IAP-25-118

How Earthquakes Become Tsunamis: Investigating Rupture Propagation into Subduction Zone Accretionary Prisms

Project Aim
Subduction zones produce Earth’s largest earthquakes with localisation of strain and rupture dynamics controlled by stress accumulation and shear strength. The largest earthquakes typically occur on the main megathrust faults. However, tsunamigenic shallow slip can also occur on shallower splay faults within the accretionary prism. Provisional data from recent International Ocean Drilling Programme (IODP) Expedition 405 (JTRACK) suggest that splay faults are active and represent a previously overlooked tsunami hazard. New core samples recovered from a ~900 m deep borehole during this expedition provide a unique opportunity to explore lithological controls on prism rupture dynamics in the Tohoku plate boundary fault zone and accretionary prism.
This project aims to deform these recovered mudstone samples during laboratory experiments that capture damage evolution and failure with simultaneous acoustic monitoring (Sound) and time-resolved X-ray imaging (Vision). Samples will be deformed under in-situ stress conditions and will characterise, for the first time, micro-scale fracture network evolution, strain localisation and rupture propagation within the different lithologies present in the accretionary prism. Complementary core-scale experiments will be conducted to characterise frictional response and bulk mechanical and micro-seismic behaviour of the samples. Results will be analysed in the context of large-scale ocean drilling and seismic information from other ongoing work, with geomechanical modelling to inform tsunami risk assessment from rupture in the accretionary prism.

Significance and Impact
The JTRACK expedition has recovered the most complete core record of the accretionary prism of an active subduction zone. Quantifying the mechanical properties and deformation characteristics of these samples using state-of-the art methods including time-resolved X-ray microtomography will provide a novel view of how the composition and state of consolidation of these samples interact with the in-situ stress conditions and deformation rates to allow strain to localise within an accretionary prism. Constraining these relationships will drive forward our understanding of subduction zone dynamics and reduce uncertainties in current earthquake and tsunami risk assessments, allowing us to establish how and when tsunamis might occur after shallow slip within the accretionary prism.

Background
The Earth’s largest and most destructive earthquakes occur in subduction zones, such as the Tohoku plate boundary. These earthquakes typically nucleate at depths of 10-50km [Bilek and Lay, 2018] on megathrust faults, and can exceed Mw 9. They present significant tsunami risk due to significant shallow slip as the fault rupture propagates from the deep seismogenic zone to the seafloor near the Japan Trench. Large crustal displacement at the seafloor displaces the ocean water above it, resulting in large tsunamis. Strain localisation and rupture dynamics are controlled by stress accumulation and the shear strength of crustal rocks. Risk of slip propagating all the way to the trench, and therefore large tsunamis, is highest when the shear strength of the subducting rocks and sediments is low. Although the largest slip occurs on the main megathrust fault, or décollement, that represents the active plate boundary, tsunamigenic shallow slip can also occur on shallower splay faults within the accretionary prism, representing an additional tsunami hazard at shallower water depths [e.g., Moore et al., 2007].
The Mw >9 Tohoku-Oki earthquake in 2011 was the 4th largest earthquake in historic timescales. Rupture initiated at a depth of ~17 km and propagated up-dip along the megathrust all the way to the trench [Kodaira et al., 2021]. It generated more than 50m of extreme shallow slip and surface displacement [Fujiwara et al., 2011] and resulted in a devastating tsunami over 40 high that inundated coastlines regionally, killing over 20,000 people and destroying $235 billion worth of infrastructure. To explore the reasons for such extreme shallow slip, two International Ocean Drilling Programme (IODP) expeditions were undertaken: JFAST in 2012 [Chester et al., 2012] and more recently JTRACK in 2024 [Kodaira et al., 2023], on which supervisors UN and AG participated.
Although the reasons for shallow slip are poorly understood, these expeditions collected core samples and other data from the Japanese subduction zone that identified a weak clay layer that exhibit velocity weakening behaviour [Chester et al., 2013; Fulton et al., 2013; Nakamura et al., 2013]. Such behaviour results in rapid fault propagation during the earthquake cycle and likely permitted shallow slip along the megathrust [Ikari et al., 2015; 2015b]. However, provisional JTRACK results (unpublished) also highlight at least two large offset (>500 m) thrust faults in the overlying accretionary prism, at least one of which shows evidence of recent activity. This suggests that thrust faults not only form at the frontal thrust, in sequence, but continue to undergo strain during accretion. This means that fault propagation does not just occur in the weak clay layer at the décollement, but also in the overlying vitric mudstones that make up most of the accretionary prism, representing a previously unidentified tsunami hazard.

Research Problem
JTRACK recovered a near complete ~900 m long sequence of cores from the accretionary prism, plate boundary and underlying oceanic crust, providing a unique and exceptional opportunity to understand the physical and lithological processes that control ongoing deformation in the prism. Moreover, new seismic data and borehole logs from JTRACK will enable the spatial distribution and thickness variability of the prism faults and facies to be mapped, highlighting the likely consequences for prism fault propagation.
The borehole cores show that the mudstones in the accretionary prism consist of varying proportions of volcanic glass/ash, siliceous microfossils and siliciclastic material, increasing in consolidation with depth (Fig. 1, right). These mudstones likely have higher shear strength than the weak clay layer at the plate boundary but thrust faults have localised and propagated through this stronger material. State-of-the-art mechanical testing of these cores while simultaneously imaging the microstructure with X-rays can shed light on the factors controlling fault nucleation and rupture dynamics in these materials.

Key Research Questions
1. How does strain localise within the accretionary prism lithologies?
2. How does the fracture network evolve during fault nucleation and reactivation?
3. How are fault nucleation and rupture dynamics affected by specific lithologies, stress conditions and deformation rates?
4. How can we scale up these findings to provide meaningful information for tsunami risk assessments from faults in the accretionary prism?

Hypothesis
Active fault propagation in the accretionary prism is dependent on the composition and state of consolidation of the mudstones.

Objectives
• Characterise the composition and microstructure of JTRACK core samples
• Conduct micro-scale deformation experiments at Heriot-Watt / University of Edinburgh and targeted in-situ x-ray imaging experiments at a synchrotron
• Conduct core-scale deformation and rotary shear experiments at University of Durham to characterise frictional, mechanical and micro-seismic behaviour
• Analyse the integrated mechanical, micro-seismic and x-ray datasets to visualise microstructural changes, test the hypothesis and address the research questions
• Integrate the experimental results with large-scale JTRACK information to develop geomechanical models of rupture propagation in the accretionary prism

This project will conduct rock deformation experiments to constrain lithological controls on rupture propagation in the accretionary prism of megathrust fault zones. Results will be integrated with other ongoing work on seismic reflection and ocean drilling borehole data to upscale the laboratory findings.
The first stage of the project will be a thorough literature review and characterisation of the JTRACK core material for experimentation. Microscopy techniques, including SEM, x-ray tomography, ultrasonic velocities and XRF/XRD, will be used to characterise the composition and microstructural textures in samples prepared for experimentation.
Experimental campaigns will involve triaxial rock deformation and fault reactivation approaches with integrated mechanical and micro-seismic monitoring. Targeted micro-scale experiments will be conducted to observe directly changes in the microstructure with time-resolved in-situ x-ray imaging at a synchrotron. A novel x-ray transparent triaxial deformation apparatus (Fig. 1, left; Cartwright-Taylor et al., 2022) will be used that allows for deformation at elevated pressure with micro-seismic and ultrasonic monitoring, with protocols optimised in-house prior to beamtime. Lab development to achieve realistic in-situ pressure and temperature conditions representative of relevant depths will be undertaken. Core-scale triaxial deformation experiments at carefully selected conditions will provide a larger scale view of rupture propagation by mapping micro-seismic event locations, while core-scale rotary shear experiments will enable observation and characterisation of the frictional response and velocity strengthening vs weakening behaviour.
Analysis of these integrated experimental datasets will include advanced image processing methods for microstructural analysis, comprising image segmentation to separate fracture networks from the rest of the microstructure and quantify changes in fracture geometry and spatial distribution, and digital volume correlation to obtain local strain field evolution. In addition, acoustic signal processing will be performed to obtain micro-seismic event history, locations, and ultrasonic velocity evolution. These properties will be integrated with each other and with the stress history to test the hypothesis and address the research questions.
To upscale these results and inform tsunami hazard assessments, there is an opportunity to relate the laboratory findings with field observations on an exposed megathrust fault zone. Fieldwork to Inuyama in Japan, a good analogue for the Japan Trench, will be undertaken to observe and characterise mega-splay faults in a Jurassic accretionary complex.
Furthermore, the laboratory findings will be integrated with large-scale findings from other ongoing projects that are investigating the structural architecture and sedimentary properties of the accretionary prism using seismic reflection data collected by the Japanese Agency for Marine Science and Technology over recent years as well as a rich database from ocean drilling boreholes recovered over recent decades, including JTRACK.
Integrated findings from the laboratory experiments, field observations and large-scale geophysical data will be used to develop and constrain geomechanical models using conventional statistical methods and artificial intelligence / machine learning methods.
In addition to peer-reviewed publications, the project will leave a legacy of unique, integrated datasets that will be published open-access.
External collaborators include Dr Ian Butler, and Dr Mike Chandler at University of Edinburgh, Dr. Butler will contribute expertise, equipment and facilities to support the targeted synchrotron experimental campaign. Dr Chandler will contribute expertise in mudstone deformation and geomechanical modelling.

Year 1

Training
· Systematic literature review methods and time management
· Sample preparation methods, and sample characterisation techniques involving acquisition, processing and analysis for x-ray tomography, ultrasonic velocity, SEM and XRF/XRD data
· Other Heriot-Watt training sessions for PhD students, e.g., proposal writing
· International Joint Workshop on Slow to Fast Earthquakes, Europe

Literature review
· Literature review on the mechanics of megathrust earthquakes in the Japan Trench, focussing on deformation in the accretionary prism.

Sample preparation, characterisation
· Prepare samples for testing
· Perform x-ray tomography, SEM, ultrasonic velocity, SEM and XRF/XRD and run analysis to characterise samples

Communication
· Synchrotron proposal – beamtime typically granted 6-12 months after application
· First year report and presentation (major review)
· Draft review paper/chapter

Year 2

Training
· High pressure experimental testing (Heriot-Watt / University of Edinburgh)
· Synchrotron beamline safety and processes (Heriot-Watt / Diamond Light Source)
· Image processing, segmentation and object geometry analysis methods
· Heriot-Watt training sessions for PhD students, e.g. project management
· ALERT Geomaterials Doctoral School and Software for Practical Analysis of Materials (SPAM) Workshop, Europe

Experimental campaign
· In-house testing to optimise experimental protocols for synchrotron campaign (Heriot-Watt / University of Edinburgh)
· Synchrotron campaign for targeted experiments with time-resolved in-situ x-ray imaging to visualise changes in the microstructure and benchmark previously inferred geomechanical behaviour (Diamond Light Source).
· Core-scale testing to characterise frictional behaviour, micro-seismic response and bulk mechanics (University of Durham)

Data integration and evaluation
· Experimental data management, processing and analysis to characterise mudstone mechanical, micro-seismic and microstructural behaviour under in-situ conditions.

Communication
· One international conference, e.g., EGU
· Second year report (minor review)
· Draft experimental data paper/chapter

Year 3

Training
· Geomechanical modelling methods
· Heriot-Watt training sessions for PhD students, e.g., time management

Integration of experimental results with large-scale geomechanical context
· Fieldwork to Inuyama, Japan
· Integration with large-scale context from seismic and core-log data
· Development and validation of geomechanical model
· Hypothesis testing
· Reproducibility experiments

Communication
· Draft hypothesis-testing, integration and geomechanical model paper/chapter

Year 3.5

Communication
· One international conference, e.g., AGU
· Thesis completion and submission

In addition to the IAPETUS training, a comprehensive training programme will be provided comprising both specialist scientific training and generic transferable and professional skills.
Key project-specific training will be given in experimental techniques, managing and analysing mechanical, micro-seismic, ultrasonic and large x-ray datasets and methods for geomechanical modelling. This will be provided by the supervisory team and technical staff at Heriot-Watt and University of Durham and external collaborators where appropriate. The student will also benefit from in-house training in writing, coding and machine learning. This specialist training will be complemented by attendance at external workshops and conferences. Attendance at two to three doctoral training schools and two to three international conferences over the course of the project will be encouraged. This attendance will supplement project-specific training, provide exposure to a broad scientific network and audience, and enabling the application of transferable and professional skills developed through relevant University-level and School courses.
The student will be part of the dynamic Marine Geohazards research group, working alongside other world-leading researchers in the Lyell Centre for Earth and Marine Science and the Institute for Geoenergy Engineering. They will be based in the Lyell Centre and the School for Energy, Geoscience, Infrastructure and Society (EGIS), both of which host annual PhD symposia for students to showcase their work, practice their communication skills and discuss their research topic with a broad audience. The student will have frequent opportunities to present their research to peers and supervisors and receive constructive feedback to improve public speaking and writing. They will also have access to the substantial computing resources of the 4D Imaging Group. Through the external collaborators, the student will also have access to research networks within University of Edinburgh and the Diamond Light Source.
The Research Futures Academy (RFA) at Heriot-Watt provides transferable skills / career development workshops to facilitate the doctorate and future research career of PhD students. The student will be able to attend workshops focusing on:
• basic skills for successful research
• research data management
• communication and dissemination skills
• strategy for publishing
• citation and impact
• equality, diversity and inclusion
In addition to supporting doctoral students to develop such competencies, the RFA provides valuable networking and collaboration opportunities between researchers across diverse disciplines. Heriot-Watt subscribes to LinkedIn Learning which provides courses in a variety of common software packages, coding languages and other transferable skills that doctoral students have access to for online learning at their own pace.
At Heriot-Watt, the student will belong a diverse, inclusive and international community with a strong research culture that celebrates effort and achievement. As part of the IAPETUS DTP cohort, the student will receive a multidisciplinary package of training focused around meeting the specific needs and requirements of each student, benefitting from the combined strength and expertise that is available across the partner organisations.