As part of River Deep Mountain AI, we are now releasing the first iteration of our Open Risk Map on GitHub. Our risk map enables stakeholders to quickly collate different pollution data sources and to get an overview of potential pollution hotspots across river catchments.

Water companies, catchment managers and other stakeholders face significant time and financial outgoings when attempting to design and implement water quality monitoring schemes within a catchment. Furthermore, it can be challenging to determine which sources of pollution represent the highest risk at different locations within a catchment, without prior water quality monitoring. 

To address these challenges, River Deep Mountain AI is providing a standardised and open approach to produce catchment-wide 'risk maps', helping to highlight areas where pollutants (e.g. phosphorus) pose a higher threat to river water quality. 

Our Open Risk Map can be used to inform where further on-the-ground walkovers may be needed and help highlight the optimal locations for targeting water quality investigations and monitoring.
 

Utilising openly available data to map phosphorous hotspots 

Our Open Risk Map is built up of individual ‘risk layers’, each of which aligns with a particular type (e.g. phosphorus) and source (e.g. combined sewer overflows, final effluent from treatment works, agricultural pollution, etc.) of pollution. Each ‘risk layer’ ingests openly available data to calculate a standardised ‘risk value’, which can be added to a baseline catchment map. Our Open Risk Map then determines how the level of risk varies throughout the river catchment system. 

The Open River Network (ORN) is used to create our baseline catchment layer within the Open Risk Map. 

In our first model iteration, we have focussed on creating a risk map of active phosphorus. Phosphorus is a high-priority pollutant that has significant impact on the ecological health of rivers. Our first model iteration ingests data for wastewater treatment works (WwTW) and combined sewer outflows (CSOs) to estimate the annual phosphorous load entering a river catchment at different locations.

This load is then converted to a concentration within the river channel and a proportional risk value is calculated using the UKTAG Revised Standard value for ‘poor’ water quality. This methodology is followed so that different pollutant types (e.g. phosphorus, nitrate, etc.) can be compared to each other later in our model development process.

Improvements for the second iteration of our Open Risk Map will introduce additional risk layers, with an initial focus on agricultural runoff and other sources of diffuse pollution. Furthermore, we hope to create risk maps for additional pollutant types (e.g. ammonia, etc.), alongside phosphorus to demonstrate the future capability of the tool. This will help enable stakeholders to gain a more holistic overview of a given catchment quickly, ahead of designing and planning detailed monitoring schemes. In addition, our Open Risk Map will provide stakeholders with a high-level pollution source apportionment overview of a catchment.
 

A collaborative and open-sourced approach

The overarching objective of River Deep Mountain AI is to bring key stakeholders involved in waterbody health together and collaboratively develop Open-Source AI/ML and remote-sensing models that can inform effective actions to tackle waterbody pollution.  

All our models will be released open source to democratise artificial intelligence and to benefit the entire water sector. The first iterations of our models are released now in May 2025, and the second iterations will be released in November 2025.

Access the first iteration of our Open Risk Map via GitHub.

River Deep Mountain AI is funded by the Ofwat Innovation Fund and consists of 6 core partners: Northumbrian Water, Cognizant Ocean, Xylem Inc, Water Research Centre Limited, The Rivers Trust and ADAS. The project is further supported by 4 water companies across the United Kingdom and Ireland.