Surface wave propagation in central Asia: Observations of scattering and multipathing with the Kyrgyzstan broadband array

Abstract

We studied the propagation of Rayleigh waves at regional distances in central Asia using a combination of array processing techniques and surface wave analysis. We present results from the detailed analysis of three representative events recorded by a 10‐station, broadband network that has been running in the central Asian country of Kyrgyzstan since 1991: an Ms = 5.1 event near Ashkhabad, Turkmenistan; an Ms = 5.8 event in south central Tibet; and the October 7, 1994, nuclear explosion at Lop Nor. We find there is a remarkable difference in the propagation characteristics of surface waves along these three paths. The path from the event in Turkmenistan is simple and is well approximated by propagation through a laterally homogeneous medium. Array processing shows the entire Rayleigh wave train stacks coherently and arrives from an azimuth close to that predicted by a great circle path. Furthermore, estimates of dispersion curves and fundamental mode signals determined for individual stations show little variation across the array. The Tibet and Lop Nor paths are completely different. We find strong evidence for complicated multipathing and scattering effects along both of these paths. We observe a three‐stage pattern in the Tibet case: (1) the early, lowest‐frequency part of the Rayleigh wave packet arrives as a coherent signal from close to the great circle path azimuth; (2) this is overpowered in the period range around 20 s by a strong multipath signal that propagates across the array from a much more southerly azimuth; and (3) periods below 20 s rapidly become incoherent, and the signal does not have a well‐defined direction of propagation. The Lop Nor path shows similar complexity. On this path there is little dispersion for measurable periods greater than 10 s, so the low‐frequency energy arrives in an Airy phase. The Airy phase stacks somewhat coherently (it stacks, but significant power is lost in the best beam), and slowness analysis shows it arrives from an azimuth 14° north of the great circle path azimuth. Like the Tibet event, the direction of propagation of the shorter periods cannot be resolved. These results are consistent with crustal structure in central Asia. The multipathing and scattering from the Tibet and Lop Nor events are not surprising given these waves propagate across highly variable crust and high topography resulting from the collision of India with Eurasia. The relative simplicity of the Turkmenistan event path reinforces this as almost all of that path traverses the relatively undeformed Turan platform of Uzbekistan.