Volume 126, Issue 7 e2021JB021879
Research Article
Free Access

Deep Structure of the Continental Plate in the South-Central Chilean Margin: Metamorphic Wedge and Implications for Megathrust Earthquakes

A. Maksymowicz

Corresponding Author

A. Maksymowicz

Departamento de Geofísica, Universidad de Chile, Santiago, Chile

Correspondence to:

A. Maksymowicz,

[email protected]

Search for more papers by this author
E. Contreras-Reyes

E. Contreras-Reyes

Departamento de Geofísica, Universidad de Chile, Santiago, Chile

Search for more papers by this author
D. Díaz

D. Díaz

Departamento de Geofísica, Universidad de Chile, Santiago, Chile

Search for more papers by this author
D. Comte

D. Comte

Departamento de Geofísica, Universidad de Chile, Santiago, Chile

Advanced Mining Technology Center, Santiago, Chile

Search for more papers by this author
N. Bangs

N. Bangs

Institute for Geophysics, University of Texas at Austin, Austin, TX, USA

Search for more papers by this author
A. M. Tréhu

A. M. Tréhu

College of Earth, Ocean and Atmospheric Sciences, Oregon State University, Corvallis, OR, USA

Search for more papers by this author
E. Vera

E. Vera

Departamento de Geofísica, Universidad de Chile, Santiago, Chile

Search for more papers by this author
F. Hervé

F. Hervé

Departamento de Geología, Universidad de Chile, Santiago, Chile

Search for more papers by this author
A. Rietbrock

A. Rietbrock

Geophysical Institute, Karlsruhe Institute of Technology, Karlsruhe, Germany

Search for more papers by this author
First published: 17 June 2021
Citations: 4

Abstract

We study the deep structure of the continental wedge along the south-central Chilean convergent margin based on a joint interpretation of wide-angle reflection and refraction seismic phases acquired between the southern part of the Maule 2010 (36°S) and the northern part of the Valdivia 1960 (41°S) earthquake rupture areas. When combined with results of previous seismic studies, our results provide new insights into the deep structure (to 15–20 km) of the overriding, continental South American plate. We observe a latitudinal variation of P wave velocity, interpreted in terms of the distribution of Permo-Triassic metamorphic complex near the coast, and deep reflectors at the base of these metamorphic units. A change in the observed deep structural style near 38°S suggests a segmentation in the interplate frictional properties between the Maule and Valdivia earthquakes rupture zones. This regional change in the continental wedge structure indicates a spatial correlation between the coseismic ruptures of large earthquakes (e.g., the 1960 and 2010 megathrust earthquakes) and the rheology/lithology along the interplate boundary.

Key Points

  • A middle continental wedge unit of relatively high Vp is interpreted as the seaward continuation of the Permo-Triassic metamorphic complex

  • Observed deep reflectors show the presence of structures at the base of the Permo-Triassic metamorphic complex

  • A change in the deep structural style, near 38°S, suggests a segmentation in the interplate frictional properties

Plain Language Summary

Recent studies suggest that upper plate structure plays a key role during large earthquakes in subduction zones. This work presents the interpretation of 2D seismic velocity models and the analysis of seismic waves reflected in the deep zone of the continental crust, obtained along the south-central Chilean subduction zone (between 36°S and 41°S). This zone includes the southern portion of the Mw8.8 Maule earthquake (2010) and the northern patch of the Mw9.6 Valdivia earthquake (1960). Seismic data were acquired during the NSF-CEVICHE (Crustal Experiment from Valdivia to Illapel to Characterize Huge Earthquakes) cruise by the American R/V Marcus Langseth in January/February 2017 and by seismometers deployed onshore. The data processing and interpretation are complemented by previous seismic information in the area. We observe a variation of seismic structure along the margin, interpreted in terms of the distribution of Permo-Triassic metamorphic units. The change in the observed deep structural style suggests differences in the frictional properties between the tectonic plates in the zones where the Maule and Valdivia earthquakes ruptured. Results highlight that a physical characterization of the continental wedge and the study of its geological evolution provides a necessary framework for understanding the complexities of large subduction earthquakes.

1 Introduction

Several studies in recent decades suggest a possible relationship between the structure of the upper plate and the rupture process of megathrust earthquakes in subduction zones. Worldwide, reflection seismic studies (e.g., Bangs & Cande, 1997; Bangs et al., 2020; Geersen et al., 2018; Kodaira et al., 2012; Kopp et al., 2001; Ranero et al., 2006; Tsuji et al., 2013), tomographic inversion of seismic velocities (e.g., Contreras-Reyes et al., 2012; Klingelhoefer et al., 2010; Martínez-Loriente et al., 2019; Nakanishi et al., 2002; Sallarès et al., 2013), and density models (e.g., Fleming & Tréhu, 1999; Hackney et al., 2006; Lücke & Arroyo, 2015; Maksymowicz et al., 20152018; Wang et al., 2004; Zhang et al., 2020) show important variations (or segmentation) in the structure of the continental wedge, along the strike and dip direction of the subduction zone. This segmentation seems to control, at least partly, the rate and clustering of the seismicity and the rupture of large earthquakes, including heterogeneities of slip and the distribution of foreshocks and aftershocks (Contreras-Reyes et al., 2010; Hicks et al., 2014; Maksymowicz et al., 2018; Moscoso et al., 2011; Tsuji et al., 2017; Wang et al., 2004).

Among the upper plate structures relevant to earthquake rupture processes, the active accretionary prisms (and/or frontal units of highly fractured rock) have a key role in the seismotectonics of the margin. Conditions along the base of accretionary prism are conducive to aseismic behavior (stable sliding) but occasionally allow coseismic rupture of large earthquakes under velocity weakening conditions (Kodaira et al., 2012; Maksymowicz et al., 2017; Scholz, 1998; Tréhu et al., 2019). These conditions evolve farther downdip, as the upper plate transitions into older continental units with higher seismic velocities and densities, which can be associated with landward reduction in porosity/fracturing, and/or lithological changes (Contreras-Reyes et al., 2015; Klingelhoefer et al., 2010; Sallarès & Ranero, 2005). The landward changes in upper plate mechanical properties could be associated with transitions between regions of stable and unstable behaviors at the base. On the other hand, Menant et al. (2019) show that the presence of basal accretionary complexes (sedimentary underplating) can introduce important variations in the frictional properties along the megathrust.

The fluid pressures within and at the base of the wedge are key parameters that control the geometry and internal deformation of the continental wedge (Cubas, Avouac, Leroy, & Pons, 2013; Cubas, Avouac, Souloumiac, & Leroy, 2013; Dahlen, 1984; Koge et al., 2014; Maksymowicz, 2015). Geophysical observations and models show that fluid transport in the subduction channel depends on numerous parameters such as the trench fill, the porosity of the subducted sediments, the geometry of subduction channel, the density and structure of the continental wedge (and thus its weight), fracturing, wedge temperature, and other factors (Calahorrano et al., 2008; Le Pichon et al., 1993; Menant et al., 2019; Ranero et al., 2008; Sage et al., 2006; Watson et al., 2019; Yamada et al., 2014). Accordingly, the spatial variations in porosity/fracturing, velocity, density, and deformation of the overriding plate can be linked with the fluid transport and fluid pressure in the subduction channel, which in turn can modify the frictional and seismogenic properties of the margin.

The interplay between the physical segmentation of the continental wedge and seimotectonics of the margin is also reflected by the spatial correlation of large marine forearc basins and low- and high-gravity anomalies with the rupture areas of large earthquakes and location of high slip patches (Bassett & Watts, 2015; Bassett et al., 2016; Reginato et al., 2020; Schurr et al., 2020; Song & Simons, 2003; Wells et al., 2003). This suggests that the variations in normal stresses over the seismogenic contact generated by large changes in the overriding plate density play a key role in the complex process of megathrust activation during the seismic cycle (Li & Liu, 2017; Maksymowicz et al., 2015; Tassara, 2010). Recently, Sallarès and Ranero (2019) performed a global analysis of published Vp models, showing a general correlation between the downward increase of Vp (above the interplate boundary) with the variations of slip of large megathrust earthquakes (and other rupture properties) as a function of depth, suggesting again, a direct role of the upper plate physical properties on the coseismic slip process.

Along the central Chilean margin, some authors have proposed a direct relationship between the structure of the overriding plate and the rupture properties of large megathrust earthquakes. For instance, a spatial correlation is observed between the morphology of the continental wedge and the coseismic slip distribution during the 2010 Mw8.8 Maule and the 1960 Mw9.5 Valdivia earthquakes, where low angles of marine continental slope (from trench to shelf break) are observed in zones of high slip (Contreras-Reyes et al., 2017; Maksymowicz, 2015). This morphological change can be related to internal deformation (and fluid pressure) of the continental wedge and frictional properties along the interplate boundary (Cubas, Avouac, Souloumiac, & Leroy, 2013; Maksymowicz, 2015). Similarly, the high slip patches of the Maule and Valdivia earthquakes are spatially correlated to low gravity/density anomalies along the marine forearc (Hackney et al., 2006; Maksymowicz et al., 2015; Wells et al., 2003). Onshore, the surface width of old metamorphic basement outcrops (Permo-Triassic paired metamorphic belt) also seems to correlate with the patches of high slip (Figure 1), suggesting that the physical structure, lithology, and the deformation process of the upper plate play a fundamental role in the seismic segmentation of the margin. Hicks et al. (2014) interpreted two large high P wave velocity anomalies of 7.8 km/s beneath the coastline along the Maule rupture area as ultramafic bodies associated with the extension and upwelling during the Triassic. These high-velocity bodies act as barriers with reduced coseismic slip and minimal postseismic activity (Hicks et al., 2014) further pointing toward the influence of the upper plate compositional heterogeneity on the earthquake rupture processes.

Details are in the caption following the image

Morphology and basement geology of the study zone. Black lines are the four marine seismic profiles acquired during the Crustal Experiment from Valdivia to Illapel to Characterize Huge Earthquakes experiment that are discussed here. Black triangles are the seismometers deployed onshore to record the marine shots. Dotted purple line indicates the position of the shelf break, whereas green and red lines are the iso-slip contours for Maule and Valdivia earthquakes, respectively (M. Moreno et al., 2012; M. S. Moreno et al., 2009). MFZ and VFZ are the Mocha and Valdivia Fracture Zones of the Nazca plate. Blue lines are the main fault systems observed inland, including the Liquiñe-Ofqui fault system (LOFS), Mocha-Villarrica fault (MVF), Lanalhue Faut Zone (LFZ), and Bío-Bío-Aluminé Fault (BBAF). Red triangles mark the location of the volcanic arc. Morphotectonic units of the forearc are indicated as AP (Arauco Peninsula), CR (Coastal Range), CD (Central Depression), and AC (Andes Cordillera).

The south-central Chilean margin is an ideal natural laboratory to study the impact of the subduction processes on the evolution of the upper plate structure, and the possible control exerted by upper plate heterogeneities on the seismotectonic process (i.e., to understand the feedback between long-term and short-term deformation process in subduction margins). We therefore investigate the internal structure of the continental wedge, at regional scale, by using 2D modeling of onshore recordings of marine airgun data acquired during the CEVICHE (Crustal Experiment from Valdivia to Illapel to Characterize Huge Earthquakes, Bangs et al., 2020; Olsen et al., 2020) project between 36°S and 41°S, complemented by previous seismic data acquired in the region. Tomographic inversion of first arrival times provides new information on along-strike changes in the velocity of the overriding plate and internal segmentation of the continental wedge. Particularly, onshore recordings of offshore shots provide information on the lower crustal structure beneath the coastal zone, which is a critical region for earthquake nucleation in this portion of the margin.

2 Tectonic Setting

The western edge of South America has undergone continuous subduction for the last ∼200 My (continuous at least since the Jurassic; Charrier et al., 2007). This active tectonic regimen has generated volcanism, subduction erosion, accretion, and crustal deformation at different temporal and spatial scales (Clift & Vannucchi, 2004). An important result of this complex geodynamic history is the generation of the main morphotectonic units along the Chilean margin. These units are the Coastal Range (CR in Figure 1), a large forearc basin (Central Depression, CD in Figure 1), and the Andes Cordillera which is the largest noncollisional orogeny observed worldwide. Rapid convergence rate between Nazca and South America plates (currently ∼6.6 cm/year, according to Angermann et al., 1999) results in the occurrence of important seismic activity, including some of the largest instrumentally recorded, and historically reported earthquakes (M. Moreno et al., 2012; M. S. Moreno et al., 2009).

The south-central Chilean margin is currently undergoing tectonic accretion (Bangs & Cande, 1997; Contreras-Reyes et al., 20102017) and has been characterized by net subduction erosion since Mesozoic times (Kukowski & Oncken, 2006). This long-term erosive process generated the eastward migration of the trench, the magmatic arc, and the deformation front. During this long period, the margin evolved by a succession of different tectonic styles (Andean Tectonic Cycle, Charrier et al., 2007). In particular, between 36° and 43°30'S, a magmatic belt developed in an extensional tectonic regime along the Central Depression and Coastal Range during the Middle-Tertiary (Muñoz et al., 2000). This was followed by a compressive stage during the Eocene–Oligocene, as indicated by inversion of Arauco Basin at ∼38°S (Becerra et al., 2013).

This study examined the southern segment of the 2010 Maule Mw8.8 earthquake rupture and the northern portion of the 1960 Mw9.5 Valdivia earthquake rupture area (Figure 1). The southern segment of the Maule event shows moderate slip (<15 m), compared to its northern segment (M. Moreno et al., 2012). In contrast, the northern segment of Valdivia earthquake is the region of greatest slip for that event (>35 m, M. S. Moreno et al., 2009). It is interesting to note that, according to Ruegg et al. (2009), the southern segment of the Maule earthquake ruptured independently during the 1835 earthquake, which was experienced by Charles Darwin during his travel to South America.

Offshore, the southern segment of the Maule earthquake rupture zone (∼36°S–38.5°S) is characterized by a relatively narrow continental slope (∼40 km wide), defined as the region between the deformation front and the shelf break (dotted magenta line in Figure 1). In contrast to the wide continental slope observed in the northern segment of Valdivia earthquake rupture zone (∼70–80 km wide, ∼38.5°S to ∼41°S). This morphological variability is correlated with a change in the slope angle. Defining the slope as a line from the deformation front to the shelf break, this angle varies from ∼4° in the southern segment of the Maule rupture to ∼2.5° in the northern segment of Valdivia rupture (Maksymowicz et al., 2015). Applying Non-Cohesive Coulomb Wedge (NCCW) theory, this observation can be interpreted as a decrease in the effective basal fiction coefficient (μb*) south of ∼38.5°S (Maksymowicz et al., 2015).

The geology of the study zone (SERNAGEOMIN, 2003) shows a latitudinal segmentation of the pre-Mesozoic basement units. The presence of a Permo-Triassic paired metamorphic belt (Western/Eastern series) is observed continuously near the coast (at the Coastal Range). However, the width of their outcrops varies along the margin (see Figure 1) showing landward prolongation of the Western/Eastern Series to the north of ∼36.5°S and to the south of ∼38°S (Hervé, 1988). The onshore limit of the metamorphic complex is not defined southward of 40°S due to the presence of the forearc basin deposits (Central Depression). Thus, the metamorphic units could form most of the forearc basement, or it could be confined near the Coastal Range. The Western and Eastern Series are interpreted as relicts of basal and frontal accretionary prisms, respectively (Willner et al., 2005), and therefore, their physical properties (density, velocity, fracturing degree, etc.) should be different from the surrounding continental basement. The mechanical properties of the forearc and/or the frictional properties of the seismogenic contact could be affected by the distribution of these metamorphic units.

To the north of the study zone (34°–35°30'S), Willner et al. (2005) show that the western portion of the Permo-Triassic paired metamorphic belt (Western Series) reaches the upper crust (lithostatic pressure of ∼200 MPa) ∼230 Ma, during a period of intense basal accretion, after which the P-T-time path of the rocks is consistent with low exhumation rate (<0.06 mm/year). These low exhumation rates are interrupted by a short period of high contractional deformation during the Middle-Cretaceous (113–80 Ma). A similar process is presented by Glodny et al. (2005) for the Western Series at 39°50', where a high exhumation rate (∼0.6 mm/year) is inferred during a basal accretionary stage (250–212 Ma), followed by a decrease in the exhumation rate to ∼0.04 mm/year. Between 36°S and 42°S, Glodny et al. (2008) show that the Coastal Range was exhumed to upper crustal levels (<3 km depth) in the Late Cretaceous. A particular case is observed in the segment between ∼37°S and 38°S (roughly the AP segment) where the Coastal Range exhibit large exhumation rates in Plio-Pleistocene times (Glodny et al., 2008).

The long-term deformation of the forearc has also been affected by the large continental fault zones observed on the surface (see blue lines in Figure 1). South of 38°S, the active large intra-arc Liquiñe-Ofqui fault system (LOFS) shows dextral displacement, which accommodates part of the likely oblique convergence between the Nazca and South America plates (Cembrano et al., 1996). Westward, an array of large fault zones with northwest strike, and conjugated northeast structures, affected the forearc (e.g., Mocha-Villarrica fault [MVF], Lanalhue Faut Zone [LFZ], and Bío-Bío-Aluminé Fault [BBAF] in Figure 1). Evidence of this structural system can be observed along the western border of the Andes Cordillera, along the Coastal Range and in seismic lines and bathymetry offshore (Melnick et al., 2009). However, the traces of these fault systems are covered by the sediments of the Central Depression, hindering the mapping of their exact continuity from the main cordillera to the coast. These continental structures show a complex kinematics, with changes along strike, but in general, the main northwest strike faults (as MVF, LFZ, and BBAF) exhibit dextral-reverse displacements (Krawczyk et al., 2006; Melnick et al., 2009; Pérez-Flores et al., 2016). LOFS has been interpreted as the transcurrent dextral limit of a continental microplate (Chiloé Microplate, Geersen et al., 2011; Melnick et al., 2009), which should involve the forearc continental crust between ∼38°S to the Chile Triple Junction at (46°S). According to Melnick et al. (2009), the northward displacement of this continental block (relative to South America) generates intense deformation in its northern limit, explaining the kinematics of large northwest strike continental faults (as LFZ), the emergence of AP (in Figure 1), and the eastward bending of the Coastal Range (Arauco orocline).

Offshore, the continental wedge appears to be segmented into several domains between the trench and shoreline. Segmentation is characterized by regions of the thrust faulting within accretionary prims (along the lower slope), compressional geometries and confined slope basins inside the middle and upper slope, and normal and inverted faults observed in the shelf region (related to the formation of larger marine forearc basins). However, the expression and extension of these deformation styles vary along the margin (e.g., Bangs & Cande, 1997; Bangs et al., 2020; Becerra et al., 2013; Geersen et al., 2011; Tréhu et al., 2019). This makes necessary to analyze the deformation style together with morphological characteristics and physical properties of the basement (Vp, density), to understand possible feedbacks between long-term deformation and mechanical properties of the crust.

A remarkable morphological feature, correlated with the limit between the Maule and Valdivia earthquake rupture areas, is the AP (in Figure 1). This uplifted and emerged part of the continental shelf has been associated with a rapid uplift since the Pleistocene (Glodny et al., 2008) and, as was mentioned above, it has been interpreted as the result of a local North-South compression generated by the dynamics of the northern limit of LOFZ (Melnick et al., 2009). Alternatively, it may be an effect of subduction of a hypothetical bathymetric high associated with the Mocha Fracture Zone (∼3.6 Ma, Folguera & Ramos, 2009). This feature appears to be an aseismic barrier to rupture propagation during earthquakes (Saillard et al., 2017).

Previous seismic profiles in the study area were acquired during SPOC (Subduction Processes Off Chile)-2001 and ISSA (Integrated Seismological experiment in the Southern Andes)-2000 projects (gray lines and gray triangles in Figure 2). Offshore, at 38.2°S, the seismic structure of the continental wedge was obtained through 2D inversion of OBS (Ocean Bottom Seismometer) data (Contreras-Reyes et al., 2008), resulting in a model with adequate resolution to determine the backstop geometry at this latitude. Other seismic refraction profiles, generated by forward modeling of marine and onshore experiments, show smooth representations (relatively low resolution) of the continental wedge at 36°S, 37°S, and 38.25°S (Krawczyk et al., 2006; Lüth et al., 2003). Despite the horizontal smoothness of these models, the offshore velocity structure shows a general landward Vp increase within the shallow continental wedge (<10 km depth). Onshore, these profiles show high Vp values below the Coastal Range (at ∼36°S) and around the volcanic arc (at 38.25°S), which could be landward extensions of offshore metamorphic units at different latitudes. Between 39°S and to 43°S, no velocity models are available. At 38.25°S, Krawczyk et al. (2006) presented an onshore reflection seismic profile, generated by explosives sources, that shows deep reflectors interpreted to results from the internal structure of metamorphic complexes below the Coastal Range and as evidence of a possible basal accretionary prism above the interplate boundary.

Details are in the caption following the image

Location of modeled seismic data. Black lines correspond to the seismic profiles acquired by NSF-Crustal Experiment from Valdivia to Illapel to Characterize Huge Earthquakes project and inverted black triangles are the short-period stations deployed during the marine seismic experiment in January/February 2017. Gray lines are the marine seismic profiles acquired in 2001 by the Subduction Processes Off Chile (SPOC) project. Inverted gray triangles and red stars are the receivers and sources deployed during onshore seismic experiment of the Integrated Seismological experiment in the Southern Andes-2000 and SPOC projects. The inverted blue triangles correspond to a subset of the seismological network that acquired marine seismic shots during the SPOC project. Blue and red lines are the iso-slip contours for Maule (M. Moreno et al., 2012) and Valdivia (M. S. Moreno et al., 2009) earthquakes, respectively.

3 Data and Methods

In order to analyze the regional variation of the continental wedge structure, we generated 2D velocity–depth tomography models along four seismic profiles (P1–P4 in Figures 1 and 2). These seismic lines include data from four of the high-resolution marine seismic profiles (MC33, MC13, MC18, and MC28) acquired during the cruise MGL1701 of the R/V Marcus Langseth for CEVICHE (Bangs et al., 2020). A 15-km-long seismic streamer was used to record about 2,500 shots along each profile. Shots were generated by a seismic airgun source array with a total volume of 6,600 in3 fired at 1,900 ± 100 psi. In coordination with the marine survey, 32 short-period seismometers provided by the Karlsruhe Institute of Technology, Germany, and the Advance Mining Technology Center at the Universidad de Chile were deployed to record seismic signal from the CEVICHE shots onshore. However, some sites were affected by anthropogenic noise, and some stations located far from the coast (at distances >100 km) did not record clear marine shot signals. Here, we use data from 24 of these land stations (black triangles in Figures 1 and 2).

We generated travel time curves by manually picking direct P wave phases on shot records recorded on the 15-km-long streamer. We picked shots at 5 km intervals (see Figure 3 and Supplementary Material) and inverted those picks to obtain a Vp model along the profile. To extend our models onshore, we included the arrival times of marine shots recorded by land stations (see Figure 3 and Supplementary Material). For profile P2, we also included data acquired during the SPOC-2001 and ISSA-2000 projects almost two decades ago (Lüth et al., 2003).

Details are in the caption following the image

Examples of common shot gathers with picked arrivals. (a) Common shot gather, corresponding to the shot number 2791 acquired by the marine streamer at line MC33 of CEVICHE project. Picks associated with refracted P wave are indicated in red. (b) Common shot gather corresponding to all shots of marine line MC33 acquired by the land station ITAT3 (profile P1). Picks associated with refracted P wave and deep reflections are indicated in red and blue, respectively. (c) Common shot gather, corresponding to the shot number 2984 acquired by the marine streamer at line MC18 of Crustal Experiment from Valdivia to Illapel to Characterize Huge Earthquakes project. (d) Common shot gather corresponding to all shots of marine line MC18 acquired by the land station TOLT3 (profile P3).

The 2D tomographic Vp models were generated by using the PROFIT algorithm (Koulakov et al., 2010). Inversions were performed with 10 iterations, horizontal and vertical spacing for velocity interpolation of 2 and 0.2 km, respectively. To control the amplitude of derived velocity anomalies and the velocity differences between adjacent nodes, we use a regularization parameter of 0.3 and a smoothing parameter of 5, respectively (see Koulakov et al., 2010 for details). Considering an estimated picking accuracy <0.05 s, this procedure resulted in models with an RMS misfit <0.1 s (RMS misfits for the starting models ranged from 1 to 2.5 s).

We generated 20 models (20 inversion processes) for each profile, starting from different initial models. Final Vp models were obtained by averaging the resulting velocity grids cell by cell (we also calculated the corresponding standard deviation at each cell). This “Monte Carlo” approach allows exploration of a broad solution space and the estimation of the inversion sensitivity under different initial conditions, by analyzing the corresponding standard deviation of the Vp (SDVp) solutions. Initial Vp models considered three depth boundaries: (a) the surface (bathymetry/tomography), (b) a boundary located 12 km below surface, and (c) a horizontal boundary at 40 km depth. Varying Vp at these boundaries in broad ranges (up to ±30% from average values), we generated 20 layering models whit different internal Vp–depth gradients (see details in the Supporting Information).

During the picking process, we noted that the onshore stations recorded reflected phases from marine shots at very long offsets (on the order of 100 km, see examples in Figures 3b and 3d). This suggests a deep source for these reflections, providing an opportunity to analyze the deep structure of the subduction zone. Motivated by the preliminary results of onshore data (P1–P4 in Figure 2), we searched for reflected P wave arrivals in all available land stations, including the available shots of SPOC-2001 lines registered by some stations deployed by ISSA-2000 project in 2001 (Krawczyk et al., 2006; Lüth et al., 2003). This previous information was organized in four additional profiles to study the reflected phases (P1b, P1c, P2b, and P2c shown in Figure 2).

Considering the inhomogenous subsurface coverage associated with the deep reflectors and the asymmetric directivity of the recorded phases, we conducted forward modeling of these observations to estimate the locations of the deep reflectors. Thus, we fit the travel times of direct and reflected phases using smooth velocity models based on our 2D tomographic inversions of profiles P1–P4 and on previous wide-angle seismic profiles in the area (Contreras-Reyes et al., 20082010). Following the approach of Zelt & Ellis (1988), we implemented a ray-tracing algorithm in MATLAB, considering bilinear velocity functions inside parallelogram elements that form several layers. A similar methodology has been used to locate upper plate reflectors in the northern Chilean area (Oncken et al., 2003).

4 Results

Figure 4 shows the velocity–depth models obtained by averaging the tomographic inversions along the profiles P1–P4. Lower panels in Figure 4 show the corresponding SDVp. Based on the ray coverage and the SDVp, we define a zone of acceptable resolution for each model (corresponding in general to values of SDVp larger than ∼0.2 km/s). This zone is used for further interpretations (magenta lines in Figure 4). We therefore conclude that the inversion of marine shots defines the velocity structure of the subduction zone down to ∼5 km below the seafloor. Nevertheless, by including the onshore recorded data, the penetration increases to about 10 km depth below the continental shelf and coastal area.

Details are in the caption following the image

Tomographic Vp–depth models obtained along profiles P1–P4. Upper panels show modeled picks (blue dot) and observed data (red dots). Middle panels shows the final Vp models obtained by the averaging of 20 inverted models (see main text for details). Cyan dots are the position of marine shots and land stations used in the inversions. Blue line corresponds to the slab geometry according to SLAB2.0 model (Hayes, 2018). DF indicates the position of deformation front. Green, and purple dotted lines indicate interpreted landward limits of the frontal and middle wedge units, while red dotted line corresponds to the interpreted limit between middle wedge unit 1 and 2 at profiles P3 and P4 (see text for details). Lower panels present the corresponding grids of standard deviation of Vp models (SDVp). Doted magenta lines are the interpreted limit of low resolution according to SDVp models and ray coverage.

At a regional scale, the continental wedge presents a clear pattern of landward increase of velocities. We identify three units, from the trench to the coast, defined by different landward velocity gradients: (a) a frontal unit of low P wave velocity with values ranging between ∼2.0 and ∼4.5–5.0 km/s, (b) a middle wedge unit with velocities between ∼4.5–5.0 and ∼6.0–6.5 km/s, and (c) relatively high velocities (>6.0–6.5 km/s) in the eastern portion of the continental shelf. In addition, a shallow unit of low velocities (<3.0 km/s) defines the continental shelf basin and slope basins with variable thickness.

Where they overlap, our model along profile P2 is consistent with a 2D Vp model presented by Contreras-Reyes et al. (2008) based on ocean bottom seismometer data and our model along profile P3 is a smoothed version of the Vp model presented in Bangs et al. (2020), which was derived by full waveform inversion of the Langseth streamer data.

Figure 5 shows the models of deep reflectors obtained by forward ray tracing. The figure presents the travel time data (observed and modeled) and the ray tracing of reflected phases and refracted phases observed in land stations (obtained RMS is ≤0.63 s). According to the models, the seismic signals generated by marine shots and recorded on land stations are mapping short segments of deep reflectors, which are extend throughout a broad portion of the marine forearc. These segments are located, in general, between 40 and 130 km landward from the deformation front. As is expected, some of these reflectors can be associated with main crustal/mantle discontinuities (continental and oceanic Moho as well as the interplate boundary). However, numerous reflectors are clearly seen within the upper plate at depths between ∼5 and ∼20 km. The modeled deep reflectors have variable geometry, ranging from subhorizontal to east-dipping inclination, and in several cases, a progressive increase of slope to the west, similar to a syncline fold.

Details are in the caption following the image

Deep reflectors modeled by 2D forward ray tracing. (a) Ray tracing along the profile P1. Upper panel presents the modeled travel times of reflected and refracted phases with magenta and cyan dots, respectively. Black dots are the observed data. Lower panel shows the ray tracing of reflected (magenta) and refracted (cyan) phases modeled to locate the segments of deep reflectors, which are highlighted with black lines. Red triangles indicate the position of land stations. The color grid shows the smoothed Vp model used for the ray tracing. (b–h) The ray tracing along the profiles P1b, P1c, P2b, P2, P2c, P3, and P4, respectively. All elements as in (a).

As is observed in Figure 5, the forward model of refracted phases (cyan dots and rays) provides constraints for the rays associated with the reflected phases, at least in the upper portion of the crust (to ∼10–15 km). In the deeper region, the uncertainty in the reflector position increases due to the tradeoff between depth and velocity. For this reason, we based the modeling on Vp profiles available near or inside the study area. However, during the modeling process, we observe that the velocity uncertainties associated with this tradeoff have a moderate magnitude. For instance, keeping unaltered the upper portion of the Vp models (∼10–15 km) to fit the refracted phases and including a completely unrealistic 1-D Vp model below, the position of the modeled reflectors located in the upper ∼20 km is altered less than 5 km vertically and horizontally, while only the deepest reflectors (located below) are shifted around 5–10 km in both directions (see an example in supplementary material). Thus, we estimate that the location of deep reflectors derived from Vp models presented here and in previous studies (Contreras-Reyes et al., 2008) have errors lower than ±5 km.

5 Interpretation and Discussion

5.1 Vp Structure of the Continental Wedge

As mentioned above, the continental wedge presents a pattern of landward increase of Vp characterized by three units, from the trench to the coast, defined by different landward velocity gradients. This structure is presented in Figure 6, where we plot the Vp values of tomographic models, along horizontal paths at constant depths of 4 and 7 km below the sea level (Figures 6a and 6b, respectively). In this analysis, we observed that the frontal low velocity unit, which extends from the deformation front to ∼40 km landwards, is characterized by a strong lateral eastward Vp gradient. This feature is probably related to a rapid landward compaction of the sedimentary units of the frontal accretionary prism. These features have also been observed by several authors in the region (Bangs & Cande, 1997; Contreras-Reyes et al., 200820102017; Moscoso et al., 2011). Recently, Bangs et al. (2020) presents a detailed interpretation of the CEVICHE multichannel seismic reflection profiles along profiles P3 and P4. At these latitudes, the frontal portion of the continental wedge (to ∼40 km from the trench) shows an atypical position of the décollement, which is located in the shallow strata of trench fill, allowing the subduction of a high fraction of incoming sediments (see also, Olsen et al., 2020). Above the décollement, sedimentary units show a deformation style similar to an antiformal stack interpreted to be due to underplating of subduction channel sediments to the base of continental wedge. In their model, the landward edge of the fontal accretionary wedge corresponds to a west-dipping interface (“backstop”) located ∼40 km from the trench, which is consistent with the change in Vp and lateral Vp gradient observed in our velocity models. The relation between the backstop position and Vp changes was observed by Bangs et al. (2020), who performed a full waveform inversion of Vp in the frontal portion of profile P3.

Details are in the caption following the image

Interpretation of continental wedge Vp segmentation in profiles P1–P4. (a) Color curves correspond to Vp values extracted from tomographic models at a constant depth of 4 km below the sea level. Segmented gray lines are interpretative trends of horizontal Vp gradients. (b) Color curves correspond to Vp values extracted from tomographic models at a constant depth of 7 km below the sea level. Segmented gray lines are interpretative trends of lateral Vp gradients.

To the east, we identify a middle wedge unit (or units), extended to 80–110 km landward from the deformation front, which is characterized by larger velocities (>4.5–5.0 km/s) and a smaller lateral velocity gradient. This can be interpreted as older lithified sedimentary materials (compared with the frontal unit) and/or a landward decrease in fracturing of the continental basement. Finally, to the east, the Vp models suggest the presence of a more competent/rigid basement unit with high Vp and small lateral variation. This upper wedge unit, is more evident in the deep region of the profiles, as can be observed in Figure 4 and by comparing Figure 6a with Figure 6b. Particularly, the limit between the middle wedge and upper wedge unit is not defined at profile P4 due to the ray coverage (Figures 4 and 6). However, the Vp values and Vp gradients in the middle wedge are similar to the observed at profile P3 (see red and green curves in Figure 6b).

Comparing the models along the different profiles, the continental structure shows an important along-strike variation. The high Vp zone of upper wedge seems to be closer to the trench in the northern profiles (Figures 4 and 6). Figure 6 shows also that the middle and upper wedge units of the northern profiles have systematically higher velocities in comparison with the southern profiles (at the same depth). If we define the marine continental wedge as the portion of upper plate that includes the active accretionary prism, adjacent to the trench, and the more compacted sedimentary units and/or fractured rocks within the middle wedge (Figure 4), we can observe that this marine continental wedge is wider along the southern profiles P3 and P4 (∼100–110 km wide) in comparison with the seismic models along the northern profiles P1 and P2 (∼80–90 km wide).

At lower scale, the middle wedge shows a different velocity structure in the northern and southern regions (Figures 4 and 6). While in profiles P1 and P2, this zone is characterized by an approximated uniform landward Vp gradient, the profiles P3 and P4 show relatively low velocities and low Vp gradients in the western segment (middle wedge unit 1 in Figure 6) and higher Vp and Vp gradient in a second segment to the east (middle wedge unit 2 in Figure 6). The limit between interpreted middle wedge units 1 and 2 is located ∼70–80 km landward from the deformation front. It is interesting to note that the Vp gradient of the middle wedge unit 2 (profiles P3 and P4) is similar to the lateral Vp gradient observed in the middle wedge at profiles P1 and P2.

A joint analysis of our velocity–depth models, position of deep reflectors, and the available geotectonic background is presented in Figure 7. In this figure, the reflectors located along the complementary profiles (P1b, P1c, P2b, and P2c) are displayed in profiles nearest to P1 or P2 and aligned with respect to the deformation front position (DF). According to the models, this portion of the Chilean margin shows interesting latitudinal changes. As discussed above, north of ∼38°S (profiles P1 and P2), the margin is characterized by a narrow continental wedge with high velocities in the middle and upper wedge. Below we discuss how this change correlates with several geological and physical changes along the margin.

Details are in the caption following the image

Joint interpretation of Vp models and segments of deep reflectors along the profiles P1 to P4. All information was horizontally aligned with respect to the deformation front (DF). Slab geometry model (SLAB2.0) of Hayes (2018). (a) Joint interpretation of Vp model at profile P1 and segments of deep reflectors located along profiles P1, P1b, and P1c. Note in Figure 3 that P1c is located far south from profile P1. Slab geometry model of Hicks et al. (2014) along profiles P1 and P1c are indicated in segmented cyan and green lines, respectively. (b) Joint interpretation of Vp model at profile P2 and segments of deep reflectors located along profiles P2, P2b, and P2c. Magenta lines correspond to upper and lower limits of the oceanic crust obtained by Contreras-Reyes et al. (2008). Segmented blue lines are the main reflectors identified in the image behind, which corresponds to the onshore reflection seismic section presented by Krawczyk et al. (2006). Black dots are the upper plate seismicity located by Haberland et al. (2006) and segmented red line is the geometry of Lanalhue Faut Zone (LFZ) inferred by Ramos et al. (2018). (c) Joint interpretation of Vp model and segments of deep reflectors located along the profile P3. (d) Joint interpretation of Vp model and segments of deep reflectors located along the profile P4.

Onshore, the outcrops of the Permo-Triassic paired metamorphic belt change position relative to the coastline. In the southern region (profiles P3 and P4), the metamorphic belt extends farther to the east than it does in the northern region (Figures 1 and 7). It is important to note that at the latitude of profile P4 the Permo-Triassic metamorphic units conform the basement of the Central Depression basin, as have been observed by drill campaigns and seismic profiles (i.e., the paired metamorphic belt extends, at least, 50 km landward from coastline in the profile P4; Jordan et al., 2001). This landward extension of the paired metamorphic correlates with a wider marine continental wedge and a decrease of the middle wedge velocities. This suggests a possible interplay between the wedge structure and surface geology.

Assuming a seaward continuation of the metamorphic belt outcrops, we can interpret the middle wedge unit in profiles P1 and P2, and middle wedge unit 2 in profiles P3 and P4, as the seaward extension of the Permo-Triassic paired metamorphic belt. The middle wedge unit 1 could be indicative of a younger unit that was not accreted farther north. In this scenario, the western limit of the Permo-Triassic metamorphic belt presents a shift of ∼80 km to the east, southward from ∼38°S that is interpreted as the same arcward rotation that is observed in the onshore outcrops of the metamorphic belt and Permian Batholith (Figures 1 and 7). On the other hand, the middle wedge unit 1 in profiles P3 and P4 can be interpreted, as a frontal and fractured portion of the same metamorphic belt. In that case, their low Vp and low landward Vp gradient could be associated with deformation process due to the kinematics of the Chiloé microplate (Melnick et al., 2009), or deformation associated with changes in internal and basal frictional properties of the continental wedge (Maksymowicz et al., 2015). Both scenarios for middle wedge unit 1 should be related to a lithological/rheological change along the interplate boundary.

The obtained models are consistent with what is observed to the north and south of the study region and show a consistent along-strike segmentation of the continental wedge structure (at least for the upper 10 km covered by our models). From 34°S to ∼35°S, three wide-angle seismic profiles show a structure that is similar to profile P1 (at 36°15'). All of the profiles are characterized by a high-velocity unit (>6 km/s) starting about 70 km landward from the trench that lies east of a middle wedge characterized by velocities between 5 and 5.5 km/s (Contreras-Reyes et al., 2017; Moscoso et al., 2011). To the south (at ∼43°S), the wide-angle seismic profile presented by Contreras-Reyes et al. (2010) show a velocity structure similar to our results from profile P4 with low velocities in the middle wedge.

Between 36°S and ∼38°S (profiles P1 and P2), velocity values higher than 4.5–5.0 km/s are reached at a distance that is ∼10 km closer to the trench than those observed in the northern region (34.5°S to ∼35°S, Contreras-Reyes et al., 2017; Moscoso et al., 2011). This observation is consistent with a general density decrease of the marine continental wedge at the high slip patch of the 2010 Maule Mw8.8 rupture zone (Maksymowicz et al., 2015). South of 39°S, the general decrease of the velocities below the middle slope is also consistent with the low marine gravity anomaly that correlates with the high slip patch of the Mw9.5 Valdivia earthquake (Wells et al., 2003).

5.2 Geometrical Pattern of the Modeled Deep Reflectors

The sparse distribution of the obtained deep reflectors segments is related to the spacing of the available onshore stations causing large offset between sources and stations and heterogeneous signal/noise ratio obtained at different seismic stations. According to this, the identified reflected phases related to a specific deep reflector are observed mostly in single stations. For this reason, only a limited portion of the deep subduction structure can be determined from the reflection data and the lateral continuity between the segments cannot be unequivocally established. However, the observation and interpretation of these sparse reflective phases are relevant taking into account the scarcity of information about the deep structure of the continental crust in subduction margins. In this context, our objective is the interpretation of the general deformation style of the continental wedge, describing the first-order pattern rather than a detailed interpretation of the specific reflectors.

Comparing with seismological models of slab geometry available in the area (SLAB2.0 model of Hayes, 2018) and interplate boundary obtained by Hicks et al. (2014), the results show evidence for the location of the subducted slab (top of oceanic crust and/or oceanic Moho) at different latitudes. On profile P2 (Figure 7b), the slab reflectors show excellent consistency with the wide-angle seismic model obtained by Contreras-Reyes et al. (2008), both of which indicate an increase in the subduction angle at ∼20 km of depth (∼5 km shallower than the SLAB2.0 model of Hayes, 2018 in the marine portion). Such bends in slab geometry are observed at different latitudes of the Chilean subduction zone, and in several margins worldwide. Potentially they can alter the stress field and seismotectonics of the megathrust (Contreras-Reyes et al., 2012; Dinc et al., 2010; Kyriakopoulos & Newman., 2016; Maksymowicz et al., 2018; and references therein). Considering the uncertainties of our forward modeling (lower than ±5 km in reflector position), this new evidence provides motivation for future work to define the slab geometry more precisely and determine its impact on the coseismic rupture of large earthquakes in the study zone.

Figure 7b shows an interpretation of the reflectors observed in the seismic experiment presented by Krawczyk et al. (2006) and Ramos et al. (2018), which is the only deep seismic section generated in the onshore forearc of the study zone and is located along the track of profile P2. Krawczyk et al. (2006) interpret the general structure of the section as an antiformal stack associated with the units of the Permo-Triassic paired metamorphic belt, and below these units, the presence of a thick zone of the subduction channel in the western portion of the profile. As we can see, there is a clear coherence between the deep reflectors observed in P2 (blue lines in Figure 7b) and the interfaces observed in the onshore section (segmented blue lines in Figure 7b). Furthermore, most of the deep reflectors observed in the other profiles show similar characteristics to the observed by Krawczyk et al. (2006), which suggests these deep structures are associated with the Permo-Triassic paired metamorphic belt, and that this association extends along a broad portion of the Chilean margin.

It is interesting to note that the modeled deep reflectors are observed within and/or below the middle wedge and upper wedge Vp units (Figure 7). Thus, according to previous interpretations, the deep reflectors could be mapping the deep interior of the Permo-Triassic paired metamorphic belt or younger units located below. The observed reflectors could also be associated with more hydrated structures in response to recent activity (at geologic time scale). Haberland et al. (2006) located upper plate seismicity around profile P2 (Figure 7b) in the zone characterized by high velocities (Vp > 6–6.5 km/s), at depths where reflectors are observed, suggesting that the intraplate seismic activity is controlled partially by the inherent structure within (or below) the Permo-Triassic paired metamorphic belt.

To complement a regional structural analysis, in Figure 8a, we project all segments of deep reflectors according to their position with respect to the deformation front. Because of the large distance between seismic profiles, the objective of this analysis is to study the structural style of the continental wedge at large scale, rather than the north-south continuity of specific reflectors. As is observed, the general geometry of the deep reflectors set is similar to a syncline fold, where deep layers of the crust are progressively uplifted westward. This kind of structural style is obtained in analog and numeric models (Gutscher et al., 1996; Menant et al., 2019; Perrin et al., 2013) and has been observed in Japan, New Zealand, and others subduction margins (Bassett et al., 2010; Calvert et al., 2011; Kimura et al., 2010; Moore et al., 1991), where the subducted material is progressively accreted to the top of the subsection channel, forming a basal accretionary complex. Accordingly, the observed structural style is consistent with the presence of an underplated units above the interplate boundary (Figure 8b), which are not observed here due to limited distribution and resolution of deep reflectors segments. These underplated units could be formed by active basal accretionary processes in the current subduction channel (as was interpreted by Krawczyk et al., 2006) and/or by deep and old units that were basally accreted to the western portion of the Permo-Triassic paired metamorphic belt (Glodny et al., 2005; Willner et al., 2005). Both cases imply changes in lithology and rheology along the dip of the seismogenic contact.

Details are in the caption following the image

Joint interpretation of deep reflectors. (a) The observed segments of deep reflectors are plotted according to their distance from the deformation front. Black line corresponds to the bathymetry of profile P1 as a reference for continental wedge geometry. Dotted blue lines show the interpretations of the deep reflectors observed in the onshore seismic section presented by Krawczyk et al. (2006) at the latitude of profile P2. (b) Interpretative schematic of the deep reflectors. Progressive seaward uplift of the deep crustal units is interpreted as a response to the presence of basal accretionary complexes (see main text for details).

Several studies of the Permo-Triassic paired metamorphic belt in the zone propose an intense basal accretionary process during Triassic times (Glodny et al., 2005; Willner et al., 2005). However, this accretionary process is older than the Andean tectonic cycle, and several geodynamic processes have modified the margin since Triassic to reach the current morphology (Charrier et al., 2007), including the uplift of the Coastal Range. In fact, the large period of subduction erosion experienced by the Chilean margin since Mesozoic times (Kukowski & Oncken, 2006) suggests that the metamorphic units of the Coastal Range (and the deep reflectors observed below) were initially in a distal position relative to the trench (mainly below extensional basins, Becerra et al., 2013). During the large period of subduction erosion, these units would have progressively approached to the trench (and to the interplate boundary) and uplifted as an effect of the contractional reactivation of the older structure and/or the basal accretion of younger sedimentary units. A local intensification of this process could promote the recent uplift of the AP.

As we can see in Figure 8a, a general view of the deep reflectors segments suggests a lateral shift of the structures, where the northern profiles show similar geometries to those in the south but have formed closer to the trench. In fact, taking as a reference the results at profile P2 (blue lines in Figure 8) complemented with the interpretation of onshore seismic profile (segmented blue lines in Figure 8), the position and geometry of the deep reflectors at profiles P1, P1b, P1c, and P2b seem to be displaced toward the trench. In contrast, the reflectors observed at profiles P2c, P3, and P4 suggest a continental structure similar to the observed at the latitude of P2. Thereby, this structural change is observed around the AP at ∼38°S. Considering the presence of a basal accretionary complex as a possible explanation of the observed structural style, the trench-ward shift of the structures in the northern portion of the margin could be associated with a trench-ward shift of the basal accretionary units (this interpretation is summarized in Figure 8b). As was mentioned above, the basal accreted units are not directly observed in this study, but the geometry of modeled reflectors motivates future studies to understand their potential role on the frictional properties along the seismogenic contact.

5.3 Continental Wedge Structure and Megathrust Earthquake Ruptures

In the short-term (at seismic cycle time scale), the latitudinal and/or downdip variation of the continental wedge structure has important mechanical implications. Comparing the results obtained in the southern segment of the Mw8.8 Maule earthquake (∼36°S–38°S) with the northern segment of Mw9.5 Valdivia earthquake (∼38°S–42°S), the first segment is characterized by relatively high velocities in the middle wedge and a position of syncline geometry of deep reflectors closer to the trench, in comparison to the second segment. These variations in the middle wedge velocities, and structural style, may reflect differences in age, lithology, fracturing, fluid content, and density of the upper plate. These different physical properties should determine variations in the interseismic and coseismic stress/strain behaviors of the upper plate (e.g., rigidity) and changes in the frictional properties of the interplate boundary (Bassett et al., 2016; Calahorrano et al., 2008; Dahlen, 1984; Maksymowicz et al., 2018; Sallarès & Ranero, 2019).

Applying the NCCW theory (Dahlen, 1984), Maksymowicz (2015) show that the continental wedge geometry (including only the lower slope and middle wedge) is consistent with a latitudinal segmentation of the effective basal friction coefficient (μb*) along the margin. In particular, μb* is lower in the northern portion of the Maule earthquake rupture zone (north of ∼35°30'S) and in the northern area of the Valdivia earthquake rupture (∼38°S–42°S) than in surrounding areas. Then, considering this simple and homogenous model of the continental wedge (NCCW), μb* decreases in segments characterized by patches of high slip during these two mega earthquakes. The physical relation between frictional parameters, as μb*, and basal accretion is an unresolved problem, but recently, the numerical models presented by Menant et al. (2019) suggest that the areas with basal accretion are associated with local zones of high friction along the interplate boundary. Considering this, the interpretation of a basal accretionary complex closer to trench in the southern segment of the Mw8.8 Maule earthquake, in comparison to the northern area of the Valdivia earthquake rupture zone (Figure 8b), is consistent with the estimated change of μb* between the two segments.

Beyond the potential presence of the basal tectonic accretion, the physical variations observed along the margin (Vp velocities and reflectors) highlight the role of the upper plate structure on the seismotectonic segmentation. In particular, we propose that an increase in fracturing/porosity (or density decrease) in the middle wedge south of ∼38°S is linked to the lateral extension of Permo-Triassic metamorphic units. This change, added to a different structural style in the southern region could control, at least in part, the normal stresses and frictional properties of the interplate boundary, generating a segmentation of seismic behavior between the southern region of the Mw8.8 Maule earthquake and the northern region of Mw9.5 Valdivia earthquake. It is worth noting that, the interplate locking rate derived from inversion of GPS data before the 2010 Mw8.8 Maule earthquake also shows a change around 38°S (M. Moreno et al., 2011). According that work, in the southern segment of Maule earthquake, the upper limit of coupled region extends ∼50 km closer to the trench compared to the southern segment (i.e., in the northern area of the Valdivia earthquake). Additionally, there is evidence to support that the southern segment of the Maule earthquake slipped independently of the other adjacent segments during the 1835 and reactivated in 2010 jointly with the rupture area of the 1928 earthquake located to the north (Ruegg et al., 2009). The correlation between the physical properties of the continental wedge (seismic velocity and geometry) and the rupture area of large earthquakes suggests a causal link between short-term deformation and changes of rheology/lithology along the interplate boundary, determined in turn, by the long-term deformation process.

6 Conclusions

  1. Along the south-central Chilean margin, we observe a middle wedge unit of relatively high seismic P wave velocity (between ∼4.5–5.0 and ∼6.0–6.5 km/s), that we interpret as the seaward continuation of the Permo-Triassic metamorphic complex.

  2. Offshore, in the southern segment of the 2010 Mw8.8 Maule earthquake, these old units seem to extend westward to the boundary with the frontal sedimentary wedge (∼40 km landward from the trench). In the northern region of the 1960, Mw9.5 Valdivia earthquake rupture zone these units extend only to ∼80 km landward from the trench, suggesting the presence of younger units and/or an increase in fracturing within a middle continental wedge province that is absent to the north.

  3. Deep reflections generated at a range of depths in the lower crust of the forearc suggest the presence of structures at the base of the Permo-Triassic metamorphic complex. An along-strike variation of the deep structural style suggests a segmentation in the interplate frictional properties between the Maule and Valdivia earthquakes rupture zones.

  4. The along-strike segmentation of lithological, rheological, and structural properties within and at the base of the continental wedge should control (at least partially) the mechanical behavior of the margin during the long-term deformation process, and also at the time scale of large earthquake sequences.

Acknowledgments

This work was funded by CONICYT/ANID under the Chilean Fondo Nacional de Desarrollo Científico y Tecnológico (FONDECYT), grant 11170047. Cruise MGL1701 of the R/V Marcus G. Langseth was funded by US National Science Foundation grants OCE-1559293 and OCE-1558867 to N. Bangs and A. Tréhu. We are deeply grateful to Ivan Koulakov for his valuable software (PROFIT) assistance. We also thank the support of CONICYT/ANID- PIA/Anillo de Investigación en Ciencia y Tecnología ACT172002 project “The interplay between subduction processes and natural disasters in Chile.” Comte thanks ANID Project AFB180004 and FONDECYT 1161806. Maksymowiz and Contreras-Reyes acknowledge the partial support of ANID FONDECYT grant 1210101 and 1170009.

    Data Availability Statement

    The readers can access our seismic results at the following link: https://osf.io/x4he3/?view_only=3b56aee748e94203a03e6a5296203c41.