Auxiliary material for Paper 2008GL034756

Rupture process of the 2007 Niigata-ken Chuetsu-oki earthquake by non-linear joint inversion of strong motion and GPS data

Antonella Cirella, Alessio Piatanesi, Elisa Tinti, and Massimo Cocco
Sismologia e Tettonofisica, Istituto Nazionale di Geofisica e Vulcanologia, Rome, Italy

Cirella, A., A. Piatanesi, E. Tinti, and M. Cocco (2008), Rupture process of the 2007 Niigata-ken Chuetsu-oki earthquake by non-linear joint inversion of strong motion and GPS data, Geophys. Res. Lett., 35, LXXXXX, doi:10.1029/2008GL034756.


Introduction

The auxiliary material contains the results of a synthetic test performed to verify the suitability of stations distribution shown in Figure S1. Despite the high number of triggered station, the azimuth coverage is limited to ~180 deg due to the off-shore location of the epicenter. Synthetic data are generated using the target rupture model displayed in Figure S2; it is obtained by assumig a regularized Yoffe function with Tacc (time of peak slip velocity) equal to 0.3 sec.Slip is concentrated only on two main asperities, A and B. Asperity A has a peak slip velocity of 3.5 m/s and a rise time of 1.5 s; asperity B has 2.5 m/s of peak slip velocity and a rise time of 2.5 s. Both asperities have rake equal to 90 deg. The rupture front propagates at 2.5 km/s, except in the upper left part of the fault, where it accelerates to nearly 3.5 km/s. We invert simultaneously the kinematic parameters (peak slip velocity,rise time and rupture time) at nodal points equally spaced along strike and dip every 3.5 km. We compute synthetic ground velocities in the frequency band 0.02/0.5 Hz and horizontal components of static displacement and we use these as our target dataset. During the inversion, the peak slip velocity is allowed to vary between 0 and 4 m/s with 0.25 m/s interval; the rise time between 1 and 4 sec at 0.25 sec step increment and the rupture time of each grid node is bounded by a rupture velocity ranging between 2 and 4 km/s. The rake angle is kept fixed. We apply a two stages nonlinear global inversion technique [see Piatanesi et al., 2007]. Figure S3 shows the inverted source model obtained by averaging a subset of the model ensemble, corresponding to those models having a cost function exceeding by 2.5% the minimum value of the cost function reached during the inversion. The comparison between target and inverted waveforms and horizontal displacements are displayed in Figure S4. The synthetic test proves that the azimuthal coverage of the selected stations is good enough to obtain reliable results.

1.2008gl034756-fs01.eps
Map of the fault geometry of the 2007 Niigata-ken Chuestu-oki, Japan earthquake. The dashed black line represents the surface projection of the fault plane adopted in this study. Black star indicates the epicenter. White triangles and inverted triangles represent K-NET (surface sensor)and KiK-net (borehole sensor) strong motion stations respectively. Black dots represent GPS stations. White dots are GPS stations not used in this study.KKNPP indicates the site of Kashiwazaki-Kariwa nuclear power plant.

2.2008gl034756-fs02.eps
Target rupture model used for synthetic test. Horizontal axis is along-strike distance measured from the epicenter.Vertical axis is down-dip distance measured from the ground surface. Top, middle and bottom panels show total slip, rise time and peak slip velocity distributions, respectively. Rupture time is shown by contour ines (in seconds); black arrows displayed in upper panel represent the slip vector. The slip patches are denoted by capital letters A and B (see Introduction for details).

3.2008gl034756-fs03.eps
Inverted  rupture model (average model from ensemble inference) from the synthetic test. Horizontal axis is along-strike distance measured from the epicenter. Vertical axis is down-dip distance measured from the ground surface.  Top, middle and bottom  panels show total slip, rise time and peak slip velocity distributions, respectively. Rupture time is shown by contour lines (in seconds); black arrows displayed in upper panel represent the slip vector. The slip patches are denoted by capital letters A and B.

4.2008gl034756-fs04.eps
(a) Comparison of synthetic ground velocities from the target rupture model (blue lines) with the inverted rupture model (red lines). Numbers with each trace are peak velocity of the synthetic line (in cm/s). Waveforms are computed in the frequency band 0.02–0.5 Hz.  (b) Comparison of horizontal GPS displacements from the target rupture model (blue arrows) with the inverted rupture model (red arrows).