Joint inversion of surface dispersion and P-wave tomography for temperature and lithology: Methodology and case study in the Eastern Alps

Seismic observations are usually inverted for seismic velocity structure (Vp and Vs). By using Markov-chain Monte Carlo (McMC) inversion (i.e. solving the forward model and comparing it to the data many times and exploring the virtual space of possible solutions), it is possible to directly invert for rock type as a categorical variable (rather than its constituent minerals and parameters). McMC also manages any non-linear relationships that rock constituents and parameters may have with velocity that could result in non-convergence of a linear inversion. 

We have developed a theoretical and software framework to perform an inversion of surface wave dispersion and P-wave tomography directly to crustal rock type and, by fixing surface and lithosphere-asthenosphere boundary temperature, temperature gradient. 

This approach constrains the inversion to petrologically valid models rather than the larger space of seismologically valid models. Additionally, knowledge of rock type helps to facilitate interpretation by inferring, from the seismic observations, the various lithologies in an unbiased manner. In the crust, the forward model for inferring Vp, Vs, and density from rock type is slow. To overcome this, we calculate a look-up table of seismic properties for crustal rocks as a function of pressure and temperature. 

We demonstrate the method with a synthetic test that shows that velocity and the silica content (mafic-felsic) of crustal layers can be reliably recovered as well as some indication of the main constituent minerals. Temperature and exact mineral assemblage are poorly constrained. A test transect of seven stations in the Eastern Alps indicate a mainly felsic upper crust with a more intermediate lower crust. Temperature, although not well constrained, shows an increase where historic magmatic activity between two major tectonic faults has been previously inferred.