PPD-WE Primary Source Wave Extraction

BCE’s PPD-WE proprietary algorithm was implemented on seismic data acquired from the downhole seismic testing (DST) application of seismic cone penetration test (SCPT). BCE’s PPD-WE is a general purpose algorithm for blind deconvolution that is particularly well suited to source wave estimation when no assumptions concerning the reflectivity can be made. For example, a minimal set of reflection coefficients is not conducive to assigning an associated statistical pdf to the reflection series. Hence real data from a SCPT is analyzed due to the sparseness of the reflection coefficients.

When processing SCPT time series data it is of paramount importance to extract the first arriving source waves from each trace so that the crosscorrelation is meaningful when obtaining relative arrival times. Post ground improvement (e.g., Vibrocompaction / Stone Columns) typically results in complex stratigraphy where source wave multiples are generated during a SCPT or any other type of DST investigation. These multiples complicate the recorded time series making the selection of interval arrival times a difficult task.

The data presented in this web page was acquired with a triaxial system configuration. A triaxial sensor configuration is utilized so that full waveform analysis can be carried out and the possibility of rod rotation can easily be taken into account. For the SH wave analysis, the preprocessing of the seismic time series data captured on the X and Y axes is three-fold. 1) Apply bandpass frequency filter. 2) Rotate the X and Y axes responses onto the full waveform axis utilizing hodograms and polarization analysis. 3) Apply exponentially decaying windows to the front and end of the seismic responses on the recorded time series so that the S/N is increased.

Figs. 1 and 2 illustrate raw X and Y axes, respectively, vertical seismic profiles (VSP) captured from a SCPT. As is evident from Figs. 1 and 2, there are source multiples present within the captured time series data. The overlapping source waves present significant difficulties in implementing the crosscorrelation technique. To address this problem the PPD-WE algorithm is utilized to extract the first arriving source waves from the preprocessed time series and subsequent to this the crosscorrelation technique is implemented so that relative arrival times can be obtained.

BEFORE:

Figure 1. Raw X axis VSP from SCPT.

Figure 2. Raw Y axis VSP from SCPT.

Fig. 3 illustrates the extracted (utilizing the PPD-WE technique) primary source waves VSP which were estimated after preprocessing the data shown in Figs. 1and 2. As is evident from Fig. 3, the PPD-WE technique did a very impressive job in extracting the primary source wavelets. The crossocorrelation technique can now be implemented so that accurate interval SH wave velocities can be obtained.

The estimated interval velocities from the time series data given in Fig. 3 is outlined in Table 1. The traces shown in Fig. 3 have very high calculated interval correlation coefficients as shown in Table 1.  It should be noted that the shear and compression wave velocities are squared in deriving the elastics constants; therefore, variations in the estimated velocities can cause appreciable errors in the calculation of the elastic constant.

AFTER:

Figure 3. PPD-WE extracted primary source waves VSP.

TABLE 1

Interval velocities (PPD-WE extracted source waves) from the crosscorrelation technique

References

1)  Baziw, E. (2007). Measurement of Low Strain Elastic Constants in Complex Stratigraphy. Submitted to the Canadian Geotechnical Journal.

2) Baziw, E. (2007). Application of Bayesian Recursive Estimation for Seismic Signal Processing. Ph.D. Thesis, Dept. of Earth and Ocean Sciences, University of British Columbia.

3)  Baziw, E. (2007). Implementation of the Principle Phase Decomposition Algorithm. IEEE Transactions on Geoscience and Remote Sensing (TGRS), vol. 45, No. 6, pp. 1775-1785, June. 2007.

4)   Baziw, E. and Ulrych, T.J. (2006). Principle Phase Decomposition - A New Concept in Blind Seismic Deconvolution. IEEE Transactions on Geoscience and Remote Sensing (TGRS), vol. 44, No. 8, pp. 2271-2281, Aug. 2006.