Clement Kostov

Internal Multiple Attenuation on Radial Gathers With Inverse-scattering Series Prediction

We present a novel workflow for pre-stack prediction and attenuation of internal multiples, applicable to data acquired with orthogonal survey geometries over geologies with mild structural variations. First, we interpolate the data pre-stack (5D) data to common midpoint (CMP) gathers with regular sampling in offset and azimuth. Such CMP gathers are suitable for noise attenuation as shown in previously published case studies. Next, we predict internal multiples in radial gathers, i.e. common midpoint and common azimuth gathers. We use the leading term of the inverse scattering series to perform the algorithm called inverse-scattering series internal multiple prediction (ISIMP) and obtain a pre-stack model of internal multiples without input of subsurface information. We follow the prediction of internal multiples with multidimensional adaptive subtraction. We apply this workflow to data from the Cooper basin, Australia. Our results demonstrate overall improvements in primary resolution and reveal weak primary events previously obscured by multiples. We note that the transformation to the radial domain has been a key enabler for this workflow.

Using Models of Attenuation in the Prediction of Internal Multiples: Methodology and Synthetic Data Examples

Summary We consider methods for prediction of internal multiples that combine events in the data through convolutions and correlations, in particular methods based on the inverse scattering series and methods that predict a subset of internal multiples related to specified boundaries in the subsurface. Such methods provide kinematically accurate models when their prerequisites are satisfied but have fundamental limitations in terms of predicting accurate amplitudes; the amplitudes are under-predicted due to approximations related to scattering and anelastic attenuation losses. We propose variations of these prediction methods that use models of attenuation in the subsurface (scattering or anelastic attenuation factors) to improve the amplitudes of the predicted internal multiples. The models of attenuation required in our approach may be derived from borehole data (logs, vertical seismic profiles) or from surface seismic data (e.g. from Q tomography) and are used in pre-processing the data input to the prediction algorithms. We illustrate the new algorithms using synthetic data for 1D earth and 3D point source. The proposed methods facilitate the adaptive subtraction of the predicted internal multiples from the data and would naturally complement workflows that estimate attenuation from the seismic data and apply Q compensation during imaging.

Attenuation of near-surface multiples on land seismic data: challenges, practical aspects and open issues

We discuss attenuation of near-surface multiples on land seismic data using examples from a pilot test survey in Northern Kuwait. While the current combination of prediction and adaptive subtraction provides essential preparation of the data for subsequent processing steps, the review of current approximations and limitations points to directions for future developments linking characterization of the near surface (possibly using multiples as signal) and the attenuation of near-surface generated multiples.

Improving Inversion Accuracy by Optimal Internal Multiple Attenuation - A Case Study from the UAE

We present a multiple attenuation case study performed over an onshore orthogonal seismic survey where strong internal multiple contamination directly affects the relative amplitude between traces of primary reflection events. Preserving the primary relative amplitude trend is essential for any inversion and reservoir characterization process. The proposed workflow starts with understanding the internal multiple contamination pattern using borehole modelling to assess the overall contamination trend and amplitude trend between traces. The modelling is also used to identify formations responsible to generate strong internal multiples in the seismic record. Next, we predict the surface seismic multiples present in the seismic record using a data-driven forward modelling algorithm for surface and internal multiples. The predicted model is then subtracted from the field data using an adaptive matching process. Our results demonstrate a superior seismic inversion and a suitable input for reservoir assessment and delineation of the thin carbonates reservoirs present in this field.

Multimodel Adaptive Subtraction and Its Application to Multimeasurement Data Acquired in Shallow-water Surveys

Many existing adaptive subtraction methods can be formulated as a parameter estimation problem and all of them, although fundamentally different, have a common restriction that the dimension of the unknown parameters must be determined in advance. We propose an adaptive subtraction framework, called multimodel adaptive subtraction (MMAS), that aims to relax this restriction as well as regularize the estimation of the parameters through a generalized information criterion (GIC). We show that MMAS can be applied to the popular least-squares adaptive subtraction (LSAS) method and call the resulting algorithm the multimodel least-squares adaptive subtraction (MMAS-LS). We further extend MMAS-LS to 3D and applied it to the multiple subtraction of a 3D data set from a multimeasurement shallow-water survey. We compare our proposed 3D MMAS-LS method with conventional 3D LSAS and observe that our proposed method is able to preserve the primary events better and achieve the same level of multiple attenuation compared to 3D LSAS, while using smaller or simpler filters.

Interbed Multiple Attenuation in Kuwait - A Minagish Case Study Revisited

This work shows a revisited multiple attenuation workflow applied over a selection of the Minagish dataset. Essential components of this workflow are extended interbed multiple prediction (XIMP), simultaneous matching between the field dataset and predicted multiple models, and improved quantitative well-constrained quality control (QC) of the final results. The XIMP and the adaptive subtraction framework allow the application of the multiple attenuation workflow before imaging, avoiding the cross-talk between primaries and internal multiples during the migration process. Noticeable improvements can be seen over the Cretaceous and Jurassic zones on pre-stack gathers, seismic stacked sections, and well ties.

Comparing Finite-difference And Kirchhoff Prestack Depth Migration

The Kirchhoff algorithm has been widely used for prestack imaging in complex areas. In general, however, Kirchhoff algorithms used in large-scale production work do not account for multipathing, amplitude variation due to significant ray spreading, or phase changes through caustics. As a result, there has been significant interest recently in wavefield extrapolation methods for prestack imaging. Here, we compare several aspects of both Kirchhoff and wavefield downward-continuation in the context of subsalt imaging, and examine the relative strengths and weaknesses of each method.