Effective river management needs better data and smarter tools, says Stig Martin Fiskå, Global Head of Cognizant Ocean.
Rivers are vital to the UK’s ecosystems, communities and economy – yet they’re in crisis. Around 85% of English rivers currently fail to meet the Water Framework Directive’s ecological health standards. Pollution from sewage, farming and urban run-off continues to degrade water quality, with three in four rivers potentially posing a risk to human health.
Efforts to reverse this decline are often hampered by fragmented responsibilities, patchy data and outdated monitoring methods. Rivers are dynamic systems, but much of the current monitoring is manual and infrequent, with snapshots captured at most monthly. Without real-time insight, we’re always playing catch-up.
Real-time monitoring and artificial intelligence (AI) can make a real difference.
Why progress is slow
Responsibility for river health is shared across utilities, regulators, landowners and local authorities, making coordination difficult. Even the definition of a “healthy river” varies from catchment to catchment.
And while we collect river data through methods like geo-reporting, this typically provides limited, time-bound readings. Rivers change constantly, across seasons, and often from one hour to the next. Static monitoring isn’t enough to reflect these changes or guide timely interventions.
Real-time data and AI in action
The combination of AI and real-time monitoring is giving us the tools we need to bridge these gaps. Autonomous sensors can be deployed across river networks to monitor water chemistry, biodiversity and pollution loads. When combined with satellite imagery, drone footage and weather data, this creates an accurate and dynamic view of river health.
Rather than investing in costly new infrastructure, AI tools can also help make better use of what already exists. Machine learning models can be applied to existing sensor networks to fill in the gaps between data points, detecting patterns and predicting flow or pollution levels between measurements. For example, a pressure drop across pipes might indicate a leak, or a rapid shift in turbidity levels might point to a pollution event upstream.
At the same time, nitrogen and phosphor pollution is continuing to cause algal blooms and oxygen depletion that impacts people, marine life and the environment. Technologies such as geospatial AI and advanced video processing can be used to flag changes in nutrient levels or identify conditions that lead to algal blooms – allowing catchment managers to respond before damage is done.
In wastewater treatment, AI is helping to improve both efficiency and outcomes. Real-time monitoring of energy-intensive treatment processes can highlight savings opportunities, while AI models can analyse sludge composition to flag potential polluters before they reach the river.
Smarter river monitoring with Northumbrian Water
At Cognizant, our collaboration with Northumbrian Water, funded by Ofwat’s Innovation Fund, is a practical example of this approach. The initiative – called River Deep Mountain AI – brings together Northumbrian Water, Cognizant Ocean and a variety of industry partners and utility companies from across the UK and Ireland. Together, we’re developing a data-driven platform to monitor and improve river health.
By combining environmental sensor data, satellite imagery and weather inputs, the platform provides a detailed view of river conditions and pollution risks. This enables faster, better-informed decisions – from prioritising investment to targeting interventions and working more effectively with stakeholders.
The project’s long-term goal is to improve both drinking water quality and the wider ecosystem health of rivers. It also provides a blueprint for broader industry adoption, offering a responsive, AI-powered understanding of river systems that has not been possible before.
At the heart of the project is a commitment to open innovation. River Deep Mountain AI aims to co-develop open-source AI, machine learning and remote-sensing models that can detect pollution drivers like agricultural runoff, storm overflows and nutrient discharges. These models – including the first bare cropland detection tool – are now publicly available via GitHub, helping the entire water sector scale smarter, data-led approaches to pollution detection and catchment management.
From point solutions to systemic change
Mitigating water pollution depends on getting the right information to the right people, at the right time. Real-time monitoring of overflows, for instance, allows water companies and local authorities to anticipate flooding risks and alert communities early, protecting people and infrastructure. It also helps utility companies to operate more efficiently, identifying leaks, equipment failures or underperforming assets faster.
Single point solutions that are currently deployed will not be enough to effectively improve the health of our rivers. Systemic change relies on creating a holistic overview of a river. Through the implementation of data-powered tools, the end goal is to create a holistic water supply management system that can not only identify but predict incidents of pollution, allowing human intervention before accidents happen.
Water companies will thus be able to move away from purely reporting on their water sources but move towards mitigating and restoring damage.
Giving our rivers a voice
We now have the tools to understand rivers more deeply than ever. By investing in the right technology and working across organisational boundaries, we can give our rivers a digital voice – one that tells us where they’re under strain and how we can help.
Our work on the River Deep Mountain AI project is a step towards that future. With shared insight and open access to data and models, we can move faster, collaborate more effectively and build healthier, more resilient rivers.
