← Back to Gallery

First Steps Towards Imaging the Antarctic’s 3D Viscosity Structure Using GPS Observations

A critical challenge of Glacial Isostatic Adjustment (GIA) modeling is seeking a combination of mantle viscosity and ice history that satisfies existing GIA observations. Current approaches vary in their complexity, but all fundamentally rely on solving the forward problem many hundreds of times. This is practical if Earth’s viscosity varies with only depth, but quickly becomes unfeasible if lateral variations in viscosity are considered. In order to limit the parameter space and ensure a reasonable mantle structure, researchers are inferring 3D mantle viscosity from other geophysical models, such as seismic tomography. However, the resulting 3D viscosity inferences may not improve the GIA model’s fit to the data because of shortcomings associated with the geophysical models, the mapping to viscosity, the rheological model, and the ice history.

We take a different approach, inspired by seismic tomography, and use gradient-based optimization along with the adjoint method to invert solid Earth deformation measured by Antarctic GPS stations to directly image 3D mantle viscosity. This iterative inversion ensures that the final GIA model is consistent with the observations. Here we focus on the setup and methodology of the inversion. First, for our starting viscosity model we adopt a 3D viscosity inference based on the shear wave speeds of the continental and global tomography models ANT-20 and GLAD-M25, respectively. These inferences are obtained using the inverse calibration scheme of Richards et al. (2020), but for ANT-20 we explore the use of additional calibration constraints to further refine the anelastic calibration parameters and also obtain a 3D inference of Q^-1. Next, we examine the viscosity sensitivity kernels for solid Earth deformation and demonstrate how differential measurements can be used to restrict the spatial extent of the kernel, allowing us to update only the Antarctic mantle and minimize the influence of errors. Finally, results from a synthetic inversion using the potential Antarctic GPS dataset will be shown.

Session: Translating Seismic Imaging into Geodynamic Understanding - I

Type: Oral

Room: Tubughnenq’ 4

Date: 5/1/2024

Presentation Time: 05:15 PM (local time)

Presenting Author: Andrew Lloyd

Student Presenter: No

Additional Authors