| In the thesis, I endeavor to explore the structure of the Earth's deep interior by analyzing seismic core phases and their related scattered energy generated by small-scale heterogeneities inside the Earth. By modeling the differential traveltimes and amplitude ratios between PKP-DF and PKP-Cdiff measured from a large dataset recorded at seismic arrays, I find that the optimum model by fitting the differential travel times has relatively low velocity at the base of the outer core as in AK135, however, the optimum model found by fitting the amplitude ratios does not exhibit this feature, and instead is closer to PREM. The discrepancy may be explained by small-scale topography on the inner core boundary (ICB) or a thin layer with relatively low Q at the base of the outer core. I also analyzed the coda waves following PKP-Cdiff to locate the lateral heterogeneity and to attempt to understand its nature. By combined modeling of the travel times, slownesses, and envelopes of the coda waves, I find a very strong heterogeneity in the lowermost mantle beneath the Amazon River in South America. To examine the strength of small-scale random heterogeneities in the mantle, I assembled a large, geographically diverse data set of PKP precursor envelopes from globally distributed international monitoring system seismic arrays. I find that the amplitudes of the precursors change from region to region, and exhibit significant variations within specific geographic regions. This may imply that the lower mantle is not perfectly mixed by mantle convection, and that compositional heterogeneities can survive in the mantle for billions of years. By modeling the globally averaged PKP precursors using a seismic phonon method, my results show that a model with random heterogeneities at a scale-length of 8 km with 0.05% r.m.s. velocity perturbation uniformly distributed throughout the lower mantle can provide a reasonable fit to the observations. On the contrary, confining the heterogeneities near the core mantle boundary (CMB) or in the D" does not yield the amplitude versus time pattern observed in our data. Furthermore, extra scattering near the CMB or in the D" is not justified. |
