Deep structures along the boundary of a collisional belt: attenuation tomography ofPandSwaves in the Greater Caucasus

Abstract

Seismic attenuation (Q−1 ) of P and S waves shows a major discontinuity from the Russian platform to the tectonically active Greater Caucasus. Broadband records show this boundary over paths ≤4° long, as revealed by the decay of amplitude spectra from a digital seismic network flanking the Greater Caucasus. We measure attenuation from individual spectra, using a non‐linear least‐squares procedure to determine an attenuation parameter (t* ) simultaneously with source parameters at frequencies between 1 and 15 Hz. The t* measurements are then inverted for spatial variations of Q−1 , with parametrizations of varying complexity. Model variance for heterogeneous structures improves by more than 30 per cent compared with homogeneous parametrizations. Site corrections also significantly improve the fit. In these inversion results, mountainous regions exhibit Q values 2–3 times lower (QS = 775 ± 75) than do the adjacent shields (QS = 2060 ± 315), showing that the discontinuity is large. For both regions, QP is roughly equal to QS . Comparison of body‐wave to coda spectra indicates that intrinsic absorption rather than scattering dominates the Q−1 measurements, at least beneath the mountains. Hence Q−1 variations may give a reasonable proxy for temperature; if so, then temperature beneath the mountains exceeds that beneath the shield by 70°–400 °C. These temperature increases may not be high enough to generate widespread partial melting beneath the mountains, but could produce regional metamorphism and could contribute substantially to isostatic compensation of the mountains. Whatever their origin, the boundary in seismic attenuation is abrupt and large between stable craton and an adjacent mountain belt, demonstrating that Q−1 is a sensitive indicator of tectonic process.

Postseismic deformation following the 1991 Racha, Georgia, earthquake

“…The dominance of afterslip between 1991 and 1994, and GPS constraints on the upper limit of surface displacements between 1994 and 1996, require a lower crustal viscosity (25 -45 km) of at least 10 18 Pa s. Assuming a temperature gradient of 20 degrees per km, typical wet quartz flow laws [e.g., Gleason and Tullis, 1995] may be ruled out for the lower crust. A somewhat higher effective viscosity would still be consistent with low seismic velocities [Hearn and Ni, 1994] and high attenuation [Kadinski-Cade et al, 1981;Sarker and Abers, 1998;Sandvol et al, 2001] in the thickened, Caucasus lower crust. However, recent modeling studies of young continental collision zones have assumed extremely weak lower crust (for example, 25% of values given by Gleason and Tullis [1995], by Pysklywec et al [2002]) to decouple the mantle lithosphere from the upper crust.…”

Comparison of Sesimic Body Wave and Coda Wave Measures of

“…For regional paths, spectral fall-off of body waves can be inverted for attenuation along with source parameters in a parametric scheme (e.g., LEES and LINDLEY, 1994;SARKER and ABERS, 1998). The spectral fitting scheme, based on the approach of HOUGH et al (1988) and BOATWRIGHT (1978), fits a three-parameter source model H( f ) to observed amplitude spectra, where…”

“…These different approaches and descriptions of attenuation do not always lead to the same results. For example, in the Caucasus, the Q s values measured from spectral fall-off (SARKER and ABERS, 1998) are 3 times larger than Q coda measured from lapse time decay (RAUTIAN et al, 1979). Large differences are also seen in southern California between the coda Q values of 200 of SINGH and HERRMANN (1983) and body wave Q values of 800 and 1025, determined by FRANKEL et al (1990) and HOUGH et al (1988), respectively.…”

Three‐dimensional attenuation variations in southern California

“…The t* measurements are inverted for spatial variations in 1/Q following the velocity inversion scheme of Roecker [1993], as modified by Sarker and Abers [1998a]. In a continuous medium, t* measurements should be related to spa- To trace rays, we assume a simple, one-dimensional velocity for the entire region with constant velocity layers ( The a priori uncertainties in the model correspond to 100% uncertainty in the a priori model, a conservative but defensible assumption that may contribute to unreasonably large uncertainties.…”