We used densities and compressional wave velocities combined from index properties, MST measurements, and a velocity log (Fig. F94) to synthesize seismograms at Site 1139. A detailed comparison of velocities from samples and logs is given in "Downhole Measurements" in the "Site 1137" chapter. Based on the coherence analysis carried out for the velocity data from Site 1137 (Fig. F98), we filtered the sample velocities from Site 1139 using a filter of 5 m length. The data were resampled simultaneously every 2 m, and data gaps >5 m were linearly interpolated. More closely spaced sampling results in the inclusion of noise in the filtered profiles. We used a robust-mode filter (i.e., a maximum likelihood probability estimator) that calculated the mode within the given data window. In addition, the median of the filtered data set was computed during filtering, and outliers, whose values exceed 2.5 times the L1 scale, were replaced with the median, as described in "Downhole Measurements" in the "Site 1137" chapter.
Downhole velocity logs are extremely noisy at this site because of bad weather conditions during logging. Further, velocities from discrete samples and downhole logs are offset for the entire length of the log, with the latter displaying velocities substantially lower than those from discrete samples, on average by >0.5 km/s (see "Physical Properties"). The origin of this offset is unknown. Within basement, below 460 mbsf, it is difficult to determine whether the two data sets are offset, as only a few sample measurements are available (Fig. F94A). No sample velocity data exist between 385 and 530 mbsf. Therefore, we include basement log velocities from 490 to 575 mbsf in a composite velocity profile used for synthesizing seismograms (Fig. F95A). In this depth interval, the smoothed velocity and density profiles correlate, with both showing substantial variations between 530 and 575 mbsf (Fig. F94C). The uppermost of the composite velocity log is based on sample velocities, except for depths from 490 to 570 mbsf, where the offset between log and sample velocities appears to be small, based on samples from 550 and 575 mbsf.
Nevertheless, artifacts likely exist where we have sparse data. The negative slope that results from linearly interpolating through the velocity data gap from 385 to 490 mbsf (i.e., between a sample velocity at 385 mbsf and a log velocity at 490 mbsf) results in an apparent velocity inversion (Fig. F95A). This inversion is most likely an artifact, as velocities from the downhole log are apparently lower than formation velocities. We make no attempt to correct the velocities. The shift may or may not be constant with depth, but given scant sample velocity information at depths between 380 and 580 mbsf, the exact nature of the shift remains unknown. Therefore, we use the logged basement velocities, as their relative variation include some information useful for computing impedance contrasts. Sample velocity data within basement exhibit three velocity inversions at 630-645, 655-660, and 665-670 mbsf (Fig. F94C), all of which result in reflections that tie to the MCS data (Figs. F95, F96).
GRAPE and discrete sample densities agree fairly well at depths above ~170 mbsf. At greater depths, all GRAPE density determinations are lower than those from discrete samples. No densities are available between 380 and 530 mbsf. Also, we have no downhole density log for this site because of bad weather conditions. We smoothed and resampled sample densities similarly to the velocities (see discussion above).
We synthesized seismograms for Site 1139 (see "Seismic Stratigraphy" in the "Site 1137" chapter). We resampled both densities and velocities every 0.5 ms as a function of two-way traveltime (TWT) (Fig. F95B), and we created profiles for impedance, reflection coefficients, and a seismic trace (Fig. F95C, F95D, F95E, F95F, respectively). The seismic trace is based on convolution with a Ricker wavelet with a peak frequency of 40 Hz. Reflection coefficients with and without multiples and transmission losses do not show distinct differences at this site, and those differences that exist should not be overinterpreted given the variable sampling rate with depth and poor quality of some data.
Despite variable data quality (Fig. F96B), we can link most major MCS reflections with the lithostratigraphy. The lower to middle Miocene section includes three reflections labeled M1-M3, mainly because of density contrasts within Unit II at ~65, 170, and 170 m. The reflections at 65 and 170 mbsf correlate with increases in CaCO3 content (see "Lithostratigraphy").
Three reflections in the mid-Oligocene section (labeled O1-O3 at 375, 340, 305 mbsf, respectively) are caused by both density and velocity contrasts. All reflections correlate with increases in CaCO3 content (see "Lithostratigraphy"). The increase in CaCO3 content related to reflection O1 marks the transition of nannofossil chalk in Unit II to a sandy packstone in Unit III.
Density and velocity data are not available from Unit III downward through Units IV and V into the underlying basement. However, the synthetic seismogram shows five prominent reflections (B1-B5) within basement (Fig. F95F). The position of the uppermost of these reflections, B5, marks the transition from a data gap to that part of the basement section where data are available. Therefore, its stratigraphic position is not meaningful. However, the upper inflection point of B5 roughly corresponds to the top of felsic basement at 460 mbsf (Fig. F95F). The four remaining basement reflections are at depths of ~615-620 mbsf (B1), ~605-610 mbsf (B2), ~535-540 mbsf (B3), and ~500-505 mbsf (B4). B4 is associated with a velocity inversion of unknown cause as a result of a lack of core recovery within Subunit 1D, B3 is caused by a velocity inversion resulting from a low-velocity volcaniclastic sand (Unit 3) between trachyte units (see "Physical Volcanology"), and B2 and B1 are associated with velocity and density contrasts within basalt flows (Units 6-17).
All reflections within basement and those identified by synthetic seismograms are deeper than those on the MCS data because the velocities and densities of the packstone overlying basement are not constrained. In particular, we believe that velocities have been underestimated, resulting in overestimated two-way traveltimes for all reflections underlying the data gap (i.e., all reflections within basement). However, the relative two-way traveltimes of reflections B1-B5 are roughly correct and correlate well with five reflections in the MCS data (Fig. F96).
The observed match between the reflections from synthetic seismogram and those on the MCS data allows us to re-evaluate the cause for the shift between velocities from downhole logs and discrete samples from 100 to 390 mbsf (Fig. F94). If we would use log velocities for synthetic seismogram construction for this depth interval, instead of sample velocities, then the synthetic seismic trace would be more than 100 ms longer compared to the traces shown on Figure F95F (see Fig. F95 for comparison of TWT-depth relationships of Sites 1139 and 1140). The resulting match of synthetic basement reflections with observed reflections in the MCS data would be extremely poor. Based on this observation, we argue that it is likely that the shift between log and sample velocities is caused by incorrect log velocities, possibly a result of an extremely large hole diameter. As no caliper log is available because of the bad weather conditions, the hole size remains unknown. The above conclusion is corroborated by a qualitative evaluation of extremely low log velocities of 1.55-1.6 km/s obtained at depths between 110 and 150 mbsf, where clays and relatively consolidated claystones were recovered (see "Lithostratigraphy"). These lithologies are more consistent with velocities around 2 km/s, as determined by velocity measurements on discrete samples (Fig. F94).