Underground » Geology
This project sought to develop procedures for:
- Interactive depth conversion through integrating borehole and seismic data so that seismic data can be transformed from the time domain to the depth domain.
- Stratigraphic and coal seam attribute interpretation of seismic reflection data sets allowing integration with borehole and other geological information.
- Output of results into a form, which can be visualised and manipulated by mine planning software.
This project developed a simple time-to-depth conversion algorithm specifically for coal seismic data where many boreholes are available for control purposes. The algorithm was successfully tested using 3D seismic data from the Sandy Creek area of Xstrata's Oaky Creek Mine. The results show that the total error in depth for the majority of this survey area were approximately 2 m even when the borehole spacing was decimated to approximately 800 m.
In the algorithm, the depth conversion is performed by stretching the time axis to the depth axis along common mid-point traces. The method requires there to be sufficient borehole control and that the subsurface horizons be gently dipping with only mild lateral velocity variations. In addition to its simplicity, one of the advantages of the algorithm is that the user can easily incorporate information from new boreholes into the time-to-depth conversion data base, thus keeping the seismic results current. There is no need to go back to the original seismic processing contractor. Once the data are converted into the depth domain, structural ambiguities in the time data are reduced, as are differences introduced through the seismic processing stages by different processing personnel and processing companies.
The key to the seismic depth conversion process lies in the estimation of the conversion velocities. This involves velocity interpolation from known locations at boreholes and other control points. A number of different interpolation methods have been investigated. These provide different results. Cross validation can help the selction of interpolation methods but care is needed. The choice should include consideration of geological as well as theoretical issues.
The depth-converted seismic data can be correlated directly with borehole data. Like the time seismic data, most importantly, horizon picks and sections can be output in data formats suitable for inclusion into other geological and mine planning software. Integration is facilitated and the seismic data can be better incorporated into daily mine operations.
From the depth-converted seismic data, it is also possible to undertake more detailed geological interpretations. Through the depth conversion process on the Sandy Creek data volume, we have accurately picked three coal seam horizons. The interactions between these are clear to see. As well, we have demonstrated the use of depth-converted seismic data in mapping a complex interaction between a normal and thrust fault system at Sandy Creek. In this case, the thrust postdates the normal fault and that the thrust levels out in the immediate roof of the working seam. Mining is not planned for this particular area, but were it to happen, roof instabilities could well result.
The development of seismic trace attribute analysis has been further developed in this project. We have made a detailed analysis of the amplitude of the seismic reflections from the German Creek working seam at Sandy Creek and established that contrary to theoretical predictions, the amplitude of the seismic events are not related to the thickness of the seam. We attribute this to the seismic processing procedures employed to produce the seismic sections.
Derivation of amplitude and other attribute data that can be related to thicknesses and lithological properties require seismic data that are processed with their amplitudes preserved - both within a trace and between traces. If this can be done, the seismic acoustic impedance algorithm that we have developed will create the opportunity for direct lithological interpretation consistent with geophysical logs and other geological data. We have demonstrated the process on a synthetic example and applied it to a section of the Sandy Creek data. While the creation of accurate depth-converted seismic data represents a major advance in the ability of the coal industry to utilise seismic data, seismic inversion on these depth-converted data now appears possible. If this can be achieved, then even greater benefits for geological mapping and prediction will be achieved.
Existing and new 3D seismic data sets from areas where mining is underway or planned should be converted to the depth domain to allow accurate assessment of coal seam levels and dips. The process developed in this project can be readily implemented with the program SeisWin. Whenever new boreholes are drilled, the depth conversion process should be redone to incorporate the new depth controls.
The depth control process allows horizons other than the target seam to be incorporated. Seismic interpretations should be made on these horizons so that the depth converted data are accurate for the entire geological section and not just the target seam. Faults generally affect more than just the target seam. Interpretations of the other horizons and the fault planes themselves allow better assessment on the nature of the faults and their interaction with other structures and the lithological section. This will allow better assessment of their potential to influence mining conditions. It is recommended that these tasks become adopted as part of routine seismic interpretation.
As a result of this project, the potential has been identified for (i) even better fault analysis based on patterns of throw and (ii) lithological interpretation and seam thickness mapping integrating borehole logs and depth-converted seismic data. To achieve the fault analysis, a systematic study into the styles of faults as revealed by seismic data and geological mapping is required. The seismic inversion and attribute analysis required for the lithological mapping requires the adoption of new seismic processing procedures prior to interpretation. Further research in both of these areas is recommended.