SWIM Quarterly Update, Q4 2025: From Data to Decisions

As SWIM enters the second half of its 18-month programme, the focus has shifted from data acquisition to intelligence: turning raw measurements into decision-ready outputs.

Machine learning pipeline delivered (WP3)

The Machine Learning Solution Package (D3.1) has been completed by consortium partner IHCantabria. The ML pipeline compares three algorithms (Support Vector Regression, Random Forest, and XGBoost) for streamflow prediction from globally available data. The SVR model achieved strong performance with R-squared = 0.92 and NSE = 0.91 at the Colombian pilot sites. A reusable Python module (SWIM.py) has been released under the GPL-3.0 open-source licence and is available on the SWIM Open Research page.

DSS design and AI integration (WP3)

The Decision Support System architecture has been defined, featuring a complete ML pipeline, an EO Water Quality processing chain, and the initial design for a multilingual LLM-based chatbot. The system will support data retrieval, statistical analysis, and anomaly detection in Spanish, German, and English.

Software architecture (WP3)

The backend architecture (documented in D3.2) runs on a dual-server configuration with PostgreSQL/PostGIS for spatial data storage and HTTPS/SSL encryption. The system is designed for scalability and security.

User interface prototyping (WP4)

Interactive UX mock-ups and the alert system prototype (D4.2) have been developed, featuring four main views: Dashboard/Home, Map and data visualisation, Alerts and Notifications, and Sensor/Site detail view.

Concept demo video (WP5)

The concept demonstration video (D5.3) has been produced, a concise overview of SWIM's capabilities and vision, targeting utility managers, field operators, investors, and policy officials. The video is available on the SWIM website.

Looking ahead

Q1 2026 will focus on system integration, iterative user testing, and preparation of the Business Innovation Plan. The AI chatbot will be developed and integrated with the DSS backend.

The SWIM project (grant agreement No. 101180055) is funded by EUSPA under Horizon Europe.

Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or EUSPA. Neither the European Union nor the granting authority can be held responsible for them.