Testing real-time bacterial-risk prediction across 20 sites.
A two-year research programme led by Yorkshire Water is gathering sensor and laboratory evidence across 20 sites to develop and test AI predictions of bacterial water quality.
The programme is supported by the Ofwat Innovation Fund through the Water Breakthrough Challenge 5 Catalyst stream.
- Status
- Current research programme
- Announced
- 22 May 2026
- Scale
- 20 sites over two years
Why 20 sites matter
A useful water-quality model must work with the conditions, sensor behaviour and sampling patterns found at its site. The programme tests whether what AI learns in one location can help it work in another, rather than assuming one model transfers everywhere.
This is a central challenge in scaling up AI-assisted monitoring: reducing repeated development effort while retaining evidence that the result is reliable for each setting.
Field monitoring and sampling across the programme
The programme will assess prediction performance at each site and test how learning from one location can help at another. It combines frequent multi-parameter sensor readings with laboratory reference samples.
The project depends on consistent field measurements as well as laboratory reference data. SOCOTEC is providing the sensors and managing the laboratory sampling programme across the 20 Yorkshire Water sites.
Those measurements and laboratory results are then brought together in DAVE to support evaluation and model development in ANNA.