The application of machine-learning to establish baselines and assess change in sensitive marine habitats
This project is based at SAMS UHI, Oban
Start date: 1 October 2024
Description: This is an exciting and unparalleled opportunity to join an international team of researchers using automated (machine-learning) based image analysis to map benthos on the Greenlandic shelf and shelf edge. The project is primarily based at the Scottish Association for Marine Science (SAMS) in Oban, Scotland but you will also be based at the Greenland Institute of Natural Resources (GINR) in Nuuk, Greenland, and will attend at least one offshore research cruise on the Greenland shelf. You will develop: 1) data analytical skills in automated benthic image analysis, 2) experience in polar and subpolar field work and, 3) a deep understanding of Arctic benthic ecology.
SAMS and GINR are highly supportive, well equipped, and internationally recognised research-orientated laboratories. The context of this PhD is the development of automated benthic imagery analysis pathways, in order to map benthos and support spatial planning, marine conservation and fishing activity on the Greenlandic shelf and shelf edge.
During your PhD you will meet weekly with your SAMS supervisory team (Tom Wilding, Director of Studies, co-supervisors Joseph Marlow [benthic ecologist] and John Halpin [marine image analysis specialist]). You’ll also have further expert input from your GINR co-supervisor Nadescha Zwerschke, who will provide advice on Greenland benthic ecology and will be your host in Greenland, and from colleagues at UHI-Shetland who will help you generate outcomes from your research. Field-work support will be provided by GINR and will take place on board the 60 m Research Vessel ‘Tarajoq’. At SAMS you’ll be joining ca 40 PhD students and will be encouraged to develop your broader skill base through further training and workshops. In Greenland you will be part of the Greenland Climate Research Centre, a highly interdisciplinary international research department of GINR which is supporting several early career researchers across different universities. You will also benefit from GINRs wider network amongst the Arctic Nations.
Data science, specifically AI-assisted image analysis, is a massive subject area and the skills you will learn will establish you as a leader in this field. Employment prospects of PhD graduates with these skills are exceptional. Previous PhD graduates (Wilding as DoS) are now in senior roles in government, consultancies and in academia. This PhD would suit a biologist/ecologist with an aptitude for programming or a data-scientist with an interest in ecology.
Your PhD will include four highly publishable areas, papers envisaged include:
- Optimisation of machine-learning to standardise & automate the analysis of benthic video imagery.
- An open-source trained machine for the automated identification of key north Atlantic /Arctic epibenthos.
- The distribution of benthic taxa along the Greenlandic shelf in relation to environmental and anthropogenic factors
- Temporal change in Arctic benthic seabed communities linked to human disturbances and climate change.
Background: benthic communities are an intrinsic part of ocean ecosystems, providing multiple ecosystem services, such as carbon storage, and acting as nurseries for commercial fisheries.
On the Greenland shelf and shelf edge, these benthic communities include significant assemblages of long-lived, ecosystem engineering species that are vulnerable to disturbance, such as sponges and corals. Change in these species is of particular economic and societal relevance to fishing communities (including those in Scotland, Greenland and the wider Arctic) that are reliant on demersal fishing, an activity which impacts the seabed. Understanding the drivers of change in benthic communities is essential to minimising and mitigating damage to this critical ecosystem and providing the evidence-base for effective management.
The GINR has collected more than 350 underwater videos of the Greenlandic shelf and shelf margins since 2015, within areas that are fished and unfished. Many of these videos have either not been analysed or been analysed using variously time-intensive methods that are liable to observer bias. You will develop an automated machine-learning based approach for the systematic analysis of these videos, producing a standardised map of benthic communities across the Greenlandic shelf. This will enable direct comparison of, for example, fished and non-fished areas, and results of this research will contribute directly to marine spatial planning in Greenland and fisheries management.
You will spend two three-month periods at the GINR in Nuuk, Greenland and attend at least one GINR research cruise (https://natur.gl/facilities/skibe/new-ship/?lang=en). This is an unparalleled opportunity to work with a renowned Arctic research community, and to gain fully-supported field-work experience in a remote and challenging environment. Critically, you will also gain a better understanding of the Greenland society that is so dependent upon the continued health and functioning of marine ecosystems. As part of the benthic team on the research cruise, you will develop expertise in the use of different sampling equipment and methodologies such as video sleds and drop-down cameras. This cruise will also be used to create the first repeat sampling stations on the Greenland Shelf. This will form the basis of future monitoring, providing a mechanism to detect change in benthic communities over time and allow for monitoring the consequences of climate change.
The PhD will build upon both SAMS expertise in arctic science and machine learning and GINRs unique dataset of seabed imagery from remote and challenging arctic areas as well as Greenland’s well equipped research-infrastructure.
Related projects currently underway by the supervisory team:
SEA-AI: https://www.sams.ac.uk/science/projects/sea-ai/
Arctic 3D: https://youtu.be/cByNe-jQWRw?si=zM9-_AYTvnqXXsbk
NS3D: https://www.sams.ac.uk/science/projects/ns3d/
INAMon: https://repository.oceanbestpractices.org/handle/11329/1864
SES: https://gcrc.gl/year/2023/project-funding-breathes-new-life-into-monitoring-greenlands-seafloor/
POMP: https://pomp-project.eu/
Supervisory team:
Dr Tom Wilding, SAMS UHI - VIEW PROFILE
Dr Joseph Marlow, SAMS UHI - VIEW PROFILE
Dr John Halpin, SAMS UHI - VIEW PROFILE
Dr Nadescha Zwershke, Greenland Climate Research Centre (part of Greenland Institute of Natural Resources) - VIEW PROFILE