As part of the EU-funded D-HYDROFLEX project, the Mauzac HPP demo is showcasing how digital technologies can transform traditional hydropower infrastructure into a flexible, secure, and sustainable pillar of Europe’s energy transition.
Demo Case at a Glance
How to reduce impact of hydropower on fish using acoustic imagery
Location: Dodogne River, France
Lead Partners: EDF, EDF R&D
Demo Goals:
- Automatic detection and identification of fish species based on acoustic videos
- Automatic quantification in real time of fish migration / fish presence
The most common ecological impacts of hydropower plants (HPP) are flow regime alteration, loss of longitudinal connectivity, fish injuries, habitat degradation, fish community composition simplification. Concerning impact on ecological continuity, different methods are used to evaluate the efficiency of fish passes to estimate the migratory species abundance and analyse changes in their within-river distributions. In particular, hydropower plants must better understand and mitigate their impact on fish migration. Within this context, the action presents an innovative digital framework designed to perform biodiversity monitoring into an automated and operational tool.
Figure 1 : Eels detection on acoustic videos and floating platform for the expérimental acquisitions (Mauzac Dam)
Detect and predict the fish migration periods is a crucial concern to mitigate the impacts of hydropower plants during key periods of life history
Challenges Faced
The demo tackled key barriers such as:
- Fish species identification and quantification: Unlike optical cameras, acoustic camera operates effectively in turbid waters and low‑light conditions, allowing continuous monitoring without disturbing fish populations, does not provide;
- Interoperability between different acoustic camera models: converting proprietary sonar data into standard video formats while preserving image quality. This ensures interoperability and allows the use of conventional image-processing techniques;
Real time processing: Extracting relevant information from acoustic videos, given their large volume, requires the development of an automated analysis method, especially to meet real-time constraints.
Innovative Solutions Delivered
- Eel Counter Software: A user-friendly human-machine interface has been operational at the Mauzac plant since December 2024. It allows operators to configure parameters (camera model, detection ranges, minimum length) and visualize results in near real-time, saved in CSV files;
- RecoFish Software (generic): Under development for other species. Activities include algorithm improvement, integration of new features (real-time processing, better resolution), extensive testing, field and laboratory data collection, integration of environmental data, and user interface improvement;
- Swimming locomotion consideration: An innovative approach has been developed to species identification based on body deformation and swimming patterns. Since acoustic data do not provide sufficient visual detail, the method models fish as deformable objects;
- User Interface: An interface integrating the tool dedicated to eel quantification has been developed.
Impact & Future Outlook
Early results show:
- Data Collection and Analysis: Gather data from multiple demonstration sites, including the Mauzac HPP and Selune River, to build a comprehensive database of fish species and their behaviors. Perform controlled experiments in laboratory settings to collect high-quality acoustic video data of fish swimming under different water velocity conditions and turbulence;
- Deep Learning Integration: Explore the use of deep learning techniques to enhance fish identification accuracy and reduce uncertainty;
- Environmental Data integration: Integrate environmental data such as water velocity, river flow, and temperature into the software to improve the accuracy of fish behavior.
Voices from the Demo
“Fish monitoring framework demonstrates how digital innovation can help environmental management in hydropower. By combining acoustic videos, advanced image processing, and behavioral modeling, it provides a scalable solution capable of delivering real time, non-intrusive, and reliable fish quantification. This approach enables hydropower operators to better reconcile energy production with ecological preservation, contributing to the long-term sustainability of the sector “, said De Oliveira Eric and Zaoui Fabrice, Researchers-Experts from EDF R&D.”
Figure 2: : EDF R&D Team – Left: Dr. Fabrice Zaoui – Right : Dr. De Oliveira Eric
To learn more, please visit https://d-hydroflex.eu/ to explore all demo sites and follow updates from the project. Follow us on LinkedIn and Twitter for the latest news and insights.
