evaluate previous models for Jamaica and to generate a
new model for the Jamaica restraining bend.
6.2.1. Previous tectonic models for Jamaica
We first consider whether previously published models for the neotectonics of
Jamaica (Fig. 13) correctly predict the island’s fault geometry, earthquake focal mechanisms,
and pattern of seismicity. Previous models for how slip is transferred are (1) left-lateral shear
across a broad, east-west striking zone that crosses the island (Burke et al., 1980; Wadge and
Dixon, 1984), (2) two right-stepping restraining bends that connect the Plantain Garden fault
and the SCFZ to the Duanvale fault (Mann et al. 1985), or (3) A series of CCW-rotating
blocks bounded by the island’s major E-striking strike-slip faults (Draper, 2008).
Burke et al. (1980) first proposed a broad zone of simple shear for Jamaica distributed
on the Duanvale, Plantain Garden-Cavaliers-Rio-Minho-Crawle River, and the South Coast
fault zones. In this model, the E-striking strike-slip faults are the primary features (Fig. 13a).
The NW-striking faults and folds are secondary features that are the result of a wide left- lateral simple shear zone that has been active for the past ~10 Myr. Additionally, Burke et al.
(1980) assert that though some major NNW faults existed prior to the Miocene (e.g.,
Wagwater Belt), most of the current active faults in Jamaica are not reactivated but rather a result of the present tectonics.
Mann et al. (1985) suggest that two restraining bends define Jamaica tectonics (Fig.
13b). They propose that compressional structures in Jamaica occur at discrete stepovers between the interacting strike-slip faults rather than along the entire length of the strike-slip faults. The first restraining bend occurs where the Plantain Garden fault steps to north to the
Duanvale fault along the western edge of the Blue Mountains. The second restraining bend 127 occurs in southwestern Jamaica where the SCFZ steps north to the Duanvale fault via the
Spur Tree, Santa Cruz, and Montpelier-Newmarket fault zones. Mann et al. (1985) interpret this restraining bend to be younger than the Plantain Garden-Duanvale bend.
Most recently, Draper (2008) proposed that ‘domino tectonics’ (Proffett, 1977) - a series of counter-clockwise-rotating blocks bounded by the major E-striking strike-slip faults on the island -characterize present deformation (Fig. 13c). Draper’s (2008) model is not in opposition to that of Mann et al. (1985), but rather provides greater structural detail to this existing model. Draper (2008) asserts that most NNW faults are reactivated Paleogene structures and that some of the E-striking faults are reactivated, but some E-striking faults have only formed since the beginning of this deformational event. Using the paleomagnetic results of Gose and Testamarta (1983) and evidence of right-lateral, horizontal slip on some faults, Draper (2008) asserts that fault-bounded blocks in Jamaica have rotated ~10º counterclockwise about a vertical axis in the past ~10 Myr in order to accommodate the present deformation in Jamaica.
None of the proposed models described above account for which faults are active or their sense of slip. Some models (e.g., domino blocks, Fig. 13c) are consistent with the widespread seismicity; some are consistent with the sense of fault slip indicated by some of the focal mechanisms (e.g., a single fault would produce NNW oriented oblique reverse focal mechanisms); some (e.g., domino blocks) correctly match the fault geometry. None, however, explains all four datasets. For example, the domino blocks model (Fig. 13c) supports widespread seismicity but incorrectly suggests dextral slip on the NNW-striking faults, which is not observed. 128 6.2.2. End-member models for restraining bends
It is conceivable that deformation in Jamaica may follow a more typical model for restraining bends observed from other tectonic settings. Restraining bends occur along most strike-slip faults, including transcurrent plate boundaries like sections of the Alpine fault in
New Zealand (Little et al., 2005) and the Big Bend along the San Andreas fault (e.g., Matti et al., 1985; Fitzenz and Miller, 2004; Rust, 1998; Dolan et al., 2007). Classification of restraining bends typically relies on the relative contribution of strike-slip and thrust faults
(e.g., Cowgill et al., 2004). End-member models of restraining bends are entirely strike-slip dominated systems (Fig. 13d, e.g., Akato Tagh fault in the Altyn Tagh system; Cowgill et al.,
2004) and entirely thrust-dominated systems (Fig. 13e, e.g., Santa Cruz bend along the San
Andreas fault; Anderson, 1990; Schwartz et al., 1990).
Jamaica does not fit into either of these end-member models. For Jamaica, a strike- slip dominated model predicts a velocity field that decreases from north to south but is constant from east to west, inconsistent with the measured geodetic velocities. Additionally, it predicts strike-slip faults parallel to the bend and no reverse faults (Fig. 13d). Conversely, the thrust-dominated model predicts decreasing velocities from east to west but constant velocities from north to south, in even greater conflict with the measured geodetic velocities.
The fault geometry of this end-member is closer to what is observed in Jamaica with the reverse faults parallel to the bend of the restraining bend (Fig. 13e). Jamaica fits best into the mixed-mode restraining bend (Fig. 13f, e.g., San Bernardino) of Cowgill et al. (2004), where oblique-slip is predicted on faults parallel to the bend. However, Jamaica lacks the symmetry around the main fault that all of these models propose. 129 6.2.3. Reactivated transfer zone model
We propose that southern Jamaica is a reactivated transfer zone. Transfer zones
develop during rifting and are strike-slip or oblique-slip fault zones, which link spatially
separated areas of extension (Faulds and Varga, 1998) (Fig. 11d). The areas of extension
consist of normal faults, which terminate at the bounding transfer zones. The normal faults locally affect the slip sense on the transfer zone and create areas of relief along the strike of the transfer zone.
Transfer zones have been documented to affect later contractional structures (e.g.,
Etheridge, 1986). We propose that in southern Jamaica, the CJFS and SCFZ are the original transfer zones and the NNW faults are the reactivated normal faults. In reactivating these faults, the two sets must interact to accommodate deformation across the restraining bend
(Fig. 11). This model suggests that none of the faults accommodates all of the relative plate motion between the Gônave and Caribbean plates.
There are four key aspects to how a reactivated transfer zone explains current deformation. First, the reactivated transfer zone model successfully explains the co-existence of active E-striking strike-slip faults and NNW-striking reverse faults. Since these features were formed together in extension, the extent of the NNW-striking faults is necessarily limited by the E-striking faults. This explains why the Montpellier-Newmarket zone only
occurs N of the CJFS. Second, the model explains the widespread seismicity and the sense
of slip suggested by earthquake focal mechanisms (Fig. 1c). In general, the NNW-striking
reverse faults record contraction and the E-striking strike-slip faults record left-lateral
translation. In areas where these two fault systems are close to one another, oblique slip
occurs to allow both sets to continue to operate. Third, the reactivated transfer zone model 130 explains how NNW reverse faults create vertical offset locally on the SCFZ and CJFS. Since the NNW-striking normal faults were originally bounded by the E-striking strike-slip faults, when the normal faults are reactivated, rocks move either up or down relative to the bounding strike-slip faults, depending on the exact geometry. In this way, the topographic, geologic, and gravity observations of local uplifts and depressions – along the same strike- slip fault – can be reconciled.
The last aspect of the reactivated transfer model is that the displacement rate along the E-striking strike-slip faults varies along strike, depending on the amount of shortening accommodated within the panel of NNW faults (Fig. 11). The SCFZ bounds deforming southern Jamaica from the Caribbean plate to the south, and the CJFS bounds the deforming southern portion of Jamaica from the Gônave microplate to the north. North of the CJFS, northern Jamaica moves at the rate of the Gônave microplate within uncertainties (Chapter
1), when the effects of elastic strain are subtracted. South of the SCFZ, all material is nearly fixed relative to the Caribbean plate (with the exception of the southward movement of all sites in the Jamaica archipelago). However, a possible E-W gradient of motion within the panel of fault-bounded blocks between the CJFZ and the SCFS occurs, where easternmost
Jamaica moves at a rate closer to the Gônave microplate, and westernmost Jamaica moves at a rate closer to that of the Caribbean plate. The amount of strike-slip movement on either of the bounding, strike-slip faults depends on how much of the east-west shortening has occurred along the panel of NNW-striking faults. Although it is difficult to determine the long-term geodetic field because of the effects of locked faults, the result is broadly consistent with this conceptual model. 131 In addition to being consistent with the deformation observed throughout Jamaica, the
reactivated transfer zone model solves a major conundrum associated with neotectonic
deformation of Jamaica. The Blue Mountains are Jamaica’s highest and most seismically
active mountains and yet exhibit a cross-range GPS velocity gradient of only 1-2 mm yr-1
(DeMets and Wiggins-Grandison, 2007; Chapter 1). In the reactivated transfer zone model, the Blue Mountains are a small-scale restraining bend that forms the eastern boundary of the reactivated transfer zone. The Blue Mountains thus accommodate relatively little contraction within the overall Jamaica restraining bend. However, their seismic activity and reverse- sense focal mechanisms and high relief are consistent with being eastern boundary of the reactivated transfer zone.
7. Conclusion
Southern Jamaica accommodates Gônave-Caribbean plate motion and has the
geometry of a restraining bend. The E-striking South Coast fault zone and the central
Jamaica fault system, two of the major strike-slip fault zones on the island, bound a panel of
NNW-striking reverse faults. As shown by geological observation and gravity studies, the
reverse faults accommodate east-west contraction across the restraining bend, whereas the E-
striking faults record local vertical displacement in addition to strike-slip offset. We propose
that a reactivated transfer zone model best describes southern Jamaica. In this model, a
Paleogene transfer zone – consisting of normal faults bounded by strike-slip faults – is
reactivated by strike-slip neotectonic deformation. Based on the pre-existing geometry of the
faults, the plate boundary in Jamaica forms a restraining bend. The E-striking faults
accommodate different amounts of strike-slip displacement along strike, as a result of the 132 increasing amount of shortening accumulated on the NNW reverse faults. This model
explains the lack of a single large-offset strike-slip fault across the island, the reason for the
existence of a restraining bend in Jamaica, the island-wide seismicity and uplift, and the lack of a clear geodetic signal near the Blue Mountains in Jamaica.
8. Acknowledgments
None of this work would have been possible without more than a decade of dedicated
support from the Jamaica Earthquake Unit of the University of West Indies, especially Paul
Williams. We thank Dr. Raymond Wright for his support and assistance in Jamaica. We also
thank Paul Mann for his insight into Jamaican geology. This project was funded by National
Science Foundation grant EAR-0609578. Figures were produced using Generic Mapping
Tools software (Wessel and Smith, 1991).
9. References
Abercrombie, R.E., Bannister, S., Pancha, A., Webb, T.H., and Mori, J.J., 2001,
Determination of fault planes in a complex aftershock sequence using two-dimensional
slip inversion: Geophysical Journal International, v. 146, p. 134-142.
Anderson, R.S., 1990, Evolution of the northern Santa Cruz Mountains by advection of crust
past a San Andreas fault bend: Science, v. 249, p. 397-401.
Arden, D.D., 1975, Geology of Jamaica and the Nicaraguan Rise, in Stehli, F.G., ed., The
ocean basins and margins: the Gulf of Mexico and the Caribbean: New York, Plenum
Press, 1975, p. 617-661. 133 Beauchamp, W., Barazangi, M., Demnati, A., El Alji, M., 1996, Intracontinental rifting and
inversion: Missour Basin and Atlas Mountains, Morocco, American Association of
Petroleum Geologists Bulletin, v. 80, p. 1459-1482.
Beauchamp, W., Allmendinger, R., Barazangi, M., Demnati, A., El Alji, M., and Dahmani,
M., 1999, Inversion tectonics and the evolution of the High Atlas Mountains, Morocco,
based on a geological-geophysical transect: Tectonics, v. 18, p. 163-184.
Burke, K., Grippi, J., and Sengor, A.M.C., 1980, Neogene structures in Jamaica and the
tectonic style of the northern Caribbean plate boundary zone: Journal of Geology, v. 88,
p. 375-386.
Chandler, V.W., McSwiggen, P.L., Morey, G.B., Hinze, W.J., and Anderson, R.R., 1989,
Interpretation of seismic reflection, gravity and magnetic data across Middle Proterozoic
Mid-Continent rift system, Northwestern Wisconsin, Eastern Minnesota and Central
Iowa: American Association of Petroleum Geologists Bulletin, v. 73, p. 261-275.
Cobbold, P.R., Meisling, K.E., and Mount, V.S., 2001, Reactivation of an obliquely rifted
margin, Campos and Santos basins, southeastern Brazil: American Association of
Petroleum Geologists Bulletin, v. 85, p. 1925-1944.
Cowgill, E., Yin, A., Arrowsmith, J.R., Feng, W.X., and Shuanhong, Z., 2004, The Akato
Tagh bend along the Altyn Tagh fault, northwest Tibet 1: Smoothing by vertical-axis
rotation and the effect of topographic stresses on bend-flanking faults: Geological
Society of America Bulletin, v. 116, p. 1423-1442, doi: 10.1130/B25359.1.
Cunningham, A.D., The Neogene evolution of the Pedro Channel carbonate system, northern
Nicaragua Rise [Ph.D. thesis]: Rice University, Houston, TX, United States (USA) 859
p. 134 De Paola, N., Mirabella, F., Barchi, M.R., and Burchielli, F., 2006, Early orogenic normal
faults and their reactivation during thrust belt evolution: the Gubbio Fault case study,
Umbria-Marche Apennines (Italy), in Butler, R.W.H., Tavarnelli, E., Grasso, M., and
Holdsworth R.E., eds., Tectonic inversion and structural inheritance in mountain belts,
Journal of Structural Geology Special Edition, v. 28, 1948-1957.
DeMets, C., and Wiggins-Grandison, M., 2007, Deformation of Jamaica and motion of the
Gônave microplate from GPS and seismic data: Geophysical Journal International, v.
168, p. 362-378, doi: 10.1111/j.1365-246X.2006.03236.x.
Dolan, J.F., Bowman, D.D., and Sammis, C.G., 2007, Long-range and long-term fault
interactions in Southern California: Geology, v. 35, p. 855-858, doi:
10.1130/G23789A.1.
Draper, G., 1998, Geologic and tectonic evolution of Jamaica: Contributions to Geology,
University of the West Indies, Mona, no. 3, p. 3-9.
Draper, G., 2008, Some speculations on the Paleogene and Neogene tectonics of Jamaica:
Geological Journal, v. 43, p. 563-572, doi: 10.1002/gj.1124.
Duncan, D., Hine, A.C., and Droxler, A.W., 1999, Tectonic controls on carbonate sequence
formation in an active strike-slip setting: Serranilla Basin, northern Nicaraguan Rise,
Caribbean Sea: Marine Geology, v. 160, p. 355-382.
Duncan, R.A., and Hargraves, R.B., 1984, Plate tectonic evolution of the Caribbean region in
the mantle reference frame, in Bonini, W.E., Hargraves, R.B., and Shagam, R., eds.,
Geological Society of America Memoir, v. 162, p. 81-93. 135 Edgar, N.T., Ewing, J.I., and Hennion, J., 1971, Seismic refraction and reflection in the
Caribbean Sea: American Association of Petroleum Geologists Bulletin, v. 55, p. 833-
870.
Erslev, E.A., 1991, Trishear fault-propagation folding: Geology, v. 19, p. 617-620, doi:
10.1130/0091-7613(1991)019<0617:TFPF>2.3.CO;2.
Etheridge, M.A., 1986, On the reactivation of extensional fault systems: Philosophical
Transactions of the Royal Society of London. Series A, Mathematical and Physical
Sciences 317(1539): Major Crustal Lineaments and Their Influence on the Geological
History of Continental Lithosphere: p. 179-194.
Eva, A.N. and McFarlane, N.A., 1985, Tertiary to early Quaternary carbonate facies
relationships in Jamaica: Transactions of the 4th Latin American Geological Conference,
Port of Spain, Trinidad, 7th-15th July, 1979, v. 1, p. 210-219.
Ewing, J., Antoine, J., and Ewing, M., 1960, Geophysical measurements in the western
Caribbean Sea and in the Gulf of Mexico: Journal of Geophysical Research, v. 65, p.
4087-4126.
Faulds, J.E., and Varga, R.J., 1998, The role of accommodation zones and transfer zones in
the regional segmentation of extended terranes, in Faulds, J.E., and Stewart, J.H., eds.,
Accommodation Zones and Transfer Zones: The Regional Segmentation of the Basin
and Range Province: Boulder, Colorado, Geological Society of America Special Paper
323.
Fitzenz, D.D., and Miller, S.A., 2004, New insights on stress rotations from a forward
regional model of the San Andreas fault system near its Big Bend in southern California:
Journal of Geophysical Research, v. 109, B08404, doi: 10.1029/2003JB002890. 136 Gomez, F., Beauchamp, W., and Barazangi, M., 2000, Role of the Atlas Mountains
(northwest Africa) within the African-Eurasion plate-boundary zone: Geology, v. 28, p.
775-778.
Gose, W.A., and Testamarta M.M., 1983, Paleomagnetic results from sedimentary rocks in
Jamaica: initial results: Journal of the Geological Society of Jamaica, v. 22, p. 16-24.
Green, G.W., 1977, Structure and stratigraphy of the Wagwater Belt, Kingston, Jamaica:
Overseas Geology and Mineral Resources, v. 48, p. 1-21.
Hastie, A.R., Kerr, A.C., Mitchell, S.F., and Millar, I.L., 2008, Geochemistry and
petrogenesis of Cretaceous oceanic plateau lavas in eastern Jamaica: Lithos, v. 101, p.
323-343, doi: 10.1016/j.lithos.2007.08.003.
Hays, W.H., 1976, Interpretation of gravity data: U.S. Geological Survey, U.S. Geological
Survey Open-File Report 76-479, 148 p.
Hill, K.C., Keetley, J.T., Kendrick, R.D., and Sutriyono, E., 2004, Structure and hydrocarbon
potential of the New Guinea fold belt, in McClay, K.R., ed., Thrust tectonics and
hydrocarbon systems: American Association of Petroleum Geologists Memoir, v. 82, p.
494-514.
Holgate, N., 1969, Palaeozoic and Tertiary transcurrent movements on the Great Glen Fault:
Scottish Journal of Geology, v. 5, p. 97-139.
Horsfield, W.T., 1974, Major faults in Jamaica: Journal of Geological Society of Jamaica, v.
14, p. 1-15.
Hose, H.R. and Versey, H.R., 1956, Palaeontological and lithological divisions of the Lower
Tertiary limestones of Jamaica: Colonial Geology and Mineral Resources, v. 6, p. 19-39. 137 Houtz, R. and Ludwig, W.J., 1977, Structure of the Columbia Basin, Caribbean Sea, from
profiler-sonobuoy measurements: Journal of Geophysical Research, v. 82, p. 4861-4867.
International Seismological Centre, 2001, On-line Bulletin, http://www.isc.ac.uk,
International Seismological Centre, Thatcham, United Kingdom.
Jackson, T.A., 1977, The petrochemistry and origin of the Tertiary volcanics in the
Wagwater Belt [Ph.D. thesis]: University of the West Indies – Mona Campus, Kingston,
Jamaica, 267 p.
Jackson, T. and Smith, T.E., 1979, Tectonic significance of basalts and dacites in the
Wagwater Belt, Jamaica, Geological Magazine, v. 116, p. 365-374.
Kashfi, M.S., 1983, Geology and hydrocarbon prospects of Jamaica: AAPG Bulletin, v. 67,
p. 2117-2124.
Kennedy, W.Q., 1946, The Great Glen Fault, Quarterly Journal of the Geological Society of
London, v. 405, p. 41-76.
Konstantinovskaya, E.A., Harris, L.B., Poulin, J., and Ivanov, G.M., 2007, Transfer zones
and fault reactivation in inverted rift basins: Insights from physical modeling:
Tectonophysics, v. 441, p. 1-26, doi: 10.1016/j.tecto.2007.06.002.
LaCoste, and Romberg, 2004, Instructional Manual: Model G & D Gravity Meters: Austin,
Texas, LaCoste and Romberg, 127 p.
Leroy, S., Mercier de Lepinay, B., Mauffret, A., and Pubellier, M., 1996, Structural and
tectonic evolution of the eastern Cayman Trough (Caribbean Sea) from seismic
reflection data: American Association of Petroleum Geology, v. 80, p. 222-247. 138 Lewis, J. and Draper, G., 1990, Geology and tectonic evolution of the northern Caribbean
margin, in Dengo, G., and Case, J.E., eds., The Geology of North America, v. H. The
Caribbean Region: Geological Society of America, Boulder, CO, p. 77-140.
Little, T.A., Cox, S., Vry, J.K., and Batt, G., 2005, Variations in exhumation level and uplift
rate along the oblique-slip Alpine fault, central Southern Alps, New Zealand: Geological
Society of America Bulletin, v. 117, p. 707-723, doi: 10.1130/B25500.1.
Mann, P., 2007, Global catalogue, classification, and tectonic origin of restraining- and
releasing bends on active and ancient strike-slip fault systems, in Cunningham, W.D.,
and Mann, P., eds., Tectonics of Strike-Slip Restraining and Releasing Bends:
Geological Society of London Special Publication, v. 290, p. 13-142, doi
10.1144/SP290.2.
Mann, P. and Burke, K., 1990, Transverse intra-arc rifting: Paleogene Wagwater Belt,
Jamaica: Marine and Petroleum Geology, v. 7, p. 410-427.
Mann, P., Calais, E., Ruegg, J.-C., DeMets, C., Jansma, P., and Mattioli, G.S., 2002, Oblique
collision in the northeastern Caribbean from GPS measurements and geological
observations: Tectonics, v. 37, 10.1029/2001TC001304.
Mann, P., DeMets, C., and Wiggins-Grandison, M., 2007, Toward a better understanding of
the Late Neogene strike-slip restraining bend in Jamaica: geodetic, geological, and
seismic constraints: in Cunningham, W.D. and Mann, P., eds., Tectonics of Strike-Slip
Restraining and Releasing Bends: Geological Society, London, Special Publications, v.
290, p. 239-253, doi: 10.1144/SP290.8.
Mann, P., Draper, G., and Burke, K., 1985, Neotectonics of the Caribbean: Reviews of
Geophysics and Space Physics, v. 22, p. 309-362. 139 Mann, P., Taylor, F.W., Edwards, R.L, and Ku, T., 1995, Actively evolving microplate
formation by oblique collision and sideways motion along strike-slip faults: an example
from the northeastern Caribbean plate margin: Tectonophysics, v. 246, p. 1-69.
Matti, J.C., Morton, D.M., and Cox, B.F., 1985, Distribution and geologic relations of fault
systems in the vicinity of the central Transverse Ranges, southern California: U.S.
Geological Survey Open-File Report 85-365, 23 p.
Meyerhoff, A.A. and Krieg, A., 1977, Petroleum prospects of Jamaica: Special Publication of
Ministry of Mining and Natural Resources, Jamaica, 131 p.
Mitchell, S.F., 2003, Sedimentology and tectonic evolution of the Cretaceous rocks of central
Jamaica: Relationships to the plate tectonic evolution of the Caribbean, in Bartolini, C.,
Buffler, R.T., and Blickwede, J., eds., The Circum-Gulf of Mexico and the Caribbean:
Hydrocarbon habitats, basin formation, and plate tectonics: AAPG Memoir, v. 79, p.
605-623.
Mutti, M., Droxler, A.W., and Cunningham, A.D., 2005, Evolution of the Northern
Nicaragua Rise during the Oligocene-Miocene: Drowning by environmental factors:
Sedimentary Geology, v. 175, p. 237-258, doi: 10.1016/j.sedgeo.2004.12.028.
Natural Disaster Research, Inc., Earthquake Unit, Mines & Geology Division, 1999,
Kingston Metropolitan Area Seismic Hazard Assessment Final Report, Caribbean
Disaster Mitigation Project, Organization of American States,
www.oas.org/cdmp/document/kma/seismic/kma1.htm, 82 pp.
Officer, C., Ewing, J., Hennion, J., Harkinder, D., and Miller, D., 1959, Geophysical
investigations in the eastern Caribbean-Summary of the 1955 and 1956 cruises, in 140 Ahrens, L.M., Press, F., Rankama, K., and Runcom, S.K., eds., Physics and Chemistry
of the Earth, v. 3, New York, Pergamom, p. 17-109.
Pindell, J.L., 1993, Regional synopsis of Gulf of Mexico and Caribbean evolution: Gulf
Coast Section Society of Economic Paleontologists and Mineralogists Foundation 13th
Annual Research Conference, Houston, TX, United States, Dec. 6-9, 1992, v. 13, p. 251-
274.
Pindell, J.L. and Barrett, 1990, Geologic evolution of the Caribbean: A plate-tectonic
perspective, in Dengo, G. and Case, J.E., eds., The Caribbean Region: The Geology of
North America, Geological Society of America, Boulder, Colorado, p. 405-432.
Pindell, J. and Kennan, L., 2001, Kinematic Evolution of the Gulf of Mexico and Caribbean:
GCSSEPM Foundations 21st Annual Research Conference Transactions, Petroleum
Systems of Deep-Water Basins, December 2-5, 2001, p. 193-220.
Proffett, J.M., 1977, Cenozoic geology of the Yerington District, Nevada, and implications
for nature and origin of Basin and Range faulting: Geological Society of America
Bulletin, v. 88, p. 247-266.
Quintana, L., Alonso, J.L., Pulgar, J.A., and Rodríguez-Fernáxndez, L.R., 2006,
Transpressional inversion in an extensional transfer zone (the Saltacaballos fault,
northern Spain), in Butler, R.W.H., Tavarnelli, E., Grasso, M., and Holdsworth R.E.,
eds., Tectonic inversion and structural inheritance in mountain belts, Journal of
Structural Geology Special Edition, v. 28, 2038-2048.
Robinson, E., 1994, Jamaica, in Donovan, S.K. and Jackson, T.A., eds., Caribbean Geology:
An Introduction: University of West Indies Publisher’s Association, Kingston, p. 111-
123. 141 Robinson, E., 2004, Coastal changes along the coast of Vere, Jamaica over the past two
hundred years: data from maps and air photographs: Quaternary International, v. 120, p.
153-161, doi: 10.1016/j.quaint.2004.01.014.
Rosencrantz, E., 1995, Opening of the Cayman Trough and the evolution of the northern
Caribbean plate boundary: Geological Society of America Annual Meeting, v. 27, p.
153.
Rosencrantz, E., and Mann, P., 1991, SeaMARC II mapping of transform faults in the
Cayman Trough, Caribbean Sea: Geology, v. 19, p. 690-693.
Rust, D., 1998, Contractional and extensional structures in the transpressive “Big Bend” off
the San Andreas Fault, Southern California, in Holdsworth, R.E., Strachan, R.A., and
Dewey, J.F., eds., Continental Transpressional and Transtensional Tectonics: Geological
Society of London Special Publication, v. 135, p. 119-126.
Sagir, A., 2001, Ancient deep faults, their reactivation and peculiarities under different
geodynamic conditions in eastern Yakutia (northeast Russia): Polaforschung, v. 69, p.
117-184.
Sandwell, D.T. and Smith, W.H.F., 1997, Marine gravity anomaly from Geosat and ERS 1
altimetry: Journal of Geophysical Research, v. 102, p. 10039-10054.
Santos, F.A.M., Almeida, E.P., Mateus, A., Matias, H.C., Matos, L., and Mendes-Victor,
L.A., 2000, Magnetotelluric study of a plio-quaternary tectonic depression: the Vilarica
basin (NE Portugal): Journal of Applied Geophysics, v. 44, p. 1-14, doi: 10.1016/S0926-
9851(99)00068-3. 142 Schwartz, S.Y., Orange, D.L., and Anderson, R.S., 1990, Complex fault interactions in a
restraining bend on the San Andreas fault, southern Santa Cruz Mountains, California:
Geophysical Research Letters, v. 17, p. 1207-1210.
Smith, T.E. and Jackson, T.A., 1974, Tertiary spilites and quartz keratophyres of the
Wagwater Belt, Jamaica, West Indies: The Bulletin Volcanologique, v. 38, p. 870-890.
Stewart, M., Strachan, R.A., and Holdsworth, R.E., 1999, Structure and early kinematic
history of the Great Glen Fault Zone, Scotland: Tectonics, v. 18, p. 326-342, doi:
10.1029/1998TC900033.
Talwani, M., Worzel, J.L., and Landisman, M., 1959, Rapid gravity computations for two-
dimensional bodies with application to the Mendocino submarine fracture zone: Journal
of Geophysical Research, v. 64, p. 49-61.
Versey, H.R. and Prescott, G.C., 1958, Progress report on the geology and groundwater
resources of the Clarendon Plains, Jamaica. Occasional paper 1, Geological Survey
Department, Kingston, 27 p.
Vos, I.M.A., Bierlein, F.P., Barlow, M.A., and Betts, P.G., 2006, Resolving the nature and
geometry of major fault systems from geophysical and structural analysis: The
Palmerville Fault in NE Queensland, Australia, in Butler, R.W.H., Tavarnelli, E.,
Grasso, M., and Holdsworth R.E., eds., Tectonic inversion and structural inheritance in
mountain belts, Journal of Structural Geology Special Edition, v. 28, 2097-2108.
Wadge, G., Brookes, S., and Royall, M., 1983, Structure models of the lower Vere plains,
Jamaica: Journal of the Geological Society of Jamaica, v. 22, p. 1-9.
Wadge, G. and Dixon, T.H., 1984, A geological interpretation of SEASAT-SAR imagery of
Jamaica, Journal of Geology, v. 92, p. 561-581. 143 Wessel, P. and Smith, W.H.F., 1991, Free software helps map and display data: EOS,
Transactions of the American Geophysical Union, v. 72, p. 441-446.
Wiggins-Grandison, M.D., 2004, Simultaneous inversion for local earthquake hypocenters,
station corrections and 1-D velocity model of the Jamaican crust: Earth and Planetary
Science Letters, v. 224, p. 229-240, doi: 10.1016/j.epsl.2004.05.009.
Wiggins-Grandison, M.D., and Atakan, K., 2005, Seismotectonics of Jamaica: Geophysical
Journal International, v. 160, p. 573-580, doi: 10.1111/j.1365-246X.2004.02471.x.
Withjack, M.O., Olsen, P.E., and Schlishe, R.W., 1995, Tectonic evolution of the Fundy rift
basin, Canada: Evidence of extension and shortening during passive margin
development: Tectonics, v. 14, p. 390-405.
Wright, R.M., 1975, Aspects of the geology of Tertiary limestones in west-central Jamaica
[Ph.D. thesis]: Stanford University, Stanford, CA, United States (USA), 320 p. 144 10. Figure captions
Figure 1. A - Jamaica is located in a restraining bend on the left-lateral strike-slip oceanic plate boundary between the Caribbean plate and the Gônave microplate. Arrow shows North
American plate motion relative to the Caribbean plate (number in parentheses is rate in mm yr-1). CSC – Cayman spreading center. Modified from DeMets and Wiggins-Grandison
(2007). 2-min seafloor bathymetry and land topography from Sandwell and Smith (1997). B -
Major faults of Jamaica modified from Wiggins-Grandison and Atakan (2005). Interpreted reverse faults are: BM – Blue Mountain fault, BR – Brompton fault, JC – John Crow fault,
PS – Pondside fault, PO – Porus fault, SCr – Santa Cruz fault, ST – Spur Tree fault, and
WMN and EMN – Western and Eastern Montpelier-Newmarket faults, respectively. Strike- slip faults are: AV – Aeolus Valley fault, CF – Cavaliers fault, DF – Duanvale fault, PG –
Plantain Garden fault, RMCR – Rio Minho-Crawle River fault, SFS - Siloah fault system,
SCFZ – South Coast fault zone, and WF – Walton fault (northern and southern strands).
Faults are overlain on a geologic map of Jamaica modified from Hastie et al. (2008). C -
Earthquakes from the International Seismological Centre for the period 2000-2010 (June); magnitudes range from 2.0 to 5.9. Earthquake focal mechanisms #1-2 are from Van Dusen &
Doser (2000), #3-4 from Wiggins-Grandison (2001), #5-48 from DeMets & Wiggins-
Grandison (2007), and #49-54 from the ISC for the period 2005-2008. Focal mechanisms
‘beach balls’ are scaled to magnitude. Black dots in the focal mechanisms indicate P axes.
Data are overlain on 90-m Space Shuttle Radar Topographic Mission (SRTM) with topography illuminated from the southwest.
145 Figure 2. A - Cretaceous inliers of Jamaica. Modified from Hastie et al. (2008). B - Cities and mountains of Jamaica mentioned in the text. Cities are italicized.
Figure 3. Jamaica GPS site velocities relative to the campaign station PIKE, with one sigma,
2D error ellipses (Chapter 1). Velocities shown in black are north of the central Jamaica fault system and Plantain Garden fault, whereas velocities shown in red are south of these faults. Data are overlain on 90-m SRTM topography illuminated from the southwest. Scale is shown in upper right.
Figure 4. A - Slope angle generated from 90-m SRTM. Gentle slopes are blue and steepest slopes are dark red. B - Slope dip direction generated from 90-m SRTM. Slopes are color- coded based on dip direction.
Figure 5. Topographic profiles across NNW mountains and along ridges. Major faults are labeled. See Figure 1 for fault abbreviations. A - West to east profile across southwestern
Jamaica based on 90m SRTM. B-E - NNW to SSE topographic profiles based on 90m SRTM along ridges of mountains.
Figure 6. A - Geologic map of western Jamaica. Modified from Wright (1975). See Figure 1 for fault abbreviations. B - Cross-section modified from Wright (1975) based on field mapping and gravity. Towns are italicized. Key corresponds to parts A and B.
146 Figure 7. Lower hemisphere equal-area net plots. Number in lower right corresponds to
number of data points. Figures A-C, E, and F show poles to bedding (circles), inferred axial surfaces of folds (great circles), and fold hinge orientations (squares) for mountain ranges in southern Jamaica. A – Santa Cruz Mountains, B – Don Figuerero Mountains, C - Brisco
Mountain, D - Kemp’s Hill faults (great circles) and slickenlines (open circles), and E and F
- two different folds from quarry within Siloah fault system.
Figure 8. Gravity transects across the Don Figuerero and Santa Cruz Mountains. A - Black
solid circles show locations of gravity stations. Open white circles are locations of gravity
stations collapsed onto transects orthogonal to the mountain ranges. Transect A-A’ crosses
the Don Figuerero Mountains and transect B-B’ crosses both mountain ranges. B - Bouguer
gravity (circles) and elevation (squares) along traverse A-A’. C - Bouguer gravity (circles)
and elevation (squares) along traverse B-B’. D - Two-dimensional gravity model across
Santa Cruz and Don Figuerero Mountains. Open circles are gravity measurements from C.
Line is modeled anomaly based on cross-section below. Faults are dashed near surface
because it is not clear if faults have faulted the younger limestone.
Figure 9. A – Two-dimensional gravity transects across the Vere Plain. Locations of
transects are shown on map and on Figure 10. For each transect, the top panel shows the
Bouguer gravity data as open circles and the line denotes the gravity anomaly based on the
cross-section in the lower panel. All transects consist of 3 rocks types: alluvium (2.0 g/cm3),
limestone (2.5 g/cm3) and volcanics (2.7 g/cm3). Lower right map - reverse faults are denoted
with black triangles on the hanging walls. “U” and “D” are used when faults are vertical. In 147 panels A and B, contours do not have a consistent contour interval because well data allow for better constraints in the northern Vere Plain. Original Versey and Prescott (1958) maps had contours in feet, but here are converted to meters. B - Structure contour map of the base of the Quaternary alluvium based on well data in the Vere Plain and the map of Versey and
Prescott (1958). Black dots show locations of wells. Non-shaded areas are Miocene limestone at the surface. Faults are based on 2D gravity transects shown in A. Dashed black rectangle shows limits of Versey and Prescott map. C - Structure contour map of the base of the limestone/top of the volcanics. Faults are based on 2D gravity transects. Contour interval is 30.5 m (1000 ft).
Figure 10. Bouguer gravity anomaly map across the Vere Plain. Gravity stations are denoted by black dots. One mGal contours. The anomalies are greatest in Versey and Prescott’s
(1958) buried valley west of Kemp’s Hill and also south of the SCFZ. Locations of two- dimensional gravity transects of Figure 9 are shown. Bold black lines show limits of limestone exposures. Trace of SCFZ is shown with a purple line. Black dashed rectangle shows limits of Versey and Prescott (1958) map.
Figure 11. A – Interaction between major faults of southern Jamaica. See Figure 1 for fault abbreviations. The SCFZ and the AV fault define the southern boundary and the CJFS, BM fault, and PG fault define the northern boundary. The BM fault is interpreted to be the eastern boundary of plate boundary deformation and forms a small-scale restraining bend resulting in the uplift of the Blue Mountains. B – Fault kinematics of southern Jamaica. Grey arrow shows displacement vectors relative to the Caribbean plate. C – Three-dimensional block 148 diagram of fault kinematics of southern Jamaica – reactivated transfer zone model. Grey
panels are the CJFS and SCFZ. D – Three-dimensional block diagram of transfer zone of
southern Jamaica during the Eocene, when the faults formed. Grey panels are the CJFS and
SCFZ.
Figure 12. Diagram showing how plate motion is accommodated at NNW-striking oblique
slip reverse faults (map view). Top – ideal fault arrangement: if the two fault sets were
orthogonal, the reverse faults would accommodate only contraction and the east-west faults
would accommodate left-lateral plate motion. Bottom – Jamaica fault arrangement: reverse
faults are misoriented by ~30º so they primarily accommodate contraction but they also have
a component of left-lateral shear.
Figure 13. Previously proposed conceptual models of the Jamaica restraining bend. A -
Broad shear zone model for Jamaica with anticlines (triangles) and synclines (crosses)
(Burke et al., 1980; Wadge and Dixon, 1984). B - Two-discrete-restraining-bends model for
Jamaica (Mann et al., 1985). C - Series of domino blocks with dextral motion on NNW
oriented faults (Draper, 2008). D and E – End-member models of restraining bends (Cowgill et al., 2004). F – Mixed-mode restraining bend (Cowgill et al., 2004).
149 150 151 152 153 154 155 156 157 158 159 160 161 162
163 Appendix 1: Geologic history of Jamaica
It is necessary to put our current understanding of the tectonics of Jamaica into context with Jamaica’s geologic history. Jamaica has experienced four major episodes of deformation: 1) Island arc development from the Early Cretaceous to early Cenozoic; 2)
Rifting of the arc in the Eocene; 3) Subsidence in the Oligocene-Miocene; and 4) Faulting and folding from the Middle Miocene to the Holocene (Meyerhoff and Krieg, 1977; Mitchell,
2003; Draper, 2008) (Fig. A1). Here, we propose that the fourth major episode could be divided into two events, by the formation of the Gônave microplate at ~5 Ma, causing a change in the active faulting of Jamaica.
Jamaica, along with Cuba, Hispaniola, Puerto Rico, the Virgin Islands, Tobago,
Margarita and Venezuela, is part of the Mesozoic Great Caribbean Arc (Pindell and Kennan,
2001). Rocks related to Cretaceous tectonism of the volcanic arc are best preserved in the
Central Inlier (Mitchell, 2003), one of the largest of the 27 structural inliers within Jamaica
(Robinson, 1994; Fig. 2a). During the Cretaceous, east-striking and steeply north dipping
reverse faults formed in the Central Inlier (Mitchell, 2003). Cretaceous units in the area have
a similar strike, but with a gentler dip, suggesting north-south shortening with reverse faults
and back-arc basin inversion (Mitchell, 2003) (Fig. A1a).
During the middle Eocene, rifting initiated in Jamaica as the Cayman spreading
center formed immediately north of Jamaica. The left-lateral Oriente and Swan Islands faults formed to the east and west, respectively, of the spreading center to accommodate strike-slip displacement (Cunningham, 1998). Subduction ceased beneath the Nicaragua Rise and transitioned from NW-SE convergence to E-W strike-slip displacement (Pindell and Barrett,
1990). As a result of the ENE-WSW elongation associated with rifting, NNW-striking faults 164 formed across Jamaica (Draper, 2008) (Fig. A1b). At this point, the area of Jamaica was a
transfer zone (Fig. A1b).
Evidence for rifting is best exposed in two inverted basins: The Wagwater Belt and
Montpellier-Newmarket zone. The Wagwater Belt occurs between the Central Inlier and the
Blue Mountains Inlier (Fig. 2a) and contains Eocene-age dacites and basalts (Smith and
Jackson, 1974; Jackson, 1977; Jackson and Smith, 1979) and at least 5.6 km of sandstones
and conglomerates (Mann and Burke, 1990). The Montpelier-Newmarket Zone, in the
western part of the island, has the thickest sequence (966 m) of the Eocene Yellow
Limestone in Jamaica (Wright, 1975). The Yellow Limestone Group unconformably overlies
the Cretaceous igneous and sedimentary rocks and at one point covered all of Jamaica except
possibly the Blue Mountains (Fig. A1b).
The entire Northern Nicaragua Rise subsided from late Oligocene to late Early
Miocene (Mutti et al., 2005) (Fig. A1c). Thick layers of limestone, specifically the Eocene
Yellow Limestone Group and Middle Eocene-Middle Miocene White Limestone Group
record subsidence in Jamaica. The White Limestone Group ranges between 600 and 2,450 m
thick, and is inferred to have covered all of Jamaica; it presently outcrops over two-thirds of
the island (Kashfi, 1983). Although pre-existing NNW-striking normal faults largely
controlled the locations of the depocenters for the White Limestone Group (Hose and Versey,
1956; Wright, 1975), syn-depositional movement on the faults is inferred to have been
minimal (Meyerhoff and Krieg, 1977).
Transcurrent motion in Jamaica started in the Miocene, coincident with reinitiation of the Cayman Spreading Center at ~20 Ma (Rosencrantz, 1995) (Fig. A1d). At this point, we propose reactivation of many of the NNW- and E-striking faults that were active during the 165 Eocene, and thus reactivate the transfer zone as a contractional feature. With the fault geometry of the transfer zone, Jamaica became a restraining bend along the plate boundary.
At ~5 Ma, the E-W elongate Gônave microplate, which is located between the North
America and Caribbean plates, is interpreted to have formed (Fig. 1a; Mann et al., 1995,
2002). Around this time, deformation ceased in northern Jamaica and localized in southern
Jamaica (Fig. A1e). This localization resulted in a smaller stepover in the overall geometry of the Jamaica restraining bend.
Figure caption
Figure A1. Geologic history of Jamaica. A – During the Cretaceous, subduction occurred NE of Jamaica. Evidence of this is preserved in 27 inliers and includes reverse faulting in the center of the island. B – In the Eocene, E-W strike-slip faults and NNW normal faults accommodate opening associated with the Cayman Spreading Center as a transfer zone in
Jamaica. The Yellow Limestone was deposited across the island, except in the Blue
Mountains. C – Subsidence of the entire Nicaragua Rise resulted in deposition of the White
Limestone across all of Jamaica from the Late Oligocene-Early Miocene. D – Transcurrent motion related to the Caribbean-North America plate boundary began in the Middle
Miocene. E-W and NNW faults are reactivated across the island, and reactivate the transfer zone as a contractional feature. E – Gônave microplate forms at ~5 Ma and deformation localizes in southern Jamaica.
166
167 Appendix 2: Other NNW ridges and faults
1. Brisco Mountain (Brompton fault)
1.1. Topography
Brisco Mountain is a NW-striking mountain in westernmost Jamaica, just north of the town of Black River (Fig. 2). It is ~10 km long and 2-3 km wide and reaches 610 m in elevation (Figs 1b and 2b). At its northern limit, an east-west valley cuts across it. Brisco
Mountain’s southern extent is just north of the town of Black River (Fig. 2b). It is similar to both the Don Figuerero Mountains and Santa Cruz Mountains, in that the western slope is steeper than the eastern slope (Fig. 4). The western slope of Brisco Mountain is ~18° and rises up from sea level to 610 m, whereas the eastern side slopes ~4-5° and only descends to
230 m elevation.
1.2. Geology
Eocene-Oligocene Bonny Gate Formation of the White Limestone Group is exposed along most of Brisco Mountain. As the mountain loses elevation toward Black River,
Miocene Newport Formation, Miocene-Pleistocene Coastal Limestone Formation, and
Quaternary alluvium are present (Fig. 4).
Bedding measurements of the Bonny Gate Formation and Newport Formation from
Wright (1975) and our own observations across Brisco Mountain show that these stratigraphic units generally dip parallel to topography (Figs 4 and 7c). These bedding measurements indicate a NNW-striking, upright fold with an axial plane oriented 333, 88°E
and a fold hinge oriented 4° ! 154 (Fig. 7c). The orientation of the axial plane and fold 168 hinge are subparallel to those of the folds of the Don Figuerero and Santa Cruz Mountains.
Wright (1975) maps the Brompton fault at the southwestern base of Brisco Mountain with a vertical dip and with the northeast-side-up (Fig. 6).
Based on topography and associated bedding, it is likely that, similar to the Santa Cruz
Mountains and western Don Figuerero Mountains, Brisco Mountain is the result of a fault- propagation fold.
2. Hill Top hill (Pondside fault)
2.1. Topography
Hill Top hill is the westernmost range in southern Jamaica. It is also the smallest range examined in this paper, at just 9 km long and 800 m wide. The ridge is at most 100 m in elevation and decreases in elevation to the north. Hill Top hill trends ~351°, and its western side is steeper than its east (Fig. 4). At the latitude of 19.70° N, a 1.70 km long elongate lake, Wally Wash Great Pond, occurs just west of Hill Top hill. Five kilometers south of Wally Wash Great Pond, Hill Top hill defines the coastline for ~2.5 km (Fig. 5). In this area, Hill Top hill is ~85 m in elevation and drops to sea level over ~300 m, with a cliff of ~10 m at the coast.
2.2. Geology
The Miocene Newport Formation outcrops along the ridge of Hill Top hill and dips gently to the east (Fig. 6). Just west of the elongate hill and surrounding the Wally Wash
Great Pond is Quaternary alluvium. Based on recent field study of both the topography and geology, the Pondside fault has east-side-up motion and juxtaposes Quaternary alluvium to the west with Miocene Newport Formation of the White Limestone Group to the east. 169
CHAPTER 4
Character of the Caribbean – Gônave – North America plate boundaries in the upper
mantle based on shear-wave splitting
B. Benford, B. Tikoff, and C. DeMets
Department of Geoscience
University of Wisconsin – Madison
1215 W. Dayton St.
Madison, Wisconsin 53706, USA
To be submitted, Geophysical Research Letters.
170 Abstract
We present new shear-wave splitting measurements of SKS, SKKS, PKS, and sSKS phases from eight stations in the northern Caribbean. Prior to this work, shear-wave splitting analysis of the northern Caribbean boundary was only evaluated at a station in Puerto Rico.
Stations that lie within several tens of kilometers of microplate boundaries have mean fast polarization directions parallel to the boundary and have delay times greater than 1 s.
Stations more than several tens of kilometers away from microplate boundaries show no evidence for an anisotropic upper mantle. Stations in Cuba and Jamaica have fast axes oriented ~100° with delay times of ~1.5 s, indicating that the east-striking left-lateral strike- slip faults that define the north and south boundaries of the Gônave microplate continue into the upper mantle. A station located in Antigua, where the North America plate subducts beneath the Caribbean plate, has a high degree of splitting with the fast axis parallel to the trench. Based on our results, the deformation related to the presence of microplates in the northern Caribbean extends into the upper mantle.
171 1. Introduction
Seismic anisotropy is a powerful tool for understanding deformation of the mantle since
it is based on the preferred orientation of minerals, primarily olivine, in response to tectonic
strain [e.g., Mainprice et al., 2000]. The upper mantle is composed primarily of olivine
(70%) and deformation of this mineral phase produces a lattice-preferred orientation, which imparts an anisotropy [Crosson and Lin, 1971; Christensen, 1984; Nicolas and Christensen,
1987]. In an isotropic upper mantle, PKS, SKS, SKKS, and sSKS waves are radially polarized, whereas in an anisotropic upper mantle, shear waves split into fast and slow components and are transversely polarized [Savage, 1999]. Seismic anisotropy analysis enables the determination of the polarization direction (!) of the fast shear wave and the delay time (!t) between the arrival of the fast and slow waves, where ! and !t are the result of the foliation and lineation in the upper mantle and !t is the result of both the degree of anisotropy and the thickness of the layer [Mainprice and Silver, 1993]. Between the depths of
100 and 200 km, Fresnel zones have radii between 40 and 60 km, so lateral resolution is only a few tens of kilometers [e.g., Alsina and Snieder, 1995; Margheriti et al., 2003]. This enables lateral variations in the degree of fabric development to be observed.
To understand the structure of the upper mantle, SKS, SKKS, PKS, and sSKS waves are ideal for several reasons. First, the observed anisotropy can be constrained to the receiver side because of the conversion to/back to an S-wave from a P-wave at the core-mantle boundary. Second, each of these waves represents a nearly vertical ray path through the mantle so the propagation direction is nearly constant [Silver and Chan, 1991; Mainprice and
Silver, 1993]. Third, these waves are radially polarized when they leave the core-mantle
172 boundary [Silver, 1996], so any transverse energy at the receiver is a result of anisotropy
[Silver and Chan, 1991].
Many aspects of active deformation in the northern Caribbean are already well understood. Obliquely convergent left-lateral slip of 19-20 mm yr-1 between the Caribbean
and North America plates [Dixon et al., 1998; DeMets et al., 2000; 2007; Lopez et al., 2006;
DeMets et al., 2010; Fig. 1], including the oblique collision of the Bahama platform with
Hispaniola [Mann et al., 1995, 2002], appears to drive westward movement of the Puerto
Rico-Virgin Islands (PRVI), Hispaniola, and Gônave microplates [Mann et al., 1995;
Chapter 2]. GPS measurements show 2.6±2.0 mm yr-1 of westward motion of the PRVI microplate relative to the Caribbean plate [Jansma and Mattioli, 2005] and 7-13 mm yr-1 of
westward Gônave microplate motion [DeMets and Wiggins-Grandison, 2007; Chapter 1],
consistent with extension across the Mona Passage between Puerto Rico and Hispaniola
[Jansma et al., 2000; Jansma and Mattioli, 2005]. In a related paper, we modeled a 126-
station GPS velocity field for the Caribbean plate and its northern boundary to better
understand the geometry of present deformation, fault slip rates, and presence of microplates
[Chapter 2]. Here, using shear-wave splitting, we provide data about the fabric of the upper
mantle at the northern Caribbean boundary and show that the deformation associated with the
presence of microplates extends into the upper mantle.
2. Data and Methods
In order to understand the nature of the Caribbean-North America plate boundary and the
associated microplates in the upper mantle, we analyzed seismograms from eight broadband
stations on the islands of Jamaica, Hispaniola, Cuba, Puerto Rico, Grand Turk of Turks and
173 Caicos, Barbuda, and St. Thomas of the U.S. Virgin Islands (Fig. 1). Five stations (ANWB,
GRTK, GTBY, MTDJ and SDDR) are from the Caribbean network, one station (SDD) is
from the Dominican Republic network, one (SJG) is from the IRIS/USGS network, and one
(STVI) is from the Puerto Rico network. Prior to this study, shear-wave splitting had only been evaluated at station SJG.
We examined events from 2008, 2009, and January of 2010. Station SDD was not
operating until the middle of 2009, so we extended its window to August of 2011. The station
SJG in Puerto Rico was part of the Russo et al. [1996] study, and so we include their time window (May 1993 – January 1994) for comparison since they used the method of Silver and
Chan [1991] to estimate splitting parameters, and we used this method as well as two other methods (described below).
For this analysis, we included events that are 85º-140º from the broadband station and have a minimum moment magnitude (Mw) of 5.5, the minimum required for a necessary signal-to-noise ratio of 5:1. For each station, we obtained between 225 and 650 events that fit these criteria. However, many of these events did not show a clear arrival of the particular phase and were discarded. The National Earthquake Information Center preliminary determination of epicenters catalog (U.S. Geological Survey) was used for event locations and origin times.
Inversion of the data to determine ! and !t was done using the Matlab-based program
SplitLab [Wüstefeld et al., 2008]. To control the quality of the results, SplitLab allows the user to process each event individually. This program uses three well-known methods simultaneously to determine the shear-wave splitting parameters !t and ! (see Fig. S1 in the
Supplementary Material), namely, the rotation correlation method [Bowman and Ando, 1987;
174 Vinnik et al., 1989], the minimum eigenvalue method [Silver and Chan, 1991], and the
minimizing transverse component method [Silver and Chan, 1991; Savage and Silver, 1993].
SplitLab also assigns a quality level to each event. Here, we present events where there is
good agreement between the results of the three methods.
We carried out 412 individual splitting measurements. Of these, 246 were non-null
measurements. In Data Sets S1 and S2 of the auxiliary material, we present the following
results for each event: the phase used, the backazimuth and angle of incidence of the event,
the manually applied filter, the determined splitting parameters and the error bars determined
from the 95% confidence interval for the three methodologies. Each event is also assigned a
quality of good, fair, poor, fair null, or good null. Supplementary Figure S1 gives an
example of good quality splitting data for an event at station MTDJ in Jamaica. These ratings
are based on the similarity of the solutions from the rotation-correlation method and the
minimum energy method [Wüstefeld and Bokelmann, 2007]. The rating is based on: 1) a
high signal-to-noise ratio, 2) a small confidence region, 3) good linearization of the
transverse component, 4) good correlation between the two shear waves, and 5) good
correlation between the two methods. Ratings of good, fair, or poor meet all five, four, or
three or fewer of these criteria, respectively. Filtering was applied manually in order to
maximize the signal-to-noise ratio. Most events were bandpass filtered using a combination
of corner frequencies typically between 0.01 and 0.2 Hz in order to remove high-frequency
and long-period signals.
During our analysis, we determined 166 “null” measurements (Data Set S2 in the
auxiliary material). Null measurements are when the shear wave has not been split, so that
there is a clear arrival on the radial component and no energy on the transverse component of
175 the seismogram. Bonnin et al. [2010] suggests three reasons for null measurements: 1) The incoming wave only traveled through an isotropic medium, 2) The polarization of the incoming wave is parallel to the slow or fast direction in the anisotropic medium, or 3) Two anisotropic layers with orthogonal fabrics that cancel out each other’s delay time exist beneath the station. Null events that have a “good” quality are events that have a high signal- to-noise ratio on the radial component and minimal energy on the transverse component.
Fair nulls are events that have minimal energy on the transverse component but not enough for observing splitting.
3. Results
In evaluating splitting at each station, we consider both fair and good non-null measurements (Figs 2 and 3 and Tables 1, S1, and S2). Below we present the mean !t and ! with 1-" uncertainties using the rotation correlation method. The means are of the combined good and fair non-null splitting results.
We first compare our results for station SJG in Puerto Rico with those of the Russo et al.
[1996] study, including events from the time window of Russo et al. (May 1993-January
1994) and from January 2008 to December 2009. Based on nine measurements, Russo et al.
[1996] determined ! was oriented 85°±1° and !t = 1.2±0.2 s. For our analysis and using 10 measurements, we determined that ! is oriented 101°±20° with !t = 1.2±1.1 s (Table 1). Our results agree within the uncertainties of Russo et al. [1996], however, our uncertainties are significantly larger than those of Russo et al. [1996]. Based on the number of measurements, our larger uncertainties are consistent with other Splitlab analyses [e.g., Barruol et al., 2011].
176 Additionally, if we calculate the mean and its uncertainty from the Russo et al. [1996] results,
using the method employed in this study, it increases the uncertainty in ! and !t by ~3 times.
We were only able to measure three splitting events at station SDD in Puerto Rico,
indicating that either the time span was not long enough or that a strong fabric is not present
beneath this station. Station STVI in the U.S. Virgin Islands (Fig. 1) records unusually high
!t and has a high uncertainty (2.9±1.2 s) and a high degree of scatter in ! (61°±54°), making
determination of a fabric beneath this station impossible at this point. Similarly, we were
unable to get any fair or good non-null or null measurements at station PUCM (70.68° W,
19.44° N, not shown in figures) located in northern Hispaniola.
The highest !t (1.50-1.56 s) and lower uncertainties in ! (~29°) occur at stations located
near active plate boundaries (Figs 1 and 3 and Table 1): station MTDJ in Jamaica and GTBY
in Cuba. MTDJ is located on the left-lateral Gônave-Caribbean plate boundary, where east- west strike-slip rates are estimated at 6.5±0.5 mm yr-1 [Chapter 1]. GTBY occurs on the left-
lateral Gônave-North America plate boundary, where east-west strike-slip rates are higher,
~14-15 mm yr-1 [Chapter 2]. Good agreement exists between ! and the plate boundary
orientation (dark thick line in Fig. 3). However, ! only agrees with absolute plate motion
(dashed line in Fig. 3) when plate motion is subparallel to the plate boundary (e.g., station
GTBY).
Station GRTK, located in the Bahamas and away from an active plate boundary (Fig. 1), has the lowest !t (0.8±0.5 s) and a high degree of scatter in ! (13°±44°) (Table 1, Fig. 3).
177 4. Discussion
Station GRTK is the station farthest from an active plate boundary. Its low !t and
high degree of scatter in ! is a good basis for comparison with stations located closer to the
plate boundaries.
4.1. Gônave microplate boundaries
The two stations along the northern and southern left-lateral E-W boundaries of the
Gônave microplate have !t = ~1.5 s with ! oriented subparallel to the boundary (Table 1,
Figs 1-3). In both instances, ! is 10-15° clockwise from the left-lateral boundary, consistent with left-lateral shear [e.g., Ramsay, 1980] at depth, suggesting that deformation at the north and south boundaries of the Gônave microplate extend into the upper mantle. This result is also consistent with the 5° obliquity in a clockwise-sense measured at the right-lateral
Caribbean-South America boundary [Russo et al., 1996]. Finally, although the boundaries move at different rates, the degree of fabric development is comparable.
At station SDDR (Fig. 1), ! is oriented parallel to the trend of the northwest mountains of central Hispaniola (Fig. 2). Station SDD, located just to the east, has no evidence for a fabric in the mantle. The presence of a fabric at SDDR and its orientation may either be a fossil fabric preserved from Cretaceous subduction [Bowin, 1966], or may be a consequence of slow relative motion between distinct Gônave and Hispaniola microplates described in
Chapter 2.
178 4.2. Caribbean – North America boundary
At station ANWB, located on the island of Barbuda on the Caribbean plate, just inside
the Lesser Antilles subduction zone, ! is oriented parallel to the plate boundary. Chapter 2
predicts 10.6±1.5 mm yr-1 of left-lateral strike-slip motion and 17.1±1.0 mm yr-1 of convergence at the subduction zone front. The observation that ! is parallel to the boundary is likely the combined result of current strike-slip motion and that at subduction-zone fronts,
! has been documented to be trench-parallel [e.g., Shih et al., 1991; Russo and Silver, 1994;
Gledhill and Gubbins, 1996; Hammond et al., 2009].
4.3. Character of microplate boundaries
The presence of fabrics at stations GTBY, MTDJ, and SDDR suggests that deformation associated with the plate boundaries extend into the upper mantle. The weaker/lack of fabric at stations located tens of kilometers from a microplate boundary (GRTK, SDD, SJG, STVI) indicates that upper mantle deformation is localized along the boundaries of the microplate.
Of particular relevance is the SDD site on Hispaniola, which indicates that the upper mantle is locally undeformed, but surrounded by deformed zones. A similar situation exists with the
SJG station on Puerto Rico, which contains only minor fabric relative to adjacent sites.
These results indicate that microplates of relatively undeformed upper mantle exist below upper-crustal microplates. The inference that crustal deformation continues, although is somewhat wider, in the underlying mantle is consistent with the relatively large delay times observed locally. That is, the microplates of the northern Caribbean region extend into the mantle.
179 4.4. Lithospheric-scale deformation at transcurrent plate boundaries
Figure 4 is a three-dimensional block diagram that depicts the inferred lithospheric
architecture beneath the northern margin of the Caribbean plate. The obliquity of the fast-
wave direction at sites GTBY and MTDJ is consistent with the left-lateral transcurrent
movement. Relative to the deformation on the Caribbean/South American plate boundary
[Russo et al., 1996], the SKS delay times are slower, likely reflecting the complicating
presence of the microplates along the northern edge of the Caribbean. The presence of
microplates results in more distributed deformation (e.g., deformation on both sides of the
microplate, rather than along a single plate boundary), which acts to lower the finite strain,
and hence the fabric development, at any point. However, the boundary is also distinct from
the pattern observed on the San Andreas fault system, which requires a two-layer anisotropy
model [e.g., Ozalaybey and Savage,1995; Hartog and Schwartz, 2001]. We attribute this difference to the difference in absolute plate motion relative to the orientation of the plate boundary [e.g., Tikoff et al., 2004].. The San Andreas system is oblique to the Pacific-North
America plate boundary, whereas the major faults (e.g., Oriente, Walton, Plantain Garden, and Enriquillo faults) bounding the Gonave microplate are parallel to the Caribbean-Gonave and Gonave-North America plate boundaries.
5. Conclusion
This work documents shear-wave splitting along the northern Caribbean boundary. Based on splitting results of SKS, SKKS, PKS, and sSKS phases at eight broadband seismometers in the northern Caribbean, we propose that deformation associated with the Caribbean-
Gônave, Gônave-North America, and Caribbean-North America plate boundaries continue
180 into the upper mantle. High degrees of splitting parallel to plate boundaries occur at 4 stations. Splitting at the two stations in Cuba and Jamaica are the result of left-lateral strike- slip motion at the northern and southern boundaries of the Gônave microplate. A station in
Hispaniola either records a fossil fabric from the Cretaceous or the present-day right-lateral transpressional boundary between the Gônave and Hispaniola microplates. The station located along the North America – Caribbean oblique subduction zone records a trench- parallel fast axis. Stations located farther from active plate boundaries do not show evidence for an anisotropic upper mantle and indicate that the boundaries of the microplates are localized in the upper mantle.
6. Acknowledgments
We thank Neal Lord and Lee Powell for assistance in data acquisition and processing. We also thank Paul Mann for his insight into Jamaica. We thank Ray Russo for his knowledge of shear-wave splitting and his assistance. Support was provided by NSF Tectonics Program
Grant 0609578.
7. References
Alsina, D. and R. Snieder (1995), Small-scale sublithospheric continental mantle
deformation: constraints from SKS splitting informations, Geophys. J. Int., 123, 431-
448.
Barruol, G., M. Bonnin, H. Pedersen, G.H.R. Bokelmann, and C. Tiberi (2011), Belt-parallel
mantle flow beneath a halted continental collision: The Western Alps, Earth Planet. Sci.
Lett., 302, 429-438, doi: 10.1016/j.espl.2010.12.040.
181 Bowin, C., 1966, Geology of the Central Dominican Republic (case history of part of an
island arc), in Caribbean geological studies: Geological Society of America Memoir 98,
edited by H. Hess, 11-84.
Bowman, J. R., and M. Ando (1987), Shear-wave splitting in the upper-mantle wedge above
the Tonga subduction zone, Geophys. J. R. Aston. Soc., 88, 25-41.
Christensen, N. I. (1984), The magnitude, symmetry and origin of upper mantle anisotropy
based on fabric analysis of ultramafic tectonites, Geophys. J. R. Astron. Soc., 76, 89-112.
Crosson, R. S., and J. W. Lin (1971), Voigt and Reuss prediction of anisotropic elasticity of
dunite, J. Geophys. Res., 76, 570-578.
DeMets, C., R.G. Gordon, and D.F. Argus (2010), Geologically current plate motions,
Geophys. J. Int., 181, 1-80, doi: 10.1111/j.1365256X.2009.04491.x.
DeMets, C., P.E. Jansma, G.S. Mattioli, T.H. Dixon, F. Farina, R. Bilham, E. Calais, and P.
Mann (2000), GPS geodetic constraints on Caribbean-North America plate motion:
Geophys. Res. Lett., 27, 437-440, doi: 10.1029/1999GL005436.
DeMets, C., G. Mattioli, P. Jansma, R.D. Rogers, C. Tenorio, and H.L. Turner (2007),
Present motion and deformation of the Caribbean plate: Constraints from new GPS
geodetic measurements from Honduras and Nicaragua, in Geologic and Tectonic
Development of the Caribbean Plate in Northern Central America: Geological Society
of America Special Paper 428, edited by P. Mann., 21-36, doi: 10.1130/2007.2428(02).
DeMets, C., and M. Wiggins-Grandison (2007), Deformation of Jamaica and motion of the
Gônave microplate from GPS and seismic data, Geophys. J. Int., 168, 362-378, doi:
10.1111/j.1365-246X.2006.03236.x.
182 Dixon, T.H., F. Farina, C. DeMets, P. Jansma, P. Mann, and E. Calais (1998), Relative
motion between the Caribbean and North American plates and related boundary zone
deformation from a decade of GPS observations, J. Geophys. Res., 103, 15,157-15182,
doi: 10.1029/97JB03575.
Gledhill, K. and D. Gubbins (1996), SKS splitting and seismic anisotropy of the mantle
beneath the Hikurangi subduction zone, New Zealand, Phys. Earth Planet. Inter., 95,
227-236.
Hammond, J.O.S., J. Wookey, S. Kaneshima, H. Inoue, T. Yamashina, and P. Harjadi (2009),
Systematic variation in anisotropy beneath the mantle wedge in the Java-Sumatra
subduction system from shear-wave splitting, Phys. Earth Planet. Inter., 178, 189-201,
doi: 10.1016/j.pepi.2009.10.003.
Hartog, R., and S.Y. Schwartz (2001), Depth-dependent mantle anisotropy below the San
Andreas fault system; apparent splitting parameters and waveforms, J. Geophys. Res.,
106, 4155-4167.
Jansma, P.E. and G.S. Mattioli (2005), GPS results from Puerto Rico and the Virgin Islands:
Constraints on tectonic setting and rates of active faulting, in Active tectonics and
seismic hazards of Puerto Rico, the Virgin Islands, and offshore areas, Geol. Soc. Am.
Spec. Paper, vol. 385, edited by P. Mann, pp. 13-30, doi: 10.1130/2007.2428(02), The
Geological Society of America, Boulder, CO.
Jansma, P.E., G.S. Mattioli, A. Lopez, C. DeMets, T.H. Dixon, P. Mann, and E. Calais
(2000), Neotectonics of Puerto Rico and the Virgin Islands, northeastern Caribbean,
from GPS geodesy, Tectonics, 19(6), 1021-1037.
183 Lopez, A.M., S. Stein, T. Dixon, G. Sella, E. Calais, P. Jansma, J. Weber, and P. LaFemina
(2006), Is there a northern Lesser Antilles forearc block?, Geophys. Res. Lett., 33,
L07313, doi: 10.1029/2005GL025293.
Mainprice, D., G. Barruol, and W. Ben Ismail (2000), The seismic anisotropy of the Earth’s
mantle: from single crystal to polycrystal, in Earth’s Deep Interior: Mineral Physics and
Tomography From the Atomic to the Global Scale, Geophys. Monogr. Ser., vol. 117,
edited by S. Karato et al., pp. 237-264, AGU, Washington, D. C.
Mainprice, D., and P. G. Silver (1993), Interpretation of SKS waves using samples from the
subcontinental lithosphere, Phys. Earth Planet. Int., 78, 257-280.
Mann, P., E. Calais, J. –C. Ruegg, C. DeMets, P. Jansma, and G. S. Mattioli (2002), Oblique
collision in the northeastern Caribbean from GPS measurements and geological
observations, Tectonics, 37, doi: 10.1029/2001TC001304.
Mann, P., F. W. Taylor, R. L. Edwards, and T. Ku (1995), Actively evolving microplate
formation by oblique collision and sideways motion along strike-slip faults: an example
from the northeastern Caribbean plate margin, Tectonophysics, 246, 1-69.
Margheriti, L., F. P. Lucente, and S. Pondrelli (2003), SKS splitting measurements in the
Apenninic-Tyrrhenian domain (Italy) and their relation with lithospheric subduction and
mantle convection, J. Geophys. Res., 108 (B4), 2218, doi: 10.1029/2002JB001793.
Nicolas, A., and N. I. Christensen (1987), Formation of anisotropy in upper mantle
peridotites – a review, in Composition, Structure and Dynamics of the Lithosphere-
Asthenosphere System, vol. 16, edited by K. Fuchs and C. Froidevaux, pp. 111-123,
American Geophysical Union, Washington, D. C.
184 Ozalaybey, S. and M.K. Savage (1995), Shear wave splitting beneath western United States
in relation to plate tectonics, J. Geophys. Res., 100, 18135-18149.
Ramsay, J. G. (1980), Shear zone geometry: a review, J. Struct. Geol., 2, 83-99.
Russo, R. M., and P. G. Silver (1994), Trench-parallel flow beneath the Nazca plate from
seismic anisotropy, Science, 263, 1105-1111.
Russo, R. M., P. G. Silver, M. Franke, W. B. Ambeh, and D.E. James (1996), Shear-wave
splitting in northeast Venezuela, Trinidad, and the eastern Caribbean, Phys. Earth
Planet. Inter., 95, 251-275.
Sandwell, D.T. and W.H.F. Smith (1997), Marine gravity anomaly from Geosat and ERS 1
altimetry, J. Geophys. Res., 102, 10039-10054.
Savage, M. K. (1999), Seismic anisotropy and mantle deformation: What have we learned
from shear wave splitting?, Rev. Geophys., 37, 65-106.
Savage, M. K., and P. G. Silver (1993), Mantle deformation and tectonics: constraints from
seismic anisotropy in western United States, Phys. Earth Planet. Inter., 78, 207-227.
Shih, X.R., R.P. Meyer, and J.F. Schneider (1991), Seismic anisotropy above a subducting
plate, Geology, 19, 807-810.
Silver, P. G., and W. W. Chan (1991), Shear wave splitting and subcontinental mantle
deformation, J. Geophys. Res., 96, 16,429-16,454.
Silver, P. G. (1996), Seismic anisotropy beneath the continents: Probing the depths, Annu.
Rev. Earth Planet. Sci., 24, 365-432.
Tikoff, B., R. Russo, C., Teyssier, and A. Tommasi (2004), Mantle-driven deformation of
orogenic zones and clutch tectonics, in Vertical Coupling and Decoupling in the
185 Lithosphere, vol. 227, edited by J. Grocott, K.J.W. McCaffrey, G. Taylor and B. Tikoff,
pp. 41-64, Geological Society of London, Special Publications, London, England.
Vinnik, L. P., R. Kind, G. L. Kosarev, and L. I. Makeyeva (1989), Azimuthal anisotropy in
the lithosphere from observations of long period S waves, Geophys. J. Int., 99, 549-559.
Wüstefeld, A., and G. Bokelmann (2007), Null detection in shear-wave splitting
measurements, Bull. Seis. Soc. Am., 97, 1204-1211, doi: 10.1785/0120060190.
Wüstefeld, A., G. Bokelmann, C. Zaroli, and G. Barruol (2008), SplitLab: A shear-wave
splitting environment in Matlab, Computers & Geosciences, 34, 515-528,
doi:10.1016/j.cageo.2007.08.002.
186
8. Figure Captions
Figure 1. Tectonic setting of the northern Caribbean. Bold white arrow shows MORVEL estimate of North America plate motion in mm yr-1 relative to the Caribbean plate [DeMets et al., 2010]. Abbreviations: CSC – Cayman spreading center, HI microplate - Hispaniola microplate, PRVI microplate - Puerto Rico-Virgin Islands microplate, and PR - Puerto Rico.
2-min seafloor bathymetry and land topography from Sandwell and Smith [1997].
Figure 2. A - Each individual good (red) or fair (blue) splitting result plotted at each station.
B – Mean of the good (red) and mean of the fair and good combined (blue) splitting results.
In both panels, the azimuth of each line indicates the trend of ! and the length of each line is proportional to the magnitude of !t. Fault traces are shown in black. C – Interpreted upper mantle fabric (shown in red) based on fast-axis orientations. Foliation is vertical and lineation is horizontal. Arrows show plate motion at boundaries. Mean good (red) and mean fair and good combined (blue) from panel B are also shown.
Figure 3. The first column shows rose diagrams for each station of good and fair measurements. The primitive circle equals 52% of the measurements in all diagrams. The second column shows rose diagrams of just the good non-nulls. The primitive equals 67% of the measurements in all diagrams. Stations SDD and STVI did not have enough good measurements to generate a meaningful diagram. In both columns, the number of measurements is shown in the lower left, the dashed line is the trend of absolute plate motion, and the solid, thick line is the trend of the nearest plate boundary. Absolute plate motion was
187 determined by using the predicted angular velocity at each location relative to ITRF08 from
Chapter 2.
Figure 4. Three-dimensional model for the lithosphere in the northern Caribbean. Fabric
(red) has a vertical foliation and horizontal lineation and is localized at borders of microplates, whereas interior of microplates have weak/no fabric.
188
189
190
191
192
193
9. Supplementary Material
Figure S1. Example of a good splitting measurement (event 2008.232) at station MTDJ. A –
Initial seismogram before analysis. Red solid line - radial component; blue dashed line - transverse component; gray zone - calculation window. Upper right panel is a stereoplot of the result. Center panel displays results for the rotation correlation method and lower panel displays the results for the minimum energy method of Silver and Chan [1991]. B –
Seismogram components in fast (red solid) and slow (blue dashed) directions after correction. C – Radial (Q, red solid) and transverse (T, blue dashed) components after correction. D – Particle motion before (blue dashed) and after (red solid) correction. E – Map of correlation coefficient in center panel and map of minimum energy values on transverse component.
Figure S2. Each individual good (red) or fair (blue) null measurement plotted as its backazimuth at each station.
Figure S3. The first column shows stereoplots of the non-nulls, in which the length of each line corresponds to the delay time, and the center of the line is the backazimuth and the inclination of the wave at the surface based on the rotation correlation method. The second column includes stereoplots of the nulls, where the symbol is plotted at the backazimuth and the inclination of the wave at the surface. Good non-nulls and nulls are plotted in black and fair non-nulls and nulls in grey.
194
195
196
197 Table S1. Non-null splitting measurements. Back- Inclin- Filter Rotation correlation Silver and Chan Eigenvalue Date Phase azimuth ation (Hz) ! "t ! "t ! "t Station: ANWB Quality: good 2008.009 SKKS 32.49 11.42 [0.01 0.1] 74.49 0.62 39< 72 <-64 0.2< 0.7 <3.2 44.5 1.4 2008.015 SKKS 254.63 11.31 [0.01 0.1] -71.37 1.23 86<-77 <-39 0.7< 1.4 <2.5 -33.4 1.7 2008.058 PKS 331.59 6.98 [0.02 0.2] -71.41 0.50 72<-72 <-37 0.3< 0.6 <1.6 -20.4 2.5 2008.110 PKS 258.35 6.03 [0.01 0.1] -64.65 1.40 -86<-60 <-33 0.9< 1.5 <2.4 80.4 3.0 2008.252 PKS 258.73 6.06 [0.02 0.2] -32.27 1.40 -48<-33 <-27 0.9< 1.4 <2.0 -41.3 1.1 2008.279 SKS 244.85 5.97 [0.01 0.1] -72.15 1.18 74<-79 <-39 0.6< 1.2 <3.2 54.9 4.0 2008.303 SKKS 45.71 11.99 [0.01 0.1] 82.71 1.78 70< 88 <-74 1.4< 2.0 <2.8 47.7 3.7 2008.312 SKKS 265.78 10.73 [0.01 0.1] -52.22 0.75 -90<-50 <-11 0.2< 0.8 <3.1 -68.2 1.0 2009.285 PKS 97.59 7.36 [0.02 0.2] 41.59 0.55 21< 44 < 78 0.3< 0.6 <1.1 27.6 0.8 2009.303 PKS 346.42 6.10 [0.01 0.1] -68.58 1.35 -90<-76 <-54 1.1< 1.7 <2.5 -31.6 1.8 2009.326 SKS 259.25 5.93 [0.01 0.1] -59.75 0.68 -Inf<-59