A Machine Learning Accelerated Spectral Matching Algorithm to Identify Harmful Algal Blooms and Other Threats to Aquaculture Operations.
Dr David McKee, University of Strathclyde; Prof Paulo A. Prodöhl, Queen's University Belfast; Dr Claire Neil, Scottish Environmental Protection Agency
Aquaculture is a vital economic driver (USD $313 billion in 2022) and a key protein source for a significant portion of the global population, making it essential for enhancing global food security. As this industry continues to grow, it is crucial to address sustainability challenges and support its efficiency. The aim of this project is to provide aquaculture operators with improved, space-generated water quality data, including information on harmful algal blooms (HABs) and other potential threats to their operations. This project will extend and validate an existing spectral matching algorithm that overcomes limitations in the performance of ocean colour data in optically complex coastal waters.
The core of the project involves a physics-based spectral matching algorithm, which we plan to operationalise using machine learning to accelerate processing for global applications. Additionally, the approach will incorporate metabarcoding of environmental DNA (eDNA) to provide species-specific information that complements optical data, enhancing the early detection and monitoring of HABs and other threats. The work will combine field and lab experiments, satellite image analysis, and algorithm development, and will be based in the University of Strathclyde's Physics Department with the Marine Optics and Remote Sensing group, in collaboration with the Fish Genetics Research Group in the School of Biological Sciences at Queen's University Belfast, Northern Ireland.
Students will receive training in essential lab skills, programming, and statistical data analysis, including bioinformatics and machine learning techniques. Upon completion, they will have developed expertise in satellite data interpretation, coding in Python/Matlab, and the practical application of machine learning. Additionally, students will gain expertise in metabarcoding of eDNA samples and the associated bioinformatics pipelines. We anticipate the production of a series of high-quality publications, providing a strong foundation for a career in academic research or industry.
