Digital images of split-core surfaces of sediments from Holes 1098A, 1098B, and 1098C were collected at the ODP Core Repository in Bremen, Germany, before postcruise sampling was done, as it spoils core surfaces for imaging purposes. Images were generated with a digital color line-scan camera mounted on a Geotek multisensor track. The camera is fixed and collects line scans of 980 pixels across the core surface for each of the three colors separately (red, green, and blue [RGB]) while the core is pushed along the rack by a motor. Individual line scans are compiled into blocks of 2048 lines and stored as one image. Given the speed at which the core was moved underneath the camera, each 980 pixel x 2048 pixel image has a theoretic stratigraphic resolution of ~120 pixels/cm. However, the camera was still in an experimental phase when the Site 1098 cores were scanned and part of the time it produced darker stripes overlaid over the real sediment color variation. These stripes, when present, have a regular wavelength of 11 pixels (0.92 mm of sediment) and limit the effective resolution of the images to ~1 mm. This resolution is still fine enough to describe the color variation within the laminated intervals of Site 1098, as dark/light couplets are generally >4 mm thick.
Images include a cloth centimeter scale that was draped on the sediment surface along the edge of the core to translate the pixel positions into a metric scale. A Kodak color reference card was scanned after each 1.5-m section to allow for calibration of color values and for correction of any-long term variation in the light source. Core surfaces were scraped carefully before scanning to smooth sediment surface as best as possible. Scratches in the sediment or other surface irregularities produce dark shadows and/or light reflections, which introduce artifacts in the registered color values in the digital image. All undisturbed pelagic intervals in all three holes were scanned. Excluded are intervals with physically disturbed sediments (core catchers, except Sample 178-1098B-5H-CC, and the top 30-60 cm of each core) (Acton et al., Chap 5, this volume) and the central portion of the thick, homogeneous turbidite between 25 and 29 meters composite depth (mcd) (Barker, Camerlenghi, Acton, et al., 1999).
Nederbragt et al. (2000) described the methods used to extract a line scan of RGB color values from each image and to translate those into L*a*b* values using a color reference card for calibration. In the L*a*b* system defined by the Commission Internationale de l'Éclairage, L* represents lightness, ranging between 0 = black and 100 = white. The other two variables describe the actual color, with a* representing green (negative values) or red (positive values) and b* representing blue (negative values) or yellow (positive values). Color values were extracted from a 20-pixel-wide strip (= 2 mm) along the stratigraphic axis in individual images and compiled into a continuous color record. The data were subsequently filtered to remove the millimeter-scale dark stripes produced as an artifact by the camera. To that purpose, a low-pass filter was applied in the frequency domain to remove all variation with wavelengths of 1 mm and less. The files were then reduced in size to one L*, a*, and b* value per millimeter by taking the median value in each millimeter interval. An average value per centimeter interval is used to describe long-term variation in color (Tables T1, T2, T3). All images were logged visually on the computer screen to measure the thickness and stratigraphic position of dark laminae, using the image analysis package NIH-Image, in combination with a macro that translates subsequent mouse clicks on the image into stratigraphic position.
The depth-age relation used here is based on the published composite depth scheme (Acton et al., Chap 5, this volume) and radiocarbon ages (Domack et al., in press). Minor modifications are made to the composite depth scale. An abrupt shift in radiocarbon ages indicates that erosion occurred below the second 1.2-m-thick turbidite, at ~33 mcd (Domack et al., in press). In Core 178-1098C-4H, more sediment appears to have been removed than in the other two holes. The best match with the other two holes of the top of Core 178-1098C-4H and the lower part of the core is obtained when a gap of 62 cm is introduced below the base of the turbidite at interval 178-1098C-4H-3, 60 cm (Fig. F2). This match is based on whole-core magnetic susceptibility records, which show a good correlation in this interval (Acton et al., Chap 5, this volume). Color (Fig. F2) and gamma-ray attenuation (GRA) bulk density show more lateral variation in this interval. The composite depth scale is also changed for Cores 178-1098A-6H and 178-1098C-5H to obtain a best-fit match for the pattern of color variation within the distinctly laminated interval at the base of Site 1098. A modified splice was constructed to sample the most representative color-variation record (Fig. F2).