SURFACE WATER INFORMATION MANAGEMENT (SWIM)

Project Information

SWIM

Grant agreement ID: 101180055

DOI

10.3030/101180055

Project closed

EC signature date 1 October 2024

Start date 1 November 2024

End date 30 April 2026

Funded under

• Digital, Industry and Space

Total cost

No data

EU contribution

€ 999 778,59

Coordinated by

O.M. OFFSHORE MONITORING LIMITED
Cyprus

Periodic Reporting for period 1 - SWIM (SURFACE WATER INFORMATION MANAGEMENT (SWIM))

Reporting period: 2024-11-01 to 2026-04-30

Summary of the context and overall objectives of the project

Water resource management faces increasing challenges due to pollution, climate change, and extreme weather events such as floods and droughts. Addressing these issues requires innovative solutions that enable accurate, real-time monitoring and informed decision-making. The SWIM (Surface Water Information Management) project responds to this need by developing a Copernicus-based application that integrates Earth Observation (EO) data, in-situ measurements, and artificial intelligence to enhance water management strategies.

The project aims to provide continuous, reliable monitoring of key ecological parameters, such as water quality, algal blooms, water quantity, and hydrological balance, while offering predictive analytics for risk assessment. The AI-powered decision support system will translate complex data into actionable insights, enabling water authorities and stakeholders to mitigate risks, optimize resource allocation, and enhance resilience to environmental changes.

SWIM aligns with global and European policy frameworks, including the UN Sustainable Development Goal 6 (Clean Water and Sanitation), the Water Framework Directive, and the Surface Waters Ordinance. By delivering a scalable, user-friendly tool, the project seeks to improve water governance and contribute to sustainable water management worldwide.

The expected impact includes better risk prediction for floods and droughts, improved water quality monitoring, and enhanced decision-making for authorities. By demonstrating its technological feasibility at TRL4, SWIM lays the foundation for further development, fostering innovation and international cooperation in Earth Observation-based water management solutions.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

SWIM (Surface Water Information Management) developed, integrated and validated a Copernicus-based decision-support platform for continuous, AI-informed monitoring of surface water. The result is an operational TRL 4 prototype that combines satellite Earth Observation (EO), real-time in-situ IoT sensing and machine learning to support decisions on water quality, water balance and natural-disaster risk.
EO and in-situ data ingestion (WP2). An automated, multi-source pipeline acquires, harmonises and pre-processes Copernicus Sentinel-1/2/3 data (through the Copernicus Data Space Ecosystem, atmospherically corrected with ACOLITE), the in-situ WAMO sensor network, and auxiliary datasets (ERA5-Land, MERIT Hydro DEM, GloFAS, HydroSHEDS, SMAP and GPM). It standardises the heterogeneous streams into unified spatial-temporal data cubes, processing roughly 5 to 10 GB of EO data per day.

WOLF calibration engine (WP2). The WAMO-OPIE Learning Framework dynamically matches satellite pixels with WAMO readings to correct atmospheric noise and mixed pixels. This reduces the uncertainty of Copernicus products and assigns a confidence flag to each value. A priority patent was filed for the dynamic water-quality calibration logic.
Three modules at TRL 4. A Water Quality Module (chlorophyll-a/NDCI, turbidity, total suspended matter, Secchi depth and algal-bloom indices) was deployed at La Fe Reservoir. A Water Balance Module predicts discharge in poorly gauged and ungauged basins from global data (SVR R²=0.92 NSE=0.91). A Natural Disaster Module provides flood and drought early warning, piloted at Las Palmas. WAMO was installed at La Fe on 12 June 2025 and transmits 19 parameters hourly via LTE.

AI-DSS, software architecture and user interface (WP3 and WP4). The system runs on a scalable OPIE REST backend (PostgreSQL/PostGIS, dual-server, HTTPS/SSL) with an ML pipeline (SVR, Random Forest, XGBoost), a multilingual LLM-based chatbot in Spanish, German and English for data retrieval, statistics and anomaly detection, and an interactive TypeScript/MapLibre/Three.js dashboard with role-based views and automated alerts. End-to-end integration across five layers was tested in April 2026 (D4.3).

Requirements and end-user engagement (WP4). More than 30 prioritised functional and non-functional requirements were gathered through fieldwork, interviews and workshops with EPM, the Municipality of Envigado and the University of Envigado, and were formalised through MoUs (D4.1 D4.2).
Dissemination, exploitation and communication (WP5). The consortium delivered the brand identity and website (theswimapp.com) social media activity, a concept demonstration video, webinars, FAIR-compliant open-data practices, an IPR assessment, a preliminary Business Innovation and Commercialisation Plan with six exploitable results and routes to market, and a future-evolution roadmap from TRL 4 toward operational deployment. Management (WP1) covered data security, GDPR compliance, the Data Management Plan and timely reporting.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

SWIM advances the state of the art by combining Copernicus EO with high-frequency in-situ IoT data into a single calibrated and auditable surface-water decision-support service. The specific advances are:
• WOLF calibration engine. A machine-learning approach that dynamically harmonises EO-derived water-quality estimates against patented WAMO in-situ measurements and attaches a confidence flag to each value, which measurably reduces the uncertainty of global Copernicus products. It is protected by a priority patent filing.
• Discharge prediction in ungauged basins. An ML pipeline (SVR, Random Forest and XGBoost with temporal feature engineering) that predicts streamflow from only globally available data (ERA5-Land, HydroSHEDS, GloFAS), reaching R²=0.92 and NSE=0.91 with no basin-specific calibration. This makes it usable in data-scarce regions.
• LLM-powered hydrological DSS. A multilingual conversational chatbot that turns complex hydrological data into plain-language answers, statistics and anomaly diagnostics, which lowers the expertise barrier for water managers.
• End-to-end, modular and auditable architecture. Calibrated, contextualised and explainable outputs are delivered through a role-based dashboard, with metadata and quality flags exposed at every step. The utility and public-sector markets require this for procurement.

Key needs to ensure further uptake and success. Further public and private funding to move from TRL 4 to TRL 7 and beyond; multi-site validation and productisation (service levels, connectors, reporting templates); completion of the patent process and dual-licensing of the GPL-3.0 and proprietary layers; access to risk capital and scale-up finance; internationalisation beyond the Colombian and EU pilots; and a supportive regulatory and standardisation framework for EO-derived water products.