Abstract

After 137 years without a great earthquake, the Mw 8.1 Pisagua event of 1 April 2014 occurred in the central portion of the southern Peru–northern Chile subduction zone. This megathrust earthquake was preceded by more than 2 weeks of foreshock activity migrating ∼3.5 km/day toward the mainshock hypocenter. This foreshock sequence was triggered by an Mw 6.7 earthquake on a reverse fault in the upper plate that strikes at a high angle to the trench, similar to well-documented reverse faults onshore. These margin-oblique reverse faults accommodate north-south shortening resulting from subduction across a plate boundary that is curved in map view. Reverse slip on the crustal fault unclamped the subduction interface, precipitating the subsequent megathrust foreshock activity that culminated in the great Pisagua earthquake. The combination of crustal reverse faults and a curved subduction margin also occurs in Cascadia and northeastern Japan, indicating that there are two additional localities where great megathrust earthquakes may be triggered by upper plate fault activity.

INTRODUCTION

Two recent great subduction earthquakes (Mw 8.8 Maule, Chile, in A.D. 2010; Mw 9.1 Tohoku-Oki, Japan, in 2011) and associated upper plate aftershocks have underscored the importance of crustal faulting triggered by megathrust events (Aron et al., 2013; Toda and Tsutsumi, 2013). However, the possibility that a foreshock on an upper plate fault might contribute to triggering a subduction great earthquake has never been clearly documented. On 1 April 2014, the Mw 8.1 Pisagua earthquake broke the central section of the Iquique gap, a segment of the Nazca–South America subduction zone that last ruptured in 1877 in an estimated Mw 8.7–8.9 event (e.g., Comte and Pardo, 1991). The 2014 earthquake was preceded by 13 months of seismic activity (Schurr et al., 2014), which intensified following the 16 March 2014 Mw 6.7 earthquake offshore of the town of Pisagua. Over the next 16 days, foreshocks progressively migrated toward the position of the Mw 8.1 mainshock.

The Mw 6.7 earthquake, the largest foreshock (Hayes et al., 2014; Schurr et al., 2014), is significant for two reasons: its location in the upper plate and the orientation of the nodal planes, which deviate significantly from the megathrust strike, as given in the U.S. Geological Survey Slab1.0 model (Hayes et al., 2012; Ruiz et al., 2014). However, to date, the role that the Mw 6.7 played in the intensification of foreshock activity has not been discussed in detail in the analyses of the Pisagua earthquake sequence (e.g., Yagi et al., 2014; Lay et al., 2014; Kato and Nakagawa, 2014; Ruiz et al., 2014; Hayes et al., 2014; Schurr et al., 2014). Based on event relocations, analysis of available focal mechanisms, and static stress change calculations, we illustrate how this major upper plate reverse faulting earthquake was the principal, if not the initial, trigger of subsequent foreshock activity on the megathrust that led to the Mw 8.1 mainshock. We use geologic and seismic data to illustrate that margin-oblique upper plate reverse faulting is widespread in the northern Chilean forearc, probably related to the bending of the orocline (Allmendinger et al., 2005), and in this case led to a great earthquake on the megathrust.

MARGIN-PARALLEL SHORTENING

Within a 250-km-long segment (19.2°S–21.6°S), numerous kilometer-scale margin-oblique active reverse faults cut the forearc overlying the concave-seaward portion of the Andean subduction zone (Fig. 1; Allmendinger et al., 2005). There are at least four sites along the coastline where reverse faults offset Quaternary marine terraces (Fig. DR1 in the GSA Data Repository1), and cumulative vertical offset of Miocene–Pliocene surfaces by single reverse faults reaches 500 m. Strikes of these reverse faults vary between 065° and 135° (j in Fig. 1), and slickenlines are oriented mainly subparallel to the dip direction, indicating the prevalence of reverse motion. Using fault plane and slickenline orientations, we estimate a mean shortening axis (P axis) for these faults as subhorizontal, trending 173°, and a subvertical extension axis (T axis) (i in Fig. 1). Kinematic data for single structures show moderate scatter of P and T axes, controlled mainly by variation in fault strike.

Between March 2010 and March 2012, a local seismological network of 21 short-period stations was installed in the Coastal Cordillera and Precordillera near 21°S with the aim of characterizing upper plate seismicity in the forearc (Bloch et al., 2014). We recorded 31 crustal earthquakes of Mw 0.6–2.7 with kinematics similar to those of the faults preserved in the geologic record (g in Fig. 1); i.e., reverse faulting on margin-oblique planes. This microseismic activity spans shallow depths (6.9 km) to the plate interface at ∼50 km depth, indicating that margin-oblique reverse faults are active throughout the crust.

PISAGUA EARTHQUAKE FORESHOCK SEQUENCE

The 1 April 2014 Mw 8.1 Pisagua mainshock and its foreshocks are collectively termed the Pisagua earthquake sequence. Schurr et al. (2014) showed a slowly decreasing seismicity rate before the largest foreshock. The time series of cumulative seismic moment release reveals that 3 earthquake clusters punctuated the background seismicity since at least 10 months before the mainshock. The pattern of decreasing of seismicity rate was interrupted immediately after the Mw 6.7 event, which represents the peak of foreshock activity (Fig. 2A).

We relocated Mw > 4 foreshocks that occurred between 16 and 31 March using the codes NonLinLoc (Lomax et al., 2000) and HypoDD (Waldhauser and Ellsworth, 2000). The relocation reveals clustering of seismicity at two depths: (1) <20 km (including the Mw 6.7 earthquake), inside the upper plate, and (2) on the interplate contact (Fig. 2B). Just a few hours after the Mw 6.7 event, the majority of seismicity migrated to the plate interface, but sporadic upper plate events also took place until at least 15 April. Our relocated distribution of foreshock latitudes shows that after the Mw 6.7 event, seismicity on the plate interface migrated northward parallel to slab strike at ∼3.3 ± 0.33 km/day and downdip at ∼1.3 ± 0.2 km/day toward the position of the 1 April Mw 8.1 event (Fig. DR2). The total migration distance of foreshocks is ∼49 km along a line of trend and plunge 004°, 08°.

FOCAL MECHANISM ANALYSIS AND REGIONAL STRUCTURES

Focal mechanisms for the Mw 6.7 earthquake were reported by the U.S. Geological Survey, the Global Centroid Moment Tensor (CMT), and the GEOFON–German Research Center for Geosciences (GFZ) (Table DR1). We derived our own focal mechanism based on moment tensor inversion from waveform analysis. All available focal mechanisms feature two nodal planes oriented obliquely to the local 341° strike of the megathrust. The average orientation of the low-angle north-dipping nodal plane is 285°, 25°N and the high-angle south-dipping nodal plane is oriented 128°, 68°W, with a P axis of trend and plunge of 213°, 27° (a in Fig. 1). The 64 summed moment tensors available for the vast majority of foreshocks after the Mw 6.7 yield a composite focal mechanism with the low-angle plane oriented 342°, 16°E, within 5° of the local strike and dip of the subduction zone from Slab1.0.

Scaling laws of Wells and Coppersmith (1994) suggest a source fault 36 km along strike and 14 km downdip for the Mw 6.7 upper plate event. The small dimensions of the reactivated fault preclude a direct connection to any mapped onshore structures. However, the orientations of the two nodal planes of the Mw 6.7 event are consistent with the strikes of reverse faults onshore (i and j in Fig. 1), as well as nodal planes of past crustal seismicity (e–g in Fig. 1). Therefore, we suggest that the Mw 6.7 earthquake may have reactivated a continuation of a structure similar to those documented onshore.

STATIC STRESS CHANGE ANALYSIS

We use static stress change calculations to explore the relationship between the upper plate Mw 6.7 earthquake and seismicity on the subduction megathrust. First, we calculated (using algorithms of Meade, 2007) the cumulative Coulomb stress change (CSC) imparted on the upper plate fault by the July 2013 and January 2014 earthquake clusters identified by Schurr et al. (2014) and found that they induced a very small CSC (<0.003 bar, or 300 Pa).

We subsequently focused on the possibility that the Mw 6.7 event triggered seismicity on the plate boundary that ultimately migrated toward the Mw 8.1 earthquake. We calculated the distribution of static CSC induced on the subduction interface by the Mw 6.7 earthquake and compared that to our relocations of subsequent megathrust foreshocks. We modeled each nodal plane of the Mw 6.7 event as a rectangular source fault using empirical scaling laws (Wells and Coppersmith, 1994) and using our derived location and the U.S. Geological Survey Mww focal mechanism (Table DR1). The regions of positive CSC up to ∼5 bar (0.5 MPa), defined with shear stress change in the local updip direction, generated by the north-dipping nodal plane contain 51% of the NonLinLoc-located aftershocks in the first 24 h after the Mw 6.7 event, while positive CSC zones calculated using the south-dipping plane contain 36% of these events (Fig. 3; Fig. DR8). However, regions of positive normal stress (unclamping) on the megathrust contain as much as 84% (for the north-dipping plane; 74% for the south-dipping plane) of these events, suggesting that reverse slip on either plane unclamped the megathrust (Fig. 3A; Fig. DR6) and facilitated the subsequent megathrust foreshocks, given the prevalence of updip shear stress accumulated during the preceding interseismic period. The January 2014 cluster of foreshocks that occurred before the Mw 6.7 earthquake (Schurr et al., 2014) also induced CSC on the megathrust of magnitude comparable to that of the upper plate event, localized around the small cluster near 20.2°S (Fig. 3). Positive CSC induced by all megathrust events between 16 and 31 March migrated northward and downdip along the megathrust, primarily reflecting increased updip shear stress induced by the foreshock sequence (Fig. 3C; Fig. DR7). The hypocenter of the Mw 8.1 Pisagua earthquake was in a region of positive CSC, just downdip from a cluster of foreshocks.

DISCUSSION

Geologic and seismic data show that the area around the Pisagua earthquake sequence is characterized by margin-parallel upper plate shortening that is expressed on short time scales in seismic records and on neotectonic time scales in the geology. This deformation regime is consistent with the obliquity of the nodal planes of the upper plate Mw 6.7 foreshock relative to the strike of the megathrust, with remarkable similarity between the kinematics inferred for this earthquake and those of the onshore seismicity and geologic fault plane analysis (Fig. 1). In particular, the P axis of the Mw 6.7 event is similar to the P axes estimated from geological structures and upper plate earthquakes previously detected in the area (i in Fig. 1). Therefore we conclude that the Mw 6.7 event represents the reactivation of one of the trench-oblique upper plate reverse faults common in this part of the forearc.

Recent work has suggested that during the Pisagua earthquake sequence, the megathrust was gradually unlocked by a propagation of long-term precursory events (Schurr et al., 2014). Our stress change analysis shows that the 16 March Mw 6.7 foreshock unclamped the megathrust, triggering the subsequent migrating sequence of foreshocks. Similar analysis indicates that the foreshocks before the middle of March did not transfer significant stress from the megathrust to the upper plate fault. In addition, the stress perturbation on the megathrust due to earthquake clusters prior to mid-March was localized compared to that promoted by the Mw 6.7 earthquake and subsequent events (Fig. 3). Therefore, we conclude that the occurrence of the Mw 6.7 earthquake primarily represents accommodation of trench-parallel shortening, and that the final conditioning of the megathrust in generating the great Mw 8.1 event was strongly influenced by the upper plate deformation regime. A key requirement of this process is the synchronicity of the earthquake cycles on the upper plate fault and the megathrust. If the megathrust had not yet reached a mature state in its interseismic period, the stress conditions on it may not have been suitable to produce a major rupture under the loading increment imposed by the upper plate faulting event. In the Iquique gap, 137 yr had elapsed since the last great earthquake, representing a complete interseismic period (e.g., Comte and Pardo, 1991).

The concave-seaward northern Chile–southern Peru margin is similar in shape to the northern Cascadia and northeast Japan Trench subduction zones, and upper plate reverse faults that strike nearly orthogonal to the plate boundary are present in each (e.g., Johnson et al., 2004; Kusunoki and Kimura, 1998). Concave-seaward segments of subduction zones can undergo interseismic shortening perpendicular to the direction of convergence, as velocity in that direction decreases toward the apex of margin curvature (Bevis et al., 2001), and this shortening may be enhanced by spatially heterogeneous interplate coupling (Rosenau et al., 2009). McCaffrey et al. (2013) showed contractional interseismic strain in the Cascadia forearc from the Olympic Peninsula to the Puget Lowlands, with shortening in the direction of plate convergence (east-northeast) but also in the perpendicular direction. Loveless and Meade (2010) estimated reverse slip on the margin-perpendicular Hidaka (Hokkiado, Japan) fold-thrust belt based on analysis of interseismic GPS velocities. Thus in each of these cases, margin-oblique reverse faults are loaded by subduction zone interseismic processes, yielding the possibility of megathrust activity triggered by seismicity on upper plate structures. This mechanism, which we have documented here for the Pisagua earthquake sequence case, encourages further study of crustal fault recurrence intervals, particularly in Cascadia, where 315 yr have passed since the great Mw 9.0 earthquake of A.D. 1700 and the seismic risk posed by both megathrust and upper plate earthquakes is large.

CONCLUSIONS

We suggest that the 1 April 2014 Mw 8.1 Pisagua subduction earthquake was mostly triggered by the 16 March Mw 6.7 foreshock on an upper plate reverse fault southwest of the mainshock epicenter. We show that static unclamping of the megathrust produced by the upper plate earthquake is probably the dominant mechanism that led to the subsequent foreshock-mainshock sequence. Following this main precursory event, progressive stressing of the subduction interface by the northeastward migration of additional foreshocks on the megathrust culminated in the partial rupture of the Iquique gap. The recent experience in northern Chile indicates that when assessing seismic hazard, earthquakes on upper plate faults should be taken into consideration, not only because of the local high-intensity shaking that they can produce, but also as possible triggering mechanisms for great earthquakes on the subduction zone interface.

We thank James Spotila, Onno Oncken, Brian Atwater, Chris Goldfinger, and two anonymous reviewers for helpful reviews, and Phil Liu and Ignacio Sepúlveda for insightful discussions. This research was supported by CONICYT/ FONDAP grant 15110017. Allmendinger acknowledges support from National Science Foundation (NSF) grants EAR-1443410, EAR-1019252, and EAR-05107852. Aron acknowledges support from CONICYT Beca Chile and NSF grant EAR-1118678. We thank the Integrated Plate boundary Observatory Chile for providing seismic waveform data.

1GSA Data Repository item 2015235, description of field geology, details of earthquake relocation and stress change modeling, Figures DR1–DR9, and Table DR1, is available online at www.geosociety.org/pubs/ft2015.htm, or on request from editing@geosociety.org or Documents Secretary, GSA, P.O. Box 9140, Boulder, CO 80301, USA.