BULK DENSITY ESTIMATION BY GAMMA RAY ATTENUATION DENSIOMETRY

Bulk density is a measure of mass per unit volume, typically expressed as grams per cubic centimeter (g/cm3). In Earth sciences, bulk density is often an indicator of changes in lithology (mineral composition, grain size, and other physical characteristics) and porosity (the spaces between mineral grains that can be filled with gas or fluid). The correlation of bulk density and porosity to other properties of rocks and sediments led to the development of the GRAPE by Marathon Oil Company during the 1960s (Boyce, 1973). The principle is based on interaction of medium-energy gamma radiation with rock or sediment by Compton scattering. The mass attenuation coefficients for most rock-forming elements are similar; therefore, the attenuation of gamma radiation can be directly related to the density of the material.

ODP started where DSDP left off. Much of the scientific expertise and laboratory equipment were transferred from DSDP, including the GRAPE system. This allowed scientists sailing on the first ODP cruise to continue collecting this type of data, which had been measured since early in DSDP. The bulk density data set as measured by the GRA systems provide a very large, densely sampled record of bulk density for >80% of the 222 km of core collected throughout the world by ODP.

Data Acquisition

ODP started operations with the DSDP GRAPE hardware and software. Basically, the GRAPE system consisted of a drive device that moved a section of core between a shielded gamma ray source (133Ba) and a shielded scintillation detector. Modifications and upgrades were made as improvements in computers and data acquisition technology became available, and scientific objectives changed. The system eventually was referred to as the GRA densiometer because the device measurements were used to calculate density rather than porosity. An MST (automated core conveying and positioning system) was installed during Leg 124E with an upgraded GRA system. This new GRA system finally retired the last of the GRAPE components as a new source (137Cs) and NaI scintillation detectors were installed.

Table T5 briefly outlines the modification history of the systems used to collect bulk density data. The acronym "GRAPE" will be used when referring to measurements taken with the original 133Ba source and detectors of the DSDP GRAPE system. Likewise, "GRA" will be used when referring to the ODP system with the 137Cs source and NaI scintillation detectors or when referring to the bulk density data set as a whole. More detailed descriptions of both DSDP and ODP bulk density measurements can be found in Boyce (1973, 1976) and Technical Note 26 (Blum, 1997).

Standard Operating Procedures

At the beginning of ODP, the Shipboard Scientists' Handbook (1990) instructed that the sections could be run through the GRAPE analyses while they stabilized for thermal conductivity measurements. After additional sensors whose measurements were temperature sensitive were added to the track, cores were stored on a rack to allow them to equilibrate to room temperature before analysis. The highest quality GRA data were made on core liners that were completely full. It was recommended that only APC cores be analyzed because APC coring routinely recovered soft sediment that filled the core liner. XCB and RCB cores were often disturbed, containing biscuits of core surrounded by drilling mud or irregular pieces of core that did not completely fill the core liner. With the older DSDP track, it would often take as long as 2 hr to take GRAPE measurements on one core. Considering the amount of time required to take these measurements and the poorer quality of data, hard rock cores and disturbed cores were not usually run in continuous mode. These cores were sometimes analyzed by GRAPE-2, a longer count density measurement taken on samples or discrete locations on a section.

Calibration

Two different calibration procedures were used during ODP (Table T6). Aluminum was chosen for the calibration standard because aluminum has an attenuation coefficient similar to common minerals. The calibration standard used until Leg 168 consisted of two aluminum cylinders of different thickness mounted in a liner. The thicker rod had a density of 2.7 g/cm3 and the thinner rod had a density of 1.00 g/cm3. With this procedure, however, the density of water was overestimated by ~11% (Boyce, 1973). A fluid correction was applied to the bulk density estimate to compensate for the overestimation of water density.

Data collection and calibration procedures set up and described by Boyce (1973, 1976) were used throughout the first part of ODP even though modifications had been made to the system. Installation of the MST and new GRA system marked a major change to both hardware and software. Documentation of calibration procedures, frequency of calibration, and the calibration parameters used for the density calculations were difficult if not impossible to find before Leg 124, when the MST was installed with a major software upgrade. After that system was installed, files were created that contained the analysis of the standard; however, these files were not always saved, and the calibration history was not documented. The software upgrade during Leg 163 resulted in a major change in the data file format. The calibration date and parameters were written in the header of the data file. From this point, the calibration history of GRA data was documented.

A new calibration procedure was implemented during Leg 169. This new procedure incorporated a two-phase standard of a telescoping aluminum rod (five elements of varying thickness) and pure water. Because of the two-phase standard, the fluid correction was no longer necessary because water was used in the calibration procedure. For a full discussion of the calibration procedures see Technical Note 26 (Blum, 1997).

Archive

Pre-Janus Archive

Most of the original GRAPE and GRA data files were archived on the ODP/TAMU servers. There was no interim database for GRA data. In a few instances, the files for a hole were concatenated into a single file. Some of these original files are no longer available, either because the scientists who concatenated the file deleted them or they were not moved onto the ODP/TAMU servers.

Migration of GRA Bulk Density to Janus

The data model for GRA bulk density can be found in "Janus GRA Densiometer Data Model" in "Appendix C." Included are the relational diagram and list of the tables that contain data pertinent to GRA, column names, and definition of each column attribute. ODP Information Services Database Group was responsible for the migration of pre-Leg 171 data to Janus. The migration of GRA bulk density was done in conjunction with MSL, PWL, and NGR MST data sets. Each change in format was documented and added to the MST migration program. Additional information about the migration of GRA data or original file formats can be requested from the IODP/TAMU Data Librarian.

As noted in the calibration section above, the raw bulk density value overestimated the density of water. The raw data files created before Leg 133 contained the uncorrected bulk density, calculated as

GRA_CALIBRATION.density_M0 + GRA_CALIBRATION.density_M1 x
ln(GRA_SECTION_DATA.meas_counts/
GRA_SECTION_DATA.actual_daq_period). (1)

After Leg 133, the Boyce density correction was applied to the density value and written in the data files. During the migration of GRA data to Janus, the Boyce density correction

Boyce corrected density = [(density – 1.128) x 1.626/1.522] + 1.024 (2)

was applied to all the bulk density values that had not already been recalculated (Legs 101–133, Site 818).

Janus GRA Data Format

GRA data can be retrieved from Janus Web using a predefined query. The GRA bulk density query Web page allows the user to extract data using the following variables to restrict the amount of data retrieved: leg, site, hole, core, section, specific run number, range in density values, depth range, or latitude and longitude range. In addition, the user can use the output raw data option in the query to extract the raw measurements and calibration parameters used to calculate the bulk density values. Because there are more than 9.2 million GRA data records in Janus, a user must restrict the amount of data requested.

Table T7 lists the data fields retrieved from the Janus database for the predefined GRA query with output raw data option turned on. The first column contains the data item, the second column indicates the Janus table or tables in which the data were stored, and the third column is the Janus column name or the calculation used to produce the value. "Description of Data Items from GRA Query" in "Appendix C" contains additional information about the fields retrieved using the Janus Web GRA query and the data format for the archived ASCII files.

Data Quality

Several things can affect the quality of GRA data. The type of cored material and drilling method used to recover the core are major factors. APC coring, used to recover softer, undisturbed sediments, routinely gives the best results because the core liner is usually full. However, the sediments can also contain gas, which creates voids in the cored material. Cores cut by XCB and RCB coring are often biscuits surrounded by drilling mud or irregularly shaped pieces. Voids, smaller diameter core, irregular pieces, and thin runny mud all give low GRA density values. Table T8 summarizes how much of the different types of core were analyzed on the GRA systems.

Even though the Shipboard Scientists' Handbook (1990) specified that GRA measurements should only be made on APC cores, the addition of other sensors to the GRAPE track and eventually the MST made it more of a temptation to run less than ideal cores through the GRA system. This did not mean that GRA density analyses on XCB and RCB cores have become more accurate. GRA values on anything but APC cores should be used with some skepticism. For more information, see Chapter 3 in Technical Note 26 (Blum, 1997) and Technical Note 36 (ODP Science Services, 2006).

A couple of mechanical factors also affect the quality of GRA density measurements. The GRA systems have always been installed on a track that either moved the section past the source and detectors or moved the source and detectors along the vertical section. Sample measurements were a function of the speed of the track and sampling time. Slight variation of track speed may account for the irregular spacing of data points.

Core sections were run through the GRA system before the liners were opened and the core curated. During the curation process, core material often shifted. In sedimentary cores, voids may have closed. Gassy cores may have small voids that continue to enlarge after analysis. Sections may not be completely full, and material may have spread throughout the liner. After curation, this material was pushed up to close voids and the section's curated length was less than originally analyzed. The effects can be seen when looking at the data for a section: (1) there are reasonable density values beyond the curated length of the section (null depth values) and (2) there are negative density values within the section indicating a measurement in a void or less than full liner.

Hard rock cores can be continuous cylinders with uniform diameter or can be broken into small irregular pieces. The curation process shifts hard rock pieces, sometimes even shifting core material from its original liner section to an adjacent section liner. Where the core material was in its liner during analysis and where it was eventually placed after curation can be very different. GRA data for these types of cores should be used with caution.

Operators may also be a source of error. Throughout ODP, the operator manually entered core information into the data acquisition program. Typographical errors or entering wrong information occasionally happened, and some mistakes were not identified. Sometimes, the scientific party noticed the error and corrected it for the data included in the Initial Reports volume, but the original files were not corrected. A lot of effort during verification of the migrated GRA data has gone into finding sections that may have been misidentified. Some runs have been renamed to different sections. The evidence for misidentification had to be conclusive. Some of the clues used to find incorrectly identified analyses are

  1. Two runs for a section and no run for the following section;
  2. Run numbers out of sequence;
  3. Two runs for a section, run numbers out of sequence (no data for that core and section in a different hole, but sequence of run numbers would be correct); and
  4. Nature of the core material (length of core, voids, or less than full liners).

NEXT