Constraining mantle heterogeneity and mantle flow using seismic and geodynamic data

In this dissertation, I have developed new 3-D models of mantle heterogeneity using seismic and geodynamic data to better understand the thermo-chemical composition of the mantle and its dynamics. First, the effect of subducting slabs in global shear wave tomography has been evaluated. By performing a synthetic test, I showed that the existence of subducting slabs could lead to serious earthquake mislocation problems. The mislocation biased travel time residuals, affected the recovery of subducting slabs in tomography, and also introduced significant artifacts into lower mantle structure. The bias was reduced if the tomography inversion used a starting model with 3-D slabs and velocity and source location were simultaneously modeled. I next developed a new tomography model by jointly inverting P and S wave seismic data. In this inversion, a 3-D thermal model of subducting slabs was used as the starting model. The new P and S models featured higher amplitude subducting slabs compared to previous global tomography results. The S to P heterogeneity ratio based on the new tomography model indicates the existence of chemical heterogeneities in the lower mantle although less than some previous studies have found. To better constrain the thermo-chemical structure, and dynamics of Earth’s mantle, I performed joint inversions using seismic and geodynamic data simultaneously. The geodynamic observations include free-air gravity, tectonic plate motions, dynamic surface topography, and excess ellipticity of the core-mantle boundary. These geodynamic observations can be related to density perturbations in the mantle assuming a known mantle viscosity model. Five different viscosity models were tested in the joint inversion. In all of these inversions, a non-thermal origin of density anomalies is required to explain the geodynamic data, though the amount of non-thermal heterogeneities varies between models. Using derived density models and their corresponding viscosity profiles, current mantle convective flow fields were also predicted. Flow fields derived using different viscosity models are similar in general. However, using density models derived from joint inversions compared to scaled seismic velocity models lead to significant differences in predicted mantle flow.