Previous   |    Next

doi:10.2204/iodp.proc.314315316.115.2009

Log-seismic correlation

Seismic observations and correlation with logging units

Based on the Kumano 3-D seismic volume (Moore et al., 2007), Site C0003 was spudded into the edge of a slope sediment sequence that overlies the thrust sheet of one splay in the megasplay fault system. In detail, this thrust sheet is itself faulted by three additional imaged faults local to the borehole, with additional subseismic-scale faulting being likely.

A prestack depth-migrated (PSDM) dip-parallel image across the site shows a slope sediment sequence with reflectors largely parallel to the seafloor lying unconformably above an older accreted sequence; the contact is a strong positive polarity (same as the seafloor) reflection (Fig. F9). Below this unconformity is a wedge-shaped sequence with an apparently subhorizontal, negative polarity reflection in the center of the wedge and a dipping negative polarity reflection at its base. Two faults that cut through the thrust sheet can be traced to near this wedge-shaped sequence. The tip of the shallower fault cuts toward the seafloor on the northwest side of the sequence; the deeper fault can be traced to near the base of the wedge, but the relationship of this fault to the reflector at the base of the wedge-shaped sequence is unclear (Fig. F9).

The basal reflection from this wedge-shaped sequence displays angular relationships with the underlying reflections. Below it is a low-reflectivity sequence with a thick fault zone at its base. A single prominent reflection within this low-reflectivity sequence lies at ~250 m LSF and could be structural or stratigraphic. The fault zone penetrated near the base of the borehole may be as thick as ~100 m. The fault zone reflectivity is complicated and laterally variable; at the borehole the highest amplitude response is near the base of the fault zone (Fig. F9).

The logging units correlate well with our seismic observations. Logging Unit I lies at the base of the slope sediment section and the logging Unit I/II contact is likely the positive polarity reflection separating the base of the slope sediments and top of the wedge-shaped sequence. The logging Unit II/III contact is the negative polarity reflection at the base of the wedge-shaped sequence. Logging Unit III spans the low-reflectivity sequence including the fault zone to the bottom.

Log-seismic correlation in zones of interest

Logging Unit I/II boundary and the base of slope sediments

The logs through the slope sediments (logging Unit I) differ strongly from logs in the underlying logging Unit II and help to explain the strong reflection that separates these sequences in the seismic data. The density (Fig. F19), gamma ray (Fig. F20), and resistivity (Fig. F21) logs show strong decreases across the boundary, whereas the neutron porosity (Fig. F22) and caliper (Fig. F23) logs show strong increases across the boundary. The sonic log (Fig. F24) included bad values for all of logging Units I and II, as discussed below. The reflection separating the slope sediments from the wedge-shaped sequence is positive; however, the density log decreases sharply across this boundary. The high caliper values throughout logging Unit II, however, suggest that the density values in Unit II are suspect because of likely washout (Fig. F9), and thus based on the strong positive reflector we anticipate that the density and velocity values within Unit II should be higher than in Unit I.

Logging Unit II and the wedge-shaped sequence

The low gamma ray and resistivity values combined with the high caliper values for logging Unit II are consistent with a sand-rich sequence (Figs. F20, F21, F23). The base of this unit is a strong negative polarity reflection requiring a decrease in velocity and/or density across the boundary (or inherent to the boundary). This observation suggests that the base of logging Unit II consists of higher density material than Unit III. Additionally there is a low-density interval ~3 m below the logging Unit II/III boundary. The combination of these features may explain the negative polarity reflection. The nature of this reflection is addressed through seismogram modeling described below.

Fault region

From 381 to 470 m LSF are a series of washouts (Fig. F9) interpreted based on the ultrasonic caliper (Fig. F23). This zone correlates with reflections that appear related to the fault zone in the thrust sheet. Interestingly, the brightest reflection within the fault-related zone of reflections lies at and just below the lowest of these washouts in a region where the densities are first quite low and then are higher (Fig. F19). Based on the seismic data, the base of the washouts in this zone is a possible location for the active part of the fault zone and the bulk of the washouts is a large damaged zone above it or perhaps a series of faults with overlapping damaged zones. The low gamma ray response across the damaged zone or series of faults suggests that either brecciation associated with faults or a sandy layer may be the cause of the washouts.

Check shot survey

Because the seismicVISION tool and its memory data were lost in the hole, we created a check shot survey using the real-time data transmitted to the surface during drilling. These data, for each depth station, consist of a 300 ms seismic trace window, comprising 15 shot records stacked downhole (Fig. F25). The tool also makes and transmits three first arrival picks and some information about their likely reliability. The operator was provided a predrilling forecast of the subsurface velocity and one-way traveltimes. During Expedition 314, these predictions were based on the prestack depth migration velocity model at each site. These predictions and the transmitted waveform are used by the operator to determine which pick, if any, is reliable. The operator may choose to accept the first pick, to accept one of the other picks, or to make a new pick from the seismic trace.

Twelve stations were occupied in Hole C0003A. For two of these stations, the waveform data were corrupted during transmission to the surface. Therefore, we used data from 10 stations ranging from 86 to 506 m LSF (Table T5). The first arrival waveforms from these stations are of high quality and we believe that the first arrival picks are nearly as reliable as those obtained using the memory data for other sites (Fig. F25). At seven of the stations, the operator accepted the first tool pick. At the other three stations, the operator made a new pick from the seismic trace. As at the other sites, corrections were made for source time and hydrophone depth. The data were used as for the other sites to obtain long wavelength interval velocities in the hole and to correct the PSDM seismic section at the borehole. The corrected PSDM section was used as the background in Figures F9, F19, F20, F21, F22, F23, and F24.

The check shot times were adjusted using the method of Lizarralde and Swift (1999), assuming a traveltime picking variance of 1 ms and a smoothing parameter of 0.2. We used a somewhat higher smoothing parameter than at other sites because we do not have the same level of confidence in the picks and because there were fewer data points. The result is a very smooth velocity curve. Although the smoothing parameter could be reduced and slightly more character teased out of the velocity-depth function, we do not think that it would be statistically justified.

We plotted the smoothed check shot interval velocities with the real-time sonic log velocities (Fig. F26). Unlike the sonic velocity data from other sites, these data were picked by the tool downhole. Only the arrival time picks were sent to the surface; no waveforms were transmitted or otherwise available for surface analysis. For the other sites, extensive reprocessing and manual interpretation of the sonic waveforms were performed. Therefore, we do not expect the real-time sonic log velocities to be reliable. The data show that above ~151 m LSF in logging Units I and II, the sonic log velocities are scattered and much faster than the check shot velocities. We suggest that the sonic velocities in this region should be ignored. Below ~151 m LSF in logging Unit III, the real-time sonic log velocities are also scattered but the check shot velocity function generally corresponds to their upper bound. Although there seems to be some relationship between the deeper real-time sonic log velocities and the check shot data, we do not think that the sonic velocities can be considered reliable.

Without the longer seismic records recorded in the seismicVISION tool, it is not possible to calculate a vertical seismic profile.

Seismogram modeling

Good quality P-wave velocity data from sonicVISION tool are not available because memory data to be reprocessed were not retrieved at this site. Furthermore, the real-time density data in logging Unit II do not reconcile with the polarity of the seismic reflection data as discussed above. These problems meant that we could not generate a meaningful synthetic seismogram from logging data at this site. We therefore tried to calculate seismograms to fit with the observation of the seismic profile. The interpolated smoothed interval velocities from check shot data were used as P-wave velocities (Fig. F27B) after slight modification within the water column and depths shallower than ~43 m LSF. The real-time density data were edited as described below. First, we removed the singular density value of 1.82 g/cm³ at 44 m LSF, which is significantly larger than the average density value of ~1.5 g/cm³ at this depth. Next, the real-time density data were edited to remove densities associated with large ADIA values. We chose a cutoff value of 10.5 inches for this purpose after testing values of 9.5, 10.0, and 10.5 inches. To correlate the calculated seismogram with the seismic section we modeled the density data at the logging Unit I/II and II/III boundaries. We assumed that density data increase from 1.62 to 1.72 g/cm³ at 71 m LSF, which is ~5 m shallower than the Unit I/II boundary, and decrease from 2.10 to 1.97 g/cm³ at the Unit II/III boundary. The density data within Unit II were not modeled and thus were interpolated linearly. The modeled density curve (Fig. F27C) was used for seismogram modeling. We used a zero-phase 256 ms wavelet retrieved from cross-line prestack time-migrated (PSTM) section at the vicinity of the site. The modeled seismogram is shown in Figure F27D. Note that the modeled seismogram senses only the density data because P-wave velocity is smoothly varying. The modeled seismogram matches with the major features of the seismic section (Fig. F27G–F27I). Both a positive reflector at the boundary between logging Units I and II and a negative amplitude reflection at the base of logging Unit II are reproduced in the modeled seismogram. The depth difference of these reflectors between the depth-converted PSTM section and the modeled seismogram was <5 m.

The modeling result supports the possibility that density values increase at the Unit I/II boundary and decrease at the Unit II/III boundary. The reflective zone at ~500 m LSF in the seismic section can be correlated with the large-amplitude wave packets in the modeled seismogram; however, we cannot exactly fit the modeled seismogram to the seismic section. This is probably caused by the low quality of real-time logging data. Note that because we assume the density and velocity data are smoothly increasing in logging Unit II, we lose any obvious reflection events within this unit.

Top of page   |    Previous   |    Next