A pilot project in south-west England will use artificial intelligence to predict water pollution before it takes place, informs the BBC.
The pilot, run by CGI IT and Business Consulting Services Company, CGI and mapping company Ordnance Survey in North Devon Biosphere Reserve, will use data from sensors in rivers and fields, combined with satellite images of local land use. It will provide information about the state of local rivers and predict when they are more vulnerable to polluting events like runoff from agricultural fields, to manage fertilizer application accordingly.
The objective is to improve bathing water quality in the coastal resort town of Combe Martin, where River Umber – which receives discharges from a wastewater treatment plant and farm runoff – flows into the sea.
An AI model has been built which integrates data from 50 sensors across the catchment area, providing information on water and soil parameters, and from rain gauges. The idea is to understand the specific triggers of pollution events, for example, a particular rainfall event that washes off pollution into waterways. In the first phase of the pilot project, the AI model used historic data to predict pollution events with an accuracy of 91.5%, according to CGI. It is hoped it can do the same with real time data, and provide advice to farmers concerning the timing of fertilizer application, based on the likelihood of it being washed into the river. It could also predict when heavy rain may overwhelm the water treatment plant and result in raw sewage discharges.
"We're starting very small here (in North Devon) … but the idea is very much to scale up and roll this out to different parts of the UK," said Mattie Yeta, CGI's chief sustainability officer.
