Using laboratory experiments to develop and test new Marchenko and imaging methods

Abstract

The Marchenko redatuming method estimates surface-tosubsurface Green's functions. It has been employed to diminish the effects of multiples in seismic data. Several such methods rely on an absolute scaling of the data; this is usually considered to be known in synthetic experiments, or is estimated using heuristic methods in real data. Here, we show using real ultrasonic laboratory data that the most common of these methods may be ill suited to the task, and that reliable ways to estimate scaling remains unavailable. Marchenko methods which rely on adaptive subtraction may therefore be more appropriate. We present two adaptive Marchenko methods: one is an extension of a current adaptive method, and the other is an adaptive implementation of a non-adaptive method. Our results show that Marchenko methods improve imaging compared to reverse-time migration, but less so than expected. This reveals that some Marchenko assumptions were violated in our experiment and likely are also in seismic data, showing that laboratory experiments contribute critical information to the development and testing of Marchenko-based methods.

Source wavelet correction for practical Marchenko imaging: A sub‐salt field‐data example from the Gulf of Mexico

“…; da Costa Filho et al . ). In the following, we demonstrate how we can use the upgoing focusing function to invert not only for the overall scale of the source wavelet but also to correct for an incorrectly estimated amplitude spectrum, i.e.…”

“…An accurate Green's function retrieval based on focusing functions, however, requires accurate knowledge of the scale of the source wavelet and its amplitude spectrum. Deconvolving an inaccurately scaled source wavelet from the acquired surface data prior to the Marchenko algorithm can lead to an instable solution of equations (1) to (4) (Ravasi et al 2015) or artefacts in the obtained Green's functions (van der Neut et al 2015b;Jia et al 2018;da Costa Filho et al 2018b). In the following, we demonstrate how we can use the upgoing focusing function f − 1 to invert not only for the overall scale of the source wavelet but also to correct for an incorrectly estimated amplitude spectrum, i.e.…”

“…While the latter approach has been successful in several pre‐salt areas, adaptively subtracting internal multiples may harm primaries (da Costa Filho et al . ). Additionally, the adaptive Marchenko implementation has the significant drawback that only the upgoing Marchenko wavefields can be obtained accurately by adaptive subtraction.…”

“…In field-data applications of the Marchenko method, in particular the latter requirement can lead to reduced quality of images because the correctly scaled source wavelet is often not known. Deconvolving an incorrectly scaled source wavelet prior to Marchenko redatuming and imaging leads to artefacts in the redatumed Green's functions (van der Neut et al 2015b;da Costa Filho et al 2018b) and subsequent images (Jia et al 2018;Staring et al 2018). To date, several different approaches have been used to estimate the correct scale of the source wavelet or to overcome this requirement.…”

“…In an alternative approach, van der Neut and Wapenaar (2016) and Staring et al (2017) subtract Marchenkopredicted overburden-related multiples adaptively. While the latter approach has been successful in several pre-salt areas, adaptively subtracting internal multiples may harm primaries (da Costa Filho et al 2018b). Additionally, the adaptive Marchenko implementation has the significant drawback that only the upgoing Marchenko wavefields can be obtained accurately by adaptive subtraction.…”