Machine learning to detect earthquakes in energy extraction (EPSRC DTP)

Supervisor(s)

Dr Andy Nowacki and Professor David Hogg

Contact Dr Andy Nowacki (A.Nowacki@leeds.ac.uk) to discuss this project further informally.

Project description

Our society depends on the safe extraction of energy from a mixture of natural resources, including heat from the ground in geothermal power production, and the exploitation of hydrocarbons.

These industrial processes inevitably lead to the occurrence of microearthquakes, which are induced by changes in stress, chemistry, fluid movement, and other factors involved with production. Whilst the earthquake may have positive consequences, such as the increased ability of steam to flow through the rocks to drive a turbine, it may also pose hazards, such as damage caused by ground shaking from larger earthquakes which are triggered. Therefore, it is vital and often mandatory to monitor these systems for seismic activity—both to mitigate risk and increase our knowledge of the subsurface.

Traditionally, monitoring for the occurrence of earthquakes has been done by manual inspection of recorded seismograms (as in Wilks et al., 2015), or basic metrics based on these recordings of ground motion (e.g., Drew et al., 2013). However, this approach requires a great deal of human intervention. More automated methods exist (e.g., Shi et al., 2018), but often require setting a number of parameters for each application or are computationally expensive, and so are sometimes not appropriate for real-time use.

Machine learning (ML) is a set of analytical techniques which has the capability to address these challenges, and is now used in a huge range of applications (e.g., Dima & Hogg, 2012). In this project, you will make progress in automating the detection, location and classification of earthquakes by using machine learning techniques to process seismic data.

You will work on a number of exciting settings and datasets, including the Aluto geothermal power station in Ethiopia and the Cotton Valley tight gas field in the southern United States. Although machine learning has been employed in seismology for many years (e.g., Dowla et al., 1990), recent advances in techniques mean rapid progress is now possible.

In this project, you will develop your skills in quantitative analysis of data and machine learning, both skills in high demand across disciplines, with the advice and support of Dr Andy Nowacki (Earth and Environment) and Prof David Hogg (Computing).

iA good candidate will have a background in geophysics, physics, maths, engineering, or a related discipline, and a keen interest in the Earth, how it works, and how we can make energy production safe and sustainable using large datasets. The successful candidate will be required to learn the fundamentals and applications of machine learning, and develop software to apply it.

You will be an integral part of the Institutes of Applied Geoscience and Geophysics and Tectonics within the School of Earth and Environment (SEE), and within the Artificial Intelligence Group in the School of Computing. As part of SEE, you will have the opportunity to travel widely to attend conferences and undertake fieldwork.

References:

Damen, D., Hogg, D.C., 2012. Detecting carried objects from sequences of of walking pedestrians. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45, 1056–1067. doi:10.1109/TPAMI.2011.205 Dowla, F.U., Taylor, S.R, Anderson, R.W., 1990. Seismic discrimination with artificial neural networks: Preliminary results with regional spectral data. Bull. Seis. Soc. Am., 80, 1346–1373. Link Drew, J., White, R.S., Tilmann, F., Tarasewicz, J., 2013. Coalescence microseismic mapping. Geophys. J. Int., 195, 1773–1785. doi:10.1093/gji/ggt331 Shi, P., Angus, D.A., Rost, S., Nowacki, A., Yuan, S., 2018. Automated seismic waveform location using Multichannel Coherency Migration (MCM)—I. Theory. Geophys. J. Int. doi:10.1093/gji/ggy132 Wilks, M., Kendall, J.-M., Nowacki, A., Biggs, J., Wookey, J., Ayele, A., Bedade, T., 2017. Seismicity associated with magmatism, faulting and geothermal circulation at Aluto Volcano, Main Ethiopian Rift. J. Volcan. Geoth. Res., 340, 52–67. doi:10.1016/j.jvolgeores.2017.04.003

Entry requirements

Applications are invited from candidates with or expecting a minimum of a UK upper second class honours degree (2:1) or equivalent, and/or a Master's degree in the relevant subject area.

If English is not your first language, you must provide evidence that you meet the University’s minimum English Language requirements.

How to apply

Formal applications for research degree study should be made online through the university's website.

If you require any further information, please contact the Graduate School Office e: apply-phd@see.leeds.ac.uk, or t: +44 (0)113 343 1634.

We welcome scholarship applications from all suitably-qualified candidates, but UK black and minority ethnic (BME) researchers are currently under-represented in our Postgraduate Research community, and we would therefore particularly encourage applications from UK BME candidates. All scholarships will be awarded on the basis of merit.