Surface wave tomography in the European and Mediterranean region

Abstract

The broad European and Mediterranean region is characterized by an extremely com- plex tectonic setting, driven by the ma jor convergence between Eurasian and African plates. A detailed model of the upper mantle in this region is fundamental to improve our understanding of its geodynamical evolution. Seismic tomography can help to ad- dress this problem modeling seismic speed anomalies, that can be related to different tectonic features, such as continental roots, rifting areas, magmatic provinces, plumes or subducting slabs. Due to high seismicity rates and dense seismograph coverage, this region has been the sub ject of many tomographic studies from regional to local scale. Traveltime high resolution models of P-wave speed anomalies [Spakman et al., 1993; Piromal lo and Morel li , 2003] have illuminated the deep structure of the mantle, but at shallow depth they often suffer from uneven ray coverage, being strongly dependent on station and epicenter distribution. Regional S-wave velocity models have been re- trieved from the analysis of surface wave group or phase velocity [Ritzwol ler and Levshin , 1998; Vil lase˜ nor et al., 2001], from waveform inversion of surface waves [Marquering and Snieder , 1996] or both surface and body waves [Marone et al., 2004]. However, the non-optimal distribution of observatories and seismic sources has affected regional to- mographic models. Global models derived from surface wave data image the large-scale structures of the region, but their resolution is insufficient to describe its complexity [Shapiro and Ritzwol ler , 2002; Boschi and Ekstr¨ om , 2002; Ritsema et al., 1999; Zhou et al., 2006; Trampert and Woodhouse , 1995]. Global models with finer parameteriza- tion on Mediterranean [Boschi et al., 2004] have been proposed and recent modeling of surface waves from ambient noise gave new insights into the shallowest European upper mantle [Yang et al., 2006]. The increased availability of high quality seismic records from permanent observato- ries and from the recent temporary deployment RETREAT in the Northern Apennines gave us the opportunity to exploit new data, that can provide new and finer constraints to the tomographic problem. We present in this thesis a new surface wave tomography study, aimed at exploiting the high sensitivity of these waves to shallow structure and their wide spatial coverage in the complex sources-stations distribution of the European and Mediterranean domain. The inverse problem of obtaining a VS three-dimensional model from analysis of surface wave can be solved in different ways. [Marone et al., 2004] use the partitioned waveform inversion of [Van der Lee and Nolet , 1997], where the 1-D average S-velocity structure along each path is first determined by non-linear waveform fitting, and in a second step the 1-D path averaged structures are combined in a damped least-squares linear inversion for a 3-D S-velocity model. [Shapiro and Ritzwol ler , 2002] in a first step estimate 2-D dispersion maps with a linear tomographic inversion of path average fundamental mode group and phase velocities, and afterwards apply a Monte-Carlo method to perform the non-linear inversion of the dispersion curves at each geographical point and retrieve the 3-D shear-velocity model. [Boschi and Ekstr¨om , 2002] carry out a single non-linear inversion of phase anomaly measurements making use of JWKB ray-theoretical sensitivity kernels computed in a reference 3-D model. [Zhou et al., 2006] invert long period fundamental mode phase delays with finite-frequency 3-D Born approximation kernels, calculated in a reference 1-D model. We will proceed with a 2 steps scheme, first inverting group path averaged speeds for a regionalized group velocity model assuming a linearized ray theoretical wave propagation. In a second step, we will use the group velocity maps as data to perform a non-linear iterative depth inversion for the local 1-D structure, accounting for the lateral variations of the Crust. This thesis presents a new model along with a discussion of the robustness and resolution of its main features. We will firstly present the group velocity measurement technique and an analysis of measurement errors (Chapter 2), then we will introduce the linear inversion of the regional data starting from a reference global model, with an accurate examination of the implication of different regularization constraints (Chapter 3). Group velocity maps will then be shown and discussed. Subsequently we will invert the group velocity for the Vs structure of upper mantle (Chapter 4). Our resulting 3-D radially anisotropic model will be discussed in detail and compared with other published global and regional models.