Overview of Research Interests

Summary

We live on a dynamic planet, whose upper layers respond in a brittle fashion to slow forcing from the underlying mantle. This takes the form of localised earthquake faulting on a large scale, and fracturing on a much smaller scale. Understanding the processes that lead to the mechanical, structural, hydraulic and chemical properties of rocks undergoing low-temperature deformation is a complex and fascinating challenge. The Earth is a complex material to start with, as a consequence of previous geological events. It is also opaque – we cannot see into its sub-surface.  The acceleration from stable crack growth to catastrophic dynamic failure is extremely rapid, non-linear, and difficult to predict for individual events. There are strong feedbacks between the elements involved. Nevertheless, complex non-linear systems often ‘self-organise’ spontaneously to produce order and pattern in the population dynamics, for example the scale-invariant or fractal geometry of fault and earthquake populations.

A quantitative understanding of such processes helps: to build our cities in a way that mitigates earthquake risk; to assess the predictability of volcanoes and earthquakes; to evaluate risk of induced seismicity associated with subsurface engineering projects; and to predict fluid flow and transport in subsurface reservoirs and aquifers, including applications to geothermal and carbon storage reservoirs to mitigate climate change and support the transition to a net zero economy.

I am interested in attacking these problems with a variety of analytical, numerical, laboratory, and field techniques, usually in collaboration with colleagues from a range of disciplines. The aim is to describe patterns of earthquakes and fractures in space and time, to work out their underlying mechanisms, and where possible to provide results in a form that can be used in a practical way.

1. Earthquake hazard.
Earthquake hazard depends on the probability of occurrence of events, the nature and strength of the seismic source, and the propagation of the disturbance to the site of a particular building or facility. During my PhD I developed a method, based on statistical mechanics and information theory, to quantify the likelihood of large damaging events that may not yet have occurred by synthesising short-term earthquake recurrence statistics with longer-term geological and tectonic information on deformation rates. More recent work has concentrated on improving earthquake hazard estimates by quantifying the effect of systematic and random uncertainties on estimates of earthquake occurrence rates, including issues of model selection and convergence of model parameters as more data become available.  This includes applying the integrated nested Laplace approximation to develop data-driven models for earthquake recurrence models in space and time, using a variety of input data including deformation rates, fault maps, slip rates, and topography, as well as earthquake catalogues, and to quantify the resulting uncertainties, with Mark Naylor and Finn Lindgren.

2. Earthquake precursors, predictability and forecasting.
The notion of of deterministic earthquake “prediction” – i.e. “specifying in advance the precise location, size and time of occurrence of an individual event, above chance” (where every word counts!) – has long been the ‘holy grail’ of Earth science. However, the extreme non-linearity of the underlying avalanche-type dynamics, the complexity of the Earth, and the lack of good data at large space and time scales make this problem extremely (perhaps inherently) intractable.  I have contributed to efforts to develop models to models for this variability in predictability in catastrophic failure rooted in fracture mechanics and the population dynamics of heterogeneous Earth materials under stress as a complex, non-linear, system, and contributed to the debate on the statistical significance of earthquake precursors, the associated earthquake predictability and the development of protocols for operational earthquake forecasting.  This includes a formal statistical evaluation of the population dynamics for natural events, mainly associated with the tendency of earthquakes to cluster in space and time.  The main applications are in assessing conditional probabilities for operational forecasting of time-dependent seismic and volcanic hazard, including testing forecasting hypotheses in real time, and on assessing the probability of inducing earthquakes by injecting or extraction fluid into/out of the Earth or mining for critical minerals.

3. Rock Physics. Along with Philip Meredith and colleagues in UCL, I developed and tested analytical and numerical models, based on time-dependent fracture and damage mechanics, to predict and test the precise nature and form of precursors to catastrophic failure observed under controlled conditions in laboratory rock samples, and to examining the extent to which such behaviour scales in space and time. The rock physics group at Edinburgh has developed a significant experimental capability to investigate the fundamental processes of catastrophic rock failure, including the effect of heterogeneity,  fluid rock interactions (including geochemical controls and signatures), and the resulting geophysical, mechanical and hydraulic signals, including acoustic emissions. We have demonstrated systematic changes in fluid permeability as a function of stress state and pore-fluid chemistry, even on relatively short time scales (hours to weeks), and developed models for environmentally-assisted crack growth and rock creep behaviour. We have also contributed to several European-funded projects to quantify fluid-rock interactions in boreholes drilled into the active Aegion fault in the Gulf of Corinth, Greece and the experimental evaluation of contaminant transport in porous fractured media. Currently the main focus is on directly imaging the deformation process in space and time in live experiments carried out in a synchrotron (x-ray ‘vision’) with contemporary acoustic monitoring (‘sound’) in a new cell designed for this purpose, leading to new insight into the relationship between the underlying processes and those we can infer from remote geophysical measurement.

4. Scaling of faults and fractures. In order to underpin the models described above a direct investigation of the structure and scaling properties of fracture systems, and how they might come about, is necessary. This includes the commonly-observed scale-free or ‘fractal’ scaling of such patterns, and how they can be correlated with mechanical (e.g. the fracture toughness), hydraulic (e.g. fluid permeability of fault gouge) and seismic (event statistics, velocity and attenuation) properties of fault and fracture systems. Along with Ferenz Kun, and colleagues in Debrecen University, we have recently provided a physical basis for the observed scaling of rupture parameters in discrete-element models for the catastrophic failure of porous media in discrete element models, involving the competition between structural disorder and the physics of grain-scale processes, producing emergent scaling relations that are remarkably similar to those observed in experiments on porous media.

5. Earth Structure. Earthquakes may be used as natural sources to illuminate the structure of the Earth at different scales, using techniques similar to medical imaging by tomography, the main difference being the lack of control in the source location. We have used this method to investigate earth structure in the form of subtle variations in seismic velocity and anelastic attenuation in the Aegean and the Pacific, in good agreement with tectonic models which have been suggested for these areas, and, in collaboration with the BGS Anisotropy project, using the scattered wave field to track changes in the pore pressure field using time-lapse seismic imaging in subsurface reservoirs.  Our work on high-frequency data recovered from processing of array data broadened the range of applicability of the ‘absorption band’ model for seismic attenuation in the Earth’s mantle.