38796 research outputs found
Sort by
Strengthening EU Innovation Policy
Get PDFThis brief presents the results from the first edition of the Regional Innovation Output Indicator (RIOI) developed by the JRC to support more informed innovation policymaking at the regional level for the EU. Combining seven indicators, the RIOI provides a snapshot of innovation output across EU regions. According to the 2025 edition, innovation output is unevenly distributed across EU regions. It is predominantly concentrated in Central and Northern European countries, with regions in Sweden, Germany, Denmark, and Finland consistently outperforming the EU27 benchmark. In contrast, most regions in Southern and Eastern European member states score below the EU27 average. The RIOI ranking is led by Stockholm (SE11), Oberbayern (DE21), Berlin (DE30), Praha (CZ01), and Ile-de-France (FR10). In contrast, the three lowest-performing regions are all located in Romania. In addition, capital city regions consistently rank among the top-performing regions within their respective country in terms of innovation output. This pattern is particularly pronounced in Southern and Eastern European member states, where capital regions systematically outperform other regions in their countries, underscoring the over-concentration of innovation activity in urban centres.JRC.S.3 - Science for Modelling, Monitoring and Evaluatio
Clean Energy Technology Observatory: Geothermal Energy in the European Union - 2025 Status Report on Technology Development, Trends, Value Chains and Markets
Get PDFGeothermal energy, a high-capacity factor renewable, remains underutilized in the EU, contributing 0.2% of electricity and 0.7% of heat production. Despite its potential for energy transition, EU policy support—via the Net Zero Industry Act, Horizon Europe, and the Innovation Fund—is critical for scaling. While deep geothermal technology is commercially viable, global R&D leadership lies with the U.S. and China, with the EU lagging in high-value patents. The EU maintains a robust value chain in resource development and drilling but there is a lack of systematic data on component manufacturing capacities. Skills shortages persist, though oil and gas expertise can partially address workforce gaps. Political calls for a dedicated EU geothermal action plan, highlights growing recognition of its strategic role.JRC.C.2 - Energy Efficiency and Renewable
Li-ion battery electrolyte vaporisation model: An experimental and computational fluid dynamics approach
No full textElectrolyte leakage from a battery cell can lead to the formation of a dangerous atmosphere in poorly ventilated and enclosed environments due to the high volatility, flammability and toxicity of common electrolyte solvents like dimethyl carbonate (DMC) and diethyl carbonate (DEC). To limit human exposure, the design of detection and warning systems is necessary. To this end, a computational fluid dynamics approach was used to create a vaporisation model to predict the evaporation rate of DMC and DEC, supported by experimental data of the evaporating solvents. All experiments were performed in an enclosed environment with no induced flow. The evaporation flux of DMC and DEC in a fully convective and diffusive environment was 0.70 and 0.15 mg/(cm2 min) at room temperature, respectively. An LES-CFD vaporisation model with a novel liquid-vapour modelling approach was used for the simulation of the evaporated liquids, demonstrating good agreement with the experimental data. The presented model iterated on previous efforts of simulating the evaporation of liquids by accounting for the changes of the evaporation rate brought on by temperature and vapour accumulation around the source, which was done by coupling evaporation, temperature and dispersion modelling dynamically. Results show that DMC was significantly affected by the changes in liquid temperature during the evaporation, highlighting the importance on modelling the local energy balance of the evaporating solvent. The dispersion model showcased good qualitative agreement with the behaviour of heavy gasses, underlining that most of the dispersion of the vapours generated by DMC and DEC is buoyancy-based.JRC.C.1 - Battery and Hydrogen Technologie
EUSO/ESDAC Newsletter No 182 - November 2025
No full textThe November edition of the EU Soil Observatory (EUSO) newsletter is now available, providing the latest highlights and insights on soil-related research and policy developments within the European Union.JRC.D.1 - Forests and Bio-Econom
On Guardrail Models Robustness to Mutations and Adversarial Attacks
No full textThe risk of generative AI systems providing unsafe information has raised significant concerns, emphasizing the need for safety guardrails. To mitigate this risk, guardrail models are increasingly used to detect unsafe content in human-AI interactions, complementing the safety alignment of Large Language Models. Despite recent efforts to evaluate those models’ effectiveness, their robustness to input mutations and adversarial attacks remains largely unexplored. In this paper, we present a comprehensive evaluation of 15 state-of-the-art guardrail models, assessing their robustness to: a) input mutations, such as typos, keywords camouflage, ciphers, and veiled expressions, and b) adversarial attacks designed to bypass models’ safety alignment. Those attacks exploit LLMs capabilities like instruction-following, role-playing, personification, reasoning, and coding, or introduce adversarial tokens to induce model misbehavior. Our results reveal that most guardrail models can be evaded with simple input mutations and are vulnerable to adversarial attacks. For instance, a single adversarial token can deceive them 44.5% of the time on average. The limitations of the current generation of guardrail models highlight the need for more robust safety guardrails.JRC.T.2 - Cybersecurity and Digital Technologie
On the application of Supervised Time Series Forest to Radio Frequency Fingerprinting
No full textRadio Frequency fingerprinting (RFF) is a method to authenticate wireless devices using their intrinsic physical features. It has been investigated by the research community in recent years, especially in combination with deep learning (DL) algorithms, which has demonstrated an excellent classification performance with the disadvantage of a high computing cost. In many studies where the original time domain representation of the signal is used, RFF can be considered a time series classification (TSC) problem and the various algorithms defined in the literature can be used to implement RFF. This possibility has been scarcely investigated in RFF literature where DL is usually preferred. A possible reason is also related to the elevated computing complexity of many TSC algorithms, which can be comparable to DL algorithms. In recent years, a new set of
computationally efficient TSC algorithms has been proposed in literature, which can be suitable for a potential application in RFF. This paper investigates two aspects: 1) the application of the recently introduced Supervised Time Series Forest (STSF) algorithm to RFF and its comparison to CNN and other classifiers, and 2) the evaluation of STSF when combined in a hybrid approach of CNN with STSF (CNN-STSF), where the activation weights calculated by the CNN are used in input to the STSF in a similar way to hybrid CNN-ML algorithms presented in research literature. The proposed approach is applied to two public data sets showing that STSF has a very competitive performance in terms of execution time and classification accuracy.JRC.E.2 - Space, Connectivity and Economic Securit
The fully-automatic Sentinel-1 Global Flood Monitoring service: Scientific challenges and future directions
No full textOne of the critical factors in operational satellite-based flood monitoring efforts is the time it takes from the acquisition of the satellite image to the delivery of the flood maps to users. Any human involvement, such as coordinating satellite acquisitions or manually interpreting images, can delay this process. To avoid such delays, a fundamentally new approach was adopted for the Sentinel-1 based Global Flood Monitoring (GFM) service: All Synthetic Aperture Radar (SAR) images acquired by the Sentinel-1 satellites in VV polarisation over land are processed entirely automatically, enabling flood maps to be delivered within eight hours of acquisition. The flood maps, along with a novel flood likelihood layer, are generated using ensemble approaches that integrate three complementary flood mapping algorithms along with reference water maps to distinguish flooded areas from permanent and seasonal water bodies. A notable feature of the service is its capability not only to depict flood-pixels evident in the Sentinel-1 images but also to provide contextual information that identifies areas where flood mapping is not feasible or
problematic due to land cover and environmental conditions. These advancements were made possible through the use of a global 20 m backscatter datacube, which has enabled the characterisation of the backscatter behaviour for approximately 379 billion land surface pixels and deriving the reference water maps and a global flood archive. The GFM service was launched in 2021 as a new component of the Copernicus Emergency Management Service (CEMS) and has quickly garnered attention from users worldwide. In this review, we offer the first comprehensive overview of the scientific accomplishments and challenges faced during the first three years of operations. This analysis discloses discrepancies between the current service capabilities and the requirements of operational users, and provides directions for future research and service improvements, anticipating the increasing availability of systematic SAR data coverage from ROSE-L and other future SAR missions.JRC.E.1 - Disaster Risk Managemen
Assessment of Ni and Mn effect on the irradiation hardening behavior of VVER-1000 model steels exposed to high fluences in the high flux reactor
No full textIn the present work, we aim at providing more data and insight related to the influence of Ni and Mn contents on the degree of irradiation hardening of Light Water Reactor RPV steels. A total of 20 model steels and realistic welds based on VVER-1000 and PWR RPVs compositions were irradiated at high flux and to high fluences in the LYRA-10 experiment, conducted in the High Flux Reactor, Petten. Among them, eight VVER-1000 model steels with 0.1 wt % Cu and systematically varied Mn and Ni contents were submitted to tensile and Vickers hardness testing for evaluation of their degree of hardening, and were characterized in detail, using Atom Probe Tomography, Transmission Electron Microscopy, Small Angle Neutron Scattering and Positron Annihilation Spectroscopy. The mechanical testing results show the clear increase in degree of irradiation hardening with the Mn and Ni contents, in particular for steels containing 1.4 wt % Mn. Microstructural observations show direct
correlation between increase in yield strength and the formation of Mn-Ni-Si solute clusters. Calculations done using classic and multiscale models confirm that the solute clusters are the main hardening features present in the irradiated RPV model steels. Furthermore, TEM and PAS results suggest that dislocation loops have a more significant role on the formation of solute clusters than on irradiation hardening of the group of materials investigated.JRC.G.I.4 - Reactor Safety and Component
An Updated Simplified Energy Yield Model for Recent Photovoltaic Module Technologies
No full textThe European Commission's Photovoltaic Geographic Information System (PVGIS) uses a simplified solar energy yield model to provide quick and reliable data on the potential performance of photovoltaic (PV) systems. This study looks at the recalibration of the model for modern module technologies, using power matrix datasets produced by the European Solar test Installation (ESTI) for seven crystalline silicon (cSi), two cadmium telluride (CdTe) and three copper indium diselenide (CIS) modules. The results show that the PVGIS power performance model with updated coefficients can provide a good description of the power output of the modern crystalline silicon (cSi) modules, with a mean absolute bias error (MABE) of less than 1% in almost all cases, against an MABE of over 3.5% with the current coefficients. The updated coefficients allow the model to better capture the improved temperature coefficients and low light performance. As a result, there will be a slight increase in the energy yield estimates. For the thin film technologies, the updated coefficients allow for a more accurate description of current data sets, but more data for modules from recent production series would be desirable to further increase the model's applicability.JRC.C.2 - Energy Efficiency and Renewable
Monitoring the FAIRness of geospatial data: Lessons learnt from the European Union
No full textThe Findable, Accessible, Interoperable and Reusable (FAIR) principles were introduced to mitigate challenges in discovering, accessing and ultimately reusing data. They still represent the backbone of current, public sector-driven geospatial data infrastructures worldwide, and Key Performance Indicators (KPIs) are used to measure the progress towards their implementation. This work reflects on the experience of the European Union (EU) geospatial data infrastructure, driven by the INSPIRE and the Open Data Directive requirements. Analysing the results of the monitoring process in the last six years, we draw a number of lessons. First and foremost, the way in which KPIs are defined steers the development of an infrastructure against specific directions, and maximising the KPIs used to measure the FAIRness is not enough. A shift would be needed to more user-centric monitoring approaches, which originate from user needs and assess the actual value generated from data reuse. The analysis also demonstrated the importance of employing automated, transparent and reproducible monitoring processes powered by open source tools, as well as the need to define an inclusive governance approach grounded on a continuous involvement, dialogue and trust with the affected stakeholders.JRC.T.4 - Data Governance and Service