Two Contrasting InSAR Studies of Recent Earthquakes in Tibet Barry Parsons Department of Earth Sciences University of Oxford John Elliott, Wanpeng Feng,, James Jackson, Zhenhong Li, Xinjian Shan, Alastair Sloan, Richard Walker, Richard Walters & Peizhen Zhang
Significant Earthquakes over the Past Two Years Earthquake Date Magnitude Population exposed to MMI VIII (USGS PAGER) Casualties = Fatalities+ Missing Casualities/ Exposed L Aquila, Italy April 2009 6.3 10,000 308 3% Haiti January 2010 7.0 3,128,000 222,500 7% Maule, Chile February 2010 8.8 3,649,000 562 0.015% Sierra El Major, Mexico April 2010 7.2 561,000 4 0.0007% Yushu, Tibet April 2010 6.9 12,000 2,700 22.5% Darfield New Zealand Christchurch New Zealand September 2010 7.0 22,000 0 0% February 2011 6.1 320,000 181 0.06% Tohoku, Japan March 2011 9.0 6,096,000 23,500 0.4%
The 2010 M w 6.9 Yushu (Qinghai, China) Earthquake The earthquake occurred at 23:49 on 13 April, 2010 (UTC; 07:49, 14 April 2010, local) Approximately 70% of houses collapsed. ~2,700 people were killed and over 12,000 injured Gyegu Monastery
Surface Ruptures of the Yushu Earthquake Earthquake occurred on the Yushu-Garzê-Xianshuihe Fault System Surface rupture traced for about 35 km with a peak slip of 1.8m
Interferograms for the Yushu Earthquake PALSAR acquisition 17 April 2010 Interferogram posted 20 April 2010 Envisat acquisition 26 April 2010 Sentinel 1A and 1B will eventually provide mean acquisition time of 3 days. To locate fault for the purpose of modelling, used azimuth and range offsets, phase gradients, interferometric coherence, and SPOT-5 imagery. Li et al, JGR (in review)
SPOT 5 Imagery for the Yushu Earthquake Panchromatic, 2.5 m resolution
SPOT 5 Imagery for the Yushu Earthquake
SPOT 5 Imagery for the Yushu Earthquake
SPOT 5 Imagery for the Yushu Earthquake
Modelling of the Interferograms Also used data from Envisat tracks 498A and 004D in the modelling 3 fault segments; 1 km resolution for the distributed slip model. Modelling showed rupture occurred along a length of about 75 km.
Surface Displacements for the Yushu Earthquake
Bodywave Modelling for the Yushu Earthquake Main points: Two peaks in the rate at which seismic moment was generated. Seismic moment is the main measure of earthquake excitation and is proportional to the amount of slip and the area on which it occurs. Best fit if assume SE propagation of rupture at ~2.5 km/s Centroid depth of 6 km
Moment Release Distribution and Surface Observations Li et al, JGR (in review) Majority of moment release above 10 km. Three slip patches. Test hypotheses that NW patch is due to the M w 6.1 aftershock.
Travel Time Differences between M w 6.1 Aftershock and Mainshock
Distributed Slip for the Yushu Earthquake N Yushu Geometry with propagation towards town at end of the fault is same as for the 2003 Bam earthquake. Slip deficit south of Yushu since last earthquake in 1896 is about 0.7m ( M w 6.5)
2008 and 2009 M w 6.3 Qaidam Basin Earthquakes Elliott et al, GRL (2011)
Faulting from ASTER Satellite Imagery
Faulting from ASTER Satellite Imagery
Envisat Interferograms for Qaidam Earthquakes Elliott et al, GRL (2011)
Evidence for Segmentation of a Fault with Depth Elliott et al, GRL (2011)
Coulomb Stress Changes due to the 2008 Earthquake Elliott et al, GRL (2011) Coulomb stress
Faulting and Focal Mechanisms for the Gowk Fault 1981 Mw 7.1 Sirch Earthquake 65 km long rupture Max 0.4 m surface offsets 18 km centroid Shallow dipping thrust 1998 Mw 6.8 Fandoqa Earthquake 20 km long rupture Max 3 m surface offsets 5 km centroid Strike slip with normal component Berberian et al., GJI (2001)
Conclusions Combining InSAR and seismology shows that the cause of the relatively large number of casualties from the 2010 Yushu earthquake was probably due to the propagation of the rupture along the fault towards the town, terminating there, combined with the vulnerability of the building stock. A similar scenario occurred for the Bam earthquake. In that case a strong motion accelerometer was available to provide independent evidence for the point of initiation of the rupture and hence the direction of propagation. The InSAR study of the closely separated 2008 and 2009 Qaidam earthquakes indicates segmentation of the fault with depth. Depth segmentation may occur where faults interact and the fault geometry is evolving. In such cases, just because a large earthquake has occurred at same location, it does not mean the immediate seismic hazard has gone away. The occurrence of earthquakes closely spaced in location and time may also lead to incompleteness in the palaeoseismological record.
Estimating the Length of the Fault from the Seismic Moment M 0 LW u Assume: = 3.2 10 10 Pa, W = 15 km, u = 1.5 m, M 0 = 2.5 10 19 N m (GCMT), then L = 35 km W =10 km and M 0 = 2.2 10 19 Nm L = 42 km
Seismic Rupture in the Bam Earthquake Bam The coloured area is the patch on the fault that slipped in the earthquake as determined by modelling the radar interferometry observations. The circle shows the distance to the initial point of rupture from Bam determined from the recording of an accelerometer in Bam. The location of rupture initiation can be identified (blue circle). The rupture propagated from south to north. Jackson et al, GJI (2006)
Slip Distributions 2008 2009