Bulletin of the Seismological Society of America, Vol. 84, No. 1, pp. 119-132, February 1994 Analysis of Short-Period Regional Phase Path Effects Associated with Topography in Eurasia by Tianrun Zhang and Thorne Lay Abstract Propagation of regional seismic phases is highly dependent on path effects, but we have a limited understanding of these effects and no general procedure for accounting for path variation influence on regional waveforms. Thus, there is strong regional variability in the effectiveness of regional wave discriminants used to identify small earthquakes and nuclear explosions. Moti- vated by many observations of correlation between surface geology and regional phase behavior, we empirically explore the relationship between short-period regional and upper mantle distance signal energy and statistics of topography along different travel paths using data for underground nuclear explosions at Semipalatinsk, Kazakhstan. We find strong linear correlations of the logarithmic rms amplitude ratio Sn/Lg (and to a lesser extent, P/Lg) with mean altitude, rms roughness, rms slope, and skewness of topography along the paths to receivers in Eurasia. This indicates that energy partitioning in the regional wave field is controlled by wave-guide structure and attenuation variations that are manifested in surface topography. This suggests that it is feasible in many cases to directly calibrate regional discriminants for path effects in terms of observable surface topography, as a surrogate for overall path properties. The relationships also help to understand the nature of regional phase propagation. Introduction Understanding short-period regional seismic wave propagation is of particular interest for nuclear explosion seismology, in which concerns have shifted to prolifer- ation and very low threshold treaty monitoring. To achieve more accurate small explosion identification, it is nec- essary to monitor small events from close distances. Be- cause of this, monitoring methods based on regional phases have become of primary importance (e.g., Bland- ford, 1982; Bennett and Murphy, 1986; Taylor et al., 1989). The crust comprises a low-velocity strongly scat- tering layer that forms the primary wave guide in which regional seismic waves propagate. Pn and Sn propagate just below the Moho discontinuity in the mantle lid, and Pg and Lg propagate within the crust. The properties of the lid and wave guide vary along almost every path and have strong influence on the propagation of these re- gional phases (e.g., Kennett et al., 1990). Many general observations of regional phase prop- agation efficiency associated with surface structure along the path have been documented since Press and Ewing (1952) reported the absence of Lg for paths that cross deep ocean basins. Examples include the absence of Lg for paths crossing the Tibetan plateau (Ruzaikin et al., 1977) and the strong Sn attenuation in the northernmost portion of the Iranian and Turkish plateaus and between the Black and Caspian Seas (Kadinsky-Cade et al., 1981). Although some general rules for propagation efficiency have been deduced, no systematic correction procedure associated with surface tectonic structure has been de- veloped. Spectral variations in Lg and Lg coda waves have been used to map crustal attenuation heterogeneity (e.g., Campillo, 1987; Campillo et al., 1993; Xie and Mitch- ell, 1990, 1991), and such models can be used to correct Lg amplitudes (e.g., Nuttli, 1986). However, blockage and signal conversion effects cannot be properly ac- counted for by such attenuation models, and only gen- eral associations of attenuation with surface structure have been derived. Rather than qualitative characterizations of the "ex- istence" or "non-existence" of one or more particular phases, studies of propagation characteristics along dif- ferent paths by looking at the relative amplitudes of Sn and Lg (Gregersen, 1984; Kennett, 1985a, 1985b; Baum- gardt, 1990) or P and S (Molnar and Oliver, 1969) pro- vide more quantitative measures of energy partitioning in the regional wave field. Using amplitude ratios min- imizes to a great extent the common effects for the phases being compared, especially the source effects. There- fore, the path effect can be partially isolated in an em- pirical fashion. In addition, some ratios, such as Pn/S, 119