Combined InSAR, Pixel Tracking, GPS, and Seismic Analysis for Slip Evolution Models of 2011 M7.1 Van, Turkey Earthquake
Fielding, Eric1; Lundgren, Paul2; Polet, Jascha3; Taymaz, Tuncay4; Owen, Susan2; Simons, Mark5; Motagh, Mahdi6; Bathke, Hannes6; Haghshenas, Mahmud6; Yun, Sang-Ho2; Yolsal-«evikbilen, Seda4
1Jet Propulsion Laboratory, California Institute of Technology, Pasadena, CA, UNITED STATES; 2JPL, Caltech, UNITED STATES; 3California State Polytechnic University Pomona, UNITED STATES; 4Istanbul Technical University, TURKEY; 5Caltech, UNITED STATES; 6GFZ, GERMANY
A large Mw 7.1 earthquake struck the area north of the city of Van in eastern Turkey on 23 October 2011. The main shock epicenter and most of the aftershocks were located south of the eastern arm of Lake Van, and extensive damage was reported in the city of Van, the city of Ercis, and many smaller towns nearby. The region is near the middle of the elevated Turkish-Iranian Plateau in the collision zone between the Arabian and Eurasian Plates. Previous geodetic studies show ongoing N-S compressional strain in the area. We analyze coseismic InSAR from two Envisat ASAR descending-track pairs and one COSMO-SkyMed (CSK) descending-track pair, along with coseismic pixel offset tracking (sub-pixel correlation) along-track displacements from the CSK pair and coseismic 3-D displacements from continuous GPS station data posted to the Geohazards Supersite.
We use SAR and GPS data to constrain a Bayesian inference method with Markov-chain Monte Carlo sampling to resolve the fault rupture location, size, and final slip distribution with posterior probability distribution estimates for the geometric parameters using a single planar fault model. The along-track (roughly N-S) displacements from the CSK data require that the main rupture dips to the north. The geodetic inversion, assuming a single planar fault plane, finds the most probable strike and dip are 259+2/-1° and 43+1/-2°, respectively. These estimates are similar to moment tensor solutions and body waveform point-source solutions. A joint inversion of the geodetic data with the seismic waveforms (body and surface waves) was then performed with the finite fault geometry fixed to the values from the geodetic inversion to estimate the kinematic fault slip evolution during the earthquake. This combined analysis shows the fault rupture proceeded upward from the hypocenter near 18 km depth with little slip shallower than 8 km depth and most of the moment release in the first 10 s. All slip was completed in 14 s. We estimate that peak slip was between 5 and 15 m at a depth close to 15 km and the area of major slip extended about 25 km along strike. Our analysis indicates that the 2011 mainshock primarily ruptured the middle crust and the shallowest part had much less slip. The slip was primarily thrust with minor strike-slip motion. Several faults, likely thrust faults, near the surface had minor amounts of slip during or shortly after the 23 October mainshock, and postseismic InSAR shows some of these faults continued to slip for several weeks after the mainshock.
Postseismic SAR images were acquired for several months after the earthquake by TerraSAR-X and COSMO-SkyMed on both ascending and descending tracks. InSAR coherence is problematic due to snow cover at the higher elevations in the winter months. Preliminary time-series analysis of the TerraSAR-X using StaMPS suggests that one or more deeper faults slipped in the first two months, in addition to the small amounts of slip on shallow thrust faults.