University of Edinburgh Featured PhD Programmes
University of Edinburgh Featured PhD Programmes
Norwich Research Park Featured PhD Programmes
Imperial College London Featured PhD Programmes
University of Reading Featured PhD Programmes

Bringing machine learning techniques to geophysical data processing

Project Description

Gravity and magnetic surveys offer a cost-effective method for investigating sedimentary basins, mineral deposits, active tectonics in the oceans, and overall crustal structure. Processing these data can be challenging due to high-frequency noise and the often irregular distribution of measurement points. The most promising data processing method is the equivalent source (or equivalent layer) technique (Dampney, 1969), which has been applied with much success to interpolation, upward-continuation, reduction to the pole, denoising, derivative calculation, and more. While much progress has been made in the field, the equivalent layer technique still suffers from two major short-comings: the difficulty in choosing configuration parameters (layer depth, regularization parameter, number of sources, etc) and the high computational load. The requirement for tuning of configuration parameters is shared with many machine learning methods. In fact, the mathematics underlying most machine learning regression problems is the same as that for equivalent layers. Many sophisticated methods exist for the automatic tuning of machine learning problems which have so far gone unused in geophysics. The nascent field to automatic machine learning (AutoML) is progressing rapidly towards fully automated pipelines, from data pre-processing to model fitting (Guyon et al., 2019). It presents a great opportunity for the adaptation of existing methods as well as the development of new ones optimized for geophysics.

Project Summary:
The goal of this project is to investigate the use of existing machine learning techniques to process geophysical data using equivalent layers, mainly: (1) parameter tuning methods for automatic configuration; (2) feature selection to improve stability and predictive power; (3) cross-validation for quantitative evaluation of performance; (4) parallel and GPU (graphics processing unit) computing to tackle the large computational load. The methods and software developed during this project can be applied to process large amounts of gravity and magnetics data, including airborne and satellite surveys, and produce data products that can enable further scientific investigations. Examples of such data products include global gravity gradient grids from GOCE satellite measurements, regional magnetic grids for the UK, gravity grids for the Moon and Mars, etc.

The appointed student will acquire the mathematical and programming skills required to undertake the project. They will be trained to develop software in a collaborative environment using GitHub and use the current best practices in software engineering. The project will be conducted following the current established norms of reproducible research, with all outputs published on the group’s GitHub page ( The project will also involve code contributions to the different open-source Python software developed by the research group, mainly Fatiando a Terra (, leading to potential impact beyond standard scientific publications. This position would suit someone with mathematical and numerical methods skills (or who is willing to learn). Some experience with computer programming in any language is desirable. The appointed student will participate in online group discussions and peer-to-peer learning.

To apply for this opportunity please visit: and click the ‘Apply online’ button.

Funding Notes

Full funding (fees, stipend, research support budget) is provided by the University of Liverpool for 3.5 years for UK or EU citizens. Formal training is offered through partnership between the Universities of Liverpool and Manchester. Our training programme will provide all PhD students with an opportunity to collaborate with an academic or non-academic partner and participate in placements.


Dampney, C. N. G. (1969). The equivalent source technique. Geophysics, 34(1), 39–53. doi:10.1190/1.1439996
Guyon, I., Sun-Hosoya, L., Boullé, M., Escalante, H. J., Escalera, S., Liu, Z., et al. (2019). Analysis of the AutoML Challenge Series 2015–2018. In Automated Machine Learning: Methods, Systems, Challenges (pp. 177–219). Cham: Springer International Publishing. doi:10.1007/978-3-030-05318-5_10

Related Subjects

Email Now

Insert previous message below for editing? 
You haven’t included a message. Providing a specific message means universities will take your enquiry more seriously and helps them provide the information you need.
Why not add a message here
* required field
Send a copy to me for my own records.

Your enquiry has been emailed successfully

FindAPhD. Copyright 2005-2019
All rights reserved.