The 2024 DANA floods created an unprecedented restoration challenge across Spain. They submerged the coastal lagoon of Albufera de Valencia under sediments, chemicals, heavy metals, and vast quantities of contaminated waste. The danger is real and very serious for the survivability of the ecosystem, as well as the ability to grow food in this 21000-hectare area. Albufera’s rice fields produce the varieties traditionally used for paella, tying the lagoon directly to how Valencians live and eat. Hence, restoring the lagoon’s health is not just a matter of environmental protection, but is essential to the survival of local communities and cultures, and requires a massive, well-coordinated effort.

To coordinate restoration efforts, a robust ecosystem health monitoring system is required. Here, the main problem lies in data heterogeneity. There are many available data sources, but they vastly differ in formats and communication protocols. Global data describing the whole Albufera is available through OpenEO as spectral bands from the Sentinel-2 satellite. Other sources provide information through IoT devices, REST APIs and CSV files. What’s more, because Albufera is a very dynamic and sensitive environment, local scientists and policy makers cannot rely on coarse datapoints or long-time averages. Real-time harmonization of heterogeneous and asynchronous data is absolutely necessary, making it an even more difficult problem.

For successful ecosystem restoration we need a solution that can turn this multi-ingredient data soup into well-structured ecological intelligence. We need a platform that would semantically unify the data, aggregate it, and turn it into a clearly understandable  indicators measuring the health of Albufera de Valencia and the progress of its restoration.

Satellite images of Albufera de Valencia, taken on October 8th and October 30th, 2024, showing the scale of the floods and their devastating effect, Source: USGS, processed by ESA, via CNN Weather

Enter: OBEREK.

As part of the bigger RURBANIVE EU project, OBEREK is building a Rural-Urban Enabler (RUE), which will serve as a trusted guardian watching over the restoration efforts and gathering all of the heterogeneous data. It will hear everything that Albufera has to say and tell us exactly what Albufera needs at any given moment. 

The data stack: What’s Albufera saying?

We take a holistic approach to guiding restoration and tracking Albufera’s health. Since it’s an "all-hands-on-deck” situation, monitoring all symptoms is essential. We use a diverse stream of information from a few vantage points:

IoT & LoRaWAN.

With the help of our partners at Universitat Politècnica de València (UPV), we are establishing a network of rugged sensors, monitoring the "vitals” of soil and water: pH, conductivity, oxygen content, and many more. This "microscopic” view takes information from 3 remote sensor sites and sends them over long ranges with minimal energy consumption via LoRa.

Copernicus – Sentinel-2.

Our eyes in the sky. The OpenEO API provides access to spectral bands, like RGB or infrared, captured by the Sentinel-2 satellite from 786 kilometers away! Through HTTP APIs, our platform retrieves the relevant bands for the whole area of Albufera and uses them to calculate relevant indices, like chlorophyll or vegetation coloring. Copernicus provides a big-picture view of the entire Albufera by default, but can also focus on specific areas.

Municipal data.

Data from locally available meteorological and water pumping stations. This is a reliable and thoroughly validated source that provides baseline knowledge on how the Albufera is feeling. Accessing this data is a tad more complex, as various sources have different access points. From authorized HTTP requests to unzipping Excel files and parsing a daily-updated CSV, the OBEREK RUE queries them all.

The Sentinel-2 satellite is capable of capturing images in over 12 spectral bands and a Scene Classification Layer, Source: ESA

The treatment: neurosymbolic AI

OBEREK’s main goal is to monitor the Albufera de Valencia in real-time. NeverBlink’s neurosymbolic platform is designed for exactly that – lightning fast data processing with dynamic knowledge graphs. But how do we make a knowledge graph from all of this data, if IoT sensors can come as JSONs, Copernicus indices as GeoTiffs, and water pumping stations share their information via CSVs? To bridge the gaps, we use semantic lifting. We don’t just convert the data, we transform it into RDF (Resource Description Framework) triples, which encode semantic information about the data alongside raw values. By assigning unique IRIs and using standardized ontologies, we restructure messy data into a machine-understandable knowledge graph. We integrate satellite imagery within the knowledge graph using our tensor-in-RDF approach, letting us manipulate multimodal data easily. This ensures that the neurosymbolic brain of our watcher not only listens to Albufera, but also understands what it’s saying, and through AI reasoning knows exactly which treatment to apply. 

Finally, the platform incorporates a persistence layer based on Apache Kafka, making the neurosymbolic pipeline virtually stateless.

The availability: you need it? It’s got it.

We use Jelly to make RDF communication in OBEREK as efficient as possible. However, interoperability and readability are also top priorities for OBEREK. The project should not just help the stakeholders and farmers, but encourage and enable data science communities around Spain to make projects around our platform. Which is why we save every useful data point, curated sensor data, metrics, and indicators, into 3 formats: our RDF serialization format – Jelly, and the industry standard JSONs and CSVs. By lowering RDF graphs and saving them to an S3 Bucket, we not only ensure that the user receives new readings in real-time, but they can access historical data with equal ease.

The understandability: you don’t need a PhD!

OBEREK distills the semantic soup of various sensor readings, satellite images and weather station JSONs into 3 clear Key Ecological Indicators (KEIs). KEI-1 and KEI-2 track the water quality and health of Albufera’s habitat. For KEI-1, we combine direct water-quality measurements (pH, turbidity, conductivity, and more) from IoT sensors with our Sentinel-2 indices and data from water pumping stations. KEI-2 works similarly, but the focus moves from water to vegetation: we track salinity and pH, data from the satellite and meteorological municipal stations. By cooperating with the domain experts over at Albufera (Fundació ASSUT), we establish the healthy target values for these readings and measure each reading against that. These differences then form our KEIs, and tell us how Albufera is feeling and if it’s getting better. The third indicator is the alignment with the community, where we ingest the communities’ opinions (polls, surveys deployed on a readable and accessible dashboard – available in 3 languages) and correlate their satisfaction about Albufera’s restoration against other KEIs.

If you’d rather not dig into the details, OBEREK has you covered: it summarizes Albufera’s health using a simple traffic-light system. In short, green means good, red means bad, and amber means “worth a closer look.” As Kevin from The Office put it: “Why waste time say lot word when few word do trick?”. 

The co-operation: everyone’s welcome!

The quad-helix model of OBEREK engagement.

OBEREK brings people together! A large, coordinated effort centered on a single unifying goal is the literal foundation of the project. The quad-helix model illustrates that perfectly. It engages the local communities, who are represented by Fundació ASSUT, the academia (led by UPV), the technological minds from NeverBlink, and the local governments and policymakers, who finally have a single, concrete source of information on which they can base their policies around Albufera.

The aftermath: what’s the future of the watcher?

Progress, people, progress! OBEREK will help us advance the technological readiness of our platform, applying it in a real-life use case. While it’s initially deployed for the benefit of the beautiful Valencian region, the whole solution is designed to be adaptable and reusable. We will publish it as a comprehensive package in the RURBANIVE Community Store, so other regions facing – or recovering from – ecological challenges can deploy the RUE as part of their toolkit. To make adoption seamless, we’ll provide clear tutorials, best practices, and thorough documentation – everything needed to implement it effectively.

Funding

The OBEREK project has indirectly received funding from the European Union’s Horizon Europe research and innovation programme under project RURBANIVE (Grant Agreement number: 101136597). Total project budget: €100,000.00. 

OBEREK's deliverables and public materials reflect only views of the OBEREK consortium. The European Commission or the RURBANIVE project are not liable for any use that may be made of the information contained therein.