We now briefly outline the layout of this thesis, which comprises a total of nine chapters, three of which are research papers reformatted for this thesis. Therefore, where appropriate, we describe the current publication status of each chapter and the contributions made by co-authors. The formatted versions of the published papers are enclosed with this thesis.
We begin by summarising current knowledge on megathrust physical heterogeneity. We review studies of subduction zone structure and past megathrust earthquakes to describe the range of physical structures that may affect seismogenic processes along the megathrust. Beyond Chapter Two, the thesis is divided into two main parts to reflect the multiple phases of study and key aims defined in the previous section. The first of these is passive seismic imaging of the subduction zone using ray- based methods; the second focuses on probing earthquake source complexity using full-waveform approaches. These topics each begin with a chapter (Chapters Three and Six) that outlines the background theory and past application of the two methods, with a particular focus on subduction zones.
1.4 Thesis organisation and publication status of chapters
In Chapter Four, we describe and interpret a preliminary seismic velocity model for central Chile based on a dataset of automatically determined seismic wave onset times. One of the key observations made in this chapter is the presence of a large, high seismic velocity anomaly lying beneath the coastline in the centre of the rupture area. We interpret this structural anomaly as to its control on the nucleation and rupture process of the Maule earthquake. This chapter was accepted for publication in Geophysical Research Letters on 14 September 2012 and published on 16 October 2012. Full citation: Stephen P. Hicks, Andreas Rietbrock, Christian Haberland, Isabelle M.A. Ryder, Mark Simons, and Andrés Tassara, The 2010 Mw 8.8 Maule,
Chile earthquake: Nucleation and rupture propagation controlled by a subducted topographic high, Geophysical Research Letters 39, L19308, doi: 10.1029/2012GL053184. Andreas Rietbrock gathered the preliminary dataset of P- and S-wave onset times and gave comments on the interpretation and discussion. Christian Haberland assisted with the analysis of the full model resolution matrix. Isabelle Ryder assisted with the comparison between seismic velocity structure and geodetic models of the Maule rupture. Mark Simons assisted with the interpretation. Andrés Tassara provided the raw Bouguer gravity model.
Based on main features of the preliminary tomographic images presented in Chapter Four, we describe an updated seismic velocity model in Chapter Five. This model is based on a higher quality seismic travel time dataset and incorporates offshore data. Compared to that of the previous chapter, our new velocity model has sharper images that allow for detailed probing of the shallow plate interface. This improvement allows for a more in-depth interpretation of absolute seismic velocities. We make direct comparisons between physical properties and megathrust behaviour at different stages of the seismic cycle. This chapter was accepted for publication in Earth and Planetary Science Letters on 26 August 2014 and published on 1 November 2014. Full citation: Stephen P. Hicks, Andreas Rietbrock, Isabelle M.A. Ryder, Chao-Shing Lee, Matthew Miller, Anatomy of a megathrust: The 2010 M8.8 Maule, Chile earthquake rupture zone imaged using seismic tomography, Earth and Planetary Science Letters, Volume 405, 1 November 2014, Pages 142-155, ISSN 0012-821X, doi: 10.1016/j.epsl.2014.08.028. Andreas Rietbrock helped to prepare the seismic data,
Chapter 1: Introduction refine the comparison between coseismic and afterslip models with seismic velocities. Chao-Shing Lee and Matthew Miller worked on the OBS deployments and assisted with the analysis of this offshore dataset.
In Chapter Seven, we apply a multiple point-source moment tensor inversion scheme to the Mw 7.1 Araucania aftershock in order to explore megathrust rupture
complexity and earthquake triggering mechanisms. By interpreting our results in the context of seismic velocity structure identified in the earlier chapters, we demonstrate how slip on the plate interface instantaneously triggered fault rupture in the overriding plate. This chapter was accepted for publication in Nature Geoscience on 13 October 2015. Citation: Stephen P. Hicks and Andreas Rietbrock: Seismic slip on an upper plate normal fault during a large subduction megathrust rupture, Nature Geoscience, November 2015, doi: 10.1038/ngeo2585. Andreas Rietbrock ran the 3-D full- waveform simulation on the high performance computing cluster and helped to interpret the results.
In Chapter Eight, we synthesise interpretations made in the preceding chapters to discuss the physical factors that influenced the Maule rupture and their relationship with tectonic structure along the South American subduction margin. We speculate what our findings may mean for future earthquake hazard in central Chile. We also compare our results from the Maule earthquake with similar studies from the 2011 Tohoku earthquake rupture zone. This comparison sheds light on some of the key controls on rupture size and slip localisation of large subduction earthquakes. We also demonstrate the effect of 3-D structural heterogeneity on seismic wave propagation by presenting some preliminary results from spectral element waveform simulations of moderately sized aftershocks of the Maule sequence.
Finally, in Chapter Nine, we present the main conclusions of the project, answering the research questions posed in this introduction and highlighting the significant contributions made toward understanding subduction megathrust earthquakes. We also provide recommendations for future work in this field of research.
Chapter 2
UPPER PLATE VERSUS LOWER PLATE:
P
HYSICAL
C
ONTROLS ON
M
EGATHRUST
E
ARTHQUAKE
R
UPTURE
PROCESSES
Earth’s largest earthquakes occur in subduction zones, where two plates collide and one sinks beneath the other. With 43,000 km of subduction plate boundaries (Wang, 2010), understanding the pattern of ruptures along these faults is vital for seismic and tsunami hazard assessment. However, we lack knowledge on the full history of fault locking and past slip in many subduction zones. In order to constrain future earthquake characteristics, a long-term aim in subduction zone research is to comprehend the physical controls of large earthquakes. There are critical questions that need to be answered on stress distribution and resulting strain accumulation in subduction zones. What influences the spatial pattern of coupling and eventual slip along the plate interface? Does the upper plate or subducting plate play a greater role in controlling earthquake processes? What controls the potential for shallow, tsunamigenic earthquakes?
The surge of large subduction earthquakes in the past decade, and the breadth of seismic and geodetic data have that detailed these ruptures (e.g. Lay, 2015) allow us to begin to answer these crucial questions. Therefore, a review of this subject is timely. Earlier attempts have been made to unravel the properties of the subducting plate interface and to assess the influence on large earthquake ruptures (Hyndman et
2.1 Structural domains above and below the megathrust
al., 1997; Ruff and Tichelaar, 1996). Kopp (2013) provides a global summary of subduction margin structural domains and their control of subduction zone earthquakes, with particular focus on the western Pacific. However, these papers do not include numerous findings from the 2010 Mw 8.8 Maule, Chile and 2011 Mw 9.0
Tohoku, Japan earthquakes. Furthermore, Kopp (2013) gives little attention to rheology and density variations in the upper plate. Wang and Bilek (2014) focussed on the role of subducted topographic relief in seismogenic processes. As described in the previous chapter, Lay et al. (2012) describe the depth distribution of rupture properties, but inferring connections with physical properties was not within the scope of this paper.
A gap still remains in integrating earthquake observations with images of physical properties to assess governing factors of the seismic cycle. Moreover, laboratory and numerical experiments are needed to directly link physical properties with fault behaviour. Here, we strictly focus on the shallowest and most strongly coupled part of the subduction zone that extends from the trench to around 50 km depth: the subduction megathrust (e.g. Lay et al., 2012). We begin by introducing the typical structural features along and around the plate interface, and how the characteristics of these vary between subduction zones. We then assess the role played by each of these domains at the different stages of the seismic cycle (which was outlined in Section 1.1).
2.1
Structural domains above and below the megathrust
We describe the main structural features that directly interact with the subducting plate interface (Figure 2.1). We start with domains of the upper plate, and gradually move from the trench through to the central forearc. The nature of the megathrust contact itself, as well as the composition and structural styles of the subducting plate, are considered. Although every subduction zone has unique physical characteristics, one unified classification divides regions of subduction into accretionary and erosive sedimentation styles (Cloos and Shreve, 1988; von Huene and Scholl, 1991; Scholl et al., 1980); (Figure 2.2).
Chapter 2: Upper Plate Versus Lower Plate: Physical Controls on Megathrust