Real-time water quality monitors are being installed at wild swimming spots and beaches across southern England to help people assess their immediate risk of getting ill from polluted water.
Wessex Water is installing sensors at three freshwater sites in Dorset, Somerset and Hampshire, plus two coastal sites in Bournemouth, after a successful pilot study at Warleigh Weir near Bath. Here, the artificial intelligence-based system correctly predicted when bacteria in the water were high 87% of the time.
Southern Water is trialling a different monitoring system at Tankerton in Kent and Langstone Harbour in Hampshire, with a further sensor expected to be launched at neighbouring Hayling Island in the near future.
Although water companies and environment regulators now test river water for markers of pollution that harm wildlife, there is no requirement to test for faecal bacteria such as E coli and intestinal enterococci, which can cause stomach upsets, unless the site is a designated bathing water.
Even here, water samples must be sent to a laboratory for analysis, and the results are not published for a week or so after the sample was taken, with the final classification not published until November. This makes it difficult to assess the immediate risk posed by sewage or agricultural runoff.
“One of the biggest criticisms of the current bathing water classification system is that people may be swimming at a location all year, and it is only when the final classification is published that they realise what the quality has been,” said Ruth Barden, the director of environmental solutions at Wessex Water.
The company decided to use Warleigh Weir as a test bed for various water quality sensors and systems after pressure from the local landowner – a keen swimmer – to disclose how its storm overflows and treatment works were affecting water quality within the River Avon, and what they were doing to improve the situation.
Barden said: “We want people to be able to enjoy and use their watercourses without putting themselves in danger. But collecting this data is also helpful for us to understand what level of impact our assets are having on water quality – if any – and what the solutions could be to rectify that.”
One of the systems it tested was an AI-based approach developed by the UK-based startup UnifAI Technology. Rather than measuring bacteria directly, it infers when levels of E coli or enterococci are high, by analysing data from real-time sensors placed upstream, that measure pH, temperature, turbidity (cloudiness), dissolved oxygen and ammonia.
Water companies are obliged to start installing such sensors at sites up and downstream of where their storm overflows and wastewater treatment works discharge into rivers, as part of the Environment Act 2021, which aims to improve water quality and protect wildlife.
During an initial training period of about six months, river water samples are repeatedly tested for bacteria levels, and the AI learns to correlate these with patterns in the sensor data. After this period, members of the public are given access to an app that gives half-hourly predictions on bacteria levels in the river, and issues a water-quality alert when bacterial levels are likely to be high.
“We never tell swimmers that it is safe to get in the water; that’s a personal decision, and there are lots of reasons why it may not be safe. But an alert means there may be a problem with the water quality,” said Dan Byles, the chief commercial officer at UnifAI.
Wessex Water is installing the system at three more swimming spots – Farleigh Hungerford in Somerset, Fordingbridge in Hampshire and Poole Park lagoon in Dorset – plus Bournemouth and Boscombe piers, with real-time swim alerts expected to be available from 2025. The company is also in discussions with landowners and river users at 20 sites across south-west England about installing the system there.
The expansion of the technology could provide new insights into how storm overflow discharges and other events correlate with reduced water quality. Byles said: “We can effectively start to build up a digital twin of the river systems.”
The campaign group River Action UK broadly welcomed the technology but said it should not deflect from the urgent need to tackle root causes of pollution.
“Technological advances in real-time monitoring of river pollution are welcome, and well overdue, especially if they can be used to warn river users of dangerous pathogen levels proactively,” said River Action’s chief executive, James Wallace.
“We need the environmental regulators to use pollution data as evidence of malpractice to underpin enforcement of the law and hold polluters to account. Only when it becomes less expensive to abide by the law than breaking it will we see the necessary investment in maintaining and upgrading sewage treatment works.”
Southern Water is investigating whether UnifAI’s machine learning models could be applied to data from the sensors it is testing at coastal sites, and use its own algorithms to estimate bacteria levels, based on the scattering of light in surrounding waters. Although these sensors have been in the water for about a year, “we’re still in the science phase and not yet providing that data to the general public”, Southern Water said.
