Éditeur(s) : HAL CCSD Mineralogical Society of America
Résumé : International audience
Accessory phases are important hosts of trace elements; allanite may contain >90% of the REE in a bulk rock. The mobility and redistribution of several trace elements, notably HREE, Th, U, and Y is thus controlled by reactions involving allanite and other REE phases, as well as several rock-forming minerals. As these elements are commonly concentrated in mature elastic sediments, a suite of impure quartzite was studied. Two eclogite facies samples from the Monometamorphic Cover Complex of the Sesia Zone (Western Italian Alps) are presented in some detail, as they reveal a remarkably rich spectrum of reaction relationships that involve REE phases.;Two allanite-forming reactions were inferred from textures and phase compositions;(1) monazite + Ca-silicate(?) + fluid -> allanite + apatite + thorite;;(2) monazite + thorite + Ca-silicate(?) + fluid -> Th-rich allanite + auerlite +/- apatite.;Petrographic observations and thermodynamic models suggest that allanite entered the HP assemblage at similar to 530 degrees C and 17-18 kbar during prograde metamorphism.;In one sample, allanite is rimmed by epidote rich in Y and HREE that grew at the expense of xenotime. Two net transfer reactions were derived;(3) xenotime + allanite + fluid -> Y-rich epidote + apatite + thorite;;(4) xenotime + allanite + fluid -> Y-rich epidote + aeschynite + thorite + (phosphate?).;Textural relationships and trace element analyses of coexisting allanite/monazite and xenotime/Y-rich epidote reveal systematic partitioning of the REE. Partition coefficients for the HREE are compatible with equilibrium fractionation, whereas those for the LREE show patterns that seem to be inherited from the precursor phases, in this case zircon with variable LREE composition.
ISSN: 0003-004X

hal-00745674
https://hal.archives-ouvertes.fr/hal-00745674
DOI : 10.2138/am.2012.3832

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
This study addresses the mechanical and chemical behavior of allanite during shear zone formation under high-pressure metamorphism. Understanding physico-chemical processes related to the retention or resetting of Pb isotopes in allanite during geological processes is essential for robust petrochronology. Dating of allanite in meta-granodiorite showing variable amounts of strain (from an undeformed protolith to mylonite) at Monte Mucrone (Sesia Zone, NW Italy) gave surprising results. Based on structural and petrographic observations the shear zones at Mt Mucrone are Alpine, yet allanite located within an eclogite facies mylonite yielded Permian ages (208Pb/232Th average age: 287 ± 7 Ma). These mm-sized allanite grains are rimmed by an aggregate of coarse-grained garnet + phengite, thought to derive from former epidote. These aggregates were immersed in a weak matrix that experienced granular flow, and they were thus chemically and mechanically shielded during Alpine mylonitization. In undeformed samples (8a and 8b), two populations of epidote group minerals were found. Allanite forms either coronas around Permian monazite or individual grains with patchy zoning. Both types yield Permian ages (208Pb/232Th age: 291 ± 5 Ma). On the other hand, grains of REE-rich clinozoisite of Cretaceous age are found in undeformed rocks. These grains appear as small fragments with embayed surface outlines and minute satellites or rims around Permian allanite. These (re)crystallized grains are Sr-rich and show mosaic zoning. These results indicate that allanite crystals retained their chemical and isotopic characteristics, and thus their Permian age, as a result of strong strain partitioning between the epidote group porphyroclasts and the eclogite facies matrix in HP-mylonites. The observed partial mobilization of Pb isotopes, which lead to the Cretaceous-aged rims or grains in undeformed samples was facilitated by (re)crystallization of allanite and not by mere Pb diffusion alone under the HP conditions.
ISSN: 0024-4937

insu-00679730
https://hal-insu.archives-ouvertes.fr/insu-00679730
DOI : 10.1016/j.lithos.2011.01.005

Éditeur(s) : HAL CCSD American Geophysical Union (AGU)
Résumé : International audience
Detailed seafloor mapping in the area east of Taiwan revealed trench-parallel stretching and folding of the Ryukyu forearc and lateral motion of the accretionary wedge under oblique convergence. East of 122ø40'E, a steep accretionary wedge is elongated in an E-W direction. A major transcurrent right-lateral strike-slip fault accommodates the strain partitioning caused by an oblique convergence of 40 ø. A spectacular out-of-sequence thrust may be related to the subduction of a structural high lying in the axis of the N-S trending Gagua Ridge. This asperity is likely responsible for the uplift of the accretionary wedge and forearc basement and may have augmented strain partitioning by increasing the coupling between the two plates. West of 122ø40'E, the low-taper accretionary wedge is sheared in a direction subparallel to the convergence vector with respect to the Ryukyu Arc. The bayonet shape of the southern Ryukyu Arc slope partly results from the recent (re)opening of the southern Okinawa Trough at a rate of about 2 to 4 cm/yr. Right-lateral shearing of the sedimentary forearc with respect to the nonlinear Ryukyu backstop generates trench-parallel extension in the forearc sediment sequence at dilational jogs and trench-parallel folding at compressive jogs. The Hoping Basin lies above a diffuse trench/trench/fault (TTF) or TFF unstable triple junction moving toward the south along a N-S transform zone which accommodates the southward drift of the Ryukyu Arc with respect to Eurasia.
ISSN: 0278-7407

hal-01261555
https://hal.archives-ouvertes.fr/hal-01261555
https://hal.archives-ouvertes.fr/hal-01261555/document
https://hal.archives-ouvertes.fr/hal-01261555/file/Lallemand_et_al-1999-Tectonics.pdf
DOI : 10.1029/1998TC900011

Éditeur(s) : HAL CCSD
Résumé : In SW Yukon – SE Alaska, the boundary between the Pacific and North America plates is characterized by a syntaxis at the transition between the Aleutian subduction to the W and the Fairweather – Queen Charlotte strike-slip faults to the SE. The relative motion is oblique to the main fault structures, and the area is marked by the Yakutat block collision. From the Chugach – Saint Elias mountains in the plate boundary zone (up to 6 000 m high) to the intraplate strike-slip faults, markers of the present-day deformation give information on its partition in the system.During my PhD, I first measure surface deformation using a dense GPS network, deployed up to 500 km inland the North America plate. After precise processing and corrections of transient effects in the area (postseismic and glacial isostatic rebound), a new residual velocity field is produced for the syntaxis area, from which I derive strain rates. Those data allow me to quantify the fault slip rates for Fairweather and southern Denali strike-slip faults, and to characterize a bi-modal deformation pattern: Along the plate boundary, the deformation is localized on large-scale structures (accretionary prism to the W, Fairweather to the E); In the syntaxis area, strain rates are the highest and the GPS data shows a diffuse intraplate deformation, similar to an indentor pattern. The Yakutat block seems to strongly drive the North America plate deformation.This indentor pattern induces strong lateral variations on the large intraplate faults: the Denali – Totschunda – Duke River system. In a second part, I realize a regional geomorphological study to characterize the role and slip rate of those faults. From very high-resolution Digital Elevation Models (~ 1 m), a detailed cartography is done. On the basis of fieldwork observations, I measure offsets of fluvial and glacial markers, which are sampled for dating. A dextral cumulative deformation is highlighted on the northern Denali Fault, where as all southern Denali is marked by vertical deformation. This study allows me to quantify new slip rates for the system Denali – Totschunda – Duke River, and to show the leading role of the Totschunda (~ 14 mm/a) and Duke River (~ 6 mm/a) faults, contrary to the Denali Fault (~ 1 mm/a) North of the syntaxis.The new tectonic model for the Yakutat collision provides an important case study for the understanding of indentor systems. The concomitance of a rigid-block deformation (to the West) and diffuse deformation (to the East), as well as the near-zero slip rate on the lithospheric-scale southern Denali Fault highlight the major control of boundary conditions and the structural heritage on the orogen deformation.
Au SO Yukon – SE Alaska, la frontière entre les plaques Pacifique et Nord-Amérique est marquée par une syntaxe à la transition entre la subduction des Aléoutiennes à l’O et le décrochement de Fairweather – Queen Charlotte au SE. Le mouvement relatif est oblique et la région est marquée par la collision du bloc Yakutat. De la chaine des Chugach – Saint Elias en frontière de plaque (jusqu’à 6 000 m d’altitude) aux grands décrochements intraplaque, les marqueurs de la déformation actuelle apportent des informations sur sa partition dans ce système.Au cours de cette thèse, je m’intéresse dans un premier temps à mesurer la déformation de surface à l’aide d’un réseau GPS dense, déployé jusqu’à 500 km à l’intérieur de la plaque Nord-Amérique. Après un travail minutieux de traitement des données et de correction des effets transitoires (rebond post-glaciaire et post-sismique), un champ de vitesses résiduelles robuste et inédit est produit au niveau de la syntaxe, dont je dérive des taux de déformation. Ces données me permettent de quantifier les vitesses de glissement les failles décrochantes de Fairweather et Denali au Sud, et également de caractériser une déformation bimodale : En frontière de plaques, la déformation est localisée sur les grandes structures (prisme d’accrétion à l’O, Fairweather à l’E) ; Au niveau de la syntaxe, les taux de déformation sont les plus importants et les données GPS mettent en évidence une déformation intraplaque diffuse, similaire à un champ attendu à l’aplomb d’un indenteur. Le bloc Yakutat semble donc contrôler fortement la déformation de la plaque Nord-Amérique.Ce champ de déformation induit des variations latérales fortes sur les grandes failles intraplaque : le système Denali-Totschunda-Duke River. Dans une seconde partie, je réalise une étude géomorphologique régionale pour caractériser le rôle et la vitesse de ces failles. À partir de Modèles Numériques de Terrain très haute résolution (~ 1 m), une cartographie de détail est réalisée. Puis une mission de terrain me permet de mesurer des décalages sur des marqueurs fluviatiles et glaciaires et collecter des échantillons pour les dater. Je mets en évidence une déformation cumulée dextre sur le segment Nord de la faille de Denali, alors que toute la partie Sud déforme verticalement la surface. Cette étude me permet de quantifier de nouvelles vitesses de failles sur le système Denali-Totschunda-Duke River, et de montrer le rôle prépondérant des failles de Totschunda (~ 14 mm/a) et Duke River (~ 6 mm/a) contrairement à la faille de Denali (~ 1 mm/a) au Nord de la syntaxe.Le nouveau modèle cinématique proposé pour la région de collision Yakutat permet d’apporter un nouvel exemple à la compréhension des systèmes d’indenteur. La concomitance d’une déformation de blocs (à l’Ouest) et diffuse (à l’Est), ainsi que l’absence de déformation sur la faille lithosphérique de Denali Sud met en évidence le contrôle majeur des conditions aux limites et de l’héritage structural dans la déformation des orogènes.
https://hal.archives-ouvertes.fr/tel-01241688

tel-01241688
https://hal.archives-ouvertes.fr/tel-01241688
https://hal.archives-ouvertes.fr/tel-01241688/document
https://hal.archives-ouvertes.fr/tel-01241688/file/manuscrit_anais_v2_optimise.pdf

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
We report chlorine (Cl) and fluorine (F) abundances in minerals, interstitial glasses, and melt inclusions in 12 andesite-hosted, spinel harzburgite xenoliths and crosscutting pyroxenite veins exhumed from the sub-arc lithospheric mantle beneath Avacha volcano in the Kamchatka Arc (NE Russia). The data are used to calculate equilibrium mineral-melt partition coefficients (Dmineral/meltDmineral/melt) for Cl and F relevant to subduction-zone processes and unravel the history of volatile depletion and enrichment mechanisms in an arc setting.Chlorine is ∼100 times more incompatible in pyroxenes (View the MathML sourceDClmineral/melt = 0.005–0.008 [±0.002–0.003]) than F (View the MathML sourceDFmineral/melt = 0.50–0.57 [±0.21–0.24]), which indicates that partial melting of mantle sources leads to strong depletions in Cl relative to F in the residues. The data set in this study suggests a strong control of melt composition on View the MathML sourceDCl,Fpyroxene/melt, in particular H2O contents and Al/(Al + Si), which is in line with recent experiments. Fluorine is compatible in Ca-amphibole in the ‘wet’ sub-arc mantle (View the MathML sourceDFamphibole/melt = 3.5–3.7 [±1.5]) but not Cl (View the MathML sourceDClamphibole/melt = 0.03–0.05 [±0.01–0.03]), indicating that amphibole may fractionate F from Cl in the mantle wedge. The inter-mineral partition coefficients for Cl and F in this study are consistent amongst different harzburgite samples, whether they contain glass or not. In particular, disseminated amphibole hosts much of the Cl and F bulk rock budgets of spinel harzburgites (View the MathML sourceDClamphibole/pyroxene up to 14 and View the MathML sourceDFamphibole/pyroxene up to 40). Chlorine and fluorine are variably enriched (up to 1500 ppm Cl and 750 ppm F) in the parental arc picrite and boninite melts of primitive pyroxenite veins (and related melt inclusions) crosscutting spinel harzburgites.Based on the data in this study, the main inferences on the behaviour of Cl and F during melting and metasomatic processes in the sub-arc mantle are as follow:(i) Melting models show that most depleted mantle protoliths of intra-oceanic arc sources can have extremely low Cl/F (0.002–0.007) before being overprinted by subduction-derived components.(ii) Chlorine has a higher percolation distance in the mantle than F. Even for small fluid or melt volumes, Cl and F signatures of partial melting are overprinted by those of pervasive percolation, which increases Cl/F in percolating agents and bulk peridotites during chromatographic interaction and/or amphibole-forming metasomatic reactions. These processes ultimately control the bulk Cl and F compositions of the residual mantle lithosphere beneath arcs, and likely in other tectonic settings.(iii) Fluxed melting models suggest that Cl enrichment in arc picrite and boninite melts in this study, and in many arc melt inclusions reported in the literature, could be related to the infiltration of high Cl/F fluids derived from subducted serpentinite or altered crust in mantle wedge sources. However, these high Cl/F signatures should be re-evaluated with new models in light of the possible overprint of pervasive percolation effects in the mantle. The breakdown of amphibole (and/or mica) in the deep metasomatised mantle at higher pressure and temperature conditions than in the slab may explain, at least in part, the positive correlations between F abundances and Cl/F in primitive arc melt inclusions and slab depth.
ISSN: 0016-7037

hal-01553610
https://hal.archives-ouvertes.fr/hal-01553610
DOI : 10.1016/j.gca.2016.10.035

Éditeur(s) : HAL CCSD Elsevier
Résumé : International audience
Melting experiments were performed on a silica-rich peridotite composition at 10-17 GPa to determine majoritic garnet-melt partition coefficients (D) for major and trace elements. Our results show that D for many elements, including Na, Sc, Y and rare earth elements (REE), varies significantly with increasing pressure or proportion of majorite component. Lu and Sc become incompatible at 17 GPa, with D decreasing from 1.5 at 10 GPa to 0.9 at 17 GPa. As predicted from lattice strain, log D for isovalent cations entering the large site of majoritic garnet exhibits a near-parabolic dependence on ionic radius. Our data are used to refine a previously published predictive model for garnet-melt partitioning of trivalent cations, which suffered from a lack of calibration in the 10-20 GPa range. Our results suggest that Archean Al-depleted komatiites from Barberton (South Africa) may have been generated by partial melting of dry peridotite at depths between 200 and 400 km. We also speculate that transition zone diamonds from Kankan (Guinea), which contain inclusions of majoritic garnet, may have formed from the partial reduction of CO2-rich magmas that subsequently transported them to the surface. This hypothesis would provide an explanation for the REE patterns of majoritic garnet trapped within these diamonds, including Eu anomalies. Finally, we show that segregation of majoritic garnet-bearing cumulates during crystallisation of a deep Martian magma ocean could lead to a variety of Lu/Hf and Sm/Nd ratios depending on pressure, leading to a range of ε143Nd and ε176Hf isotope signatures for potential mantle sources of Martian rocks.
ISSN: 0024-4937

hal-00767567
https://hal.archives-ouvertes.fr/hal-00767567
DOI : 10.1016/j.lithos.2012.06.013

Éditeur(s) : HAL CCSD Wiley
Résumé : International audience
We report in situ LA-ICP-MS trace element analyses of silicate phases in olivine-bearing chondrules in the Sahara 97096 (EH3) enstatite chondrite. Most olivine and enstatite present rare earth element (REE) patterns comparable to their counterparts in type I chondrules in ordinary chondrites. They thus likely share a similar igneous origin, likely under similar redox conditions. The mesostasis however frequently shows negative Eu and/or Yb (and more rarely Sm) anomalies, evidently out of equilibrium with olivine and enstatite. We suggest that this reflects crystallization of oldhamite during a sulfidation event, already inferred by others, during which the mesostasis was molten, where the complementary positive Eu and Yb anomalies exhibited by oldhamite would have possibly arisen due to a divalent state of these elements. Much of this igneous oldhamite would have been expelled from the chondrules, presumably by inertial acceleration or surface tension effects, and would have contributed to the high abundance of opaque nodules found outside them in EH chondrites. In two chondrules, olivine and enstatite exhibit negatively sloped REE patterns, which may be an extreme manifestation of a general phenomenon (possibly linked to near-liquidus partitioning) underlying the overabundance of light REE observed in most chondrule silicates relative to equilibrium predictions. The silicate phases in one of these two chondrules show complementary Eu, Yb, and Sm anomalies providing direct evidence for the postulated occurrence of the divalent state for these elements at some stage in the formation reservoir of enstatite chondrites. Our work supports the idea that the peculiarities of enstatite chondrites may not require a condensation sequence at high C/O ratios as has long been believed.
ISSN: 1086-9379

hal-01217095
https://hal.archives-ouvertes.fr/hal-01217095
DOI : 10.1111/maps.12481

Éditeur(s) : HAL CCSD Oxford University Press (OUP)
Résumé : The Oman ophiolite consists of several massifs cropping out along a 500 km long band trending NW-SE along the coast of Oman; it is one of the best exposed sections of oceanic crust and mantle in the world. There is a gradient in igneous processes and composition in the ophiolite, with the northern massifs recording a polygenetic igneous history involving an increasingly important subduction component, whereas the southern massifs were formed primarily via a mid-ocean ridge basalt (MORB)-like, single-stage process at a submarine spreading ridge. In this study we use geochemical data from Wadi Tayin, which is one of the southern massifs of the Oman ophiolite, to constrain the composition and genesis of oceanic crust and upper mantle. The Wadi Tayin harzburgites are residues of partial melting that are as depleted as the most depleted mid-ocean ridge peridotites. They have low middle- to heavy rare earth element ratios, most probably reflecting melting close to and beyond the exhaustion of clinopyroxene. Like many abyssal peridotites, the Wadi Tayin samples show enrichment in highly incompatible elements, giving rise to U-shaped MORB-normalized trace element patterns. We favor the idea that this is caused by enrichment of highly incompatible elements along grain boundaries in peridotite, which may be the result of near-equilibrium partitioning between grain boundaries and grain interiors, rather than disequilibrium processes. Equilibrium partitioning between crystals and grain boundaries during melting and melt extraction is also our preferred explanation for ubiquitous high Pb contents relative to Ce and La in our samples as well as in most abyssal peridotites. The Wadi Tayin samples record substantial variability in terms of osmium isotopic composition and platinum group element (PGE) concentrations. The more radiogenic nature of the dunites and impregnated peridotites compared with the residual harzburgites may be due to relatively high Os-187/Os-188 in melt transported through the dunites. The presence of residual peridotites with whole-rock osmium isotopic compositions less radiogenic than MORB may be explained by melting of a veined (upwelling) mantle forming mixed melts in which much of the Os derives from veins with high Re/Os in a matrix of previously depleted peridotite. An important result of this study is that the shallowest samples are refertilized (i.e. anomalously enriched in incompatible elements), and record relatively high metamorphic closure temperatures. These observations suggest that migrating melt underwent crystallization during rapid cooling in the uppermost mantle during and immediately after ridge magmatism. A variety of geothermometers all yield the result that the stratigraphically highest harzburgites equilibrated at higher temperature than the deeper ones. The systematic decrease in closure temperature with increasing depth below the Moho transition zone probably reflects systematic variation in cooling rate as a function of depth in the mantle section. We hypothesize that the refertilization and higher closure temperatures recorded by the uppermost mantle samples are linked. More rapid cooling led to higher closure temperatures, and to partial crystallization of migrating melts in the shallowest part of the mantle section, yielding slightly elevated abundances of elements such as Ca and Na, and incompatible trace elements.
ISSN: 0022-3530

hal-00456056
https://hal.archives-ouvertes.fr/hal-00456056
DOI : 10.1093/petrology/egp077

Éditeur(s) : HAL CCSD Mineralogical Society of America
Résumé : International audience
Hydrogen distribution between nominally anhydrous minerals (NAMs) of a garnet-lherzolite under subsolidus conditions has been investigated. Separated NAMs from a garnet-peridotite from Patagonia (Chile) are annealed together (olivine, orthopyroxene, clinopyroxene, and garnet) using a piston-cylinder at 3 GPa and 1100 °C using talc-pyrex cell assembly for 10, 25, and 100 h. The talc-pyrex assembly provides enough hydrogen in the system to re-equilibrate the hydrogen concentrations at high pressure. The three coexisting nominally anhydrous minerals (NAMs, i.e., olivine, orthopyroxene, and clinopyroxene) were successfully analyzed using FTIR. The resulting hydrogen concentrations exceed significantly the initial hydrogen concentration by a factor of 13 for olivine and a factor of 3 for both pyroxenes. Once mineral-specific infrared calibrations are applied, the average concentrations in NAMs are 115 ± 12 ppm wt H2O for olivine, 635 ± 75 ppm wt H2O for orthopyroxene, and 1214 ± 137 ppm wt H2O for clinopyroxene, garnet grains are dry. Since local equilibrium seems achieved over time (for 100 h), the calculated concentration ratios are interpreted as mineral-to-mineral hydrogen partition coefficients (i.e., Nernst’s law) for a garnet-peridotite assemblage. It yields, based on mineral-specific infrared calibrations, DOpx/Ol = 5 ± 1, DCpx/Ol = 10 ± 2, and DCpx/Opx = 1.9 ± 0.4. While DCpx/Opx is in agreement (within error) with previous results from experimental studies and concentration ratios observed in mantle-derived peridotites, the DPx/Ol from this study are significantly lower than the values reported from mantle-derived xenoliths and also at odd with several previous experimental studies where melt and/or hydrous minerals co-exists with NAMs. The results confirm the sensitivity of hydrogen incorporation in olivine regarding the amount of water-derived species (H) in the system and/or the amount of water in the coexisting silicate melt. The results are in agreement with an important but incomplete dehydration of mantle-derived olivine occurring at depth, during transport by the host magma or during slow lava flow cooling at the surface. The rapid concentration modification in mantle pyroxenes also points out that pyroxenes might not be a hydrogen recorder as reliable as previously thought.
ISSN: 0003-004X

hal-01622132
https://hal.archives-ouvertes.fr/hal-01622132
DOI : 10.2138/am-2017-6089

Éditeur(s) : HAL CCSD Wiley
Résumé : International audience
We report trace element analyses from mineral phases in chondrules from carbonaceous chondrites (Vigarano, Renazzo, and Acfer 187), carried out by laser ablation inductively coupled plasma-mass spectrometry. Results are similar in all three meteorites. Mesostasis has rare earth element (REE) concentrations of 10-20 × CI. Low-Ca pyroxene has light REE (LREE) concentrations near 0.1 × CI and heavy REE (HREE) near 1 × CI, respectively. Olivine has HREE concentrations at 0.1-1 × CI and LREE around 10−2 × CI. The coarsest olivine crystals tend to have the most fractionated REE patterns, indicative of equilibrium partitioning. Low-Ca pyroxene in the most pyroxene-rich chondrules tends to have the lowest REE concentrations. Type I chondrules seem to have undergone a significant degree of batch crystallization (as opposed to fractional crystallization), which requires cooling rates slower than 1-100 K h−1. This would fill the gap between igneous calcium-aluminum-rich inclusions (CAIs) and type II chondrules. The anticorrelation between REE abundances and pyroxene mode may be understood as due to dilution by addition of silica to the chondrule melt, as in the gas-melt interaction scenario of Libourel et al. (2006). The rapid cooling rate (of the order of 1000 K h−1) which seems recorded by low-Ca pyroxene, contrasted with the more diverse record of olivine, may point to a nonlinear cooling history or suggest that formation of pyroxene-rich chondrule margins was an event distinct from the crystallization of the interior.
ISSN: 1086-9379

hal-00795434
https://hal.archives-ouvertes.fr/hal-00795434
DOI : 10.1111/maps.12005

Éditeur(s) : HAL CCSD
Résumé : National audience
We propose a method to solve the problem of clustering individuals described by mixed variables structured in blocks. It is a hierarchical method with two levels similar to Hierarchical Principal Component Analysis presented by Wold, based on the mixed topology maps (MTM). The first step consists in establishing, for each block of mixed variables, a clustering representing a synthesis of local individuals at the block level. The second step is to re-apply MTM on the weighted results of the first level. We obtain a global clustering of individuals, consensus among the partitions from level 1. The proposed method is illustrated on data from the national survey in the French housing stock carried out in 2003-2005 by the Observatory for Indoor Air Quality (OQAI).
Fouille de Données Complexes, complexité liée aux données multiples et massives (FDC)
bordeaux, France

hal-00748841
https://hal.archives-ouvertes.fr/hal-00748841
https://hal.archives-ouvertes.fr/hal-00748841/document