Sub-basalt imaging is often poor quality with the seismic data collected offering little information beneath the so-called "masking layer." Standard acquisition and bulk data processing parameters seem to do little to alleviate the problem. Detailed examination of each individual basalt covered region is required to determine image quality and reliability for that particular area. In the picture above Deep basalt. A synthetic model, for use in the 2D viscoelastic finite difference scheme, contains a masking layer at depth. The water layer at the surface has been stripped off to save on computation time. The red stars indicate the shot locations and the receivers are also located at the surface. The colorbar on the right shows the P-wave velocities in ms-1. Each of the boundaries are labelled, I = bottom layer one, II = bottom layer two, TB = top basalt, BB = bottom basalt and TAR represents a target layer at depth. The basalt layer is on average 80 m thick.
It is necessary to pay close attention to the local geostatistics and attenuation properties, of not only the masking layer but also of the overlying layers, when analyzing the image quality in a particularly seismically difficult and complex region.
One such region that causes major difficulties for the collection of seismic data is the Irish Atlantic Margin. This area remains relatively unexplored, with the exception of the Porcupine Basin. It is a harsh region, which is plagued with extremely difficult weather conditions for both drilling and seismic. This, coupled with the water depth and extremely poor quality of the seismic data in some of the West of Ireland basins leaves it a somewhat undesirable place for exploration.
Although there is no control over the weather, recent technological advancements have proved that the water depths (200-2,500 meters within the Rockall Basin) should not necessarily act as a deterrent. This leaves the problem of data quality. The main reason for the poor quality of the seismic data is the presence of Tertiary volcanics in the form of lava flows and sills. These act as masking layers, degrading the quality of the image beneath.
The problem
It is often believed that these basalt masking layers effectively act as total barriers to the seismic energy, however, recent findings show that this is not the case and that significant transmission through to deeper layers does occur. In addition to the high impedance contrast associated with these masking layers, there is also a high degree of heterogeneity. This heterogeneity has a detrimental effect on the amount of seismic energy that is transferred to depth. In the photo above, - Shallow basalt. The basalt (masking) layer is located near the surface (Top Basalt = TB and Bottom Basalt = BB) with two target layers beneath (TAR1 and TAR2).
Seismic wave attenuation further reduces the available energy, and this, coupled with mis-stacking brought about by rugosity at the top of the basalt layers, results in sub-basalt structures being particularly difficult to image. In order to image target structures and subtle features beneath these layers, adequate energy must be available at depth to ensure a sufficiently high signal-to-noise ratio.
It is known that the masking effect of high impedance layers is sensitive to the layer thickness versus seismic wavelength relationship. Analogous to this problem is the layer continuity and how the width of the gaps in the masking layer, if any, compare to the source frequency. Therefore, the source frequency is obviously very important at the acquisition stage.
Another very important parameter for sub-basalt imaging is the highly heterogeneous nature of the material. It is this parameter which has a large influence on the quality and reliability of seismic data beneath the masking layer. Recent work has shown that the internal characteristics of the basalt itself can degrade sub-basalt image quality. If the velocity variations within the basalts are large, this can lead to strong seismic wave scattering, which is not corrected for in migration. Improved sub-basalt imaging requires innovative processing and/or acquisition techniques.
Recent methods
Solutions designed to help overcome the masking effect of these high-velocity basalts have included long-offset seismic programs. These surveys have been designed to record converted waves to image sediments below the volcanics. Conventional long offsets have utilized the wide aperture recording (WAR) technique, which can be expensive and has several limitations, although some areas have seen improved result windows. In the photo above A brute stack of the data collected in the shallow basalt model. Although consisting of a single continuous layer in the model, the basalt appears discontinuous in the brute stack. There are strong lateral variations in the intensity of the response from each of the reflectors (labelled Basalt, TAR1 and TAR2). As a result, it is difficult to follow any of the reflectors over a distance of 300 m or more. Note the quality of the reflector images are much poorer than those imaged in the deep basalt model.
Longer offset data did help with the imaging problem as it allowed for the detection of converted waves (P to S) at the basalt/sediment interface. Conversion occurs when the P-waves are incident on the top basalt at angles beyond the critical angle and the phenomenon is known to be more prominent when the P-wave velocity in the sediment layer above the basalt is very similar to the S-wave velocity in the basalt layer itself.
The wave is transmitted, reflected as an S-wave and reconverted at the top basalt interface. These converted waves were seen to give the most information below the basalt. A problem associated with this method meant that demultiple techniques used on the long offset data also removed primary energy. Other techniques have since been developed at the processing end to try to isolate the P-wave energy. However, some of these processes take time and the turnaround from acquisition to interpreter can increase accordingly.
Alternative techniques
For the last few years, Fastnet GeoLabs have been approaching the problem from the perspective of trying to understand the causes associated with poor image quality in the region. Determination of the effects of layer heterogeneity on the wavefield, using geological scenarios based on seismically complex areas, were examined.
Based on the results, investigations into how image quality might be improved by varying acquisition parameters were conducted and assessment of image quality variability as a function of processing parameters ensued. The aim of the project was to offer an assessment of the probable reliability of currently available field seismic data in a basalt covered problem area.
Close attention was played, throughout, to the role of frequency content in sub-basalt imagery. Heterogeneity was also seen to play a major role and therefore was introduced with realistic geostatistics into a visco-elastic full wavefield finite difference simulator. Snapshots of the wavefield in time show that strong velocity fluctuations within the basalt leads to strong seismic wave scattering. The behavior of the wavefield in each layer is strongly determined by the range of velocity fluctuations or the standard deviation.
Unlike other velocity models, the input velocity models used for these finite difference simulations try to incorporate as much of the medium heterogeneity, as possible, in order to capture all the variability. This variability cannot be calculated on a point-to-point basis and therefore has to be defined statistically. The heterogeneity is extremely important in the scattering problem and has to be included if the medium is to have the same scattering properties as the real earth.
Therefore, the velocity fluctuations within each layer will be statistically defined based on the mean velocity and the fluctuations about this mean derived from borehole data, where available, or from analogous boreholes. The simulations have also shown that the variations associated with the severe rugosity of the basalt interfaces is, as important if not, more important than the heterogeneity within the layer. Realistic degrees of roughness can be obtained from the real seismic data available and included in the velocity models.
The viscoelastic finite difference scheme is used to test the effects of highly heterogeneous basalt layers on underlying target structures. Velocity models with realistic velocity fluctuations and topography variations are generated and the synthetic data processed like real data. Multifold migrated CMP stacked sections are then produced. Sweeps of realistic source frequencies and different processing flows can be tested on each of the models.
Two synthetic generic models based on Irish offshore data are presented. These incorporate either a deep or shallow basalt layer in highly heterogeneous media. Velocity fluctuations are statistically defined based on the scaling statistics for equivalent geologies. Rough interface topographies showing realistic rugosity have been included. The models are statistically realistic, highly heterogeneous, and incorporate wave scattering and focusing/defocusing characteristics.
In the deep basalt brute stack, all interfaces are clearly identified. These appear laterally discontinuous on the section. It should be noted that there clearly is transmission through the basalt, therefore sub-basalt reflectors on available real seismic data are most probably caused by real sub-basalt structures. However, on the section the basalt appears to show a high degree of lateral variability, which results from minor changes in thickness and topography in the actual model. Therefore the geometry and continuity of reflections from sub-basalt structures are severely distorted by the overlying basalt.
The shallow basalt brute stack shows how the shallow basalt has a detrimental effect on our ability to image at depth. There is greater effective screening of apparent seismic energy than with the deep basalt. In general, energy transmission is very poor, however, it was found that there is some low frequency transmission through the basalts, whereas high frequencies are almost totally screened and do not penetrate the basalt.
Summary of results
The spectral characteristics of the source and receiver could prove crucial in the sub-basalt imaging problem. Determination of the attenuation factors in the layers above is needed to calculate the frequency spectrum of the wavefield when it encounters these highly scattering layers. High frequency spectral content of the wavefield may be detrimental in areas which include very heterogeneous shallow layers. Synthetic testing of the effect of the geometry, including rugosity, of the masking layers will help determine the reliability of structures visible, if any, in the sub-basalt region in question.
Detailed reprocessing of field data might improve image quality in these difficult areas, however, it may be that new technologies such as character matching and pattern recognition are required to aid the detection of subtle reflectors in poor S/N ratio sub-basalt environments.
The scheme presented can be used to test current interpretations of specific areas and the results obtained will obviously be useful when investigating the quality and reliability of currently available seismic data from any basalt covered areas. Also the relative effectiveness of streamer versus OBC data may be tested synthetically, using this method, before major expense is utilized in the field.
The masking effect of the volcanic layers is severely affecting exploration activities in frontier areas, such as the Rockall Trough off the west coast of Ireland. An innovative deterministic modeling technique, such as the one presented, is needed to add quality control to the seismic data collected and improve confidence in areas with potentially significant undiscovered hydrocarbons.
Acknowledgement
The authors would like to thank Statoil Exploration (Ireland) Limited for allowing presentation of this work.
