HAL: Hyper Article en Ligne
Not a member yet
3159010 research outputs found
Sort by
FlowEO: Generative Unsupervised Domain Adaptation for Earth Observation
No full textInternational audienceThe increasing availability of Earth observation data offers unprecedented opportunities for large-scale environmental monitoring and analysis. However, these datasets are inherently heterogeneous, stemming from diverse sensors, geographical regions, acquisition times, and atmospheric conditions. Distribution shifts between training and deployment domains severely limit the generalization of pretrained remote sensing models, making unsupervised domain adaptation (UDA) crucial for real-world applications. We introduce FlowEO, a novel framework that leverages generative models for image-space UDA in Earth observation. We leverage flow matching to learn a semantically preserving mapping that transports from the source to the target image distribution. This allows us to tackle challenging domain adaptation configurations for classification and semantic segmentation of Earth observation images. We conduct extensive experiments across four datasets covering adaptation scenarios such as SAR to optical translation and temporal and semantic shifts caused by natural disasters. Experimental results demonstrate that FlowEO outperforms existing image translation approaches for domain adaptation while achieving on-par or better perceptual image quality, highlighting the potential of flow-matching-based UDA for remote sensing
Simultaneous joint inversion of synthetic seismic and ground penetrating radar data with petrophysical variable change
No full textInternational audienceIn this work, we address the characterization of porosity and water saturation in a synthetic model of a shallow alluvial subsurface using frequency electromagnetic and seismic data. The inversion method employs a Gauss–Newton scheme, where the Jacobian of the merged seismic and electromagnetic data is formulated with respect to the spatially heterogeneous petrophysical parameters. This is made possible by introducing realistic petrophysical relationships, which significantly reduce the number of unknowns in the inverse problem and incorporate a strong prior correlation between the information contained in both data types regarding the subsurface composition. The results obtained show that this simultaneous joint petrophysical inversion produces reconstructions with clear improvements compared to independent petrophysical inversion. Indeed, it greatly enhances the spatial resolution of subsurface mapping, as well as the quantitative estimation of porosity and saturation
SMARTscape: An agent-based model to analyse rural landscapes multifunctionality
No full textInternational audienceLand-use changes and behavioural adaptation will be essential in global efforts to address environmental crises. For this, landscape multifunctionality will become critical to support a wide range of ecosystem services (ES), land productivity, and economic profitability. This study aims to explore pathways towards achieving sustainable food production while enabling multifunctionality. We developed SMARTscape, an empirically stylized agent-based model (ABM) developed to simulate interactions between land-use patterns, ES supply, and decision-making across a range of stylized landscapes inspired by New Zealand rural systems. SMARTscape combines three agent behavioural types—business-as-usual (BAU), profit-oriented (PO), and environment-oriented (EO)—with three decision-making rules of increasing social complexity. Agents interact with four collective entities (neighbours, peers, industry bodies, government), allowing simulation of bottom-up and top-down influence mechanisms. We designed a factorial experiment of 60 scenarios, from the combination of five landscape compositions, three decision rules, four agent-population behaviour mixes, and tested cluster size effect (on 10 cluster sizes from a very fragmented landscape to very aggregated) on 15 out of the 60 scenarios. Results show that land-use intensity and fragmentation are primary drivers of multifunctionality. However, profit-oriented behaviours can outperform environment-oriented strategies in intensive contexts, particularly through adoption of perennial systems, and behavioural diversity enhances trade-off management. Combined decision-making consistently led to higher multifunctionality, highlighting the importance of social influence. SMARTscape demonstrates the value of integrating social and spatial dynamics in a flexible modelling platform, offering both theoretical insights and a tool for exploring sustainable land-use transitions in data-poor but policy-relevant contexts
Charles III (1543-1608) et la régence de Christine de Danemark (1545-1552) puis de Nicolas de Lorraine (1545-1559)
No full text
RECHARGE, a model of potential recharge of aquifers applied to mainland France
No full textInternational audienceCalculating aquifer recharge provides a means of estimating the renewable fraction of groundwater resources, which is often difficult to quantify. This paper introduces the RECHARGE method, developed to calculate potential groundwater recharge from precipitation infiltration, and its application across mainland France over an extended historical period.The method relies on a simple soil water budget approach to estimate effective precipitation, using meteorological data and a spatial parameter that accounts for land cover and allows the seasonal variability of evapotranspiration to be reflected. An effective precipitation infiltration ratio (EPIR) is then derived for catchments with homogeneous geological lithology, based on linear regressions involving the baseflow index and a GIS-derived parameter. Given the low interannual variability of the baseflow index, the EPIR is assumed to remain constant over time and is subsequently used to convert effective precipitation into potential recharge at the scale of all groundwater bodies in mainland France.To validate this approach, annual effective precipitation estimates were compared for 556 selected catchments, both with observed annual river flows and with outputs from the physically based SURFEX model. The calculated potential recharge was also evaluated at both annual and seasonal scales for the entire French territory, using SURFEX as a reference. Results demonstrate that the RECHARGE model can effectively estimate annual and seasonal potential aquifer recharge. It is suitable for large-scale applications without requiring detailed knowledge of aquifer properties. Future improvements are envisioned, particularly to enhance monthlyscale accuracy in mountainous regions
Flash vacuum expansion technology for small-scale production of fruit puree: development and quality assessment
No full textInternational audienceFruit processing at small scale often lacks access to efficient, integrated technologies capable of ensuring product quality and safety. Flash Vacuum Expansion emerges as a promising alternative to conventional thermal treatments by coupling rapid steam heating with instant decompression, enabling simultaneous pasteurization and cooling, tissue disintegration, and deaeration. This study presents the design, construction, and validation of an optimized Flash Vacuum Expansion prototype tailored for small- and medium-scale agroindustries. The system integrates six unit operations—blanching, pasteurization, mashing, cooling, deaeration, and pulping—of which the last four are conducted under vacuum conditions. Heating is achieved through a screw conveyor blancher equipped with direct steam injection, while the vacuum operations are synchronized using double-valve airlocks. Performance was assessed using açai and Andean blackberry, evaluating microbial inactivation, physicochemical and rheological properties, and energy efficiency. FVE-treated purees showed complete microbial inactivation (<10 CFU/g), significant increases in pulp yield and extraction efficiency, reduced insoluble solids, enhanced pigment release, and superior rheological consistency. Energy consumption analysis revealed specific energy consumption (SEC) values as low as 0.97 MJ/kg—markedly lower than conventional processing benchmarks. The proposed FVE system represents a scalable and energy-efficient solution for producing high-quality fruit purees
Analytical solution and parametric design of bio-PCM-based passive BTMS for cylindrical lithium-ion cells under lumped model assumptions
No full textInternational audienceThe present study proposes a parametric investigation of a passive battery thermal management system (BTMS) utilizing bio-based phase change materials (bPCMs). The thermal network approach allocates to the cell and the bPCM two thermal nodes to accurately capture the surface and core temperatures of the cell as well as the two concentric layers of the bPCM. A novel analytical solution to the thermal network model is introduced for the first time in the context of cell-bPCM configuration. The resolution is implemented through the zero-order hold discretisation technique, which involves piecewise time integration. Validation against experimental and computational fluid dynamics data under variable load demonstrates the model's predictive capability. Notably, this work features an extensive parametric analysis, examining both bPCM and cell thermo-physical parameters, thereby providing new insight into their effects on thermal performance over consecutive charge-discharge cycles. Results indicate that a 6 mm bPCM layer thickness was identified as optimal, providing a balance between thermal performance and system compactness. A lower melting point within the operating range leads to earlier activation of latent thermal absorption. The heat of fusion showed diminishing benefits beyond 200 kJ/kg, while bPCM thermal conductivity mainly improved internal homogeneity rather than peak suppression. Variations in bPCM density showed negligible impact on peak cell temperature but influenced thermal storage capacity. Furthermore, the study encompasses the effects of battery format, where the cell radius was found to be inversely proportional to temperature spikes observed, while cells with higher heat capacity showed improved resilience to thermal spikes. Notably, increasing the ambient heat transfer coefficient from 5 to 100 W/m2.K significantly enhances heat dissipation to the environment and promotes thermal recovery of the bPCM between cycles, reducing peak cell temperatures by up to 3 °C. Additionally, analysis of the BTMS under realistic driving conditions (WLTC, JC08, CLTC, NEDC, UDDS) underscores the system's ability to maintain the cell operating temperature within its optimal range (<36.2 °C), with temperature differences below 6 °C across all driving scenarios examined. This work provides a scalable tool for BTMS design and sizing, facilitating the integration of sustainable solutions into electric vehicles
Convergence rate for the coupon collector's problem with Stein's method
No full textInternational audienceThe functional characterization of a measure, an essential but delicate aspect of Stein's method, is shown to be accessible for stable probability distributions on convex cones. This notion encompasses the usual stable distributions \textit{e.g.} Gaussian, Pareto, \textit{etc.} but also the max-stable distributions: Weibull, Gumbel and Fréchet. We use the definition of max-stability to define a Markov process whose invariant measure is the stable measure of interest. In this paper, we focus on the Gumbel distribution and show how this construction can be applied to estimate the rate of convergence in the classical coupon collector's problem