HAL Portal AU (University of Avignon)
Not a member yet
35713 research outputs found
Sort by
Le Service National d'Observation Gravimétrie SNOG
No full textInternational audienceLe Service National d’Observation Gravimétrie (SNOG) assure le suivi des réseaux d’observations et contribue aux services de l’Association internationale de géodésie (AIG). Il assure ainsi le suivi du réseau permanent gravimétrique (RPG) et des réseaux répétés gravimétriques (RRG). Le RPG est en particulier constitué de cinq stations multi-instrumentées, dont quatre colocalisant des mesures absolues avec des mesures de variations temporelles de la gravité à l’aide de gravimètres supraconducteurs permanents. Le SNOG contribue également au bureau central de deux services scientifiques internationaux de l’IAG: le Bureau Gravimétrique International (BGI) et le Service international de géodynamique et des marées terrestres (IGETS). Les données que le SNO récolte et distribue sont utilisées par différentes communautés pour observer les variations de masses à toutes échelles (surface, croûte, manteau...) ainsi que pour vérifier la stabilité verticale des références (marégraphes, etc.). En constante évolution pour répondre aux besoins de ses communautés scientifiques et les nouveaux déploiements technologiques, nous présentons les différentes composantes du SNOG, les missions d’observations qu’il pilote et les perspectives de développement, en particulier pour ce qui concerne la gravimétrie spatial
Data-driven dimensional analysis applied to single-phase natural circulation
No full textInternational audienc
FUTURISKS à Mayotte : comprendre les risques côtiers pour mieux s’y préparer: Anticiper l’impact des vagues et du climat sur la barrière de récif de Mayotte
No full textPoster aux journées portes ouvertes de l'université de MayotteFUTURISKS en bref : Un programme national sur l’ensemble des territoires d’outremer français (Antilles, Mayotte, la Réunion, la Nouvelle Calédonie, la Polynésie française).Observer & modéliser : mesures terrain + modélisations numériquesComprendre les risques, anticiper la submersion et l’érosionUn projet collectif : Chercheurs + acteurs locaux = stratégies d’adaptation durables pour les îles tropicales
Decrypting the Breeding Biology of the Elusive and Declining Tahiti Petrel <i>Pseudobulweria rostrata</i>
No full textInternational audienceAmong procellariids, the Tahiti Petrel Pseudobulweria rostrata is one of the most endangered and least known species. Its global populations are declining, yet demographic and ecological studies remain scarce. Understanding the breeding cycle and behavior of Tahiti Petrel adults and chicks is essential to develop effective protection measures. To address this knowledge gap, Tahiti Petrel colonies of various sizes were studied at three sites in New Caledonia, with extensive sampling at two main sites and limited sampling at a third. A total of 157 burrows were monitored for up to two years using endoscopic cameras and camera traps. This allowed the depiction of the breeding phenology, reproductive success, frequency of adult nest visits, chick behavior, first emergence, and fledging dates. During the study, 75% of the identified burrows were visited by Tahiti Petrels. Egg-laying peaked in December but occurred year-round, indicating aseasonal breeding by Tahiti Petrels in New Caledonia. The average breeding cycle was 329 ± 11.6 days, including an average incubation period of 55.7 ± 0.9 days and an average chick-rearing period of 110.7 ± 5.6 days. Parents visited nests every 1.3 days on average during chick-rearing. After the chick’s first emergence, which typically occurred 31 days before fledging, adult visitation decreased. Chicks did not show defensive behavior against predators, and most chicks fledged after nine days without feeding. Breeding success was 50% at a predator-free site and 32% at a site with invasive predators. These findings suggest high sensitivity to disturbance and depredation, contributing to the species’ decline in New Caledonia and elsewhere in the world. Together with previous studies conducted in New Caledonia, these results provide crucial information for the implementation of adapted conservation measures for this declining species
Marine spatial planning initiatives off the coast of large coastal cities. An assessment in the Northwestern Mediterranean
No full textInternational audienceMarine Spatial Planning (MSP) is fundamental to the management of marine resources and space. Many countries have drawn up national marine spatial plans that, as with land spatial planning, need to be transposed and adapted at local level, particularly in areas of significant socio-economic interest. Major coastal cities and adjacent marine areas are critical places for local MSP. However, there has been little investigation of MSP implementation in these priority contexts. To narrow this gap, this study analyses how the local authorities of seven major coastal cities of the Northwestern Mediterranean, located in France, Italy and Spain, are including the sea in their planning strategies, and what they are doing in terms of MSP. Examining their public policies and spatial planning strategies sheds light on whether and how they are transposing MSP, and with what objectives. Results show that local implementation of MSP in these cities is highly heterogeneous, influenced both by national policies and local initiatives, and still rare. Among the seven cities, only Barcelona and Marseille stand out for their initiatives to plan and manage their marine area. This calls into question the impact of the European and national maritime policies enacted over the past decades to promote MSP
Matérialités médiévales en Auvergne : appréhender le quotidien à travers les sources de la pratique. Un dossier en cours
No full textInternational audienc
Reconnaissance automatique d’écriture et sources historiques: Limites et nouvelles perspectives
No full textInternational audienceThe aim of this article is not to present unpublished research, but to review the work we have been carrying out since 2019 at the École nationale des chartes, then at INRIA Paris and the CNRS, on Automatic Text Recognition (ATR), and more particularly on its application to medieval manuscripts. This article provides an overview of our research, together with references for exploring in detail specific aspects of our past and current work
Exploring Organic-mineral Evolution in Planetary Analogs: Insights from Hydrothermal Alteration of Aromatic Molecules and Mafic Minerals
No full textInternational audienceAbstract Organic-mineral interactions are crucial drivers of diversity in abiotic systems on planetary surfaces. Despite their significance, the evolution of these systems, particularly the role of organic molecules in newly formed minerals, remains underexplored. In this study, we proposed new experiments on analogs to explore the interactions between organic molecules likely present in planetary environments and primary minerals. We examine interactions between aromatic benzyl-group compounds and igneous minerals (olivine, feldspar) under mid-temperature aqueous conditions, analogous to hydrothermal systems hypothesized to exist on early Martian and terrestrial environments. Using multiscale techniques—gas chromatography, infrared spectroscopy, mass spectrometry, X-ray diffraction, microscopy, and adsorption studies—we analyzed the analogs after 45 days at 100°C. The results reveal that the presence of minerals influences the distribution of newly formed organic compounds, for example, by promoting the formation of organic chelates. Minerals altered in the presence of organics formed secondary phases, such as phyllosilicates and amorphous materials resembling serpentines or smectites. Organic compounds impacted dissolution rates, secondary mineral parageneses, and porosity, enriching the diversity of hydrated mineral phases compared to mineral alteration without organic matter. Significant carbon and nitrogen were found filling mineral porosity (up to 7 wt% C), modifying physical properties compared to systems without organics. These findings highlight the pivotal role of organics in shaping mineralogy on planetary surfaces and underscore the need for broader studies of organic-mineral analogs to improve interpretations of in situ and remote organic detections in extraterrestrial samples
Special issue: Biodiversity and ecosystem services in European vineyards: Innovative approaches for a sustainable future
No full textInternational audienc
Analyse de graphes complexes pour détecter la corruption dans les marchés publics
No full textPublic procurement plays an essential role in the functioning of institutions, representing around 15 % of global GDP. In theory, procurement procedures are designed to ensure transparency, competition, and efficiency. In practice, they are often complex, hard to interpret, and exposed to risks such as collusion, favoritism, or corruption. In this context, the use of large volumes of available data opens new ways to detect fraud, as a complement to traditional methods, including econometric approaches.In this perspective, the DeCoMaP project (Detection of Corruption in Public Procurement), funded by the French National Research Agency, aimed to develop detection tools combining legal, economic, and computational expertise, using real data from French public contracts. Conducted as part of DeCoMaP, this thesis addresses two major methodological challenges: the low reliability of existing databases, and the lack of modeling of the relationships between economic actors. To tackle these issues, we use graph-based models to represent the interactions between buyers and suppliers in public procurement.We start by building two original datasets, FOPPA and BeauAMP, based on official publications related to French public contracts, both at national and European levels. This work involves extensive data processing, including entity disambiguation. The resulting datasets enable reliable large-scale relational analyses and surpass existing sources in terms of completeness, consistency, and ease of use. They provide a solid foundation for studying public procurement in France and can benefit both researchers and public decision-makers.From these graphs, we aim to distinguish usual procurement networks from those showing atypical or potentially suspicious patterns. To do this, we extract discriminative patterns: subgraphs that appear more frequently in one class than in the other. A major challenge lies in selecting the most relevant patterns for this task. We conduct a systematic study of 38 quality measures from the literature, comparing their behavior and performance. Our results show that some measures are unstable, while others, simple but robust, perform well. We also apply clustering to the patterns to reduce redundancy and better organize the pattern space. These results form a useful benchmark to guide the choice of quality measures in future work.Based on these findings, we develop the PANG framework (Pattern-based Anomaly detection in Graphs), which includes all steps of the process: pattern extraction, selection, representation, and graph classification. We evaluate it both on standard public datasets for graph classification and on the FOPPA dataset. Results show that PANG achieves performance that is comparable to, or better than, existing methods. By relying on interpretable patterns, it helps give meaning to the detected configurations and supports both economic and institutional interpretation.This work contributes to better fraud detection and aims to strengthen transparency and integrity in public procurement procedures.Les marchés publics jouent un rôle essentiel dans le fonctionnement des institutions, représentant environ 15 % du PIB mondial. En théorie, les procédures sont conçues pour garantir transparence, concurrence et efficacité. En pratique, elles sont souvent complexes, peu lisibles, et exposées à des risques comme la collusion, le favoritisme ou la corruption. Dans ce contexte, l'exploitation des grands volumes de données disponibles permet d'envisager de nouvelles manières de détecter les fraudes, en complément des méthodes classiques, notamment économétriques.Dans cette perspective, le projet DeCoMaP (Détection de la Corruption dans les Marchés Publics), financé par l'Agence nationale de la recherche avait pour but de concevoir des outils de détection combinant expertise juridique, économique et informatique, à partir de données issues des marchés publics français. Menée dans le cadre du projet DeCoMaP, cette thèse cible deux verrous méthodologiques importants : la faible fiabilité des bases de données existantes, et le manque de prise en compte des relations entre les acteurs économiques. Pour y répondre, nous adoptons une modélisation en graphes, afin de mieux représenter les interactions entre acheteurs et fournisseurs dans les marchés publics.Nous commençons par construire deux bases de données originales, FOPPA et BeauAMP, en nous appuyant sur les publications officielles relatives aux marchés publics français, diffusées à l'échelle nationale et européenne. Ce travail repose sur un traitement approfondi des données, comprenant notamment la désambiguïsation des entités. Les bases ainsi obtenues permettent des analyses relationnelles fiables à grande échelle, tout en surpassant les sources existantes en qualité et en facilité d'exploitation. Elles constituent une base solide pour l'étude des marchés publics en France, et peuvent bénéficier aussi bien aux chercheurs qu'aux décideurs publics.À partir de ces graphes, nous cherchons à distinguer les réseaux de marchés habituels de ceux qui présentent des configurations atypiques, voire suspectes. Pour cela, nous extrayons des motifs discriminants : des sous-graphes apparaissant préférentiellement dans une classe plutôt que dans l'autre. L'un des défis majeurs réside dans la sélection des motifs les plus pertinents pour cette tâche. Pour cela, nous conduisons une étude systématique de 38 mesures de qualité issues de la littérature, en comparant leur comportement et leur performance. Nos résultats montrent que certaines mesures sont instables, tandis que d'autres, simples mais robustes, offrent de bonnes performances. Nous introduisons également un clustering des motifs pour limiter la redondance et structurer plus efficacement l'espace des patterns. L'ensemble de ces résultats constitue un benchmark utile pour guider pour guider le choix de mesures de qualité dans de futurs travaux.En nous appuyant sur les résultats précédents, nous concevons le framework PANG (Pattern-based Anomaly detection in Graphs), qui regroupe toutes les étapes du processus : extraction, sélection et représentation des motifs, puis classification des graphes. Nous l'évaluons à la fois sur des jeux de données publics standards en classification de graphes, ainsi que sur la base FOPPA. Les résultats montrent que PANG atteint des performances comparables, voire meilleures, que celles des méthodes existantes. Grâce à des motifs faciles à interpréter, il permet de donner du sens aux configurations détectées, et d'en proposer une lecture à la fois économique et institutionnelle.Ce travail contribue à une meilleure détection des irrégularités, avec l'objectif de renforcer la transparence et l'intégrité des procédures d'attribution dans les marchés publics