A field data example of Marchenko multiple elimination

Internal multiple reflections have been widely considered as coherent noise in measured seismic data, and many approaches have been developed for their attenuation. The Marchenko multiple elimination (MME) scheme eliminates internal multiple reflections without model information or adaptive subtraction. This schemewas originally derived from coupled Marchenko equations, but it was modified to make it model independent. It filters primary reflections with their two-way traveltimes and physical amplitudes from measured seismic data. The MME scheme is applied to a deepwater field data set from the Norwegian North Sea to evaluate its success in removing internal multiple reflections. The result indicates that most internal multiple reflections are successfully removed and primary reflections masked by overlapping internal multiple reflections are recovered. INTRODUCTION Standard migration schemes map all of the reflections in the data into reflectors in the model domain if they are all primary reflections. These schemes assume that all events in the measured data are reflected only once in the subsurface. Because of this assumption, the measured seismic data should be preprocessed before being migrated by standard migration schemes. Therefore, multiple reflection elimination schemes play a crucial role for standard migration schemes. To date, several approaches have been developed to deal with multiple reflections. Some focus on free-surface-multiple reflections, whereas others focus on internal multiple reflections. Free-surface-multiple reflections can be strong enough to cause artifacts in the image from marine and land data such that much attention has been attracted from industry and academia. Free-surfacemultiple elimination (SRME) (Verschuur et al., 1992) and estimation of primaries by sparse inversion (EPSI) (van Groenestijn and Verschuur, 2009) are the two schemes that have been widely accepted as robust tools for free-surface-multiple attenuation in industry. For SRME, all orders of free-surface-multiple reflections are predicted and a minimum-energy criterion is used to subtract predicted events from the measured data. The EPSI scheme replaces the two-stage processing of SRME, prediction, and adaptive subtraction by an inversion scheme based on the full-waveform inversion approach (van Groenestijn and Verschuur, 2009). Both have achieved success on field data sets. Another strategy is to image the primary and freesurface multiple reflections simultaneously (Brown and Guitton, 2005; Whitmore et al., 2010; Verschuur and Berkhout, 2011; Wang et al., 2014, 2017; Lu et al., 2015), where free-surface multiple reflections give extended illumination of the subsurface. However, crosstalk is present in the resulting image as coherent noise. Less effort has been devoted to deal with internal multiple reflections. As pioneers, Araújo et al. (1994) derive an internal multiple attenuation scheme from the inverse scattering series (ISS). This is the first data-driven scheme that was developed by Weglein et al. (1997) and modified by Ten Kroode (2002) and Löer et al. (2016). Internal multiple elimination (IME) is a layer-related scheme extended from SRME (Berkhout and Verschuur, 1997). The IME scheme downward extrapolates shot records to a virtual surface and attenuates internal multiple reflections related to that surface. Therefore, velocity information is required for its implementation. The ISS and IME schemes have been demonstrated on numerical and field data sets (Matson et al., 1999; Verschuur and Berkhout, 2005; Luo et al., 2011). Adaptive subtraction is needed for both schemes to achieve a multiple-attenuated data set because of the approximate nature of the predicted events. Using internal multiple reflections in imaging is done via full wavefield migration, a dataconsistent closed-loop scheme (Berkhout, 2014). Davydenko and Verschuur (2018) present a field data application. Recently, Marchenko redatuming schemes have been proposed to remove internal multiple reflections and create images free from artifacts (Slob et al., 2014; Wapenaar et al., 2014). Meles et al. (2015) combine convolutional interferometry with the Marchenko scheme to give an internal multiple reflection attenuation scheme. Manuscript received by the Editor 26 May 2019; revised manuscript received 5 September 2019; published ahead of production 30 October 2019; published online 9 January 2020. Delft University of Technology, 2628 CN Delft, The Netherlands. E-mail: l.zhang-1@tudelft.nl (corresponding author); e.c.slob@tudelft.nl. © 2020 Society of Exploration Geophysicists. All rights reserved. S65 GEOPHYSICS, VOL. 85, NO. 2 (MARCH-APRIL 2020); P. S65–S70, 6 FIGS. 10.1190/GEO2019-0327.1 D ow nl oa de d 01 /0 9/ 20 to 1 45 .9 4. 67 .7 1. R ed is tr ib ut io n su bj ec t t o SE G li ce ns e or c op yr ig ht ; s ee T er m s of U se a t h ttp :// lib ra ry .s eg .o rg / Staring et al. (2018) propose to attenuate the first-order internal multiple reflections using an adaptive Marchenko double-focusing method. Model information and adaptive subtraction are required for the implementation of these schemes. Zhang and Staring (2018) modify a Marchenko multiple elimination (MME) scheme (van der Neut and Wapenaar, 2016), which in theory removes all orders of internal multiple reflections without model information or adaptive subtraction. The MME scheme has been extended to also account for transmission loss in primary reflections and free-surface multiple reflections (Zhang and Slob, 2019). Thus, free-surface and internal multiple reflections can be removed and transmission loss in primary reflections can be compensated for in one step without model information or adaptive subtraction. In this paper, the MME scheme is applied to a deepwater field data set from the Norwegian North Sea. It is the first field data example to validate its capabilities for removal of internal multiple reflections without model information or adaptive subtraction. The paper is organized as follows. In the “Theory” section, we give a brief overview of the theory of the MME scheme. The detailed theory can be found in Zhang and Staring (2018). In the “Field example” section, we apply the MME scheme to a field data set for internal multiple reflection elimination. The performance of the MME scheme is analyzed in the “Discussion” section, and we end with our conclusions.