Henning Hoeber

Frequency Dependent, Structurally Conformable Filtering

At low frequencies, data have inherently poor spatial resolution, and are thus inherently over-sampled by 3D marine data. We recently demonstrated (Whitcombe & Hodgson, 2005; 2007) that large spatial filters can be applied to post-migration low-frequency data without degrading signal: a large spatial averaging filter can be thought of as spatial stacking, which will not smear the data provided the operator size is comparable to the spatial wavelength. This technique is particularly useful for inverted data, which relies heavily on low frequency information. These lower frequencies enhance the imaging of thick pay intervals, and help avoid the occurrence of false high impedance beds within the middle of such low impedance layers. Additionally, the extra low frequencies improve the linearity of the calibration & calculation of seismic net pay from the layer SNA (sum of negative amplitudes) attribute.

Multivariate Geostatistical filtering of time-lapse seismic data for an improved 4-D signature.

4-D technology is moving into an accelerated phase with several successful projects to date. Acquisition and processing imprints are considered negligible with standard 3-D processing but become critical when computing time-lapse seismic differences. We introduce a spatial co-filtering geostatistical technique into the 4-D processing sequence to remove acquisition imprints and other uncorrelated noise that increases the repeatability and optimises the 4-D signature. Examples from two fields from the North-Sea, Draugen (Norske Shell) and Forties (BP) are presented.

Improved 4D seismic processing: foinaven case study

Using a case history from West of Shetlands, we study the combined impact of improved acquisition and improved processing on 4D time-lapse repeatability, as well as on 3D imaging. We show how, following a carefully planned 4D acquisition, new processing technologies improve the repeatability and interpretability of our three seismic vintages. In the first part of this paper we compare the repeatability of two vintages acquired in a non-4D friendly fashion to that of 4D-purpose data. We show how processing these vintages with an improved 4D sequence, i.e. better regularization and 4D binning, with pre-stack time Kirchhoff imaging, increases the repeatability. In the second part of the paper we also show comparisons of data processed with the new Kirchhoff imaging based 4D sequence to the same data analyzed with a 4D DMO sequence from four years ago.

Stabilizing the AVO Gradient

A technique to reduce the noise in AVO gradient data is presented. This filtering technique is a post-processor to a traditional AVO fit. It has its roots both in time windowed AVO (Ratcliffe & Adler, 2000), and also in hodogram technology (Keho et al., 2001). Both of these approaches utilize the principle that, for noise-free data, we expect a constant relationship between the intercept A, and gradient B for each point in the wavelet associated with a discrete reflection event. In B vs. A crossplot space these points would plot along a trajectory passing through the origin and characterized by a specific angle, which we call χ, and measure anti-clockwise from the positive A axis.

Simultaneous multivintage time-shift estimation

Time-shift estimation is a key step in seismic time-lapse processing as well as in many other signal-processing applications. We consider the time-shift problem in the setting of multiple repeat surveys that must be aligned consistently. We introduce an optimized least-squares method based on the Taylor expansion for estimating two-vintage time shifts and compare it to crosscorrelation. The superiority of the proposed algorithm is demonstrated with synthetic data and residual time-lapse matching on a U. K. continental shelf data set. We then discuss the shortcomings of cascaded time alignment in multiple repeat monitor surveys and propose an approach to estimate simultaneous multivintage time shifts that uses a constrained least-squares technique combined with elements of network theory. The resulting time shifts are consistent across all vintages in a least-squares sense, improving overall alignment when compared to the classical flow of alignment in a cascaded manner. The method surpasses the cascaded approach, as noted with sample synthetic and three-vintage U. K. continental shelf time-lapse data sets.

Robust Residual Gather Flattening

Well-aligned or “flat” events are key for high quality amplitude versus offset or amplitude versus angle fitting. In the context of marine 3D and 4D processing we have recently reported (MacKenzie et. al, 2004) on our efforts to improve gather quality with automated dense higher order moveout correction (including improved filtering methods). Here, we show how residual non-flatness of the data that cannot be further improved with the velocity picking tools, can nevertheless be corrected for prior to AVO. This gather pre-conditioning can be applied in a time-variant manner and is therefore able to correct for conflicting time-shifts. We also show that polarity reversal events are correctly flattened.

On the Use of Geostatistical Filtering Techniques In Seismic Processing

When redundancy of seismic data exists factorial cokriging enables the estimation of (1) a common part, based on the common spatial behavior, and (2) the differences relative to the common part of the input data. Coléou (2002) first introduced the automatic implementation of factorial co-kriging (AFACK) as a filtering technique for the time-lapse (4D) processing sequence. It was specifically designed to optimize the critical time-lapse information such as the repeatability and the 4D seismic signature. However, over the last two years we have been developing new applications of this technique in very different processing environments. For example, applications providing data reduction, such as stacking or AVO and EI analysis, have a direct interest in the common part of consecutive offset cubes. Furthermore, we have successfully applied AFACK to more specialized problems such as the merging of OBC and streamer data or the decomposition of wide-azimuth data for fracture characterization.

Foinaven: 4D processing comes up trumps

A good 4D image shows the effect on the seismic of production-related changes from within and sur-rounding a reservoir. It is well understood that the quality of 4D images is driven by two main ingredients: the quality of the acquisition and the quality of the processing. While 4D acquisition has been taking the limelight recently, the evolution of 4D processing algorithms has been steadily, and perhaps quietly, enhancing the quality of 4D images, including data acquired before the advent of carefully designed, dedicated 4D surveys. Using data from the Foinaven Field, one of three BP-operated fields in the West of Shetlands area, this paper shows how such improvements in processing technology over the last decade have enhanced the quality of the 3D imaging, the 4D seismic difference image, and improved the 4D repeatability.

Improved Time Delay Estimation

We compare cross-correlation based time delay estimation to two alternative algorithms: a time-domain least squares Taylor method and a time-domain based bi-coherence analysis. For short-window seismic analysis and uncorrelated noise the Taylor method performs best. This is demonstrated on synthetics and with a North Sea 4D data example for which we find a significant reduction in NRMS (13%) with the improved algorithm, when compared to cross-correlation time-alignment.

A New Approach to Reducing Multiple Leakage on Time Lapse Datasets

The standard processing solution in time-lapse analysis to the problem of non-repeatable noise sources, both coherent and random, is to perform the best denoise on all of the vintages independently. For this purpose, 4D QC measures and, of course, inspections of the seismic 4D differences are used to find the optimum parameter solution. This is currently also best practice in the removal of multiple energy, as there are many reasons for multiples to be non-repeatable, in particular changes in the source and receiver locations, changes in the water-layer and variations in the acquisition wavelet. Finding the optimal multiple subtraction operators on each vintage so as to obtain the least leakage of multiple energy on the 4D difference is notoriously difficult. In this paper we introduce a novel inversion scheme which designs shaping operators for the optimum adaptation of the multiple models to the individual vintages whilst simultaneously optimizing the subtractions for minimum leakage on the 4D difference.