Non-linear globalP-wave tomography by iterated linearized inversion

Abstract

Summary Recent advances in global imaging have lead to tomographic mantle models with regional scale details. To improve these models further, we have extended the usual linearized approach to traveltime tomography to non‐linear tomography. Here ‘non‐linear’ means that seismic ray bending due to inferred velocity heterogeneity is taken into account in an iterative method in which inversion steps are alternated with 3‐D ray tracing to update ray paths and traveltimes. As a starting point for our non‐linear inversion we have used the mantle model of Bijwaard et al.(1998) and ray tracing is performed following Bijwaard & Spakman (1999a). We have not attempted a full exploration of the non‐linear nature of the traveltime inverse problem. This would at the very least require tests with different starting models and the relocation of all events in each of these models. The main results are as follows. We observe no overall dramatic change in anomaly patterns, but subtle changes on the global mantle scale lead to a small increase in variance reduction and model amplitudes. These small changes together with very similar resolution estimates for the linear and non‐linear inversions do not allow us to investigate formally possible model improvements. However, expected non‐linear effects such as the focusing of structures and a baseline shift towards lower velocities indicate an improved solution, which is also more consistent with expected physics than a fully linearized inversion. Apart from that, some very strong changes occur in distinct upper mantle regions such as below Japan, Tibet, South America, Europe and Tonga–Fiji, where 3‐D ray bending effects are substantially larger than in the deeper mantle since model amplitudes fall off rapidly with depth. In the lower mantle, how‐ever, increased focusing effects can be observed that may prove important for detailed interpretations.

Crustal thickness and images of the lithospheric discontinuities in the Gibraltar arc and surrounding areas

“…Recent tomographic images from first arrivals and surface waves using data from a similar stations distribution as we use here show a high-velocity body in the upper mantle (Bezada 2013;Palomeras et al 2014) similar to the structures found in previous tomographic studies (Blanco & Spakman 1993;Calvert et al 2000;Bijwaard & Spakman 2000). This high-velocity body, possibly corresponding to oceanic lithosphere, has a slab-like structure dipping steeply towards the east.…”

“…The occurrence of intermediate depth seismicity suggests that there are deep heterogeneities in the upper mantle as confirmed by tomographic studies (e.g. Serrano et al 1998;Bijwaard & Spakman 2000;Bezada 2013). A big part of the shallow seismicity is located in southeastern Iberia and northeastern Morocco and it is related with the Trans-Alboran shear zone (Fig.…”

“…This continental Iberian slab seems to be connected to the oceanic slab revealed by regional tomographic studies underneath the Alboran Sea (Blanco & Spakman 1993;Bijwaard & Spakman 2000;García-Castellanos & Villaseñor 2011;Bezada 2013;Palomeras et al 2014). With our network, we cannot observe the continuation of the Iberian slab offshore.…”

Global Imaging of the Earth's Deep Interior: Seismic Constraints on (An)isotropy, Density and Attenuation

“…Overall, the resolution is more even over the globe, but that is not to say that it is known precisely. With a few notable exceptions (e.g., Bijwaard & Spakman, 2000;Panning & Romanowicz, 2006), all classical P-and S-velocity models are obtained by a linear inversion using expression (11.2). The most important tool to assess the amplitude and shape of the models obtained by linearized inversions is the resolution operator, provided that the influence of the data errors is small (Equation 11.3).…”

Virtual seismometers in the subsurface of the Earth from seismic interferometry

“…However, the global array of permanent seismometers that record seismic energy is confined almost exclusively to land-based sites. This limits the resolution of subsurface images, and results in relatively few local measurements from areas of great geological and tectonic interest (for example, the mid-ocean ridges and the Tibetan plateau) 1 To interrogate the Earth's subsurface at depths greater than a few kilometres, traditional seismology analyses seismic wave energy from earthquakes. Other energy recorded in seismograms, such as ambient Earth oscillation, is considered noise and is excluded from analysis.…”