A pilot project in Devon, southwest England, aims to utilize artificial intelligence (AI) to predict pollution and take preventive measures.
The initiative, led by computer systems company CGI and mapping experts Ordnance Survey, aims to enhance water quality at the seaside resort of Combe Martin.
Sensors placed in rivers and fields will collect data on local river conditions, rainfall, and soil quality.
This information will be combined with satellite imagery of the area’s land use. By analyzing this data, the AI system will predict when the local river system is most susceptible to pollution sources like agricultural runoff.
This will enable appropriate actions, such as requesting farms to delay fertilizer application.
The project is being piloted in the North Devon Biosphere Reserve, a protected area encompassing natural habitats, farmland, and small towns, covering 55 square miles (142 square kilometers).
During initial testing, the AI system demonstrated over 90% accuracy in predicting pollution events.
Combe Martin, known for its longstanding concerns about bathing water quality, stands to benefit from this project. The town’s water quality rating, which received a ‘good’ designation last year by the Environment Agency, is mainly attributed to dry weather conditions.
In typical years, the rating has been ‘poor,’ leading to swimming advisories and potential negative impacts on local businesses.
According to Andy Bell from the North Devon Biosphere Reserve, the main culprit for water pollution is the River Umber, which carries pollution from a sewage treatment plant and agricultural runoff.
The AI project emphasizes the importance of real-time information to tackle this issue effectively.
About 50 connected sensors are being deployed across the catchment area to gather real-time data, monitoring indicators like water acidity, ammonia levels, dissolved oxygen, and water clarity.
Ordnance Survey is contributing mapping expertise to integrate this information with location-specific data and satellite imagery.
The hope is that the AI system will provide actionable insights, such as advising farmers to refrain from fertilizing fields when heavy rainfall is forecasted.
While addressing raw sewage discharges during heavy rainfall is more complex, the AI system may still assist in predicting such events, albeit with potential limitations due to infrastructure capacity.
Having successfully completed the initial phase using historical data, the AI model is now being tested in real-world conditions.
If proven effective, the project aims to scale up and implement similar initiatives across different regions of the UK.
