Project team
UQ – Associate Professor Sarit Kaserzon
Exeter – Associate Professor Edward Keedwell
Project description
Background
Maintaining water security remains one of the greatest challenges globally. Accordingly, providing clean drinking water is a UN goal for sustainable development (SDG6). Incidents of drinking water contamination have increased over recent years, challenging current water resource management. Growing numbers of water contamination scenarios are reported (e.g. PFAS contamination in drinking water supplies), exacerbated by climatic events (i.e., floods, algal blooms, increased pollution). Concerningly, current monitoring practices involve several costly and disparate analytical techniques and only target a very limited number of regulated contaminants. The rate at which chemicals enter waterways far outpaces current regulatory methodologies. Therefore, strategies that can timely identify environmental and human hazards are paramount for adequate risk management.
Aims/Objectives/Approach/Deliverables
This project aims to place a student at the forefront of innovation and technology by developing the capabilities to enable robust identification of chemical threats in water systems, that is fit-for-purpose and adaptable to changing climate and environmental stressors. Such a tool does not exist, but is required to support water authorities, environmental and health protection regulators and water laws. Starting at UQ year 1, samples will be run using established HRMS methods at QAEHS and used to generate the data to build, train and test the ML models in Exeter (years 2-3). Key deliverables include:
- Obtain training set data from Australian/UK water treatment plants to establish ‘typical baseline chemical fingerprints’ (ground truth data), including time-series data.
- Develop train and test ML models based on HRMS water quality fingerprinting.
- Develop the anomaly detection mechanism using the sampling cycle, HRMS, ML and fine-tuning.
- Stress-test using different water baseline parameter scenarios (i.e. post floods with extremely turbid water or when water sources are mixed).
Main deliverable will be a highly novel PhD Thesis with several publications highlighting an open access ML tool, ready for validation in large test case applications, starting with water authority collaborators, followed by other facilities. Future commercialisation would also be a possibility.
Expertise/Facilities
QAEHS consistently maintains high research outputs and success in major Australian and international research grants. It operates through a state-of-the-art laboratory with instrumentation equipped for trace-micropollutant analysis (14xGC and LC-MS/MS Incl. 4xHRMSs), with a highly supportive environment for PhD students (~40 PhD’s from 15 countries). Facilities at Exeter include three complementary aspects; the Department of Computer Science (DCS), Centre for Water Systems (CWS) and the Centre for Resilience in Environment, Water and Waste (CREWW). In REF2021, 95% of research outputs in the DCS were rated internationally excellent, 41% as world-leading, with a cohort of ~70 PhD’s. Computationally, Keedwell’s group includes access to 2xhigh-powered workstations and server and the ISCA supercomputing facility.
Both UQ/Exeter groups have established long-term meaningful collaboration with the Australian/UK water industries. E.g. several technologies developed by CI Kaserzon’s team are today applied by industry in Australia and globally (>$9M funding). While CI Keedwell’s team have a proven track record working with the water industry on improving industrial knowledge systems and applying ML optimisation to solve problems in the water sector (>£5M funding; EPSRC, Innovate UK, EU and industry).
Contact
Questions about this project should be directed to Associate Professor Sarit Kaserzon k.sarit@uq.edu.au
