The resolving power of Antarctic GNSS observations

Understanding Antarctica's past and future ice dynamics critically depends on our ability to accurately model the solid Earth's response to surface mass loading. This response strongly depends on the assumed mantle rheology and ice loading history, both of which are highly uncertain for Antarctica—a region of complex 3D Earth structure and limited observational constraints. Local geodetic observations such as GNSS may provide invaluable constraints on these uncertain parameters. However, to rigorously invert GNSS data, one must quantify their intrinsic resolving power, or sensitivity, to both the 3-D mantle viscosity field and ice mass history. The large (effectively infinite) parameter space associated with these inputs makes the usual computation of partial derivatives unfeasible for this purpose.

Here, we use adjoint modelling to compute sensitivity kernels of Antarctic GNSS observations. Adjoint modelling has the major advantage of requiring only two calculations (the so-called forward and adjoint problems) to generate these kernels. We present a suite of such kernels, and use them to quantify non-uniqueness associated with previous inferences of mantle viscosity based on GNSS observations. These studies were largely focused on radial crustal motions, and we explore the improvement in resolving power obtainable by incorporating horizontal motions. Finally, we also discuss the unique sensitivity of radial and horizontal motions to variations in the adopted ice history. Our results provide for an important first step toward rigorous inversions of the GNSS observations as a tool for monitoring the stability of Antarctica's ice sheets.