The DIVINE results that go beyond the current state of the art work lie in the following areas:
- Delivery of the DIVINE Reference Architecture that provides a reliable architectural framework for agricultural data spaces, addressing matters of data-space federation, interoperability, related policies, etc.
- Delivery, deployment and evaluation of the DIVINE agricultural data space ecosystem (ADSE) with facilities ensuring increased transparency in data sharing, data trust & sovereignty, data traceability & usage monitoring
- Delivery and usage of AIM+, a semantic agri-data model that supports semantic interoperability, while it integrates policies and regulations that concern data sharing and governance. The model and its components are involved and contribute to ongoing standardization efforts.
- Achievement of enhanced semantic interoperability and greater transparency on data sharing & usage
- Delivery of framework supporting the performance evaluation of the DIVINE facilities and the data sharing costs, benefits, risks, and added values using KPIs in four areas: (i) agronomic performance; (ii) economic performance, (iii) environmental sustainability and (iv) societal impact.
- A self-service dashboard framework (catering web and mobile devices) has been developed, capable of integrating a number of service front-ends, transformation operators and visualisation widgets, implementing proven AI-based decision making approaches providing actionable advices while users will be able to provide their feedback on the DIVINE services' performance.
- Delivery of innovative (AI-based) models involved in various agricultural operations such as the following: NDVI Super-resolution Model (Significantly improves the resolution of NDVI images, enabling more precise identification of terrain features and vegetation cover); Soil Organic Carbon Estimation Model (Estimates soil organic carbon levels using satellite imagery, a crucial indicator of soil fertility and water retention capacity); Vegetation Classification Model (Differentiates trees from other vegetation and assesses their health status, critical information for forest management and environmental conservation); NDVI Forecasting Model (Predicts NDVI values for grassland five days in advance, allowing for proactive irrigation scheduling, water management and grass clearing activities);
- Support for advanced Decision Making (via a system that integrates all data sources and model outputs to provide real-time decision support for various agricultural practices, namely grass weeding, irrigation, and soil health management), etc. In this respect, improvement in efficient farming practices towards sustainability is delivered; reduction of raw materials such as fertilisers and greenhouse gas emissions, especially carbon; delivery of near real time calculation of CAP relevant indicators based on agricultural practices recorded in the farm's calendar on a farm level and on a group of farms level; optimisation of resource usage and farming practices; etc.
- Delivery of stakeholder-validated policy framework for agri-data sharing and governance. Input collected from external stakeholders confirmed the relevance and applicability of DIVINE's policy-related findings, with the regulatory analysis and governance outputs receiving positive feedback from practitioners and policy-oriented actors.
- Contribution to future CAP discussions. Stakeholder interactions revealed convergent expectations around data governance and agricultural policy, positioning DIVINE as a potential reference framework and working example for future Common Agricultural Policy (CAP) discussions at EU level.
- Application for three patents within DIVINE: (1) Seed Selection Support for Intelligent Next Sow Decisioning (UCD/SETU) [This patent is focused on selecting seeds that best address the local conditions and historical performance of a specific farm, based on a specialised seed data repository.], (2) Yield Predictor Tuning for Seed Selection Support (UCD/SETU) [This patent describes a management system for crop planting to support highly localised yield predictions.] and (3) A Computer Implemented Method for Generating Cloud-Free Satellite Imagery and Extracting Data Therefrom, Data Processing System and Computer Program (VICOM) [This patent overcomes remote sensing surveillance challenges. Specifically, it enables the generation of cloud-free satellite imagery and the subsequent extraction of high-quality data.]
