Don't miss our weekly PhD newsletter | Sign up now Don't miss our weekly PhD newsletter | Sign up now

  Imaging the Earth’s Interior with advanced machine learning methods using Geological, Geophysical, Geochemical or Geobiological information

   School of Geosciences

This project is no longer listed on and may not be available.

Click here to search for PhD studentship opportunities
  Prof Andrew Curtis  No more applications being accepted  Funded PhD Project (Students Worldwide)

About the Project

Please read the information provided at: this will tell you everything you need on how to apply.

Project background

We often wish to image or monitor changes in the Earth’s subsurface using measurements made on the ground surface. For example, the subsurface distribution of fluids, stress and heat are all affected by the injection and extraction of fluids in the Earth’s subsurface, which might be done to harvest geothermal energy, or to store gasses such as CO2, Hydrogen or methane in the subsurface rock pore space. They are also affected by natural phenomena such as magma movement & earthquakes. Being able to image and monitor changes in the Earth’s subsurface is of global interest to academia, industry, governments.

Subsurface imaging typically uses Geophysical measurements. These often involve injecting seismic or electromagnetic energy into the subsurface and recording waves that emerge at the surface after being scattered by variations in fluid or rock properties in the subsurface. 

Imaging large and complex subsurface volumes is difficult, and highly computational. There is a pressing need to develop novel algorithms that introduce geological and other information (in addition to standard geophysical data) to improve images, and to evaluate our uncertainty in the answers to key scientific and applied questions that affect subsurface decision-making.

This PhD project will develop novel methods to image and monitor the Earth’s subsurface. It will create ways to assess our state of knowledge or uncertainty in the results, and apply the methods to the above applied academic and industrial activities, focussed on energy transition applications. 

The projects will be designed according to each applicant’s interests and capabilities. Each will use wave theory and other appropriate physics, inversion methods, uncertainty analysis, geological, geophysical, geochemical and geobiological modelling and data analysis, machine learning & high performance computation in varying proportions. 

A successful applicant will join the dynamic research team, the “Edinburgh Imaging Project” (, and will have time and guidance to learn all necessary skills.

Research questions

  • How can images of the Earth’s subsurface be improved?
  • To what extent do geochemical and geobiological data improve subsurface information?
  • Can we develop novel methods to estimate the uncertainty in subsurface information, particularly if many parameters define our images (i.e., in high-dimensional problems)?
  • How can we best deploy subsurface imaging to answer practical scientific questions connected to the Earth?
  • How do we create value from the imaging and uncertainty results?


Initially you will learn about state-of-the-art imaging and inversion methods that have been developed in various fields of research including geophysics, mathematics and machine learning. You will learn about geological process models, geochemical and geobiological data, and how they relate to a shared model of the Earth’s subsurface. Then you will develop novel methods of your own, which you will test on both synthetic (artificial, and therefore controlled) and real data sets. This will be carried out together with your primary supervisor, and as part of a research team, and you may draw on collaborations with industrial partners.

Year 1 – Read background literature and test existing imaging codes. We will focus the project on a practical problem designed to be of interest to you, around which you will develop novel algorithms for imaging and uncertainty assessment. Computational tests of these algorithms will form the basis of your first published paper.

Year 2 – Introduce real data or more challenging artificial test data to assess the algorithms under realistic conditions. Optimise the algorithm design to focus on specific questions of interest to Earth scientists and industry. This will form the basis of your second paper.

Year 3 – Apply the methods to a range of real problems, to improve decision-making around energy transitional applications in the academic and industrial areas described above. This will form the basis of your third paper.


A comprehensive on-going training programme will be designed for you personally, comprising both specialist scientific training and generic transferable and professional skills. You will receive training in processing and acquisition of a range of geophysical, geological and other relevant data types, and in the latest imaging methods using active and passive methods. You will develop expertise in high performance computing, and learn about key areas of mathematics, machine learning, and a range of methods from other fields. You will be trained to present your results professionally to industrial and academic partners at least every six months, gaining confidence in the way that you communicate complex scientific information.


Students with a strong background including a B. Sc. Hons. level degree in quantitative Earth science, geophysics, physics, mathematics or related subjects are encouraged to apply. Demonstrated ability to write complex computer code will be an advantage. 

Environmental Sciences (13) Geology (18)

Funding Notes

Fully Funded PhD project for National or International students with Enhanced Stipend.


Please read the information provided at: - this will tell you everything you need on how to apply.

Where will I study?

Search Suggestions
Search suggestions

Based on your current searches we recommend the following search filters.