Efficient 2D multiple attenuation using SRME with curvelet-domain subtraction

Surface Related Multiple Elimination (SRME) usually suffers the issue of either over-attenuation that damages the primaries or under-attenuation that leaves strong residual multiples. This dilemma happens commonly when SRME is combined with least-squares subtraction. Here we introduce a more sophisticated subtraction approach that facilitates better separation of multiples from primaries. Curvelet-domain subtraction transforms both the data and the multiple model into the curvelet domain, where different frequency bands (scales) and event directions (orientations) are represented by a finite number of curvelet coefficients. When combined with adaptive subtraction in the time–space domain, this method can handle model prediction errors to achieve effective subtraction. We demonstrate this method on two 2D surveys from the TAiwan Integrated GEodynamics Research (TAIGER) project. With a careful parameter determination flow, our result shows curvelet-domain subtraction outperforms least-squares subtraction in all geological settings. We also present one failed case where specific geological condition hinders proper multiple subtraction. We further demonstrate that even for data acquired with short cables, curvelet-domain subtraction can still provide better results than least-squares subtraction. We recommend this method as the standard processing flow for multi-channel seismic data.

The Role of Legacy Seismic in Enhancing the Image in the Current Seismic Data – Abu Dhabi: Case Study

Acquiring surface seismic data can be challenging in areas of intense human activities, due to presence of infrastructures (roads, houses, rigs), often leaving large gaps in the fold of coverage that can span over several kilometers. Modern interpolation algorithms can interpolate up to a certain extent, but quality of reconstructed seismic data diminishes as the acquisition gap increases. This is where vintage seismic acquisition can aid processing and imaging, especially if previous acquisition did not face the same surface obstacles. In this paper we will present how the legacy seismic survey has helped to fill in the data gaps of the new acquisition and produced improved seismic image. The new acquisition survey is part of the Mega 3D onshore effort undertaken by ADNOC, characterized by dense shot and receiver spacing with focus on full azimuth and broadband. Due to surface infrastructures, data could not be completely acquired leaving sizable gap in the target area. However, a legacy seismic acquisition undertaken in 2014 had access to such gap zones, as infrastructures were not present at the time. Legacy seismic data has been previously processed and imaged, however simple post-imaging merge would not be adequate as two datasets were processed using different workflows and imaging was done using different velocity models. In order to synchronize the two datasets, we have processed them in parallel. Data matching and merging were done before regularization. It has been regularized to radial geometry using 5D Matching Pursuit with Fourier Interpolation (MPFI). This has provided 12 well sampled azimuth sectors that went through surface consistent processing, multiple attenuation, and residual noise attenuation. Near surface model was built using data-driven image-based static (DIBS) while reflection tomography was used to build the anisotropic velocity model. Imaging was done using Pre-Stack Kirchhoff Depth Migration. Processing legacy survey from the beginning has helped to improve signal to noise ratio which assisted with data merging to not degrade the quality of the end image. Building one near surface model allowed both datasets to match well in time domain. Bringing datasets to the same level was an important condition before matching and merging. Amplitude and phase analysis have shown that both surveys are aligned quite well with minimal difference. Only the portion of the legacy survey that covers the gap was used in the regularization, allowing MPFI to reconstruct missing data. Regularized data went through surface multiple attenuation and further noise attenuation as preconditioning for migration. Final image that is created using both datasets has allowed target to be imaged better.

Demonstrating multiple attenuation with model-driven processing using neural networks

Some of the key tasks in seismic processing involve suppressing multiples and noise that interfere with primary events. Conventional multiple attenuation on seismic prestack data is time-consuming and subjective. As an alternative, we propose model-driven processing using a convolutional neural network trained on synthetically modeled training data. The crucial part of our approach is to generate appropriate training data. Here, we compute a generic data set with pairs of synthetic gathers with and without multiples. Because we generate the primaries first and then add multiples, we ensure that we have perfect target data without any multiple energy. To compute generic and realistic training data, we include elements of wave propagation physics and implement a randomized flexibility of settings such as the wavelet, frequency content, degree of random noise, and amplitude variation with offset effects with each gather pair. A fully convolutional neural network is trained on the synthetic data in order to learn to suppress the noise and multiples. Evaluations of the approach on benchmark data indicate that our trained network is faster than conventional multiple attenuation because it can be run efficiently on a modern GPU, and it has the potential to better preserve primary amplitudes. Multiple removal with model-driven processing is demonstrated on seismic field data, and the results are compared to conventional multiple attenuation using a commercial Radon algorithm. The model-driven approach performs well when applied to real common-depth point gathers, and it successfully removes multiples, even where the multiples interfere with the primary signals on the near offsets.

A review of OBN processing: challenges and solutions

In the last decade, a significant shift in the marine seismic acquisition business has been made where ocean bottom nodes gained a substantial market share from streamer cable configurations. Ocean bottom node acquisition (OBN) can acquire wide azimuth seismic data over geographical areas with challenging deep and shallow bathymetries and complex subsurface regimes. When the water bottom is rugose and has significant elevation differences, OBN data processing faces a number of challenges, such as denoising of the vertical geophone, accurate wavefield separation, redatuming the sparse receiver nodes from ocean bottom to sea level and multiple attenuation. In this work, we review a number of challenges using real OBN data illustrations. We demonstrate corresponding solutions using processing workflows comprising denoising the vertical geophones by using all four recorded nodal components, cross-ghosting the data or using direct wave to design calibration filters for up- and down-going wavefield separation, performing one-dimensional reversible redatuming for stacking QC and multiple prediction, and designing cascaded model and data-driven multiple elimination applications. The optimum combination of the mentioned technologies produced cleaner and high-resolution migration images mitigating the risk of false interpretations.

Memorial

Virgil Bardan was known for his contributions to seismic data acquisition and digital data processing related to inversion, sampling, and multiple attenuation. His numerous publications and erudite presentations, in a career that extended for more than 45 years, established him as a leader in exploration geophysics.