Source parameters of the 1 October 1995 Dinar (Turkey) earthquake from SAR interferometry and seismic bodywave modelling

The 1 October 1995, M-s = 6.1 Dinar earthquake ruptured a 10 km section of the NW-SE Dinar-Civril fault. There are discrepancies between the published source parameters from seismic data, with seismic moments in disagreement by over a factor of two. We use both SAR interferometry and seismic bodywave modelling to determine earthquake source parameters. An interferogram generated from ERS-1/2 SAR imagery spanning the event, and separated by 5 months, is used to derive source parameters by a downhill simplex inversion with multiple Monte-Carlo restarts. We model the displacements in the satellite line of sight, initially using uniform slip on a rectangular dislocation in an elastic half-space. The resultant model fault plane agrees in strike and location with the observed surface break, but systematic residuals exist in the line-of-sight deformation field, resulting in a r.m.s. residual of 20 mm in the interferogram. The residuals are reduced if the depth distribution of slip is allowed to vary spatially in four segments along a continuous fault plane. Our best-fitting solution? with a r.m.s. misfit of 8 mm, reveals two distinct areas of slip on the fault plane (strike 145 degrees, dip 49 degrees, rake 270 degrees): a main rupture slipping by 1.44 m between depths of 1 and 8 km, becoming deeper to the SE and matching the observed surface rupture, and an along-strike continuation to the NW of the same fault plane, but between depths of 8 and 13 km and not associated with a surface break. The total geodetic moment (4.5 +/- 0.1 x 10(18) Nm) is more than twice as large as published seismic moments based on the inversion of P-waveforms alone, but close to the Harvard CMT moment (4.7 x 10(18) Nm). We use SH-waveforms, in addition to the P-waves used previously, to determine an alternative seismic source mechanism. SH-waves constrain the depth to be shallower than solutions based on P-waves alone, agreeing with the depths from the interferometric inversion and resulting in a larger moment (3.1 +/- 0.4 x 10(18) Nm) than the previous bodywave estimates (2, 2.1 x 10(18) Nm). The CMT moment reduces in magnitude to a similar size (3.3 x 10(18) Nm) if the centroid depth and fault dip are constrained to the values determined from bodywave modelling and interferometry. Thus, the geodetic moment is 40% bigger than the moment determined from seismology. (C) 1999 Elsevier Science B.V. All rights reserved.