Acquérir des informations sur les forts séismes passés est crucial pour anticiper les caractéristiques des forts séismes futurs. Une partie des traces laissées par les forts séismes passés sont enfouies dans les premiers mètres du sol et sont en général révélées par des tranchées de quelques mètres de profondeur ouvertes à travers les failles sismogènes. Bien que pertinente, cette méthode est destructive. L'objectif a été de développer une nouvelle forme de paléosismologie, non destructive, basée sur l'imagerie géoradar pseudo 3D, capable de retrouver ces traces enfouies des séismes passés. Dans ce travail, cinq sites d’étude sont présentés, situés le long de failles actives décrochantes de Nouvelle Zélande. Notre nouvelle approche débute, dans un premier temps, par l’analyse ‘classique’ de la morphologie de surface à partir de données LiDAR et de MNT GPS haute résolution. Ceci nous permet d’identifier l'ensemble des marqueurs morphologiques préservés à la surface et les déplacements horizontaux qu’ils ont enregistrés. Dans un second temps, l’analyse des profils GPR pseudo-3D acquis en chacun des sites révèlent des réflecteurs principaux dans les premiers 5-10 m du sol recoupés par un grand nombre de marqueurs morphologiques, partiellement ou totalement invisibles en surface. La plupart de ces marqueurs enfouis sont coupés et décalés par la faille considérée. Les mesures de ces décalages fournissent des collections denses de déplacements cumulés sur chacune des failles investiguées avec généralement un nombre de mesures effectués en sub-surface 10 à 20 fois plus important qu’en surface et couvrant une plus large gamme de valeurs. L’application sur la faille de Hope de cette approche a notamment permis de mettre en évidence un déplacement latéral caractéristique de 3.2 ± 1 m lors des 30-35 derniers forts séismes. Ce travail démontre le potentiel de l'imagerie géoradar pseudo-3D à détecter une partie de l'histoire sismique des failles et, ce faisant, à fournir des informations sur les caractéristiques des forts séismes passés. / Collecting information on past strong earthquakes is crucial to anticipate the characteristics of the future strong earthquakes that threaten us. A part of the traces left by the past earthquakes remains hidden in the first few meters of the ground. Until now, paleoseismological trenches across faults have been used to search for these traces. Though relevant, this method is destructive and allows, at best, detecting the few most recent events. The objective of my PhD work, done in the framework of the ANR project CENTURISK, was to develop a novel form of paleoseismology, of geophysical type, based on multi-frequency, pseudo-3D GPR surveys. The idea is to image at high-resolution the architecture of the first ≈ 10 m of the ground over wide areas along active faults, in order to detect the possibly buried traces, especially the offsets, produced by the last 10-20 strong earthquakes on the fault. We have first developed the approach by adapting the acquisition and processing of GPR data to the selected targets. We have then applied the approach on some of the largest active strike-slip faults in New Zealand, where sedimentation conditions are ideal. Twelve sites were investigated, 5 of them are presented in this work. At each site, we first analyzed the surface morphology in the greatest detail on LiDAR data and high resolution GPS DEMs. This analysis allowed us to identify all the morphological markers preserved at the ground surface, and being offset by the fault. We measured these surface offsets, doing so collecting a dense population of cumulative displacement values. We then surveyed each site with 40-60, 100 and 250 MHz, hundreds of meters long GPR profiles, parallel to the fault and regularly spaced by 5-10 m on either side of the fault trace. At each site, the processing of the GPR data revealed a large number of buried markers – palaeosurfaces and incision features, hidden in the first 5-10 m of the ground. Most of the buried markers were observed cut and laterally displaced by the fault, and these offsets could be measured. The measures provide a dense collection of cumulative offsets on each investigated fault, generally 10-20 times more than ever reported. To analyze these dense surface and sub-surface data collections, we used statistical methods made to define and retain only the best constrained offset values. These best values are separated by slip increments that are directly related to the successive coseismic slips that we search. The entire analysis revealed that the offsets measured in the sub-surface fill the gaps in the surface record, and that the surface offsets are systematically lower than those measured in the sub-surface on the same markers. Additionally, the buried record is longer than the surface record. Applied to the Hope Fault, our novel approach allowed identifying the last 30-35 strong earthquakes that broke the fault, each had produced a lateral offset at surface of 3.2 ± 1 m and got a magnitude ≈ Mw 7.0-7.4. Applied to the Wellington Fault (at Te Marua site), the approach allowed identifying a minimum of 15 past strong earthquakes, each had produced a lateral offset at surface of 3.7 ± 1.7 m and got a magnitude ≈ Mw 6.9-7.6. My PhD work thus confirms the great potential of pseudo-3D Ground Penetrating Radar survey to detect a significant part of the fault seismic history, and thus to provide critical information to determine the displacements and magnitudes of the past strong earthquakes on faults. Applied to seismogenic faults worldwide, in complement to surface approaches, the geophysical GPR paleoseismology should help better assessing seismic hazard.
| Identifer | oai:union.ndltd.org:theses.fr/2013GRENU001 |
| Date | 24 January 2013 |
| Creators | Beaupretre, Sophie |
| Contributors | Grenoble, Manighetti, Isabelle, Garambois, Stéphane |
| Source Sets | Dépôt national des thèses électroniques françaises |
| Language | French |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation, Text |
Page generated in 0.0019 seconds