Seamount Subduction and Earthquakes

. Seamounts are ubiquitous features of the seafloor that form part of the fabric of oceanic crust. When a seamount enters a subduction zone, it has a major affect on forearc morphology, the uplift history of the island arc, and the structure of the downgoing slab. It is not known, however, what controls whether a seamount is accreted to the forearc or carried down into the subduction zone and recycled into the deep mantle. Of societal interest is the role seamounts play in geohazards, in particular, the generation of large earthquakes.

(b) Perspective view showing the trench-parallel fault that splits the seamount vertically and offsets its once-flat top into two gently tilted surfaces, A and B. Based on swath bathymetry data in Lallemand, et al. (1989), Dominguez et al. (1998) and http://www. utdallas.edu/~rjstern/pdfs/TrenchProof.pdf The red-filled star shows the approximate location of the 1982 M w = 7.0 earthquake (Mochizuki et al., 2008). (c) Deep seismic structure (Nishizawa, et al. 2009). The light brown contour lines show the P-wave velocity in km s -1 . The estimated extent of the surface volcanic load and underlying flexed oceanic crust are shown as brown and grey shaded regions, respectively. 166 This article has been published in Oceanography, Volume 23, Number 1, a quarterly journal of The Oceanography Society. © 2010 by The Oceanography Society. All rights reserved. Permission is granted to copy this article for use in teaching and research. Republication, systemmatic reproduction, or collective redistirbution of any portion of this article by photocopy machine, reposting, or other means is permitted only with the approval of The Oceanography Society. Send all correspondence to: info@tos.org or Th e Oceanography Society, PO Box 1931, Rockville, MD 20849-1931 affect the morphology, structure, and vertical motion history of the forearc region between the trench axis and the volcanic arc. Seamount subduction may also influence the degree of coupling between the overriding and subducting plates and may affect the seismicity, especially the size and frequency of large earthquakes (Kelleher and McCann, 1976). Recently, Nishizawa et al. (2009) and Das and Watts (2009) demonstrated that seamounts play a major role in controlling the rupture history of large earthquakes, in particular, acting as either barriers (e.g., Kodaira et al., 2000) or as asperities (e.g., Husen et al., 2002) (Mogi and Nishizawa, 1980) show that it is bisected by a steep, trench-parallel normal fault that has vertically offset its flat top by ~ 1.5 km ( Figure 1). Similar-trending faults have been reported from other seamounts in the vicinity of the Mariana-Izu Bonin trenches (Fryer and Smoot, 1985) and Tonga-Kermadec trench (Coulbourn et al., 1989). Only a few deep seismic studies of trench seamounts exist, but recent surveys over Daiichi-Kashima (Nishizawa et al., 2009)

INTRODUCTION
The seafloor is littered with seamounts, most of which are small (Hillier and Watts, 2007). Satellite-derived gravity anomaly data, however, suggest that there maybe as many as 12,000 large seamounts that rise > 1.5 km above the depth of the surrounding seafloor (Wessel et al., 2010). Irrespective of whether these seamounts are growing up during their volcanic construction or are sinking, once they become inactive, they are being carried along by plate motions and will eventually be subducted (Staudigel and Clague, 2010). Indeed, there are many present-day examples of seamounts that are in the throes of being subducted.
Niue Seamount (DuBois et al., 1975) in the Southwest Pacific Ocean, and Christmas Seamount (Woodroffe et al., 1990) in the Indian Ocean, for example, are being uplifted as they ride the flexural bulge seaward of the Tonga and Java-Sumatra trenches. Osbourn (Lonsdale, 1986)

FOREARCS AND BURIED SEAMOUNTS
Many forearcs comprise a thick sequence of highly deformed sediment (the so-called accretionary wedge) that has been scraped off the subducting plate and structurally deformed by folding and thrusting. Others, however, have less sediment, and the depth to the crystalline basement is relatively shallow.
A good example of a subducted seamount is Muroto in the Nankai accretionary wedge offshore Southwest Japan.
Seismic data show that this seamount is ~ 50-km wide at its base, ~ 25-km wide at its top and 4-km high, and is underlain by flexed oceanic crust (Kodaira, et al., 2000). Other forearcs where seamounts have been imaged include Mediterranean Ridge (von Huene et al., 1997), Cascadia (Wells et al., 2009), and Mariana (Oakley et al., 2007(Oakley et al., , 2008 Quepos Earthquake M w = 6.9, 1999 Figure 2. Perspective view of the Costa Rica margin showing the morphology of the subducting Cocos oceanic plate and the overriding North American Plate, constructed using swath bathymetry data acquired onboard MV Sonne. Seamounts on the subducting plate align with landslide scars on the overriding plate, suggesting that subducting seamounts are able to "plow through" the forearc and generate large-scale slumps and slides. The red-filled stars show the projected locations of 1990 M w = 7.0 and 1999 M w = 6.9 earthquakes (Bilek et al., 2003). Based on von Huene et al. (2000) In addition to modifying the upper shallow part of the forearc, seamount subduction may modify its lower deep part (Dominguez et al., 1998). For example, it may cause tectonic erosion, steepening any overlying thrusts and folds and releasing sediments and fluids into the underlying subduction zone (Bangs et al., 2006), and it may modify the subsidence and uplift history of a forearc (Lallemand and Le Pichon, 1987;Lallemand et al., 1989;Dominguez et al., 1998;Oakley et al., 2008). A seamount chain, rather than an isolated seamount, might induce a "deformation wave" that alternately inflates and deflates a forearc as the seamounts are progressively carried into the subduction zone by plate motion (Laursen et al., 2002).

Thickness of the Subduction Channel
Seismic images of the subduction channel off Ecuador (Sage et al., 2006) show a layer of poorly consolidated and intensively sheared sediment that has been dragged down by the subducting plate beneath the overriding plate. Cloos and Shreve (1996) suggest that it is the ratio of the relative thickness of the sediments within the channel to the height of a seamount that determines its fate. If the seamount is high relative to channel thickness, it might be decapitated at shallow depths (Figure 3a), whereas if it is low relative to channel thickness, it may be carried some distance and decapitated at greater depths when it jams up against the roof of the channel (Figure 3b).

Relative Strength of the Subducting and Overriding Plates
It has long been accepted that the negative buoyancy (i.e., sinking) of oceanic lithosphere initiates subduction (see Koppers and Watts, 2010

Trench Axis
Trench Axis DEEP DECAPITATION Figure 3. Schematic diagram illustrating the Cloos and Shreve (1996) model for seamount subduction. (a) Mariana-type case where a relatively large seamount is carried into a subduction zone with a relatively thin subduction channel. The seamount is truncated at relatively shallow depths and, hence, low confining pressures, so there is little or no seismicity. (b) Chilean-type case where a relatively small seamount is carried into a subduction zone with a relatively thick subduction channel. The seamount, in this case, is truncated at relatively deep depths and high confining pressures, which leads to seismicity.

Internal Structure of Subducted Seamounts
Seismic studies reveal that the volcanic edifice of a seamount builds up and out on a gently flexed, uniformly layered, oceanic crust. It is easy to envisage how such a structure would deform when it reaches a trench because the flexed layers of the edifice and underlying crust will be rotated and sheared as they enter a subduction zone. Deformation may be facilitated by the fact that the interface between the edifice and oceanic crust may already have acted as a décollement surface during volcano growth (Got et al., 2008). Some seamounts and oceanic islands, however, have complex internal structures with dense volcanic cores (e.g., Tenerife; Canales et al., 2000)

Buoyancy of Subducted Seamounts
Oceanic flexure studies show that seamounts are compensated differently at depth, depending on whether they formed on a weak plate near a midocean ridge or on a strong plate in a plate interior (see Koppers and Watts, 2010).
As a result, seamounts will be in different states of crustal buoyancy as they enter trenches ( Figure 4). For example, a seamount on a strong plate would be regionally supported and, hence, have had much of its compensation removed before its arrival at a trench. Because it is less buoyant, such a seamount would be less likely to lift up the forearc and, thus, more weakly coupled to the overriding plate. A seamount formed on a weak plate, however, is more locally compensated and so retains more of its buoyancy and is more likely to jam a subduction zone.

SEAMOUNTS AND THE RUPTURE HISTORIES OF LARGE SUBDUCTION ZONE EARTHQUAKES
It has been more 35 years since Kanamori (1971) first suggested that large earthquakes (i.e., M w > 8.0) at convergent plate boundaries reflect the degree of coupling between the overriding and subducting plates. In his view, some arc systems were strongly coupled (e.g., Aleutian) and generated large earthquakes when they finally slipped, while others (e.g., Izu-Bonin and Mariana) were weakly coupled and large earthquakes were either rare or absent.   . Simple model for the crustal structure of two seamounts-one formed offridge and the other on-ridge-that are about to enter a subduction zone. (a) Off-ridge seamounts are more regionally compensated so that much of their support would have already been removed by subduction. These seamounts are therefore less buoyant (small orange arrow) and thus less well coupled to the forearc when they enter a subduction zone. (b) On-ridge seamounts are locally more compensated so that little of their support would have been removed. These seamounts are therefore more buoyant (large orange arrow) and thus better coupled to the forearc. Kelleher and McCann (1976) noted, however, that many arc-trench systems were highly segmented with regard to their seismicity. Some segments were active on historical time scales, while others were not (so-called "seismic gap").
Because the seafloor varies spatially in roughness, Kelleher and McCann (1976) suggested that there might be a link between large earthquakes and the morphology of the subducting oceanic plate. In particular, they noted that the size and frequency of large earthquakes were reduced in regions of thick oceanic crust where aseismic ridges and other "bathymetric highs" (i.e., seamount chains) were about to enter a subduction zone. Cloos (1993) argued that only the thickest continental "blocks" (30 km and thicker) would be buoyant enough to resist subduction. However, Cloos and Shreve (1996) pointed out that large seamounts could be subducted to great enough depths to cause an earthquake if the sediment in the subduction channel is sufficiently thick (Figure 4). In such a case, subducted seamounts act as a strong asperity (or sticking point), such that the fault ruptures sporadically when frictional resistance is overcome. Scholz and Small (1997) Kodaira et al. (2000) used an integrated data set of earthquake aftershock relocations, seismic refraction, and swath bathymetry data to suggest that seamounts acted as a barrier to the rupture zone that generated the 1946 M w = 8.2 Nankai earthquake, while Husen et al. (2002) suggested that seamounts acted as an asperity during the 1990 M w = 7 Gulf of Nicoya earthquake. Bilek et al. (2003) (Das and Kostrov, 1990;Robinson et al., 2006 Das and Kostrov (1990) and Das and Watts (2009). The red-filled star shows the region where the rupture initiated. The arrow shows the convergence direction between the Pacific and Eurasian plates. The solid black lines show the bathymetry along a trench-parallel profile constructed from the GEBCO grid. The figure shows that the region of highest slip during the earthquake aligns with a number of small seamounts that rise up to ~ 800 m above the depth of the surrounding seafloor.