Éditeur(s) : HAL CCSD
Résumé : poster
International audience
Locations for earthquake recorded in the Lesser Antilles subduction zone are processed separately by regional observatories, NEIC and ISC. There is no earthquake location catalog available compiling all available phase arrival data. We propose a new best complete earthquake catalog by merging all available phase arrival data for better constrains on earthquake locations. ISC provides the phase arrival data of 29243 earthquakes (magnitude range from 1.4 to 6.4) recorded by PRSN (Porto Rico), SRC (British West Indies), and from FUNVISIS (Venezuela). We add phases data from IPGP observatories for 68718 earthquakes from magnitudes 0.1 to 7.5 (OVSG, Guadeloupe, recorded 53226 earthquakes since 1981, and OVSM, Martinique, recorded 29931 earthquakes since 1972). IPGP also provides the accelerometer waveform data of the GIS-RAP network. We achieved automatic picking on the GIS-RAP data using the Component Energy Correlation Method. The CECM provides high precision phase detection, a realistic estimation of picking error and realistic weights that can be used with manual pick weights. The CECM add an average of 3 P-waves and 2 S-waves arrivals to 3846 earthquakes recorded by the GIS-RAP network since 2002. The final catalog contains 84979 earthquakes between 1972 and 2013, 24528 of which we compiled additional data. We achieve earthquake location using NonLinLoc, regional P and S waves data and a set of one dimensional velocity models. We produce improved locations for 22974 earthquakes (better residuals, on equal or larger arrival dataset) and improved duration magnitudes for 6258 earthquakes (using duration data and improved locations). A subset of best constrained 15626 hypocenters (with more than 8 phases and an average RMS of 0.48_0.77s) could be used for structural analysis and earthquake local tomography. Relative locations are to be applied in order to image active faulting. We aim to understand coupling in the seismogenic zone as well as triggering mechanisms of intermediate depth seismicity like fluid migration beneath the volcanic arc.
Geophysical Research Abstracts
Vienne, Austria

hal-01059207
https://hal-brgm.archives-ouvertes.fr/hal-01059207

Éditeur(s) : HAL CCSD
Résumé : The Lesser Antilles is an area of high volcanic and earthquake activity, characterized by a 1000 km convergence zone resulting from the Atlantic plate subduction under the Caribbean plate with a slow convergent motion (2 cm/year). With five years of available data, the CDSA data base presents a more homogeneous vision of Lesser Antilles arc seismicity and allows detecting low seismic activity zones, in the north near the Virgin Islands, and in the south between St-Lucia and Grenada. The accuracy improvements of location is used to better study the relationship between tectonic structures and seismicity, to better define the subduction slab geometry, and to analyse the spatial variations of the Gutenberg-Richter law b-value. Concerning the 21 november 2004 Les Saintes earthquake, EMS98 intensities in Les Saintes islands were evaluated from an individual request and hypocenter relocalisations of the Les Saintes earthquake and the biggest aftershock of 14 february 2005 were made using a method of master/slave in order to contrain the seismic source of the mainchock. As seismic hazard assessment depends from strong ground motion generated by earthquakes, the last part of this work concerns modeling accelerometric records of the Les Saintes earthquakes, combining a composite source model with an empirical Green Function technique. In our study, we used Les Saintes aftershocks as Empirical Green Function.
L'arc des Petites Antilles situé sur la bordure nord-est de la plaque Caraïbe, résulte d'un phénomène de subduction, les plaques américaines plongeant sous celle de la Caraïbe avec une vitesse de 2 cm/an. Les Antilles Françaises représentent la région française où le risque sismique est le plus important. Ce travail dont l'objectif est d'améliorer la connaissance de l'aléa sismique dans les Antilles Françaises, repose sur les données du Réseau Accélérométrique Permanent (RAP) et sur celles du « Centre de Données des Antilles Françaises ». L'analyse de ces données permet de mieux contraindre la sismicité de l'arc. Deux zones de très faible sismicité sont mises en évidence : au nord près des Iles Vierges et au sud entre Sainte-Lucie et Grenade. D'autres points sont aussi résolus : imagerie de la subduction le long de l'arc, relations entre la sismicité superficielle et la tectonique active, variations de la pente (b-value) de la loi de Gutenberg-Richter. Le séisme majeur du 21 novembre 2004 Mw=6.3 est au cœur des deux derniers chapitres. Plusieurs aspects y sont présentés : macrosismicité avec l'évaluation d'intensités EMS98 obtenues du dépouillement de formulaires individuels recueillis pour les Iles des Saintes, et relocalisation par la méthode de maître/esclave du choc principal et de sa plus forte réplique du 14 février 2005, de façon à mieux contraindre l'imagerie de la source du choc principal, étape primordiale pour la dernière partie portant sur la modélisation des signaux du choc principal des Saintes. Nous tentons dans cette dernière partie de modéliser les enregistrements de mouvements forts avec une approche semi-empirique en se basant sur un modèle stochastique large-bande et proposons l'utilisation de plusieurs Fonctions de Green Empiriques (FGE) sélectionnées parmi les répliques du séisme des Saintes.
https://tel.archives-ouvertes.fr/tel-00409021

tel-00409021
https://tel.archives-ouvertes.fr/tel-00409021
https://tel.archives-ouvertes.fr/tel-00409021/document
https://tel.archives-ouvertes.fr/tel-00409021/file/these_BengoubouMendy_2008.pdf

Éditeur(s) : HAL CCSD Oxford University Press (OUP)
Résumé : International audience
Energy dissipation is observed on seismic data when a wave propagates through a porous medium, involving different frequency regimes depending on the nature of rock and fluid types. We focus here on the role of partial fluid saturation in unconsolidated porous media, looking in particular at P-wave phase velocity and attenuation. The study consists in running an experiment in a sand-filled tank partially saturated with water. Seismic propagation in the tank is generated in the kHz range by hitting a steel ball on a granite plate. Seismic data are recorded by buried accelerometers and injecting or extracting water controls the partial saturation. Several imbibition/drainage cycles were performed between the water and gas residual saturations. A Continuous Wavelet Transform applied on seismic records allowed us to extract the direct P-wave at each receiver.We observe an hysteresis in phase velocities and inverse quality factors between imbibition and drainage. Phase velocities and inverse quality factors are then jointly inverted to get a final poro-visco-elastic model of the partially saturated sand that satisfactorily reproduces the data. The model formulation consists in generalizing the Biot theory to effective properties of the fluid and medium (permeability and bulk modulus) in order to properly explain the phase velocity variation as a function of the saturation. The strong level of attenuation measured experimentally is further explained by an anelastic effect due to grain to grain sliding, adding to Biot's losses. This study shows that fluid distribution at microscopic scale has strong influence on the attenuation of direct P-waves at macroscopic scale and confirms that seismic prospection may be a powerful tool for the characterization of transport phenomena in porous media.
ISSN: 0956-540X

hal-00779920
https://hal.archives-ouvertes.fr/hal-00779920
https://hal.archives-ouvertes.fr/hal-00779920/document
https://hal.archives-ouvertes.fr/hal-00779920/file/Barriere_et_al_HAL.pdf
DOI : 10.1111/j.1365-246X.2012.05691.x

Éditeur(s) : HAL CCSD
Résumé : International audience
We present a novel approach towards signature recognition from spatio-temporal data. The data is obtained by recording gyroscope and accelerometer measurements from an embedded pen device. The idea of Universum learning was previously presented by Vapnik and recently popularized in machine learning community. It assumes that the decision boundary of a classifier lies close to data with high uncertainty. The quality of the final classifier strongly depends on a way how to choose the Universum data and also on the representation of original data. In our paper we use a novel approach of Universum learning to classify signature data, also we present our novel idea how to sample the Universum data. At last, we also find more effective representation of the signature data itself compared to the baseline method. These three novelties allow us to outperform previously published results by 4.89% / 5.58%.
17th Biennial Conference of the International Graphonomics Society
Pointe-à-Pitre, Guadeloupe

hal-01165925
https://hal.univ-antilles.fr/hal-01165925
https://hal.univ-antilles.fr/hal-01165925/document
https://hal.univ-antilles.fr/hal-01165925/file/IGS_2015_submission_38.pdf