(Tim Brown and C. Liner)
See also: Spectral decomposition at Dickman (2)
In an ongoing effort to effectively identify geological features within our seismic volume it is important to have reliable filtering capabilities. Filtering techniques have been applied to our seismic volume using both Seismic MicroTechnology (SMT) and Seismic Unix (SU) to aid in identifying geological anomalies. However, significant differences arise when attempting to use narrow band-pass filters within SMT, especially at low frequencies.
After applying a trapezoidal filter centered on 5 Hz (Figure 1 lower left), it is clear that residual frequencies still reside in the data. The frequency spectrum of the 5 Hz data supports this observation (Figure 1 lower right).
Application of a similar isolation filter in SU yields something closer to a pure 5 Hz decomposition. This 5 Hz attribute was imported into SMT and the results are displayed in Figure 2. The seismic section shows a low-frequency character expected of a 5 Hz decomposition, and this is supported by its associated frequency spectrum (Figure 2 lower right).
We compare the results from the two filtered outputs to emphasize these are not esoteric differences. Figure 3 shows an 848 ms time-slice from the SU 5 Hz data, clearly indicating linear features present in the center portion of the survey area. However, these linear features are absent in the SMT 5 Hz time-slice (Figure 4). This discrepancy between the two data sets is problematic because using narrow-band filters at low frequencies in SMT becomes an unreliable method to enhance these important features that may be related to fracturing. In our case, it is important to have a functioning isolation filter because the linear features are only evident at these lower frequencies when using spectral decomposition.