As part of River Deep Mountain AI, we are now releasing the first open source remote-sensing model capable of detecting bare winter cropland, which represent potential sources of pollution in river catchments.

In England, agriculture is one of many sources of pollution impacting water quality. For instance, around 60% of nitrate, 25% of phosphorus and 75% of sediment in water bodies is estimated to originate from agriculture1. These pollutants, together with others, contribute to the fact that only 14% of water bodies in England meet the “good ecological status” defined by the Water Framework Directive.

When assessing the pollution risk of a river catchment, it’s often necessary to do walkovers next to the river to detect areas with increased run-off risks. These walkovers are time-consuming, inefficient and ideally need to be undertaken during wet weather to give an accurate understanding of potential risks.

To try and help reduce the financial costs associated with time-consuming walkovers, we are developing several remote sensing-based models utilising computer vision and satellite imagery to detect potential sources of pollution. Now we are releasing the first one, focused on detecting bare croplands.

These models will help enable catchment managers and other stakeholders, to quickly pinpoint some of the potential sources of pollution within a catchment, and conduct targeted investigations.
 

Detecting bare cropland 

Bare cropland detection is useful for identifying areas with greater runoff risk, where pollutants like fertilisers, pesticides, and sediment are more likely to be transported into nearby water bodies during periods of rainfall. Identifying bare cropland can assist in pinpointing potential pollution hotspots and implementing targeted measures to mitigate water contamination.

Our remote sensing-based model focuses on identifying bare cropland, primarily utilising the Bare Soil Index (BSI). To support and enhance the accuracy of BSI, we also analyse Soil-Adjusted Vegetation Index (SAVI), Normalized Difference Moisture Index (NDMI), and Normalized Difference Snow Index (NDSI). This methodology allows for monitoring across multiple time periods during a year, ensuring that changes in land cover are accurately tracked and managed. The method demonstrates an accuracy exceeding 80% ensuring the model can efficiently detect bare cropland across any given date range, offering a reliable tool for environmental monitoring and management.

Future improvements will focus on refining the model's applicability and increasing detection reliability through additional data sources and advanced modelling techniques. We also plan to explore other features and parameters that might contribute to more accurate land cover detection, ultimately aiming to develop a model capable of providing comprehensive insights into land use and its impact on water quality.
 

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 models that can inform effective actions to tackle waterbody pollution.  

All our models will be released open source to democratise artificial intelligence and 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 versions of our AI and remote sensing-based models open source 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.