Research

My research develops probabilistic methods for imaging the Earth’s subsurface from incomplete and noisy observations. I am particularly interested in methods that make uncertainty explicit, computationally practical and scientifically useful.

Bayesian geophysical imaging

Geophysical inverse problems are often non-unique: many models can explain the same observations. I develop Bayesian inversion methods that represent this ambiguity rather than reducing it to a single best-fitting model. The resulting posterior distributions support more transparent interpretation and better-informed decisions.

Efficient uncertainty quantification

Exact Bayesian inference is usually too expensive for realistic seismic and geophysical problems. My work explores variational inference, normalising flows and other scalable approximations that preserve useful uncertainty information while making high-dimensional inference feasible.

Seismic imaging and wavefield methods

I work on seismic tomography, reverse-time migration and full waveform inversion. A central theme is combining physical wave-propagation models with modern statistical inference to improve imaging quality and quantify confidence in recovered structures.

Monitoring and environmental applications

These methods have applications in time-lapse monitoring of carbon storage, groundwater characterisation and near-surface imaging. In these settings, uncertainty estimates are not an optional extra: they are essential for interpreting change and assessing risk.

Selected themes

• Bayesian and variational inference for geophysical inverse problems
• Uncertainty quantification and interrogation in seismic imaging and tomography
• Full waveform inversion and seismic wavefield modelling

For related research outputs, visit Publications.