• Login
• Admin

Hawley-Sibbett, Luke Richard (2023) Exploring novel algorithms for an improved ERT monitoring workflow. PhD thesis, University of Nottingham.

Preview

PDF (Final thesis, including typo corrections, as approved by examiners) (Thesis - as examined) - Requires a PDF viewer such as GSview, Xpdf or Adobe Acrobat Reader Available under Licence Creative Commons Attribution. Download (17MB) | Preview

Abstract

Electrical Resistivity Tomography (ERT) is a technique for estimating the resistivity of the subsurface. Long term ERT monitoring is increasingly used to estimate changes in resistivity over time, but data processing workflows are not fully developed. In this thesis, new methods are explored to improve important elements of the workflow: the regularisation of time-lapse inversions, and the detection of data quality anomalies which can be missed by conventional methods.

One of the limitations of ERT inversion is that regularisation methods are often biased towards either sharp or smooth changes in resistivity in both the temporal or spatial domains. A novel alternating-minimisation inversion algorithm is developed, using Total Generalised Variation regularisation, a functional which is able to represent both smooth and sharp changes better than existing methods. Unlike minimum gradient support type approaches, it does not depend on a pre-defined threshold in the smoothing function which limits smooth behaviours to a specific resistivity or gradient range. Comprehensive parameter testing of the algorithm demonstrated that it converges reliably, and outperforms conventional TV and l2 regularisation functions for a test model when properly configured. However, the solutions are significantly influenced by the initial model. The smooth behaviour is also limited near the model boundaries. Potential limitations of the finite difference approach used in this implementation are discussed, and alternatives are proposed for future improvements.

Existing data quality measures such as reciprocal errors and contact resistance measurements are insensitive to certain sources of noise. This work is motivated by a case study where a damaged cable connector led to short circuiting errors. Short-circuiting errors have the potential to occur at any monitoring installation, as they are a consequence of any physical damage to the cables. A Principal Component Analysis (PCA) control chart method is developed in order to detect the onset of these errors, which went unnoticed until months later. Previous data from the site was used to establish a PCA model. Deviation from the model is measured using a Q2 statistic. The new method is able to successfully detect the onset of these shorting errors, using both resistivity and reciprocal error data. Therefore, this approach may be used where reciprocal error data is unavailable. Suggestions are made for the further development of this method though testing with a wider array of data.

Item Type:	Thesis (University of Nottingham only) (PhD)
Supervisors:	Garibaldi, Jonathan M. Pridmore, Tony Chambers, Jonathan E. C. Wilkinson, Paul B.
Keywords:	Electrical Resistivity Tomography, Inverse Problems, Geoelectrical Imaging, Regularisation, Electrical Resistivity Monitoring
Subjects:	Q Science > QA Mathematics > QA 75 Electronic computers. Computer science
Faculties/Schools:	UK Campuses > Faculty of Science > School of Computer Science
Item ID:	74048
Depositing User:	Hawley-Sibbett, Luke
Date Deposited:	17 Apr 2024 09:38
Last Modified:	17 Apr 2024 09:38
URI:	https://eprints.nottingham.ac.uk/id/eprint/74048

Actions (Archive Staff Only)

Edit View