Search

Estimation of primaries by sparse inversion (EPSI) is an iterative method that effectively separates primaries and surface-related multiples, especially in shallow water. Multiple attenuation in shallow water is challenging, mainly because of acquisition limitations. We propose a strategy for EPSI with the following objectives: (1) creates an alternating picking method for the first iteration of EPSI that correctly separates primaries and multiples and also expedites the convergence in some cases and (2) picks only strong shallow reflectors to alleviate the cost while attacking most of multiples generated by those reflectors . We applied our method to two synthetic data sets. Then, we tested the EPSI method on a complex field data set and demonstrated that it can effectively attenuate multiples.

The Point Thomson field, located on the North Slope of Alaska, covers the transition zone from onshore to a frozen lagoon. This complexity in the near-surface poses many challenges to proper imaging of the reservoir. Unground ice on the lagoon causes very poor signal-to-noise ratio (S/N) over a critical portion of the field. The rapid lateral variation in permafrost thickness introduces strong lateral velocity gradients in the near-surface. Onshore, the presence of ice lakes create highly localized anomalies. We present a workflow that we used to address these near-surface issues during the recent reprocessing of the vintage low-fold data. Aided by modern imaging technology for noise attenuation, statics, and velocity model building, we were able to address many of these issues and produce a significantly improved image.

Although wavefield extrapolation techniques are well developed for P-wave seismic imaging, ray based migration algorithms are still the workhorse for converted-wave (PS-wave) depth imaging. Full (exact) elastic-wave reverse-time anisotropic migration (RTM) has not been widely adopted for reasons of computational and workflow efficiency, despite its potential to deliver accurate sub-surface images in complex geological settings by directly solving the elastic wave equation. Even (approximate) converted-wave RTM in anisotropic media, using separate finite-difference propagators for quasi-P and quasi-S waves, is limited in applicability for algorithmic reasons. Here, we introduce an alternative converted-wave anisotropic RTM, using a low-rank decomposition of mixed-domain space-wavenumber propagators for quasi-P and quasi-S waves. These operators are formal integral solutions of the pure-mode wave equations which guarantee stable and dispersion-free time extrapolation for coarse time steps in anisotropic, heterogeneous media. The pure-mode extrapolators are attractive for both PS-wave structural imaging and velocity analysis. An ocean bottom cable synthetic example illustrates the effectiveness of low-rank PS-wave RTM when compared against state-of-the-art Gaussian beam and finite difference RTM algorithms.

4D seismic processing is designed to minimize non-repeatable noise while preserving real 4D signal. Coherent noise, such as from surface and interbed multiples, has different character from one vintage to another. Mitigating for these unwanted effects in a vintage-independent manner leads to suboptimal 4D results. In Zabihi et al., 2012, a method for reducing residual 4D multiple leakage was proposed. Khalil et al., 2013 expanded its applicability and put it in a feasible workflow. We extend this workflow to a multi-vintage setting including legacy and modern 4D acquisitions with very different 4D noise characteristics.

A precise knowledge of the seismic source far-field signature is required for accurate source de-signature. Near-field hydrophone data can be used to provide good quality signatures but are not always available. We describe a method of extracting the far-field signature from inversion of direct arrivals in the seismic data. We present the application of the methodology to a deep water survey offshore Gabon. The method can be applied to legacy surveys where near-field hydrophone data is unavailable for a more broadband result.

Changes in fracture direction across interfaces can have an important impact on PS-wave reflection coefficients in azimuthally anisotropic media. Extending conventional joint inversion with P-waves to include amplitude variations with azimuth (AVAz) would use radial and transverse reflection coefficients (RPSV and RPSH, respectively), and present a number of challenges. One of these is that layerstripping (LS) must be done either prior to or during joint inversion with fast PS1- and slow PS2-waves for azimuthal rock properties (e.g., fractures or stress directions). Also, null amplitude directions of RPSH can be shifted in azimuth when fracture direction changes across the interface, and will be different from the actual fracture direction. Thus, a full waveform inversion (FWI) approach with-out registering PS-waves to P-wave time could be more practical than attempting to align them prior to joint AVAz inversion.

We presented a joint tomography flow for OBN and streamer WAZ data in deep water Gulf of Mexico. This tomography provides the method to utilize surface offset gathers from surveys with different shot and receiver datums together for joint tomography. By incorporating both OBN and streamer data in tomography, the aperture and coarse sampling limitations of OBN can be overcome, and the full azimuth coverage, better low frequency, longer offsets, and higher S/N of OBN data can benefit tomography. Subsequently we are able to build a better velocity model.

We discuss methods to quantify the impact, reliability and value of low frequencies as provided in modern towed streamer broadband acquisitions. Acquisition, processing and inversion all have a role to play in creating reliable low frequency data. In this paper, focusing on marine data, we discuss various aspects of low frequency technology, associated uncertainties and QC methods. We address two key questions: Can broadband deliver the low frequencies? What value do they have? We show that scanning for the crossover frequency at well locations, where the background model is optimally known, is a useful way to visualize the impact of, and quantify the value of the low frequencies. Uncertainties due to unknowns such as wavelet errors at low frequencies and optimal regularization parameters such as sparseness constraints are discussed. Using this method, with a North Sea 3D multi-client data example, we show that broadband data provides valuable information, compared to conventional data that has been broadband processed.

Thrust complex imaging in the Timor Trough suffers from the fault shadows due to strong lateral velocity variation. We demonstrate a new workflow to tackle this. Broadband seismic data were acquired with high signal-to-noise ratio of low frequency. With broadband input, FWI derived better velocity model at the shallow water thrust area where the reflection tomography has limitation. Compared to conventional tomography which has difficulty in addressing the sharp velocity boundary properly, fault constrained tomography (FCT) uses the interpreted fault planes as constraint for inversion and benefits from better low frequency penetration in the severe fault shadows. Broadband seismic and depth imaging with FWI and FCT make a step change over the thrust complex areas.