Open-source AI Models Support Water Quality Monitoring
The collaborative innovation project, River Deep Mountain AI (RDMAI), has made a groundbreaking announcement with the open-source release of a suite of artificial intelligence and machine learning (AI/ML) models. These models promise to revolutionize how water quality data is collected and utilized, marking a significant advancement in environmental technology.
This initiative, funded by Ofwat’s Water Breakthrough Challenge and spearheaded by Northumbrian Water, with Spring Innovation as the knowledge-sharing partner, is a cross-sector effort. RDMAI focuses on creating open-source, scalable AI tools aimed at combating waterbody pollution and enhancing river health. By harnessing data from a multitude of sources, including citizen science and satellite imagery, the project team has successfully constructed these sophisticated models.
The release of AI/ML and remote-sensing models on the open-source platform GitHub represents the project’s first significant achievement, following the completion of its development and initial testing phases. During this process, the project team aggregated datasets from various sectors, conducted experiments with AI/ML models, and engaged in co-creation sessions with partners and stakeholders.
These models and datasets are designed to support:
At this initial stage, feedback is welcomed to further refine and improve the models as the project evolves.
The water environment in the UK is currently facing significant challenges due to factors such as population growth, climate change, pollution from diverse sources, and nutrient overload. Currently, only 14% of English rivers meet the Water Framework Directive standards for good ecological status.
Launched in July 2024, River Deep Mountain AI is set to tackle these challenges by developing open-source, scalable AI/ML models that uncover pollution patterns and provide actionable insights for safeguarding waterbodies.
The project is a collaboration with numerous partners, including ADAS, Anglian Water, Cognizant, Northern Ireland Water, South West Water, Stream, The Rivers Trust, Google, WRc, Wessex Water, and Xylem.
George Gerring, project lead at Northumbrian Water, expressed, “We have developed a robust set of capabilities utilizing artificial intelligence, machine learning, generative AI, and remote sensing to analyze and predict the variables impacting waterbody health. The open-source release on GitHub makes these models accessible to citizens, researchers, water organizations, and NGOs. Feedback on these early releases is crucial for refining and building upon our current achievements.”
Angela MacOscar, head of innovation at Northumbrian Water, added, “Data on waterbody health is currently fragmented and difficult to access. The RDMAI team is focused on deriving maximum actionable information from existing data sources. By integrating data from environmental sensors, satellite imagery, and citizen science, we are bridging data gaps in waterbody health and enabling quicker, more effective interventions. The open-source nature of these models heralds a significant shift in collaborative efforts to tackle environmental challenges.”
Stig Martin, global head of ocean at Cognizant, remarked, “This project exemplifies the power of research and development, showcasing the potential of technology in addressing complex, large-scale environmental issues. Our commitment to transparency is evident in making this project open-source, allowing everyone to understand the system’s construction and insight generation. Being a part of a collaboration that is actively building tools to drive change is incredibly fulfilling.”
The project is now entering phase three, which concentrates on model enhancement, validation in new catchments, and assessing the potential for scaling across the UK. The improved models are anticipated for release in November.
For more information, visit the GitHub page for RDMAI at https://github.com/Cognizant-RDMAI.
