Improving Models of Earth Structure in Greenland

Abstract

The work presented in this thesis aims to enhance our knowledge of Earth structure and composition beneath Greenland and the surrounding region. Improving Earth structure and composition constraints are important to understand and separate the evolution of ice sheet change and solid Earth response signals. In addition, it provides valuable information about the tectonic evolution and mantle dynamics of this region. We generated a new, regional, high-resolution Rayleigh wave phase velocity model for Greenland and the surrounding region using the teleseismic two-station interferometry technique. The new phase velocity map improved the constraints on the velocity structure of the crust and the shallow upper mantle. The azimuthal anisotropy analysis also provided valuable insight into the geologic evolution and lattice-preferred orientation (LPO)/mantle flow. Significant heterogeneity was observed in our results, and we found a strong correlation between inferred geological features and isotropic and anisotropic velocities. While the thermal properties are usually inferred only from seismic data, we constrained the thermochemical properties of the shallow upper mantle beneath Greenland and surrounding regions via multiple geophysical data sets. We adopted the LitMod software, which applies the Bayesian probabilistic approach to invert for the radial thermochemical structure of the lithosphere and shallow mantle. We combined the inferred seismic data with other geophysical data sets to perform a joint inversion for temperature and composition, from which properties such as viscosity can be determined. Our results indicate large LAB-depth variations across Greenland, reflecting multiple geologic structures and thermal processes. Our LAB-depth result supports the W-E passage of the Iceland hotspot with the inferred thin lithosphere in the mid-section of Greenland. We examined the impact of lateral variations in Earth viscosity on the GIA model fits to paleo RSL and VLM data. We used the inferred regional temperature models to generate structures of the lithosphere and upper mantle in Greenland. We used a machine learning technique to emulate the 3D GIA models to examine a wide range of parameter sets (UMV-LMV-LT). We generated 50 Earth model realizations from inferred temperature profiles to quantify the uncertainty corresponding to the Earth viscosity structure. Our 3D model indicates the significant influence of laterally variable viscosity on RSL and VLM predictions compared to 1D model results. Overall, the 3D model fits the paleo RSL and contemporary VLM rates data better than the 1D model. Incorporating recent and short-term changes in ice loading with the Huy3 deglacial model improved the VLM data-model fit greatly. Uncertainty in modelled RSL and VLM rates data originating from the Earth viscosity structures was significant and could accommodate the residuals at many sites.