Because of the low S-wave velocity in these formations (below the compressional velocity of the borehole fluid), S-wave information must be extracted from the flexural wave at its low-frequency cutoff (Winbow, 1985). The excitation of this mode depends on the S-wave slowness of the formation, the borehole size, and the resonant frequency of the flexural mode from the DSI tool. Modeling of the flexural mode poles under the conditions encountered at Site 1173 is shown in Figure F4. A strong resonant peak of the flexural mode at ~780 ms/ft occurs below 1000 Hz. Energy decreases at slowness values away from this peak and at frequencies above 2500 Hz. Other minor peaks occur at low slowness values and are due to scattered energy from the dipole source in the DSI tool. Thus, in order to eliminate the effect of resonant tool energy, the raw dipole waveforms were filtered over a frequency range from 250 to 750 Hz to extract the flexural mode only. This bandpass filter was run three times to reduce windowing edge effects.
The modeling of the flexural wave also shows that dispersion only occurs above ~500 Hz under these conditions. Correction of the flexural wave dispersion is therefore not required for these data (e.g., Brie and Saiki, 1996). Other late-arriving modes in the dipole waveform, such as fluid wave and Stoneley modes, are eliminated by selecting a variable processing time window length of 9,999-11,499 ms and a fixed slowness range from 280 to 1080 ms/ft. Bandpass filtering and time windowing are performed to extract the flexural mode of the dipole waveforms from both logging passes. The S-wave slowness is determined automatically by picking the arrival time of the flexural wave from the filtered data at 16 receivers (orthogonal dipole receivers at the eight stations). In general, pass 2 data have higher signal-to-noise due to the use of the lower-frequency dipole source. The signal-to-noise energy ratio is sufficiently high to obtain results from 226 to 350 mbsf for pass 1 and over the entire interval for pass 2.
Below 350 mbsf, however, only receiver stations 1-5 were selected in the pass 2 data and 32 points/cycle were used for phase picking. The flexural mode is so weak in this interval that the farthest three stations along the tool array have insufficient signal-to-noise ratios for picking and were eliminated before processing. The weak flexural mode energy in this interval also limited the processing to ~80% of the first period and required that a higher sampling density be used. Testing indicates that using only five stations and the shorter sampling period agrees well with the result of normal processing in other intervals where the flexural wave signal is strong.