W.-C. Yu $^{a}$, L. Wen $^{a}$ and F. Niu $^{b}$
$^{a}$ Department of Geosciences, State University of New York, Stony Brook, NY USA. $^{b}$ Department of Terrestrial Magnetism, Carnegie Institution of Washington, Washington, DC, USA
The seismic velocity structure in the Earth's outer core is important for understanding its dynamics and composition. There, however, is not a unified model in the outer core, especially at the bottom of the outer core (e.g., AK135, PREM). Neither are many studies on the lateral variation in seismic structure there. From seismological point of view, different seismic structures employed in the outer core will also directly affect our inference of seismic structures in the inner core. To study seismic velocity structure in the outer core, we have assembled a vast amount of PKP observations from the seismic data recorded in the Global Seismic Network (GSN) from 1990 to 2000 and in many dense regional seismic arrays. Our datasets include the PKP waveforms (PKiKP-PKPBdiff) observed around the caustic distance range (141 - 145 degrees), the differential PKPab-PKPbc travel times, PKPbc/PKPab amplitudes, and the PKPbc-PKPdf differential travel times at the distance range of 146 - 158 degrees. Of particular interest are the velocity gradient and its lateral variation at the bottom of the outer core. Preliminary observations and conclusions are: (1) around the caustics (141-145 degrees), the PKiKP-PKPBdiff travel time residuals in general exhibit an "East-West" hemispherical difference. Most of the residuals are larger for the observations sampling the "western" hemisphere than those sampling the "eastern" hemisphere. The observations sampling the "western" hemisphere can be explained by 1D models with either an AK135-like velocity gradient at the bottom of the outer core, or a positive P velocity gradient in an order of a few percent at the bottom of the mantle. On the other hand, PREM fits the data better for those sampling the "eastern" hemisphere. (2) at the distance range of 147-158 degrees, for the data sampling the "western" hemisphere, the PKPab-PKPbc differential travel times, PKPbc/PKPab amplitudes, and PKPbc-PKPdf differential travel times, can be better fit by a model with an AK135-like velocity gradient at the bottom of the outer core. The PKPab-PKPbc differential travel times exclude the alternative 1D models appealing to a positive P velocity gradient in an order of a few percent at the bottom of the mantle beneath the "western" hemisphere to explain the PKPBdiff-PKiKP waveform data around the caustics. For the observations sampling the "eastern" hemisphere, PREM explains the observed PKPab-PKPbc differential travel times and amplitudes well, and invoking an AK135-like velocity gradient at the bottom of the outer core increases the misfits to the data sampling this "hemisphere". In summary, there appears that the observations sampling the "western" hemisphere can be better explained by an AK-135 like model in the outer core, while those sampling the "eastern" hemisphere can be just well explained by PREM and require no AK135-like velocity gradient at the bottom of the outer core. The observations also show scatter, in both differential travel time and amplitude. For these observations, we perform a preliminary study on the effects of the mantle structures based on a tomographic model by Grand. While the tomographic model we used is capable of producing the magnitude of the observed travel time scatter depending on the vs-vp scaling assumed, it cannot explain most of the observations. Tomographic corrections with large vs-vp scalings (e.g., 0.5) make the observations sampling the "eastern" hemisphere even more scattered in the both distance ranges. Further investigation is still needed to clarify the contributions from the seismic structures in the mantle.