Xidatan Permanent GPS (XPEG) : Un réseau GPS permanent pour étudier une faille décrochante : la faille du Kunlun IPGP : Y. Klinger, C. Romieu, P. Tapponnier Geosciences Azur : J.M. Nocquet, M. Ueno Chinese Earthquake Administration : Xu Xiwey romieu@geoazur.unice.fr
CONTEXTE TECTONIQUE Poussée de l’Inde : 4cm/an 3 cm vers le NE au Tibet Central Mesures GPS (articles….) :vitesse Tibet accomodée soit de façon répartie soit sur qqs failles : essentiellement Altyn Tagh et Kunlun Kunlun : de 0.5cm/an à 1.1 cm/an
(modified from Tapponnier et al., 2001) Kokoxili: Nov 2001, M 7,8, 400 km de rupture Manyi: aout 1997, M 7.6: 150km de rupture 1937 M =7.5 longueur rupture ? (D’après Klinger et al, 2005 et Van der Voerde et al, 2002)
Crédit Yann Klinger
Séisme de Kokoxili : D ~ 4 - 5 m Kokoxili :D ~ 4 - 5 m Vitesse sur la faille du Kunlun = 11.5mm/y Période de récurrence pour un tel évènement : 350 à 450 années? Sur le segment de Xidatan, il ne s’est rien passé depuis au moins 300 ou 400 ans Sur Kokoxili : D ~ 4 - 5 m, les vitesses de déplacement mesurées par luminescence ou par datation carbone montrent des vitesses sur la Kunlun Fault de 11,5 mm/y cela implique un temps de retour de l’ordre de 400 ans? Sur Xidatan pas d’événement connu au cours des 4 ou 5 derniers siècles: bientôt ? D’après Van der Woerd et al, 2000 D’après Klinger, 2006
Objectifs scientifiques Mesurer la déformation associée au chargement de la faille en période pré-sismique Au premier ordre, quel est la vitesse de glissement actuelle ? Quelle est la profondeur de blocage ? Observe-t-on un chargement constant ? Voit-on des signaux transitoires et d’éventuels précurseurs ? Etudier la rupture Capture le déplacement co-sismique statique Capture le déplacement dynamique
En période présismique: chargement, vitesse actuelle et profondeur de blocage
Des signaux transitoires?
Où en est on? La reconnaissance de 13 des 14 points a été faite en septembre – octobre 2006 Site 20
Critères de choix Critères classiques GPS :bedrock, masques, accessibilité (pb du gel et du permafrost), communication En fonction des objectifs scientifiques et des modèles de dislocation
Matériel utilisé Antennes Trimble Zéphyr Récepteurs Trimble NetRS PC durcis PIP6 (stockage de données et communication avec Pékin)
Données Les données à 30s seront systématiquement rapatriées et serviront au calcul du champs de vitesse Les données à 10Hz seront conservées sur site et ne seront rapatriées qu’en cas d’évènement sismique
10 Hz : pourquoi ? Un exemple d’utilisation de données GPS haute fréquence : le séisme de Miyagi-oki (Mw = 7.2) en 2005
DATA DESCRIPTION The 2005 Miyagi-oki earthquake (Mw=7.2) Pacific Ocean 0549 0550 0031 50 km Denali, Hokkaido, Mw ~ 8 25 s Duration 42 km Depth 142.11° E Longitude 38.16° N Latitude Epicentre 11:46 JST (02:46 UTC) Time 16/08/2005 Date We studied the 2005 Miyagi-oki Earthquake (Mw=7.2), which occurred about a year ago, 350 km north of Tokyo, and 50 km off the coast. This earthquake was much smaller than the Denali and the Hokkaido quakes, thus we need to study smaller ground movement. The specification is summarised in this table. We analysed 1‑Hz GEONET data observed on the day of the earthquake. The GEONET is a GPS network operated by the Geographical Survey Institute (GSI) of Japan. More than 1200 GPS stations are distributed approximately each 20 km. We processed data using different methods for a period of 1 h. I will show some example of our results obtained from the stations where accelerometers are co-located within a range of 2 km. Our approach is single-receiver method but we compared the results with relative mode. The station 0628 is used as reference for relative velocity determination. 350 km north of Tokyo 50 km off the coast Tokyo 0628 GEONET 1-Hz GPS data (GSI of Japan) Dense network (separation ~ 20 km) Processed for 1 hour
RESULTS 1 - Single point vs. Relative mode - 0550 - Velocity N-S 0550 - Velocity N-S Relative (30 km) Single point Relative (461 km) Single point 2 cm/s 2 cm/s These images show the comparison of velocity and acceleration between relative and single point method. I show here N-S and U-D components for the station 0550, the closest station to the epicentre. The distance was about 50 km. The velocity and acceleration are unfiltered. We see that seismic waves arrived at the station, 15-20 second from the earthquake time and lasted about 15 seconds. Relative method cancel out most of common errors such as satellite orbits, propagation errors in the atmosphere and provide more accurate velocity estimation for a moving object with respect to the fixed stationary reference station. However, in this case, relative mode does not look better than single point method. After the passage of seismic waves, there is no movement and thus they are considered as noise. Single point velocity N-S 0550 - Acceleration N-S 0549 0550
RESULTS 2 - GPS vs. Accelerometers (0.01 Hz < f < 0.2 Hz) - (m/s-2 ) 0550 – MYG011 (49 km, Δ0.6 km) 0031 – AKT015 (216 km, Δ1.8 km) (m/s-2 ) x10-4 8 E-W E-W 0.01 -0.01 -8 x10-4 Single point velocity-derived acceleration looked better than that obtained from relative velocity. In order to validate the velocity-derived acceleration, we compared the GPS acceleration with that of accelerometers located in the vicinity of GPS stations. In order to compare the two results, both the results were band-pass filtered with frequencies between 0.01 and 0.2 Hz. Thicker lines are for accelerometer. The images on your left are the 3 components of acceleration observed at Station 0550, the closest site to the epicentre. The distance from the epicentre was 49 km and the distance between the GPS and accelerometer stations was 0.6 km. In the E-W component, the peaks corresponding to the arrival of seismic waves exist in both and fit well each other. The N-S component is less good for the amplitude correspondence at the time of the shock but the other epochs are corresponding fairly well. The U-D component also shows good correspondence of the peaks but for other epochs, the high- frequency oscillation is more significant in GPS acceleration. The magnitude of oscillation is less than 5 mm/s2, which corresponds to the noise of GPS signals. The images on your right show the same comparison but for Station 0031, located 216 km from the epicentre. The distance between the GPS and accelerometer stations was 1.8 km. The magnitude of acceleration is much smaller as seismic signals attenuate over the path. The images show good correspondence in horizontal components. The vertical component is less good but the magnitude of measured acceleration is 4 mm/s2, which is beyond the limit of GPS measurements for the vertical component. We could say that the single point results were quite good. 0.01 10 N-S N-S -0.01 -10 x10-4 0.01 6 U-D U-D -0.01 -4
RESULTS 3 - Position by integration vs. kinematic positioning - Position by single point velocity integration Position by single point velocity integration 0550 E-W 0550 - 0549 E-W 2 cm 2 cm The images on your left show the displacement obtained by integration for Station 0550 and 0549, respectively. We selected the stations to compare with kinematic positioning. We removed residual receiver clock drifts and made integration. I show the E-W component which had more significant impacts on these stations. We can see the ground movement during the shake. There is still some problem due to the satellite clock drifts. We need to work on it. The images on your right are baseline displacement. The upper image is displacement by integration and the lower one is by relative kimematic positioning. We see the similar trend on both images. The results are preliminary but we could see the static co-seismic displacement in the order of 1 cm in both the images. It is necessary to choose a reference site located far from the rover in order to avoid the signals containing the seismic waves reached two stations almost at the same time. Position by single point velocity integration Relative kinematic positioning - L1 0549 E-W 0550 - 0549 E-W
Merci