Present-Day Crustal Deformation Along the Philippine Fault in Luzon

Journal of Asian Earth Sciences 65 (2013) 64–74

Contents lists available at ScienceDirect

Journal of Asian Earth Sciences

journal homepage: www.elsevier.com/locate/jseaes

Present-day crustal deformation along the Philippine Fault in Luzon, Philippines ⇑ Shui-Beih Yu a, , Ya-Ju Hsu a, Teresito Bacolcol b, Chia-Chu Yang c, Yi-Chun Tsai a, Renato Solidum b a Institute of Earth Sciences, Academia Sinica, Taipei, Taiwan b Philippine Institute of Volcanology and Seismology, Quezon City, Philippines c Institute of Geophysics, National Central University, Taoyuan, Taiwan article info abstract

Article history: The Philippine Fault results from the oblique convergence between the Philippine Sea Plate and the Sun- Available online 26 January 2011 da Block/Eurasian Plate. The fault exhibits left-lateral slip and transects the Philippine archipelago from the northwest corner of Luzon to the southeast end of Mindanao for about 1200 km. To better understand Keywords: fault slip behavior along the Philippine Fault, eight GPS surveys were conducted from 1996 to 2008 in the Philippine Fault Luzon region. We combine the 12-yr survey-mode GPS data in the Luzon region and continuous GPS data GPS velocity in Taiwan, along with additional 15 International GNSS Service sites in the Asia-Paciﬁc region, and use the Crustal strain GAMIT/GLOBK software to calculate site coordinates. We then estimate the site velocity from position Interseismic deformation time series by linear regression. Our results show that the horizontal velocities with respect to the Sunda Dislocation model Block gradually decrease from north to south along the western Luzon at rates of 85–49 mm/yr in the west–northwest direction. This feature also implies a southward decrease of convergence rate along the Manila Trench. Signiﬁcant internal deformation is observed near the Philippine Fault. Using a two dimensional elastic dislocation model and GPS velocities, we invert for fault geometries and back-slip rates of the Philippine Fault. The results indicate that the back-slip rates on the Philippine Fault increase from north to south, with the rates of 22, 37 and 40 mm/yr, respectively, on the northern, central, and southern segments. The inferred long-term fault slip rates of 24–40 mm/yr are very close to back-slip rates on locked fault segments, suggesting the Philippine Fault is fully locked. The stress tensor inversions

from earthquake focal mechanisms indicate a transpressional regime in the Luzon area. Directions of r1 axes and maximum horizontal compressive axes are between 90° and 110°, consistent with major tec-

tonic features in the Philippines. The high angle between r1 axes and the Philippine Fault in central Luzon suggests a weak fault zone possibly associated with ﬂuid pressure. Ó 2011 Elsevier Ltd. All rights reserved.

1. Introduction (Chamot-Rooke and Le Pichon, 1999) showed the Sunda Block to be a distinct entity and rotates clockwise with respect to the Eur- The Philippine archipelago is a deformed orogenic belt resulting asian plate (Chamot-Rooke and Le Pichon, 1999; Michel et al., from the collage and collision of blocks of oceanic and continental 2001). Simons et al. (2007) used a decade (1994–2004) of GPS data afﬁnities (Karig, 1983). It is wedged between two converging to characterize the Sunda Block boundaries and derived the rota- plates: the oceanic northwest-moving Philippine Sea Plate in the tion pole at 49.0°N–94.2°E, with a clockwise rotation rate of east and the Sunda (Sundaland) Block/Eurasian Plate in the west 0.34°/Myr. The convergence rate of about 80–90 mm/yr between (Fig. 1). The east-dipping Manila Trench forms part of its western the Philippine Sea Plate and the Eurasian Plate has been reported boundary and together with the Negros-Sulu-Cotabato Trench, ab- from the plate model (Seno et al., 1993) and Global Positioning Sys- sorbs the convergence along the western side. The northwestward tem (GPS) (Yu et al., 1999). The oblique convergence between two motion of Philippine Sea Plate is absorbed in part, by the subduc- plates is decomposed into a trench-parallel component of 20– tion of west-dipping Philippine Trench and the East Luzon Trough 25 mm/yr on the Philippine Fault (Barrier et al., 1991) and a in the east, and in another by the Philippine Fault. Recent data de- trench-perpendicular component of 40–90 mm/yr on the Philip- rived from GPS (Rangin et al., 1999; Simons et al., 1999; Kreemer pine and Manila Trench (Megawati et al., 2009). et al., 2000; Bacolcol et al., 2005) and earthquake slip vectors The Philippine Fault is a sinistral strike-slip fault which transects the Philippine archipelago from north to south for about 1200 km. In spite of its recognition as a major geological structure ⇑ Corresponding author. Address: Institute of Earth Sciences, Academia Sinica, and sources of destructive earthquakes (Ms 7.5 1973 Ragay Gulf P.O. Box 1-55, Nankang, Taipei 115, Taiwan. Tel.: +886 2 2783 9910x416; fax: +886 earthquake; Ms 7.9 1990 Luzon earthquake; Ms 6.2 2002 Masbate 2 2783 9871. earthquake), a number of characteristics, e.g., precise fault location, E-mail address: [email protected] (S.-B. Yu).

Fig. 1. Tectonic setting in the Philippines. The black barbed lines denote the major trenches. The Philippine Fault (solid line in the middle) transects the Philippine archipelago from the northwest corner of Luzon to the southeast end of Mindanao for about 1200 km. segmentation, fault slip rates, seismicity, and earthquake recur- with slip rates of 23 and 36 mm/yr found in Masbate and Leyte rence intervals, are poorly understood. (Fig. 1), respectively (Bacolcol, 2003; Bacolcol et al., 2005). On the The ﬁrst quantitative measurement along the Philippine Fault other hand, the slip rate of the Philippine Fault is about 17– was in Mindanao Island wherein a left-lateral displacement of 31 mm/yr in the Luzon area (Yu et al., 1999). Geological and pa- about 28 km was found (Gervacio, 1971). Since then, studies on leo-seismological investigations indicate that the slip rate on the the motion of the Philippine Fault have been proposed using vari- Philippine Fault near central Luzon is generally between 9 and ous approaches (Acharya, 1980; Karig, 1983; Hirano et al., 1986; 17 mm/yr (Daligdig, 1997), which is lower than the value computed Pinet, 1990; Barrier et al., 1991; Aurelio, 1992; Duquesnoy et al., from GPS data. This discrepancy will be discussed in Section 5. 1994; Galgana et al., 2007). Based on GPS measurements, Duques- The area of interest in this study is the segment of Philippine noy et al. (1994) infer the slip rate of 26 ± 0.1 mm/yr on the creep- Fault in the Luzon Island, north of the Philippine archipelago. Lu- ing section of the Philippine Fault near the Leyte Island, consistent zon is part of the N–S trending Luzon arc, a 1200 km chain of 66 S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74

Fig. 2. An example of ITRF2005 GPS position time series for east, north and vertical components at station LUZE. Black dots with error bars show observed daily coordinate values and one standard deviation, respectively. Red dots are model predictions from linear regression. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) mostly late Tertiary to Quaternary volcanics that extends from the 45 km-long to the north of the mapped rupture based on geodetic Coastal Range of Taiwan (24°N) to Mindoro (13°N). In Luzon, the constraints. Daligdig (1997) examined geomorphic and paleoseis- Philippine Fault acts as the tectonic boundary that separates the mic data and found a recurrence interval of about 300–400 yrs Northern Luzon volcanics and Eastern Luzon metamorphics from along the Digdig Fault. the Zambales-Angat ophiolites (Karig, 1983; Karig et al., 1986). Using satellite imagery, digital elevation models, and geophysi- The Philippine Fault branches into several splays in central Luzon, cal data, Galgana et al. (2007) delineated six tectonic blocks in Lu- including the San Jose Fault, the San Manuel Fault, the Gabaldon zon and utilized a combination of earthquake slip vectors and GPS- Fault and the Digdig Fault (Nakata et al., 1977). derived horizontal surface velocities to invert for block rotations Historic records indicate large events with M 7 on the Philip- and elastic strain accumulated on the fault. They found that block pine Fault in the Luzon area occurred in 1901, 1937, and 1973 rotations can explain the majority of regional deformation in Lu- (Acharya, 1980). The 1973 event occurred along the Guinyangan zon; while fault-locking strain still makes a significant contribu- Fault near 13°N in southern Luzon (Morante and Allen, 1973; Mor- tion to the observed GPS velocity field near the fault. The ante, 1974). The region has at least seven major events in the last Philippine Fault is locked to partly-locked in the Luzon area and two centuries with a recurrence interval of about 65 yrs (Besana the locking depth is about 25 km. and Ando, 2005). The most recent large earthquake on the Philip- In this paper, we compute velocities of survey-mode GPS sites pine Fault, the Ms 7.9 July 16, 1990 earthquake, occurred on the using data collected between 1996 and 2008 and discuss the Digdig Fault segment in central Luzon and ruptured for about implications to regional tectonics. We use an interseismic strain 120 km. The average left-lateral slip is 5.4 m from seismic inver- accumulation model (Savage, 1983; Matsu’ura et al., 1986) to in- sion (Yoshida and Abe, 1992) and 5.5–6.5 m from geodetic inver- vert for the long-term slip rates (i.e., block motion) and back-slip sion (Silcock and Beavan, 2001). Both studies infer the bottom rates (i.e., slip deficits, presumably to be repaid in some forth- depth of coseismic rupture is about 20 km. Additionally, Silcock coming earthquakes) on various segments along the Philippine and Beavan (2001) reported another fault segment of about Fault. Additionally, we use earthquake focal mechanisms to con- S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74 67

Table 1 ITRF2005 station velocities in the Taiwan–Luzon region (1996–2008).

Site Lat (°) Lon (°) Ve (mm/yr) Vn (mm/yr) Vu (mm/yr) Time period (yr)

BGB1 15.6561 121.2144 36.6 ± 1.3 16.4 ± 0.8 0.7 ± 1.9 2000.3–2008.6 BRG1 18.5203 120.6008 48.5 ± 0.4 9.8 ± 0.4 17.5 ± 2.2 1996.2–2004.3 BTS3 20.4383 121.9628 38.5 ± 0.9 27.9 ± 0.6 10.0 ± 3.7 1996.2–2000.4 CMN2 14.1344 122.9827 36.2 ± 0.3 28.5 ± 0.2 1.0 ± 0.6 1998.1–2008.5 CMS2 13.7609 123.2861 30.4 ± 0.5 26.4 ± 0.4 2.1 ± 2.8 1998.1–2006.9 COCO 12.1883 96.8340 47.5 ± 0.4 50.6 ± 0.6 1.6 ± 0.7 1996.8–2009.0 CRIS 15.7829 121.0595 39.5 ± 0.9 13.1 ± 0.3 1.2 ± 0.7 1999.2–2008.6 DAEJ 36.3994 127.3745 26.2 ± 0.3 11.3 ± 0.2 1.0 ± 0.8 1999.2–2009.0 DARW 12.8437 131.1327 35.3 ± 0.4 58.8 ± 0.4 1.4 ± 0.9 1996.5–2009.0 GUAM 13.5893 144.8684 9.5 ± 0.4 3.1 ± 0.7 0.4 ± 0.9 1996.0–2009.0 HOB2 42.8047 147.4387 14.4 ± 0.5 55.5 ± 0.3 0.8 ± 0.6 1996.3–2009.0 IISC 13.0212 77.5704 40.0 ± 0.7 35.2 ± 0.4 0.0 ± 1.0 1996.2–2009.0 IRKT 52.2190 104.3162 23.2 ± 0.4 5.9 ± 0.3 0.4 ± 1.0 1996.0–2009.0 KAYT 13.9869 120.9777 21.1 ± 1.8 1.1 ± 1.3 2.3 ± 2.2 1999.2–2002.1 KDNM 21.9494 120.7820 18.2 ± 0.2 3.6 ± 0.1 6.0 ± 0.6 1996.0–2009.0 KUNM 25.0295 102.7972 28.2 ± 0.4 18.9 ± 0.3 1.1 ± 0.6 1998.8–2009.0 LUZA 14.8775 120.1953 25.2 ± 0.3 1.4 ± 0.2 0.4 ± 1.1 1996.2–2007.0 LUZB 15.3728 120.5171 26.5 ± 0.3 1.5 ± 0.2 0.8 ± 1.1 1996.2–2008.6 LUZC 16.3880 120.5682 34.1 ± 0.3 5.6 ± 0.2 1.0 ± 1.1 1996.2–2008.7 LUZD 17.5509 120.4556 41.7 ± 0.3 9.4 ± 0.2 2.9 ± 1.5 1996.2–2006.9 LUZE 15.5610 121.0970 33.5 ± 0.4 6.5 ± 0.2 1.6 ± 0.7 1996.2–2008.7 LUZF 15.8136 121.1131 39.2 ± 0.3 18.2 ± 0.2 0.6 ± 0.8 1996.2–2008.6 LUZG 16.6075 121.4821 40.8 ± 0.4 27.7 ± 0.2 2.3 ± 1.2 1996.2–2006.9 LUZH 17.7174 121.8038 46.7 ± 1.0 26.1 ± 0.5 3.4 ± 1.6 1996.2–2006.9 LUZI 18.2458 121.9279 45.1 ± 0.4 29.1 ± 0.4 2.8 ± 2.3 1996.2–2004.3 LUZL 14.6201 121.2080 29.0 ± 0.4 4.1 ± 0.3 3.2 ± 1.9 1996.2–2007.0 LUZN 13.7861 122.0605 13.6 ± 0.6 12.9 ± 0.6 15.3 ± 4.8 1998.1–2006.9 LUZP 13.8145 120.9753 15.2 ± 0.4 2.9 ± 0.4 1.4 ± 1.7 1998.1–2006.9 PABL 15.7856 121.0615 35.3 ± 1.2 12.2 ± 0.7 1.0 ± 1.1 1999.2–2008.6 PERT 31.8020 115.8853 38.7 ± 0.4 57.3 ± 0.2 4.9 ± 1.1 1996.5–2009.0 PHIV 14.6522 121.0587 29.1 ± 0.2 4.9 ± 0.4 7.0 ± 0.7 1998.1–2008.7 PIMO 14.6357 121.0777 30.2 ± 0.2 5.2 ± 0.4 1.8 ± 0.8 1999.2–2009.0 PLYN 15.5620 121.1008 33.8 ± 1.5 8.7 ± 0.4 0.3 ± 1.5 2000.3–2008.6 PUERa 10.086 118.851 32.0 ± 1.0 13.6 ± 0.7 1.8 ± 3.5 1994.9–1998.9 PUGA 15.7261 121.0407 34.0 ± 1.0 11.1 ± 0.6 1.0 ± 1.0 1999.2–2008.6 QZN6 14.1882 121.7287 21.7 ± 0.6 8.4 0.2 1.9 ± 0.7 1998.1–2004.3 S01R 23.6553 119.5924 29.9 ± 0.1 12.8 0.1 0.2 ± 0.4 1996.0–2009.0 S102 22.0372 121.5582 38.8 ± 0.3 39.3 0.3 6.6 ± 0.6 1996.0–2009.0 S104 22.8208 121.1894 23.7 ± 0.1 31.6 0.2 2.3 ± 0.4 1996.0–2009.0 S23R 22.6450 120.6062 22.9 ± 0.1 10.6 0.1 0.2 ± 0.3 1996.0–2009.0 SHAO 31.0996 121.2004 30.4 ± 0.5 13.7 0.3 1.9 ± 0.6 1996.0–2009.0 SIBH 13.5190 123.0135 20.9 ± 2.3 18.3 1.7 0.7 ± 1.5 2005.8–2008.5 TNSM 20.7026 116.7246 29.7 ± 0.2 12.9 0.1 0.2 ± 0.7 1996.3–2009.0 TOW2 19.2693 147.0557 28.6 ± 0.4 56.1 0.4 0.2 ± 0.6 1996.5–2009.0 TSKB 36.1057 140.0875 4.0 ± 0.3 7.1 0.3 0.9 ± 0.7 1996.0–2009.0 UP02 14.6553 121.0595 29.6 ± 0.2 4.9 0.3 4.7 ± 0.9 1997.4–2007.0 USUD 36.1331 138.3620 0.4 ± 0.3 9.3 0.3 1.6 ± 0.7 1996.0–2009.0 VRC4 13.5665 124.3368 31.7 ± 2.2 37.4 1.8 1.6 ± 1.4 2002.1–2007.0 WUHN 30.5317 114.3573 30.1 ± 0.3 12.8 0.3 3.5 ± 1.0 1996.1–2009.0

Ve, Vn, Vu are east, north and up velocity components and uncertainties quoted are standard errors. a Data from Simons et al. (2007). duct stress tensor inversions and illustrate the stress status ceeding GPS surveys were done annually until 2000. Three more along the Philippine Fault. Our study aims to delineate a com- campaigns were conducted in 2004, 2006, and 2008, respectively. prehensive deformation feature on the Philippine Fault in the Lu- In all of these surveys, most of the sites were occupied continu- zon area by combining available geological, geodetic, and seismic ously for 2–3 days using dual-frequency, geodetic GPS receivers, data. with a 30-s sampling rate. The collected GPS data are processed by GAMIT/GLOBK soft- 2. GPS data acquisition and processing ware packages, version 10.3 (Herring et al., 2009) using standard procedures based on double-difference phase observables, includ- The survey-mode GPS network in Luzon was established in late ing tropospheric and ionospheric modeling. We fix the Interna- 1995 and first measured in 1996 by the Institute of Earth Sciences, tional GNSS Service (IGS) final precise ephemerides in the Academia Sinica, Taiwan in collaboration with the Philippine Insti- parameter estimation. The GPS data used in the processing in- tute of Volcanology and Seismology, Philippines. Initially, the net- cludes 30 Luzon sites, 15 permanent IGS sites in the Asia-Pacific re- work was composed of 15 stations, including 10 newly set up sites, gion, and 6 sites from Taiwan continuous GPS Array (Yu et al., two National Mapping and Resource Information Authority (NAM- 1999) in southern Taiwan. These sites were integrated during the RIA) sites, and three Geodynamics of South and Southeast Asia processing to obtain a more accurate and consistent regional defor- (GEODYSSEA, Wilson et al., 1998; Michel et al., 2001) sites (Yu mation pattern of the Luzon Island. et al., 1999). Since then 16 stations were added to densify the net- Fourteen IGS sites (COCO, DAEJ, DARW, GUAM, HOB2, IISC, IRKT, work in central and southern Luzon between 1998 and 1999. Suc- KUNM, PERT, SHAO, TOW2, TSKB, USUD, WUHN) with long obser- 68 S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74

Table 2 vation period. Outliers and anomalous data are removed prior to Station velocities w.r.t. Sunda Block in the Taiwan–Luzon region (1996–2008). the final estimation of velocities. The best-fitting ITRF2005 velocities and their standard errors, station coordinates, and observation Site Ve (mm/yr) Vn (mm/yr) V (mm/yr) Azi (°) time periods for all sites are listed in Table 1. We then transform BGB1 69.2 ± 1.3 30.6 ± 0.8 75.7 ± 1.6 293.9 the ITRF2005 velocities into a Sunda Block-fixed reference frame BRG1 81.6 ± 0.4 23.8 ± 0.4 85.0 ± 0.5 286.3 BTS3 71.8 ± 0.9 42.4 ± 0.6 83.4 ± 1.1 300.6 (Simons et al., 2007). The horizontal velocities are shown in Table 2 CMN2 68.3 ± 0.3 43.3 ± 0.2 80.9 ± 0.3 302.4 and Fig. 3. CMS2 62.4 ± 0.5 41.4 ± 0.4 74.9 ± 0.7 303.5 The station velocities with respect to Sunda Block are about 49– COCO 24.9 ± 0.4 55.3 ± 0.6 60.7 ± 0.7 24.3 89 mm/yr, in the west–northwest (WNW) to northwest (NW) CRIS 72.1 ± 0.9 27.2 ± 0.3 77.1 ± 0.9 290.7 DAEJ 7.3 ± 0.3 5.0 ± 0.2 8.9 ± 0.3 304.2 directions. At the Batanes Islands (BTS3) and the northwestern cor- DARW 11.6 ± 0.4 76.3 ± 0.4 77.1 ± 0.6 8.7 ner of Luzon Island (BRG1) velocities are 83–85 mm/yr in the GUAM 39.9 ± 0.4 24.1 ± 0.7 46.6 ± 0.8 301.1 directions of 286–301°. Along the western coast of Luzon, the sta- GUIN 55.7 ± 2.0 35.8 ± 1.3 66.2 ± 2.4 302.8 tion velocities reduce southward gradually, they are 78 mm/yr, in HOB2 1.6 ± 0.5 77.1 ± 0.3 77.1 ± 0.6 1.2 287° at Santa (LUZD); 70 mm/yr, in 286° at Baguio City; 59 mm/ IISC 7.1 ± 0.7 31.7 ± 0.4 32.5 ± 0.8 12.6 IRKT 12.4 ± 0.4 1.9 ± 0.3 12.6 ± 0.5 278.7 yr, in 282° at Subic (LUZA) and 49 mm/yr, in 283° at Batangas KAYT 53.3 ± 1.8 13.0 ± 1.3 54.9 ± 2.2 283.7 (LUZP). This spatial variation of surface velocities also implies a KDNM 51.8 ± 0.2 10.4 ± 0.1 52.9 ± 0.2 281.4 southward decrease of convergence rate along the Manila Trench. KUNM 7.2 ± 0.4 11.7 ± 0.3 13.8 ± 0.5 211.6 Additionally, the station velocities on the east coast of Luzon are LUZA 57.6 ± 0.3 12.5 ± 0.2 59.0 ± 0.4 282.2 LUZB 59.1 ± 0.3 15.4 ± 0.2 61.1 ± 0.4 284.6 generally more northerly-directed than the corresponding sites on LUZC 66.8 ± 0.3 19.6 ± 0.2 69.7 ± 0.3 286.3 the west coast of Luzon and there are no significant changes on LUZD 74.7 ± 0.3 23.4 ± 0.2 78.2 ± 0.4 287.4 rates southward. The GPS velocities are 89 mm/yr, in the direction LUZE 66.1 ± 0.4 20.7 ± 0.2 69.2 ± 0.4 287.4 of 299° at the northeastern corner of Luzon (LUZI); 89 mm/yr, in LUZF 71.8 ± 0.3 32.4 ± 0.2 78.8 ± 0.4 294.2 297° at Tuguegarao (LUZH); 85 mm/yr, in 300° at Santiago (LUZG); LUZG 73.5 ± 0.4 42.0 ± 0.2 84.7 ± 0.5 299.8 LUZH 79.6 ± 1.0 40.5 ± 0.5 89.3 ± 1.1 297.0 and 81 mm/yr, in 302° at Daet (CMN2). The relative velocities be- LUZI 78.0 ± 0.4 43.6 ± 0.4 89.4 ± 0.5 299.2 tween four station pairs, LUZI-BRG1, LUZH-LUZD, LUZG-LUZC, LUZL 61.3 ± 0.4 18.3 ± 0.3 64.0 ± 0.5 286.6 and CMN2-LUZP are 19.8 mm/yr, in 10°; 17.6 mm/yr, in 344°; LUZN 45.7 ± 0.6 27.4 ± 0.6 53.3 ± 0.8 300.9 23.3 mm/yr, in 343°; and 38.1 mm/yr, in 327°, respectively. These LUZP 47.3 ± 0.4 11.2 ± 0.4 48.6 ± 0.5 283.3 PABL 67.9 ± 1.2 26.4 ± 0.7 72.8 ± 1.4 291.2 relative motions indicate a general picture of strain accumulation PERT 25.8 ± 0.4 69.6 ± 0.2 74.3 ± 0.5 20.4 across the Philippine Fault. PHIV 61.4 ± 0.2 19.1 ± 0.4 64.3 ± 0.4 287.3 In order to investigate the internal deformation in the Luzon PIMO 62.5 ± 0.2 19.4 ± 0.4 65.5 ± 0.5 287.3 arc, the Luzon ITRF2005 velocities are also transformed to a refer- PLYN 66.3 ± 1.5 22.8 ± 0.4 70.1 ± 1.5 289.0 ence frame fixed at Subic (LUZA), a GPS site located in western Lu- PUERa 0.7 ± 1.0 0.2 ± 0.7 0.7 ± 1.2 105.9 PUGA 66.6 ± 1.0 25.2 ± 0.6 71.2 ± 1.2 290.7 zon (Fig. 4). For comparison and to be used in the following QZN6 53.9 ± 0.6 22.8 ± 0.2 58.5 ± 0.7 293.0 modeling studies, part of the velocity data published in Galgana S01R 4.1 ± 0.1 0.9 ± 0.1 4.2 ± 0.2 281.9 et al. (2007) are also included in Fig. 4. Their results are in general S102 72.4 ± 0.3 53.6 ± 0.3 90.1 ± 0.4 306.5 agreement with ours, while we have a denser spatial coverage and S104 57.4 ± 0.1 45.8 ± 0.2 73.4 ± 0.2 308.6 longer time history of GPS data across the Philippine Fault. S23R 56.7 ± 0.1 3.4 ± 0.1 56.8 ± 0.2 273.4 SHAO 4.1 ± 0.5 0.6 ± 0.3 4.1 ± 0.6 277.7 We find that there are no significant velocity jumps across two SIBH 52.9 ± 2.3 33.1 ± 1.7 62.4 ± 2.9 302.1 sites of the Philippine Fault: along San Jose Fault (between PUGA TNSM 4.1 ± 0.2 0.2 ± 0.1 4.1 ± 0.2 266.6 and CRIS), the relative velocity rate is 5.9 ± 2.1 mm/yr; along Dig- TOW2 5.8 ± 0.4 77.6 ± 0.4 77.8 ± 0.5 4.3 dig Fault (between CRIS and LUZF), the velocity rate is TSKB 35.2 ± 0.3 12.8 ± 0.3 37.5 ± 0.4 290.0 UP02 61.9 ± 0.2 19.1 ± 0.3 64.8 ± 0.3 287.1 5.1 ± 1.5 mm/yr (Fig. 4). This feature may imply near-surface aseis- USUD 32.0 ± 0.3 10.2 ± 0.3 33.6 ± 0.4 287.7 mic creeping is absent on this portion of the Philippine Fault. VRC4 63.6 ± 2.2 52.7 ± 1.8 82.6 ± 2.8 309.6 Other known active faults found in the area also show insignif- WUHN 5.1 ± 0.3 1.1 ± 0.3 5.2 ± 0.4 257.7 icant velocity changes across the fault (Fig. 4): Ve, Vn are east and north velocity components, V and Azi are rate and azimuth of horizontal velocity, respectively. Uncertainties quoted are standard errors. – East Zambales Fault (between LUZA and LUZB): 3.1 ± 1.1 mm/yr. a Data from Simons et al. (2007). – Valley Fault System (between PHIV and LUZL): 0.8 ± 1.0 mm/yr. vation history surrounding the studied region are adopted as the 4. Dislocation model stabilization sites in GLOBK processing. The positions of these reference stations are constrained to their 2005 International Terres- Using GPS velocities derived in the previous section, we inves- trial Reference Frame (ITRF2005; Altamimi et al., 2007) tigate the strain accumulation on the Philippine Fault. We use coordinates. The station positions, variance–covariance matrices, the interseismic crustal deformation model (Savage, 1983; Mat- and other parameters from the GAMIT solution are combined in su’ura et al., 1986) such that the interseismic velocity field can GLOBK to produce estimates of station coordinates for other sites be represented as the sum of rigid block motion at long-term rates in the ITRF2005 reference frame. The position time series of sta- across the fault and back-slip rate (slip deficits) on the locked fault tions are then extracted from the GLOBK output files. Fig. 2 shows (negative dislocation). Because of an irregular spatial distribution an example of ITRF2005 position time series for east, north and up of GPS sites in the Luzon area, we only choose three transects components of station LUZE. across the Philippine Fault wherein GPS data is sufficient to con- strain fault slip rates. These transects from north to south are 3. Velocity field shown in Fig. 5, Transect AA0 is E–W directed and covered the northern Luzon region between 16°N and 17°N, Transect BB0 is We perform a least squares linear fit individually to position NE–SW directed with the width of 100 km and located in the mid- time series of three components (north, east, and up) at each GPS dle portion of the Luzon Island wherein the 1990 Ms 7.9 earth- station and estimate its average station velocity during the obser- quake occurred, and Transect CC0 is NE–SW directed with the S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74 69

Fig. 3. Station velocities with respect to Sunda Block (red vectors). Error ellipses indicate 95% confidence intervals of GPS velocities. DDF and PHF denote the Digdig Fault and the Philippine Fault, respectively. Color indicates the topography. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) 2 width of 130 km and covered the southern portion of the Luzon Is- X1=2 land. This region has experienced many large earthquakes in the Fðs; mÞ¼ ðGðmÞs dÞ ð1Þ past (Besana and Ando, 2005). We project GPS data along these 2-D transects and estimate P where 1=2 is the inverse square root of the data covariance ma- back-slip rate, fault width, dip, long-term slip rates using an elastic trix; G(m) are Green’s functions, which depend on the fault geom- half-space model (Okada, 1985). The positions of our model faults etry parameters m; s is slip rate; d is the observed GPS velocities. approximately follow the surface trace of the Philippine Fault. An The fit to the data is quantified from the mean of the normalized initial value of fault dip is assumed to be 90°. We search for the square residuals, v2; defined as the chi-square divide by the number optimal long-term slip rates and fault width using a grid search ap- r of data points. A value of 1 for v2 means that the model fits the data proach. In these 2-D models, we only consider the motion parallel r within uncertainties on average. to the Philippine Fault. The data are inverted using a weighted Due to limitations of our knowledge of fault parameters, we in- least-squares approach by minimizing the following functional: vert for fault geometry, long-term fault slip rate, and back-slip slip 70 S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74

Fig. 4. Station velocities with respect to LUZA. Error ellipses indicate 95% confidence interval of GPS velocities. Red vectors are velocities derived in this study and blue vectors are velocities reported by Galgana et al. (2007). DDF: Digdig Fault; EZF: East Zambales Fault; GF: Gabaldon Fault; PHF: Philippine Fault; SJF: San Jose Fault; SMF: San Manuel Fault; VFS: Valley Fault System; VPSF: Verde Passage-Sibuyan Sea Fault. (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) rate using a gird search approach. We search for values of long- back-slip rate and fault width are 22 mm/yr and 15 km, respec- term rates from 15 to 45 mm/yr, fault widths from 10 km to tively (Table 3 and Fig. 6a). However, a poor spatial coverage of 40 km, and fault dip from 60° to 90° based on apriori constraints GPS sites in the far field may limit the resolution on the fault param- from previous studies (Acharya, 1980; Barrier et al., 1991). The eters. In Transect BB0, the back-slip rate and fault width are 37 mm/ inversions show that the modeling results are sensitive to the yr and 15 km, respectively (Fig. 6b). The spatial distribution of GPS long-term slip rates but less sensitive to the fault width. However, sites in Transect BB0 is the best among three profiles; therefore, the our method does not directly provide correlation coefficients be- fault parameters are well determined. In Transect CC’, the back-slip tween all fault parameters. In order to estimate uncertainties of rate and fault width are 40 mm/yr and 28 km, respectively (Fig. 6c). fault parameters, we apply a bootstrap method by re-sampling The preferred fault dip is 70°, different from a nearly 90° dip ob- GPS data to generate 1000 synthetic data sets and compute fault tained in Transects AA0 and BB0. The deformation near Transect CC0 parameters. The starting fault parameters are obtained from the re- is more complex than other two sections and is possibly affected sults of grid search. Estimates of optimal fault parameters and their by the collision tectonics between Palawan and Mindoro (Rangin 95% confidence regions are given in Table 3. et al., 1999). Fig. 5 shows the seismicity between 1977 and 2009 from US Geological Survey (USGS)/National Earthquake Informa- 5. Results and discussion tion Center (NEIC) catalog. The seismicity near the Philippine Fault system in central Luzon is mainly related to the 1990 Ms 7.9 Luzon 5.1. Fault parameters earthquake and its aftershocks, while the seismicity activity is absent on the other segments of the Philippine Fault. The optimal models for three transects generally fit the data The coupling ratio is defined as the back-slip rate relative to the 2 0 long-term slip rate. If the ratio is large (close to 1.0), meaning fault with the values of vr between 1 and 2.6. In Transect AA , the S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74 71

Fig. 5. Locations of three transects and seismicity in the Luzon area. The seismicity between 1977 and 2009 is from USGS/NEIC catalog. Color circle indicates earthquake focal depth and its size is proportional to the magnitude. PHF: Philippine Fault. (For interpretation of the references to color in this ﬁgure legend, the reader is referred to the web version of this article.)

Table 3 Optimal fault parameters and 95% conﬁdence regions in three transects derived from elastic dislocation models.

2 Back-slip rate (mm/yr) Dip (°) Locking depth (km) Long-term rate (mm/yr) vr AA0 22 (17–35) 85 (79–90) 15 (6–17) 24 (20–25) 1.0 BB0 37 (33–47) 89 (80–90) 15 (10–26) 37 (32–45) 1.9 CC0 40 (33–50) 70 (52–90) 28 (20–56) 40 (33–52) 2.6

is fully locked, it implies the earthquake potential is high in the quake. The 1990 Ms 7.9 earthquake produced a 110 km-long area. In these three transects, the back-slip rates are close to the surface rupture and 5–6 m left-lateral slip near the Transect BB’ long-term fault slip rates (Table 3), suggesting that the Philippine (Yoshida and Abe, 1992; Silcock and Beavan, 2001). The mainshock Fault is nearly fully locked in the Luzon area. The inferred fault slip is likely to induce viscoelastic relaxation of the lower lithosphere. rates increase from north to south, in agreement with the distribu- Our preferred slip rate of 24–40 mm/yr on the Philippine Fault tion of historic large earthquakes and the results of fault block can be represented as an upper bound. Additionally, we compute models reported in Galgana et al. (2007), which shows the slip the recurrence interval of about 162 yr near the rupture area of rates near Transects AA0-BB0-CC0 are 17–27, 25–29, and 29– 1990 earthquake (Transect BB0) based on the average coseismic slip 40 mm/yr, respectively. Our estimate of the slip rate of about of 6 m (Yoshida and Abe, 1992; Silcock and Beavan, 2001) and the 37 mm/yr near the 1990 rupture zone is similar to 42 mm/yr de- interseismic slip rate of 37 mm/yr. The return period of the 1990 rived from an elastic model (Beavan et al., 2001). Additionally, type event derived from the ratio of peak slip to fault slip rate in the inferred fault slip rates are 20–25 mm/yr based on plate kine- this study can be represented as the lower bound of M 7 earth- matics (Barrier et al., 1991) and 35 mm/yr from GPS measurements quakes in the seismic hazard analysis. (Rangin et al., 1999; Thibault, 1999; Yu et al., 1999). However, fault slip rates inferred from geodetic data are larger than geologic slip 5.2. Stress tensor analysis rates. Geological and paleo-seismological investigations suggest 0 that the slip rate on the Philippine Fault (near Transect BB ) is gen- To give a comprehensive understanding of deformation in the erally between 9–17 mm/yr and the recurrence interval is 300– seismogenic zone, we investigate earthquake focal mechanisms 400 yrs (Daligdig, 1997). The discrepancy between the long-term between 1977 and 2009 from USGS/NEIC catalog to reveal stress and short-terms slip rates presumably results from the contamina- status at depths. We choose events near the Luzon area (12– tion of postseismic deformation associated with the 1990 earth- 19°N, 118–125°E) with focal depths less than 40 km (Fig. 7a). The 72 S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74

Fig. 6. GPS velocities parallel to the Philippine Fault along three transects indicated in Fig. 5. (a) Transect along line A–A0; (b) transect along line B–B0; (c) transect along line A–A0. Solid circles with error bars are the observed velocities and the red curves with circles represent the model predictions. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) earthquake magnitude falls in the range of 4.7–7.7. Most events oc- cur on the western branch of the Philippine Fault (16.5°N, 121°E) in northern Luzon and on the Verde Passage-Sibuyan Sea Fault (near 13–14°N, 120–122°E, VPSF in Fig. 4). We use the stress tensor Fig. 7. Earthquake focal mechanisms and results of stress tensor inversions. (a) inversion method proposed by Michael (1984, 1987) which as- Focal mechanisms from USGS/NEIC catalog with focal depth less than 40 km. Color beach balls indicate different fault types shown on bottom left corner. Major faults sumes that the slip on the fault plane occurs in the direction par- are shown in black lines. Colors indicate the shaded relief topography and allel to the direction of resolved shear stress. Figs. 7b and 8 bathymetry. (b) Results of the stress tensor inversions. Squares, triangles, and present inversion results based on a moving–window approach circles represent three principal stress axes, r1, r2 and r3, in equal-area projection on the 0.5°-spacing grid. We include all events within a 1° 1° of the lower hemisphere. The best solution is marked by large symbols with white rectangle centered at the node and estimate the stress tensor when outlines. The small symbols show the distribution of stress axes within 95% confidence region. (For interpretation of the references to colour in this figure there are at least 10 earthquakes within a given rectangular box legend, the reader is referred to the web version of this article.) (Fig. 7b). The present-day deformation shows predominately strike-slip faulting in Luzon with the stress ratio less than 0.5 with geological structures. The r1 axes are trending in the direction (Fig. 8a), indicating a transpressional regime and corresponding between WNW–ESE and E–W, while r2 axes are close to vertical S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74 73

Fig. 8. The stress ratios and directions of maximum horizontal compressive stress axes. (a) Gray color and texts denote the ratio of principal stress difference W from the stress tensor inversion. (b) Directions of maximum horizontal compressive stress axes. The numbers indicate azimuths. oriented. Additionally, we compute the maximum horizontal com- uous GPS Array and IGS sites in the Asia-Paciﬁc region, we derive a pressive direction of about 90–110° in the Luzon area (Fig. 8b), new velocity ﬁeld in the Taiwan–Luzon region. The GPS velocities consistent with major tectonic features in the Philippines (Barrier with respect to Sunda Block move at rates of 49–89 mm/yr in the et al., 1991; Aurelio et al., 1997; Rangin et al., 1999). west–northwest to northwest directions in Luzon. We observe a

Directions of r1 axes are sub-perpendicular to the Philippine gradual decreasing of station velocity from 85 mm/yr at the north- Fault in central Luzon (Fig. 7b). A similar distribution of r1 axes west corner of Luzon to 49 mm/yr at Batangas of southwestern Lu- has been found along the San Andreas Fault in California (Zoback zon. This feature also suggests a southward decrease of converging et al., 1987; Mount and Suppe, 1992). The r1 axes should be ori- rate along the Manila Trench. Using a two dimensional elastic ented about 30° to the fault plane for a strong fault (Byerlee, dislocation model, we infer the fault slip rate falls in the range of

1978), thereby the high angle between r1 axes and fault plane in 24–40 mm/yr and increases from northern to southern Luzon. central Luzon suggests a weak fault zone. The mechanism for the The Philippine Fault is nearly fully locked in the Luzon area. We weak fault might be related to high fluid pressure lowering the also perform stress tensor inversions using earthquake focal mech- effective normal stress on a fault and decreasing shear stress (Hub- anisms and find a transpressional stress regime in Luzon. The bert and Rubey, 1959; Sleep and Blanpied, 1992; Hardebeck and r1 axes are in the direction of 90–110°, at high angle to a nearly Hauksson, 1999). The fluid in the western central Luzon possibly N–S trending fault strike. This implies the segment of the release from the slab below the forearc region and facilitate partial Philippine Fault in central Luzon is possibly weak. melting and active volcanism (Defant et al., 1988). However, the total number of earthquake focal mechanism is insufficient to re- Acknowledgements solve the spatial variation of stress axes near the fault zone. The analysis on small to moderate-sized earthquake focal mechanisms We thank many colleagues at the Philippine Institute of Volca- from the Luzon seismic network is required in the future. nology and Seismology and the Institute of Earth Sciences, Acade- Based on modeling results in this study, we suggest to densify mia Sinica who participated in Luzon GPS surveys. We also thank the present GPS network in the Luzon area. Additional sites are re- Ministry of the Interior (MOI) of Taiwan and International GNSS quired to confidently resolve the long-term slip rates of the faults Service (IGS) community for providing the continuous GPS data comprising the Philippine Fault system. Constructing new sites in this study. Many figures in this paper were generated using along known active structures like the Valley Fault System and East the generic mapping tools (GMT) (Wessel and Smith, 1998). We Zambales Fault (Fig. 4) plays a crucial role in seismic hazard anal- are indebted to two anonymous reviewers for their constructive ysis. Measurements from various data sets would be gathered and suggestions and comments. This study was financially supported eventually used towards a development of a comprehensive pro- by Academia Sinica and National Science Council of Taiwan under gram in evaluating seismic risks in the Philippines. Grant NSC 98-2119-M-001-030. This is a contribution of the Insti- tute of Earth Sciences, Academia Sinica, IESAS1507.

6. Conclusions References

Using the survey-mode GPS data collected on eight campaigns Acharya, H.K., 1980. Seismic slip on the Philippine fault and its tectonic from 1996 to 2008 and continuous GPS data from Taiwan Contin- implications. Geology 8, 40–42. 74 S.-B. Yu et al. / Journal of Asian Earth Sciences 65 (2013) 64–74

Altamimi, Z., Collilieux, X., Legrand, J., Garayt, B., Boucher, C., 2007. ITRF2005: a new Part I. Source characterization and the resulting tsunami. Journal of Asian Earth release of the International Terrestrial Reference Frame based on time series of Sciences 36, 13–20. station positions and earth orientation parameters. Journal of Geophysical Michael, A.J., 1984. Determination of stress from slip data - faults and folds. Journal Research 112, 19–21. of Geophysical Research 89, 1517–1526. Aurelio, M., 1992. Tectonique du segment central de la faille philippine, étude Michael, A.J., 1987. Use of focal mechanisms to determine Stress - a control study. structurale, cinématique et évolution géodynamique, Ph.D., Univ. Paris 6, Paris. Journal of Geophysical Research 92, 357–368. Aurelio, M., Barrier, E., Gaulon, R., Rangin, C., 1997. Deformation and stress states Michel, G.M., Yu, Y.Q., Zhu, S.Y., Reigber, C., Becker, M., Reinhart, E., Simons, W., along the central segment of the Philippine Fault: implications to wrench fault Ambrosius, B., Vigny, C., Chamot-Rooke, N., Le Pichon, X., Morgan, P., tectonics. Journal of Asian Earth Sciences 15, 107–119. Matheussen, S., 2001. Crustal motion and block behaviour in SE-Asia from Bacolcol, T., 2003. Etudes geodesiques de la faille philippine a partir des donnees GPS measurements. Earth and Planetary Science Letters 187, 239–244. GPS, Thèse de Doctorat d’état, Université Pierre et Marie Curie. Morante, E.M., 1974. The Ragay Gulf earthquake of March 17, 1973, southern Luzon, Bacolcol, T., Aguilar, A., Jorgio, G., de la Cruz, R., Lasala, M., 2005. GPS constraints on Philippines. Journal of the Geological Society of the Philippines 28, 69–93. the Philippine fault slip rate in Masbate Island, Central Philippines. Journal of Morante, E.M., Allen, C.R., 1973. Displacement of the Philippine fault during the the Geological Society of Philippines 60. Ragay Gulf earthquake of March 17, 1973. Geological Society of America Barrier, E., Huchon, P., Aurelio, M., 1991. Philippine fault – a key for Philippine Bulletin 5, 744–745. kinematics. Geology 19, 32–35. Mount, V.S., Suppe, J., 1992. Present-day stress orientations adjacent to active Beavan, J., Silcock, D., Hamburger, M., Ramos, E., Thibault, C., Feir, R., 2001. Geodetic strike-slip faults: California and Sumatra. Journal of Geophysical Research 97, constraints on postseismic deformation following the 1990 M-s 7.8 Luzon 11,995–91,2013. earthquake and implications for Luzon tectonics and Philippine Sea plate Nakata, T., Sangawa, A., Hirano, S., 1977. A report on tectonic landforms along the motion. Geochemistry Geophysics Geosystems 2. doi:10.1029/2000GC000100. Philippine fault zone in Northern Luzon Philippines. Science Reports 7th Series- Besana, G.M., Ando, M., 2005. The central Philippine Fault Zone: location of great Geography 27, 69–93. earthquakes, slow events, and creep activity. Earth Planets and Space 57, 987– Okada, Y., 1985. Surface deformation to shear and tensile faults in a half space. 994. Bulletin. Seismological Society of America 75, 1135–1154. Byerlee, J., 1978. Friction of rocks. Pure and Applied Geophysics 116, 615–626. Pinet, N., 1990. Répartition des mouvements decrochants et chevauchements le Chamot-Rooke, N., Le Pichon, X., 1999. GPS determined eastward Sundaland motion long de frontière NW de la plaque Mer des Philippines, Paris. with respect to Eurasia confirmed by earthquakes slip vectors at Sunda and Rangin, C., Le Pichon, X., Mazzotti, S., Pubellier, M., Chamotrooke, N., Aurelio, M., Philippine trenches. Earth and Planetary Science Letters 173, 439–455. Walpersdorf, A., Quebral, R., 1999. Plate convergence measured by GPS across Daligdig, J., 1997. Recent Faulting and Paleoseismicity Along the Philippine Fault the Sundaland/Philippine sea plate deformed boundary: the Philippines and Zone, North Central Luzon, Philippines, Ph.D. Thesis, Kyoto University. eastern Indonesia. Geophysical Journal International 139, 296–316. Defant, M.J., Deboer, J.Z., Oles, D., 1988. The western central Luzon volcanic arc, the Savage, J.C., 1983. A dislocation model of strain accumulation and release at a Philippines – 2 arcs divided by rifting. Tectonophysics 145, 305–317. subduction zone. Journal of Geophysical Research 88, 4984–4996. Duquesnoy, T., Barrier, E., Kasser, M., Aurelio, M., Gaulon, R., Punongbayan, R.S., Seno, T., Stein, S., Gripp, A.E., 1993. A model for the motion of the Philippine Sea Rangin, C., Bautista, B.C., Delacruz, E., Isada, M., Marc, S., Puertollano, J., Ramos, Plate consistent with NUVEL-1 and geological data. Journal of Geophysical A., Prevot, M., Dupio, A., Eto, I., Sajona, F.G., Rigor, D., Delfin, F.G., Layugan, D., Research 98, 17941–17948. 1994. Detection of creep along the Philippine fault – 1st results of geodetic Silcock, D.M., Beavan, J., 2001. Geodetic constrains on coseismic rupture during the measurements on Leyte-island, central Philippine. Geophysical Research Letters 1990 M-s 7.8 Luzon, Philippines, earthquake, Geochemistry Geophysics 21, 975–978. Geosystems 2, 2000GC000101. Galgana, G., Hamburger, M., McCaffrey, R., Corpuz, E., Chen, Q.Z., 2007. Analysis of Simons, W.J.F., Ambrosius, B.A.C., Noomen, R., Angermann, D., Wilson, P., Becker, M., crustal deformation in Luzon, Philippines using geodetic observations and Reinhart, E., Walpersdorf, A., Vigny, C., 1999. Observing Plate Motions in S.E. earthquake focal mechanisms. Tectonophysics 432, 63–87. Asia: Geodetic Results of the GEODYSSEA Project, edn, American Geophysical Gervacio, F.C., 1971. Geotectonic development of the Philippines. Journal of the Union, Washington, DC, ETATS-UNIS. Geological Society of the Philippines 25, 18–38. Simons, W.J.F., Socquet, A., Vigny, C., Ambrosius, B.A.C., Abu, S.H., Promthong, C., Hardebeck, J.L., Hauksson, E., 1999. Role of fluids in faulting inferred from stress Subarya, C., Sarsito, D.A., Matheussen, S., Morgan, P., Spakman, W., 2007. A field signatures. Science 285, 236–239. decade of GPS in Southeast Asia: resolving Sundaland motion and boundaries. Herring, T.A., King, R.W., McClusky, S.C., 2009. Documentation for the GAMIT Journal of Geophysical Research 112. Analysis Software, Release 10.3, Massachusetts Institute of Technology, Sleep, N.H., Blanpied, M.L., 1992. Creep, compaction and the weak rheology of major Cambridge, MA, 183pp. faults. Nature 359, 687–692. Hirano, S., Nakata, T., Sangawa, A., 1986. Fault topography and quaternary faulting Thibault, C.A., 1999. GPS Measurements of Crustal Deformation in the Northern along the Philippine Fault zone, Central Luzon, the Philippines. Journal of Philippine Island Arc, MS Thesis, Indiana University. Geography 95, 1–23. Wessel, P., Smith, W.H.F., 1998. New, improved version of generic mapping tools Hubbert, M.K., Rubey, W.W., 1959. Role of fluid pressure in mechanics of overthrust released. EOS Transactions AGU 79 (47), 579. faulting. I. Mechanics of fluid-filled porous solids and its application to Wilson, P., Rais, J., Reigber, C., Reinhart, E., Ambrosius, B.A.C., Le Pichon, X., Kasser, overthrust faulting. Geological Society of America Bulletin 79, 115–166. M., Suharto, P., Majid, D., Yaakub, D.A., Almeda, R., Boonphakdee, C., 1998. Study Karig, D.E., 1983. Accreted terranes in the northern part of the Philippine provides data on active plate tectonics in Southeast Asia region. Eos, archipelago. Tectonics 2, 211–236. Transactions AGU 79 (545), 548–549. Karig, D.E., Sarewitz, D.R., Haeck, G.D., 1986. Role of strike-slip faulting in the Yoshida, Y., Abe, K., 1992. Source mechanism of the Luzon, Philippines earthquake evolution of allochthonous terranes in the Philippines. Geology 14, 852–855. of July 16, 1990. Geophysical Research Letters 19, 545–548. Kreemer, C., Holt, W.E., Goes, S., Govers, R., 2000. Active deformation in eastern Yu, S.B., Kuo, L.C., Punongbayan, R.S., Ramos, E.G., 1999. GPS observation of crustal Indonesia and the Philippines from GPS and seismicity data. Journal of deformation in the Taiwan–Luzon region. Geophysical Research Letters 26, Geophysical Research 105, 663–680. 923–926. Matsu’ura, M., Jackson, D.D., Cheng, A., 1986. Dislocation model for aseismic crustal Zoback, M.D., Zoback, M.L., Mount, V.S., Suppe, J., Eaton, J.P., Healy, J.H., deformation at Hollister, California. Journal of Geophysical Research 91, 2661– Oppenheimer, D., Reasenberg, P., Jones, L., Raleigh, C.B., Wong, I.G., Scotti, O., 2674. Wentworth, C., 1987. New evidence on the state of stress of the San-Andreas Megawati, K., Shaw, F., Sieh, K., Huang, Z.H., Wu, T.R., Lin, Y.N., Tan, S.K., Pan, T.C., fault system. Science 238, 1105–1111. 2009. Tsunami hazard from the subduction megathrust of the South China Sea: