Context and objectives

The Dranse de Bagnes Valley aquifer is equipped with nine sensors that measure its level, temperature, and conductivity. The objectives of the project are to:

• Automatically detect sensor measurement errors
• Issue an alert when the groundwater level is too low
• Estimate missing groundwater level values
• Explore predictions of future groundwater levels

Objectives

QleanUP (Holistic approach towards high quality urban water) offers a comprehensive approach to:

• Develop innovative sensors for nutrients, heavy metals, pathogens, microplastics, and PFAS,
• Create soft sensors and digital twins capable of predicting water quality and identifying sources of pollution,
• Design data-driven risk management plans, integrating short- and long-term scenarios.
• Deploy an interoperable digital platform to support real-time decision-making.
• Involve citizens and local stakeholders in data collection and the co-creation of solutions.

Results

Automatic Detection of Measurement Errors and Missing Values

A program was implemented to detect measurement errors (mainly value jumps). It can be useful for detecting errors in real time for sensors that transmit data every hour. SARIMAX was used to estimate missing values based on the values of other sensors while accounting for seasonality.

Low-Level Alerts

A statistical analysis is used to detect low groundwater levels. The level averaged by month is compared with those of available previous years. The lowest months are shown in red, average months in yellow, and the highest in green. Figure above shows the monthly situation for 2023, when the aquifer level was lower than usual.

Prediction

SARIMA methods can also be used to predict the future level of the aquifer. However, we did not obtain good predictions. Better predictions would require a more in-depth sensor-by-sensor analysis, taking into account other parameters and variables, such as seasonal weather forecasts.