SHORE-BASED INTERPRETATION OF DOWNHOLE MEASUREMENTS AT ODP HOLE 735B³

A shore-based interpretation of wireline logs from ODP Hole 735B was conducted after Leg 176. The aim is to interpret logging data for a continuous lithologic reconstruction and to make the results available to the shipboard scientific party as a basis for further investigation. Similar studies were performed on several sites drilled during Leg 173 (Delius et al., 1998). Interpretation of downhole measurements provides information about the lithology, stratigraphy, and structure of the drilled lithology. This is especially important in holes with low core recovery. In holes with good core recovery, log interpretation may be based on a core-log correlation. Lithologic interpretation of logging data from Hole 735B was conducted to (1) identify the major rock types, (2) characterize the rock types by their in situ physical properties, and (3) reconstruct a continuous Electrofacies-Log (EFA-Log) lithologic profile.

Methods

The EFA-Log interpretation technique used in this study is a numerical procedure based on a core-log correlation to support geological interpretations. The calibration of the EFA-Logs was conducted before Leg 176 on the basis of the Leg 118 core and log data. After Leg 176, the calibration was checked with the new log data, and by using discriminant analysis, the interpretation was extended to the newly drilled section. These two steps, calibration and discriminant analysis, are explained below.

Calibration

Information about the lithology was taken from the core descriptions in Robinson, Von Herzen, et al. (1989) and the lithostratigraphy from Dick et al. (1991a). The aim of the calibration is (1) to determine the principal relations between the physical logs and the core lithology, (2) to find out which rock types can be distinguished using the logs (this is especially important in holes such as Hole 735B, in which most rock types are defined by mineralogical criteria), and (3) to define the electrofacies, the set of log responses that characterizes a rock type and distinguishes it from others (Serra, 1984). To derive relations between the lithologic characteristics and the tool responses, the data were displayed as two- or three-dimensional crossplots. Trends in the crossplots reflect the dependence of the tool responses on mineralogical, geochemical or structural variations in the rocks. In this way, definite log values are assigned to lithologic characteristics and each rock type is classified by an electrofacies (Pechnig et al., 1997).

After the calibration based on the Leg 118 data, the assignment of the electrofacies was checked with the Leg 176 log data. Only slight differences, mainly caused by different generations of tools used in the two legs, were observed. For example, during Leg 118 the Schlumberger compensated neutron tool was used to measure neutron porosity whereas during Leg 176 the Schlumberger APS was applied. The two tools are described in Eberli, Swart, Malone, et al. (1997). The transfer of the electrofacies to the newly drilled depth intervals was done by discriminant analysis. To do this, seven depth intervals characterized by well-defined, homogeneous electrofacies and a good correlation between cores and logs were chosen as key intervals. The key intervals are 85-95, 105.5-118, 130-139, 179-187, 242-295, 300-317, and 320-350 mbsf.

Discriminant Analysis

The discriminant analysis serves the following two objectives. First, it assesses the accuracy of a predefined classification. In this study, the predefined classification corresponds to the classification carried out in the calibration step. Second, the discriminant analysis gives a prediction for the classification of unclassified cases (Backhaus et al., 1989). This second objective is used here to predict and classify the new depth intervals. The analysis uses the logs to calculate linear discriminant functions that best separate the different electrofacies. These functions are used to calculate the probability for electrofacies classification of a particular depth point. The following Leg 176 logs have been included into the discriminant analysis: electrical resistivity deep (LLd), neutron porosity (APLC), density (RHOB), photoelectric factor (PEF), capture cross section (SIGF), compressional wave velocity (V_p), shear wave velocity (V_p), and the caliper log from the FMS tool. Descriptions of the tools are given in "Principles and Uses of the Tools" (in "Downhole Logging") in the "Explanatory Notes" chapter. Further information about the use of the discriminant analysis for the interpretation of well-logging data can be found in Doveton (1994).

After the transfer, the result of the discriminant analysis was checked and corrected. Attention was focused on depth intervals with borehole enlargements, alternating layers of small thickness, and low calculated probabilities for classification. Finally, a continuous synthetic lithologic log, the EFA-Log was established for the entire logged interval of the hole.

Results

We defined four electrofacies during the calibration with the Leg 118 data and transferred these electrofacies onto the entire logged interval of Leg 176. Figure F138 shows the very good agreement between the logging data from Leg 118 and Leg 176. The data sets displayed in Figure F138A and F138B show nearly identical ranges and trends. The definition of the electrofacies with individual value ranges or trends in the logs is the same in the Leg 118 data, which were used for the calibration, and in the Leg 176 data.

The electrofacies can be described and characterized as follows. Quantitatively, the most important electrofacies makes up 73% of the logged interval. The correlation with the cores shows that this electrofacies comprises a variety of gabbros. The classification of these rocks by the shipboard petrologic teams is based on varying amounts of plagioclase, olivine, orthopyroxene, and clinopyroxene (see "Rock Classification" [in "Igneous Petrology"] in the "Explanatory Notes" chapter, this volume). These differences in mineralogy do not significantly affect the physical properties of the rocks; therefore, most of the gabbroic rocks cannot be resolved as individual electrofacies. This electrofacies is named olivine gabbro because olivine gabbro is the most abundant rock type. It is characterized by high electrical resistivities; the average value of the LLd is 3400 m (Table T20). However, the values show a broad range from 100 to 34,000 m. The log density values range from 2.7 to 3.1 g/cm³; the average value is 2.87 g/cm³. These broad ranges in resistivity and density are probably related to variations in mineral content or secondary mineralization.

The oxide-rich gabbros form 19% of the EFA-Log. This electrofacies comprises rocks that are classified as different types of oxide gabbros in the core stratigraphy. Rocks with disseminated oxides could not be assigned to this electrofacies. The characteristic feature of the oxide-rich gabbroic rocks is the presence of Fe-Ti oxide minerals. The presence of these minerals causes the electrical resistivity to decrease because of electronic conduction (Pezard et al., 1991) and the density to increase because of the higher density of the ore minerals. Thus, electrical resistivities are very low, with an average value for the LLd of 135 m. The density measurements are high and the values range from 2.7 to 3.3 g/cm³ with an average of 3.03 g/cm³. An additional log that shows very characteristic responses for oxide gabbros is the photoelectric factor. The photoelectric factor depends on the average atomic number of the constituents of the rocks and is thus an indicator for mineralogy. The oxide-rich gabbros show higher values of the photoelectric factor than the olivine gabbros. The average value in the oxide-rich gabbros is 5.6 barns/e^-. In the olivine gabbros, it is 4 barns/e^-.

Altered zones form 5% of the EFA-Log. These rocks are characterized by intermediate neutron porosities with an average value of 9%. The neutron porosity log is sensitive to the hydrogen content of the formation. In these gabbroic rocks, hydrogen is bound in hydrous alteration minerals or in seawater that fills fractures. Frequently occurring hydrous alteration minerals are talc, clay minerals, smectites, and amphiboles (Robinson, Von Herzen, et al., 1989). Electrical resistivities are intermediate as well. The average value of the LLd is 920 m. The density values are >2.5 g/cm³, with the average value being 2.7 g/cm³.

Fractured zones are of minor occurrence and make up 3% of the EFA-Log. They are identified by high neutron porosities (average = 23%), intermediate to low electrical resistivities (LLd average = 320 m), and low density values (average = 2.41 g/cm³). Fractured zones are associated with depth intervals of reduced core recovery and borehole enlargements. The rocks are mostly disintegrated and saturated with seawater so that the log responses are strongly influenced by the presence of seawater. Some of the depth intervals that have been classified as fractured zones correspond to fault zones (e.g., at 560 mbsf; see "Structural Geology," "Physical Properties," and "Downhole Logging").

The electrical resistivity, density, neutron porosity, and photoelectric factor are the most important logs to distinguish between the electrofacies. Figure F138 outlines the specific trends in the electrical resistivity and density for each electrofacies. Olivine gabbros show high electrical resistivities and density values that are typical of gabbros. In oxide-rich gabbros, the log responses clearly reflect the presence of the conductive and very dense oxide minerals. Electrical resistivity decreases, whereas the density increases. Altered and fractured zones are influenced by the presence of alteration minerals and seawater that fills fractures. Thus, electrical resistivities and densities are intermediate to low.

In Figure F139 the EFA-Log summarizes the electrofacies stratigraphy for the logged interval compared to the actual lithostratigraphy (Dick et al., 1991a; see "Igneous Petrology" and the "Core Descriptions" contents list). The core observations show much greater variability than the EFA-Log because the logging data are unsuitable to distinguish differences in rock types based on slight variations in primary silicate mineralogy, grain size (e.g., the various types of microgabbro), or other macroscopic features (e.g., patchy olivine gabbro). Thus the 21 rock types observed on board were reduced in the EFA-Log to oxide-rich gabbro and olivine gabbro.

A very good correlation exists between the different types of oxide-rich gabbros in the core stratigraphy and in the EFA-Log. This not only applies for the correlation of single layers in the two profiles. The EFA-Log gives a very good estimate of the total amount of oxide-rich gabbros in the logged interval. Oxide-rich gabbros make up 19% of the EFA-Log and 17.7% of the core stratigraphy. However, in Figure F139, oxide gabbros and disseminated oxide gabbros have the same pattern so that in some depth intervals, these correlations are not obvious. This is especially the case between 170 and 223 mbsf, where most of the rocks are disseminated oxide gabbros (Robinson, Von Herzen, et al., 1989). Rocks containing disseminated oxides can not be distinguished from olivine gabbros. The threshold value of 2% oxide content applied by the shipboard petrographic team to distinguish between disseminated and oxide-rich gabbros ("Metamorphic Petrology" in the "Explanatory Notes" chapter) apparently corresponds to the detection limit of the logs. Besides the lithologic information, the EFA-Log gives information about structural aspects such as alteration and fracturing. Altered zones often give hints for paleofluid circulation, whereas fractured zones can be related to recent pathways. This additional information may help to reconstruct a comprehensive model of the local geology.