Seismic Structures in the Earth’s Inner Core Below Southeastern Asia

Abstract

Documenting seismic heterogeneities in the Earth’s inner core (IC) is important in terms of getting an insight into its history and dynamics. A valuable means for studying properties and spatial structure of such heterogeneities is provided by measurements of body waves refracted in the vicinity of the inner core boundary (ICB). Here, we investigate eastern hemisphere of the solid core by means of PKPBC–PKPDF differential travel times that sample depths from 140 to 360 km below its boundary. We study 292 polar and 133 equatorial residuals measured over the traces that probe roughly the same volume of the IC in both planes. Equatorial residuals show slight spatial variations in the sampled IC volume mostly below the level of 0.5 %, whereas polar residuals are up to three times as big, direction dependent and can exhibit higher local variations. The measurements reveal fast changes in seismic velocity within a restricted volume of the IC. We interpret the observations in terms of anisotropy and check against several anisotropy models few of which have been found capable of fitting the residuals scatter. We particularly quantify the model where a dipping discontinuity separates fully isotropic roof of the IC from its anisotropic body, whereas the depth of isotropy–anisotropy transition increases in southeast direction from 190 km below Southeastern Asia (off the coast of China) to 350 km beneath Australia. Another acceptable model cast in terms of localized anisotropic heterogeneities is valid if 33 largest polar measurements over the rays sampling a small volume below Southeastern Asia and the rest of polar data are treated separately. This model envisages almost isotropic eastern hemisphere of the IC at least down to the depth of 360 km below the ICB and constrains the anisotropic volume only to the ranges of North latitudes from 18° to 23°, East longitudes from 125° to 135° and depths exceeding 170 km. The anisotropy strength in either model is about 2 %. Further effective pursuit of the models presents challenges in terms of resolution and coverage and basically requires a significant dataset extension.