Portail "HAL-Francophonie Afrique et Océan Indien"
Not a member yet
43052 research outputs found
Sort by
Retrodictive Forecasting: A Proof-of-Concept for Exploiting Temporal Asymmetry in Time Series Prediction
No full textWe propose a retrodictive forecasting paradigm for time series: instead of predicting the future from the past, we identify the future that best explains the observed present via inverse MAP optimization over a Conditional Variational Autoencoder (CVAE) [1]. This conditioning is a statistical modeling choice for Bayesian inversion; it does not assert that future events cause past observations. The approach is theoretically grounded in an information-theoretic arrow-of-time measure related to stochastic thermodynamics. The symmetrized Kullback-Leibler divergence between forward and time-reversed trajectory ensembles provides both the conceptual rationale and an operational GO/NO-GO diagnostic for applicability. We implement the paradigm as MAP inference over an inverse CVAE with a learned RealNVP normalizing-flow prior and evaluate it on six time series cases: four synthetic processes with controlled temporal asymmetry and two ERA5 reanalysis datasets [2] (North Sea wind speed and solar irradiance). The work makes four contributions: (i) a formal retrodictive inference formulation; (ii) an inverse CVAE architecture with flow prior and Forward-Inverse Chaining warm-start; (iii) a model-free irreversibility diagnostic; and (iv) a falsifiable validation protocol with pre-specified predictions across controlled synthetic and real-world processes. All four pre-specified predictions are empirically supported: the diagnostic correctly classifies all six cases; the learned flow prior improves over an isotropic Gaussian baseline on GO cases; the inverse MAP yields no spurious advantage on time-reversible dynamics; and on irreversible GO cases, it achieves competitive or superior RMSE relative to forward baselines, with a statistically significant 17.7% reduction over a forward MLP on ERA5 solar irradiance (Diebold-Mariano test, p < 0.001). These results provide a structured proof-of-concept that retrodictive forecasting can constitute a viable alternative to conventional forward prediction when statistical time-irreversibility is present and exploitable
Energy import dependence in Madagascar: A decoupling analysis
No full textWe analyze Madagascar's structural dependence on energy imports using annual and quarterly data spanning 2007 to 2024. Our methodological framework integrates the Tapio decoupling index, Logarithmic Mean Divisia Index (LMDI) decomposition, Autoregressive Distributed Lag (ARDL) modeling, and Granger causality testing. The results show persistent expansive negative decoupling throughout the sample period, which demonstrates that economic expansion drives a disproportionate rise in energy import volumes relative to GDP growth. The LMDI decomposition reveals a systematic inverse symmetry between price and intensity effects. During periods of low international prices, the economy temporarily improves efficiency, yet these gains prove unsustainable. When prices surge, energy intensity simultaneously increases. This pattern reflects structural rigidity in Madagascar's energy system, as the economy fails to sustain efficiency gains during price shocks and instead reverts to more import-intensive production methods. The ARDL estimates confirm this dependency and yield long-run and short-run GDP elasticities of 1.043 and 1.256, respectively. Granger causality tests support the energy-led growth hypothesis and reveal unidirectional flows from energy prices and imports to economic growth in the long run. In the short run, positive shocks in energy prices cause positive changes in GDP, while negative shocks in energy prices cause negative changes in GDP. Short-run GDP shocks, however, cause fluctuations in energy imports of the opposite sign, providing further evidence of the decoupling status. These empirical findings demonstrate that Madagascar's economy remains locked into a high-dependency trajectory where growth is constrained by external energy market volatility
Analyser la vulnérabilité différentielle des territoires côtiers à l'échelle de l'île de La Réunion
No full textInternational audienceIle tropicale du sud-ouest de l’océan Indien, La Réunion est fortement exposée aux risques côtiers (érosion, submersion marine, inondations) en contexte de forçages paroxysmaux (cyclones, houles australes) dont l’intensité et la récurrence risquent d’être, selon les projections du GIEC, amplifiées par le changement climatique.Ces aléas interviennent dans un contexte de disparités territoriales, où les dynamiques socio-spatiales et les modes d’habiter différenciés participent à la production de vulnérabilités contrastées à l’échelle de l’île. Documenter et qualifier ces vulnérabilités dans une perspective systémique est l’objectif des Observatoires du littoral (Meur-Ferec et al., 2020). Cette recherche décline les indicateurs spatiaux de vulnérabilité systémique articulant composantes physiques, sociales et institutionnelles du risque, et les croise avec des indicateurs de vulnérabilité sociale contextualisés (Johnson et al., 2023) afin d’examiner comment les inégalités sociales et territoriales se traduisent spatialement face aux aléas naturels et interagissent avec les politiques publiques d’aménagement du territoire, notamment urbanisme et prévention des risques.Elle s’inscrit dans trois sites situés dans des secteurs fortement impactés par les risques côtiers et compris dans des périmètres Politiques de La Ville identifiés comme vulnérables à partir d’indicateurs socio-économiques (Cutter et al., 2003). Les entretiens auprès des gestionnaires et spécialistes des risques côtiers constituent le point de départ de l’analyse : identification d’enjeux, perceptions et pratiques de gestion.Enfin, la démarche cartographique mobilise le système d’information géographique (SIG) et les données maillées de l’observatoire du littoral, ainsi que des données socio-économiques de l’INSEE et des données qualitatives issues d’entretiens. Elle cherche à éclairer les logiques de production d’une vulnérabilité différenciée (Blaikie et al., 2004) sur les littoraux réunionnais et à questionner la cohérence des politiques publiques à l’échelle locale. Les cartes produites constituent un appui pour confronter ces discours aux données disponibles et mettre en évidence les contrastes spatiaux et des dynamiques différenciées
Archeointensity Database and Geomagnetic Field Reference Curves for South America Over the Past 5 Millennia
No full textInternational audienceThe study of variations in geomagnetic field intensity over time and space is crucial to understand the evolution of the geodynamo and its interactions with the Earth's surface. In this work, we introduce the SAGEOMAG (South America GEOMAGnetic) database, a comprehensive and updated repository of geomagnetic field intensity records derived from archeological artifacts and volcanic materials over the past 5 millennia and adherent to the FAIR (Findable, Accessible, Interoperable and Reusable) principles. A key advantage of the SAGEOMAG database is the inclusion of statistical parameters at the specimen level for data quality control, as well as the data categorization according to the hierarchical level of record (site, fragment or specimen). Additionally, we propose new intensity references curves for South America (SARIC), modeled with a bootstrap algorithm that minimizes the influence of outliers and estimates the probability density function without needing explicit prior shape considerations. For records without intensity errors and age uncertainty details in the original studies, standard errors and uncertainties were assigned based on log‐normal distributions. Our findings demonstrate that the modeling of master curves is highly sensitive to the quality filters applied to the data and the hierarchical level of the records. This study aims to improve the organization of geomagnetic data for South America and contribute to regional and global archeomagnetic field modeling, thereby enhancing our understanding of the geomagnetic field evolution and especially the South Atlantic Anomaly (SAA)
Evaluating the Effects of Transdisciplinary Research in the Context of Marine and Coastal Social-Ecological Systems. Preliminary insights from a systematic literature review
No full textInternational audienceDespite growing recognition of evaluation as a vital driver of reflection, learning, and unlearning in transdisciplinary research (TdR) and science–policy interfaces (SPI), how, with whom, and when such evaluation should be carried out remains a matter of considerable debate. Against this background, our work seeks to clarify both the potential benefits and the persistent challenges of evaluating TdR and SPI, with a focus on marine and coastal contexts. We screened an initial corpus of 3,722 peer-reviewed papers published globally. From this sample we identified 16 studies that explicitly applied and explained evaluation approaches to TdR or SPI in marine or coastal settings. These studies serve as the basis of our conference presentation. Across this sample, existing evaluations tend to emphasise short-term outcomes rather than longer-term impacts. Although qualitative methods have generated valuable contextual insights, there remains a need for more systematic and mixed-method approaches to distinguish project-specific effects from external influences. Participatory evaluation, despite its strong conceptual alignment with TdR and SPI, remains underutilised, even though it could enhance legitimacy, collaboration, and transformative potential. While evaluations frequently demonstrate outcomes such as knowledge exchange and network-building, these rarely translate into sustained collaboration, collective action, or tangible environmental change. Moreover, few studies critically assess the evaluation frameworks themselves or capture the long-term transformative impacts of projects. Our methodological approach, though limited in scope at this stage of the project, provided important insights into diverse evaluation practices and how research evaluation can contribute to transformative change. These initial findings form the foundation for a more comprehensive and comparative analysis across a more diverse set of social-ecological systems. Building on our results, we also aim to develop a tailored evaluation framework for the BRIDGES research programme – a 10-year inter- and transdisciplinary initiative in the southwestern Indian Ocean on fisheries transformations and ocean conservation
Linking energy service access and human capabilities to assess energy justice in the rural Sahel
No full textInternational audienceThe Sahel covers more than 3 M km² across eleven countries and accounts for only one-tenth of the world’s per capita energy consumption, yet it remains largely unexplored in the energy justice literature. Energy justice offers a suitable analytical framework for studying disparities in energy access and assessing corresponding public policies. However, apprehending the real benefits of energy related infrastructures (e.g., electric grid, roads, telecommunications network) requires to go beyond access to new energy carriers. We thus propose to enrich the energy justice framework by considering (1) access to energy services and (2) the capability-based theory of human well-being. Drawing on field observations and interviews in rural Senegal, we show that (1) benefits from new infrastructures still remain unevenly distributed: though opening up rural areas and providing local populations with new energy services, specific communities (e.g., semi-nomadic, low-income) are often left aside; (2) social tensions may arise locally due to resource management, as new or more productive energy services become available. We conclude that energy policies should not only focus on infrastructures (regional energy access), but also consider the corresponding end-use equipment (local energy service access), the heterogeneity of populations’ practices and lifestyles, and the potential side effects of energy policies on other sectors
Next-Gen IoT localization: When quantum-SSA-Markov hybridization meets energy efficiency for robust, accurate, and sustainable positioning in smart environments
No full textInternational audienceThe Internet of Things (IoT) is transforming communication and data exchange across sectors such as healthcare and transportation, where efficient and precise localization and high-resolution sensing are critical requirements for applications ranging from asset tracking and autonomous navigation. Existing localization techniques often suffer from accuracy limitations, excessive energy consumption, and latency issues due to the dynamic and resource-constrained environments in which IoT devices operate. These limitations are exacerbated by the dense deployment of IoT devices, multipath signal propagation, and environmental interference, which can compromise the reliability of location-based services. This paper presents a novel hybrid model that integrates quantum mechanics for state representation, Markov processes for predictive modeling, and the Salp Swarm Algorithm (SSA) for optimizing nodal positions and speeds. Our Quantum-SSA-Markov model significantly enhances localization accuracy and efficiency in IoTbased wireless sensor networks (WSNs), outperforming existing techniques in energy consumption, latency, signal-to-noise ratio (SNR), and data flow rate. Experimental results confirm that this innovative approach provides a scalable and reliable solution for next-generation IoT systems, particularly in scenarios that require robust real-time localization
Ashes to art: The glassmaking traditions of Königsfelden’s medieval windows
No full textInternational audienceThe chemical composition and optical properties of the stained-glass windows in the Königsfelden Abbey church (Aargau, Switzerland), which are among the oldest and best-preserved examples from the 14th century in the region, were analysed on site using portable X-ray fluorescence (pXRF) and optical absorption spectroscopy. A subset of loose fragments was further characterised by Raman spectroscopy and LA-ICP-MS, and these data were used to improve the calibration of the results obtained with in situ pXRF. A machine-learning classifier (GLORIA) was applied to assess the variability of the glasses and their provenance relative to a reference database of European stained-glass windows. The results demonstrate that the Königsfelden corpus consists of potash-based forest glasses, mostly attributed to Central Europe (59%), with additional inputs from the Rhine region (32%) and Northwestern France (9%). While the colouring mechanisms for cobalt blue, copper red, manganese purple, silver yellow and iron sulphide amber are relatively well understood, the colouring techniques of a distinctive bluish grey and rare Pb-rich green glass ( ∼10 wt% PbO) are unusual and have no clear published parallels. One hypothesis for the high PbO contents in green glass is that a CuO-PbO frit was deliberately employed to facilitate the dissolution of metallic copper in the silicate melt. The blue-grey glass, unique in hue and composition, appears to result from the combination of a manganese-cobalt glass with an additional, yet unidentified, contribution. Our study thus provides new insights into medieval glass recipes, trade networks, and workshop practices, while also validating the potential of calibrated pXRF combined with reference standards and machine learning for heritage science applications
Effets de la Corruption sur la qualité de l'environnement : cas des pays en développement
No full textThis thesis analyses the effects of corruption on environmental quality in developing countries, using a panel of 70 developing countries covering the period 2000–2022. It begins with a regional assessment of corruption and environmental quality in Sub-Saharan Africa, the Middle East and North Africa, Latin America and the Caribbean, as well as Asia. This assessment highlights that corruption constitutes a major risk to the environment and that these countries are particularly vulnerable to the effects of climate change and biodiversity loss.The study then distinguishes between the direct and indirect effects of corruption on environmental quality. To identify these effects, it employs robust econometric methods, including system GMM and structural equation modelling (SEM) mediation techniques. The empirical results indicate that corruption does not have a significant direct effect on the environmental indicators considered. However, it acts indirectly through several channels, notably economic growth, international trade, urbanisation, demographic dynamics, dependence on fossil fuels, agriculture, and the level of democracy.Furthermore, the estimates do not support the environmental Kuznets curve hypothesis for the developing countries in the sample.Overall, corruption primarily influences environmental quality through indirect channels. In terms of public policy, these results underscore the need to strengthen anti-corruption institutions, improve transparency, accelerate the energy transition, and consolidate environmental governance to mitigate its harmful effects.Cette thèse analyse les effets de la corruption sur la qualité de l’environnement dansles pays en développement, à partir d’un panel de 70 pays en développement couvrant la période 2000-2022.Elle commence par dresser un état tat des lieux régional de la corruption et de la qualité de l'environnement en Afrique subsaharienne, dans la région Moyen-Orient et Afrique du Nord,en Amérique latine et dans les Caraïbes, ainsi qu’en Asie. Ce diagnostic met en évidence que la corruption constitue un risque majeur pour l’environnement et que ces pays sont particulièrement exposés aux effets du changement climatique et à la perte de leur biodiversité. L’´etude distingue ensuite les effets directs et indirects de la corruption sur la qualitéde l’environnement. Pour identifier ces effets, elle mobilise des méthodes économétriquesrobustes, notamment le GMM en système et par la médiation par équations structurelles(SEM). Les résultats empiriques indiquent que la corruption n’exerce pas d’effet direct significatif sur les indicateurs environnementaux considérés. En revanche, elle agit demanière indirecte à travers plusieurs mécanismes, notamment la croissance économique,le commerce international, l’urbanisation, la dynamique démographique et la dépendance aux énergies fossiles, l’agriculture, le niveau de démocratie.Par ailleurs, les estimations ne confirment pas l’hypothèse de la courbe environnementalede Kuznets pour les pays en développement de l’´echantillon.En définitive, la corruption influence principalement la qualité de l’environnement par des canaux indirects. Sur le plan des politiques publiques, ces résultats soulignent la nécessité de renforcer les institutions de lutte contre la corruption, d’améliorer la transparence,d’accélérer la transition énergétique et de consolider la gouvernance environnementaleafin de limiter ses effets délétères