Previous   |    Next

doi:10.2204/iodp.proc.341.204.2017

Methods and materials

Physical property data

Expedition 341 collected core and logging data on the continental shelf and slope at Sites U1420 and U1421 (Fig. F1). During drilling, the shipboard science party performed measurements of various physical properties at multiple scales and resolutions. After the core came on deck, the whole-round core passed through the Special Task Multisensor Logger (STMSL) where gamma ray attenuation (GRA) bulk density and magnetic susceptibility measurements were collected. After a period of temperature equilibration, the core was passed through the Whole-Round Multisensor Logger (WRMSL), which recorded GRA bulk density, magnetic susceptibility, and compressional wave velocity (using the P-wave velocity logger [PWL]). Natural gamma ray (NGR) measurements were taken before the splitting the cores into working and archive halves. The archive half then passed through the Section Half Multisensor Logger (SHMSL) where reflection spectroscopy, colorimetry, magnetic susceptibility, and laser split-core surface analyses were performed. From the working halves, point measurements of compressional wave velocity were taken with the P-wave caliper (PWC) instrument and discrete samples were collected for moisture and density (MAD) measurements, which consisted of bulk density, water content, porosity, and grain density. This process is described in detail in the “Methods” chapter (Jaeger et al., 2014a) and is schematically illustrated in Figure F2. Core recovery was hindered at both sites by large clasts that jammed recovery equipment and resulted in 139.91m of recovered core (14%) over the 1020.8 m interval at Site U1420 (see the “Site U1420” chapter [Jaeger et al., 2014b]). At Site U1421, three holes were drilled, with Hole U1421A reaching total penetration of 1432.4 mbsf (driller’s seafloor depth = 729.7 m), and total recovery from all holes at Site U1421 was ~176 m.

At Sites U1420 and U1421, shipboard scientists deployed a wireline tool string, which included NGR, sonic velocity, and resistivity tools (see the “Site U1420” and “Site U1421” chapters [Jaeger et al., 2014b, 2014c]). As a result of concerns about borehole stability during logging, the density tool with its radioactive source was not included in this tool string; thus, no in situ density measurements were made at Sites U1420 and U1421. The logged interval, excluding the bottom-hole assembly (BHA), at Site U1420 was 89–288 m wireline log matched depth below seafloor (WMSF) due to borehole instability (see the “Site U1420” chapter [Jaeger et al., 2014b]). At Site U1421, the logged interval was below the BHA at 92 m WMSF and extended to 695 m WMSF. At Site U1421, we used vertical seismic profile (VSP) constraints from six stations in the borehole to constrain the TDR between 1.278–1.641 s two-way traveltime (TWT) and 284.7–687 m WMSF (Table T1).

More details about core handling, logging data collection, and processing methods as well as tool string and instrument technical data can be found in the “Methods” chapter (Jaeger et al., 2014a). Table T2 summarizes site data including location and physical properties measurement availability by source. Logging summaries and plots are shown in Figure F3. Depth scales and datums are described in “Stratigraphic correlation” in the “Methods” chapter (Jaeger et al., 2014a). We approached the well tie process based on the methods of White and Simm (2003) in a few steps, including the development of initial velocity models, synthetic seismic generation, wavelet extraction, generation of a new synthetic seismic trace, and finally, conservative visual seismic-to-synthetic matching.

Data conditioning and compilation

In general our strategy was to favor a log-centric approach to the data conditioning for two reasons: (1) the core data seemed to be noisy and had more outliers in comparison to the logging data and (2) we expect the in situ logged values to be more consistent with conditions reflected in the seismic data than measurements derived post situ. Core-based and in situ logging data may be different for a number of reasons, such as sampling rate, measurement depth of investigation, datum mismatch, borehole and core irregularity, and coring practices (e.g., see the “Methods” chapter [Jaeger et al., 2014a] and Daigle and Piña, 2016). For the purposes of this study, we applied minimal or no specific corrections for such factors in preparing the data compilations. Shipboard core-based measurements were collected in a variety of methods including track-based sensors, which measure properties of the whole or split-core; by analysis of discrete volumetric samples; and by using point-based instruments. Instrument response functions, the effects of core recovery technique (advanced piston coring [APC], extended core barrel [XCB], etc.), and a volumetric-based correction are discussed in Walczak et al. (2015). At Sites U1418 and U1417, the correction applied by Walczak et al. (2015) corrected for sediment compaction, variable recovered core diameter, and changes in the gas content of the sediment column but reduced the variance of NGR and bulk density measurements by ~20% and 50%, respectively, and after correction the largest changes in GRA bulk density were correlated with changes in coring technique. In order to preserve variability in the final synthetic seismic traces, our compilation treated all core-based data with equal weight and without adjustment for instrument effects or core recovery technique.

Site U1420 density and velocity compilation

At Site U1420, the bulk density data we included in the compilation were derived from WRMSL measurements where outliers (<1 g/cm³) were removed from the data set and then smoothed by a 19-sample moving average; typically this resulted in a moving window of 45 cm. STMSL bulk density data were not available at this site, but these measurements were at a lower spatial resolution than WRMSL GRA bulk density measurements, used similar sensors, were not given time for temperature equilibration, and were generally used for guiding drilling operations during the expedition (see the “Methods” chapter [Jaeger et al., 2014a]). Velocity data consisted of a logged interval with the Dipole Shear Sonic Imager (DSI) and PWC core measurements. We cleaned the PWC data of outlier measurements <1.5 km/s. We compiled log data on the WMSF depth scale, which is based on matching a spike in the gamma ray log with the seafloor. Core data were compiled on the core depth below seafloor, Method A (CSF-A), depth scale, which is an initial scale based on the advancement of the drill string during coring (see “Stratigraphic correlation” in the “Methods” chapter [Jaeger et al., 2014a]). Core-based CSF-A and logging WMSF depth scales were used together directly, which may introduce error because of a number of drilling effects (see the “Methods” chapter [Jaeger et al., 2014a]), which we also discuss briefly here.

Site U1421 density and velocity compilation

At Site U1421, we compiled bulk density data from core for the STMSL, WRMSL sources, and from Holes U1421A–U1421C, which we cleaned of outliers beyond 2 standard deviations of the mean for each instrument source. Next, we culled any remaining measurements <1 g/cm³. Finally, we smoothed these compiled density data with a 19-sample moving average (45 cm) on the core composite depth below seafloor, Method B (CCSF-B), depth scale, which was developed by splicing Holes U1421A–U1421C accounting for ship heave and applying an affine value to adjust for postrecovery expansion (see “Stratigraphic correlation” in the “Methods” chapter [Jaeger et al., 2014a]). Compressional wave velocity data at Site U1421 included both core and logging data. We combined core-based velocity measurements from the PWL, WRMSL, and PWC on the CCSF-B depth scale. The sonic log was used on the WMSF depth scale, and after removal of outliers beyond 3 standard deviations from the moving 40 m average, we combined it with the core-based measurements, which were shifted so that their mean was equal to the mean value from logging data in the overlapping interval. This shift to the core measurements, an increase of 53 m/s, served as a simple method to calibrate the wireline-logged and core-derived measurements in such a way to approximate in situ logged measurements (Fig. F4).

Data fitting

In Matlab R2015a software, we fit the cleaned and processed data described above by spline approximations of various smoothness. We employed a strategy of aggressively fitting the data in constrained intervals, which tended to create artificial data variability and introduce unrealistic predictions, such as bulk density <1 g/cm³, or negative velocity estimates in unconstrained intervals. To overcome these artifacts of overfitting in large poorly constrained gaps and often also where the only source was core-based data, smoother splines were used for interpolation. To create data sets comparable to those generated in the logging operations and create continuous physical property records for the full drilled interval, we resampled the compiled smoothed data at 0.1524 m spacing, consistent with the logging rate. The final interpolated curves, goodness of fit characteristics, and original measurements are shown in Figures F5, F6, F7, F8, and F9 and discussed in Results.

Initial velocity models

At Site U1420, the initial velocity model was constructed directly from the compiled velocity curve, which included both logging and core data. We created initial interval velocity models for each site using the continuous velocity curves. The primary role of calibration at this site, as seen in Figure F10A, is to initially match the seafloor between the velocity curve, the seismic line (259 m WMSF and 330 ms TWT, respectively), and the interval velocity model at a 5 m sample spacing. At Site U1421, we used results from the VSP experiment to create an initial velocity structure against which to calibrate the velocity curve (Fig. F10B). We used the measured depth of the seafloor and the TWT of the positive seafloor arrival (729.7 m WMSF and 980 ms, respectively) in the seismic line with the time-depth matches of the check shots (Table T3) to create the initial seven-layer velocity model. In Petrel 2015 software, we performed calibration of the velocity curve to the VSP by addition of a linearly interpolated drift value between check shots as well as the interval velocity calculations.

Synthetic seismic generation

At each site, we iteratively generated a synthetic seismic trace; an initial convolution with calculated well reflectivity series and a Ricker wavelet provided a rough initial calibration. Next, any necessary bulk shift was applied to obtain a depth match between the seafloor in the drilling and seismic data. We used the Roy White wavelet extraction method at each site in Petrel 2015 to estimate source wavelet information using the continuous physical property curves and trace recordings near the hole. This method refines the tie by using cross-correlation of the hole reflectivity and seismic trace amplitudes to extract a deterministic wavelet (White, 1980). The algorithm produces a wavelet by using the ratio of cross-correlation of the reflectivity and seismic trace in the frequency domain (by fast Fourier transform) to the autocorrelation of reflectivity plus white noise. Spectral properties of wavelets used in this analysis are summarized in Figure F11. We convolved these deterministically extracted wavelets with the reflectivity series to create the synthetic seismic traces used in the synthetic-to-seismic matching process.

At Site U1420, we generated the synthetic seismic at the hole location first by convolution with the Ricker wavelet in Figure F11 (128 ms length and central positive peak at zero), and then any necessary bulk shift was applied to assure that the data aligned such that the seafloor time-depth match occurred near the positive seafloor reflection. Next, we performed convolution with a deterministically extracted wavelet (Fig. F11) according to the White method (White, 1980). Figure F12 shows the synthetic trace in the hole, nearby seismic traces from Line GOA 2505, interval velocity, and calculated reflection coefficients in both TWT and meters below sea level on the seismic depth scale (SSL). At Site U1421, we generated the synthetic seismic trace illustrated in Figure F13 using the deterministically extracted wavelet at this location after convolution with the same Ricker wavelet, as previously described, and bulk shift to maintain alignment of the top of the log with the positive reflection of the seafloor.

Seismic-to-synthetic matching

The traditional well tie method correlates the seismic recording and the synthetic seismic trace at the hole location (White, 1997). We made matches by visually identifying key horizons in the synthetic seismic trace generated in each hole and events in the survey near the hole location. In order to finally calibrate the hole and seismic, it is common practice to execute these matches by small adjustments to the velocity model in order to increase the correlation between the synthetic trace and the seismic data (White and Simm, 2003). In this way the velocity model, and thus the TDR, depends initially on the velocity log, or velocity log and check shots in the case of Hole U1421A, and the final velocity model is determined by the applied adjustments from the visual matches (e.g., White and Simm, 2003). We applied the calibration procedure described above as the basis for the development of a TDR at each site (Tables T1, T4).

Top of page   |    Previous   |    Next