In this and subsequent site chapters, we combine physical volcanology and igneous petrology discussions. We sought to determine the types of eruptive activity and the nature of the eruption environments involved in the formation of the volcanic basement successions recovered during Leg 197. In our initial description of each core we used the nongeneric scheme outlined in Figure F11, but as our knowledge and understanding of the basement lithostratigraphy improved, we adopted a more generic volcanic classification based on the scheme outlined in Figure F12. The overall procedures used here are similar to those developed by the Leg 183 Shipboard Scientific Party (2000).
The emphasis of our shipboard studies was to produce an integrated picture of the style of volcanism and environmental setting of each site drilled. This was achieved by systematic rock descriptions and identification of key volcaniclastic and igneous textures that are diagnostic of specific modern physical processes. Additional information presented in the "Lithostratigraphy," "Physical Volcanology and Igneous Petrology," "Downhole Measurements," and "Alteration and Weathering" sections of each site chapter were important in producing the final interpretation.
The physical description of volcanic rocks and deposits recovered during Leg 197 was a multistage process. The first step involved defining boundaries of lithologic units by either visual identification of contacts or by inference from changes in igneous or volcaniclastic textures. This was followed by general description of the characteristic igneous textures, including macroscopic flow textures and structures and petrography (Fig. F11).
Our initial justification for placing unit boundaries changed as our understanding of the site improved, especially in light of data obtained from chemical analysis of identified lithologies and from physical properties and downhole measurements, which reflected major physical changes in the core (e.g., fragmental vs. coherent). Internal unit contacts (i.e., lava lobe boundaries or obvious bedding planes in volcaniclastic units) were documented wherever discernible, but more subtle boundaries, such as those between individual lava lobes, were usually not broken out despite their possible significance to the volcanological interpretation.
We examined individual lithologic units in more detail to determine their characteristic volcanological architecture (i.e., types and number of lithofacies as well as the processes responsible for their formation) (Figs. F12, F13). The observed lithofacies associations laid the foundation for a genetic volcanological interpretation of each lithologic unit (i.e., types and number of lava flows/lobes or depositional subunits). We also attempted to synthesize the observations from the different units, combining the data from the "Physical Volcanology and Igneous Petrology" and "Downhole Measurements" sections of each site chapter. This examination was particularly useful for comparing different sites.
In the site chapters we use nongenetic terminology in the description of each unit unless the origin of a particular unit is known beyond a doubt. Consequently, genetic phrases such as "chill zone," "pillowed," "hyaloclastite," or "flow-top breccia" are only used after proper lithologic identification was achieved. A unit-by-unit description and interpretation with explanations of the criteria used to separate the different units follows.
Traditionally, mafic lava flows have been divided into pillow, pahoehoe, and a'a flows (Macdonald, 1953, 1967) (Table T2). This morphological division is important because (1) the mode of lava emplacement for a'a and pahoehoe flows is fundamentally different and (2) despite similar emplacement mechanisms for pahoehoe and pillow lavas, they form in vastly different environments. A'a flows move like the treads on a bulldozer and are typified by a thermally inefficient mode of emplacement in open channels where they disrupt and mix their upper crusts. Pahoehoe and pillow lava flows, on the other hand, are characterized by insulating transport and growth by sequential lobe-by-lobe emplacement. The lava is transported in internal pathways (or lava tubes) to the active flow fronts, where they advance by inflating a lobe with a continuous crust, much like filling a rubber balloon with water (Walker, 1991; Hon et al., 1994).
A wide range of intermediate flow types occur between these two end-member types. Most subaerial transitional flow types have "pahoehoe" in their names: for example rubbly pahoehoe, slab pahoehoe, and toothpaste pahoehoe (Macdonald, 1953, 1967; Rowland and Walker, 1987; Keszthelyi and Thordarson, 2000; Keszthelyi et al., 2000), but some are closer in character to a'a than to pahoehoe (Table T2).
Determining lava type is relatively straightforward if centimeter-scale morphologic features of the flow tops and bottoms are recovered and well preserved. A'a flows are characterized by angular, spinose clinker at both the flow tops and bottoms, whereas pahoehoe and pillow flows are characterized by smooth tops and bottoms (Figs. F14, F15). Transitional flows show some of the characteristics of both a'a and pahoehoe lava flows (Table T2; Fig. F16). However, these features are most susceptible to erosion and alteration and are often not recovered in drill core.
Most subaerial lava flows are vesicular, although their porosity as well as vesicle shapes and size distributions vary widely. However, the vesiculation patterns (i.e., vesicle abundance, shape, and size distribution) vary systematically between flow types and typically exhibit enough constancy within each type that it is possible to use this property in conjunction with changes in petrographic textures and flow structures to make a positive identification (Table T2).
A'a clinker primarily forms by the rupture of molten lava at the flow surface (Fig. F15). This can happen only where strain rates are high and/or the viscosity of the fluid lava is high (Peterson and Tilling, 1980). This tearing is a result of the non-Newtonian behavior of lava under these conditions. Large (10-200 cm wide) "arms" or "fingers" of lava from the massive interior of the flow commonly extend into the breccia, especially along well-defined shear zones (Lockwood and Lipman, 1980). The vesicle abundance in a'a flows is typically low (<20 vol%), and the vesicles tend to be distorted, drawn out into elongate shapes by stretching or viscous deformation (Macdonald, 1953, 1967). A'a flows also commonly contain partially resorbed pieces of entrained crust within the massive lava interior of the flow. These typically appear as fist-sized clots of small, highly distorted vesicles within the nonvesicular part of the flow. Entrainment of the breccia clasts into the interior of the flow is a powerful heat loss mechanism for a'a flows (Crisp and Baloga, 1994), driving more rapid crystallization of the flow interior (Crisp et al., 1994) and leading to more efficient degassing.
Pahoehoe lavas are usually highly vesicular, with bulk porosity in excess of 20 vol% (range = 20-60 vol%) (Fig. F14). Thin pahoehoe lobes are often vesicular throughout and exhibit a gradual coarsening in vesicle size from lobe margins to the interior (Wilmoth and Walker, 1993). Thicker pahoehoe lobes are typically characterized by the threefold structure of vesicular upper crust, a dense core, and a thinner vesicular lower crust (Aubele et al., 1988; Thordarson and Self, 1998). To avoid confusion with the drilled core, we use the term massive lava interior instead of core throughout this volume. For other key structures of pahoehoe lavas we use the terminology from Thordarson and Self (1998) (Tables T3, T4). Key vesicle features are horizontal vesicular zones in the upper crust, horizontal vesicle sheets defining the boundary between the upper crust and the massive interior of the flow, vesicle cylinders, and pipe vesicles at the base (Fig. F14). The vesicles in the uppermost 1-10 cm (i.e., top of the upper crust) and lowermost 10-50 cm (i.e., the lower crust) of a pahoehoe flow are usually relatively small (0.1-3.0 mm in diameter) and highly spherical. However, the vesicular zones in the upper crust can vary in thickness from 0.5 to 200 cm and typically contain spherical vesicles 2-50 mm in size (Fig. F14). Commonly, they are marked by relatively sharp upper boundaries and grade downward into a zone of relatively low vesicularity. Vesicle sizes usually coarsen downward across each zone. The vesicles in the upper crust are thought to represent bubbles formed during lava emplacement caused by periodic drops in the internal hydrostatic pressure, which are due to sequential surface breakouts at the active lava front.
Pipe vesicles, vesicle cylinders, and horizontal vesicle sheets represent structures produced during lava emplacement by gas-driven melt segregation from the viscous part of the basal crust after 35%-50% crystallization (Thordarson and Self, 1998; T. Thordarson, unpubl. data). These features represent a population of vesiculation structures that are very different from the vesicular zones of the upper crust of pahoehoe flows (Fig. F17). Pipe vesicles are typically 0.3-0.7 cm in diameter, 3-15 cm in height, most commonly occur at the base of lobes, and are reliable indicators of flow margin. Pipe vesicles may grade into vesicle cylinders, which in turn lead to 1- to 20-cm-thick horizontal vesicle sheets at the base of the upper vesicular crust (Thordarson and Self, 1998). Vesicle cylinders and horizontal vesicle sheets typically contain vesicular segregated differentiates that can have micropegmatite textures and are synonymous with segregation veins (e.g., Goff, 1996). The vesicles in these segregation structures are usually 1-5 mm in size with highly irregular outlines and commonly exhibit evidence of growth by coalescence. We paid special attention to occurrences of segregated melt structures because their more evolved composition (richer in K than host lava) and feldspar-dominated mineralogy makes them ideal targets for 40Ar-39Ar dating.
We can use the thickness of the upper crust of an inflated pahoehoe lobe to calculate the duration of active flow. The thickness of the upper vesicular crust is interpreted as a direct measurement of the amount of lava that solidified while new lava was being injected into the flow. A cooling model allows the translation of this thickness into time. We estimate the duration of active flow using the Hon et al. (1994) empirical cooling model as adapted by Thordarson (1995). This technique and its limitations and errors are described in more detail in Self et al. (1997, 1998) and Thordarson and Self (1998).
Transitional lavas fall into three broad categories:
Lava flows that entered water or erupted subaqueously may form pillow basalt, hyaloclastites, or peperites. The latter two are volcaniclastic sediments that form by quenched fragmentation. Pillow basalt formed in deep water (1000 m depth) can usually be distinguished from pahoehoe flows by the thicker glass rinds, significantly lower vesicularity (<5 vol%), and the presence of marine sediment between the pillows. However, pillow basalt formed in the upper 1000 m of the water column often consists of moderately to highly vesicular lava, and thus care needs to be taken to differentiate it from compound pahoehoe lavas constructed from decimeter- to meter-sized lobes. Larger subaqueous flows may form sheetlike flows (Ballard and van Andel, 1979), and their morphology is controlled by local flow rate, which is primarily a function of slope and eruption rate (Gregg and Fink, 1995).
After defining the main lava sequences, we examined each unit in more detail to record its internal architecture. Although we made every effort to have unit boundaries reflect individual lava packages, the term "unit" should not be considered synonymous with "lava flow" in this volume.
We made general observations during the initial lithologic description of the rocks, such as intervals of coherent and fragmental lava facies along with gross changes in vesicularity and crystallinity (Fig. F11). Often, time did not permit detailed examination of all the lobes and we subjected only a subset to full scrutiny. This was followed by more systematic description of units concentrated on documenting vertical changes in crystallinity and vesicularity as well as occurrence and orientation of macroscopic vesiculation structures. Definitions and abbreviations of the categories used in reporting the type, size, and abundance of these properties are given in Table T5.
We paid special attention to vesiculation features by recording (1) volume percentage, (2) average size range, (3) shape, and (4) grading (fining upward or coarsening upward) of the vesicles at intervals appropriate for the variability shown in the core (typically across 10- to 100-cm intervals). The volume fraction of vesicles was estimated using visual percentage estimate charts and was grouped into four categories: nonvesicular (<1%), sparsely vesicular (1%-5%), moderately vesicular (5%-20%), and highly vesicular (>20%). The average vesicle size is a measure of what appeared to be a typical vesicle size. Special attention was paid to gradual changes in vesicle size because such size grading when coupled with changes in groundmass texture (i.e., fining of grain size and appearance of feathery or dendritic crystal forms) was very helpful in locating lobe boundaries when quenched glassy lobe margins were not recovered. We estimated vesicle sphericity using the charts produced for describing sedimentary rocks. During these measurements, we also noted evidence of growth by coalescence, occurrence, and orientation of elongated vesicles.
Finally, during the synthesis of the various data sets, we estimated the thickness of the various units and flows. It is very important to note that ODP curation procedures assume that the top of recovered material is from the top of the cored interval. This can lead to large (up to 9 m) differences between the curated depth and the actual depth. Downhole logging measurements were required to produce more accurate estimates of flow thicknesses.
In fragmental lava facies we documented volcaniclastic textures using standard sedimentological techniques (i.e., clast vs. matrix modal proportions, clast size, shape, sorting, and lithology) (Fig. F11). Of particular interest was the presence (or absence) of features diagnostic of viscous or quenched fragmentation while the lava was hot. These include clasts engulfing fragments of other (earlier) clasts, welding, glassy margins surrounding the clasts, extensions of the interior of the flow into the breccia, entrained clasts within the interior of the flow, and the presence of a basal breccia. We used changes in clast morphology and crystallinity as well as changes in vesicle abundance and shape to identify clast types within the breccia.
Our identification and interpretation of the lava types involved three basic steps: (1) we initially tied the observed lava facies associations and structures to physical processes, (2) the emplacement style of individual flows was then inferred, and (3) the environmental setting (e.g., subaerial or submarine) for the whole sequence was assessed. The lava flow types and the criteria for their identification is listed in Tables T3 and T4. Our interpretations relied most heavily on observations from active volcanism in Hawaii and on studies of the physical volcanology of subaerial and subaqueous lava flows in Iceland and the Columbia River Basalt Group in the United States (e.g., Swanson, 1973; Walker, 1991; Mattox et al., 1993; Hon et al., 1994; Keszthelyi and Self, 1998; Keszthelyi et al., 2000; Thordarson, 2000; Thordarson and Self, 1998).
We used "volcaniclastic" as a nongenetic term for any fragmental aggregate of volcanic parentage containing >60% volcaniclastic grains and <40% other types of clastic and/or biogenic material. This definition is broader than that for pyroclastic deposits because the term "pyroclastic" strictly applies only to products of explosive volcanic activity and includes hydrovolcanic deposits formed by explosive interaction between magma and water and nonexplosive quenched fragmentation (i.e., hyaloclastite and peperite). This definition also includes epiclastic sediment (the volcanic detritus produced by erosion of volcanic rocks); the term "volcaniclastic" does not imply any active volcanism at the time of deposition.
The subclassification of volcaniclastic sediment followed here differs from the standard ODP classification (Mazzullo et al., 1988) in that we adopt a descriptive (nongenetic) terminology similar to that employed during ODP Leg 183 (Coffin, Frey, Wallace, et al., 2000). Unless a pyroclastic origin for sediment could be defined, we described deposits of volcanic provenance (volcaniclastic) according to the classification scheme for clastic sediment, noting the dominance of volcanic grains. We followed the clastic textural classification of Wentworth (1922) to separate the various volcanic sediment types and sedimentary rocks (according to grain size) into volcanic gravel (>2 mm), volcanic sand (2-0.0625 mm), volcanic silt (0.0625-0.0039 mm), and volcanic clay (<0.0039 mm). For coarse-grained and poorly sorted consolidated volcaniclastic sediment such as those produced by gravity currents, we applied the terms volcanic breccia (angular clasts) or volcanic conglomerate (rounded clasts) and used lithologic or structural modifiers for further description.
We classified volcanic sediment by designating a principal name and major and minor modifiers. The principal name defines its grain-size class (e.g., gravel, sand, silt, or clay). Relative proportions of the vitric (glass), crystal (mineral), and lithic (rock fragment) components of the sediment are used to determine additional modifiers in the name and are placed before the principal name. For example, volcanic sand composed of 75% glass, 5% feldspar crystals, and 20% lithic fragments is named vitric-lithic volcanic sand. Volcanological features were recorded on both volcaniclastic sediment and igneous rock VCD forms (see the "Core Descriptions" contents list). We adopted the classification scheme of Fisher and Schmincke (1984) where the evidence for a pyroclastic origin was unequivocal. In these instances we used the grain-size terms volcanic breccia (>64 mm), lapilli (lapillistone [2-64 mm]), and ash (tuff [<2 mm]). Sedimentary structures included graded bedding, cross-bedding, planar laminations, foreset bedding, dune forms, and ripples.
To describe important mineralogic and structural features in both the archive and working halves, we examined core sections containing igneous rocks prior to cutting with a diamond-impregnated saw. Each piece was numbered sequentially from the top of the core section and labeled on the outside surface. Pieces that could be fitted together were assigned the same number and were lettered consecutively (e.g., 1A, 1B, 1C, etc.). Plastic spacers were placed between pieces with different numbers. The presence of a spacer may represent a substantial interval of no recovery. If it was evident that an individual piece had not rotated about a horizontal axis during drilling, an arrow was added pointing to the top of the section.
Nondestructive physical properties measurements, such as color, imaging, and natural gamma ray emission were made on the core before it was split (see "Physical Properties"). After the core was split, lithologic descriptions were made of the archive half and the working half was sampled for shipboard physical properties measurements (see "Physical Properties"), magnetic studies (see "Paleomagnetism and Rock Magnetism"), thin sections, and XRD and ICP-AES instrumental analyses. The archive half was described on the VCD form and was photographed. To minimize contamination of the core with platinum-group elements and gold, scientists and technicians removed jewelry from hands and wrists prior to handling.
We used VCD forms to document each section of the igneous rock cores. The left column on the form, adjacent to the core photograph, represents the archive half. A horizontal line across the entire width of the column denotes a plastic spacer. Oriented pieces are indicated on the form by an upward-pointing arrow to the right of the piece. The key to the symbols, colors, and other notations used on the VCDs can be seen in Figure F18.
Locations of samples selected for shipboard studies are indicated in the column headed "Shipboard Studies," with the following notations:
Structural features are noted in the "Volcanic Structures" column and include vesicularity (based on vesicle content):
The nature of the groundmass is broadly represented in the "Groundmass Grain Size Crystallinity" column on the VCD to indicate the presence of different lava lobes in a given core section. The following notation is used for crystallinity:
The following notation is used for grain size:
The column "Phenocryst %" is used to represent a visual estimation of abundance and variation throughout the core section using the following notations (based on phenocryst content):
The "Alteration" column is used to denote the presence of veins by "V" and gives an estimate of the degree of alteration as follows (based on the percent of rock formed by alteration products):
Further details can be found in "Alteration and Weathering".
We subdivided the core into consecutively numbered lithologic units (denoted in the "Lithologic Unit" column on the VCD) on the basis of changes in color, structure, brecciation, grain size, vesicle abundance, mineral occurrence and abundance, and the presence of sedimentary interbeds. Intercalated sediment horizons and volcanic breccias were often designated as "a" and the underlying volcanic rock as "b" within the same unit, unless definitive evidence was available that allowed us to break these out as separate units.
Written descriptions accompany the schematic representation of the core sections and include the following:
We examined thin sections from the core intervals noted on the VCD forms to complement and refine the hand specimen observations. In general, the same terminology was used for thin section descriptions as for the visual core descriptions. The percentages of individual phenocryst, groundmass, and alteration phases were estimated visually, and textural descriptions are reported in "Thin Sections" for each site (see the "Core Descriptions" contents list). The textural terms used are defined by MacKenzie et al. (1982). For some porphyritic basalt, the thin section and visual core descriptions may differ slightly, typically because small plagioclase laths in a rock with seriate texture are visible only in thin section. Thus, a rock described visually as olivine-plagioclase-phyric may be plagioclase-olivine-phyric in the thin section description. Similarly, vitric and lithic components may be estimated differently in volcaniclastic hand samples; therefore, thin section examination may result in modification of a rock name. We generally described at least one thin section per defined unit.
We selected representative samples of major lithologic units for shipboard ICP-AES analysis. Large whole-rock pieces were first cut with a diamond-impregnated saw blade and ground on a diamond wheel to remove surface contamination. Samples were washed in an ultrasonic bath containing methanol for ~10 min, followed by three consecutive ~10-min washes in an ultrasonic bath containing nanopure deionized water, and then dried for ~12 hr in an oven at 110°C. The cleaned whole-rock samples (~20 g/cm3) were reduced to fragments <1 cm in diameter by crushing between two disks of Delrin plastic in a hydraulic press followed by grinding for ~5 min in a Spex 8510 shatterbox with a tungsten carbide barrel. The sample powders were weighed on a Scientech balance and ignited to determine weight loss on ignition (LOI).
We weighed 0.100 ± 0.002-g aliquots of the ignited whole-rock powders and mixed them with 0.4000 ± 0.0004 g of Li metaborate (LiBO2) flux that had been preweighed on shore. Standard rock powders and full procedural blanks were included with the unknowns for each sample run. All samples and standards were weighed on the Cahn Electro balance. Weighing errors are conservatively estimated to be ±0.00001 g.
Mixtures of flux and rock powders were fused in Pt-Au crucibles at 1050°C for 10-12 min in a bead sampler NT-2100. A 10-µL aliquot of 0.172-mM aqueous lithium bromide (LiBr) solution was added to the mixture before fusion as an antiwetting agent to prevent the cooled bead from sticking to the crucible. Cooled beads were transferred to 125-mL polypropylene bottles and dissolved in 50 mL of 2.3-M HNO3 by shaking with a Burrell wrist action bottle shaker for 1 hr. After digestion of the glass bead, all of the solution was filtered to 0.45 µm into a clean 60-mL widemouthed polypropylene bottle. Next, 2.5 mL of this solution was transferred to a plastic vial and diluted with 17.5 mL of 2.3-M HNO3 to bring the total volume to 20 mL. The solution-to-sample dilution factor for this procedure is ~4000. Dilutions were conducted using a Brinkman Instruments dispensette (0-25 mL).
Major (Si, Ti, Al, Fe, Mn, Mg, Ca, Na, K, and P) and trace (Zr, Y, Sr, Ba, Ni, Cr, Sc, V, Co, Cu, and Zn) element concentrations of powder samples were determined with the JY2000 Ultrace ICP-AES. The JY2000 sequentially measures characteristic emission intensities (with wavelengths between ~100 and 800 nm). ICP-AES protocols for dissolution and analysis of rock powders were developed by Murray et al. (2000; see also Shipboard Scientific Party, 2001). The hard rock analytical procedure was refined during Leg 197. The elements analyzed, emission lines used, and the specific analytical conditions for each sample run during Leg 197 are provided in Table T6.
The JY2000 plasma was ignited 30 min before each run to allow the instrument to warm up and stabilize. After the warm-up period, a zero-order search was performed to check the mechanical zero of the diffraction grating. After the zero-order search, the mechanical step positions of emission lines were tuned by automatically searching with a 0.002-nm window across each emission peak using the University of Massachusetts Kilauea basalt laboratory standard K-1919, prepared in 2.3-M HNO3. The only exception is P, which was automatically searched by using a single-element standard. During the initial setup, an emission profile was collected for each peak, using the K-1919 standard, to determine peak-to-background intensities and to set the locations of background points for each element. The JY2000 software uses these background locations to calculate the net intensity for each emission line. The photomultiplier voltage was optimized by automatically adjusting the gain for each element using the standard (BHVO-2, JB-1a, BIR-1, or K1919) with the highest concentration for that element. Before each run, a profile of the K-1919 standard was collected to assess the performance of the instrument from day to day. A typical sample run lasted ~12-14 hr, depending on the number of samples and replicate analyses.
All ICP-AES data presented in the site chapter reports were acquired using Mode 2 of the JY2000 software, except for Fe, Mg, Mn, Ba, Cr, Sc, V, and Y data, which were acquired in Mode 5. In Mode 5, the intensity at the peak of an emission line is measured and averaged over a given counting interval, repeated three times sequentially. Mode 2 fits a Gaussian curve to a variable number of points across a peak and then integrates to determine the area under the curve. The parameters for each run are given in Table T6. Each unknown sample was run at least twice, nonsequentially, in all sample runs.
A typical ICP-AES run includes (1) a set of three certified rock standards (BHVO-2, BIR-1, and JB-1a) (Table T7) run at the beginning, middle, and end of the sample run; (2) up to 11 unknown samples; (3) a drift-correcting sample (the K-1919 standard) analyzed every fourth sample position; and (4) a blank solution run near the beginning, middle, and end of each run. A 2.3-M HNO3 wash solution was run for a minimum of 90 s between each of the samples and standards.
Following each sample run, the raw intensities were transferred to a data file and data reduction was completed using a spreadsheet to ensure proper control over standardization and drift correction. Once transferred, intensities for all samples were corrected for the full procedural blank. A drift correction was then applied to each element by linear interpolation between drift-monitoring solutions run before and after a particular batch of samples. The interpolation was calculated using the lever rule. Following blank subtraction and drift correction, concentrations for each sample were calculated from the average intensity per unit concentration for the U.S. Geological Survey (USGS) standard BHVO-2, which was analyzed twice during the run.
Estimates of accuracy and precision for major and trace element analyses were based on replicate analyses of BHVO-2, BIR-1, and JB-1a standards, the results of which are presented in Table T7. In general, run-to-run relative precision by ICP-AES was better than 2% for the major elements. Run-to-run relative precision for trace elements was generally <5%. Exceptions typically occurred when the element in question was near the detection limit of the instrument (see Table T6 for instrument detection limits).