ALBERT

Description: 〈span〉〈div〉ABSTRACT〈/div〉We have developed a novel approach to image vertical structures using multiply scattered waves. This method requires only a smooth seismic velocity model and data recorded at the surface. Previous methods to image near-vertical interfaces or faults using migration methods all require prior information about small-scale details in the seismic velocity model to infer the locations of multiple-scattering wave interactions. We used waves that had their last scattering interaction with near-vertical interfaces, while their other scattering points might be anywhere in the earth, including at the free surface. Our algorithm then images the final scattering point using a time-reversed mirror-style imaging condition, so we refer to the method as time-reversed mirror imaging. Artifacts in the images produced have clear causes and can be filtered out by stacking over shots and including contributions from multiples. Our numerical examples demonstrate the successful application of the method for staircase structures and a section of the Marmousi model. They also reveal a new way to diagnose errors in the smooth or reference velocity model used. In addition, our method can be used to image point scatterers in active seismic surveys or for event location in passive surveys.〈/span〉

Description: 〈span〉〈div〉Abstract〈/div〉The resolution limit of seismic data is an intricate issue that depends not only on frequency and data quality (signal-to-noise ratio) but also on the tools and technology used to analyze seismic response. In this context, the subject of thin-bed delineation is extremely significant for coal-laminated (causing large acoustic impedance contrasts) clastic sequences of the Western Onshore Basin, India. Most of the clastic reservoirs in the area are of subseismic resolution (below 10 m in thickness) due to the low dominant frequency available in seismic data (19–35 Hz). This is where improving seismic resolution is essential for a detailed structural and stratigraphic interpretation. We have implemented a modified workflow with which, by using state-of-the-art techniques of time-frequency decomposition and cepstral analysis, significant seismic bandwidth extension has been achieved. This in turn yields improved vertical resolution of the seismic data with better geologic interpretability. Our approach is named the “syn-cepstral method” after its two integral constituents — synchrosqueezing transform and cepstral analysis. Applying the syn-cepstral method produces better well-to-seismic ties and resolves additional events in comparison to the original seismic data. The validity of syn-cepstral methodology has been demonstrated by 1D and 2D modeling studies followed by application to a 3D seismic data set from the Western Onshore Basin of India. The improvement in thin-bed delineation arising from the increased bandwidth of the resultant data has been validated by well-to-seismic ties and amplitude map interpretation. Thus, while thin clastic reservoir beds in the logs show no discernible presence in the original seismic data, upon application of the syn-cepstral method, the resultant seismic data show improved interpretability of these units.〈/span〉

Description: The presence of injected $${\mathrm{CO}}_{2}$$ in the Utsira Sand at the Sleipner site, Norway, is associated with a high negative P-wave velocity anomaly; that is, a low postinjection velocity and a strong seismic response. Time-lapse seismic imaging of $${\mathrm{CO}}_{2}$$ injection at Sleipner is thus a viable monitoring tool of the injected $${\mathrm{CO}}_{2}$$ . The work flow usually involves conventional seismic processing, including stacking, and results in seismic images. Multiple reflections, interference effects such as tuning, and the velocity pushdown effect due to $${\mathrm{CO}}_{2}$$ injection render these seismic images ambiguous in terms of the localization and the quantification of the $${\mathrm{CO}}_{2}$$ in the Utsira Sand. Nonetheless, seismic images often form the basis for analyses that aim to quantify the injected $${\mathrm{CO}}_{2}$$ . We employed elastic 2D full waveform inversion to invert prestack seismic Sleipner data from preinjection (1994) and postinjection (1999) and compared the resulting postinjection P-wave velocity model with the corresponding seismic image. We found that the high-amplitude reflections in the seismic image do not everywhere coincide with low postinjection P-wave velocities. Drawing extensive and integrated conclusions is out of our scope, because this would require full control over the seismic data processing and a more comprehensive forward modeling. For instance, modeling should be done in 3D and an adequate anelasticity formulation should be added. However, the waveform inversion scheme we used accounts for all the aforementioned elastic propagation effects. The results therefore suggested that the exclusive use of seismic images to quantify $${\mathrm{CO}}_{2}$$ could be revised and full waveform inversion should be added to the analysis toolbox.

Description: By solving the Marchenko equations, one can retrieve the Green’s function (Marchenko Green’s function) between a virtual receiver in the subsurface and points at the surface (no physical receiver is required at the virtual location). We extend the idea behind these equations to retrieve the Green’s function between any two points in the subsurface, i.e., between a virtual source and a virtual receiver (no physical source or physical receiver is required at either of these locations). This Green’s function is called the virtual Green’s function, and it includes all primary, internal, and free-surface multiples. Similar to the Marchenko Green’s function, this virtual Green’s function requires the reflection response at the surface (single-sided illumination) and an estimate of the first-arrival traveltime from the virtual locations to the surface. These Green’s functions can be used to image the interfaces from above and below.

Description: We developed a method of moveout correction in the $$\tau \hbox{ - }p$$ domain to tackle some of the problems associated with processing wide-angle seismic reflection data, including residual moveout and normal-moveout stretching. We evaluated the concept of the shifted ellipse in the $$\tau \hbox{ - }p$$ domain as an alternative to the well-known concept of the shifted hyperbola in the $$t\hbox{ - }x$$ domain. We used this shifted-ellipse concept to address the problem of residual moveout caused by vertical heterogeneity in the subsurface. We also addressed the stretching problem associated with dynamic corrections by combining selected strips from a set of constant-moveout stacks generated using a shifted-ellipse equation. Application of this method to a wide-angle data set from the Faeroe-Shetland Basin provided an enhanced image of the subbasalt structure.

Description: We applied the Helmholtz tomography technique to 6.5 hours of continuous seismic noise record data set of the Valhall Life of Field network. This network, that has 2320 receivers, allows us to perform a multifrequency, high-resolution, ambient-noise Scholte wave phase velocity tomography at Valhall. First, we computed crosscorrelations between all possible pairs of receivers to convert every station into a virtual source recorded by all other receivers. Our next step was to measure phase traveltimes and spectral amplitudes at different periods from crosscorrelations between stations separated by distances between two and six wavelengths. This is done in a straightforward fashion in the Fourier domain. Then, we interpolated these measurements onto a regular grid and computed local gradients of traveltimes and local Laplacians of the amplitude to infer local phase velocities using a frequency dependent Eikonal equation. This procedure was repeated for all 2320 virtual sources and final phase velocities were estimated as statistical average from all these measurements at each grid points. The resulting phase velocities for periods between 0.65 and 1.6 s demonstrate a significant dispersion with an increase of the phase velocities at longer periods. Their lateral distribution is found in very good agreement with previous ambient noise tomography done at Valhall as well as with a full waveform inversion P-wave model computed from an active seismic data set. We put effort into assessing the spatial resolution of our tomography with checkerboard tests, and we discuss the influence of the interpolation methods on the quality of our final models.