4. the odp color digital imaging system: color logs of

Kennett, J.P., Baldauf, J.G., and Lyle, M. (Eds.), 1995Proceedings of the Ocean Drilling Program, Scientific Results, Vol. 146 (Pt. 2)

4. THE ODP COLOR DIGITAL IMAGING SYSTEM: COLOR LOGS OF QUATERNARY SEDIMENTSFROM THE SANTA BARBARA BASIN, SITE 8931

Russell B. Merrill2 and John W. Beck2

ABSTRACT

Digital color logs of cores from Leg 146, Holes 893A and 893B, have been taken from images captured during January1993, within days after the cores were split and described. The images were captured and color analyses performed on theOcean Drilling Program (ODP) color digital imaging system, which was assembled from relatively inexpensive, off-the-shelfcomponents. The images were used to calculate sedimentation rates by fitting chronological data from Hole 893A to void-cor-rected depths determined by eliminating all voids mapped from the images as >l cm in length measured downcore. Color mea-surements were made at intervals between 0.22 and 1.0 mm in length, and then Commission Internationale de VEclairage(CIE) 1931 chromaticity values were computed. Results plotted within CIE chromaticity space lie in close proximity to theMunsell 5Y hue plane, confirming that the instrumental color analysis technique has produced results consistent with those ofthe human observers who described colors visually. Significant periodicities occurring at 12, 17, 31, and 90 years correlate withsunspot activity cycles, suggesting that color variations may reflect global climatic forcing functions. Linear correlationbetween color data sets from Holes 893 A and 893B suggests that as much as 1.2 m of material present at or near the top of Hole893B may not have been recovered from Hole 893A, and that there is a 70-cm depth discrepancy in the opposite direction at51.5 meters below seafloor (mbsf). The paper recommends that the 1931 CIE chromaticity system be used routinely fordescribing colors of geological materials because it readily accommodates the mathematical manipulations required for statisti-cal and time-series analyses and avoids the subjectivity and other weaknesses inherent in the Munsell Color System.

INTRODUCTION

The growing importance of high resolution paleoceanographicstudies in recent years has aggravated the need to reconstruct compre-hensive, continuous columnar sections from well logs and recoveredcores. Ideally, these reconstructions should be completed within afew hours of recovering the cores or obtaining the well log data; how-ever, reality is that reconciling the various depth measurements asso-ciated with multiple holes and multiple tools is a major challenge,especially when core recovery is not equal to 100 percent of thedrilled interval.

Color is a readily observable physical property, and one which haslong been used by geologists to correlate beds visible in outcrop. Un-fortunately, the colors of most fresh marine sediments are too ephem-eral, and color differences between layers typically are too subtle andtoo abundant for quantitative work to be accomplished reliably by hu-man inspection. Overcoming these problems would allow the colorproperties of the sediment column to be studied as proxy records ofchanges in the climatic, depositional, erosional, and consolidation en-vironments that have affected local events. Readily comparable datasets derived from closely spaced measurements also allow standardcorrelation techniques to be used to adjust depths between holes andto detect missing intervals for the guidance of the shipboard partyduring drilling, as well as in subsequent scientific studies.

The Ocean Drilling Program (ODP) color digital imaging systemis a candidate instrument for inclusion on a future multi-sensor track(MST) for split core analyses on board the JOIDES Resolution,

'Kennett, J.P., Baldauf, J.G., and Lyle, M. (Eds.), 1995. Proc. ODP, Sci. Results,146 (Pt. 2): College Station, TX (Ocean Drilling Program).

2Ocean Drilling Program, Texas A&M University Research Park, 1000 DiscoveryDrive, College Station, TX 77845-9547, U.S.A.

should funds become available for development of such an MST. Thesystem is based entirely upon relatively inexpensive off-the-shelfcomponents, so it can be deployed in any laboratory.

Like the shipboard archival photographs, images captured withthe ODP digital imaging system can be used long after the cruise forstudies of structures, textures, bioturbation, and other physical as-pects of cores visible in the split surface. Unlike archival photo-graphs, in which emulsion dyes often change significantly within afew years of exposure, digital images preserve color information in-definitely, so an ephemeral physical property can now be studiedquantitatively during and a cruise and after the cruise is over. Becausecolor can be measured objectively and quantitatively from digital im-ages at any desired interval (minimum size is a single pixel), the sys-tem offers data appropriate for a variety of uses. Colors can bemeasured essentially continuously along the length of a core, provid-ing a dense data set suitable for core-core correlation or time-seriesanalysis, which then can be smoothed for inspection or plotting atspacings comparable with analyses made at discrete intervals, such assample-derived geochemical parameters or measurements from thewhole-core MST. Depending upon the physical condition of thecores and the resulting color data quality, core-core correlations maybe performed aboard the vessel in near real-time or may require sig-nificant post-cruise processing.

The Munsell Color System, which has been used traditionally todescribe sediment colors, is a qualitative system dependent upon vi-sual comparison of sample color chips by observers working underideal (but rarely achieved) lighting conditions. Data recorded in theMunsell System cannot readily be manipulated or compared mathe-matically and, inevitably, vary in quality as observers and observa-tional conditions change. These problems have contributed to thedifficulty of comparing and using recorded color observations for sci-entific interpretations. One of the objectives of this study, therefore,has been to evaluate an internationally recognized color classification

R.B. MERRILL, J.W. BECK

system, the Commission Internationale de UEclairage (CIE) 1931chromaticity system, for use in describing geological materials.

This is the first report of an investigation that applies digital videocapture and digital still camera technology to the making of quantita-tive, continuous color measurements of fresh marine sediment coresand to the adjustment of depth scales for drilling and coring artifacts.It also is the first attempt to use the CIE 1931 chromaticity system torecord sediment color properties so that they can be manipulated intime-series analyses and correlated with other physical propertiesmeasurements. Preliminary results of the study have been describedby Rack et al. (1993) and by Rack et al. (1994). This paper describesthe ODP color digital imaging system in its role as a color loggingtool, along with the application of that tool to Quaternary sedimentsrecovered from the Santa Barbara Basin at Site 893. The color dataare correlated with the Site 893 magnetic susceptibility data by Rackand Merrill (this volume).

BACKGROUND

Site 893 is located at 34°17.25'N, 120°02.19'W at a water depthof 576.5 m in the Santa Barbara Basin, approximately 20 km south ofSanta Barbara, California. During December 1992 two holes weredrilled 10 m apart: 21 cores to a depth of 196.5 m below the seafloor(mbsf) from Hole 893A; 8 APC cores were recovered from Hole893B, reaching a total depth of 68.8 mbsf. The cores were returned tothe Gulf Coast Repository, College Station, Texas, without beingsplit.

The scientific party met at the Repository during January 1993 tosplit, describe, and sample the cores. Sediments recovered from Site893 consist largely of clays, silty clays, and sands intermixed withbiogenic material (carbonate and opal). See Kennett, Baldauf, et al.(1994) for a complete description.

Color Measurement

Colors have always been recognized as important properties ofnatural materials, including rocks and soils. Color differences areamong the most easily observed physical properties, but because theyare difficult to describe unambiguously and comprehensibly to oth-ers, few geologists have attempted quantitative studies. Clearly, itwould be desirable to be able to study color differences quantitative-ly, like other physical properties, in order to use them as guides tosedimentary processes and as keys to intercore depth correlation.

A number of systems have been devised for describing colors ofmaterials by comparing them to widely distributed sets of color chips(Wyszecki and Stiles, 1982). Of these, the Munsell Color System hasbecome widely accepted by geologists and has been employed in de-scribing Deep Sea Drilling Project (DSDP) and ODP cores since thebeginning of the drilling program. The Munsell Color System isbased upon the work of Albert H. Munsell (1929, 1963; Billmeyerand Saltzman, 1981), who assembled a collection of color samplesbelieved to represent equal intervals of visual perception and deviseda three-coordinate system (hue, value, chroma) to describe all possi-ble colors. Munsell hue is expressed by a combination of numbersand letters (e.g., 2.5Y) which describes a color's property of beingYellow, Red, Green, Blue, Purple or binary mixtures thereof. Mun-sell value describes the color's lightness relative to a gray scale rang-ing from white to black, while Munsell chroma corresponds to thedegree of difference between a color (other than black, white or gray)and a gray of the same value or lightness. Within the Munsell System,therefore, a red-brown might be described as 5YR 5/6, where 5YR isthe hue, 5/ is the value, and /6 is the chroma.

The color-measurement tool traditionally employed on board thedrilling vessel has been a 5- × 8-inch notebook, Munsell Soil ColorCharts (Munsell Color, Macbeth Division, Kollmorgen Corporation,

Baltimore, MD, U.S.A.), containing 12 charts and 300 color chips,from which scientists describing cores attempt to select the closestmatch or to interpolate values. The system is simple to use, easilycomprehensible, and requires minimal training. Unfortunately, mea-surements are labor-intensive, hard to replicate, and difficult to ma-nipulate mathematically.

Practical difficulties in applying the Munsell technique (or similarchip comparison systems) include variability in observers' color per-ceptions; lighting (the Munsell Color System requires that the illumi-nation be daylight balanced, whereas the description table aboard theJOIDES Resolution is lit with cool-white fluorescent lamps); and al-teration of the color reference chips due to age, handling, and expo-sure to mud and water. Generally, sediment colors observed on thedescription table are ephemeral, changing rapidly as the surface ofthe split core oxidizes, so it is difficult to reproduce a color observa-tion only hours after making it, even if the same set of color chips canbe employed.

If the observational difficulties in applying the Munsell ColorSystem to deep-sea sediment cores could be overcome, then we mustdeal with the fact that the system does not describe colors in a phys-ical continuum, but only in terms of members of a physical set of col-or chips. Sediment samples that appear to match members of the chipset almost certainly contain coloring agents with spectral characteris-tics different from those of the chip (i.e., the sediment and the chipform a metameric pair), so the match is valid only under the preciseconditions of the original observation, including the color tempera-ture of the lights and the identity of the observer.

Inherently, comparison-based color description systems likeMunsell cannot provide quantitative data of the quality produced, forexample, by the magnetic susceptibility, P-wave, GRAPE, Hamiltonvane shear, or other physical properties instruments routinely used onboard ship.

CIE Color Space

The group that sets the standards for measuring and describingcolor is the Commission Internationale de VEclairage. This commis-sion has established a number of applicable standards, the choice ofwhich depends upon the precise nature of the measurement and con-ditions of observation. The standards appropriate for use with theODP digital imaging system are the CIE 1931 Standard ColorimetricObserver, CIE Standard Sources D (Daylight Simulators), and CIE1931 Standard Colorimetric System (Wyszecki and Stiles, 1982),which will be employed throughout this paper.

CIE 1931 chromaticity (x, y, Y) space is illustrated in Figure 1.This three-dimensional shape bounds the region of observable colorswithin CIE tristimulus space. Within CIE chromaticity space, colorsare described uniquely and continuously in terms of the three coordi-nates x, y, and z. These dimensions are related to the CIE primary tri-stimulus values (X, Y, Z) as follows:

X

y =

Because these coordinates sum to 1, a tristimulus value, usually Y,must also be specified to define each color uniquely. By convention,Y is plotted perpendicular to the x, y-surface (parallel to the z-axis)and is thought of as defining the luminance of the color.

Intuitively, luminance (Y) is analogous to value in the MunsellSystem or to brightness (a perfect white reflector would have Y =100), while the (x, y) coordinates describe the other properties of acolor, which are in some degree analogous to Munsell hue and chro-

(X

(X

(X

+ Y-Y

+ Y-Z

+ Y -

h Z ) '

h Z ) '

HZ) '

ODP COLOR DIGITAL IMAGING SYSTEM

= 100

Figure I. CIE 1931 chromaticity (x, y, Y) space located within CIE tristimu-lus (X, Y, Z) space. Modified after Agoston (1987).

1.0

Loci of constant Munsell Hues inCIE 1931 (x,y)-chromaticity space

0.8

Figure 2. Projection of Munsell hues into the CIE (x, y) chromaticity coordi-nate plane. Heavy line envelopes base of the three-dimensional (x, y, Y)shape seen in Figure 1. Munsell hues occupy surfaces within this shape, loca-tions of which were determined empirically by Newhall et al. (1943). Filleddiamond marks location of the D5000 light source employed in this study.

ma. For reference, Munsell hue surfaces are projected onto the (x, y)chromaticity plane in Figure 2. Very simplistically, increasing valuesof y correlate with increasing yellow content, while x ranges betweenblue and red.

The clear advantages that the CIE chromaticity system offers overthe Munsell System are that effects of changing illuminants and/orobservers are calculable so measurements are easily verifiable and re-producible with different instruments, and normal mathematical rela-tionships apply so that statistical manipulations are valid.

There is no free lunch, of course. The principal drawback of theCIE chromaticity system is that it distorts some regions of colorspace, much as a Mercator projection distorts portions of a map re-mote from its center. This distortion is irrelevant to most geologicalapplications, but there are some awkward moments. For example, allthree tristimulus values (X, Y, Z) of the color black equal zero by def-inition so true black lies anywhere and everywhere in the x, y-plane(z, Y = 0) of the chromaticity diagram. Colors close to black can bedifficult to characterize uniquely. Coal geologists may wish to avoid

ambiguity by selecting a coordinate system with more emphasis onshades of gray and black, but most geologists will be well served bythe CIE chromaticity system.

Instruments

Four instrument systems capable of making color measurementshave been employed on ODP cores recently: a multi-wavelengthscanner (Leg 138; Mix et al., 1992), a Minolta CM-2002 color scan-ner (Leg 145; Rea et al., 1993), an OptoTech color digitizer (Schaafand Thurow, in press; see also Schaaf and Thurow, this volume), andthe ODP color digital imaging system (Leg 146; Kennett, Baldauf, etal., 1994), which is the subject of this paper. The prototype multi-wavelength scanner deployed on Leg 138 measured reflectance at511 different wavelengths from a 2-cm-wide spot centered in the splitcore. The normal interval between measurements was 4 cm, whichproved adequate for correlating depths among adjacent holes. Thelarge number of wavelengths measured extends the capability of thisinstrument well beyond color measurement, enabling users to distin-guish lithologic components by differential reflectance (Mix et al.,1995). The authors reported results in terms of relative reflectancesin red, green, and blue portions of the spectrum rather than in termsof color, but it should be straightforward to compute color valuesfrom their data if calibration procedures were followed.

The Minolta scanner, which now is routinely available on theJOIDES Resolution, is a computer peripheral that resembles a gunthat can be aimed at a target of interest and triggered to report colorreflectance (31 wavelength channels between 400 and 700 nm) invarious standard formats, including Munsell and CIE chromaticity.The output of the instrument can be directed to an IBM PC-clone forcapture.

The OptoTech 8-bit color scanner (described by Schaaf andThurow, in press) requires three passes over the target core, each witha different filter, in order to capture three monochrome images whichthen are merged to form a 24-bit color image. This device would beat a disadvantage aboard the ship because of vibration; however, it isappropriate for use on shore. Schaaf and Thurow (this volume) haveused this instrument to capture 8-bit images of the Site 893 cores inshades of gray.

ODP Digital Imaging System

The imaging system consists of four components: lights, cameras,computers, and software. Cores or samples to be imaged are placedon an illuminated copy stand so that an image can be formed withineither a video camera or a digital still camera. The image is capturedwith a microcomputer, and then stored on archival media. Images canbe processed later with a variety of image processing packages, in-cluding the ODP color measurement software discussed in this paper.All electronic components are available off-the-shelf. Satisfactorysubstitutes are available from these and other manufacturers at a widevariety of costs, depending upon the resolution and accuracy re-quired.

The geometry of lights and camera (Fig. 3) conform to the CIE 45/0 standard for viewing and illumination.

Cameras

The ODP digital imaging system is equipped with two cameras:

1. A SONY DXC-750MD NTSC camera equipped with threeCCDs (768 × 493 array) and a Fujinon TV.Z 4 × 7.5 lens. Thiscamera is preferred for imaging on board a ship because it de-livers the red (centered at 700 nm), green (546 nm), and blue(436 nm) images simultaneously, reducing the chance that vi-bration will adversely affect registration of the images. SONYdeclines to publish more detailed information on wavelength

47


Camera

Figure 3. Geometry of the ODP digital imaging system complies with CIE(45/0) conditions for illumination and viewing. Video or digital still camerafeeds image to a PC-clone or to a Macintosh microcomputer, where it is pro-cessed and archived.

sensitivity; however, it is available to customers who sign anondisclosure agreement.

2. A LEAF (Southboro, MA) digital still-camera back, mountedon a Hasselblad ELX camera body equipped with a Hasselblad50mm Distagon lens, is used to capture close-up images onshore. This camera, which has a digital resolution of 2048 ×2048 42-bit pixels, employs a filter-wheel to generate red,blue, and green images sequentially. The filters are KodakWratten #25, #58, and #47B. Detailed filter specifications arereported in Kodak (1990).

Illumination

Illumination for the SONY video camera version of the systemconsists of two tungsten lamps filtered to a color temperature of5020°K. The color temperature of light incident on the core samplesranges between 4300°K and 4600°K, with variations resulting prima-rily from the aging of bulbs. Tungsten bulbs age rapidly, with signif-icant color temperature variations observable within hours. Somevariation in color temperature can be accommodated by frequentrecalibration (in this study, calibration images were captured betweenevery section of core); however, the difficulty of finding stable, day-light-color-balanced lighting sources is a real problem for this kind ofwork. Most commercially available lights are designed for film pho-tography, which is more tolerant of lighting variations than is color-imetry.

UV-corrected, daylight-balanced strobes (5500°K) are used to il-luminate the copy stand when the LEAF digital still camera is in-stalled. Strobes are much less prone to change color temperature withage than are the tungsten bulbs.

If desired, data sets can be corrected for nonuniform light distri-bution by fitting curves independently to each channel of color inten-sity data derived from a uniform surface, such as an unexposedphotographic emulsion. These curves can then be ratioed with dataderived from samples in order to accomplish the correction. Data re-ported in this paper were not corrected for the effects of nonuniformlight distribution because the magnitude of such corrections was ob-served to be less than noise.

Computers

The SONY video camera is supported by a Truevision (Indianap-olis, IN) AT-Vista4-Mb video capture board equipped with a 10-MbAT/VMX auxiliary memory board and mounted in an IBM 486/DX66 clone. As of this writing, Truevision has merged with Raster-

Ops (Santa Clara, CA). The LEAF digital camera back is supportedby a Macintosh Quadra 800 equipped with a RasterOps 8/24XLi vid-eo board.

Captured images are stored in extended TGA (TGA 2.0) formaton magnetic or magneto-optical discs for immediate use, and on 4-mm DAT tape or ISO-9660 CD-R discs for archival storage. Digitalimages can be extremely large: images produced by the SONY cam-era system may be as large as 3 Mb, and those produced by the LEAFcamera are 26 Mb each. LEAF images are stored in 48-bit TIFF 6.0format initially, but are transformed to 24-bit TGA 2.0 images for usewith the ODP color measurement system.

Computer Software

The ODP Color Measurement software, which supports digitalimage capture and color analysis on the 486 clone, is written in Mi-crosoft C and Fortran, and makes use of proprietary libraries writtenby Truevision and IMSL. Details of algorithms are available from theauthors if desired. Images from the LEAF camera are captured withproprietary software from LEAF, and are then converted to TGA 2.0formats in ADOBE Photoshop.

Core images used for most of this study were captured with theSONY DXC-750MD camera and the Truevision AT-Vista captureboard operating at a resolution of 1008 × 486 16-bit pixels, with 5 bitsof color information allocated to each channel. At this resolution, thesystem can distinguish 32,768 different colors; however, quantiza-tion effects are observable in data from areas with similar colors. Itwould have been better to use 24-bit pixels, but storage logistics pre-cluded this option at the time. The LEAF camera was used to capturehigh-resolution images used in fine-scale correlation.

METHODS

During the post-Leg 146 sampling party of January 1993, coresfrom Site 893 were split, scraped gently with a razor blade to removesurface disturbances, and imaged. Using the SONY video camera, a1008 × 486-pixel 16-bit color image was captured of each 10-cm in-terval of the split core, overlapping at both ends of each interval sothat each image covers a 13.5-cm length of core. Each pixel spansabout 134 µm along the length of the core and 210 µm along the di-ameter. The image capture process was not automated, so approxi-mately two weeks were required to complete the task. Because of thisdelay, it was impossible to image all of the cores while perfectlyfresh, so all surfaces were allowed to oxidize prior to imaging.

Images of color calibration standards were captured between ev-ery core section in order to minimize possible effects of drifting illu-minant color temperatures caused by aging tungsten lamps. Primarycalibration standard was a set of red, green, blue, white, and gray col-or chips supplied by MacBeth (Kollmorgen Corporation, Baltimore,MD), supplemented by the Kodak Color Separation Guide and GrayScale (CAT 1527654). Images of 12 British Ceramic Research Asso-ciation (BCRA) Ceramic Color Standards, calibrated by the Hem-mendinger Color lab (Princeton, NJ), were captured at less frequentintervals to ensure that the primary color standards remained stable.All standards remained stable throughout the period of image cap-ture.

Using the ODP color system (v. 1.1) software, color measure-ments were made at 1-mm intervals downcore through all images ofHole 893B cores and through all images of Hole 893 A, Cores 1H and4H-11H. Measurements were made at 0.6-mm intervals downcorethrough all images of Hole 893A, Cores 2H, 3H, and 12H-21H. Ateach individual point, the red, green, and blue values were measuredfor each pixel in a 9-pixel kernel centered on the measurement point.Pixels that differed from the mean value of their respective colorchannel by more than one standard deviation were abandoned in or-


der to improve discrimination against edges of the abundant voidsand cracks, and adjusted means were calculated from remaining val-ues. The adjusted means were recorded as the red, green and blue(RGB) values of that point.

Using a print of each digital image as a guide, the data were editedto remove points attributable to voids, scrape marks, drilling distur-bances, and similar problems. Measurements from the "orphan" sec-tions (pieces recovered from the catwalk after they had been ejectedby gas pressure when the cores were first cut) were included only ifthe sections were in excellent condition, because physical distur-bance seriously perturbs color measurements. Depths were closed uponly in the case of the voids listed in Table 1; otherwise, bad datapoints were eliminated without shortening the depth scale, so the col-or data are compatible with other physical property data compiled onthe void-corrected depth scale. Gaps due to the elimination of baddata points were filled by linear interpolation before analyzing thedata set as a time series.

RGB values were converted to CIE tristimulus (XYZ) values viathe following relationship (Rogers, 1985):

6 5

ybcb

- xr - yr)Cr( 1 - xg - yg)Cg( 1 - xb - yb)Cb

where (xr, yr), (xg, yg), and (xb, yb) are CIE chromaticity coordinatesfor the red, green, and blue channels (primaries) of the camera,respectively. The primary chromaticity coordinates may be availablefrom the manufacturer for some cameras; however, they and thecoefficients (Cr, Cg, Cb) remain constant for a given image or set ofimages, so they can be calculated from measurements of knowncolor standards, which was the procedure followed here. Becauseimages of the standards were captured between every core section,any changes which may have occurred in the optical path or minorvariations in illumination are accounted for in the coefficients. Tri-stimulus values were then converted to CIE chromaticity coordi-nates (x, y, Y) as discussed earlier.

RESULTS

Final versions of the color data sets for Holes 893A and 893B(Appendix) will be found in spreadsheets, stored as tab-delimited textfiles suitable for import into Kaleidagraph, located on this volume'sCD-ROM.

Because the color digital images are high-resolution maps of thecores, their first application was in locating all voids greater than 1cm in length downcore, so that the voids could be eliminated fromdepth calculations made by the entire shipboard party. Using theODP digital imaging system, each void gap was measured to thenearest 0.5 mm. Preliminary versions of the resulting void catalogand revised top-of-section depths were prepared by Merrill and Rack(Kennett, Baldauf, et al., 1994). Corrected versions are included hereas Tables 1 and 2. These tables were made available to all investiga-tors who have contributed to this volume. Additionally, void-correct-ed depths of all Hole 893A and 893B samples and ages of Hole 893Asamples (based on the work of Ingram and Kennett, this volume)were calculated and supplied to these investigators so that analyticalresults could be plotted on a common scale. Relevant age-depth tiepoints, resulting curve-fit parameters, and corresponding sedimenta-tion rates are listed in Table 3.

CIE chromaticity data for Holes 893 A and 893B are shown in Fig-ure 4, projected into the (x, y) plane, and are projected into the or-thogonal (x, Y) and (y, Y) planes in Figures 5 and 6. It is immediatelyreassuring to note the close agreement between these data and those

recorded in the barrel sheets by the sedimentologists who describedthe major lithologies in the freshly split cores as ranging in color fromMunsell 5Y 3/1 to 5Y 6/3, with the majority in the neighborhood of5Y 4/2 (Kennett, Baldauf, et al., 1994). We get color values from thedigital measurements that agree with those we get by visual inspec-tion. It is also pleasing that the digital measurements found colors thatare only slightly darker (displaced to lower Munsell values) thanthose observed in freshly split cores, given the inevitable delay ofseveral days between description and image capture. Oxidation ofcore surfaces during this period appears to have had little effect uponthe digital color measurements.

The data from Holes 893A and 893B are in general agreement;however, there is a noteworthy "tail" observed in the (x, y) data forHole 893B that is not present in the Hole 893A data. Points in this tailextend parallel to the 5Y 3/ line (Fig. 4) at luminance values (Fig. 6)which are consistently lower than the remainder of the 893B or 893 Acore. The tail is entirely made up of measurements made at depthsshallower than 11.5 mbsf. This tail may reflect different water con-tents, different oxidation levels, or other geochemical gradients char-acteristic of sediments near the mud line, but none of thesemechanisms explains why the tail is seen in Hole 893B but not inHole 893A.

Figures 7 and 8 show the chromaticity data from Holes 893A and893B, respectively, plotted against depth below seafloor corrected bythe removal of voids per Table 1. Relatively low values of luminance(Y) are seen to be typical of turbidite sequences, as might be expectedbecause the coarse, granular textures of turbidite intervals absorblight; however, the relationship has not been quantified. A verticalblack bar parallel to the Hole 893 A luminance plot in Figure 7 marksthe interval of the thick turbidite between 113.9 and 116.7 mbsf.

A vertical dashed bar within the Hole 893A y-chromaticity plot(Fig. 7) parallels Core 16H, which is enlarged in Figure 9. Within thefirst three sections, thin, light olive gray beds intercalated with darkolive gray beds are clearly visible in the archival photograph (Fig.10). These intercalated light and dark beds appear in the color data(Fig. 9) as peaks of increased luminance correlated with decreased (x,y)-chromaticity. Lower in the core, intercalated beds give way tomore massive, structureless sediment that is considerably less vari-able in color. At the scale at which these data (one point per 0.5 cm)are plotted, the intercalated beds can be mapped easily, and thechanging lithology is clearly visible.

Power Spectra

Using linear interpolation, a continuous data set for Hole 893A,Cores 1H-5H, was extracted from the edited but unaveraged colordata set and correlated with the age data of Ingram and Kennett (thisvolume) to yield a time series data set with x, y, Y values every 0.2years. Power spectra derived from this data set, which contained144,660 sets of values ranging in age from 30 years to 30,000 years,are shown in Figure 11.

Numerous peaks appear in all three spectra with prominent exam-ples at 12, 17, 31, and 90 years. It is noteworthy that cyclical varia-tions in one coordinate are not always matched by variations in theother two coordinates. For example, there is a moderate peak in theY data at 23 years which is arguably missing from the x and y spectra,while the prominent peak which appears in the Y data between the 90and 138 year peaks is much less visible in y and is almost completelymissing from the x spectrum. Clearly, factors which affect sedimentcolors operate independently of one another and are reflected differ-ently by the individual CIE chromaticity coordinates.

Peak positions common to the three spectra agree approximatelywith reported solar sunspot activity cycles (Landscheidt, 1987) at 11,22, 31, and 86 years, which indicates that colors of sediments of theSanta Barbara Basin respond to high frequency global climatic forc-ing functions. This contrasts with the observation of Gardner and


Table 1. Catalog of voids greater than 1 cm observed in digital images of cores from Site 893.

Top of voidwithin section

(cm)

146-893 A-1H-4, 120.52H-1, 13.42H-2, 71.32H-3, 33.52H-3, 1202H-3, 141.32H-4, 15.32H-4, 35.72H-4, 49.82H-4, 81.42H-4, 96.12H-5, 8.62H-5, 23.82H-5, 56.52H-5, 64.22H-5, 73.62H-5, 103.92H-5, 124.12H-5, 142.92H-6, 25.82H-6, 38.42H-6, 65.53H-1,46.83H-3, 793H-3, 90.13H-4, 139.33H-5, 353H-5, 623H-5, 67.43H-5. 76.63H-6, 20.94H-1,62.54H-1,85.54H-2, 22.64H-2, 50.64H-2, 124.94H-3, 704H-3, 97.14H-3, 108.84H-4, 39.84H-4, 61.84H-4, 884H-4, 104.94H-4, 130.44H-5, 69.54H-5, 80.14H-5, 108.64H-5, 113.84H-5, 1164H-6, 164H-6, 68.84H-6, 106.94H-6, 128.54H-7, 6.75H-1.99.55H-1, 108.35H-1, 120.15H-2, 3.75H-2, 18.25H-2, 29.65H-3, 76.45H-3,91.55H-3, 110.75H-4, 115H-4, 17.65H-4, 88.55H-4, 122.35H-5, 32.45H-6, 23.85H-6, 33.75H-6, 47.25H-6, 51.75H-6, 69.95H-6, 82.35H-6, 97.55H-6, 139.36H-1,2.76H-1.21.26H-1,43.86H-1,70.56H-1,97.86H-2, 10.96H-2, 31.56H-2, 104.86H-2, 126.86H-3, 68.76H-3. 85.46H-3, 115.8

Top of void(mbsf)

5.706.638.739.84

10.7010.9211.1611.3611.5011.8211.9612.5812.7413.0613.1413.2313.5413.7413.9314.2514.3814.6516.4719.7819.8921.8822.3322.6022.6522.7423.6826.1326.3627.2127.4928.2429.1929.4629.5830.3830.6030.8631.0331.2932.1832.2932.5732.6232.6533.1333.6634.0434.2634.5536.0036.0836.2036.5336.6736.7838.7238.8739.0639.5539.6140.3240.6641.2442.6342.7342.8642.9143.Q943.2243.3743.7944.5344.7144.9445.2145.4846.0946.3047.0347.2548.1648.3248.63

Void length(cm)

.51.54.32.73.32.12.32.22.325.35.31.94.72.2

.32.81.82.63.12.212.73.115.42.4

.82.17.44.7

21.71.33.9

.54.52.23.32.32.99.12.728.932.91.11.11.4

1.22.32.843.67.52.52.23.67.64.62.32.43.753.92.11.93.42.21.512.26.26.85.42.54.49.52.9

1511.55.33.82.23.6

I9.7


(cm)

6H-4, 22.16H-4, 55.86H-5, 20.56H-6, 52.96H-6, 78.46H-6, 90.77H-2, 50.27H-2,78.17H-2, 96.27H-3, 7.17H-3, 21.17H-3, 52.27H-3, 67.37H-3, 102.47H-3, 123.67H-3, 134.37H-4, 07H-4, 30.37H-4, 437H-4, 67.97H-4, 92.67H-6, 07H-8.41.18H-2, 135.28H-2, 139.58H-3,7.58H-3, 37.58H-3.44.28H-3, 85.18H-3, 109.48H-3, 126.18H-3, 135.28H-3, 141.28H-4, 11.98H-4, 21.98H-4, 37.98H-4, 96.38H-4, 1168H-4, 130.98H-5, 118.78H-6, 79.68H-6, 128.58H-7, 5.48H-7, 10.78H-7, 28.38H-7, 52.68H-7, 118.78H-8,4,28H-8, 12.48H-8, 36.18H-8, 65.49H-1,60.59H-U75.29H-1, 116.19H-1, 128.19H-1, 134.99H-2, 379H-2, 82.29H-2, 88.99H-2, 100.89H-2, 112.79H-3, 11.59H-3, 20.39H-3, 24.49H-3, 33.29H-3, 58.89H-3, 669H-3, 80.19H-3, 929H-4, 12.49H-4, 20.69H-4, 48.79H-4, 100.69H-4, 120.59H-4, 128.39H-5, 60.59H-7, 5.49H-7, 17.39H-7, 19.59H-7, 31.89H-7, 73.29H-7, 90.310H-l,010H-1, 102.710H-1, 129.510H-2, 36.110H-4, 93.410H-6, 20.410H-7, 35.6

Top of void(mbsf)

49.1849.5150.6452.4452.7052.8254.5554.8355.0155.6155.7556.0656.2156.5656.7856.8857.0157.3157.4457.6857.9360.0563.4165.0065.0565.2265.5265.5966.0066.2466.4166.5066.5666.7566.8567.0167.5967.7967.9469.2670.3670.8571.1171.1671.3471.5872.2472.6672.6672.9073.1973.6173.7574.1674.2874.3574.8575.3075.3675.4875.6076.0776.1676.2076.2976.5576.6276.7676.8877.5677.6477.9278.4478.6478.7279.5380.5980.7180.7380.8681.2781.4482.5083.5383.8084.3586.4988.7489.70

Void length(cm)

1.83.41.74.94.43.7

19.513.210.51.64.12.96.497.62.85.83.44.56.53.92.74.92.63.77.35.96.2

14.96.314.83.36.26.3

10.12.42.41.93.55.34.13.43.24.536.64.44.47.34.1

12.12.82.72.82.33.34.61.64.73.74.824.61.22.31.74.14.31.822.12.722.71.11.31.31.310.84.23.14.341.212.22


Table 1 (continued).


(cm)

11H-1.011H-1,81.211H-1,97.411H-4, 35.511H-4, 102.711H-4, 111.912H-1,46I2H-1, 120.2l2H-2,49.812H-2, 84.512H-2,9812H-3, 20.212H-4,45.912H-4, 87.312H-6, 72.812H-6, 84.312H-6, 88.212H-7, 20.313H-1, 19.113H-1.37.113H-1,43.813H-1,95.913H-3, 117.113H-4, 013H-4, 29.113H-8, 34.914H-1, 104.214H-8, 32.214H-8, 43.514H-8, 79.815H-15H-15H-;16H-16H-16H-16H-16H-16H-16H-16H-17H-17H-

,38.4,75.8

>, 28.7, 14.4,21.9,32.3,50.3,87.3, 113, 134.3

>, 117.8,7.9,9.8

17H-3, 79.817H-6, 84.5I7H-6, 86.217H-6, 95.718H-1,54.318H-2, 23.318H-4, 45.218H-5, 57.718H-5, 68.918H-5, 143.318H-6, 25.718H-6, 105.818H-6, 110.118H-7, 12.219H-1, 10.819H-1, 15.419H-2, 4.519H-2, 11.819H-2, 42.119H-3, 2819H-3, 80.319H-5, 135.819H-6, 83.720H-1, 11.720H-l,2820H-1, 110.920H-1, 117.720H-3,41.320H-6, 103.320H-6, 116.320H-6, 124.721H-1, 7.621H-1, 103.921H-2, 36.221H-2, 148.821H-3,41.7

146-893B-1H-1, 65.91H-CC7.32H-1,0.02H-4, 36.33H-1.5.73H-1, 81.83H-1, 138.83H-3, 114.33H-4 83.3

Top of void(mbsf)

92.0092.8192.9796.8197.4997.58

101.96102.70103.49103.84103.97104.76106.51106.93108.73108.85108.89109.70111.19111.37111.44111.96113.84114.16114.45120.47121.54128.77128.88129.24130.38130.76131.78139.64139.72139.82140.00140.37140.63140.84142.18149.08149.10152.82157.51157.52157.62158.55160.23163.53165.15165.26166.01166.31167.12167.16167.69168.11168.15169.56169.63169.93171.26171.79175.37176.34177.62177.78178.61178.68182.41187.52187.65187.73187.08188.04188.46189.58190.20

0.662.192.307.22

11.8612.6213.1915.6117.12

Void length(cm)

1.62.55.65.84.88.8

10.75.83.35.812.95.82.54.61.39.41.1

14.11

114.61.1

17.21.39.73.41.71.51.3

27.611.14.425.71.48.63.15.41.511.33.51.12.21.14.94.55.2

:

:

;

:

t

.5

.1

.5

.65.35.8

.5

.3t.8.4.2.4

2.91.14.71.39.57.112.11.3

15.425.20.61.7

5.417.714.26.95.23.8


(cm)

3H-4, 97.53H-4, 145.53H-5, 10.83H-5, 31.43H-5, 49.03H-5, 54.63H-5, 61.73H-5, 79.23H-5, 93.93H-5, 114.93H-5, 130.63H-5, 141.23H-6, 19.13H-6, 53.03H-7, 16.34H-2, 103.44H-2, 120.84H-3, 127.24H-3, 130.94H-3, 145.84H-4, 23.44H-4, 53.74H-4, 98.84H-5, 54.94H-5, 87.54H-6, 6.64H-6, 18.44H-6, 146.34H-7, 7.44H-7, 18.55H-1,65.15 H-1,99.05H-1, 122.65H-2, 3.45H-2, 28.15H-2, 49.55H-2, 62.75H-2, 73.25H-3, 52.15H-3 133 35H-3, 135.75H-4, 55.55H-4, 85.55H-4, 95.95H-4, 119.85H-5, 19.65H-5, 61.55H-5, 101.65H-6,24.1 '5H-6, 60.25H-6, 109.45H-7, 27.66H-3, 34.36H-3, 80.36H-3, 98.66H-3, 124.36H-3, 139.86H-4, 18.36H-4, 28.36H-5, 21.56H-5, 27.96H-7, 12.36H-7, 23.37H-2, 23.17H-3, 31.87H-3, 53.67H-3, 62.67H-3, 95.37H-4, 7.98H-1, 13.18H-1,37.68H-1,58.78H-l,70.58H-2, 35.58H-2, 88.18H-3, 62.58H-3, 124.78H-4, 21.88H-4, 43.58H-4, 83.78H-4, 128.18H-5, 40.48H-5, 68.58H-5, 71.08H-6,78.28H-6, 131.3

Top of void(mbsf)

17.2617.7417.8918.0918.2718.3218.3918.5718.7218.9319.0819.1919.4519.7920.9223.8324.0025.5725.6025.7526.0226.3226.7727.8328.1528.8428.9630.2430.3530.4631.4531.7932.0332.3432.5932.8032.9433.0434.3335.1435.1635.8536.1536.2536.4936.9837.4037.8038.5238.8839.3740.0343.6444.1044.2944.5444.7044.9745.0746.4946.5649.3949.5051.5153.1053.3153.4053.7354.3559.4359.6859.8960.0161.1561.6762.9163.5463.9964.2164.6165.0665.6765.9565.9767.5468.07

Void length(cm)

1.82.22.82.72.81.22.56.39.87.83.63.96.49.11.11.72.41.51.11.22.616.33.40.841.42.41.14.8

11.421

1.41.81.67.59.11.81.41.81.33.77.16.61.15.72.64.82.335.31.41.8

11.98.12.32.41.11.23.61.32.41.22.84.94.79.62.92.9

13.713.51.12

43.146.3

5.61.21.62.44.9

1224.6

1.21.11.45.4

Note: Voids may be curated empty or filled with styrofoam.

51


Table 2. Depths (mbsf) of tops of sections from Site 893, adjusted by de-letion of voids greater than 1 cm.

Table 3. Chronology and sedimentation rates in Hole 893A.

Section depth

146-893 A-1H-1,0.001H-2, 1.481H-3, 2.991H-4, 4.50lH-5,5.981H-CC, 6.242H-1,6.502H-2, 8.002H-3, 9.442H-4, 10.862H-5, 12.222H-6, 13.492H-7, 14.922H-CC, 15.703H-1, 16.003H-2, 17.473H-3, 18.973H-4, 20.413H-5, 21.853H-6, 23.213H-7, 24.653H-CC 34.404H-1,25.504H-2, 26.764H-3. 28.164H-4, 29.584H-5, 30.824H-6, 32.224H-7, 33.654H-CC, 34.405H-1,35.005H-2, 36.355H-3, 37.685H-4, 39.075H-5, 40.415H-6, 41.865H-7, 43.075H-CC, 43.666H-1,44.506H-2, 45.646H-3, 46.906H-4, 48.246H-5, 49.676H-6, 51.136H-7, 52.496H-CC, 53.347H-1,54.007H-2, 54.057H-3, 55.117H-4, 56.237H-5, 57.567H-6, 59.037H-7, 60.497H-8, 61.967H-CC, 62.798H-1,63.508H-2, 63.658H-3, 65.088H-4, 66.078H-5, 67.228H-6, 68.688H-7, 70.078H-8, 71.328H-CC, 71.969H-1,73.009H-2, 74.259H-3, 75.559H-4, 76.789H-5, 78.139H-6, 79.589H-7, 79.749H-8, 81.129H-CC, 82.011 OH-1,82.5010H-2, 83.8710H-3, 85.35

Section depth

10H-4, 86.9210H-5, 88.4010H-6, 89.8810H-7, 90.6710H-CC, 92.1411H-1,92.0011H-2, 93.3911H-3, 94.8911H-4, 96.3611H-5, 97.6511H-6, 98.41I1H-CC99.8812H-1, 101.5012H-2, 102.8312H-3, 104.2912H-4, 105.7612H-5, 107.1712H-6, 107.6312H-7, 108.9612H-8, 109.3412H-CC, 110.7713H-1, 111.0013H-2, 112.1613H-3, 112.3613H-4, 113.7713H-5, 115.0913H-6, 116.5813H-7, 118.0613H-8, 119.5513H-CC, 120.2914H-1, 120.5014H-2, 122.0214H-3, 123.5614H-4, 125.0514H-5, 126.5314H-6, 126.7714H-7, 126.9314H-8, 128.4114H-9, 129.8714H-CC, 130.2815H-1, 130.0015H-2, 131.2115H-3, 132.7315H-4, 134.2615H-5, 135.8715H-6, 137.4815H-7, 138.97I5H-CC, 139.7416H-1, 139.5016H-2. 140.6916H-3, 142.1616H-4, 143.8016H-5, 145.3316H-6, 146.8516H-7, 148.3316HCC, 149.1817H-1, 149.0017H-2, 150.4617H-3, 151.9917H-4, 153.5417H-5, 155.1117H-6, 156.6017H-7, 158.0517H-CC, 158.4318H-1, 158.5018H-2, 159.9518H-3, 161.4018H-4, 162.9818H-5, 164.4318H-6, 165.8818H-7, 167.3218H-CC, 167.7419H-1, 168.00I9H-2. 169.49I9H-3, 170.8819H-4, 172.4019H-5, 173.88

Section depth

19H-6, 175.3619H-CC, 176.6520H-1, 177.5020H-2, 178.8420H-3, 180.3420H-4, 181.7620H-5, 183.2520H-6, 184.7520H-7, 186.2320H-CC, 186.7621H-1, 187.0021H-2, 187.9221H-3, 189.5521H-4, 191.0721H-5, 192.6321H-6, 194.1321H-CC, 195.24

146-893B-1H-1,0.001H-2, 1.431H-CC, 2.072H-1,2.302H-2, 3.742H-3, 5.272H-4, 6.782H-5, 8.332H-6, 9.822H-7, 11.322H-CC, 11.873H-1, 11.803H-2, 13.133H-3, 14.623H-4, 16.093H-5, 17.493H-6* 18.543H-7, 19.813H-CC, 20.604H-1, 21.304H-2, 22.804H-3, 24.254H-4, 25.714H-5, 27.104H-6, 28.564H-7, 29.984H-CC, 30.735H-1,30.835H-2, 31.975H-3, 33.255H-4, 34.695H-5*, 36.005H-6, 37.365H-7, 38.735H-CC, 39.556H-1,40.306H-2, 41.806H-3, 43.306H-4, 44.526H-5, 45.996H-6, 47.436H-7, 48.936H-CC, 49.697H-1,49.807H-2, 51.287H-3, 52.757H-4, 54.027H-5, 55.487H-6*, 56.977H-7, 58.137H-CC, 58.998H-1,59.308H-2, 60.498H-3, 61.098H-4, 62.518H-5, 63.778H-6, 65.008H-7, 66.428H-CC. 67.24

Sample/event

Top of section (?)5H-7, 1 lpStage 3.3-3.31Stage 4.22^.23**Stage 5d/5eStage 5e (5.51)Termination IIStage 6.41

Depth at baseof interval

(mbsf)

-1.2143.1773.8990.49

113.89116.35140.69147.46156.87195

Age at base ofinterval

(ka)

28.89651659090

116122.56127161.61

Sedimentationrate in interval

(mm/yr)

1.541.391.190.94

*0.941.032.12110

Note: 3-cm paleontological whole-round sections taken routinely from core catchers(CC) and 10-cm interstitial-water whole-round sections taken from alternate coresin Hole 893B (marked by an asterisk) are NOT treated as voids in this compilation,although no material remains in liner.

Notes: Depths are corrected for gas voids (Table 1). I4C data for 16 samples (Ingram andKennett, this volume) between 3.62 and 43.17 mbsf were fitted with a curve (r2 =0.998): age = a + b (depth)' where a = 0.474657733, b = 527.3684544, and c =1.059172067. Ages were assigned to depths below 43.17 mbsf by linear interpola-tion between the ISO and pollen events (Heusser, this volume). For the purposes ofthis curve fit, zero age was assumed to occur at zero depth. Assuming zero age tooccur at -1.21 mbsf in Hole 893A, as suggested by correlation between holes withthe color data, the curve parameters were adjusted to be a = 30.08622645, b =522.3079487, and c = 1.061531766, with r = 0.995. *The interval between 113.89and 116.35 mbsf consists of a single turbidite layer, presumed to have been depos-ited instantaneously.

Dartnell (this volume) that variations in paleoproductivity and othercarbon measures do not reflect low frequency climatic forcing func-tions, such as Milankovitch cycles. Schaaf and Thurow (this volume)also saw a correlation with the 86-year secular solar cycle in theirgray scale measurements, which they attributed to variations in ter-rigenous flux rates and in sizes of summer phytoplankton assemblag-es; however, their data did not show the correlations with the higher-frequency cycles seen in this study.

Core-core Correlation

The upper 50 m of luminance (Y) data from each hole are plottedin Figure 12, where the depth scale for Hole 893A has been shifted togreater depths in the column by 1.21 m to correlate (r = 0.683; signif-icant at the 99% confidence level) with the data from Hole 893B. Anargument, based on the color data alone, can be made that approxi-mately 1.2 m of material is present at or near the top of the Hole 893Bthat was not recovered from Hole 893A. Peaks on which this correla-tion is based are listed in Table 4. Correlated peaks within the upper10 m of each hole are shown in Figure 13.

Schaaf and Thurow (this volume) reported that 2.9 m of materialmay be missing from the top of Hole 893A relative to Hole 893B,more than double the 1.2 m observed in this study. Rack and Merrill(this volume) correlated the color data with the magnetic susceptibil-ity data in considerable detail, and conclude that approximately 70cm is missing from the top of Hole 893 A, relative to Hole 893B, thatadditional material (75 to 100 cm) may be missing from Hole 893Abetween Cores 1H and 2H, and that between 1.5 and 2.0 m of materialmay be missing from between Cores 2H and 3H in both holes. Cor-relation of multiple data sets improves the clarity of the view tremen-dously.

During the January 1993 splitting party, the scientific party no-ticed strong similarities in the spacing of fine laminations within in-tervals between 50 and 53 mbsf in both holes, and proposed that theseintervals may be correlative (Kennett, Baldauf, et al , 1994; Figure12). Close-up photographs of portions of these intervals are shown inFigure 14, together with CIE luminance (Y) data gathered fromLEAF high-resolution images which were captured in August 1994,19 months after the cores had been split and exposed to air. The lam-inations pictured in figure 11 of the site chapter can still be seen clear-ly in Figure 14. Linear correlation (r = 0.588; significant at the 99%

52


0.5

y 0.4

O S

5GY

- là

893A 100 r-893A

0.5893B

y o.4 -

0.3

/δGY '

^ ^ ^ ^ 5YR

0.3 0.4x

0.5

Figure 4. CIE 1931 chromaticity data for Holes 893A and 893B projectedinto the (x, y) plane of Figure 1. Data points shown are averages that attenu-ate the data set to one point every 1.25 cm.

confidence level) between these intervals was achieved by shiftingthe depth scale of the Hole 893A material relative to that of Hole893B upward (i.e., to shallower depths) by approximately 0.7 m. Thisdepth shift is more than half the magnitude of and opposite in direc-tion to the shift observed in shallower portions of the cores; however,it is impossible to tell from these data whether the depth difference isdue to loss of material during drilling or to differential compaction orerosion. Given the gassy nature of these cores, and the observed loss-es of material during their opening on the catwalk, it is more likelythat this depth difference is caused by drilling problems than by nat-ural phenomena.

CONCLUSIONS

The ODP color digital imaging system is a useful tool for loggingcolor changes in cores as a function of depth, although there remains

Y 50 -

Figure 5. Projection of CIE 1931 CIE chromaticity data (1.25 cm averageddata set) for Hole 893A into (x, Y) and (y, Y) coordinate planes. Heavy lineis the projection envelope of the three-dimensional CIE chromaticity space(MacAdam, 1935). Munsell 5Y hue surface is projected as a dashed enve-lope, with values shown as horizontal dashed lines.

room for improvement in automation of the data collection and edit-ing processes. The instrument captures and stores images in digitalformats that are not subject to the degradation commonly observed inphotographic films, so that useful studies of colors, lithologies, struc-tures, textures, bioturbation and other phenomena visible in imagesof the split-core surface can be conducted long after the original coreshave oxidized, faded, and desiccated. Standard image analyticaltools, such as ImagePro or NIH Image, can manipulate these imagesso they form an archive which will have value indefinitely.

The images can be used to map voids in detail, so that depth datafor other data sets (MST and discrete sample analyses) can be cor-rected automatically to eliminate spurious data and so that they canbe compared on comparable depth axes. Dense color data sets can begathered from such images, averaged or smoothed to suit the needs


893A

V1

\ 3/

893A1 i •

' Hi•. $§£

j i

• J « •

B•H$r:.•.

16

SC p

Lum

n

n

I•

i i

893B1

i i

4// 5 Y <

-

-

893B

0.32 0.42x

0.52 0.52 0.32

Figure 6. Enlargements of relevant portions of Figure 5, these plots are projections of 1931 CIE chromaticity data (1.25 cm averaged data set) for Holes 893Aand 893B into the (x, Y) and (y, Y) coordinate planes. Munsell 5Y hue surface is projected as a heavy solid line, with the 3/ and 4/ values plotted as horizontaldashed lines.

of the study, and plotted at scales comparable with data from discretemeasurements, so that correlations can be made with physical andchemical properties.

The Munsell Color System is inappropriate for use with instru-mental color measurements because it depends upon both the humaneye and very specific observational conditions. The CIE 1931 chro-maticity system is a better choice because it was designed for usewith instrumental techniques, and because it permits mathematicalmanipulation of the chromaticity (x, y, Y) data, including statisticaland time-series analyses which cannot be performed within the Mun-sell Color System. Projection of Munsell hue and value boundariesinto CIE chromaticity space confirms that the colors measured digi-tally within a few days of splitting agree closely with the MunsellSystem colors identified by the sedimentologists who performed thevisual descriptions.

Correlation of the color data between Holes 893A and 893B sug-gests that as much as 1.2 m of material may be present at or near thetop of Hole 893B which was not recovered from Hole 893A, and thatthe intervals between 50 and 53 m thought to be correlative on the ba-sis of visual inspection are in fact correlative. Correlation of the colordata with other continuous physical property data sets improves theinterpretation of both data sets (e.g., Rack and Merrill, this volume).

Through digital imaging, which can be accomplished with off-the-shelf equipment available for relatively modest cost, color prop-

erties of geological materials, particularly of fresh marine sediments,can be measured and studied quantitatively in concert with otherphysical and chemical properties. Once the images have been cap-tured, color analyses can be conducted in near real-time or delayedindefinitely without loss of information. Because the color data set isderived from images which remain available to the scientist while heor she is processing the data, it becomes a simple process to eliminateartifacts resulting from drilling disturbances, gas voids, or other un-natural disturbances of the split core surface. The density of colormeasurements is limited only by the scale of the image (e.g., mm2 ofsample per pixel), which is a choice made at the time of image cap-ture; thereafter, measurements can be made as densely as the scientistwishes, including the very high density needed for time-series or sta-tistical manipulations, and then can be decimated as required forcomparison with other data sets. Resulting color measurements canbe correlated with a variety of other sample, MST, and well-log mea-surements.

ACKNOWLEDGMENTS

We wish to thank Mr. Bradley Cook, who assisted with collectingthe core images. Ms. Randy Merrill has generously spent many longhours pulling the color data out of the images. Without the generous

54

0.38 0.42 0.34 0.46

200


Luminance (Y)10 12 14 16

Figure 7. CIE 1931 chromaticity data for Hole 893A (1.25 cm averaged data set) plotted against void-corrected depth (Tables 1, 2). Data have been smoothedwith a 5-point running average. Solid vertical bar in luminance plot corresponds to thick turbidite between 113.9 and 116.7 mbsf. Dashed vertical bar in y-chro-maticity plot parallels Core 146-893A-16H, which is enlarged in Figure 9.

0.34 0.38 0.42

-S- 40 -

0.34

80 -

100

Luminance (Y)8 10 12 14 16

Figure 8. CIE 1931 chromaticity data for Hole 893B (1.25 cm averaged data set) plotted against void-corrected depth (Tables 1, 2). Data have been smoothedwith a 5-point running average.

help of these people, this study would not have been possible. Wealso thank Alan Mix and Michael Schaaf for useful reviews. Finan-cial support for development of the ODP color digital imaging sys-tem has been provided by the Ocean Drilling Program of Texas A&MUniversity under Contracts 64-84 and 1-94 with the Joint Oceano-graphic Institutions, Inc.

REFERENCES

Agoston, G.A., 1987. Color Theory and its Application in Art and Design(2nd ed.): Berlin (Springer-Verlag).

Billmeyer, F.W., Jr., and Saltzman, M., 1981. Principles of Color Technol-ogy: New York (Wiley).

55


Kennett, J.P., Baldauf, J.G., et al., 1994. Proc. ODP, Init. Repts., 146 (Pt. 2):College Station, TX (Ocean Drilling Program).

Kodak, 1990. Handbook of Kodak Photographic Filters: Rochester, NY(Eastman Kodak Co.), Kodak Publ. #B-3.

Landscheidt, T., 1987. Long-range forecasts of solar cycles and climatechange. In Rampino, M.R., Sanders, J.E., Newman, W.S., Königsson,L.K. (Eds.), Climate: New York (Van Nostrand Reinhold).

MacAdam, D.L., 1935. Maximum visual efficiency of colored materials. J.Opt. Soc. Am., 25:361-367.

Mix, A.C., Harris, S.E., and Janecek, T.R., 1995. Estimating lithology fromnoninclusive reflectance spectra: Leg 138. In Pisias, N.G., Mayer, L.A.,Janecek, T.R., Palmer-Julson, A., and van Andel, T.H. (Eds.), Proc.ODP, Sci. Results, 138: College Station, TX (Ocean Drilling Program),413-427.

Mix, A.C., Rugh, W., Pisias, N.G., Veirs, S., Leg 138 Shipboard Sedimentol-ogists (Hagelberg, T., Hovan, S., Kemp, A., Leinen, M., Levitan, M.,Ravelo, C) , and Leg 138 Scientific Party, 1992. Color reflectance spec-troscopy: a tool for rapid characterization of deep-sea sediments. InMayer, L., Pisias, N., Janecek, T., et al., Proc. ODP, Init. Repts., 138 (Pt.1): College Station, TX (Ocean Drilling Program), 67-77.

Munsell, A.H., 1929. Munsell Book of Color: Baltimore (Munsell ColorCo.).

, 1963. A Color Notation: Baltimore (Munsell Color Co.).Newhall, S.M., Nickerson, D., and Judd, D.B., 1943. Final report of the

O.S.A. Subcommittee on the spacing of the Munsell Colors. J. Opt. Soc.Am., 33:385-418.

Rack, F.R., Gardner, J.V., Merrill, R.B., and Heise, E.A., 1994. Proxyrecords of Quaternary climate change from the Santa Barbara Basin:Ocean Drilling Program Site 893. Am. Geophys. Union, Spring Meet.,Baltimore. (Abstract)

Rack, F.R., Heise, E.R., and Merrill, R.B., 1993. Continuous logs of mag-netic susceptibility and color reflectance in Late Quaternary sedimentcores from the Santa Barbara Basin. Am. Geophys. Union, Fall Meet.,San Francisco. (Abstract)

Rea, D.K., Basov, I.A., Janecek, T.R., Palmer-Julson, A., et al., 1993. Proc.ODP, Init. Repts., 145: College Station, TX (Ocean Drilling Program).

Rogers, D.F., 1985. Procedural Elements for Computer Graphics: NewYork (McGraw-Hill).

Schaaf, M., and Thurow, J., in press. A fast and easy method to derive high-est-resolution time-series-datasets from drill cores and rock samples.Sediment. Geol.

Wyszecki, G., and Stiles, W.S., 1982. Color Science: Concepts and Methods,Quantitative Data and Formulae (2nd ed.): New York (Wiley).

Date of initial receipt: 31 August 1994Date of acceptance: 7 April 1995Ms 146SR-285

0.33 0.36 0.39 0.42 0.37

140

142 -

y0.4 0.44

Luminance (Y)8 12

-S144

Q.

Q

1461-

148 -

150

Figure 9. CIE 1931 chromaticity (x, y, Y) data from Core 146-893A-16H, plotted at one point per 0.5 cm, and smoothed with a 5-point running average.


CORE;;°

6H

PHOTOGRAPHED

Figure 10. Archival photograph of Core 146-893A-16H. Note the intercalated beds of dark and light olive gray clay visible in the upper three sections, and themore massive clay beds in Section 5 and below. The lithologic variations are clearly visible in the color data plotted in Figure 9, with light gray beds corre-sponding to positive digressions in Y and to negative digressions in x and y values.

57


500

250 -

500

E 250 -

Luminance (Y)12 2

1 1 1 1 1

- Ill J 1

1 1

31

1 ' 1

A k23

i i i i

17

1 • '

12

I i

y ;

•

3UU '

250 •

90

138

i • • •

31

1 1 1 1 1

17

I ' '

12

x -

-

-

-

0.005 0.01 0.015Frequency

0.02

Figure 11. Power spectra of CIE chromaticity data for Cores 146-893A-1Hthrough -5H, which cover the last 30 ka. Sample interval is 0.2 year, soNyquist frequency is 2.5 cycles/year. Prominent peaks are denoted with peri-odicity in years.

Luminance (Y)4 6 8 10 12 14 4 6 8 10 12 14

Figure 13. Luminance (Y) peaks correlated between holes in the upper 10 mof Holes 893A and 893B, plotted at original depth, prior to correlation. SeeTable 4 for precise depths of peaks correlated, and Figure 12 for a plot of theupper 50 m, after correlation. Dashed straight lines connect pairs of peaksused in the linear correlation. A shift of 1.2 m applied to depths in Hole 893 Ais required to correlate these peaks.

Table 4. Depths (mbsf) correlated between Holes 893A and 893B in Fig-ures 12 and 13.

893A

1.49032.30153.26574.00254.8395.25556.9519.9766

12.31417.66334.41242.007

893B

2.72063.60544.55675.39146.28286.83958.5781

11.96914.14419.40135.48643.343

Figure 12. Correlation between luminance (Y) data from the upper 50 m sec-tions of Holes 893A and 893B. Depths of peaks from Hole 893A were lin-early correlated with those of Hole 893B using Analyseries (provided by L.Labeyrie). 4357 data points were used from each hole, smoothed with a 5-point running average to reduce the impact of minor events. Resulting corre-lation coefficient (r) is 0.683. Correlation peak picks are detailed in Table 4and Figure 13.

893A6H, 6

893B7H, 2


893A 893BLuminance (Y)

4 6 8 10 1251.55

-fa.•—I

%:>

U 3

;

en

GO

en

oo

t o

: : ;

• • : ; i

4:*

- 51.65Q.ΦQ

- 51.7

51.75

Figure 14. Images and high-resolution CIE 1931 luminance (Y) data from interval 146-893A-6H-6, 37.1-53.4 cm, correlated with those from interval 146-893B-7H-2, 29.6-46.1 cm. These images were captured in color (2048 × 2048 × 42-bit pixels) with the LEAF digital camera back. The unsmoothed luminancedata shown on the right were taken from color versions of the images printed on the left. These intervals are part of the finely laminated portions of Subunit 1Cfor which the scientific party proposed correlation on the basis of visual similarity (Kennett, Baldauf, et al., 1994, figure 11, p. 30). To accomplish the correla-tion, Hole 893A depth values have had to be shifted upward in the column by 0.707 m ± 0.0018 m, resulting in a correlation coefficient of 0.588 between thetwo data sets (690 points each). Measured luminance values from Hole 893A have been offset (by -5) in this figure to facilitate comparison with the Hole 893Bdata.

59

4. the odp color digital imaging system: color logs of

Documents