Source and Propagation Effects of Rayleigh Waves From

A reference point equalization method has been developed which enables 'the separation of source and propagation effects of surface waves. The method works on seismic events located in a small source region, which allows us to assume that all events share the same path effects to a given receiver. Two important steps in the method are initialization and iteration. Initialization obtains the first "reference events" in order to compute initial estimates of phase velocity and attenuation coefficient. Iteration simultaneously refines the propagation parameters and determines the source parameters of new earthquakes in the vicinity of the reference point. This method was applied to earthquakes in the Pamir mountains, Central Asia -(reference point: 39.58N, 73.55E). In the initialization step, the method of Weidner and Aki (1973) was applied to obtain focal depths and revise fault plane parameters of the first two earthquakes. The residuals obtained from fitting the observed amplitude ratios and phase differences indicate that the crust and upper mantle in Central Asia is more laterally heterogeneous than near the ocean rifts, the site of Weidner and Aki's e periment. We computed heterogenity quotients of .46 vs .80 x 10 Napier2/km and .93 vs 2.4 x l0-4radian 2/km for ocean versus continent as a measure of the increasing scatter in amplitude and phase of 20-60sec Rayleigh waves due to lateral heterogeneities. To determine source parameters in the iteration, we applied the linear moment tensor inversion on Rayleigh wave complex source spectra. The presence of random additive errors in the complex spectra does not pose difficulties for recovering reliable source parameters using the linear inversion method. However, amplitude magnification errors in the complex spectra will lead to over-estimation of the moment tensor elements and phase incoherency will lead to under-estimation. In applying this method to our dataset, it was necessary to modify the straight least squares inversion method because of its sensitivity to even a few bad data points. The residuals obtained from the repeated application of the moment tensor inversion over trial focal depth showed two minima: one minima at depths less than 20km and the other at depths greater than 70km. The values of the residuals at these minima were close enough to cast doubt on the determination of focal depth. One way to resolve this ambiguity is to compare the geometry of the moment tensor obtained for shallow and deep focus inversions with observed P-wave polarities. The focal depths of eight out of the nine events in our dataset were found to be shallow, between 5-15km. Their principal compressive stress axes are aligned north-south and nearly horizontal, consistent with the interpretation of plate tectonics in Central Asia. The moment tensor inversion generally gives three-couple force systems having significant non-zero intermediate component. However, in light of errors in our data and the similarities of the double couple models to the three-couple models, it can not be established convincingly that these results are caused by departures of source from the double couple model. We interpret the propagation parameters in terms of lateral variation of phase velocity and Q on the Eurasian continent. A phase velocity regionalization is proposed involving five continental provinces: Indian Shield, Northern Platforms, Coastal Plains, Tectonic and Plateau. Phase velocities on the Indian Shield are 20% higher than velocities on the Plateau province at 26sec period and 5% higher at 90sec period. Stable provinces in Eurasia are found to have significantly higher phase velocity than tectonic provinces out to 150sec period. Interpretations of the phase velocities on the Indian Shield show a lithosphere thickness of about 120km, considerably thicker than the lithosphere under the Northern Platforms (n75km). The lowest shear velocity in the upper mantle is found under the Tectonic province with a value about 4.3km/sec over depths between 83-240km. The crustal thickness of the Plateau province is as great as 70km provided that shear velocity in the lower half of the crust is about 3.8km/sec. The upper mantle structure under this province is very similar to James' (1971) for the Andes mountains region. Surface wave amplitudes on the Eurasian continent are strongly affected by horizontal refraction as well as intrinsic Q of the medium. Average Rayleigh wave Q has very different character on paths over northern platforms and tectonic provinces east and west of the reference point. Under platforms Q is found to increase with depth from a Q between 200-300 in the crust to