A new technique to construct crustal 3-D Vs models: Implementation of Ray Tracing, Model Parameterization and Inversion
Abstract
We develop a new tool in which we exploit P-to-S converted waves to construct a fully 3-D shear-wave velocity model of the crust. This technique requires a dense seismological network to map the usually less-studied S-wave velocities.
For the forward model, we implement an accurate ray-propagator which respects Snell's law in 3-D at any interface geometry. First, we employ an existing ray-shooting tool (Knapmeyer, 2004) to calculate P-ray geometry in a global velocity model (iasp91) to arrive at the station. Then, starting from the piercing point at the local Moho, we propagate an S-wave to the surface: this reaches the surface several km away from the station. We therefore adjust the ray parameter to make the corresponding crustal S-waves arrive at the station. A synthetic precision test with flat Moho and constant crustal velocity shows that the converted waves reach the station within 10 m. A real case using local velocity structure and complex Moho geometry yields a mean distance of 150 m (median 40 m) from the station. We parameterize the model grid using square cells in X and Y directions and define its size based on the ray coverage map of actual data. For a 20-year dataset at 120 stations recording more than 300 000 traces, the mesh size is typically 50x50 km. For the depth of the interfaces, we consider a multi-layer crustal model with vertically flexible nodes. For each layer we define an S-wave velocity at the top and at the bottom: this allows to accommodate both velocity gradients within and velocity jumps between layers. The individual velocity profile along each trace is extracted from the 3-D initial model and the velocity model is updated during the inversion. We envisage to manage the inversion by the stochastic Neighbourhood Algorithm (NA, Sambridge, 1999), looking at the ensemble of models that sample the good data-fitting regions of a multidimensional parameter space. We plan to test our approach first with a 1-layer Vs model and then introduce intra-crustal discontinuities (e.g. Conrad). Our first focus region is the Central Alps, where a well-defined Moho map (Spada et al., 2013) and a high-resolution P-wave velocity-model (Diehl et al., 2009) are available. We plan to extend our study to the entire Alpine domain in frame of the AlpArray project (Hetényi et al., 2018).- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2018
- Bibcode:
- 2018AGUFM.S31D0535C
- Keywords:
-
- 7260 Theory;
- SEISMOLOGYDE: 7270 Tomography;
- SEISMOLOGYDE: 7290 Computational seismology;
- SEISMOLOGY