PhD studentship: Optimising Groundwater Monitoring

We are looking for a highly motivated student with significant statistical experience to work on further developing the GWSDAT package to include groundwater monitoring optimisation. The selected candidate will take the lead in this project, develop efficient routines to design and optimise groundwater monitoring networks and implement these routines in GWSDAT so that they become available for use in the groundwater industry. This project is the result of a collaboration between the University of Glasgow and Shell Global Solutions and allows the student to work both in a vibrant academic setting as well as working with industry hydrogeologists and data scientists.

Requirements

The candidate should be either:

  • a graduate in hydrology, geosciences or environmental sciences and have a strong interest in data science or
  • a graduate in data science with a strong interest in environmental science and geoscience.

Good communication skills in English are required.

Funding available

The studentship will cover Home/EU fees for 3.5 years as well as a stipend in line with the Research Council doctoral stipend level (£15,285 p.a. for 2020/21, not considered taxable income by HMRC).

Background

Industrial assets such as refineries or factories need close monitoring of groundwater to protect human health and the environment. It is a process that is managed by contracting environmental consulting companies, which periodically collect, analyse and interpret the groundwater monitoring site data sets. It is important that these results are presented in a clear way that benefits the understanding of contaminant migration in the subsurface.

Shell and the School of Mathematics and Statistics at the University of Glasgow developed software to aid the interpretation of groundwater monitoring data (Ground Water Spatiotemporal Data Analysis Tool, GWSDAT). GWSDAT is a user-friendly Microsoft Excel tool that interfaces with the statistical program R to statistically interpolate the groundwater monitoring data. For almost 10 years, GWSDAT has been applied routinely by Shell and its environmental consultants to assess its sites.

The use of open-source software in developing GWSDAT was important because it allowed complete transparency so that users and environmental regulators could see the code and techniques allowing full scrutiny and continuous improvement. Thanks to its open source foundation, GWSDAT has now become a standard, globally-adopted software application in the environmental monitoring industry. GWSDAT has already been downloaded over 10,000 times and is used by many engineers and scientists, environmental regulators, students and researchers around the world. The source code of GWSDAT is available from CRAN at https://cran.r-project.org/package=GWSDAT and from GitHub at https://github.com/levvers/GWSDAT.

Project details

Our previous research has shown the clear benefits of using spatio-temporal models for modelling groundwater contamination data rather than the prevalent, purely spatial models which fit temporally independent spatial models to each sampling time. By using spatio-temporal models we have shown that that spatio-temporal models can increase efficiency markedly so that, in comparison with repeated spatial analysis, spatiotemporal models can achieve the same level of performance but with smaller sample sizes.

Taking groundwater samples is expensive both in terms of time and money and so it is vitally important that as much information as possible is gained when a set of samples is collected. One aspect of this is the sampling design. This project will focus on spatio-temporal design of monitoring networks and sampling schemes for modelling contamination in groundwater. Up until now the design aspect of this problem has only briefly been considered, with the current spatio-temporal models being built using historical sampling information from real sites which were sampled with little design guidance. Therefore, by combining spatio-temporal models with optimal sampling designs there is the potential to increase efficiency further.

Another aspect is the modelling. We have already seen that using spatio-temporal models outperform repeated spatial analysis, but domain-specific models might further increase efficiency, requiring fewer samples and/or providing more detailed predictions, possibly also further into the future.

Some questions we will try to answer during this project are,

  • How can the sampling process be improved? How can we provide automatic guidance as to where the `best' position for new well would be? Similarly, can we provide automated guidance as to which wells contribute little information and are thus redundant?
  • Can we adapt the modelling so that we make more efficient use of the data and the geophysical understanding of the underlying processes?

A focus of the project is to develop methods that can be used in the day-to-day operations in industry. Thus, the project will involve developing computationally efficient algorithms and implementations so that the computations can be carried out in realistic time on a standard computer, so that an implementation can be included in GWSDAT. The users of GWSDAT are not data scientists, so another important aspect of the project will be the development effective ways of communicating the results of the method to users using suitable interactive plots, charts and tables of results.

This project offers the opportunity of significant methodological research work in addition to the application of the developed methods to data provided by Shell. Another major cornerstone of the project will be the development of software.

Key deliverables

  • Further develop and compare criteria for assessing the effectiveness of different spatio-temporal designs for different types of models (notably spline-based models and kriging).
  • Explore how our understanding of the underlying geophysical processes can be incorporated into the modelling.
  • Develop computationally efficient algorithms for model fitting and for computing and/or approximating different optimality criteria as well as optimising these.
  • Demonstrate and investigate the effectiveness of methods and strategies proposed using real groundwater monitoring data sets.
  • Implement the proposed methods in R and integrate the implementation into GWSDAT, so that it can be used by non-statisticians.
  • Document and communicate research findings and promote best practice to the company.

Research training and student experience

The successful candidate will be based in the School of Mathematics and Statistics at the University of Glasgow. The school has a vibrant and diverse community of postgraduate research students ( 60 students, of which 25 in Statistics). The School is based in a new building (built 2017). Postgraduate students have their own desks and computers in a shared office and access a dedicated large communal area for postgraduate research students.

In their first year, the successful candidate will take courses offered by the Academy for PhD Training in Statistics (APTS, http://www.apts.ac.uk) , undertake the generic skills and employability training offered by the University, and participate in postgraduate away-days which provide general research training, such as thesis writing in LaTeX and computational skills.

The project also offers insight into the operation of a large multinational company, further underpinning employability prospects on graduation. The student will have the opportunity to collaborate with Shell's Digital Centre of Expertise (https://www.shell.ai) which is part of Shell's Technology business unit which offers the opportunity to lead the transformation of an industry.

The student would be expected to be willing to travel regularly to the industrial partner's offices in London and Rotterdam once the travel restrictions put in place to address the COVID-19 pandemic will have been lifted. The cost of travel and accommodation will be met by the industrial partner.

How to apply

Please send your application, before June 29th , by email to ludger.evers@glasgow.ac.uk. Please include the following in your application.

  • a cover letter of at most two pages explaining why you are interested in the project and indicating what skills and ideas you would contribute to the project;
  • an up-to-date curriculum vitae;
  • two academic references;
  • evidence of a first degree in a relevant subject such as Mathematics, Statistics, Computing, Data Science or Hydrology (with good coding skill), which is of a good Upper Second Class or First Class standard, or equivalent (can be in progress provided you will complete the degree before September 2020); and
  • certification of proficiency in English (if required).

We anticipate to hold interviews in the week starting July 6th. You will be interview by the two academic supervisors and a representative of the industrial partner.

Informal enquiries

For further information and informal expression of interests please don't hesitate to contact the prospective academic supervisors.

Dr Ludger Evers: ludger.evers@glasgow.ac.uk
Dr Marnie Low: marnie.low@glasgow.ac.uk

published: 15 June 2020     Please mention EARTHWORKS when responding to this advertisement.