GNGTS 2019 - Atti del 38° Convegno Nazionale

GNGTS 2019 S essione 3.3 741 from teleseismic earthquakes allow investigating deeper regions of the Earth (e.g. Zhou et al. , 2012). The two-station method (e.g. Meier et al. , 2004; Soomro et al. , 2016) has been succesfully employed in regional studies for retrieving earthquake-based (EQ) measurements (e.g. Darbyshire and Lebedev, 2009; Kästle et al. , 2018). This method provides interstation phase velocities of surface waves and allows a direct comparison of AN and EQ measurements. A number of studies have reported a discrepancy between AN and EQ phase velocities, with the latter being, on average, ~1% higher than the former for a wide range of periods (Yao et al. , 2006; Kästle et al. , 2016). A ~1% error in observed phase-velocity values would result in a similar error on estimates of shear velocity, based on the same data, at crustal and upper mantle depths (Boschi & Ekström, 2002); these, in turn, could be interpreted in terms of significant thermal and/or compositional anomalies (e.g. Cammarano et al. , 2003; Diaferia and Cammarano, 2017). The origin of the systematic AN-EQ discrepancy is not yet understood; differences in AN vs. EQ sensitivity kernels, overtone contamination (Soomro et al. , 2016) and off-path propagation of the earthquake between epicenter and stations have all been invoked as possible explanations. In this study, we explore in some depth the latter hypothesis, i.e. that deviations of the teleseismic Rayleigh-wave propagation paths from the first-order prediction (the great circles connecting source and receivers) might result in a significant overestimate of EQ-estimated phase velocity. Method. Earth’s lateral heterogenity results in deviations of the propagation path from the great circle connecting source and receivers (Fig. 1). This phenomenon has been widely observed at both short and long periods (see, for example, Laske, 1995) and can lead to overestimation of phase velocity. While several implementations of the one-station method account for this effect (e.g. Laske and Masters, 1996), deviations from the great-circle paths are often neglected in the two-station one (e.g. Darbyshire et al. , 2004). This is in agreement with the fact that EQ velocity estimates are systematically higher than AN ones. Fig. 1 - Sketch representing a theoretical (black curve) and observed (red curve) propagation paths of a surface wave in case of wavefront distortion due to structural heterogenities, and/or earthquake misalignment with respect to the receiveres. θ 1 and θ 2 denote the apparent arrival angles of the surface wave a stations 1 and 2, respectively.