Optimizing measurement geometry for seismic near-surface full waveform inversion
Open access
Datum
2017-09Typ
- Journal Article
Abstract
Full waveform inversion (FWI) is an increasingly popular tool for analysing seismic data. Current practise is to record seismic data sets that are suitable for reflection processing, that is, a very dense spatial sampling and a high fold are required. Using tools from optimized experimental design (ED), we demonstrate that such a dense sampling is not necessary for FWI purposes. With a simple noise-free acoustic example, we show that only a few suitably selected source positions are required for computing high-quality images. A second, more extensive study includes elastic FWI with noise-contaminated data and free-surface boundary conditions on a typical near-surface setup, where surface waves play a crucial role. The study reveals that it is sufficient to employ a receiver spacing in the order of the minimum shear wavelength expected. Furthermore, we show that horizontally oriented sources and multicomponent receivers are the preferred option for 2-D elastic FWI, and we found that with a small amount of carefully selected source positions, similarly good results can be achieved, as if as many sources as receivers would have been employed. For the sake of simplicity, we assume in our simulations that the full data information content is available, but data pre-processing and the presence of coloured noise may impose restrictions. Our ED procedure requires an a priori subsurface model as input, but tests indicate that a relatively crude approximation to the true model is adequate. A further pre-requisite of our ED algorithm is that a suitable inversion strategy exists that accounts for the non-linearity of the FWI problem. Here, we assume that such a strategy is available. For the sake of simplicity, we consider only 2-D FWI experiments in this study, but our ED algorithm is sufficiently general and flexible, such that it can be adapted to other configurations, such as crosshole, vertical seismic profiling or 3-D surface setups, also including larger scale exploration experiments. It also offers interesting possibilities for analysing existing large-scale data sets that are too large to be inverted. With our methodology, it is possible to extract a small (and thus invertible) subset that offers similar information content as the full data set. Mehr anzeigen
Persistenter Link
https://doi.org/10.3929/ethz-b-000190403Publikationsstatus
publishedExterne Links
Zeitschrift / Serie
Geophysical Journal InternationalBand
Seiten / Artikelnummer
Verlag
Oxford University PressThema
Inverse theory; Waveform inversion; Seismic tomographyOrganisationseinheit
03953 - Robertsson, Johan / Robertsson, Johan
Anmerkungen
It was possible to publish this article open access thanks to a Swiss National Licence with the publisher.