2.3.1
Complex interplay between upper and lower plates
Thus far, we have considered the roles played by each structural domain separately. In reality, the seismic response of the megathrust results from the complex intertwining and feedback between many geological structures. For example, fracture zones can dam axial turbidity flows (Cloos and Shreve, 1988), affecting the characteristics of the subduction channel at depth. Furthermore, Dean et al. (2010) and Geersen et al. (2013) show that a subducting fracture zone at the Sumatra trench causes along-strike variations in the composition and thickness of the subduction channel. These variations in sediment properties may result in segmentation of the margin due to differences in thermal structure and frictional characteristics. Like fracture zones, seamounts and horst-and-graben features on the seafloor can also strongly influence
2.3 Limitations and remaining questions
sediment flux into the subduction zone (Kopp, 2013). In this case, such topographic anomalies enhance erosion of the upper plate (von Huene and Ranero, 2003; Scholl et al., 1980). Therefore, subduction of seafloor topography may shallow the up-dip limit of seismogenesis, resulting in a megathrust that is more conducive to long-duration tsunami earthquakes (e.g. Bell et al., 2014). Finally, in interpretations of upper plate structures, it is generally assumed that forearc morphological features, such as peninsulas are directly related to deeper structures within the forearc. However, workers have suggested that ppeninsulas may result from deeper processes, causing uplift of the forearc. For example, coastal features along the Kamchatka margin may be related to ridges and fracture zones on the oceanic plate (Bürgmann et al., 2005); the Raukumara peninsula in New Zealand may be caused by underplated sediment (e.g. Bassett et al., 2010).
In a reversal of roles, the subduction channel may affect the integrity and structure of subducted seamounts. Cloos and Shreve (1996) argue that the geometry and thickness of the subduction channel determines the depth at which seamounts may truncate or fully decapitate. At erosive margins, where the subduction channel is thin at shallow depths, early truncation occurs against the strong upper plate framework. On the other hand, at accretionary margins, seamounts remain intact to greater depths, where confining pressure is higher, leading to a strong, highly-coupled plate interface. Moreover, Ruff (1989) suggests that thick subduction channels smooth out roughness along the megathrust, allowing for larger rupture areas to develop. Geometrical complexities on the plate interface can damage the base of the upper plate, generating networks of faults at the base of the overlying crust that affect the mechanical behaviour of the megathrust (Armijo and Thiele, 1990; Melnick et al., 2009). As well as seamount subduction, rupture on upper plate faults may enhance subduction erosion through mass wasting on the forearc (Tsuji et al., 2013).
The complex feedback processes between sediment, subducting topography and upper plate faulting described above is well illustrated by the central Ecuador subduction margin (Sage et al., 2006); (Figure 2.4). Here, seamount subduction enhances erosion of the outer wedge, which in turn, increases sediment flux to the
Chapter 2: Upper Plate Versus Lower Plate: Physical Controls on Megathrust Earthquake Rupture Processes coupling allows the development of seaward-dipping normal faults in the overlying forearc (Figure 2.4).
Figure 2.4: Schematic view of the main structural features of the central Ecuador subduction zone showing interplay between fluid-rich sediment lenses, seamounts, upper plate weakening, and plate coupling. Redrawn after Sage et al. (2006).
2.3.2
Time-dependence of the seismic cycle
As we have shown, the seismogenic behaviour of the megathrust depends on the complex interplay between many geological structures. Moreover, we have assumed the seismogenic response of these structures remains constant over time. Compared with the characteristic timescale of the seismic cycle (~100s of years), interseismic locking distributions provide a time-limited snapshot of plate interface coupling. It is often assumed that the degree of locking is long-lived; however, studies have shown that locking can rapidly change with time. For example, interseismic strain in the north-east Japan subduction zone rapidly increases (Avouac, 2011) and decays (Nishimura et al., 2004) because of nearby moderate–large earthquake ruptures. Furthermore, palaeogeodetic observations from the Sumatra subduction zone suggest that strong locking builds up within a matter of decades before a large megathrust earthquake (Meltzner et al., 2015). This time-dependent coupling may result from transient phenomena such as slow slip and seismic tremor. We do not discuss such transient processes in this thesis, but other detailed reviews exist on this topic (e.g. Beroza and Ide, 2011; Schwartz and Rokosky, 2007).
2.3 Limitations and remaining questions
Although plate interface coupling distributions are related to physical properties of the megathrust region, time-dependence of these properties should not be overlooked. The seismogenic character of a geological structure depends on the magnitude of dynamic shear stress at the time and elastic energy carried by the tip of the propagating rupture (Müller and Landgrebe, 2012). Consequently, the seismogenic response of structural anomalies not only depends on static frictional properties, but also on time-dependent stochastic variations in friction and shear stress (dynamic heterogeneity). These factors depend on past coupling and rupture history of a portion of the plate interface; this may explain why some structures may exhibit dual behaviour, changing their seismogenic character from one seismic cycle to another (Kodaira et al., 2006).
2.3.3
Moving toward finer scale structures in the upper plate
As described above, there have been many studies focussing on the seismogenic character of subducted seamounts. We believe that such large attention to seamounts has arisen as a result of two main factors. First, based on ocean bathymetry, seamounts clearly enter subduction zones, playing a vital role in large earthquake ruptures. Second, the large size of seamounts allows for their well-defined identification in subsurface images. Examples of seamounts acting as asperities tend to be historic earthquakes, such as the 1946 Nankaido (Kodaira et al., 2000), the 1947 Poverty Bay, New Zealand (Bell et al., 2014) and 1990 Nicoya, Costa Rica earthquakes. Due to the time when these earthquakes occurred, they have weakly constrained slip distributions; therefore, in our view of these examples, it is not easy to pinpoint an asperity to a comparatively small seamount. Higher-resolution rupture models are required to test the seamount asperity theory. Furthermore, for recent, well-recorded megathrust earthquakes, there has been little evidence to support the presence of subducted topographic heterogeneity within high slip regions. For instance, in the maximum slip area of the 2011 Tohoku, Japan earthquake, there is little evidence to suggest that the presence of seamounts (Wang and Bilek, 2014).
There appears to be a growing trend that as the resolving capability of seismic source inversions and subsurface imaging improves, attention is shifting toward finer-
Chapter 2: Upper Plate Versus Lower Plate: Physical Controls on Megathrust Earthquake Rupture Processes has been documented on several occasions, there is no evidence for rapid stress transfer between upper plate faults and the megathrust. This may arise because faulting networks in the overriding plate are likely more complex than those in the oceanic plate, which has relatively uniform characteristics between different subduction zones. The structure of the upper plate is dependent on unique geological histories, so it is difficult to find systematic behaviour of overriding plate structures between subduction zones. Looking ahead, greater geophysical datasets and finer resolution images will help to resolve the detailed structure of the overriding plate.