Advances in target imaging of deep Earth structure
Abstract:A new generation of global tomographic models (Lekić and Romanowicz, 2011; French et al, 2013, 2014) has emerged with the development of accurate numerical wavefield computations in a 3D earth combined with access to enhanced HPC capabilities. These models have sharpened up mantle images and unveiled relatively small scale structures that were blurred out in previous generation models. Fingerlike structures have been found at the base of the oceanic asthenosphere, and vertically oriented broad low velocity plume conduits extend throughout the lower mantle beneath those major hotspots that are located within the perimeter of the deep mantle large low shear velocity provinces (LLSVPs). While providing new insights into our understanding of mantle dynamics, the detailed morphology of these features, requires further efforts to obtain higher resolution images.
The focus of our ongoing effort is to develop advanced tomographic methods to image remote regions of the Earth at fine scales. We have developed an approach in which distant sources (located outside of the target region) are replaced by an equivalent set of local sources located at the border of the computational domain (Masson et al., 2014). A limited number of global simulations in a reference 3D earth model is then required. These simulations are computed prior to the regional inversion, while iterations of the model need to be performed only within the region of interest, potentially allowing us to include shorter periods at limited additional computational cost.
Until now, the application was limited to a distribution of receivers inside the target region. This is particularly suitable for studies of upper mantle structure in regions with dense arrays (e.g. see our companion presentation Clouzet et al., this Fall AGU). Here we present our latest development that now can include teleseismic data recorded outside the imaged region. This allows us to perform regional waveform tomography in the situation where neither earthquakes nor seismological stations are present within the region of interest, such as would be desireable for the study of a region in the deep mantle. We present benchmark tests showing how the uncertainties in the reference 3D model employed outside of the target region affects the quality of the regional tomographic images obtained.