Anat Canning

Reducing 3-D acquisition footprint for 3-D DMO and 3-D prestack migration

The acquisition patterns of 3-D surveys often have a significant effect on the results of dip moveout (DMO) or prestack migration. When the spatial distribution of input traces is irregular, results from DMO and migration are contaminated by artifacts. In many cases, the footprint of the acquisition patterns can be seen on the migrated section and may result in incorrect interpretation. This phenomena also has a very significant effect on the feasibility of conducting amplitude variation with offset (AVO) analysis after 3-D prestack migration or after 3-D DMO, and also may affect velocity analysis. We propose a simple enhancement to migration and DMO programs that acts to minimize acquisition artifacts.

Effects of irregular sampling on 3-D prestack migration

In a direct application of Kirchhoff migration each trace is added to the migrated volume by spreading the data along impulse response curves with suitable change in the amplitude and shape of the wavelets. Overlapping impulse responses form the correct answer where they form an envelope and are supposed to cancel elsewhere. For uniform data (constant offset, constant azimuth, constant midpoint spacing and constant velocity) the cancellation is excellent and the reflectors have the correct amplitude. This paper shows how failure to meet any of these constancy criteria results in noise. Modified summation procedures can reduce the noise at the expense of a loss of resolution.

Another look at the question of azimuth

The question of azimuth seems to be an emotional issue in 3-D seismic exploration. Some people believe that 3-D surveys should contain a wide range of source‐receiver azimuths so that the subsurface is sampled from all directions and the acquisition is closer to a “true” 3-D survey. Others, on the other hand, argue the case for narrow azimuth acquisition. A few years ago, Malcolm Lansley, then with Halliburton Geophysical, gave an excellent presentation regarding the question of azimuth at an SEG summer workshop. In this talk (which was recently presented to the Houston Geophysical Society), Lansley provided an extensive summary of the advantages and disadvantages of wide azimuth surveying, and listed many aspects that one should consider when deciding between narrow and wide azimuth acquisition.

A two‐pass approximation to 3-D prestack migration

A two-pass approximation to 3-D Kirchhoff migration simplifies the migration procedure by reducing it to a succession of 2-D operations. This approach has proven very successful in the zero-offset case. A two-pass approximation to 3-D migration is described here for the prestack case. Compared to the one-pass approach, the scheme presented here provides significant reduction in computation time and a relatively simple data manipulation scheme. The two-pass method was designed using velocity independent prestack time migration (DMO-PSI) applied in the crossline direction, followed by conventional prestack depth migration in the inline direction. Velocity analysis, an important part of prestack migration, is also included in the two-pass scheme. It is carried out as a 2-D procedure after 3-D effects are removed from the data volume. The procedure presented here is a practical full volume 3-D prestack migration. One of its main benefits is a realistic and efficient iterative velocity analysis procedure in three dimensions. The algorithm was designed in the frequency domain and the computational scheme was optimized by processing individual frequency slices independently. Irregular trace distribution, a feature that characterizes most 3-D seismic surveys, is implicitly accounted for within the two-pass algorithm. A numerical example tests the performance of the two-pass 3-D prestack migration program in the presence of a vertical velocity gradient. A 3-D land survey from a fold and thrust belt region was used to demonstrate the algorithm in a complex geological setting. The results were compared with images from other 2-D and 3-D migration schemes and show improved resolution and higher signal content.

Feathering correction for 3‐D marine data

The organization of 3D multi-fold seismic data on a regular grid is commonly performed by binning the data in midpoint space prior to stacking, An alternative approach using DMO & DMO-1 is designed to reorganize the data on a regular grid and remove cable feathering. With this approach the 3D data are conditioned for pre-stack processing, especially for inline pre-stack migration. An enhancement of the DMO algorithm accounts for the irregular distribution of traces in 3D surveys and minimizes DMO artifacts. The spatial distribution of input traces is measured in a preliminary stage. This measurement is passed as additional information to the DMO program, and is used to correct for the irregularity. The preprocessing program operates only on the coordinates and serves as an additional tool for evaluating the spatial sampling.

Azimuthai AVA analysis using full‐azimuth 3D angle gathers

Azimuthal AVA workflow based on full-azimuth 3D angle gathers in depth is presented. The benefits of this approach are described and demonstrated, as well as the AVAZ inversion algorithm. An example from fractured carbonates reservoir from the North Sea is presented. Unique preprocessing and inversion workflow is discussed, illustrating the significance of appropriated data preconditioning to the reservoir detection. Final results are also analyzed using attribute crossplotting.

Global multiwell wavelet estimation

Wavelet estimation by matching the synthetic seismogram, calculated from well data, with seismic data at the well location is a problematic procedure. The objective of the process is to find a wavelet which is representative of the seismic data while the extraction is performed at a single location. Moreover, often the match obtained between the seismic trace and the synthetic trace is quite poor. Automatic wavelet extraction procedures may optimize the correlation between the seismic and the synthetic trace, but may result in a wavelet which is not an adequate representation of the seismic data. Wavelet extraction at different wells often provides different results, a situation that illustrates the local character of the procedure. Therefore, it is sometimes preferable to compromise the correlation values in order to extract a consistent wavelet which provides reasonable correlation at all wells, and is representative of the seismic data.

4-D Cross-Equalization and Offset Equalization Using A Neural Networks Approach

Global variations between different vintage datasets are traditionally minimized using a combination of deterministic wavelet matching techniques and crosscorrelations. In this paper we propose a statistical approach to cross-equalization. For this purpose we assume that the differences between the two datasets can be very complex, but they are global for the data. This means that overall the same differences can be detected at any location in the dataset except in the vicinity of the reservoir. Based on this assumption we analyze the difference between the two datasets at a region away from the production zone. We then find a single operator which maps one dataset to the other. Finally, we apply this operator to the whole input dataset and obtain two equalized datasets where the difference due to production can be easily mapped. We use the Neural Networks approach to find this operator. It allows us to define a non-linear, multidimensional operator, which can handle complex mapping of one dataset to the other, and does not rely on any deterministic theory to explain the differences between the

Amplitude Inversion of Reflectivity Type AVO Attributes

Summary Conventional acoustic impedance inversion techniques can be used to transform AVO attributes or post stack seismic data to micro-models of the corresponding physical parameters. This means that a section or volume of some physical parameter P can be built by applying amplitude inversion to the section (volume) of reflectivity ΔP/P. P is not necessarily acoustic impedance, but could also be P-wave velocity, S-wave velocity or the ratio Vp/Vs. These types of data can be used for interpretation, geological model building, reservoir characterization, etc. Here we present a method for amplitude inversion that is based on an eiegenvector basis expansion, and show its application to AVO attributes.

Some practical aspects of amplitude recovery before AVO and inversion

This paper is being presented at a special session honoring J.H.F. Gardner who was my mentor. Here I discuss a phenomenon which we studied together in 1992. At that time we noticed that migration amplitudes are highly affected by the acquisition geometries, and that migration introduces very significant amplitude distortions to the data. We suggested ways to solve these problems during the migration. Since then, I have often received data for AVO inversion which was naively migrated and contaminated by this artifact. Re-migration was not an option in many of these cases. Here I suggest a practical solution which can be applied after migration and before AVO. It is a very heuristic approach, but is practical for many cases where re-migration is not available.