Caractérisation de la déformation active par l’Interférométrie Radar (InSAR) : Failles sismiques aveugles et cachées de l’Atlas Tellien (Algérie) et du Rif (Maroc) le long de la limite des plaques Afrique-Eurasie.
BELABBES, Samir (2008) Caractérisation de la déformation active par l’Interférométrie Radar (InSAR) : Failles sismiques aveugles et cachées de l’Atlas Tellien (Algérie) et du Rif (Maroc) le long de la limite des plaques Afrique-Eurasie. Thèses de doctorat, Université Louis Pasteur.
Plein texte disponible en tant que :
This thesis aims at the study of crustal deformation caused by moderate and large earthquakes in the Tell Atlas(northern Algeria) and the Rif Mountains (northern Morocco). My work included mainly the identification and characterisation of active faulting using the method of Synthetic Aperture Radar interferometry (InSAR). The active deformation and related seismicity is essentially concentrated along the Africa – Eurasia plate boundary where large and moderate earthquakes result from the plate convergence. The surface deformation associated with moderate-size earthquakes is, however, very often hardly visible in the field because because the deformation occur on blind faults (e.g., El Hoceima 2004, Mw 6.4; Ain Temouchent 1999, Mw 5.7) or to hidden faults beneath Quaternary deposits or offshore along the Mediterranean coastline (e.g., Zemmouri 2003; Mw 6.8). The InSAR application to these complex tectonic zones provides an opportunity to obtain the surface field displacement (to the cm scale) and related seismic rupture parameters. In this thesis, made of five chapters I used the SAR data of ERS-1, ERS-2 and ENVISAT European satellites, as well as the data of RADARSAT Canadian satellite. In the first chapter I present the basic principles and physical fundamentals of InSAR analysis and the Poly3Dinv modelling code. InSAR provides access to the phase difference between two radar images of the earth surface taken from two different times and related to the travel-time of the radar signal at the same target point. The interferogram is obtained after the phase correction due to the topography, the atmospheric artefacts and the pixel image. The interferogram shows fringes that illustrate the earth surface changes in between two images in the line of sight (LOS) direction, with a half wavelength accuracy of the radar signal (λ/2 = 2.83 cm for C band in the case of ERS, Envisat and Radarsat satellites). SAR data were processed to obtain interferograms using Sarscape, ROI_PAC and Doris softwares. The modelling code Poly3Dinv that was used to invert surface deformation (InSAR and field measurements) computes dislocation by means of triangular surface(Thomas, 1993 ; Maerten et al., 2005). These surfaces allow us to disregard voids and overlaps inevitable in the case of rectangular surface elements (Okada, 1992). This modelling approach is useful in the case of ruptures with complex geometries and provides a better constraint on the modelling of near-field geodetic data (Maerten et al, 2005). The obtained results describe the physical characteristics of seismogenic faults and related slip distribution. Chapter II describes the seismotectonics framework of the Tell Atlas and Rif Mountains. The Tell Atlas has been regularly affected by damaging earthquakes in the past (e.g.; Alger, 1365 and in 1716 Io = X MKS ; Oran in 1790 Io = XI-X MKS). More recently, several large earthquakes have occured in north Africa and the two major events are the El Asnam (10/10/1980, Mw 7.3) and Zemmouri (21/05/2003, Mw 6.8) earthquakes. Although moderate, other main earthquakes were responsible of severe damage in the Tell Atlas (e.g., 10/08/1994 at Mascara, Mw 5.7; 21/12/1999 at Ain Temouchent, Mw 5.7; 10/11/2000 at Beni Ourtilane, Mw=5.6). The Rif Mountains and the Al Hoceima region experienced two major earthquakes in 26/05/1994 and 24/02/2004 with Mw 6.0 and 6.4, respectively. Moreover, we observe that the seismicity in northern Morocco is mostly related to strike-slip focal mechanism solutions. The seismicity in NW Algeria is related to NE-SW trending active foldrelated- faults and shows focal mechanisms with reverse faulting and ~ 2 mm/yr NNW-SSE shortening rate in the Tell Atlas due to the Africa-Eurasia plate convergence(Meghraoui & Doumaz, 1996). Chapter III presents the InSAR analysis applied to the active deformation of the Tell Atlas and details the study of the Zemmouri 2003, Ain Temouchent 1999 and Mascara 1994 earthquakes. The Zemmouri thrust earthquake, the largest recorded event in the Algiers region since 1716, shows a coastal epicentre and induced ~ 0.50 m uplift along 50-km-long shorelin (Meghraoui et al., 2004). The offshore fault location has been the subject of a large debate with the coastal uplift suggesting a possible surface (sea bottom) rupture at 5 – 10 km from the coastline and 1.75 × 1019 Nm geodetic moment (Meghraoui et al., 2004) in agreement with the analysis of waveform inversions and related source time functions (Delouis et al., 2004). Other studies using sea bottom morphology coupled with seismic profiles and models infer a fault scarp at 15 to 30 km from the shoreline (Deverchère et al., 2005 ; Semmane et al., 2005). Knowing that InSAR may provide some constraints in the rupture parameters, I proceeded with the image analysis of SAR (IS2 of ENVISAT) and ST4 of RADARSAT. Despite the low coherency factor I have obtained 4 interferograms with clear fringes: 2 ENVISAT interferograms in the ascending and descending modes cover the western epicentral region, and 2 RADARSAT interferograms in the descending mode cover the entire Zemmouri earthquake area. The extent of the interferograms from Cap Matifou (west of epicentre) to Tighzirt (east of epicentre) allowed a more accurate identification of the coseismic surface deformation consistent with the observed and measured coastal uplift. The maximum coastal uplift near Boumerdes-Figuier corresponds to the highest fringe gradient visible on interferograms (14 on ENVISAT and 16 on RADARSAT). RADARSAT interferograms being more sensible to horizontal movements due to their 36° incidence angle (23° for Envisat IS2), show offset and inverted fringes east of Boumerdes. The ascending ENVISAT interferogram also shows two offset fringes south of Cap Matifou with a similar orientation of the RADARSAT interferograms. Inverted and offset fringes in the western epicentral region correspond to a minimum 5.66 cm of lateral slip in the LOS. The modelling results obtained from 27 fault planes regularly located from the coastline to 18 km offshore indicate that the best fit of InSAR data combined with coastal uplift is a curved fault plane at 8 km from the coast. Hence, the modelled rupture is 50-kmlong, strikes N 65°, has an average dip of 40° SE and shows a maximum of 4.7 m reverse slip at depth. The area of maximum slip indicates two patches east and west of the hypocenter location that confirm the bilateral rupture propagation obtained by Yagi, 2004 and Delouis et al., 2004. In addition, the rupture location and characteristics are in good agreement with the aftershocks distribution and tomographic study (Ayadi et al., 2008). Horizontal movements observed on interferograms and on minor cracks in the field may be explained by 0.15 m slip at depth on the conjugate Thenia fault. Recent InSAR studies of the Niigata-Ken Chuetsu-Oki (Japon, Mw 6.8) and San Simeon in California (2003, Mw 6.5) display comparable complex surface deformation related with seismic ruptures (Toda et al., in preparation; Wicks, 2006). Our results of the Zemmouri earthquake InSAR analysis have been submitted to the J. Geophys. Res. After a first revision. The moderate sized Ain Temouchent earthquake (Mw 5.7) shows a reverse focal mechanism but with no surface ruptures and no recorded aftershocks due to the absence of a local seismic network. Hence, the precise epicentre location, extent of deformed zone and related fault rupture remained unknown. Although the moderate magnitude coupled with the low coherence of the earthquake area did not favour an InSAR analysis, I considered recent successful examples of interferograms obtained from the 1997 Umbria-Marche earthquake sequence in Italy (Mw 5.7; Crippa et al., 2006). The analysis of SAR images of the 1999 Ain Temouchent earthquake area did indicate deformation fringes showing NE-SW trending lobe and uplift in agreement with the focal mechanism solution. Using the gradient of fringes in its SE area I inferred a 20-km-long, 16-km-wide and 32° NW dip of the fold-related fault rupture. The forward model obtained from the best fit of surface deformation yields 5.1 × 1017 Nm of seismic moment and a maximum 1 m reverse slip at ~ 6 km depth. Field observations showing an active NE-SW trending fold and related fault in the epicentral area corroborates the modelled InSAR data. This work have been submitted to the Journal of Seismology and a revised version is under review. The 1994 Mascara earthquake did not exceed Mw 5.7 and did not show surface ruptures. Here again, despite the low coherency of SAR images an interferogram obtained from ERS-1 data shows NE-SW trending lobe of fringes. The lobe shape and narrow fringe gradient to the SE suggest a NE-SW trending, 20-km-long rupture dipping NW. This active structure as determined by the InSAR study appears to be very comparable to the Ain Temouchent fold-related fault. Chapter IV describes the InSAR analysis of the 1994 and 2004 Al Hoceima earthquakes (Mw 6.0, 6.4, respectively) of the Rif Mountains. Although the 2004 earthquake reached 5.2 × 1018 Nm (Global CMT) with a strike-slip mechanism, no surface ruptures were clearly identified. The study and modelling of SAR Envisat data in ascending and descending modes provide a good constraint of the blind fault rupture. The best fit solution corresponds to a N 130° trending, 21-km-long and 16.5-km-wide strike-slip fault that imply 6.8 × 1018 N m geodetic moment with 2.7 m slip at ~ 7-km-depth. The results of this study are published in the Bulletin of the Seismological Society of America (BSSA) in 2006. The combined study of the 1994 and 2004 earthquake interferograms revealed the existence of two blind conjugate ruptures (N 23° with left-lateral slip and N 130° with right-lateral slip) in the Rif Mountains. The results are published in Earth and Planetary Science Letters (EPSL) in 2006. Chapter V presents a synthesis of InSAR studies of large and moderate earthquakes associated with active folding of the Tell Atlas of Algeria. I observe that the correlation between active fold-related faults and fringe lobes of interferograms provides a good constraint of the coseismic rupture dimension, its location (not always determined by seismic networks) and geodetic moment. Other late Quaternary fold-related faults in the Tell Atlas may have generated earthquakes and produced surface faulting and deformation would have shown comparable interferograms with fringe lobes. The location of potential active folds in the Tell Atlas and the modeled fringes from the 1980 El Asnam earthquake data allow a discussion on the physical characteristics of thrust faults and extent of surface deformation for moderate and large earthquakes. In conclusion, and even if the studied earthquakes did not present the most favourable conditions for the InSAR analyses, field displacements of blind or hidden earthquakes are successfully extracted from interferograms. The results obtained for the Zemmouri earthquake emphasize the existence of a hidden offshore fault with complex coseismic rupture (modelled curved shape that corresponds to conjugate faults) and confirm the coastal uplift measured in the field. The InSAR contribution to the study of moderate earthquakes is in our case decisive since it substantially improves the determination of the earthquake location, and the fault rupture identification and characterisation. In perspective, this thesis shows the contribution of InSAR for earthquake studies along the plate boundary in north Africa. Since outstanding InSAR results were obtained for earthquake magnitudes as low as Mw 4.5 in the Zagros Mountains of Iran (Lohman & Simons, 2005),active tectonic structures of the Sahara Atlas (e.g., the 1960 Agadir earthquake, Ms 5.9) may present similar conditions and obtain interferograms for relatively low magnitudes. Another application would be the use of PS-InSAR (Permanent Scatterer InSAR) using reflectors in regions with slow deformation, in comparison with GPS measurements. These studies will perhaps offer a better knowledge on the seismic cycle of convergent tectonic domains. In appendix, I add an article published in Earth and Planetary Science Letters (EPSL) : (Cakir, Z., A. M. Akoglu, S. Belabbes, S. Ergintav, M. Meghraoui, 2005) in which my contribution is also my first initiation to the treatment of SAR images and the analysis of slow aseismic deformation measured with radar interferometry (namely the creeping section of the Gerede segment of the North Anatolian Fault with 8 +/- 3 mm/yr right-lateral movement).
Administrateurs de l'archive uniquement : éditer cet enregistrement