New algorithm can predict water quality

New COWI tool can be used for assessing pesticide vulnerability of groundwater. That will provide more accurate information that will be helpful when determining the location of new wells.

Since the 1990s, pesticide occurrence in groundwater has caused challenges for an increasing number of water utilities and waterworks, leading to wells being closed. New pesticides and breakdown products result in new problems for water utilities, and of an unprecedented scale.

COWI has developed a new tool that will allow us to plan future water catchment without getting pesticides in our drinking water. Using machine learning, we have developed an algorithm that predicts the risk of pesticide or breakdown product occurrence in groundwater at a specific geographical point.

The algorithm is fed public environmental data that describes soil and groundwater conditions. That includes analysis data on pesticides and breakdown products, general groundwater chemistry and hydrological parameters. This data is linked with other significant factors such as precipitation, soil conditions and mapping of land use. Data is compiled into a series of descriptive parameters that form part of the final model. The descriptive parameters were carefully selected to provide the optimum basis for the final team. The tool consists of a GIS layer for every aquifer that shows the risk of pesticide occurrence at every point.

Measuring data can be adjusted to different geographies across Denmark, where different pesticides have been used for various purposes. We hope that the tool will contribute to future planning and assessments related to pesticide contamination.

The project is funded by COWIfonden and carried out in cooperation with Greater Copenhagen Utility (HOFOR), VCS Denmark (Vandcenter Syd) and Aalborg Forsyning. The new tool will be tested in 2020 in collaboration with four Danish water utilities.