Constraining Mars Crust and Mantle Structure From Waveform Inversion of Fundamental and Higher Mode Surface Waves

Date: 4/25/2019

Time: 09:30 AM

Room: Vashon

One of InSight’s primary objectives is to constrain Mars crustal and mantle structure. To achieve this goal, we will use waveforms recorded by InSight’s broadband seismic instrument to invert the fundamental and, when available, the higher mode surface waves.

To measure higher mode dispersion, we developed a waveform modeling technique based on a reversible jump Markov Chain Monte Carlo method within a hierarchical Bayesian framework. It can result in two outcomes: posterior probability density functions (PPDFs) for VS models of Mars interior that represent the average structure for the source-receiver path, and fundamental and higher mode dispersion curves with uncertainties, which can be seen as secondary data that can be inverted jointly with other datasets to further constrain Mars internal structure.

To prepare for the mission, we tested our technique on a blind waveform dataset generated by members of the Mars Structure Service team. The location, source parameters, noise level, and interior model were unknown to the rest of the team. Using source parameter estimates obtained by the Mars Quake Service team, we were able to fit fundamental mode Rayleigh waves at 25-50s period and obtain VS models down to 150 km depth. We found trade-offs between VS and the source parameters, but the range of VS models and the group and phase velocities did not significantly depend on whether source parameters were included in the inversion. Our group velocities and VS models were in good agreement with those obtained by other InSight team members with other techniques. It demonstrates that, although trade-offs exist, our models and measurements together with their uncertainties are robust with respect to the source parameters. It also shows that it will be important to jointly invert for VS and source parameters to get reliable uncertainties. This will be especially important when dealing with real Mars data as we may not always get reliable source parameter estimates.

Presenting Author: Caroline Beghein

Authors