Previous   |    Next

doi:10.2204/iodp.proc.314315316.113.2009

Log-seismic correlation

Overall logging unit correlation

Logging Unit I corresponds to the hemipelagic sediments between the seafloor and ~200 m seismic depth below seafloor (SSF) (Fig. F29). Logging Subunit IB corresponds to a thin low-amplitude reflection overlying the strong positive reflection at 200 m SSF that defines the base of logging Unit I. Logging Unit II corresponds to a zone of southeast-dipping, generally low amplitude reflections across these boundaries. Reflections intersect the borehole at the logging Subunit IIA/IIB and IIB/IIC boundaries, but there is no change in the general character of the reflections. The boundary between logging Units II and III correlates with a change in reflectivity from low amplitude above the boundary to high amplitude below the boundary. Logging Subunit IIIA corresponds to a series of high-amplitude, laterally continuous reflections that appear on the southeast side of the borehole that are cut off by the inferred fault that intersects the borehole at the base of this subunit.

Check shot survey data

Check shot data at Site C0001 were acquired at 16 depths in Holes C0001C (1 depth) and C0001D (15 depths) (Fig. F47; Table T8). These data sample depths from the seafloor to 635 m LSF. Attempts to acquire data at six stations deeper in Hole C0001D were unsuccessful because the battery in the seismicVISION tool failed. Between 3 and 20 (typically 10) air gun array shots were fired at each station. Noisy traces and traces with poor first arrival waveforms were deleted. The remaining traces were filtered (trapezoidal, minimum phase, and 30-40-150-200 Hz band-pass) and stacked to produce the traces shown in Figure F47. The first arrival wavelet is unambiguous on all traces, although noise makes identifying the true first break difficult on a few of the traces.

The first arrival times were picked manually as illustrated in Figure F47. These are the “raw first arrival” times in Table T8. We applied a damped least-squares inversion to the observed depth-time data (Lizarralde and Swift, 1999). This inversion determines a smooth velocity versus depth curve by varying the arrival times by amounts that are within their uncertainty. We estimated the uncertainty of the arrivals to be ~0.2 ms. We used a damping coefficient of 0.05 because it produced a chi squared (χ²) value consistent with the optimal balance between over- and underfitting the data. The smoothed interval velocities and adjusted arrival times are shown in Table T8. The improvement in estimated interval velocities, indicated by the smoothness of the curve and the general downward increase of velocity (Fig. F48), is dramatic. The accompanying changes in arrival times are very small. We used the smoothed arrival picks and the tool depths as the check shot curve, which we then used for synthetic seismogram preparation.

The smoothed interval velocity curve and sonic P-wave velocity data are generally similar (Fig. F49). The largest mismatch is from 0 to 175 m LSF, where the check shot curve is ~60–80 m/s faster than the sonic log. In this interval, the processed sonic log data did not show an arrival that could be reliably separated from the mud arrival (see “Data and log quality”). Therefore, we interpret that the check shot velocity is a better representation of the long wavelength velocity depth function in this depth range. Both the check shot and sonic log velocities show a general leveling of velocity at ~500 m LSF then a resumption of the downward increasing trend below ~550 m LSF.

We were not able to construct a meaningful vertical seismic profile (VSP) using these data. We tried a number of filtering and gain combinations but could not identify coherent upward traveling reflections. We attribute this to noise from the vibration of the drill pipe in the current and other sources of downhole noise.

Density log data

Density data, along with P-wave velocity data, are necessary to calculate a synthetic reflection seismogram. At Site C0001, density data were logged using the adnVISION tool; however, the tool was damaged during the operation and we could not retrieve any data from its memory until the final days of the expedition, too late to be used for this analysis. During the LWD operation, a subset of the data was sent to the surface by mud pulse from the bottom in real time (see the “Expedition 314 methods” chapter). Figure F50A shows these real-time bulk density data (RHOB), which were obtained in Hole C0001D from the seafloor to ~500 m LSF. However, the data in the 15–45 m LSF interval are noisy. To improve the log for use in synthetic seismogram generation, we spliced in the real-time density data from Hole C0001C from above the seafloor to 49 m LSF and assumed a constant density to the sea surface (Fig. F50B). The density curve of Hole C0001C was shifted up by ~1 m to correlate with that of Hole C0001D at the seafloor.

The merged Hole C0001C/C0001D real-time density log is well behaved to just above a 200 m LSF zone of interest that correlates with a strong positive impedance reflector on the 3-D seismic data. The density log, however, varies strongly, with obviously incorrect (as low as 1.1 g/cm³) density values at this boundary. During synthetic seismogram generation, these variations produce a series of emergent reflections just below the boundary. The real-time density log below the 200 m LSF boundary continues to be noisy; based on the real-time ultrasonic caliper data the real-time densities below this boundary may be suspect (see “Data and log quality”).

To produce a more realistic density log that does not produce spurious reflections on the synthetic seismogram at the 200 m boundary and below, we created a pseudodensity log derived from the thermally corrected ring resistivity–derived porosity log (see “Physical properties”). To obtain a formation matrix density, the real-time density-derived porosity data were cross-plotted against the ring resistivity porosity data; where the values were approximately equal to each other, it was assumed that the values were real. The corresponding logging density values were then used to derive a matrix density of 2.64 g/cm³, and the fluid density was assumed to be that of seawater (1.03 g/cm³). The matrix and fluid density values along with the resistivity-derived porosity were then used to derive the pseudodensity curve, based on the following equation:

where ρ_b is the pseudodensity, ρ_ma is the matrix density, ρ_f is the fluid density, and ϕ is porosity. The resulting pseudodensity log is a smoother curve than the real-time density log (Fig. F50C).

We merged the spliced real-time density curve (Fig. F50B) and this pseudodensity curve (Fig. F50C) into a preferred density curve designed to produce the fewest spurious reflections in the resulting synthetic seismogram. To create the composite density curve, we used the spliced real-time data from the sea surface to just above the unrealistically low density values in the 200 m LSF zone and pseudodensity data for the 200 m zone and from this zone to the bottom of the hole (Fig. F50D).

Sonic log data

The sonicVISION tool is designed for use in higher velocity materials than those commonly encountered while drilling slope and accreted sediments. The Schlumberger DCS representative processed the raw sonic waveforms on board using a range of filters and interpreted the formation slowness based on a mixture of mixed process (MP) wide and leaky-P mode processed waveforms (see the “Expedition 314 methods” chapter). The resulting slowness values are most reliable in the areas of low noise and where the velocities are distinct from the borehole mud velocities, which are in this case close to that of water. For example, in Hole C0001D the slowness values above ~175 m LSF (Fig. F49) are suspect, as are the values in zones with poor hole conditions or low signal-to-noise ratio. The quality of these logs was qualitatively determined by examining the waveforms and coherency plots, referencing the method of processing used for the picking, and assessing whether the data are either noisy or close to the mud arrival. Areas where hole conditions were poor and/or drilling operations may have affected the quality of the derived slowness values are detailed in “Data and log quality.”

The picked sonic values produced a sonic log that was then calibrated using the check shot data for the site. For Hole C0001D, the seismicVISION tool stopped working after an observation at 635.2 m LSF and acquired no data deeper in the hole. We base the time-depth relationship below the deepest check shot values by assuming a linear drift curve from ~635 m LSF to total depth (Fig. F51A). The explanation of the use of drift curves to calibrate sonic logs is available in the “Expedition 314 methods” chapter. The calibrated sonic log differs most significantly from the picked sonic log in the shallow areas (shallower than ~175 m LSF) in which the calibrated curve has significantly higher velocities.

Synthetic seismogram

Using the favored density log (Fig. F51C), created by splicing the real-time density from Holes C0001C and C0001D, the pseudodensity curve, and the calibrated sonic log (Fig. F51B), corrected using the long-wavelength component of the time-depth relationship determined by check shots (Fig. F51A), we generated a synthetic seismogram (Fig. F51D). The synthetic seismogram in general agrees well with the reflectivity of the 20 cross-line traces on Inline 2480 in the vicinity of the borehole (Fig. F51G) for the regions from seafloor to ~192 m LSF. However, the 8 m zone of low densities and slightly lower velocities just above 200 m LSF produces a negative polarity reflection in the synthetic seismogram that does not agree with the seismic data that instead exhibit a strong positive polarity reflection with evidence of composite tuned reflection character at its base. Additionally, from 200 to ~528 m LSF, the synthetic seismogram exhibits a series of reflections not present in the seismic data that are largely transparent over this interval. The ~25 m cycles observed in the sonic log over this region may be the cause of this anomalous reflectivity (see “Structural geology and geomechanics” for discussion of the sonic log). In the synthetic seismogram, a series of strong reflections exist within the low-velocity zone from ~523 to 640 m LSF before transitioning to occasional reflectivity below this zone. There is a general increase in reflectivity in the seismic data around the depths of the low-velocity zone and through the bottom of the hole. One additional anomalous reflection is present in the synthetic seismogram at ~901 m LSF that is generated by the change to higher velocities in the deepest parts of the hole; however, this change may be a function of increased signal-to-noise ratio rather than a true change in velocity, as evidenced by the lack of such a strong reflection at that depth on the seismic data.

Strong reflector at 200 m LSF

A prominent reflector is observed in the 3-D prestack time-migrated seismic volume (Moore et al., 2007) at a depth of ~3.2 s two-way traveltime around Site C0001, which separates the package of reflective slope sediments from an underlying seismically transparent unit (Fig. F46). This reflector approximately correlates with the boundary between logging Units I and II at ~200 m LSF. The reflector is traceable in the cross-lines of the 3-D seismic volume across the entire ~12 km width of the volume (Fig. F52). This reflector exhibits a strong positive polarity waveform (Fig. F53), similar to that expressed at the seafloor, which requires an increase in acoustic impedance (velocity × density) with depth.

The synthetic seismogram computed using the calibrated sonic log and the best available density log (Fig. F51) shows a negative polarity reflection at the ~200 m LSF boundary (Fig. F54A), which does not match the observed reflector in the 3-D seismic data. To test the sensitivity of this reflector to the specific densities within the 8 m zone above the ~200 m LSF boundary, we created a series of models based on the pseudodensity curve with density lows to 1.4 g/cm³ at the 200 m LSF boundary in an attempt to match both the low resistivity and low density values recorded in that zone (Fig. F54B–F54D). All of these tests produce a negative polarity reflection. Even removing the 8 m zone entirely produces a small negative polarity reflection simply because of the density decrease (by 0.13 g/cm³) at the logging Unit I/II boundary (Fig. F54E). With the strong possibility that the 8 m zone had poor hole conditions (see description of the real-time caliper in “Data and log quality”) and the observation of a velocity and density increase just above this zone, we ran an additional test where the 8 m zone is instead a density high (Fig. F54F). This model more appropriately reproduces the strong positive reflector from the top of the zone but still produces an erroneously large second reflector off the base of the zone. Our final and best synthetic model at this site is one where the top of the 8 m interval is a density peak and the base of the interval returns to lower density values over ~16 m (Fig. F54G). Both this model and the previous model use a 2.1 g/cm³ density value based on the hemipelagic muds of Site 808. Our final model results in a positive reflection at the top of the 8 m thick zone and a complicated response at the base of the 200 m LSF layer, consistent with the local seismic reflection character.

Top of page   |    Previous   |    Next