High-Resolution Lg Q Model for Eastern Eurasia

Abstract:

We analyze large numbers of regional seismograms recorded on about 260 broadband permanent stations to measure Q of crustal Lg waves in eastern Eurasia using two-station based methods. Extracting reliable Q from observed spectra is difficult in that they are tightly coupled with source spectra, site response and geometric spreading. The two-station (TS), as well as its improvements, reversed two-station (RTS) and reversed two-event (RTE) provides the most reliable methods in estimating Q as they are able to efficiently cancel out source spectra, as well as site responses and instrument responses from spectral ratio of observed seismograms, but they all require restricted recording geometries. Our recording stations consist of backbone national seismic stations across China, regional stations in the western China, and IRIS stations around China, forming a densely and evenly distributed virtual seismic network that makes these two-station based methods desirable. After strict data quality screening, thousands of interstation path-variable Q measurements are collected. They are used to invert for Lg Q_o and η and yield tomographic models of lateral variation.

The most striking low Q_o region is in and around the Tibetan plateau where Q_o is generally below 200. The low Q_o regions also include the Qaidam and Sichuan basins, probably reflecting thick sedimentary structure on the top of crust. The regions in north and west of China, including the Tarim basin, the Ordos basin and Inner Mongolia, have Q_o around 500. The Q_o values continue increasing to 600 – 800 in the Tianshan, the Altai, and Mongolia in the north. To the eastern China, Q_o values are around 550 – 600 in the Sino-Korean craton, Yangtze craton and South China blocks.

The major distribution of η values concentrates on 0.33 – 0.47 with a mean at 0.45±0.21. This average of η is higher than that derived from previous study using TS method, but close to the mean using Lg coda or joint inversion of source spectra and Lg Q. However, such comparison is superficial because our study region is not exactly the same as the other studies, and the estimates of η usually contain higher uncertainties than Q_o. Moreover, the ratio of site responses derived from RTS method exhibits complicated patterns with frequencies, and makes it a non-ignorable term in estimating Q_o and η.