HAL-Ecole des Ponts ParisTech
Not a member yet
42743 research outputs found
Sort by
Graded index Silicon Germanium photonics circuits in the mid-infrared
No full textInternational audienceA photonics platform based on graded SiGe waveguides has been developed for operating in the mid-IR range, with experimental demonstrations up to 11 µm wavelength. Applications are related with absorption spectroscopy and sensing in the fingerprint region, and with free space communications. After an overview of the platform, the presentation will focus on : (i) integrated high speed modulators operating in a wide spectral range, based on free carrier plasma dispersion effect, (ii) integrated resonators operating around 8 µm wavelength with quality factors beyond 10 5 . We will also show that mid-IR graded SiGe photonics circuits can be fabricated on industrial-scale 200 mm wafers, and that low propagation losses can be obtained in a wide spectral range of the mid-IR. These results thus pave the way for a scalable silicon compatible mid-infrared platform
Reconstructed global monthly burned area maps from 1901 to 2020
No full textInternational audienceFire is a key Earth system process, driving variability in the global carbon cycle through CO2 emissions into the atmosphere and subsequent CO2 uptake through vegetation recovery after fires. Global spatiotemporally consistent datasets on burned area have been available since the beginning of the satellite era in the 1980s, but they are sparse prior to that date. In this study, we reconstructed global monthly burned area at a resolution of 0.5° × 0.5° from 1901 to 2020 using machine learning models trained on satellite-based observations of burned area between 2003 and 2020, with the goal of reconstructing long-term burned area information to constrain historical fire simulations. We first conducted a classification model to separate grid cells with extreme (burned area ≥ the 90th percentile in a given region) or regular fires. We then trained separate regression models for grid cells with extreme or regular fires. Both the classification and regression models were trained on a satellite-based burned area product (FireCCI51), using explanatory variables related to climate, vegetation and human activities. The trained models can well reproduce the long-term spatial patterns (slopes = 0.70–1.28 and R2 = 0.69–0.98 spatially), inter-annual variability and seasonality of the satellite-based burned area observations. After applying the trained model to the historical period, the predicted annual global total burned area ranges from 3.46×106 to 4.58×106 km2 yr−1 over 1901–2020 with regular and extreme fires accounting for 1.36×106–1.74×106 and 2.00×106–3.03×106 km2 yr−1, respectively. Our models estimate a global decrease in burned area during 1901–1978 (slope = -0.009×106 km2 yr−2), followed by an increase during 1978–2008 (slope = 0.020×106 km2 yr−2), and then a stronger decline in 2008–2020 (slope = -0.049×106 km2 yr−2). Africa was the continent with the largest burned area globally during 1901–2020, and its trends also dominated the global trends. We validated our predictions against charcoal records, and our product exhibits a high overall accuracy in simulating fire occurrence (>80 %) in boreal North America, southern Europe, South America, Africa and southeast Australia, but the overall accuracy is relatively lower in northern Europe and Asia (<50 %). In addition, we compared our burned area data with multiple independent regional burned area maps in Canada, the USA, Brazil, Chile and Europe, and found general consistency in the spatial patterns (linear regression slopes ranging 0.84–1.38 spatially) and the inter-annual variability. The global monthly 0.5° × 0.5° burned area fraction maps for 1901–2020 presented by this study can be downloaded for free from https://doi.org/10.5281/zenodo.14191467 (Guo and Li, 2024)
Monitoring the Effects of Alkalinity on the Self-sealing of the Opalinus Clay from the Lower Sandy Facies (LSF) of Mont Terri Site Through Hydraulic Conductivity Test and μ-CT Observation
No full textInternational audienceClaystones are considered as potential host rocks for deep geological waste repositories. During excavation, a Damaged Zone () is created. This is expected to be sealed with underground water seepage. During the operational phase, the degradation of concrete structures will act as a source of alkaline plumes increasing the of the ground water, further affecting the sealing of the fractures within the . This study focuses on the effect of alkalinity on the self-sealing of the Opalinus Clay from the lower sandy facies () of Mont Terri site, considered as potential host rock in Switzerland. Both hydraulic conductivity and X-ray computed microtomography (μ-CT) tests were performed. Series of initially unsaturated and artificially fractured Opalinus Clay samples were exposed to the synthetic water or the concrete pore water. It was found that the hydraulic conductivity of the initially unsaturated samples exposed to the synthetic water varies non-monotonically with time. While at first the hydraulic conductivity increased suggesting the generation of secondary fractures with hydration, a decrease was then observed underlining the self-sealing. After days of exposure, the hydraulic conductivity was one order of magnitude lower than the one at days and two orders of magnitude greater than the one for the undisturbed material. Based on µ-CT observations, the self-sealing of the Opalinus Clay after days of exposure to the synthetic water and to the alkaline solution was of and , respectively, suggesting that the self-sealing of the investigated claystone occurs even under high
On the formulation and implementation of mixed mode I and mode II extrinsic cohesive zone models with contact and friction
No full textAn extrinsic cohesive zone model for mixed mode I and mode II fracture that encompasses contact and Coulomb friction is developed in the framework of nonsmooth mechanics. The model is extended to include the effects of dynamics with impact and sliding, and is discretised in time so that it can be written as a linear complementarity problem (LCP). The LCP is proved to have a solution, subject to a condition on the size of the time-step. Finally, we study the behaviour of the LCP system numerically, by observing the response of a simple test geometry to rapid loading, and observe the numerical method reproduces analytically predicted and experimentally observed behaviours, without requiring impracticably small time-steps
Utilisation de Wikidata et Wikisource pour référencer l’œuvre poétique de Marceline Desbordes-Valmore
No full textNational audienceDepuis janvier 2023, la Société des études Marceline Desbordes-Valmore propose sur son site web une base de données dédiée à l'œuvre poétique de cette figure majeure du mouvement romantique. La construction de cette base de données, qui fournit le texte des poèmes tout en référençant les manuscrits, éditions, partitions de mises en musique de ces poèmes, leurs traductions ou encore des enregistrements de versions lues ou chantées, s'est appuyée sur plusieurs communs numériques du mouvement Wikimedia. Plusieurs recueils de poèmes ont été relus par la communauté Wikisource francophone, sous l'impulsion d'un projet communautaire, le Défi 5000, et d'ateliers de relecture proposés par l'association Le deuxième texte. Un travail spécifique d'ajout sur Wikidata de fiches sur les compositrices de mises en musique de poèmes de Desbordes-Valmore a été mené, notamment en croisant la base de données collaborative avec la base de données de compositrices "Demandez à Clara". Enfin, la liste des poèmes établie par Marc Bertrand dans ses éditions, en 1973 puis en 2007, de l'œuvre poétique de Marceline Desbordes-Valmore, a été utilisée comme base de travail pour la création systématique, dans Wikidata, d'un élément de type œuvre pour chaque poème, auxquelles sont progressivement rattachées les diverses éditions des poèmes relus sur Wikisource. Ceci permettra notamment de disposer d'identifiants pérennes pour les poèmes lors de la publication de données de la recherche liées à cette œuvre littéraire, mais aussi d'associer de riches métadonnées aux poèmes, par exemple pour identifier les dédicataires ou les personnes dépeintes par les poèmes
Universal Multifracals characterization of Intensity-Duration-Frequency curves in Northern Italy
No full textInternational audienceQuantifying rainfall extremes and their evolution is important in hydrologic risk analysis and design. It usually relies on the use of intensity-duration-frequency (IDF) curves. Numerous approaches have been developed over the years to estimate them empirically from point measurements. Here we suggest to implement and further develop a framework initially introduced by Bendjoudi et al. (1997), in which IDF curves are theoretically derived from a Universal Multifractals (UM) framework. It is a mathematically robust framework which relies on the physically based notion of scale invariance, inherited from the governing Navier-Stokes equations. It is parsimonious since it relies only on three parameters with physical interpretation. UM have been extensively used to characterize and simulate geophysical fields extremely variable over wide range of space-time scales such as rainfall. The method is here applied to 5-min resolution data from rain gauges located in a complex topography area in North-eastern Italy, with a temporal coverage of 25-35 years. For each station, UM parameters are estimated through an analysis carried out on the whole available continuous time series, which enables to ultimately derive IDF curves. Comparison with extreme value approaches as well as spatial variations of the obtained results over the studied area will be discussed. References:Bendjoudi H., Hubert P., Schertzer D., Lovejoy S., 1997, Interprétation multifractale des courbes intensité-durée-fréquence des précipitations, Comptes Rendus de l'Académie des Sciences - Series IIA - Earth and Planetary Science, 325, 5, 323-326,https://doi.org/10.1016/S1251-8050(97)81379-
Blunt Extension and Dynamic Generation of Multifractal Cascade Fields Tree for Rainfall Drop Trajectories Simulations
No full textInternational audiencePrecipitation variability at small space-time scales significantly influences hydrological processes, particularly in heterogeneous environments such as urban areas. Building on established methodologies for generating universal multifractal cascade fields, we propose an alternative approach that optimizes memory efficiency while maintaining the fidelity and flexibility of high-resolution simulations. Our method generates cascade fields dynamically, we call it Cascade Tree, which reduces memory usage by over 100 times compared to precomputing and storing full datasets. This improvement complements existing techniques by offering a scalable option for real-time applications. To further enhance the realism of the simulated fields, we integrate the blunt extension of universal multifractals, which smooths transitions between far branches in Cascade Tree and addresses non-conservativeness in a computationally efficient manner. By leveraging GPU acceleration, we achieve rapid computation of cascade fields, enabling their use in simulating complex phenomena such as rainfall dynamics in turbulent wind fields. The method is applied to simulate 3D trajectories and velocities of raindrops in a high-resolution multifractal turbulent wind field, using real wind field data to improve the applicability of the results. Our simulations capture the spatial and temporal variability of rainfall and demonstrate the dispersion of over 100,000 raindrops across scales relevant to radar pixels and urban catchment hydrology. This work provides new tools for exploring rainfall-driven processes, with applications ranging from downscaling radar precipitation data to refining hydrological response models. By complementing established methods with a memory-efficient and GPU-accelerated framework, our approach bridges the gap between drop-scale dynamics and catchment-scale impacts
Low-Complexity Approach to Intelligent SHM by Combining Machine Learning Models Using Single-Sensor Data
No full textInternational audienceStructural Health Monitoring (SHM) systems, when applied to large civil engineering structures such as bridges, process high-volume data and run computationally intensive algorithms, which typically require important processing power to ensure low inference latency to enable real-time monitoring and rapid decision-making. In this study, we propose a novel resource-efficient approach to optimizing SHM for civil structures. The methodology integrates lightweight machine learning models that rely exclusively on single-sensor data, enabling deployment at the sensor level (smart sensors). By aggregating outputs from multiple sensors, the approach captures spatial information, introduces diversity, and benefits from an averaging effect, significantly improving overall performance compared to individual sensor-based predictions. This single-sensor strategy ensures low computational complexity while maintaining high accuracy, making it particularly suitable for resource-constrained environments. To evaluate the effectiveness of the proposed methodology, we applied it to the Z24 benchmark dataset, a widely recognized SHM resource for civil structures. The objective was to classify various damage scenarios based on data collected from accelerometers deployed on the bridge. The results demonstrate competitive performance with minimal computational complexity, highlighting the scalability and suitability of such an approach for large-scale SHM applications. Ultimately, this study underscores the potential of resource-efficient SHM solutions, contributing to developing sustainable and intelligent monitoring systems.</div
Segmenting France Across Four Centuries
No full textInternational audienceHistorical maps offer an invaluable perspective into territory evolution across past centuries, long before satellite or remote sensing technologies existed. Deep learning methods have shown promising results in segmenting historical maps, but publicly available datasets typically focus on a single map type or period, require extensive and costly annotations, and are not suited for nationwide, long-term analyses. In this paper, we introduce a new dataset of historical maps tailored for analyzing large-scale, long-term land use and land cover evolution with limited annotations. Spanning metropolitan France (548,305 km^2), our dataset contains three map collections from the 18th, 19th, and 20th centuries. We provide both comprehensive modern labels and 22,878 km^2 of manually annotated historical labels for the 18th and 19th century maps. Our dataset illustrates the complexity of the segmentation task, featuring stylistic inconsistencies, interpretive ambiguities, and significant landscape changes (e.g., marshlands disappearing in favor of forests). We assess the difficulty of these challenges by benchmarking three approaches: a fully-supervised model trained with historical labels, and two weakly-supervised models that rely only on modern annotations. The latter either use the modern labels directly or first perform image-to-image translation to address the stylistic gap between historical and contemporary maps. Finally, we discuss how these methods can support long-term environment monitoring, offering insights into centuries of landscape transformation. Our repository is publicly available on GitHub
Reallocating Wasted Votes in Proportional Parliamentary Elections with Thresholds
No full textInternational audienceIn many proportional parliamentary elections, electoral thresholds (typically 3-5%) are used to promote stability and governability by preventing the election of parties with very small representation. However, these thresholds often result in a significant number of "wasted votes" cast for parties that fail to meet the threshold, which reduces representativeness. One proposal is to allow voters to specify replacement votes, by either indicating a second choice party or by ranking a subset of the parties, but there are several ways of deciding on the scores of the parties (and thus the composition of the parliament) given those votes. We introduce a formal model of party voting with thresholds, and compare a variety of party selection rules axiomatically, and experimentally using a dataset we collected during the 2024 European election in France. We identify three particularly attractive rules, called Direct Winners Only (DO), Single Transferable Vote (STV) and Greedy Plurality (GP)