Suppression of surface-related and internal multiples is an outstanding challenge in seismic data processing. The former is particularly difficult in shallow water, whereas the latter is problematic for targets buried under complex, highly scattering overburdens. We propose a two-step, amplitude- and phase-preserving, inversion-based workflow, which addresses these problems. We apply Robust Estimation of Primaries by Sparse Inversion (R-EPSI) to suppress the surface-related multiples and solve for the source wavelet. A significant advantage of the inversion approach of the R-EPSI method is that it does not rely on an adaptive subtraction step that typically limits other de-multiple methods such as SRME. The resulting Green's function is used as input to a Marchenko equation-based approach to predict the complex interference pattern of all overburden-generated internal multiples at once, without a priori subsurface information. In theory, the interbed multiples can be predicted with correct amplitude and phase and, again, no adaptive filters are required. We illustrate this workflow by applying it on an Arabian Gulf field data example. It is crucial that all pre-processing steps are performed in an amplitude preserving way to restrict any impact on the accuracy of the multiple prediction. In practice, some minor inaccuracies in the processing flow may end up as prediction errors that need to be corrected for. Hence, we decided that the use of conservative adaptive filters is necessary to obtain the best results after interbed multiple removal. The obtained results show promising suppression of both surface-related and interbed multiples.

No abstract

Over the last few years, machine learning has become more and more a topic of interest in the seismic industry. In seismic interpretation like fault/salt dome detection (Amin et al. 2015, Guitton et al. 2017) and velocity picking (Smith 2017), there already have been successful implementations for some years now. Recently, machine learning was introduced in seismic processing algorithms like denoising, regularization and tomography (Araya-Polo et al. 2018) as well. In this abstract a deblending algorithm is proposed that utilizes supervised machine learning algorithms. The method combines the two main functionalities of supervised learning, classification and regression to achieve maximum control on the deblending process. First, blended acquisition and conventional deblending methods are discussed, followed by an introduction to machine learning algorithms and how these machine learning methods can contribute to improve existing deblending algorithms. Finally, synthetic data examples are shown to illustrate the machine learning deblending approach.

The simultaneous firing of marine sources can provide a significant uplift in terms of acquisition efficiency and data quality enhancement. However, the seismic interference resulting from one or more ‘other’ sources needs to be well understood and the appropriate processing strategies need to be developed for the method to fulfil its promise. In this paper, a modified inversion approach is presented for the effective separation of sources in marine simultaneous shooting acquisition. The method aims to distribute all energy in the simultaneous shot records by reconstructing the individual shot records at their respective locations. The method is applied to a wide azimuth data set acquired in the Gulf of Mexico where two sources out of four in total were fired simultaneously. Results demonstrate that the individual sources can be separated satisfactory, both at the prestack and post‐stack level.

No abstract