Portal de Periódicos da CAPES

Seismogenic characteristics of the Northern Mariana shallow thrust zone from local array data

2011; Wiley; Volume: 12; Issue: 12 Linguagem: Inglês

10.1029/2011gc003853

1525-2027

Erica Emry, Douglas A. Wiens, Hajime Shiobara, Hiroko Sugioka,

Geological and Geochemical Analysis

Geochemistry, Geophysics, GeosystemsVolume 12, Issue 12 Free Access Seismogenic characteristics of the Northern Mariana shallow thrust zone from local array data Erica L. Emry, Erica L. Emry [email protected] Department of Earth and Planetary Sciences, Washington University in St. Louis, Campus Box 1169, One Brookings Drive, St. Louis, Missouri 63130-4899, USASearch for more papers by this authorDouglas A. Wiens, Douglas A. Wiens Department of Earth and Planetary Sciences, Washington University in St. Louis, Campus Box 1169, One Brookings Drive, St. Louis, Missouri 63130-4899, USASearch for more papers by this authorHajime Shiobara, Hajime Shiobara Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, JapanSearch for more papers by this authorHiroko Sugioka, Hiroko Sugioka Institute for Frontier Research on Earth Evolution, JAMSTEC, 2-15 Natsushima-Cho, Yokosuka-city, Yokosuka 237-0061, JapanSearch for more papers by this author Erica L. Emry, Erica L. Emry [email protected] Department of Earth and Planetary Sciences, Washington University in St. Louis, Campus Box 1169, One Brookings Drive, St. Louis, Missouri 63130-4899, USASearch for more papers by this authorDouglas A. Wiens, Douglas A. Wiens Department of Earth and Planetary Sciences, Washington University in St. Louis, Campus Box 1169, One Brookings Drive, St. Louis, Missouri 63130-4899, USASearch for more papers by this authorHajime Shiobara, Hajime Shiobara Earthquake Research Institute, University of Tokyo, 1-1-1 Yayoi, Bunkyo-ku, Tokyo 113-0032, JapanSearch for more papers by this authorHiroko Sugioka, Hiroko Sugioka Institute for Frontier Research on Earth Evolution, JAMSTEC, 2-15 Natsushima-Cho, Yokosuka-city, Yokosuka 237-0061, JapanSearch for more papers by this author First published: 09 December 2011 https://doi.org/10.1029/2011GC003853Citations: 14AboutSectionsPDF ToolsRequest permissionExport citationAdd to favoritesTrack citation ShareShare Give accessShare full text accessShare full-text accessPlease review our Terms and Conditions of Use and check box below to share full-text version of article.I have read and accept the Wiley Online Library Terms and Conditions of UseShareable LinkUse the link below to share a full-text version of this article with your friends and colleagues. Learn more.Copy URL Share a linkShare onFacebookTwitterLinkedInRedditWechat Abstract [1] The Northern Mariana seismogenic zone has no shallow thrust earthquakes larger than Ms 7.4 in the historical seismological record and is traditionally considered 'decoupled' or 'aseismic'. During the 2003–2004 Mariana Subduction Factory Imaging Experiment, we recorded local shallow earthquakes throughout the central and northern regions of the Mariana forearc using an array of terrestrial broadband and ocean bottom seismographs. Accurate locations for both the 2003–2004 local seismicity as well as earthquakes with Global Centroid Moment Tensor (GCMT) solutions from 1976 to 2008 were obtained using the hypocentroidal decomposition relocation method and a local velocity model. Additionally, focal mechanisms for the largest 2003–2004 earthquakes were determined using regional waveform inversion. Thrust faulting earthquakes occur along the Mariana megathrust between depths of 20–60 km, showing that the lack of great shallow thrust earthquakes does not result from a narrow seismogenic zone and that most seismicity occurs where the downgoing plate contacts the overriding mantle wedge. Clusters of small plate interface earthquakes with Ml 1.6–4.7 occur within patches 100–120 km west of the trench at depths of 30–45 km. Furthermore, the larger GCMT earthquakes (Mw 4.9–5.8) occur mostly updip and downdip of the patches of smaller earthquakes recorded by our local array and is suggestive of changes in the fault properties with depth. Clusters of small, forearc earthquakes occur discontinuously along the length of the Mariana subduction zone, showing that Northern Mariana is variable both along the strike of the margin and with depth along the seismogenic zone. We propose that the lack of great (Mw > 8) thrust faulting earthquakes is due in part to the variable frictional heterogeneity along the megathrust. Key Points Plate interface seismicity in Northern Marianas extends to depths of 60 km Seismicity is variable along strike and along dip of the seismogenic zone Aseismicity of Mariana subduction zone is due to variable fault strength 1. Introduction [2] The Mariana subduction zone is commonly considered to be the aseismic end-member on a spectrum of subduction zones, with the opposite end represented by the Chilean and Alaskan margins, where megathrust earthquakes approaching Mw 9.5 are feasible [Kanamori, 1977; Uyeda and Kanamori, 1979]. To date, no great (Mw > 8.0) shallow thrust earthquakes have been recorded and accurately located in the Mariana subduction zone, although historical records do contain evidence for infrequent moderate-sized (Mw > 7.0) shallow events and large earthquakes for which magnitude estimates are absent or unreliable (Table 1). In a global study comparing seismic slip coefficients, the Mariana Islands was determined to have a seismic coupling coefficient of 0.002, meaning that only 0.2% of the slip between the Pacific and Philippine plates could be accounted for by historical records of large thrust earthquakes [Pacheco et al., 1993]. The results imply that either a large percentage of interplate slip is accommodated through stable, aseismic slip or that the subduction zone is due for a giant earthquake every hundred years. Table 1. Large Shallow Earthquakes (M ≥ 7 or Guam MMI ≥ 8) in the Mariana Forearc 1825–2011 Event Date Time (UTC) Latitude (°N) Longitude (°E) Depth (km) Type of Slip Magnitude and/or Intensity 1 Apr 1825 n/a n/aa n/aa n/a n/a 8 MMIb 2 May 1834 n/a n/aa n/aa n/a n/a 8 MMIb 3 25 Jan 1849 14:56b,c n/aa n/aa n/a n/a M 7.5d/9 MMIb 4 16 May 1892 21:10b,c n/aa n/aa n/a n/a M 7.5d/8 MMIb 5 22 Sept 1902 1:46:30e 18.0e 146.0e n/a n/a Ms 7.4f/9 MMIb 6 23 Mar 1913 20:47.3 24g 142g 80g n/a mb 7.1h 7 24 Oct 1930 20:15:11 18.5g 147g 35g n/a Ms 7.0h 8 28 Jan 1931 21:24:03 11g 144.75g 35g n/a Ms 7.1h 9 24 Feb 1934 6:23:40 22.5g 144g 35g n/a Ms 7.1f 10 17 Jan 1940 01:15:00 17g 148g 80g n/a mb 7.3i 11 28 Dec 1940 16:37:44 18e 147.5g 80g n/a mb 7.3i 12 14 Jun 1942 03:09:45 15g 145g 80g n/a mb 7.0h 13 25 May 1950 18:35:07 13g 143.5g 90g n/a mb 7.0h 14j 8 Aug 1993 8:34:49.3 13.06 145.31 59.3 Thrust Mw 7.7 15j 12 Oct 2001 15:02:23.3 12.88 145.08 42.0 Thrust Mw 7.0 16j 26 Apr 2002 16:06:13.9 13.15 144.67 69.1 Thrust Mw 7.0 a Strong shaking felt on island of Guam, earthquake locations unknown. b Maso [1910]. c Indicates Local Guam Time. d Soloviev and Go [1974]. e Gutenberg [1956]. f Pacheco and Sykes [1992]. g Gutenberg and Richter [1954]. h Abe [1981]. i Abe and Kanamori [1979]. j Global GCMT catalog, Dziewonski et al. [1981]. [3] Recent devastating magnitude nine earthquakes in regions previously thought to have little potential for great earthquakes have caused reassessment of such seismic/aseismic classifications. Sumatra, for example, had a coupling coefficient of 0.007 determined by Pacheco et al. [1993], yet produced a Mw ∼9.1 earthquake in 2004 [e.g., Lay et al., 2005]. The recent 2011 Northeast Japan earthquake also occurred in a region where magnitude nine earthquakes were thought to be impossible. Clearly a much better understanding of the factors controlling the seismic characteristics of the subduction zone thrust interface is needed. [4] Previous studies present two main hypotheses to explain the lack of large earthquakes in the Mariana and other 'aseismic' subduction zones. The first suggests that the shallow intersection of the plate interface with a serpentinized mantle wedge narrows the seismogenic width so that large earthquakes are not possible [Hyndman et al., 1997; Peacock and Hyndman, 1999; Hyndman, 2007]. This idea proposes that seismic slip is limited to where the underthrusting plate contacts the overriding forearc crust, which occurs in the Central and Northern Mariana Islands at a depth of about 15 km [Takahashi et al., 2007]. The second hypothesis for lack of large earthquakes in Mariana is that the plate interface is very weakly coupled due to the geometry of the subduction zone and predominance of horizontal tensional tectonic stresses in the region, as evidenced by seafloor spreading in the back arc [Scholz and Campos, 1995]. [5] In the absence of large underthrusting earthquakes, the depth extent of the seismogenic zone may be identified through use of microseismicity or by geodesy [Schwartz and DeShon, 2007]. Seismic studies are the most feasible option in an island arc setting, given that the geodetic signal of plate coupling and strain accumulation occurs offshore, where geodetic studies require extremely expensive ocean-bottom GPS technology. Through seismic studies, we can examine the pattern of seismic release along the thrust zone and obtain a rough estimate of how much surface area could rupture seismically if it were all to slip at once. [6] In this paper we use shallow earthquake locations and focal mechanisms for small earthquakes recorded by a temporary local array of land and ocean bottom seismographs deployed during 2003–2004 to better understand the Northern Mariana shallow thrust region [Pozgay et al., 2007]. Previously, the Northern Mariana subduction zone had only been studied using larger earthquakes detectable teleseismically, since there are few permanent seismic stations in this region. The local recordings provide for study of much smaller earthquakes with much greater location precision, allowing us to answer basic questions about the characteristics of this unusual subduction zone. 2. Background Geological Setting 2.1.1. Tectonic Setting and History [7] The Mariana-Izu-Bonin system is a young subduction zone extending from Japan in the north to Guam in the south that first formed about 43 Ma [Stern et al., 2003]. The Mariana Islands are the southern portion of the island chain, where a strong, outward curvature of the arc separates it from the mostly North-South Izu-Bonin section. Throughout the history of the Mariana Islands, the volcanic arc has split twice – the remnant arcs from these rifting events constitute the Kyushu-Palau ridge and the West Mariana Ridge [Stern et al., 2003]. Evidence from paleomagnetism suggests the orientation of the Izu-Bonin-Mariana system was originally East-West and gradually rotated clockwise to its current North-South orientation [Hall et al., 1995; Hall, 2002]. The convergence rate between the Pacific plate and the forearc of the Mariana subduction zone from GPS measurements at stations along the island arc and forearc rise show that the arc and forearc are moving east relative to the rest of the Philippine Sea plate, due to active extension in the back-arc basin [Kato et al., 2003]. The rate of subduction beneath the northern part of the Mariana Islands near Agrihan is 35–45 mm/yr, while the rate of subduction in the south near Guam is 60–70 mm/yr [Kato et al., 2003]. The angle of convergence of the Pacific plate beneath the Mariana forearc is 83° West of North (Figure 1) [Kato et al., 2003]. Figure 1Open in figure viewerPowerPoint Array geometry for the 2003–2004 Mariana Subduction Factory Imaging Experiment. (top) Complete array of stations along with bathymetry. (bottom left) Stations that returned good data for the period before August 2003, prior to failure of the MPL4n OBS. (bottom right) Station geometry following August 2003. Stations are indicated in both plots by colored triangles (dark blue: Guralp 40T; cyan: STS2; red: Japanese PMD OBS; dark gray: MPL4o OBS; and light gray: MPL4n OBS). Pacific plate convergence beneath the Mariana Forearc is shown by thick black arrows; rate of convergence as determined by Kato et al. [2003]is noted above each arrow. Thick red lines show the location of the back-arc spreading center. Inset map in Figure 1 (top) shows all subduction trenches (blue lines), spreading centers (red lines) and transform boundaries (green lines) in the vicinity of the Mariana subduction zone and the Philippine Sea. 2.1.2. Forearc Morphology [8] The western portion of the forearc in our study region is flat, covered in volcaniclastic and pelagic sediments, and spans about 2/3 of the forearc seafloor. The region is cut by normal faults that run roughly parallel to the trend of the volcanic arc and the trench, the presence of which indicate that the Mariana forearc is under tension [Stern and Smoot, 1998]. The eastern portion of the forearc is cut by numerous, small normal faults in a mostly southwest to northeast orientation [Stern and Smoot, 1998]. In this highly deformed part of the forearc, a number of large serpentinite seamounts, unique to the Izu-Bonin-Mariana subduction zone are present [Stern et al., 2003]. The serpentinite seamounts in the Mariana forearc are located at 15–90 km distance from the trench and are formed through serpenitite mud volcanism. The minerals ejected reflect increasing pressure as distance from the trench increases, suggesting that materials are being ejected from progressively deeper depths [Fryer and Salisbury, 2006]. The Big Blue seamount is located at 70 km to the west of the trench [Oakley et al., 2007] and is the largest seamount located on the Mariana forearc. The tectonic instability of the Mariana forearc is responsible for the presence of the active serpentinite seamounts, as fluids expelled during subduction are able to move upward through the extensively faulted forearc and erupt at the surface [Fryer et al., 1999; Stern et al., 2003; Fryer and Salisbury, 2006]. [9] The presence of serpentinite mud volcanoes is frequently used as evidence that the underlying Mariana mantle wedge is serpentinized. Geochemical work from Benton et al. [2004] found that expelled fluids were not derived directly from the slab, but rather, had interacted with mantle wedge materials prior to being emitted from the seamount. Receiver functions calculated beneath Saipan and Tinian Islands further to the south reveal a pronounced low velocity layer in the mantle wedge at 40–55 km depth, leading to the conclusion that mantle serpentinization is prevalent at these depths [Tibi et al., 2008]. Pozgay et al. [2009] suggest mantle serpentinization in the forearc beneath the seamounts as an explanation for an observed zone of high attenuation. Results from surface wave phase velocities in the Northern Mariana Islands also showed a low velocity anomaly in the forearc between Celestial and Big Blue Seamounts [Pyle et al., 2010], postulated to be mantle serpentinization. Similarly, P and S velocity tomography by Barklage [2010] revealed a region of an unusually high 1.95–2.0 Vp/Vs ratio; this combined with modeling by Hacker et al. [2003] suggests that the forearc mantle wedge is ∼30–60% serpentinized. While the evidence for a serpentinized mantle wedge in Mariana is substantial, its spatial distribution is not well constrained and its relation to shallow forearc seismicity is not completely understood. Seismicity [10] The most recent, large, shallow thrust earthquakes have occurred in the Southern Mariana Islands near Guam (Figure 2 and Table 1, events 14–16) and began with a Mw 7.7 earthquake on August 8, 1993 [Campos et al., 1996]. Depth estimates for this earthquake vary from 41.5 km [Campos et al., 1996] to 74.5 km [Harada and Ishibashi, 2008], and the focal mechanism is consistent with a shallow dipping thrust earthquake [Campos et al., 1996]. In 2001 and 2002, two large Mw 7.0 earthquakes occurred nearby the location of the large 1993 earthquake. The large 1993 Guam earthquake was initially interpreted as rupture along the plate interface [Campos et al., 1996] and supported the interpretation that the Southern Mariana plate interface is more strongly coupled than the Northern plate interface [Scholz and Campos, 1995]. However, more recent studies suggest that 1993, 2001, and 2002 Guam earthquakes occurred in the subducting Pacific plate and thus do not represent seismic slip along the megathrust [Tanioka et al., 1995; Harada and Ishibashi, 2008]. Figure 2Open in figure viewerPowerPoint Locations for Events 5–16 listed in Table 1. All events are Mw or Ms ≥ 7.0, with depths less than 100 km, and are in or nearby the Mariana forearc. Epicenters of earthquakes occurring in 1902–1950 are indicated by yellow circles. Numbers within the yellow circles correspond to event numbers 5–13 listed in Table 1. Epicenters of thrust earthquakes occurring in 1951–2011 are indicated by red circles, and numbers within the red circles correspond to event numbers 14–16 in Table 1. [11] A number of potentially shallow thrust earthquakes with magnitudes greater than 7.0 occurred from 1900 to 1950. Many of these events, listed by Gutenberg and Richter [1954] as shallow or intermediate depth earthquakes, have revised magnitudes between Ms 7.0–7.4 [Abe and Kanamori, 1979; Abe, 1981; Pacheco and Sykes, 1992] and occur along the entire length of the Mariana forearc (Figure 2 and Table 1, events 6–13). Although some of these events are classified as intermediate depth earthquakes, and a few are located in the outer rise of the Pacific slab, they are included in the record due to the possibility of poor event locations in the early 1900s – up to 1° laterally and 30 km in depth for the best-located events, with earthquakes at 40–100 km depth being particularly problematic [Gutenberg and Richter, 1954]. In addition to uncertainty in earthquake locations and depths, magnitude estimates for these events have been calculated and revised numerous times [Gutenberg and Richter, 1954; Gutenberg, 1956; Richter, 1958; Abe and Kanamori, 1979; Abe, 1981; Abe and Noguchi, 1983; Pacheco and Sykes, 1992]; the most recent magnitude revisions are listed in Table 1. [12] During 1825–1892, four large earthquakes and tsunamis are known to have affected the island of Guam. Estimates for the intensity of shaking on the island of Guam as compiled by Maso [1910] are included for all earthquakes occurring in 1825–1902 (Figure 2 and Table 1, events 1–5), but earthquake location, depth, and slip are unknown for the earliest events. Large, shallow thrust earthquakes often create tsunamis; however large extensional earthquakes in the bending Pacific plate at the Mariana trench have also produced tsunamis [Satake et al., 1992; Yoshida et al., 1992]. Therefore although significant damage and records of tsunamis on Guam exist, these tsunamis may not have been generated by shallow thrust earthquakes. [13] The seismic record used by Pacheco et al. [1993] to compute seismic coupling coefficients along this margin included only two large events: 1902 Ms 7.4 occurring near 18°N, 146°E and 1934 Ms 7.1 occurring near 22.5°N, 144°E (Figure 2 and Table 1, events 5 and 10). No shallow thrust earthquakes larger than Ms 7.4 have been recorded and clearly located in the central and northern parts of the Mariana Islands during 1897 to 2010 [Gutenberg and Richter, 1954; Abe and Kanamori, 1979; Abe, 1981; Pacheco and Sykes, 1992]. Given the relationship for seismic coupling and fault parameters used by Pacheco et al. [1993], the absence of earthquakes larger than Ms 7.4 over the last ∼110 years requires that a giant earthquake the size of the great Chilean or Alaskan earthquakes (Mw > 9) occur in order to seismically release the accumulated strain (Table 2, first row). Even assuming that every earthquake listed in Table 1 is a shallow thrust earthquake, the resulting seismic coupling coefficient is 0.0076 and requires that a Mw 9.39 earthquake occur every ∼110 years in order to seismically release all accumulated strain (Table 2, second row). Although it is difficult to preclude this, most previous studies assume that the absence of earthquakes results from aseismic slip rather than an impending great megathrust earthquake [e.g., Uyeda and Kanamori, 1979]. Table 2. Parameters Affecting Seismic Slip Coefficient and Maximum Earthquake Magnitudea Margin Length, L (km) Seismogenic Width, W (km) Time, T (yrs) Convergence Rate, p (mm/yr) Cumulative Moment, Moi(N-m) Seismic Slip Coefficient, α Moment Deficit (N-m) Magnitude Deficit 1 1280 74 90 30 2.36 E20 0.0018 1.28 E23 9.34 2 1280 74 110 30 1.20 E21 0.0077 1.55 E23 9.39 3 1280 100 110 50 1.20 E21 0.0034 3.51 E23 9.63 4 560 100 110 40 4.62 E20 0.0037 1.23 E23 9.33 a Relationship used by Pacheco et al. [1993] for seismic slip coefficient: α = where p is plate convergence rate and rate of seismic slip: s = . For all calculations, rigidity (μ) is 5.0E10 . Parameters and results in the first row are those calculated by Pacheco et al. [1993] for the entire 1280 km length of the Mariana subduction zone. The second row incorporates a longer record and all shallow (depth < 100 km), M ≥ 7.0 seismicity nearby the Mariana forearc. The third row assumes a 100 km seismogenic width and recent (averaged along the margin) plate convergence rate from Kato et al. [2003]. The fourth row assumes the margin between 15 and 20°N and moment from all events in Table 1 at those latitudes. All calculations require that moment equivalent to a magnitude 9+ earthquake be released in order to compensate for the 110 years of Pacific plate convergence. The Marianas Seismogenic Zone and Aseismic Slip [14] The two proposed explanations for Mariana aseismicity represent inherently different physical processes: reduction of normal force between the plates [Scholz and Campos, 1995] or reduced frictional strength between the plates due to rheological or fault zone properties [Hyndman et al., 1997; Peacock and Hyndman, 1999; Hyndman, 2007]. In this section we review what is known about the Mariana shallow thrust zone in the context of the proposed explanations for the absence of great earthquakes. Updip Limit of Seismogenic Zone [15] Very little is known about the location of the updip limit in the Mariana seismogenic zone – previous studies of coupling have assumed a 10 km updip limit depth for all subduction zones [Pacheco et al., 1993]. The onset of seismogenesis in continental subduction zones is classically perceived to begin near the base of the accretionary wedge, due to the compaction and cementation of sediments or presence of stronger crustal materials [Byrne et al., 1988; Marone and Scholz, 1988; Byrne and Fisher, 1990; Moore and Saffer, 2001]. However, the Mariana island arc lacks an accretionary wedge. Hypotheses that the updip limit could be controlled by the phase transition of weak smectite clays to stronger illite clays [Vrolijk, 1990; Moore and Saffer, 2001] were found to not strongly affect onset of seismogenesis [Saffer and Marone, 2003]. More recently, the updip limit is thought to be controlled by decreasing pore pressure and fluid flux with depth as fluid producing diagenetic changes, such as the opal to quartz, smectite to illite, or hydrocarbon maturation cease [Oleskevich et al., 1999; Moore and Saffer, 2001; Spinelli and Saffer, 2004]. Other diagenetic and low-grade metamorphic processes, such as pressure solution with subsequent quartz cementation, and zeolite-facies metamorphism with resulting cementation, are thought to strengthen the downgoing slab sediments [Moore and Saffer, 2001]. [16] Regardless of the underlying physical cause for the updip limit, there appears to be a correlation between the 100–150°C isotherm and the onset of thrust seismicity in subduction zones [Hyndman and Wang, 1993; Oleskevich et al., 1999], although it is unclear whether the transition results directly from temperature or from other factors [Saffer and Marone, 2003]. In Costa Rica, a change in the age and temperature of the subducting seafloor correlates with a measurable offset in the location of the updip limit [Harris and Wang, 2002; Newman et al., 2002; DeShon et al., 2006; Schwartz and DeShon, 2007]. In the Mariana Islands, recent geochemical work by Hulme et al. [2010] estimates the temperature conditions beneath Big Blue Seamount to be greater than 200°C. Given this and our current understanding of the initiation of seismogenesis, the updip limit should occur east of Big Blue Seamount [Hyndman and Wang, 1993; Oleskevich et al., 1999]. Downdip Limit of Seismogenic Zone [17] The transition from unstable slip producing earthquakes to ductile deformation beyond the downdip limit of the seismogenic zone has traditionally been interpreted as due to increasing temperature [Hyndman and Wang, 1993; Tichelaar and Ruff, 1993; Hyndman et al., 1995; Hyndman et al., 1997; Harris and Wang, 2002]. In continental subduction settings, the downdip limit was suggested to correspond to the 350–400°C isotherm with a transitional region of stable slip extending to 450°C [Hyndman et al., 1995]. The downdip limit in regions such as the Mariana Islands, where the overriding plate has a thin crust and the downgoing plate contacts the forearc mantle, is suggested to correspond to higher temperatures, near 550°C [Tichelaar and Ruff, 1993]. [18] An alternate explanation suggests that the downdip limit is the boundary between overriding crust and serpentinized mantle wedge below, explained by aseismic layered serpentinite, brucite, and talc minerals within the mantle wedge [Hyndman et al., 1997; Peacock and Hyndman, 1999; Harris and Wang, 2002; Seno, 2005]. This supposition relies on laboratory experiments indicating that these materials show stable sliding behavior at seismogenic depths [Reinen et al., 1991; Moore et al., 1997; Hilairet et al., 2007; Moore and Lockner, 2007]. Earthquake producing slip depends on which serpentinite polymorph is present at that depth; antigorite, brucite and talc were found in one study by Moore and Lockner [2007]to be velocity-strengthening, while lizardite and chrysotile were velocity-weakening at experimental temperatures. Thermal modeling suggests that lizardite may be the dominant phase at shallow depths in the Mariana mantle wedge [Wada and Wang, 2009]. Variability of the Plate Interface Seismogenic Zone [19] Some studies indicate variability in the sizes and characteristics of rupture with depth along the seismogenic width of subduction zones [e.g., Hyndman et al., 1997; Bilek and Lay, 2000]. The subduction zones of Kermadec, Solomon, and Kamchatka exhibit a bimodal depth distribution of shallow thrust earthquakes [Pacheco et al., 1993; Hyndman et al., 1997], which has been explained by serpentinization at the shallowest mantle depths [Hyndman et al., 1997]. In this model, the subducting plate slides aseismically while in contact with the serpentinized part of the mantle wedge but transitions back to stick-slip behavior deeper in the mantle wedge, where serpentinites are no longer stable and where the plate contact is not yet in the ductile deformation regime. In the Mariana Islands, serpentinites were similarly used to explain the seemingly narrow seismogenic width, although no second, deep seismogenic zone was observed [Hyndman et al., 1997]. Source time durations for select circum-Pacific subduction zones, not including the Mariana Islands, show a general trend of decreasing, normalized rupture time with increasing depth of plate interface earthquake [Bilek and Lay, 1999, 2000]. The results were interpreted to be indicative of an increase in rigidity due to compaction and de-watering of subducting sediments [Bilek and Lay, 1999, 2000]. [20] Along-strike variability in interplate coupling as indicated by spatial distribution of shallow earthquakes [Hasegawa et al., 2007] and earthquake rupture characteristics [Ammon et al., 2005] has been noted to some extent in almost all subduction zones. Some subduction zones clearly show different degrees of locking versus stable sliding along strike [Freymueller et al., 2008]. In the Mariana Islands, GPS data from the outer forearc are not available, so observations of creep and measures of interseismic locking cannot be obtained. The southern region may be more strongly coupled than the northern region [Scholz and Campos, 1995]; however this conclusion depends on the interpretation of large, shallow earthquakes in the historical records [Pacheco et al., 1993] as well as the controversial 1993 Guam earthquake [Tanioka et al., 1995; Harada and Ishibashi, 2008]. Observations of small earthquakes during a 2001 ocean bottom seismograph experiment in the Mariana Islands reveal distinct clusters of earthquakes - indicating that the plate interface may be slipping regularly in some regions, but may be either locked or slipping aseismically in other regions along the length of the subduction zone [Shiobara et al., 2010]. [21] Along-strike changes in shallow thrust seismicity have been explained due to effects of subducting oceanic seamounts or other bathymetric highs [Tanioka et al., 1997; Bilek et al., 2003; DeShon et al., 2003; Shinohara et al., 2005; Bilek, 2007]. To the north, the Magellan Seamount Cluster and Dutton Ridge within the East Mariana Basin intersect the Mariana and Bonin trenches and are composed of Cretaceous volcanic seamounts (∼100–120 Ma) [Smith et al., 1989]. Intersecting the Mariana and Yap trenches in the south are the seamounts and islands of the Caroline Island chain, which is made up of you