Skip to main content Link Search Menu Expand Document (external link)

SPIN ESR 2.3: Next-Generation Physics-based earthquake forecasts


Host institution: British Geological Survey (BGS) drawing

Supervisors:

Supervisors: Margarita Segou, Brian Baptie (BGS, UK), Ian Main & Andrew Curtis (University of Edinburgh, UK)

General information

We are seeking highly motivated candidates for a fully funded PhD studentship at the British Geological Survey and the University of Edinburgh. Depending on the interests of the successful candidate, the project will either explore the use of physics-based simulations to develop the next generation of earthquake forecast models or generate new images of the Southern Alpine Fault in New Zealand, and use the results to model future hazard and earthquake triggering.

This PhD position is one of the 15 Early Stage Researcher (ESR) positions within the SPIN project. SPIN is an Innovative Training Network (ITN) funded by the European Commission under the Horizon 2020 Marie Sklodowska-Curie Action (MSCA).

SPIN will focus on training 15 PhD candidates in emerging measurement technologies in seismology. We will research the design of monitoring systems for precursory changes in material properties, all while optimizing observation strategies. The unique interdisciplinary and inter-sectoral network will enable PhDs to gain international expertise at excellent research institutions, with a meaningful exposure of each PhD to other disciplines and sectors, thus going far beyond the education at a single PhD programme.

Project descriptions

Option 1: Next-Generation Physics-based earthquake forecasts_

Earthquakes show clustering in space and time, as illustrated by the aftershocks triggered by large events. Empirical descriptions of clustering explain many features observed in seismicity catalogues, and they can be used to construct forecasts that indicate how earthquake probabilities change over the short term. However, these statistical approaches do not offer significant improvement for the physical triggering mechanisms that govern earthquake occurrence. The complexity and heterogeneity imaged in the structure and stress field of the Earth makes any direct interpretation of laboratory experimental results challenging. Now, the theory of static stress transfer combined with the laboratory derived rate-and-state law that describes the seismicity response to a stress perturbation is able to describe the stress-mediated fault interactions within a testable framework.

Recent work shows that physics-based models match or even exceed the performance of empirical approaches when applied to aftershock sequences. The most important elements of improved performance in these approaches come from the consideration of heterogenous faulting networks and stress states. The challenge behind the development of empirical and physics-based forecasts lies largely in their interpretation since short-term earthquake probabilities for future large magnitude events remain low in an absolute sense (< 1% per day).

Here, we seek to push the limits of physics-based approaches in earthquake forecasting by including improved time-dependent representation of stress (transient deformation, pore pressure effects etc) to achieve an evolving physics-based model that will inform us about large-scale processes that occur in real Earth. In this project you will develop earthquake forecast models in large scale based on physics-based simulations aiming to improve our process-based understanding of earthquake triggering. Those large-scale processes may inform us about future experiments in the lab motivating further rock physics research. The framework will generate, evaluate, optimize and discriminate earthquake forecasts based on robust statistical modelling and validation.

Option 2: Combining tomographic imaging, earthquake triggering and seismic hazard: application to the Southern Alpine Fault, New Zealand.

Understanding the potential for inter-earthquake triggering is an important component of improving earthquake forecasts. You will use a new seismometer array to image the subsurface beneath the Southern Alpine Fault, and use the results to model future hazard and triggering.

Damaging earthquakes occur when the tectonic stress that builds up across locked zones of faults is suddenly released by rapid rupture and slip along one or more faults. The energy released travels outward from the fault in the form of seismic waves. That energy eventually arrives at locations of potential hazard such as cities and infrastructure, and at locations of other faults which may also be critically stressed and are therefore primed for rupture. These waves propagate through the Earth’s subsurface, and how they focus or defocus their energy depends on the subsurface seismic velocity structure. If large-amplitude patterns of energy arrive in cities or at critical infrastructure then severe damage may occur; if the energy arrives at other critically-stressed faults, those faults may be triggered into rupture causing additional earthquakes. Either scenario may cause a threat to human life. It is therefore necessary both to understand the potential slip patterns and magnitudes that may occur along a fault, and how energy will propagate from that fault in both the short and long term to vulnerable locations of potential hazard or triggering (add general reference).

The Southern Alpine Fault of New Zealand provides an important example of potential hazard to cities and infrastructure. It has a historical record of rupturing in very large earthquakes, and according to the average time between past events, a future earthquake is relatively likely to occur within the next few decades. Seismic data is being recorded on a revolutionary new array of seismometers that, for the first time, spans the majority of the on-shore length of the fault (left panel in the Figure above). Your research will focus on this opportunity: you will help to collect data from the seismometers, you will analyse them in your Ph. D., and you will work with seismologists in Edinburgh and Wellington to contribute to the physics-based assessment of seismic hazard and potential earthquake triggering mechanisms in New Zealand.

ESR 2.3 illustration

_Left: New array of seismometers along Southern Alpine Fault, New Zealand (white circles). Coloured rays show how fault slip will be projected to predict shaking of the major cities of Wellington, Christchurch and Queenstown. Other arrays of seismometer shown by spots and crosses. Top-left inset shows a cross-section of seismicity along the fault, and the inferred depth at which the locked fault begins to creep. Right: inter-seismometer paths available for seismic tomography._

You will initially use novel probabilistic tomographic methods (Zhang et al., 2020) to create images of the seismic velocity structure along and around the fault. Interpreting this structure in conjunction with other geological and physical measurements and processes will allow models of fault mechanics, and potential slip scenarios to be refined. You will then use the data set represented in the right panel of the Figure above to construct tomographic images of the velocity structure beneath the South Island of New Zealand. These models will allow the patterns of focussing of seismic waves to be assessed across the island for any slip scenario. Finally you will assess the stress perturbations that may be created both by the fault slip itself, and by those propagating waves, to evaluate the potential for damage to critical infrastructure or cities, and for triggering of earthquakes on other known or unknown faults (Segou 2020).

Key Research Questions

1. What is the seismic structure of the Southern Alpine Fault? 
2. Will seismic waves from earthquakes on the fault focus energy on vulnerable locations?
3. Is it likely that the seismic waves will trigger earthquakes on other faults in New Zealand?

Methods and Plan of Activities.

  • Year 1: Learn about the structure and tectonics of New Zealand, and help to service the array of seismometers as necessary. Learn about short and long term earthquake triggering mechanisms using retrospective analysis of well documented events such as the Christchurch earthquake sequence of 2011, as case studies. Learn about seismic interferometry and tomographic methods, and identify weaknesses and advantages of current approaches for the particular survey geometry in the Figure. During a short industrial placement to RealTimeSeismic, learn about earthquake detection algorithms and apply them to the recorded data streams. Develop novel imaging methods as necessary, conduct computational tests to evaluate their success, and apply to the first batch of data from the new array to image the fault zone. This work will form the basis of your first scientific paper from this project, to be written in Year 2.
  • Year 2: Perform tomography across the South Island of New Zealand using the data set depicted on the right of the above Figure. Collaborating with other SPIN PhD candidates during a short placement at the Dublin Institute for Advanced Studies in Ireland, assess the potential to use similar methods on a smaller scale problems, around volcanoes or other regions of evolving stress.
  • Year 3: Use the tomographic models to simulate wave propagation through models developed in other SPIN projects for linear and nonlinear materials, to assess the potential focussing and energy redistribution due to waves from future slip scenarios in particular on vulnerable locations, and assess the likelihood of triggering other earthquakes. Compare the pattern of results to those from existing statistical forecasting models, and contribute your new knowledge to the on-going effort to update hazard maps for New Zealand.

Further Reading: For general information about earthquake forecasting see:

  • X. Zhang, C. Roy, A. Curtis, A. Nowacki, B. Baptie, 2020. Imaging the subsurface using induced seismicity and ambient noise: 3D Tomographic Monte Carlo joint inversion of earthquake body wave travel times and surface wave dispersion. Geophys. J. Int., 222, pp.1639–1655, doi: 10.1093/gji/ggaa230 – pdf at https://blogs.ed.ac.uk/curtis/publications/