Research

PhD research

My research focused on developing statistical models to describe and understand earthquakes that are caused by gas extraction in the Netherlands.

Human-induced earthquakes are usually smaller in magnitude and fewer in number than their tectonic counterparts. This low-data setting is a challenge to statistical modelling and necessitates the inclusion of domain-expert knowledge. If this low-data setting was not challenging enough, changes in gas extraction and earthquake detection lead to further complications and inefficiencies if standard modelling approaches are used.

My PhD research developed statistical methodology to make most efficient use of the limited available data and extended existing earthquake models to improve understanding of these induced seismic events.

You can find a copy of my thesis on github.

PhD outputs

Publications

Varty, Z., Tawn J.A., Atkinson P.M. and Bierman S. (2021). Inference for extreme earthquake mangitudes accounting for a time-varying measurement process. (Submitted, preprint on arXiv)

Conferences and workshop contributions

Date Event Location
Jun 2022 M_max workshop Amsterdam, NL.
Jan 2021 CRG Extremes workshop Remote.
May 2020 STOR-i time-series and spatial statistics workshop Remote.
Sept 2019 Interfaces in extreme value theory workshop Lancaster, UK.
Sept 2019 Royal Statistical Society conference Belfast, UK.
Aug 2019 International statistical seismology workshop (StatSei11) Hakone, JPN.
Jul 2019 GRASPA (Italian Environmetics Society) Pescara, IT.
Jan 2019 STOR-i annual conference Lancaster, UK.
Jan 2018 STOR-i annual conference Lancaster, UK.

Past projects

A review of simulated annealing techniques: Simulated annealing is a metahuristic technique mainly used for combinatorial optimisation. Applications, parallelisation and extensions of the technique were reviewed.

Inference on censored networks: Networks are censored when existing nodes or edges are not observed. Methods for inference under different types of missingness were explored. Master’s project supervised by Dr. Christopher Nemeth.

Computionally intensive methods for modelling houshold epidemics: Approximate Bayesian Computation was utilised to allow inference on disease models with intractable likelihoods. Master’s dissertation supervised by Prof. Peter Neal.