New Seismic Hazard Maps for Puerto Rico and the U.S. Virgin Islands

a) a) a) Charles Mueller, Arthur Frankel, M.EERI, Mark Petersen, a) M.EERI, and Edgar Leyendecker

The probabilistic methodology developed by the U.S. Geological Survey is applied to a new seismic hazard assessment for Puerto Rico and the U.S. Virgin Islands. Modeled seismic sources include gridded historical seismicity, subduction-interface and strike-slip faults with known slip rates, and two broad zones of crustal extension with seismicity rates constrained by GPS geodesy. We use attenuation relations from western North American and worldwide data, as well as a Caribbean-speciﬁc relation. Results are presented as maps of peak ground acceleration and 0.2- and 1.0-second spectral response acceleration for 2% and 10% probabilities of exceedance in 50 years (return periods of about 2,500 and 500 years, respectively). This paper describes the hazard model and maps that were balloted by the Building Seismic Safety Council and recommended for the 2003 NEHRP Provisions and the 2006 International Building Code. ͓DOI: 10.1193/1.3277667͔

INTRODUCTION We present results of a new probabilistic seismic hazard assessment for Puerto Rico and the U.S. Virgin Islands (PRVI). The study area (Figure 1) is located along the boundary between the northeastern Caribbean Sea and the Atlantic Ocean, at the intersection of the Greater and Lesser Antilles island chains. These islands demarcate the active boundary between the North America and Caribbean tectonic plates, with the North America plate moving west-southwestward relative to the Caribbean plate at a rate determined from GPS geodesy of 18–20 mm/yr (Jansma et al. 2000, Mann 2005). West of PRVI, the two plates move along typical transform structures (the east-west-striking, left- lateral Oriente and Enriquillo-Plantain Garden faults and related structures). Southeast of PRVI, the North America plate subducts orthogonally westward beneath the Caribbean plate, forming a typical volcanic island arc (the north-south-trending Lesser Antilles). Plate inter- action near PRVI, in contrast, occurs in a 250-km-wide, east-west-trending zone of complex transpressional deformation, delimited by the Puerto Rico Trench in the north and the Muer- tos Trough in the south. Puerto Rico and the Virgin Islands lie on a shallow submarine bank within this complex deformational zone. The region is home to approximately four million U.S. citizens and has a long history of destructive earthquakes. Historical records show that major earthquakes have struck PRVI several times during the past 500 years, although the locations and sizes of events that have occurred more than a few decades ago are poorly known. Major earthquakes

a) United States Geological Survey, P.O. Box 25046 MS 966, Denver, CO 80225

Bahama

Bank 19.4 mm/yr North American Plate

20 PT Main Rdg

SFZ BunceFZ H SFZ BowinFZ MC EPGFZ MP PR 18 VI AP MT

Caribbean Plate 16

deformation front 14 strike-slip fault normal fault

12 -74 -72 -70 -68 -66 -64 -62 -60

Figure 1. Map of North American—Caribbean plate boundary region. Inset box shows area of seismic hazard analysis. Arrow shows motion of North American plate relative to Caribbean plate from Jansma et al. (2000). Structural elements generalized from Grindlay et al. (1997), Dolan et al. (1998), LaForge and McCann (2005), and Grindlay et al. (2005). Abbreviations: H=Hispaniola, PR=Puerto Rico, VI=Virgin Islands, PT=Puerto Rico Trench, MT=Muertos Trough, MP=Mona Passage, AP=Anegada Passage, SFZ=Septentrional fault zone, EPGFZ =Enriquillo-Plantain Garden fault zone, and MC=Mona Canyon.

have damaged Puerto Rico in 1615, 1670, 1787 (estimated moment magnitude, Mw, about 8.0–8.2, thought to have ruptured the plate interface south of the Puerto Rico Trench, offshore northern Puerto Rico), 1844, 1846, 1867 (estimated Mw about 7.3, Anegada Passage, offshore southeastern Puerto Rico), 1875, 1906, 1918 (Mw 7.2–7.4, Mona Passage, offshore northwestern Puerto Rico), 1943 (Mw 7.6–7.8, plate interface, offshore northwestern Puerto Rico), and 1946 (Mw 7.8–8.0, plate interface, offshore northeastern Hispaniola). The 1867 and 1918 earthquakes were accompanied by destructive tsunamis. McCann (1985), Pacheco and Sykes (1992), Dolan and Wald (1998), McCann (2003a), Doser et al. (2005), and LaForge and McCann (2005) provide more information about PRVI historical earthquakes. Seismicity near PRVI is primarily related to 1) highly oblique subduction of the North America plate beneath the terranes of the plate boundary zone along the main plate interface south of the Puerto Rico Trench and 2) the interactions of several prob- able microplates within the complex boundary zone. Geodesy and seismicity data suggest the existence of a Puerto Rico-northern Virgin Islands microplate that is relatively rigid and seismically quiescent internally (Masson and Scanlon 1991, Jansma et al. NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 171

2000, Jansma and Mattioli 2005). Most of the major seismogenic structures are located offshore; thus, estimates of activity rates can be highly uncertain, often based on indirect evidence such as seismicity patterns and focal mechanisms, bathymetry and shallow sea- floor seismic imagery, regional geodesy, kinematic reasoning, and tectonic analogs. Seismic sources related to deformation along the main plate boundary include megathrust faulting along the plate interface, southward-deepening intraslab faulting within the subducting North America plate, and strike-slip faulting on several structures that strike subparallel to the Puerto Rico Trench north and northwest of Puerto Rico (Figure 1). These include the Septentrional fault, a plate-boundary structure in central Hispani- ola that extends eastward across the northern Mona Passage toward Puerto Rico, as well as the Bunce fault zone (formerly North Puerto Rico Slope fault zone) and the Bowin fault zone (formerly South Puerto Rico Slope fault zone) and related structures. Other sources include two broad zones of roughly east-west tectonic extension, one west of Puerto Rico roughly coincident with the Mona Passage, and one southeast of Puerto Rico roughly coincident with the Anegada Passage. Extension within these zones is thought to be related to differences in rates of eastward motion (relative to North America) of crustal blocks south of the main plate boundary (e.g., McCann et al. 1987, Jansma et al. 2000); the Caribbean plate moves eastward relatively unrestricted, while blocks within the boundary zone are restricted by relatively high-standing tectonic fea- tures like the Bahama Bank and Main Ridge (Figure 1). The Great Northern and Great Southern Puerto Rico faults, major left-lateral strike-slip systems active on Puerto Rico from the early Cretaceous to the early Miocene are now considered largely quiescent, although they seem to be associated with very small earthquakes, and may represent in- herited zones of weakness (McCann 1985). Prentice et al. (2000) determine a recurrence rate for one onshore fault, the South Lajas fault in southwestern Puerto Rico, that they consider to be currently active. Several other candidate faults have been identified in western Puerto Rico, but not yet fully evaluated. Unspecified crustal faulting (and the intraslab faulting) is modeled with gridded historical seismicity (below).

METHODOLOGY AND HAZARD MODEL We apply the probabilistic seismic hazard methodology developed by the U.S. Geo- logical Survey (USGS) as described by Frankel et al. (1996, 2000, 2002) and Petersen et al. (2008), and present maps of probabilistic ground motions: peak ground acceleration (PGA), and 0.2-second and 1.0-second spectral response acceleration (5%-damped) for 2% and 10% probabilities of exceedance in 50 years, corresponding to return times of about 2,500 and 500 years, respectively. Like others in the USGS seismic hazard series, these maps are intended to inform building codes and other structural-design guidelines, as well as earthquake-mitigation plans and emergency-response plans. Design maps derived from the hazard maps (Leyendecker et al. 2000) have been approved by vote of the BSSC for the NE- HRP Recommended Provisions for Seismic Regulations for New Buildings and Other Struc- tures, 2003 Edition (BSSC 2004), and accepted for the ASCE 7 guidelines (ASCE 2005) and the 2006 updates of the International Building Code and International Residential Code (ICC 2006a and 2006b). 172 MUELLER ET AL.

20 Hispaniola Seg Puerto Rico Seg

BFZ SFZ H

C ST V SJ SLF PR 18

Mona Extnl Zn Anegada SC Extnl Zn

HISTORICAL SEISMICITY (M>=4.5,1963-2001) 0 =< depth (km) < 50 50=

Figure 2. Map showing modeled seismic hazard sources. Diagonal pattern=subduction segments (sparse=Puerto Rico seg., dense=Hispaniola seg.). Vertical pattern=Mona and Anegada extensional zones. Abbreviations: SFZ=Septentrional fault zone, BFZ=Bowin fault zone, SLF =South Lajas fault, H=Hispaniola, PR=Puerto Rico, V=Vieques, C=Culebra, SC=St. Croix, ST=St. Thomas, SJ=St. John.

The USGS methodology is based on 1) historical seismicity generalized using expo- nential magnitude distributions with regionally determined b values and 2) specific fault and subduction sources with geologically or geophysically constrained slip-rate or recurrence information. Where there is reason to suspect that the seismicity or fault components of the model are incomplete, they can be supplemented with sources based on geodetic or other deformation data. Epistemic uncertainty in attenuation relations and source parameters is accommodated using logic trees. Earthquakes are assumed to occur randomly in time; the mapped probabilistic ground motions represent mean, time- independent seismic hazard. Specific elements of the PRVI hazard model (Figure 2) include gridded and smoothed seismicity with Mw greater than or equal to 4.5 from 1963 to 2001 divided into four depth ranges, megathrust faulting along the main subduction interface south of the Puerto Rico Trench, the Septentrional fault, the Bowin fault, the South Lajas fault, and Mona Passage, and Anegada Passage extensional zones with seismic activity rates determined from GPS geodesy. Because some sources overlap spatially, magnitude ranges are adjusted to avoid double counting hazard contributions. Like LaForge and McCann (2005), we exclude the Bunce fault because it is located more than 100 km NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 173 from the north shore of Puerto Rico and only a few kilometers south of the deformation front of the Puerto Rico Trench subduction interface (Dolan and Wald 1998). We also exclude the Muertos Trough and related fault structures, offshore southern Puerto Rico, because their rates of activity are poorly known and probably small. Proximity to the south coast of Puerto Rico, however, makes them important targets for future research. (LaForge and McCann 2005 include the Muertos structure in their model on a “cannot preclude” basis with a slip rate of about 0.5 mm/yr; it can be an important contributor at longer return periods for sites in southern Puerto Rico.) We use ground-motion-prediction relations for a uniform NEHRP B/C-boundary (firm-rock, VS30=760 m/s) site condition. For shallow-seismicity (depth less than 50 km), extensional-zone, and crustal-fault sources, we use a mix of relations for crustal earthquakes derived from western North American and worldwide data. For deeper-seismicity (primarily intraslab) and subduction-interface sources, we use relations based on worldwide data. (Spe- cific ground-motion relations are discussed below for each hazard-model element. In each case, the overall approach is consistent with the 2002 update of the USGS national seismic hazard maps. The so-called NGA relations were not available when this work was com- pleted.) A preliminary version of a ground-motion relation based on Caribbean data by Mo- tazedian and Atkinson (Motazedian 2003, called MA2003 hereinafter) is included for shallow-seismicity, extensional-zone, and crustal-fault sources. (Roughly two-thirds of the earthquakes in their dataset are shallower than 50 km; we use MA2003 for crustal sources only.) We adjust their relation for NEHRP B/C site conditions using factors derived from equations in Boore et al. (1997). High-frequency ground motions from the MA2003 relation are greater than those from comparable western North American relations (see “Discussion” section below), so pending further evaluation, MA2003 receives about one half the weight of other relations in the ground-motion logic tree. MA2003 has been revised for publication (Motazedian and Atkinson 2005), but the results presented here—and considered by the BSSC and other code groups—are based on the preliminary version. With a site grid spacing of 0.05 degree in latitude and longitude, we compute hazard curves for approximately 16,000 sites within the area mapped in Figure 2.

CATALOG AND GRIDDED SEISMICITY A uniform catalog is constructed from four sources: 1) a compilation by Engdahl and Villasenor (2002) of large global earthquakes, many carefully relocated (called EV, downloaded from ftp site ciei.colorado.edu/pub/user/engdahl/Handbook/CENT.CAT); 2) the International Seismological Centre global catalog (ISC, downloaded from the USGS National Earthquake Information Center); 3) the USGS Preliminary Determination of Epicenters global catalog (PDE, downloaded from the USGS National Earthquake In- formation Center); and 4) the Panamerican Institute of Geology and History catalog (IPGH, downloaded from the Web site of the Middle America Seismograph Consortium: midas.uprm.edu). Aware of well-documented past problems with magnitude determina- tions (e.g., LaForge and McCann 2005), we choose not to include data from the Puerto Rico Seismic Network catalog. The four source catalogs are reformatted, concatenated, and chronologically sorted. We remove duplicates and dependent events generally following the approach described by Mueller et al. (1997); duplicates are removed using 174 MUELLER ET AL. the above source-catalog preference order, and aftershocks and foreshocks are removed (to insure statistical independence in the hazard analysis) using the algorithm of Gardner and Knopoff (1974). The ﬁnal PRVI catalog is judged to be complete down to Mw 4.5 from 1963 through 2001; it includes 15 EV events, 354 ISC events, 137 PDE events, and 20 IPGH events. The catalog is divided into four subcatalogs by depth: 0–50 km (shallow), 50–80 km, 80–120 km, and greater than 120 km. Spatial relationships of the deeper-than-50-km hypocenters suggest that these earthquakes are primarily associated with (located within) the southward-subducting North American plate. The seismicity data, both shallow and deeper, are generally consistent with a b value of 1.0, which is subsequently applied for all hazard model elements. Gridded 10a values are computed for each subcatalog using a maximum-likelihood method (Weichert 1980) and are spatially smoothed using a two-dimensional Gaussian function with a 50-km correlation distance for shallow seismicity and a 30-km correlation distance for deeper seismicity (for details of the methodology see Frankel et al. 1996 and Petersen et al. 2008). These grids are used to compute the hazard contribution from seismicity for the distance range 0 to 200 km and Mw range 5.0 to 7.0. For grid cells located within the Mona and Anegada extensional zones, the maximum magnitude is reduced from Mw 7.0 to 6.5 to avoid overlapping hazard contributions (below). For shallow seismicity, we use MA2003 plus four ground-motion relations for crustal earthquakes with coefﬁcients for reverse-oblique (or averaged reverse-slip/strike-slip) faulting: Boore et al. (1997) for VS30=760 m/s, Sadigh et al. (1997) for rock, Campbell and Bozo- rgnia (2003) adjusted for VS30=760 m/s, and Abrahamson and Silva (1997) for rock. MA2003 receives 10% weight and each of the other four, 22.5% in the ground-motion logic tree. For deeper seismicity we use two intraplate ground-motion relations: Youngs et al. (1997) and Atkinson and Boore (2003), each with 50% weight.

SUBDUCTION INTERFACE We subdivide the subduction interface south of the Puerto Rico Trench into Puerto Rico and Hispaniola segments (Figure 2), generally following published ideas about tectonic segmentation, the estimated sizes of past megathrust earthquakes, and regional variations in seismic coupling (Frankel et al. 1980, Dolan and Wald 1998, LaForge and McCann 2005). Interface earthquakes on the Puerto Rico segment are modeled as characteristic, ﬂoating Mw-7.9 ruptures with 190-year recurrence times, corresponding to approximately 20% seismic coupling for a GPS-determined slip rate of about 17 mm/yr between North America and PRVI (Jansma et al. 2000). Interface earthquakes on the Hispaniola segment are modeled as characteristic, ﬂoating Mw-8.0 ruptures with 200-year recurrence times, corresponding to approximately 80% seismic coupling. Historically, the 1943 Mw-7.6–7.8 northern Mona Passage earthquake and the 1946 Mw-7.8–8.0 northeastern Hispaniola earthquake ruptured the Puerto Rico and Hispaniola subduction segments, re- ͑ ϳ ͒ spectively (Dolan and Wald 1998), and the great earthquake of 1787 Mw 8.0–8.2 probably ruptured the Puerto Rico subduction segment (McCann 1985). Modeled updip and downdip rupture edges correspond to plate-top depths of 10 and 40 km, respectively. We apply two ground-motion relations: Youngs et al. (1997) and Sadigh et al. (1997), each with 50% weight out to a crossover distance of 58 km, beyond which the Youngs relation receives NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 175 full weight. The Sadigh et al. (1997) relation, developed from crustal earthquake data, offsets a tendency for relations based on the empirical interface-earthquake dataset alone to under- predict ground motions close to simulated large thrust events (Youngs et al. 1997).

MONA PASSAGE AND ANEGADA PASSAGE EXTENSIONAL ZONES The Mona Passage between Puerto Rico and Hispaniola coincides with a broad zone of active crustal extension. Bathymetry, sub-sea seismic imaging, and focal-mechanism data are all suggestive of normal faulting on generally north-south-trending structures with east-west-directed extension. The most prominent bathymetric feature in the Pas- sage, the north-south-trending Mona Canyon, is thought to be normal-fault controlled, and was probably the site (Mercado and McCann 1998, Doser et al. 2005) of the 1918 Mw-7.2–7.4 tsunamigenic earthquake that shook northwestern Puerto Rico with large losses of life and property (LaForge and McCann 2005). We assign a rate of east-west extension of 5mm/yrbased on GPS geodesy results (Jansma et al. 2000, Jansma and Mattioli 2005), assume a b value of 1.0, and prorate faulting uniformly into each grid cell in the zone using a method described by Frankel et al. (1996) for computing hazard in areal zones. The result- ing 10a grids are smoothed using a 20-km Gaussian correlation distance, softening the sharp zone boundary. The zone shape (Figure 2) is similar to that suggested by McCann (1998).It also serves as a proxy for increased seismicity in southwestern Puerto Rico that is seen in some maps of small earthquakes (e.g., LaForge and McCann 2005, Puerto Rico Seismic Net- work at http://redsismica.uprm.edu), but not reflected in our catalog. Hazard is computed for the distance range 0–200 km and Mw range 6.5–7.5. For grid cells located within 10 kilometers of the trace of the South Lajas fault, the maximum magnitude is reduced from Mw 7.5 to 7.0 to avoid overlapping hazard contributions. The model predicts the oc- currence of one Mw-6.5-or-larger earthquake in the zone about every 50 years; our catalog contains two Mw-6.5-or-larger earthquakes and nine Mw-6.0-or-larger earthquakes since 1860, including the 1918 event. Ground motions are computed with the same group of five relations used for shallow seismicity (above), using coefficients for strike-slip faulting, and adding a relation from Spudich et al. (1999) for extensional tectonic regimes. MA2003 receives 10% weight, and each of the other five 18% in the ground-motion logic tree. The east-northeast-striking Anegada Passage, offshore from southeastern Puerto Rico, coincides with a series of fault-bounded basins and a zone of tectonic transtension where displacement from the eastern end of the Muertos Trough is thought to be trans- ferred to the Puerto Rico Trench (e.g., Jansma et al. 2000). GPS data show negligible motion near Virgin Gorda, but up to 2mm/yrof extension farther west between sites in eastern Puerto Rico and St. Croix. This is consistent with active extension along mapped northeast-trending structures in the Whiting and Virgin Island Basins and with regional seismicity patterns (Jansma and Mattioli 2005). We assign 1mm/yrof east-west extension to the Anegada Passage zone. The zone shape (Figure 2) follows McCann (1993), as extended northeastward beyond the Virgin Islands (McCann 2003b). The model predicts the occur- rence of one Mw-6.5-or-larger earthquake in the zone about every 300 years; historically, ϳ the 1867 Mw 7.3 earthquake occurred in the Anegada Passage. Other details of the hazard analysis are similar to those described above for the Mona Passage extensional zone. 176 MUELLER ET AL.

Table 1. Septentrional fault scenarios

Mom. mag Slip rate (mm/yr) Implied recurrence (yr) Comment

7.3 9 70 too frequent 7.3 5 130 too frequent 7.3 2 310 OK? 8.0 9 770 OK 8.0 5 1400 OK 8.0 2 3600 OK

EASTERN SEPTENTRIONAL FAULT AND BOWIN FAULT ZONE The Septentrional fault system extends east-southeastward, subparallel to the Puerto Rico Trench, from northern Hispaniola toward Puerto Rico (Figure 2). This transitional feature accommodates an increasing partition of left-lateral strike-slip motion with increasing distance west of PRVI, until it becomes a major plate boundary structure in central Hispaniola (Prentice et al. 1993, 2003) with upwards of 9mm/yrof slip. A major concern in assessing the seismic hazard of Puerto Rico is estimating the importance of the eastern reach of the Septentrional fault and its possible eastward extension across Mona Can- yon. Deformation at the longitude of Puerto Rico may be taken up along other strike-slip structures that lie farther north, closer to the Puerto Rico Trench, such as the Northeastern Hispaniola Slope fault and the Bunce fault zone (Grindlay et al. 1997, Dolan et al. 1998, LaForge and McCann 2005, ten Brink 2003, ten Brink et al. 2004), structures that do not pose a significant ground-motion threat to Puerto Rico. Unfortunately, it is difficult to esti- mate the relative importance of these offshore structures directly. The eastern reach of the Septentrional fault apparently intersects Mona Canyon near 67.5 degrees west longitude, but it does not seem to offset the canyon, and its possible extension eastward is unknown. The 350-km total length of the fault correlates with Mw 8.0 using the relations of Wells and Coppersmith (1994). Prentice et al. (2003) find a slip rate of 9mm/yrin central Hispaniola from paleoseismic analysis. LaForge and McCann (2005) ar- gue, however, that the slip rate and maximum magnitude must decrease along the eastern reach of the fault, and use Mw 7.3 and a slip rate of 2mm/yrin their model. There are no direct estimates of the slip rate, and about the only constraint on the parameters required for the seismic hazard calculation is the observation that there have been no major earthquakes associated with the eastern Septentrional fault historically. In Table 1 we consider several “end member” rupture scenarios and use this weak constraint to test their plausibility. Re- currence times implied by the two high-slip-rate, Mw-7.3 scenarios are too short; the surviv- ing four scenarios are combined as characteristic, floating ruptures in a logic tree with equal 0.25 weights. Ground motions are computed with the same group of five relations used (above) for shallow seismicity, using coefficients for strike-slip faulting. Estimates of the slip rate on the Bowin fault zone range from 9mm/yr(as the eastward extension of the Septentrional fault) to 0mm/yr(the transverse component of deformation is fully accommodated along structures farther to the north). LaForge and McCann (2005) assign it a slip rate of 1mm/yrin their model but note that it may be inactive. The NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 177

160-km total length of the fault correlates with Mw 7.6, using the relations of Wells and Coppersmith (1994). Following LaForge and McCann (2005), we assign a slip rate of 1mm/yr(corresponding to a recurrence time of 3,900 years) and use the same suite of ﬁve ground-motion relations that were used for the Septentrional fault. The sensitivity of estimated hazard in Puerto Rico to the choice of slip rate on the Bowin fault zone is discussed below.

SOUTH LAJAS FAULT Prentice et al. (2000) studied the east-west-striking South Lajas fault in southwestern Puerto Rico, ﬁnding evidence for two surface-rupturing earthquakes in the last 7,500 years. The 50-km total length of the fault (LaForge and McCann 2005) correlates with Mw 7.0 using the relations of Wells and Coppersmith (1994). We include the fault in the hazard model as a characteristic source with Mw 7.0 and a recurrence time of 3,500 years, using the same suite of ﬁve ground-motion relations that were used for the Septentrional fault.

RESULTS AND DISCUSSION Maps of probabilistic PGA, 0.2-second spectral acceleration (5% damping) and 1.0-second spectral acceleration (5% damping), including the contributions from all modeled sources, for an exceedance probability of 2% in 50 years (about 2,500-year return time), are shown in Figure 3. Corresponding maps for an exceedance probability of 10% in 50 years (about 500-year return time) are shown in Figure 4. (Results for 0.3-second spectral acceleration are not shown; spatial patterns are similar to the 0.2-second maps, with 0.3-second hazard about 15% smaller for both 2%-in-50-yr and 10%-in-50-yr exceedance probabilities.) The ranges of probabilistic ground motions for PRVI are roughly comparable to those estimated for the Basin and Range province of the western United States. LaForge and McCann (2005) conducted a seismic hazard analysis for sites near the four corners of the island of Puerto Rico. For western Puerto Rico our PGA values agree with theirs to within 5%–10%. For eastern Puerto Rico our PGA values are systemati- cally 10%–15% greater than theirs. Clusters of seismicity near and north of the Virgin Islands (Figure 2) are strong contributors to the hazard in eastern Puerto Rico and the Virgin Islands in our model (below), whereas background seismicity is distributed into a uniform areal zone in the LaForge and McCann approach. The sensitivity of our results to the choice of slip rate on the Bowin fault zone is explored in Figure 5. The maps show probabilistic PGA from all modeled sources, com- paring 9-mm/yr (440-year recurrence) and 1-mm/yr (3,900-year recurrence, used in the final model) characteristic scenarios. Near the fault (upper left-hand corner of each map) there are significant PGA differences, but closer sources dominate the hazard in Puerto Rico, and differences there are negligible (note the 50%-g contour that cuts across the northwest corner of the island). In another sensitivity test, we compare maps (not shown) using different weights for the preliminary MA2003 attenuation relation. For 0.2 seconds, ground motions with 50% weight are significantly greater than motions with 10% weight (used in the final model), as 178 MUELLER ET AL.

120 (%g) 80 70 60 18.5 50 55 40 50 45 40 35 18 30 25 20 15 17.5 10 5 PGA 0

120 (%g) 80 70 60 18.5 55

40 50 45 40 35 18 30 25 30 20 15 17.5 10 5 1.0-s SA 0

300 (%g) 250 200 150 18.5 120 100 90 10 80 0 70 18 60 50 40 80 30 17.5 100 20 10 0.2-s SA 0

-67 -66.5 -66 -65.5 -65 -64.5

Figure 3. Probabilistic ground motions with 2% probability of exceedance in 50 years (about 2,500-year return period). NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 179

120 (%g) 80 70 60 18.5 55 50 45 40 35 18 30 20 25 20 15 17.5 10 5 PGA 0

120 (%g) 80 70 60 18.5 55 50

20 45 40 35 18 30 25 15 20 15 17.5 10 5 1.0-s SA 0

300 (%g) 250 200 150 18.5 120

60 100 90 80 70 18 60 50 40 40 30 17.5 20 10 0.2-s SA 0

-67 -66.5 -66 -65.5 -65 -64.5

Figure 4. Probabilistic ground motions with 10% probability of exceedance in 50 years (about 500-year return period). 180 MUELLER ET AL.

120 (%g) 80 70 60 18.5 50 40 55 50 45 40 35 18 30 25 20 15 10 17.5 5 PGA, 1 mm/yr BowinFZ 0

120 (%g) 80 70 60 18.5 50 40 55 50 45 40 35 18 30 25 20 15 10 17.5 5 PGA, 9 mm/yr BowinFZ 0

-67 -66.5 -66 -65.5 -65 -64.5

Figure 5. Compare 1-mm/yr (used in final model) and 9-mm/yr rate scenarios for Bowin fault zone (3,900-yr and 440-yr recurrence, respectively) for PGA with 2% probability of exceedance in 50 years (about 2,500-year return period). The only significant differences lie in the northwest corner of the map, near the Bowin FZ (see Figure 2). much as 60% greater near the southwest corner of Puerto Rico where hazard is dominated by the Mona Passage extensional zone and South Lajas fault sources, both of which use MA2003. Although we might expect less attenuation in PRVI than in California, concern over the large high-frequency motions accounts for the smaller weight given to MA2003 in the final model—pending further evaluation of the revised relation (Motazedian and Atkin- son 2005). 1.0-second ground-motion differences are small. NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 181

1 seismicity d<50 San Juan seismicity d>=50 subdctn pr seg 0.1 subdctn hsp seg mona extnl zn anegada extnl zn septentrional flt 0.01 south lajas flt bowin flt all sources

0.001 exceedance / yr

0.0001

1e-05 1 Mayaguez

0.1

0.01

0.001 exceedance / yr

0.0001

1e-05 1 Charlotte Amalie

0.1

0.01

0.001 exceedance / yr

0.0001

1e-05 0.005 0.01 0.02 0.05 0.1 0.2 0.5 1 2 5 PGA (g)

Figure 6. PGA hazard curves for San Juan, Mayaguez, and Charlotte Amalie. Horizontal lines correspond to exceedance probabilities of 10% in 50 years (about 0.0021 annual or 500-year return time) and 2% in 50 years (about 0.0004 annual or 2,500-year return time). 182 MUELLER ET AL.

Figure 6 shows hazard curves with contributions from individual sources for three sites. At San Juan, in the ground-motion range between 2,500- and 500-year return times (about 0.0004 and 0.002 annual exceedance rates, respectively), shallow seismicity, deep seismicity, and Puerto Rico-segment subduction account, about equally, for essentially all of the hazard. At Mayaguez (and also at Ponce, not shown), seismicity and the Mona Passage extensional zone are main contributors at 500-year return time, with the Mona Passage source increasingly dominant at longer return times. At Charlotte Amalie (St. Thomas), shallow and deep seismicity dominate the hazard about equally, with much smaller contributions from the Puerto Rico-segment subduction and the Anegada Passage extensional zone. Fi- nally, we note that near Pta. Aguila in southwestern Puerto Rico (results not shown) the contribution from the South Lajas fault equals that of the Mona Passage extensional zone for 2,500-year return time and dominates the hazard for longer return times, demonstrating the importance of targeting other potentially active onshore faults for future research.

ACKNOWLEDGMENTS We gratefully acknowledge discussions with Roland LaForge, Bill McCann, Carol Prentice, Christa von Hillebrandt, Gail Atkinson, Dariush Motazedian, Steve Harmsen, and C. B. Crouse. LaForge and von Hillebrandt reviewed the paper. We received valuable input from participants of the Workshop on Seismic and Tsunami Hazards of Puerto Rico held in San Juan in March 1999, and feedback on a preliminary version of the hazard model from members of the TS-1 subcommittee of the Building Seismic Safety Commission chaired by C. B. Crouse. The ﬁrst author gratefully acknowledges the en- couragement and forbearance of Associate Editor Ivan Wong. The illustrations in this paper were produced using the Generic Mapping Tools (GMT) software (Wessel and Smith 1991).

REFERENCES Abrahamson, N. A., and Silva, W. J., 1997. Empirical response spectral attenuation relations for shallow crustal earthquakes, Seismol. Res. Lett. 68, 94–127. American Society of Civil Engineers (ASCE), 2005. Minimum Design Loads for Buildings and Other Structures, ASCE 7-05, Reston, Virginia. Atkinson, G. M., and Boore, D. M., 2003. Empirical ground-motion relations for subduction zone earthquakes and their applications to Cascadia and other regions, Bull. Seismol. Soc. Am. 93, 1703–1729. Boore, D. M., Joyner, W. B., and Fumal, T. E., 1997. Equations for estimating horizontal response spectra and peak acceleration from western North American earthquakes: a summary of recent work, Seismol. Res. Lett. 68, 128–153. Building Seismic Safety Council (BSSC), 2004. NEHRP Recommended Provisions for Seismic Regulations for New Buildings and Other Structures, 2003 Edition, Provisions and Com- mentary, prepared by the Building Seismic Safety Council for the Federal Emergency Man- agement Agency, FEMA Publication No. 450, Washington D.C. Campbell, K. W., and Bozorgnia, Y., 2003. Updated near-source ground motion (attenuation) relations for the horizontal and vertical components of peak ground acceleration and acceleration response spectra, Bull. Seismol. Soc. Am. 93, 314–331. NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 183

Dolan, J. F., Mullins, H. T., and Wald, D. J., 1998. Active tectonics of the north-central Carib- bean: Oblique collision, strain partitioning, and opposing subducted slabs, in Dolan, J. F., and Mann, P., eds., Active Strike-Slip and Collisional Tectonics of the Northern Caribbean Plate Boundary Zone, Geological Society of America Special Paper 326, Boulder Colorado, 1–61. Dolan, J. F., and Wald, D. J., 1998. The 1943–1953 north-central Caribbean earthquakes: Ac- tive tectonic setting, seismic hazards, and implications for Caribbean—North America plate motions, in Dolan, J. F., and Mann, P., eds., Active Strike-Slip and Collisional Tectonics of the Northern Caribbean Plate Boundary Zone, Geological Society of America Special Paper 326, Boulder Colorado, 143–169. Doser, D. I., Rodriguez, C. M., and Florez, C., 2005. Historical earthquakes of the Puerto Rico—Virgin Islands region (1915–1963), in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological Society of America Special Paper 385, 103–114. Engdahl, E. R., and Villasenor, A., 2002. Global seismicity—1900–1999, in Lee, W. H. K., Kanamori, H., Jennings, P. C., and Kisslinger, C., eds., International Handbook of Earth- quake and Engineering Seismology, International Association of Seismology and Physics of the Earth’s Interior (IASPEI), 665–690. Frankel, A., McCann, W. R., and Murphy, A. J., 1980. Observations from a seismic network in the Virgin Islands region: tectonic structures and earthquake swarms, Bull. Seismol. Soc. Am. 85, 2669–2678. Frankel, A., Mueller, C., Barnhard, T., Perkins, D., Leyendecker, E., Dickman, N., Hanson, S., and Hopper, M., 1996. National Seismic Hazard Maps: Documentation June 1996, USGS Open-File Report 96–532,70pp. Frankel, A. D., Mueller, C. S., Barnhard, T. P., Leyendecker, E. V., Wesson, R. L., Harmsen, S. C., Klein, F. W., Perkins, D. M., Dickman, N. C., Hanson, S. L., and Hopper, M. G., 2000. USGS National Seismic Hazard Maps, Earthquake Spectra 16, 1–19. Frankel, A. D., Petersen, M. D., Mueller, C. S., Haller, K. M., Wheeler, R. L., Leyendecker, E. V., Wesson, R. L., Harmsen, S. C., Cramer, C. H., Perkins, D. M., and Rukstales, K. S., 2002. Documentation for the 2002 Update of the National Seismic Hazard Maps, USGS Open-File Report 02–420,33pp. Gardner, J. K., and Knopoff, L., 1974. Is the sequence of earthquakes in southern California, with aftershocks removed, Poissonian? Bull. Seismol. Soc. Am. 64, 1363–1367. Grindlay, N., Mann, P., and Dolan, J., 1997. Researchers investigate submarine faults north of Puerto Rico, EOS Trans. Am. Geophys. Union 78, 404. Grindlay, N. R., Mann, P., Dolan, J. F., and van Gestel, J-P., 2005. Neotectonics and subsidence of the northern Puerto Rico—Virgin Islands margin in response to the oblique subduction of high-standing ridges, in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological Society of America Special Paper 385, 31–60. International Code Council (ICC), 2006a. International Building Code, 2006 Edition, Interna- tional Code Council, Inc., Country Club Hills, Illinois. ––—, 2006b. International Residential Code, 2006 Edition, International Code Council, Inc., Country Club Hills, Illinois. 184 MUELLER ET AL.

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 6, 1021–1037. Jansma, P. E., and Mattioli, G. S., 2005. GPS results from Puerto Rico and the Virgin Islands: Constraints on tectonic setting and rates of active faulting, in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological So- ciety of America Special Paper 385, 13–30. LaForge, R. C., and McCann, W. R., 2005. A seismic source model for Puerto Rico, in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological Society of America Special Paper 385, 223–248. Leyendecker, E. V., Hunt, R. J., Frankel, A. D., and Rukstales, K. S., 2000. Development of maximum considered earthquake ground motion maps, Earthquake Spectra 16, 21–40. Mann, P., 2005. Introduction, in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological Society of America Special Paper 385, 1–12. Masson, D. G., and Scanlon, K. M., 1991. The neotectonic setting of Puerto Rico, Geol. Soc. Am. Bull. 103, 144–154. McCann, W. R., 1985. On the earthquake hazards of Puerto Rico and the Virgin Islands, Bull. Seismol. Soc. Am. 75, 251–262. ––—, 1993. Seismic Hazard Map for Puerto Rico, Report to The Seismic Safety Commission of Puerto Rico,60pp. ––—, 1998. Tsunami Hazard of Western Puerto Rico from Local Sources: Characteristics of Tsunamigenic Faults, Technical Report to the Department of Marine Sciences, University of Puerto Rico. ––—, 2003a. Catalog of felt earthquakes for Puerto Rico and neighboring islands 1492–1899 with additional information for some 20th-century earthquakes, Seismol. Res. Lett. 74, 199. ––—, 2003b. Personal communication. McCann, W. R., Joyce, J., and Lithgow, C., 1987. The Puerto Rico platelet at the northeastern edge of the Caribbean plate, EOS Trans. Am. Geophys. Union 68, 1483. Mercado, A., and McCann, W. R., 1998. Numerical simulation of the 1918 Puerto Rico tsunami, Natural Hazards 18, 57–76. Motazedian, D., 2003. Personal communication. Motazedian, D., and Atkinson, G., 2005. Ground-motion relations for Puerto Rico, in Mann, P., ed., Active Tectonics and Seismic Hazards of Puerto Rico, the Virgin Islands, and Offshore Areas, Geological Society of America Special Paper 385, 61–80. Mueller, C., Hopper, M., and Frankel, A., 1997. Preparation of Earthquake Catalogs for the National seismic Hazard Maps: Contiguous 48 States, USGS Open-File Report 97–464,36 pp. Pacheco, J. F., and Sykes, L. R., 1992. Seismic moment catalog of large shallow earthquakes 1900 to 1989, Bull. Seismol. Soc. Am. 82, 1306–1349. Petersen, M. D., Frankel, A. D., Harmsen, S. C., Mueller, C. S., Haller, K. M., Wheeler, R. L., Wesson, R. L., Zeng, Y., Boyd, O. S., Perkins, D. M., Luco, N., Field, E. H., Wills, C. J., and Rukstales, K. S., 2008. Documentation for the 2008 Update of the United States National Seismic Hazard Maps, USGS Open-File Report 2008-1128, 60 pp. plus appendixes. NEW SEISMIC HAZARD MAPS FOR PUERTO RICO AND THE U.S.VIRGIN ISLANDS 185

Prentice, C. S., Mann, P., Taylor, F. W., Burr, G., and Valastro, S., 1993. Paleoseismicity of the North American—Caribbean plate boundary (Septentrional fault), Dominican Republic, Ge- ology 21, 49–52. Prentice, C. S., Mann, P., and Burr, G., 2000. Prehistoric earthquakes associated with a Late Quaternary fault in the Lajas Valley, southwestern Puerto Rico, EOS Trans. Am. Geophys. Union 81, F1182 (abstract). Prentice, C. S., Mann, P., Pena, L. R., and Burr, G., 2003. Slip rate and earthquake recurrence along the central Septentrional fault, North American—Caribbean plate boundary, Domini- can Republic, J. Geophys. Res. 108(B3), 2149, doi:10.1029/2001JB000442, 17 pp. Sadigh, K., Chang, C.-Y., Egan, J. A., Makdisi, F., and Youngs, R. R., 1997. Attenuation relationships for shallow crustal earthquakes based on California strong motion data, Seismol. Res. Lett. 68, 180–189. Spudich, P., Joyner, W. B., Lindh, A. G., Boore, D. M., Margaris, B. M., and Fletcher, J. B., 1999. SEA99: A revised ground motion prediction relation for use in extentional tectonic regimes, Bull. Seismol. Soc. Am. 89, 1156–1170. ten Brink, U., 2003. Personal communication. ten Brink, U., Danforth, W., Polloni, C., Andrews, B., Llanes, P., Smith, S., Parker, E., and Uo- zumi, T., 2004. New seaﬂoor map of the Puerto Rico Trench helps assess earthquake and tsunami hazards, EOS Trans. Am. Geophys. Union 85 349–354. Weichert, D. H., 1980. Estimation of earthquake recurrence parameters for unequal observation periods for different magnitudes, Bull. Seismol. Soc. Am. 70, 1337–1356. Wells, D. L., and Coppersmith, K. J., 1994. New empirical relationships among magnitude, rupture length, rupture width, rupture area, and surface displacement, Bull. Seismol. Soc. Am. 84, 974–1002. Wessel, P., and Smith, W. H. F., 1991. Free software helps map and display data, EOS Trans. Am. Geophys. Union 72, 441, 445–446. Youngs, R. R., Chiou, S.-J., Silva, W. J., and Humphrey, J. R., 1997. Strong ground motion attenuation relationships for subduction zone earthquakes, Seismol. Res. Lett. 68, 58–73. (Received 21 April 2009; accepted 5 May 2009͒