Post-doc subject title : Joint seismic and gravity data inversion.

Roland Martin and Sébastien Chevrot
GET/Géosciences Environnement Toulouse CNRS UMR 4463 – Université Paul Sabatier – Observatoire Midi Pyrénées

Major Results

During the Post-doc of Nian Wan, a teleseismic inversion framework called Teleseisflow has been developed under Python and based on Spectral Element techniques. Density reconstruction has been improved at depth using gravity inversion constrained by seismic velocities obtained previously with Teleseisflow.


In contrast to seismic velocities, density is usually not well retrieved by seismic tomography. While gravity data are sensitive to density variations, inverting gravity data alone leads to poorly constrained solutions. Therefore, the joint inversion of gravity and seismic data is a promising way to reveal Earth’s 3D structure of density and seismic wave velocities. Seismic data considered in the joint inversion methods were so far often limited to simple observables (e.g. arrival times) and very few studies focused on full waveforms. In this work, we developped a workflow for the collaborative inversion of teleseismic waveforms and gravity anomalies. Through a series of synthetic tests using models derived from tomographic models of the western Pyrenees, we found that an efficient strategy for the joint inversion of seismic and gravity data is to first perform seismic inversion and then gravity inversion. The 3D full waveform inversion of teleseismic P waves is based upon the open source Python package TeleSeisFlows.


The gravity inversion is based upon the parallel TOMOFAST3D-x code (Martin et al. 2020) and provided good first results to retrieve densities at depth in the westernmost profile crossing Mauleon’s Basin: the 3D density anomalies are aligned along the main strike of the Pyrenees accordingly to Wang et al. (2016). Based on this idea, we built in a second step an automatic tool for full waveform inversion (FWI) to show its capability over traditional (i.e. travel time or phase related) seismic inversion in resolving heterogeneous 3D density and elastic structures. We investigated the influence of the seismic models through FWI on the results of the collaborative inversion. We show on the Figure (Wang et al. 2020), in presence of a subduction zone and a basin on top, how seismic velocities can be recovered by FWI and how density can be improved with gravity.