AI insights for river management

Project summary

River Deep Mountain AI (RDMAI) is a large, collaborative Ofwat Innovation Fund project involving RSK Group companies WRc and ADAS, as well as Cognizant Ocean, Xylem, The Rivers Trust, Stream and five supporting water companies: Northumbrian Water, Wessex Water, Anglian Water, Northern Ireland Water and South West Water. The project has delivered eight open AI machine learning (ML) models that can help inform the management of our waterbodies. These models demonstrate the viability of using AI/ML to inform water quality management and offer significant benefits to the water sector.

The background

Healthy rivers support public health, local economies and long-term environmental resilience. In 2024, only 14% of rivers in England, and 44% in Wales, are considered to have a ‘good’ or better ecological status. The Cunliffe Review, published in July 2025, emphasises the importance of digital technology and data in improving water management and decision-making. The water sector needs innovative solutions to ensure the health of our rivers and clean water for future generations. With the increasing accessibility of artificial AI/ML in recent years, there is an opportunity to exploit it for improved insights on waterbody pollution.

To address this need, RDMAI has delivered eight open AI/ML models that can help inform the management of our waterbodies. The project was launched at the Northumbrian Water Innovation Festival in July 2024, progressing through stakeholder engagement and testing in early 2025, and the final model was released in February 2026.

The models are:

• Open Flow Model
• Open E. coli model
• Open Orthophosphate model
• Open Slurry Tank Detection
• Open Bare Cropland Detection
• Open Poached Land Detection
• Open Risk Map
• Open Anomaly Detection Model

WRc has built a strong, ongoing relationship with Northumbrian Water through the successful delivery of RDMAI, which has directly led to further collaboration on a £2 million Ofwat Innovation Fund project, Open Catchment Intelligence, reinforcing a long-term partnership focused on innovation and improved water quality management.

Multidisciplinary approach

RDMAI brought together a unique combination of skills across WRc and ADAS, integrating expertise in water quality, catchment management, data science, agronomy and environmental monitoring. WRc led on river water quality domain expertise, model review and practical application, ensuring outputs were robust, relevant and aligned with real-world operational needs.

ADAS contributed specialist knowledge in land use, agriculture and environmental systems, particularly supporting feature engineering and interpretation of remote sensing outputs. This close collaboration enabled the development of models across a full spectrum of environmental and hydrological areas, all of which were grounded in practical application, demonstrating the value of an integrated, multidisciplinary approach that maximises the strengths of each partner.

The challenges

The project addressed a number of sector-wide challenges, including the rapid growth in environmental data volumes and the difficulty in extracting meaningful, actionable insights. Ensuring data quality, consistency and accessibility across multiple sources was a key obstacle, alongside integrating AI and ML approaches into a traditionally risk-averse industry. It was also necessary to ensure that models were not only scientifically credible but also practical, interpretable and trusted by end users. Coordinating inputs and requirements across a large and diverse partnership of organisations added further complexity.

The solutions

These challenges were addressed through the development of eight open source AI and machine learning models, designed to transform complex data sets into clear, actionable insights. A strong emphasis was placed on transparency, usability and validation, including independent benchmarking to build confidence in model performance. Close engagement with stakeholders throughout the project ensured that outputs were aligned with real-world needs, while the use of open data platforms improved accessibility and integration. The combination of domain expertise, advanced analytics and continuous collaboration enabled the successful deployment of innovative yet practical digital tools.

The impact

WRc and ADAS played a central role in ensuring that the project delivered meaningful, real-world impact, shaping the development of models that are both scientifically robust and operationally relevant. By combining technical innovation with practical application, the project has accelerated the adoption of AI-driven approaches within the water sector. RDMAI is helping to transform how water quality is monitored and managed, enabling more proactive, data-led decision-making. The open source nature of the outputs ensures long-term value, supporting ongoing innovation, wider collaboration and tangible improvements to environmental and community outcomes.

The models offer significant savings in money and time. For example, the Open Anomaly Detection Model could save a water company operating 4000 water quality sondes probes as much as £42,000 per year; the Open Risk Map can save the water industry up to around £5000 per year per user on software licence fees; used as an alternative to laboratory measurements, and the Open E. coli model shows potential money and time savings of as much as £200,000 and 6000 days, respectively, for a 2000-point data set.

The project has received the following awards:

• 2024 Utility Week Awards Collaborative Excellence Award winner
• 2025 Utility Week Awards Unlocking Data Award finalist
• 2025 CIWEM Urban Drainage Group Annual Conference Best Paper Award