Early Detection and Effective Management of Olive Fruit Fly: A Field Demonstration

How can digital technologies help olive growers detect pest threats earlier and make better decisions in the field?

This question was at the heart of a field demonstration organised by NILEAS, as part of the AgriDataValue project, specifically Use Case 3.4, which focuses on olive fruit fly control.

The event showcased the use of digital olive fruit fly traps and environmental sensors as complementary technologies for improving pest monitoring and supporting more effective management of the olive fruit fly (Bactrocera oleae), one of the most significant threats to Mediterranean olive production.

Digital traps enable the monitoring of olive fruit fly populations, providing valuable information on pest activity and helping growers identify potential risks at an early stage. At the same time, environmental sensors collect data on conditions such as temperature and humidity, which can influence pest development and activity. By combining pest monitoring data with environmental information, these technologies can provide a more comprehensive understanding of conditions within the olive grove, supporting timely, targeted, and evidence-based pest management decisions.

Beyond the technological demonstration, the field event provided an opportunity to explore the practical application of these tools in everyday olive farming and discuss their potential to improve pest management efficiency while reducing unnecessary pesticide applications. The combination of digital traps, environmental sensors, and farmers’ practical knowledge represents an important step towards smarter and more sustainable olive production. Furthermore, the field event highlighted how digital technologies can transform pest monitoring into actionable knowledge, helping farmers protect their crops while making more efficient use of agricultural resources.

Read More

AgriDataValue participated at the 10th European Big Data Value Forum (EBDVF) 2026.

AgriDataValue participated in the 10th European Big Data Value Forum (EBDVF) 2026, held from September 30 to October 1 at the Dexcom Stadium in Galway, Ireland. As an official associated Irish EU Presidency event supported by the European Commission and Fáilte Ireland, the forum brought together Europe’s data, AI, and research communities under the theme “Building Europe’s AI and Data Backbone: Strategy, Investment and Impact.”

AgriDataValue representatives from ITA and Netcompany participated in the dedicated session “From Agrifood Data Spaces to Real Business Value.” This session addressed the transition from theoretical data frameworks to practical commercial applications within the agricultural and food sectors.

Discussions centered on real-world pilot experiences, emerging business models, and strategies for scaling sustainable data-driven services. The session provided a platform for industry leaders, researchers, and public sector experts to discuss secure, collaborative data-sharing ecosystems for European agriculture.

Read More

From Orchards to Algorithms: AgriDataValue at the 33rd IGU-CSRS Colloquium in Genoa

In July, the University of Lodz team presented the AgriDataValue project at the 33rd Colloquium of the International Geographical Union Commission on the Sustainability of Rural Systems, hosted by the University of Genoa.

This year’s theme, „Rural Systems in an Age of Local and Global Mobilities” brought together scholars from the Global North and South to discuss how flows of people, climate, capital and culture are reshaping rural areas. Presentation of the AgriDataValue project added one more flow to that map: the mobility of knowledge.

Karolina Dmochowska, Marcin Wójcik, Paulina Tobiasz-Lis and Patrycja Grzyś asked what happens to farming knowledge that was never written down. Fruit growers learn to read water stress in a leaf, a crack in the soil or the colour of the sky over decades of apprenticeship in one place. As Europe’s farm population ages and young people move to cities, that chain is breaking, and a lifetime of know-how can disappear when a grower retires.

At Wiatrowy Sad near Głowno (Poland), one family has been growing apples for more than 60 years. There is no irrigation on these rolling moraine slopes, so knowing when trees are running short of water has always depended on the grower’s eye and memory of past seasons. In AgriDataValue, the UniLodz team is working with the orchard to translate that judgement into data: a weather station and soil-moisture probes feed a machine-learning model that forecasts water-stress risk before the trees show it, and explains why. The aim is not to automate the grower’s intuition but to give it a form that can outlast one working life and be shared with the next generation.

The teams’s central argument is that explainable AI should work as a dialogue between generations, not as a replacement for the experienced grower. The algorithm needs the grower’s validation, and the grower gains a language that can be passed on to young and returning farmers.

This is a starting point, not a closed result. Over the next three growing seasons (2026–2028), the researchers will study what is gained and lost when tacit knowledge becomes explicit, whether young farmers actually adopt it, whether explainable AI builds more trust than black-box tools, and who risks being left behind by the digital divide.

Many thanks to the organisers at the University of Genoa and to everyone who joined the discussion!

Read More

From data to field decisions: AgriDataValue fertilisation model enters practical testing in leek

After three years of development, the AgriDataValue leek fertilisation model is taking an important step from research to practice. In 2026, the model is being tested in a field trial at Inagro, where data from soil, weather and drone observations will be transformed into practical fertilisation advice.

Putting the model to the test

The trial was planted on 2026 and consists of five fertilisation strategies, each repeated four times. By comparing these approaches under the same field conditions, the researchers can assess how well the AgriDataValue model performs against established recommendations and common advisory practice.

The unfertilised plots provide a clear baseline and are expected to show the effect of nitrogen shortage on crop development and yield. Two reference treatments represent current practice: one follows an existing fertilisation model, while the other is based on advice from an Inagro crop adviser. The final two treatments are guided by the AgriDataValue approach. Both received a limited mineral fertiliser application of 50 kg of active nitrogen per hectare in June, after which further decisions are supported by data and model predictions.

Data collection in the field

A nearby weather station continuously provides the meteorological data needed by the model. Soil nitrate samples are needed to know the nitrogen already available to the crop, while drone imagery captures the crop growth and spatial differences in crop growth and development across the trial field.

Within AgriDataValue, Siemens uses these inputs to develop machine-learning predictions for key indicators, including biomass at harvest and total nitrogen uptake by the leek crop. The model will translate multiple streams of information into knowledge that can support timely and better-informed decisions.

Towards fertilisation that balances yield and nitrate losses

For Inagro advisers, the trial provides an opportunity to evaluate whether data-driven recommendations can complement their experience and help tailor fertilisation more closely to actual crop needs.

For farmers, this could mean using nitrogen more efficiently, reducing unnecessary inputs and nitrate leaching to the environment while maintaining crop quality and profitability.

As the season progresses, the five treatments will be monitored and compared. Yield, crop quality, nitrogen uptake and model accuracy will reveal how the AgriDataValue strategies perform under real field conditions. The results will help refine the model and bring it another step closer to reliable use in leek production.

Read More

Smart Irrigation: Integrating Soil Moisture Sensors, QMS and Drip Technology

Every drop is evidence-based

Soil moisture sensors provide insights, QMS Water translates the data into irrigation recommendations, and drip irrigation enables targeted implementation. The integration of these components represents a significant step towards smarter and more evidence-based water management.

Precision irrigation begins with reliable measurement, but only truly delivers value when data is converted into irrigation that matches crop requirements. Within AgriDataValue, Delphy is investigating how soil moisture sensors, QMS Water and drip irrigation together form a single data-driven chain.

Using water more intelligently

The availability of fresh water is coming under increasing pressure. At the same time, the agricultural sector is expected to produce more efficiently, whilst using natural resources carefully. Precision irrigation offers opportunities in this regard. As part of the European AgriDataValue project, we are investigating how digital technology can support growers in making better irrigation decisions. The key lies not in any single technology, but in the integration of soil moisture sensors, the QMS Water decision-support system and drip irrigation. Together, these components form the basis for data-driven water management.

Soil moisture sensors: insight into the root zone

In practice, irrigation is often based on experience, field observations and an assessment of weather conditions. Soil moisture sensors provide continuous, objective information about the soil’s moisture status.

•          the amount of water available to the crop;

•          the rate at which the soil dries out;

•          the distribution of moisture across the root zone;

•          the effect of rainfall and previous water applications.

The quality of the measurements is crucial in this regard. A representative location, a suitable sensor type, correct placement and a stable data connection are required to make the measurements usable for irrigation control.

QMS Water: from measurement data to irrigation advice

Individual sensor measurements become more valuable when assessed in context. QMS Water combines soil moisture data with crop evapotranspiration, crop water requirements, rainfall and weather forecasts. This results in advice on the timing and volume of irrigation. This shifts decision-making from reacting to visible drought stress to looking ahead based on current moisture reserves and expected developments.

Drip irrigation: targeted application

Drip irrigation delivers water close to the root zone and enables small, frequent applications. This aligns well with digital recommendations that can be translated, on a plot-by-plot or irrigation-section-by-section basis, into specific system settings.

•          targeted application in the root zone;

•          flexible watering in small increments;

•          options for section-by-section control;

•          a useful basis for further automation.

 

The data chain at a glance

Added value is created when measurement, analysis, advice and implementation are organised as a single, coherent process.

1          Soil moisture sensors

Continuous measurement of moisture levels in the root zone.

▼

2          QMS Water

Combination of sensor data, crop requirements, evaporation, precipitation and weather forecast.

▼

3          Irrigation advice

Advice on when to irrigate and what water application is appropriate.

▼

4          Drip irrigation

Targeted water application in the root zone or per section.

▼

5          Feedback

New measurements show how the soil and crop are responding and inform the next decision.

The strength lies in the integration

A sensor alone provides a measurement. A model alone provides a calculation. An irrigation system alone supplies water. By linking these three components, a practical decision-making chain is created in which data can directly contribute to crop management.

The ultimate aim is a closed-loop control system. Soil moisture sensors provide new measurements; QMS Water processes this information into a recommendation; the drip irrigation system carries out the watering; and the next measurement shows the effect of this. Reliable data and agronomic oversight remain essential.

Learning from practice

The field trials also highlight areas for further development. When sensor data is temporarily unavailable or the system has technical limitations, additional assessment in the field remains necessary. This is precisely why validation under real-world conditions is important: not only the model, but the entire chain must function reliably.

In the AgriDataValue pilot for onions, soil moisture, weather data and the amount of irrigation applied are monitored. This data is used to test the functioning of the data chain and to further validate the model.

AgriDataValue: building practical digitalisation

The AgriDataValue project focuses on connecting and utilising agricultural data. For Delphy, this means bringing together sensor technology, crop production knowledge and practical irrigation. The aim is not to collect more data, but to enable better decisions for growers and advisers.

Read More

From Sensor to Value: The Journey of Agricultural Data in AgriDataValue

AgriDataValue aims to be a game changer in the Smart Farming Digital transformation using an innovative Agri Environment Data Space. The project implements a specialized digital platform to upscale sensor data for EU-wide monitoring of production and environmental data, combining artificial intelligence frameworks and agricultural knowledge. The first step towards this goal is measuring relevant production characteristics and collecting data. As data collection forms the foundation of the entire process, a key challenge is the efficient acquisition and integration of different data streams. At ILVO’s dairy and pig farming pilots (19 and 22), data collection has evolved from a time-consuming manual task into an efficient and automated process. Before data can be shared, analysed, or used by AI models, it first needs to be collected, validated, and integrated. This is exactly where the ILVO pilot farms contribute to the AgriDataValue vision.

Although not always visible for the untrained eye, vast amounts of data are being collected at the ILVO experimental pig and dairy farms. On the dairy farm, each cow has her own story. From the birth of the calf onwards, each aspect of the cow’s life is documented throughout her life. Its unique identification, age, pedigree and genetic characteristics form the start of a productive life producing milk for human consumption. Cow 956, a 25-month old heifer, started off with creating new life by giving birth to a female calf on July 5th. Currently she’s producing 34 kg of milk with 4.03 % protein and 4.68 % fat, eating 28.5 kilograms of roughage feed and 7.3 kg of concentrates. Through activity and rumination monitoring, the cow’s health is continuously monitored, and every medical hurdle is being tracked on the way. As one may have noticed by now, the cow is potentially generating lots of data, presuming these aspects of her life can be measured and monitored.

An interplay of technological systems and software systems actually measures and provides the data on these characteristics, creating data flows through different streams to be connected. As demonstrated in the overview below, many sensor systems have been implemented on the farm to obtain this goal. Making use of as many commercially available solutions as possible, as a farmer would in practice, these systems are fragmented and not connected well, providing individual data streams with specific data characteristics. Therefore, to make these data easily useable and accessible for research and innovation, custom applications are necessary to correctly obtain data from each data collection instance, and merge them into useable records of cow data logged in a centralised database.

On the pig farm, several life stories are unfolding at the same time. Sow 34 has just given birth to 15 piglets yesterday after being inseminated only once, whereas sow 28 gave birth to 17 but only 16 survived the delivery. Her previous litter is doing well, but the recently weaned piglets from sow 21 are having some digestive issues, thus reducing their feed intake and daily growth. In the fattening pig barn, some of the older groups are growing well and nearing market readiness, supporting farm profitability through favourable carcass weight and dressing percentage outcomes. For now, however, they are still emitting certain amounts of ammonia into the air. Once again, data is coming from all sides at once, and dedicated procedures are necessary to bring everything together.

Therefore, ILVO has been working extensively on centralised databases, creating our own custom SmartFarm database. With dedicated SmartCow and SmartPig applications, data collection, merging, and centralisation is automated and facilitated, making data available for research and data sharing, enabling meta-analyses and the application of artificial intelligence and machine learning techniques. The road from data capture to data value starts on the farm. Behind every data-driven innovation lies a practical reality: data first needs to be measured, collected, and connected. With its custom SmartFarm infrastructure, ILVO is helping to make that process more efficient, reliable, and ready to support research, innovation and data sharing initiatives such as AgriDataValue.

Read More

AgriDataValue – CrackSense Data Stories 360° [ Vidcast ]

AgriDataValue and sister project CrackSense, are joining forces! In the third episode of CrackSense Data Stories 360°, Sandra Kolarić (CrackSense) talks with Nikos Arvanitis (AgriDataValue, Synelixis S.A.) about the intersection of artificial intelligence and agriculture. The discussion highlights how data-driven tools—ranging from smart irrigation to pest and disease detection—are driving smarter agricultural practices.

Watch the video here: https://www.youtube.com/watch?v=fr6kbsQODn8

Read More

Practical AI Meets Data Sovereignty: Inside AGRIDATAVALUE’s Integrated Agricultural Platform

European agriculture faces a dual challenge: boosting farm productivity while adhering to strict environmental targets. Modern farms generate vast streams of information from soil sensors, drones, and satellites. However, data often remains fragmented across closed vendor software due to privacy concerns and technical incompatibilities.

The Horizon Europe AGRIDATAVALUE project resolves this friction. The platform connects disparate agricultural systems into a unified, federated ecosystem while guaranteeing that farmers maintain full ownership of their data.

Bridging Complex Data and Real-World Farming

AGRIDATAVALUE integrates two complementary standards to handle information efficiently:

  • – Agriculture Information Model (AIM): Translates field-level details—such as parcel boundaries and crop health—into a common digital language
  • – International Data Spaces (IDS): Wraps this content in standardized usage rules, ensuring data is shared only under terms set by the farm owner

This architecture allows independent regional platform instances—operated by cooperatives, tech providers, or public agencies—to connect without storing all European farm data in a single central database.

Practical Tools for Daily Farm Decisions

At Month 36, the platform’s Decision Support System delivers Artificial Intelligence (AI) models across 20 European pilot use cases:

  • – Pest and Disease Warnings: Fungal risk predictors in vineyards and infestation models in olive groves allow farmers to apply treatments only when necessary, cutting pesticide costs.
  • – Targeted Weed Control: Drone cameras identify weed patches in row crops, guiding spot-spraying rather than blanket-treat fields.
Figure 1: Weed and crop detection results
Figure 1: Weed and crop detection results
  • – Livestock Welfare: Wearable equipment track cattle activity, flagging health anomalies and predicting calving times to ensure timely intervention.

Built for Trust and Security

Adopting AI requires clear, understandable results. AGRIDATAVALUE embeds trust directly into its technical design:

  • – Federated Learning (FDML): Instead of uploading raw field measurements to a cloud server, machine learning models train locally on edge devices. Only model updates are transmitted, preserving data confidentiality.
  • – Dynamic AI (DynAI): Connects localized, standalone models to the platform, allowing custom farm algorithms to run alongside federated networks.
  • – Explainable AI (XAI): Farmers do not have to follow unexplained advice blindly. The system generates simple summary cards explaining why a decision was made.
  • – Blockchain Traceability (CHAINTRACK): Registers supply chain milestones—from crop harvest to food processing—on a tamper-proof ledger, confirming product origin for consumers.
Figure 2: How AGRIDATAVALUE protects privacy—raw farm data remains on local devices, while only anonymized AI model updates are shared with the central platform.

By uniting edge analytics, data sovereignty, and human-interpretable AI, AGRIDATAVALUE provides European agriculture with a practical path toward sustainable, data-driven farming.

Read More

Why Data Sharing in Agriculture Needs a Common Structure, and What the ADV Data Model Contributes

Modern farms generate data from many independent sources: soil and weather sensors; farm management software; machinery; drone and satellite imagery; and veterinary or laboratory systems. Each of these systems tends to describe the same real-world thing, a parcel, a crop, an animal or an observation, in its own way, using its own field names, units and formats. When a cooperative, an advisory service or a public authority wants to combine information from several of these sources, for example to plan irrigation, forecast a pest outbreak or report on Common Agricultural Policy (CAP) eco-scheme compliance, that mismatch has to be resolved somewhere, usually through one-off, manual integration work. This is one of the main reasons agricultural data, despite being collected in growing volumes, is not used to its full potential: the effort needed to make it usable elsewhere often exceeds the effort needed to collect it in the first place.

Data spaces, not a single platform

AgriDataValue addresses this challenge by building what its most recent architecture update calls a “platform of platforms”: rather than moving all agricultural data into one central system, independent platforms, sensor networks and pilot environments stay where they are and connect through common interoperability mechanisms, an approach closer to a data space than to a conventional integration project. A data space, in the sense defined by International Data Spaces (IDS) approach, is not a single piece of software but an ecosystem in which participants exchange data under agreed rules while keeping control over their own resources. For that to work in practice, participants need to align on four layers: technical (including syntactic), so systems can exchange data at all; semantic, so the data carries the same meaning for everyone; organizational, so roles and processes are agreed; and legal, so usage terms are clear. Agriculture, with its mix of small and large actors, public and private data holders and cross-border pilots, is a demanding test of exactly that alignment, and much of it starts with something deceptively basic: a common way to describe what a dataset contains, who may use it and under what conditions.

The ADV Data Model: a DCAT Application Profile for agriculture

This is the gap the AgriDataValue (ADV) Data Model is built to close. The Data Catalog Vocabulary (DCAT) is a World Wide Web Consortium (W3C) standard for describing datasets so that they can be found and understood across different catalogs. Because DCAT is deliberately generic, communities that need more specific structure typically define a DCAT Application Profile (DCAT-AP) on top of it, in the same way the European Commission maintains a DCAT-AP for public sector data portals or a geospatial extension, GeoDCAT-AP, for spatial data. The ADV Data Model, sometimes called AgriDCAT-AP within the project, follows the same pattern for agriculture: a DCAT Application Profile for describing and sharing agricultural datasets through data spaces, currently at version 3.0.0.

Rather than inventing a new agricultural vocabulary, the model combines two layers. A governance wrapper, built on DCAT together with the Open Digital Rights Language (ODRL), covers what any data space needs regardless of domain: a description of the dataset for catalog discovery and a machine-readable usage policy attached to it. The domain content layer then describes the agricultural data itself using vocabularies the sector already maintains: the W3C’s SOSA vocabulary (Sensor, Observation, Sample and Actuator) for sensor and earth-observation measurements; the European Telecommunications Standards Institute’s (ETSI) Smart Applications REFerence ontology, extended for agriculture as SAREF4AGRI, for parcels, crops and livestock; concepts carried over from the FOODIE project (Farm-Oriented Open Data In Europe) for field interventions and alerts; and the Food and Agriculture Organization’s AGROVOC thesaurus for controlled terms. Seven ready-made profiles cover observations, weather data, soil analysis, parcel and crop information, interventions, animals and alerts, each with a Shapes Constraint Language (SHACL) definition that can validate a dataset before it is shared and a matching JavaScript Object Notation for Linked Data (JSON-LD) template. A set of ready-to-use ODRL policy templates, for example open access, attribution required, research-only or time-limited, means a data provider can attach a usage condition without drafting one from scratch. The resulting dataset descriptions and policies remain exchangeable through the Dataspace Protocol (DSP), the interoperable protocol for data space interactions, or through any other DCAT-compatible catalog.

What this contributes to interoperability

The model’s contribution can be set against the four layers of interoperability that data space standards use to describe what makes exchange work: technical (including syntactic), semantic, organizational and legal. Reusing SOSA, SAREF4AGRI, FOODIE and AGROVOC keeps the semantic layer aligned with the same vocabularies behind the Agriculture Information Model (AIM) from the DEMETER project, which AgriDataValue’s own reference architecture treats as the domain-semantics counterpart to the IDS Information Model. Expressing profiles as JSON-LD with SHACL validation gives the technical layer an automatic conformance check instead of a manual one. Documented bridges to AgGateway’s ADAPT Standard (Agricultural Data Application Programming Toolkit) for farm-equipment data and to the FIWARE platform’s context-data model mean the organizational layer does not require every pilot or platform to abandon tools already in use. The bundled ODRL policies give the legal layer a concrete, machine-readable way to express data sovereignty, the principle that a provider keeps control over how its data is used, rather than leaving it to a separate contract. The model itself is published under a Creative Commons Attribution 4.0 license (CC BY 4.0) and maintained openly on GitHub, so its profiles, validation rules and policy templates can be reused, adapted or extended outside AgriDataValue.

None of this makes agricultural data sharing effortless. A profile can only standardize how data is described and governed; it cannot replace the trust-building, pilot testing and stakeholder engagement that AgriDataValue’s wider work carries out across its pilot sites, 23 of them at present, spanning different sectors and countries. What a profile like this provides is more modest and, for that reason, more durable: a common, tested, openly licensed structure that other agricultural data-sharing efforts do not have to design from zero. IDSA, as a partner in the AgriDataValue consortium, sees the ADV Data Model as a concrete example of how a sector-specific DCAT Application Profile can be built consistent with data space principles, one that other domains facing similar fragmentation may find worth a closer look.

References

1.  I. Chrysakis, A. Turkyamali, K. Railis, Y. Oikonomidis and T. Zahariadis, “Integrating Agricultural Semantic Models with Data Spaces for Enhanced Interoperability,” 2025 6th International Conference in Electronic Engineering & Information Technology (EEITE), Chania, Greece, 2025, pp. 1-6, doi: 10.1109/EEITE65381.2025.11166007. https://ieeexplore.ieee.org/document/11166007

2. AgriDataValue Consortium. ADV Data Model (ADV-Data-Model repository), version 3.0.0. https://github.com/agridatavalue/ADV-Data-Model

3.  AgriDataValue. “From Semantic Interoperability to Operational Intelligence.” 15 July 2026. https://agridatavalue.eu/index.php/2026/07/15/from-semantic-interoperability-to-operational-intelligence/

4.  AgriDataValue. “Enhancing Agricultural Data Models and Semantic Interoperability.” 17 March 2025. https://agridatavalue.eu/index.php/2025/03/17/enhancing-agricultural-data-models-and-semantic-interoperability/

5.  World Wide Web Consortium (W3C). Data Catalog Vocabulary (DCAT), Version 3.

6.  European Commission, Interoperable Europe. DCAT Application Profile for Data Portals in Europe (DCAT-AP).

7.  International Data Spaces Association. International Data Spaces Reference Architecture Model (IDS-RAM).

Read More

ADV Pilot 5: Video Presentation  

We are thrilled to share a new video presenting AgriDataValue Pilot #5, taking place in Belgium! This innovative pilot features five plots integrated with cutting-edge smart farming technologies.

The Challenge

Traditional fertilization and pest control methods can easily lead to resource wastage and excessive chemical use if they are not carefully calibrated to the soil’s exact needs. Without real-time insights, farmers risk over-applying inputs, which impacts both profitability and the environment.

The Solution

To closely track soil dynamics, we have deployed multiple on-site sensors to monitor key soil characteristics. Additionally, we are using drone technology to map out weeds, uploading these aerial images directly to the AgriDataValue platform. By harnessing this data, the pilot delivers optimized, targeted irrigation and fertilization, alongside advanced prediction models for harvests and crop diseases. This results in a massive reduction in wasted water, fertilizers, pesticides, and energy. Ultimately, Pilot #5 aims to optimize crop production and to increase environmental sustainability.

Watch the full video insight here: https://www.youtube.com/watch?v=W0i5nbKDDhs

Read More