Bull Volcanol (1999) 61 : 64–82

Q Springer-Verlag 1999

ORIGINAL PAPER

G. A. Smith 7 M. J. Grubensky J. W. Geissman

Nature and origin of cone-forming volcanic breccias in the Te Herenga Formation, Ruapehu, New Zealand

Received: 1 October 1998 / Accepted: 28 December 1998

Abstract Volcanic breccias form large parts of composite volcanoes and are commonly viewed as containing pyroclastic fragments emplaced by pyroclastic processes or redistributed as laharic deposits. Field study of cone-forming breccias of the andesitic middle Pleistocene Te Herenga Formation on Ruapehu volcano, New Zealand, was complemented by paleomagnetic laboratory investigation permitting estimation of emplacement temperatures of constituent breccia clasts. The observations and data collected suggest that most breccias are autoclastic deposits. Five breccia types and subordinate, coherent lava-flow cores constitute nine, unconformity-bounded constructional units. Two types of breccia are gradational with lava-flow cores. Red breccias gradational with irregularly shaped lava-flow cores were emplaced at temperatures in excess of 580 7C and are interpreted as aa flow breccias. Clasts in gray breccia gradational with tabular lava-flow cores, and in some places forming down-slope-dipping avalanche bedding beneath flows, were emplaced at varying temperatures between 200 and 550 7C and are interpreted as forming part of block lava flows. Three textural types of breccia are found in less intimate association with lava-flow cores. Matrix-poor, well-sorted breccia can be traced upslope to lava-flow cores encased in autoclastic breccia. Unsorted boulder breccia comprises constructional units lacking significant exposed lavaflow cores. Clasts in both of these breccia types have paleomagnetic properties generally similar to those of the gray breccias gradational with lava-flow cores; they indicate reorientation after acquisition of some, or all, magnetization and ultimate emplacement over a range of temperatures between 100 and 550 7C. These brec-

Editorial responsibility: D. A. Swanson Gary A. Smith (Y) 7 Michael J. Grubensky 7 John W. Geissman Department of Earth and Planetary Sciences, University of New Mexico, Albuquerque, NM 87131-1116, USA Fax: c505 277 8843 e-mail: gsmith6unm.edu

cias are interpreted as autoclastic breccias associated with block lava flows. Matrix-poor, well-sorted breccia formed by disintegration of lava flows on steep slopes and unsorted boulder breccia is interpreted to represent channel-floor and levee breccias for block lava flows that continued down slope. Less common, matrixrich, stratified tuff breccias consisting of angular blocks, minor scoria, and a conspicuously well-sorted ash matrix were generally emplaced at ambient temperature, although some deposits contain clasts possibly emplaced at temperatures as high as 525 7C. These breccias are interpreted as debris-flow and sheetwash deposits with a dominant pyroclastic matrix and containing clasts likely of mixed autoclastic and pyroclastic origin. Pyroclastic deposits have limited preservation potential on the steep, proximal slopes of composite volcanoes. Likewise, these steep slopes are more likely sites of erosion and transport by channeled or unconfined runoff rather than depositional sites for reworked volcaniclastic debris. Autoclastic breccias need not be intimately associated with coherent lava flows in single outcrops, and fine matrix can be of autoclastic rather than pyroclastic origin. In these cases, and likely many other cases, the alternation of coherent lava flows and fragmental deposits defining composite volcanoes is better described as interlayered lava-flow cores and cogenetic autoclastic breccias, rather than as interlayered lava flows and pyroclastic beds. Reworked deposits are probably insignificant components of most proximal cone-forming sequences. Key words Volcanic breccia 7 Autoclastic breccia 7 Composite volcano 7 Emplacement temperature 7 Ruapehu

Introduction Early conceptualizations of composite volcanoes featured layered deposits dipping radially away from a central summit vent, with many layers extending to the

65

base of the cone (Macdonald 1972; Williams and McBirney 1979). These views were heavily influenced by observations of shallowly dissected flanks of active cones and of crater walls around young summit vents (Cotton 1944; Perret 1950; Rittmann 1962). Field sketches of crater-wall deposits were divided simply into coherent bands of lava and fragmental deposits, or breccias (Fisher 1958). Cinder-rich breccias were appropriately interpreted as pyroclastic, i.e., composed of fragments produced by explosions in the vent (Fisher 1961, 1966), and composite volcanoes (syn-stratovolcanoes) have generally been viewed as made of interlayered lava flows and pyroclastic deposits. In the absence of deep flank exposures, crater walls were also utilized to describe the general composition of cone flanks. More recent studies at Arenal, Costa Rica (Borgia et al. 1983, 1988), and at Ruapehu, New Zealand (Hackett and Houghton 1989), suggest that flank deposits can be composed of both pyroclastic and autoclastic breccias, the latter being composed of fragments produced by disruption of flowing lava (Fisher 1961). If fragments in the majority of cone-forming breccias are not formed by in-vent explosions, then there is considerable need for revising the details of deposit type and geometry in models of composite volcanoes. We present the results of field and paleomagnetic study conducted to evaluate breccia clast and deposit origins on Ruapehu. Field interpretations of brecciaclast origins are challenging because andesite pyroclasts, commonly resulting from vulcanian eruptions, are characteristically nonvesicular and, therefore, difficult to distinguish from autoclasts. In contrast, basaltic and rhyolitic pyroclasts are typically highly vesicular and generally distinguished from dense to moderately vesicular autoclasts (Heiken and Wohletz 1984, 1991; Fisher and Schmincke 1984). Progressive thermal demagnetization of the natural remanent magnetization (NRM) can be utilized to evaluate breccia-clast thermal histories and to determine how, or if, a volcanic fragment moved relative to a local and fixed magnetic field direction, during or after cooling and attending magnetization blocking (e.g., Hoblitt and Kellogg 1979; Downey and Tarling 1991; Tamura et al. 1991; Clement et al. 1993). These thermal demagnetization characteristics can yield insights into breccia origin (Grubensky et al. 1998). The Pleistocene Te Herenga Formation is the oldest of four cone-forming sequences that compose Ruapehu (Hackett 1985; Hackett and Houghton 1989). The studied deposits characterize the central- and flank-vent lithofacies association and the proximal cone-building lithofacies association in the Hackett and Houghton (1989) model for composite cones. Despite excellent exposure on Ruapehu, breccia-deposit origins are unclear, because they lack sedimentary structures diagnostic of specific transport processes. We describe new observations of lateral spatial relations between breccias and nonbrecciated lava (lava-flow cores), and of deposit contacts that are more detailed than those pre-

viously published. The new field observations are sufficiently consistent with the paleomagnetic data reported herein to warrant interpretation of breccias of the Te Herenga Formation as largely autoclastic deposits.

Study site Ruapehu is an andesitic composite volcano in the Taupo Volcanic Zone magmatic arc of North Island, New Zealand. Ruapehu was constructed over the past 250c k.y. in four cone-building episodes (Hackett 1985), the oldest of which produced the Te Herenga Formation (Fig. 1). Sampling and observations were conducted along the west side of Pinnacle Ridge (Fig. 1B). The ridge is approximately 350 m high and forms a roughly northward-dipping homocline. Primary dips of 20–307 are typical of most deposits, but local deformation of beds by plugs and north/south-striking, vertical, basaltic-andesite dikes has resulted in dips of as much as 757 (Hackett 1985; Brooker 1992). Field observations and sampling focused on the southern exposures of the Te Herenga Formation along Pinnacle Ridge between The Great Pinnacle and the head of Skipper’s Canyon (Fig. 1B). This transect includes proximal cone-building lithofacies deposits but avoids hydrothermally altered material of the central- and flank-vent lithofacies exposed beneath and south of the Great Pinnacle (Hackett and Houghton 1989).

Methods Field observations Detailed field observations of relationships within and between nonbrecciated lava layers (lava-flow cores) and diverse breccia types comprise a primary source for new ideas concerning origins for many deposit types. Observations concerning deposit geometry and the nature of contacts between constructional assemblages (Grubensky et al. 1998) across the entire study area were made by examining outcrops and also by mapping panoramic photographs of Pinnacle Ridge taken from an opposing ridge. The photograph-based maps (Fig. 2) facilitate comparisons to similar maps of steep exposures through other composite volcanoes (Williams 1942; Rittmann 1962; Aramaki 1966; Hackett 1985; Grubensky et al. 1998). Lava-flow cores and breccia clasts are similar in whole-rock chemistry, mineralogy, and texture (Hackett 1985). Thus, our study focused on approximate clast-size ranges, approximate median clast sizes, and estimates of sorting and of abundance of matrix for description and differentiation of breccia types. Unconformity-bound breccia layers form constructional units numbered I–IX, from oldest to youngest (Fig. 2). Breccia-deposit origins were interpreted using field characteristics and later compared, for consistency,

A

30

B

ad Ro

Crater Lake

Whaka papanu i

Bruce

Study area

head of Skipper's Canyon

2797 m x

N

Str.

66

36

ge Rid

km

le

Extent of diagram, Fig. 2

n ac

N 5

0

Pin

34 27 46

Whakapapa Fm. North Island Taupo Volcanic Zone x Ruapehu

< 15 ka

Mangawhero Fm. 15-60 ka

Wahianoa Fm. 120-250 ka

Te Herenga Fm. > 250 ka

Fig. 1 A Generalized map of Ruapehu showing four cone-building formations. B Detailed map of the Pinnacle Ridge area showing geographic features referred to in text and location of study area within the Te Herenga Formation. (Modified from Hackett and Houghton 1989)

with paleomagnetic data collected from clasts of the same deposit. In most cases paleomagnetic data from specific deposits within constructional units are assumed to be representative of all deposits within the same unit because of the consistency of field characteristics for breccias within each unit relative to the range of breccia characteristics in the Te Herenga Formation. Paleomagnetic sampling, data acquisition, and data analysis Paleomagnetic sampling and data analysis were performed using methods of Grubensky et al. (1998) and described further in Grubensky (1996). Sampling at nine well-exposed localities involving five breccia types (described herein) and 15 sites of lava-flow cores was conducted with either a standard 2.4-cm gasoline-powered rock drill or a portable 1.2-cm electric drill. Cores (independent samples) were oriented prior to removal using a magnetic compass, sun compass, or both, using procedures described by Butler (1992). Between one and three samples were obtained from as many as 45 clasts in each breccia. Five to 11 samples were removed from adjacent lava-flow cores where accessible. Sampling of lava-flow cores provides the best possible de-

Whakapapa Fm. Te Herenga Fm. The Great Dikes and plugs 0 Hydrothermal alteration

Pinnacle 500 meters

termination of the ambient field at the time of emplacement of each breccia sampled (Grubensky et al. 1998). Sample sites were selected in areas distant from intrusions both to avoid structural complexity and possible thermal remagnetization. Although some degree of structural tilt may affect sample localities, the magnitude of such deformation cannot be objectively distinguished from moderate initial dips on the upper cone; thus no effort was made to correct for such tilt. Natural remanent magnetizations (NRM) were measured using a 2-G Enterprises cryogenic magnetometer or a Schonstedt SSM-1 A spinner magnetometer. One specimen from all clasts was thermally demagnetized in air at 50 or 25 7C steps to peak temperatures of 580 7C in a Schonstedt TSD-1 furnace or a custom-built largevolume thermal demagnetizer (Grubensky 1996). Bulk susceptibility was measured between steps to detect changes in magnetic mineralogy due to heating in air using a Sapphire Instruments SI-2 susceptibility bridge. Thermal demagnetization (TD) data were plotted on modified orthogonal demagnetization diagrams and inspected for the presence of one or more interpretable magnetization components, defined by a progressive change in magnetization intensity over several demagnetization steps concurrent with a uniform change in magnetization direction. The direction of each magnetization and the approximate temperature interval required to unblock it were determined using leastsquares method (Kirschvink 1980). Unanchored lines were fit to three, and typically several more, demagnetization steps; those lines with maximum angular deviations of more than 157 represented magnetizations that

67 Fig. 2 Interpretive sketch, drawn from panoramic photographs, of the west face of Pinnacle Ridge. Unconformity-bounded constructional units indicated by roman numerals, and paleomagnetic sampling sites by arabic numerals. Nature of breccias found in each constructional unit is summarized in Table 1

could not be isolated and were rejected. Inspection of difference vectors and remanence vectors on stereographic projections was employed to evaluate the extent of overlap of unblocking temperatures of discrete magnetizations within individual specimens. Directions of magnetization(s) isolated over relatively constant ranges of unblocking temperatures from clasts of the same deposit were tested for nonrandom distribution using spherical statistics (Watson 1956). We refer to single-component (SC) magnetizations as those that constitute 1 95% of the NRM and unblock progressively to the maximum laboratory unblocking temperature for magnetite. Such magnetizations, of thermal origin and carried in magnetite, which are well grouped around an estimated mean direction, are indicative of collective emplacement of heated clasts at temperatures above the temperature range required to unblock the magnetization in the laboratory (Néel 1955). For deposits characterized by clasts with two or more well-defined partial thermoremanences (pTRMs; Théllier 1941), a successful application of the emplacement temperature estimate requires that at least one pTRM (usually that with the lowest unblocking temperature) common to all specimens of, or a subset of, the sample population be well grouped and consistent with the ambient field at the time of emplacement. Random distributions of TRMs indicate emplacement at temperatures below the lower end of the TRM unblocking temperature spectrum (which typically varies from ~100 to 300 7C) and may indicate an ambient temperature of emplacement (Aramaki and Akimoto 1957; Hoblitt and Kellogg 1979). Additional laboratory experiments on replicate specimens of representative clasts were conducted to determine the composition and grain size of the carriers of magnetization. The NRM of selected specimens was demagnetized in peak alternating fields (AFD) of 130 milliTesla (mT) using an automated degaussing system configured to a 2-G Enterprises superconducting rock magnetometer. Modified orthogonal demagnetization diagrams were used to compare results with thermaldemagnetization data.

Field observations of breccias Distribution of constructional units The west face of Pinnacle Ridge is divided into nine constructional units each largely composed of one breccia type (described herein) with, or without, lava-flow cores (Fig. 2). Contacts between constructional units are high- to low-angle unconformities that strike

68

north–south nearly parallel to the ridge and dip east or west. In at least two locations the dip of an unconformity is nearly vertical with more than 20 m of exposed relief on the surface. Abrupt terminations of units I and V (Fig. 2) are the result of these steep erosional contacts with units II and VIII, respectively. Reconnaissance observations on the east side of Pinnacle Ridge indicate that most constructional units are not represented there. Only the ridge-capping unit IX and unit I are exposed on both sides of the ridge. Breccia types The Pinnacle Ridge section is dominated by thickly bedded ( 1 2 m) to massive breccia, although a few constructional units have well-developed stratification. Clasts exhibit a moderate range of vesicularity (F0–20 vol.%) and a broad range of oxidation coloration. Lava-flow cores are at least minor constituents of most constructional units. Rather than describe all nine constructional units and their internal variations separately, five specific breccia types, each of which is at least locally present within more than one unit, are described. Table 1 summarizes the distribution of breccia types within constructional units. Because lava-flow cores at Pinnacle Ridge are an important component of many of the units, the two contact styles between lavaflow cores and breccia, gradational and nongradational, serve as a primary classification for the breccias.

abruptly over distances of a few meters. Separate lenses of lava on the scale of a meter long and isolated in red breccia are also common. The orientations of continuous lava cores are variable over distances less than 20 m and range between gently dipping and conformable with the crude sense of layering in the adjacent breccia, to steeply northward or eastward dipping. Discontinuous cores tend to have steeper dips than those of the layering in surrounding deposits. Breccia clasts range from poorly to moderately vesicular and are texturally identical to associated flow cores (Fig. 3). This breccia type is typically massive and clast supported, and many exposures have weakly developed normal grading of clasts. Clasts close to cores are angular and spinose, with ragged edges and shallow tension gashes oriented perpendicular to clast faces. Clasts distant from cores are less angular and more intensely oxidized with increasing distance from the core. Dense clasts average 1–20 cm across but are locally up to 1–2 m across. Contacts between lava cores and red breccia are irregular and characterized by angular apophyses of dense lava extending 5–50 cm into the breccia. The coherent lava along the contact is commonly marked by tension gashes a few centimeters long oriented perpendicular to the contact. Gray breccia with tabular lava-flow cores. Tabular lava-flow cores with 2- to 4-m-thick breccia interbeds

Breccias gradational with lava-flow cores Red breccia with irregularly shaped cores. Red breccia on Pinnacle Ridge is everywhere intercalated with lavaflow cores, which are 1–2 m thick and change thickness

Table 1 Breccia types within constructional units Constructional unit

Breccia types a

I

Unsorted boulder breccia; matrix-rich, wellsorted tuff breccia Matrix-rich, stratified tuff breccia Red breccia with irregularly shaped lava-flow cores (lower half); gray breccia with tabular lava-flow cores (upper half); matrix-poor, well-sorted breccia; matrix-rich, stratified tuff breccia Gray breccia with tabular lava-flow cores; matrix-poor, well-sorted breccia Red breccia with irregular lava-flow cores; gray breccia with tabular lava-flow cores Matrix-rich, stratified tuff breccia; matrixpoor, well-sorted breccia Unsorted boulder breccia Matrix-poor, well-sorted breccia Unsorted boulder breccia

II III

IV V VI VII VIII IX a

In order of abundance

Fig. 3 Red breccia associated with irregular lava-flow cores. Lava-flow core is approximately 1.5 m thick. Note irregular base of the flow core and spinose protuberances from the core base representing early stages of flow fragmentation. Also note the abundance of matrix in the flow breccia

69

are common in the upper half of constructional unit III, are present throughout unit IV, and are present locally in unit V. Tabular lava-flow cores are nonvesicular and have little or no associated oxidized breccia. Where present, reddened breccia is at most 10 cm thick and laterally discontinuous. Both lava-flow cores and breccia clasts are dominantly gray in color. Downslope terminations of these flow cores are typically blunt, whereas upslope core terminations are gradual pinchouts between gray breccias or other tabular cores. Near the downslope termination of many tabular cores is an upward transition from flow-banded nonbrecciated core, folded and flow-banded nonbrecciated lava core, breccia with flow-banded clasts, and finally to fines-rich breccia composed of poorly sorted granulated clasts (Fig. 4). Coarser fragments in all parts of these gradational zones are nonvesicular, flow banded, and angular to subrounded. Matrix (fragments smaller than F1 cm) composes 30–50% of the breccia and includes crystals of pyroxene, plagioclase, and sand-sized, angular lithic fragments identical in mineralogy and texture to the coarser clasts. Only small changes in resistance to weathering accompany these gradations. Generally, gradational core-breccia contacts locally exhibit sharp juxtaposition of flow-foliated lava-flow core and poorly sorted, matrix-rich gray breccia. Gray breccias beneath tabular cores are layered and dip 5–107 more steeply than the overlying core and its brecciated top. One or more of the layers may include

Fig. 4 Field sketch of block lava and adjacent lava flows, including a vertical gradation from flow-foliated lava-flow core top to a mixture of sand- and pebble-sized autoclasts

discontinuous 1- to 2-m-thick lava-flow cores similar to the main, overlying core. Block-rich, matrix-poor beds alternate with those rich in sandy matrix and crude inverse grading is common. Clasts and matrix compositions are identical to those that form the gray breccia gradational with lava-flow core tops. Breccias lacking gradational contacts with lava-flow cores Unsorted boulder breccia. Boulder breccias, dominated by clasts 0.2–4 m across, are widespread and compose much of constructional units I, VII, and IX, but exposures in unit VII are the most extreme and have the largest clasts (Fig. 5). Clasts are subangular with smooth faces. Red oxidation coloration in clasts is common, but gray clasts, typically flow banded, are dominant. Prismatic fractures, indicative of in situ cooling and contraction, are present in some clasts (Fig. 5). The sandy matrix is gray or yellow and consists of holocrystalline lithic fragments and phenocrysts of plagioclase and pyroxene. Vesicular fragments are absent in the matrix, but some of the centimeter-sized clasts are highly vesicular and strongly colored by oxidation. These deposits are part of Hackett’s (1985) heterolithologic tuff breccia. Boulder breccia is clast supported throughout, with interstitial sandy matrix composing 10–30% of the deposit. Layers 1–7 m thick are distinguished by changes in clast sizes, proportion of matrix to clasts, and presence of weakly developed bedding planes. Unit VII is particularly noteworthy for concentrations of gray, poorly vesicular boulders. The largest

Overlying lava-flow core

Flow-banded clasts identical to lava-flow core

Overlying lava-flow basal breccia

5% matrix

Contact

15-20% matrix 60-70% matrix Flow-banded clasts identical to lava-flow core Lava-flow core with local flow banding Contact Underlying lava-flow carapace breccia

1m

3m

70

concentration measures 20!20 m and has blocks that average 2-m across. Clasts are mineralogically and texturally similar to lava-flow cores in adjacent constructional units. Many of the clast concentrations are irregular in shape and one has its longest dimension oriented almost vertically. Other clast concentrations are single-boulder-high trains that impart a sense of stratification to this otherwise massive breccia facies.

Fig. 5 Unsorted boulder breccia. Arrow points to prismatically fractured block approximately 80 cm long. Note contrast between zones of shattered blocks with little matrix (top) and areas that are matrix rich (center and right edge)

Matrix-poor, well-sorted breccia. Sorted breccias are minor constituents of units III and IV and are the only fragmental deposits in unit VIII (Fig. 6). This breccia type is noted for its exceptional sorting of clasts and the lack of clast diversity, particularly in terms of oxidation coloration. All clasts are poorly to moderately vesicular and dark gray. Average clast size is 25–30 cm, and the maximum is 1 m across. Subrounded as well as subangular and angular clasts are present. Each layer is clastsupported, massive, and has less than approximately 5% matrix. In contrast to interstices of other breccia types at Pinnacle Ridge, sorted breccias have locally unfilled interstices. Where found, matrix is most common near clast–clast contacts, suggesting local generation by contact abrasion. The matrix material is gradaFig. 6A, B Matrix-poor, well-sorted breccia. A Thin lava-flow cores encapsulated in breccia within the southern part of constructional unit VIII. B Close view of matrix (lens cap is 5 cm in diameter) showing paucity of matrix and relatively limited range of clast sizes

71

tional with clasts at these contacts, and small fractures near clast margins are gradational between coherent clasts and fragmental matrix. The matrix consists of mottled gray and yellow sand with holocrystalline textures similar to adjacent clasts. Many coarse fragments in unit VIII include a set of near-vertical, parallel fractures. These deposits are part of Hackett’s (1985) monolithologic tuff breccia, although most areas of most deposits lack adequate ash-size matrix to be called tuff breccias. Constructional unit VIII is present between units VII and IX (Fig. 2), which are dominated by unsorted boulder breccia. This interval is noteworthy for thin, discontinuous lava-flow cores along contacts between breccia layers (Fig. 6). The deposit is wedge-shaped overall, thinning from south to north and pinching out between units VII and VIII. Each breccia layer pinches out northward from a maximum thickness of 10–15 m. The lava cores are approximately 1 m thick and also wedge out northward (downslope). The breccia clasts are lithologically identical to the associated lava-flow cores in terms of mineralogy and color, although the range of vesicularity of the clasts is, on average, slightly greater than that of the lava-flow cores. Contacts between lava-flow cores and adjacent clasts are sharp. Another exemplary exposure of sorted breccia in unit III consists of one of the steeply inclined, coarser layers beneath a tabular lava-flow core. Jigsaw-fractured clasts are present locally in this deposit, and matrix sand separates clasts. The orientation of fractures is highly variable within individual clasts. Similar layers are less common in unit IV, but there they separate thin tabular lava-flow cores with which they are concordant. Each sorted breccia in both units III and IV is 5–10 m thick and is less extensive downdip than the associated flow core(s). Matrix-rich, stratified tuff breccia. Constructional units II and VI are composed entirely of strongly bimodally textured, matrix-rich breccias that contrast starkly with other deposits at Pinnacle Ridge (Fig. 7). Deposits contain 0–30% of moderately vesicular to dense lithic clasts and less common scoria clasts. Clasts average 5–20 cm across, with a maximum diameter of approximately 1 m. Clast shapes range from subrounded to subangular and are mostly gray, plagioclase-pyroxenephyric andesite. Red or brown clasts colored by oxidation are abundant in some beds but are everywhere subordinate to dense, gray clasts. Hydrothermally altered yellow-brown clasts are present locally. Fracture patterns consistent with either cooling in situ or clast–clast impact during deposition are absent. The matrix is remarkably well sorted, 0.5–2.0 mm in grain size, and along with coarser clasts imparts a striking bimodal grain size to the deposits. The matrix is light brown, hypohyaline and holocrystalline lithic grains with plagioclase and pyroxene crystals. Most beds are massive, although some feature reverse or normal grading of clasts and a few contain

Fig. 7 Matrix-rich, well-stratified tuff breccia in constructional unit II. Pack, 70 cm high, rests on contact between two thick beds interpreted to represent debris-flow deposits composed of autoclastic (and pyroclastic?) blocks in a remobilized tephra matrix

small-scale stratification. Individual beds vary from locally clast supported to matrix supported. Imbrication of clasts is, at best, poorly developed locally, and the matrix is consistently massive. Thickness of clast-bearing beds range from approximately 10 cm to 2 m. Each layer is continuous downslope over distances of less than 15 m. Clast-poor, sandy beds are conspicuous but are less than 50 cm thick, pinch over short distances of 5 m or less, and typically have shallower dips than adjacent, clast-rich layers. These sandy beds invariably contain diffuse planar laminations and scour-and-fill bedding. Downslope terminations of the coarse deposits tend to be steep and blunt compared with sharp pinchouts of sandy layers. Where modern gullies are cut parallel to deposit strike, exposures show that many coarse beds are also lens shaped transverse to flow direction. Hackett (1985) maps these units as heterolithologic tuff breccia.

Paleomagnetic results and interpretations Both lava-flow cores and breccia clasts are similar in terms of NRM intensity. Specimen-moment NRMs from breccia clasts and lava-flow cores range from 1!10 –1 to 5!10 2 Am 2 and average 3!10 1 Am 2. The range in NRM intensities for lava-flow cores is narrow and lies within a single to half an order of magnitude of the average. For most clasts in all breccia deposits, thermal demagnetization to 580 7C typically unblocks more

72

N, Up

N, Up

40

A

B

30

Site

N

N/No

R

D7

I7

k

a95

6 7 8 11 12 13 15 16 18 19 22 23

9 6 11 7 9 8 7 7 7 8 8 5

9/9 6/6 11/11 7/7 9/9 8/8 7/7 6/7 6/7 8/8 7/8 4/5

8.8848 5.9587 10.6023 6.9411 8.9802 7.9871 6.9536 5.8817 5.8828 7.9854 6.7943 3.9921

37.6 47.4 30.8 15.4 19.0 36.1 25.7 46.3 14.2 19.5 24.6 35.2

–56.5 –53.1 –57.3 3.5 –54.6 –61.3 –54.1 –63.8 –58.2 –56.7 –45.6 –79.7

78.1 145.3 27.7 118.8 454.6 618.2 150.9 50.7 51.2 546.1 34.0 509.0

6.2 6.1 9.3 6.0 2.6 2.4 5.3 10.4 10.4 2.5 11.4 4.7

30

35 mT

A/m

400

55 mT

NRM

20

75 mT

o

500

o

580

55 mT

10120

mT

o

500

120 mT

580o

0

20 mT

95 mT 105 mT

200o o 300 o 400

10

20 mT 30 mT 45 mT

o

20

13 mT

18 mT

300

10

20

E, Horiz.

A/m

E, Horiz. 0

A/m

10

N C

TAFD

Table 2 Paleomagnetic site-mean data from lava-flow cores. N number of independent samples collected; N/No ratio of number of samples used in calculating site mean directions to total measured from N samples; R vector sum of N unit vectors; D7, I7 in situ site mean declination and inclination direction; k best estimate of concentration parameter; a95 radius of 95% confidence cone

NRM 8 mT

o

Lava-flow cores Principal-components analysis of thermal and alternating-field data for specimens from lava-flow cores yielded single-component magnetizations that are similar in direction and intensity at the site level (Fig. 8A, B; Table 2). Magnetizations are well grouped and generally yield estimated means that are northward and moderately negative in inclination (Fig. 8C; Table 2). The high fields required to randomize these magnetizations indicate remanence carriers dominated by pseudo-single- and single-domain magnetite. Comparison of estimated mean magnetization directions in adjacent lava-flow cores and breccias facilitates interpretation of the thermal history of the breccia fragments.

40

NRM 100o 200o

A/m

than 95% of the NRM, implying magnetite as the principal carrier of the remanence. Approximately 10–15% of the specimens retain 15–60% of the NRM, suggesting a contribution of hematite to the remanence. In flow cores, at least 99% of the NRM is removed by 580 7C, implying magnetite as the primary carrier of the remanence. Alternating-field demagnetization (AFD) to peak fields of 130 mT randomizes, on average, all but 5–15% of the NRM from lava-flow cores and clasts in all five types of breccia. For lava-flow cores, results of both alternating field and thermal demagnetization are fully consistent. For breccia clasts at the same sites, the limited amount of AFD carried out yields behavior similar to that of thermal demagnetization. In some cases, however, the behavior is not fully consistent between the two techniques. Most notably, AFD may fail to resolve the two components of remanence defined by thermal methods. We interpret this behavior as showing strong overlap in coercivity spectra of magnetizations acquired in thermal blocking.

PDFD

x

*

n=9

E

Fig. 8A–C Examples of demagnetization characteristics of lavaflow core at site 13. Modified orthogonal demagnetization diagrams of A thermal demagnetization data and B alternating-field demagnetization data. Closed symbols and open symbols are projections of the magnetization onto the horizontal and vertical planes, respectively. The star is the natural remanent magnetization (NRM), measured at room temperature prior to demagnetization. C Equal-area projection of sample magnetization directions from site 13 isolated in thermal demagnetization. Open symbols plot in the upper hemisphere. Quaternary time-averaged field direction (TAFD) and present-day field direction (PDFD) are also shown and plotted in the upper hemisphere

Breccias Analysis of demagnetization data from breccia clasts relies essentially on thermal demagnetization data, because coercivity spectra cannot be directly converted to actual or laboratory unblocking temperatures. For each site with both single-component (SC) magnetization clasts and dual-component (DC) magnetization clasts, the first-unblocked magnetizations from dual-component magnetization clasts are displayed separately, because of their potential to yield either pTRMs useful for emplacement-temperature estimates, or viscous re-

73

manent magnetizations (VRM) of no consequence to this study. For DC clasts, data for the last removed (i.e., higher range of unblocking temperatures) magnetizations are combined with the single-component magnetizations prior to statistical analysis, because we assume that these magnetizations partially share a common, high range of unblocking temperatures (Table 3). Reliable emplacement temperature estimates require the presence of a first-removed, well-grouped magnetization. In the subsequent discussion the following abbreviations are used for the sake of brevity: FRpfirst-removed magnetization from dual-component magnetization clast; LRplast-removed magnetization from dualcomponent magnetization clast; SCpthe magnetization isolated from a single-component magnetization clast; TDpthermal demagnetization; AFDpalternating-field demagnetization; PFDppresent field direction; and TAFDptime-averaged field direction. Red breccia with irregularly shaped lava-flow cores. Data from two sampling sites (sites 9 and 14; Tables 3 and 4) in this breccia type suggest emplacement at elevated temperature. Randomly oriented low laboratory unblocking temperature FRs at site 9 are interpreted as VRMs acquired prior to sampling, perhaps as a result of lightning-induced isothermal magnetization. Aside from this VRM, the breccia clasts have a single-component magnetization, unblocked by 580 7C that is well grouped in direction and coincides with the magnetization direction of adjacent lava flows (Fig. 9). These data indicate a high temperature of emplacement ( 1 580 7C) and little or no reorientation of the breccia clasts after they acquired their remanence. Gray breccia associated with tabular lava-flow cores. A single paleomagnetic sample site (site 5) from breccia of this type consists of a mixture of SC and DC clasts (Fig. 10). Both the SC and DC-LR magnetizations are randomly oriented (Fig. 10C), indicating that the breccia clasts were reoriented after acquiring their remanence. Laboratory unblocking temperature spectra are highly varied for the clasts with SC magnetizations, with minimum values for the SC clasts from 175 to 5507. These clasts might have been emplaced at higher-thanambient temperatures. First-removed magnetizations of DC clasts are dispersed about an estimated mean direction that does not overlap that of either the TAFD or the PFD (Fig. 10C). These magnetizations are not random, however, but are clustered in two populations, one of northwest declination and negative inclination and the other of northeast declination and positive inclination. Curiously, this distribution overlaps the mean direction for the overlying lava-flow core (site 6). These subgroups may both be of viscous origin. Alternatively, the northeast and positive magnetization group may correspond to clasts that blocked some of their magnetization at elevated temperature in a field comparable to that recorded by the overlying lava-flow core.

N

A

15

X

Site 14 single-component magnetizations

14

*

16

E

n = 19

8

B

X

*

9

Site 9 high-temperature components (SC, DC-LR)

7

E n = 25

Fig. 9 Equal-area projection of sample magnetization directions for red breccia associated with irregularly shaped lava-flow cores (conventions and symbols as in Fig. 8) at A site 14 and B site 9 (not including DC-FR interpreted as VRM). Sample directions are well grouped, and estimated site mean directions (squares) and 95% cones of confidence (ellipses) for breccia clasts are indistinguishable from those of adjacent lava flows (sites 7, 8, 15, and 16)

The interpretation of a thermal history for clasts at site 5 is ambiguous. Clearly, however, clasts were not emplaced at temperatures as high as those experienced by the red breccias. Clasts may have been emplaced over a wide range in temperature (conceivably from ambient to nearly 5507). Clasts did move after acquiring at least some of their magnetization, of relatively low laboratory unblocking temperature. Unsorted boulder breccia. Unsorted boulder-breccia deposits at sites 10 and 21 contain clasts displaying either single- or dual-component magnetizations based on TD data (Tables 3, 4). At both sites the SC and DCLR magnetizations are randomly distributed or have unacceptably large 95% cones of confidence for reliable interpretation of a meaningful distribution. DC-FR magnetizations, however, are more closely grouped and display cones of confidence of associated estimated mean directions that overlap those determined for directly underlying and overlying lava flows (Fig. 11).

Autoclastic

Autoclastic Pyroclastic– autoclastic Autoclastic

Autoclastic

Autoclastic

Pyroclastic– autoclastic Pyroclastic– autoclastic Autoclastic

Deposit type

35

27

19 34

45

24

33

22

13

No. of samples b

18

8

19 16

11

19

14

17

11

SC

17

19

0 18

34

5

19

5

2

DC

5 22 19 33 5 24 34 45 19 18 34 19 27 17 35

13

s’

FR LRcSC FR LRcSC FR LRcSC FR LRcSC SC FR LRcSC FR LRcSC FR LRcSC

LRcSC

Magnetization c

3.8499 11.5941 8.8623 6.3385 1.2492 24.7124 24.0133 7.7963 18.7491 13.8560 3.9940 13.9638 5.9088 10.5053 9.2316

1.1740

R

b

One deposit per site Total number of clasts sampled in deposit c Magnetization(s) used in statistical analysis d Samples in distribution are to 5% certainty. All other distributions are non-random to 5% certainty

a

21

20

14 17

10

Unsorted boulder breccia Red breccia Matrix-rich, stratified tuff breccia Matrix-poor, wellsorted breccia Unsorted boulder breccia

Red breccia

9

5

3

Matrix-rich, stratified tuff breccia Matrix-rich, stratified tuff breccia Gray breccia

Breccia type

1

Site a

066.0 078.7 024.8 n.a. n.a. 023.6 359.6 n.a. 106.8 354.2 n.a. 345.3 n.a. 025.4 038.8

n.a.

D7

–79.8 –68.9 –24.9 n.a. n.a. –59.7 –59.3 n.a. –49.3 –48.4 n.a. –68.3 n.a. –65.5 –43.5

n.a.

I7

48.2 30.4 37.5 n.a. n.a. 3.2 16.2 n.a. 4.0 19.5 n.a. 20.8 n.a. 29.2 42.0

n.a.

a95

4.3 1.8 1.6 n.a. n.a. 86.9 3.4 n.a. 75.7 4.3 n.a. 3.8 n.a. 2.6 1.1

n.a.

k

CC overlaps TAFD and PDF CC overlaps PDF but not TAFD CC overlaps TAFD and PDF Random distribution d Random distribution d CC overlaps PDF but not TAFD CC overlaps TAFD and PDF Random distribution d CC does not overlap TAFD or PDF CC overlaps TAFD and PDF Random distribution d CC overlaps TAFD and PDF Random distribution d CC overlaps TAFD and PDF CC does not overlap TAFD or PDF

Random distribution d

Comments on Fisherian mean

Table 3 Site characteristics and paleomagnetic site-mean data for volcanic breccias. SC number of clasts with one magnetization; DC number of clasts with dual-component magnetizations; s’ number of samples used in statistical analysis; R vector sum of s’ unit vectors; D7, I7 in situ site-mean declination and inclination direction; a95 radius of 95% confidence cone; k best estimate of concentration parameter; FR first-removed magnetization of multi-magnetization specimen; LR second-removed magnetization of multimagnetization speciment; SC single magnetization removed from a specimen; PDF present-day field; TAFD Quaternary time-averaged field direction; CC radius of 95% confidence cone for Fisherian mean

74

75

A 10

550o

20

30

A/m

40

50

E, Horiz. 60

A/m

W, Horiz. 20

10

dual-component 10 magnetization

single-component magnetization

10 20

B

A/m

580 o

20

500o

30 o

400 300o o 200 NRM

30 40

550o

100o

S, Down

40

580o

A/m

Fig. 10A–C Examples of demagnetization characteristics of a gray breccia associated with tabular lava-flow cores at site 5. Modified orthogonal demagnetization diagrams of TD data for clasts exhibiting A single-component and B dual-component magnetizations at site 5. Isolated magnetization directions are highlighted by arrows. C Equalarea projections of magnetization directions isolated in TD showing contrast in grouping between low-unblocking-temperature component magnetizations from two-component specimens (DC-FR) and highunblocking temperature component magnetizations (DCLR) and single-component (SC) directions. Open symbols and closed symbols plot in the lower and upper hemispheres, respectively. Site mean direction and projected cone of 95% confidence for adjacent lava-flow core (site 6) shown for comparison. Symbols and conventions as in Figs. 8 and 9. Note clustering of DC-FR magnetization into two groups, one of which is similar in direction to the magnetization characteristic of the adjacent lava flow (site 6)

NRM

500o

70

o

425

150o

80 225o

C

S, Down

N

N

n = 19

X

*

n = 33

X 6

*

6 E

low -tem perature m agnetizations (D C -FR )

These data indicate heterogeneous emplacement temperatures for the sampled breccias, with some clasts emplaced at temperatures of approximately 3007 (site 21) or approximately 5507 (site 10) and others likely at or close to ambient temperature. The greater dispersion of magnetization directions for breccia clasts than for samples from related lava-flow cores was also noted by Grubensky et al. (1998) and may relate to minor slumping, sliding, or compaction of the breccia after initial emplacement of the deposit. Matrix-poor, well-sorted breccia. Site 20 was sampled in this breccia type and yielded mostly dual-component magnetizations (Tables 3, 4). The upper limit of unblocking temperatures for FRs ranges from 100 to 550 7C, and lower limits of unblocking temperatures of LRs overlap slightly with FRs. LRs and SCs are statistically random, whereas FRs are dispersed but non-random and yield an estimated mean direction that is statistically indistinguishable from that of underlying (site 18) and overlying (site 19) lava-flow cores, the TAFD, and the PFD.

high-tem perature m agnetizations (SC , D C -LR )

As with the unsorted boulder breccias, the breccia at this site consists of clasts emplaced over a range of temperature and the data imply cooling, and acquisition of all or some remanent magnetization, before penultimate incorporation in the breccia. Matrix-rich, stratified tuff breccia. Clasts from matrixrich, crudely to well-stratified breccia at sites 1, 3, and 17 yield paleomagnetic results that also suggest heterogeneous emplacement temperatures, but there is a high likelihood that most clasts were emplaced at relatively low temperatures. Notably, two of the three sites (Table 4) are dominated by single-component clasts with random or very poorly grouped magnetization directions (Fig. 12). Even the low-temperature, FR remanence in DC clasts is poorly grouped (Fig. 12). The magnetization directions are not entirely random and are similar to expected directions, so some clasts likely were deposited at elevated temperatures. The proportion of such hot clasts, compared with those emplaced at or near ambient temperatures, is difficult to determine.

76

A

B

N

N

22 X

*

22 X

21 23

*

21

23 E

n = 35 high-temperature magnetizations (SC, DC-LR)

Fig. 11A, B Equal-area projections of magnetization directions isolated in TD data from unsorted boulder breccia at site 21. A High-unblocking temperature component directions from twocomponent specimens (DC-LR), directions from single-component magnetization clasts (SC), estimated mean direction and confidence cone based on both types of directions (21), and site mean directions and confidence cones for sites in adjacent lavaflow cores (sites 22 and 23). B Low-unblocking temperature component directions from two-component specimens (DC-FR), calculated mean direction and confidence cone (21), and site mean directions and confidence cones for sites in adjacent lava-flow cores (22 and 23). Symbols and conventions as in Figs. 9 and 10. Although the mean direction for the SC and DC-LR magnetizations overlaps with those of the adjacent lava flows, the confidence cone is too large to permit an unequivocal interpretation. Tighter clustering of DC-FR data are consistent with emplacement at elevated temperatures but below the maximum blocking temperature

Interpretations of breccia types Pinnacle Ridge is composed of volcanic rocks erupted from a vent at or south of The Great Pinnacle (Hackett and Houghton 1989) and emplaced on the northern flanks of a rugged composite cone. Volcanic rocks exposed on the west side of the ridge have been divided into nine unconformity-bounded constructional units. The strongest dissection, revealed by high-angle unconformities, preceded deposition of units II and IV. The exposed heights of the high-angle segments of these unconformities are presently a small fraction of their total relief, which may be of the order of the height of modern Pinnacle Ridge, approximately 350 m. The absence of units II, III, IV, V, VII, and VIII on the east flank of Pinnacle Ridge indicates the older unconformity is also laterally extensive. Repeated dissection in this sector of the cone suggests a history marked by inactive periods following initial construction and subsequent reconstructions of the cone. Widths of the steep-walled ravines associated with the mapped unconformities are unknown, but the presence of west- and east-dipping deposits that include lava-flow cores is probably a result of V-shaped canyons cut rapidly by surface runoff. Lo-

n = 17 low-temperature magnetizations (DC-FR)

cal primary dip directions perpendicular to the overall northward dip of the Te Herenga Formation suggest that overall northward transport was also accompanied by minor east- or west-directed flow, where the margins of flows crossed interfluves into adjacent ravines. Moderate, primary dips and steep unconformities indicate that the slope angle of substrates for deposits of Te Herenga Formation were highly variable, probably exerted some influence on flow emplacement, and resulted in heterogeneities in deposit structure and texture for individual breccia deposits. Red breccias with irregularly shaped lava-flow cores The red breccias and intimately associated lava-flow cores represent aa flows. The association of a dense lava core encased in spinose, vesicular clasts is consistent with aa as described by Macdonald (1953, 1967). Red-breccia clasts and matrix at Pinnacle Ridge are autoclasts, products of granulation of associated lava-flow cores during downslope movement. Marginal gashes along margins of cores record the early stages of the shredding of rigid lava to produce the characteristic clinkery autoclasts of aa flows (Rowland and Walker 1990). Decrease in clast angularity with distance from lava-flow core suggests abrasion and granulation of clasts as the lava, including its encapsulating breccia, continues downslope. Increasing reddening with distance from lava-flow cores indicates high-temperature oxidation as clasts came into contact with air in the interstices of the developing breccia. Paleomagnetic data indicate that clasts and associated lava-flow cores moved and cooled together at temperatures initially above the maximum laboratory unblocking temperatures (ca. 580 7C). The complex intermingling of thin flow cores split from a single, thick core updip are consistent with collective high-temperature emplacement of breccia with multiple cores. The absence of intervening sedimentary rocks, and the statistical similarity of estimated mean directions for close-

SC DC–LR DC–FR

19 5 5

1 130 1 130 50 130 130 40

580 580 200–400 580 580 225–550 580 580 100–525

Matrix-rich, stratified tuff breccia 11 Site 1 SC 2 DC–LR 2 DC–FR

Site 17

Site 3

SC DC–LR DC–FR SC DC–LR DC–FR

17 5 5 16 18 18

130 130 10

580 580 100–550

Matrix-poor, well-sorted breccia 8 Site 20 SC 19 DC–LR 19 DC–FR 1 130 1 130 30

130 130 20

580 580 100–300

SC DC–LR DC–FR

18 17 17

Site 21

130 130 10–20

580 580 150–550

130 130 10

130 130 5

580 580 150 580 580 200–550

130

580

Clast were reoriented during or after remanence acquisition; emplacement temperatures ranged 100–550 7C for different clasts

Clasts were reoriented during or after remanence acquisition; emplacement temperatures ranged 100–300 7C

Clasts were reoriented during or after remanence acquisition; emplacement temperatures ranged 150–550 7C

Clasts were reoriented during or after remanence acquisition; emplacement temperatures ranged 200–5507

Clasts emplaced at temperature 1 580 7C; FR magnetization for clasts with DC magnetization is of viscous origin

Clasts emplaced at temperature 1 580 7C

Interpretation

All results are non-random but very poorly grouped

Clasts were reoriented during or after remanence acquisition; emplacement temperatures 100–525 7C

Most or all clasts reoriented after remanence acquisition; two clasts possibly emplaced at 200–400 7C, but emplacement may actually have been at ambient temperature All results are non-random but very poorly grouped; Emplacement temperatures ambiguous and possibly CC overlaps with overlying lava-flow core (site 13) very heterogeneous for different clasts Random Random Insufficient data

Random Random Well grouped, indistinguishable from underlying (site 18) and overlying (site 19) lava-flow cores

Random Random Dispersed but with CC that overlaps underlying (site 11) and differs from overlying flow core (site 12) Non-random, but very large CC Non-random, but very large CC Well grouped, indistinguishable from underlying (site 22) and overlying (site 23) lava-flow cores

Random Random Dispersed into two groups that overlap mean direction for overlying lava-flow core (site 6)

Well grouped and indistinguishable from overlying (site 16) and underlying (site 15) lava-flow cores Well grouped and indistinguishable from overlying (site 8) and underlying (site 7) lava-flow cores Random

Maximum Grouping unblocking field strength (AFD; mT)

11 34 34

Unsorted boulder breccia Site 10 SC DC–LR DC–FR

Not gradational with lava-flow cores

Gray breccia; tabular lava-flow cores 14 Site 5 SC 19 DC-LR 19 DC-LR

Site 9

Red breccia; irregular lava-flow cores Site 14 SC 19

Gradational with lava-flow cores

Breccia type Magnetization No. of Maximum and site (TD) clasts (TD) unblocking temperatures (TD; 7C)

Table 4 Summary of demagnetization data for interpreting emplacement temperatures of breccia clasts. SC single component magnetization; DC dual-component magnetization with first removed (FR) and last removed (LR) magnetizations separated; AFD alternating field demagnetization; TH thermal demagnetization; CC 95% confidence cone

77

78 Fig. 12 Equal-area projections of magnetization directions isolated from TD data from A site 3 and B site 17 in matrixrich, well-stratified breccia. Mean direction and confidence cone from lava-flow core (site 13) adjacent to site 3 is also shown

N

A

site 3

X

*

13 3

high-temperature magnetizations (SC, DC-LR)

E

n = 22 B

N

N

site 17 X

X

*

17

* E

high-temperature magnetizations (SC, DC-LR)

n = 18

n = 35 low-temperature magnetizations (DC-FR)

ly spaced breccia and cores at sites 14 and 9, are consistent with collective emplacement rather than with discrete, successive emplacement of individual flows of autobreccia and lava-flow cores.

temperature of magnetite, followed by incorporation into a flow-front breccia apron, where they then cooled to ambient temperature and acquired a lower-unblocking-temperature magnetization.

Gray breccia with tabular lava-flow cores

Matrix-poor, well-sorted breccia

Tabular lava-flow cores associated with thick, coarsegrained, gray breccias were probably emplaced as block-lava flows. These flows were prone to developing a carapace of unconsolidated, smooth-sided angular clasts and matrix that rode loosely atop the core. Avalanching of this breccia carapace from active flow fronts produced crudely layered basal breccia (Fig. 13). Paleomagnetic data from breccia at site 5, shed off the top of the flow at site 6, are consistent with an origin atop or within the front of an active, block-lava flow. Single-component magnetization clasts were emplaced at, or below, the unblocking temperature range (the lower end of which is between 175 and 550 7C) of each clast. Dual-component magnetizations were acquired as a result of cooling of carapace clasts to temperatures below the maximum laboratory unblocking

Matrix-poor, well-sorted breccia is most likely also associated with the emplacement of block-lava flows and represents rubble-strewn flow-front aprons. The fragments are interpreted as autoclasts, because they are strictly monolithologic and are identical texturally and mineralogically to closely associated lava-flow cores. The paucity of matrix other than sand formed by clast–clast grinding, and the uniform size of clasts, are both consistent with descriptions of block-lava-flow deposits by Williams (1942). Individual breccia intervals interlayered with lava-flow cores thin abruptly downslope away from the termination of the associated lavaflow cores, indicating a genetic relationship to the coherent lava rather than a separate pyroclastic or sedimentary facies. The associated thin lava-flow cores were emplaced between breccias and are among the

79

Matrix-rich, stratified tuff breccias

Fig. 13 Diagrammatic representation of the formation of crudely stratified breccia below an advancing block-lava-flow core by avalanching of carapace breccia down the flow front, with addition of breccia blocks created at the flow front

thinnest of the tabular type at Pinnacle Ridge. We interpret each lava-flow core as a lobe separated from the parent flow and each breccia layer as an avalanche deposit of a carapace breccia. Paleomagnetic data support a complex origin for the clasts in these deposits. Clasts with DC magnetizations may have been separated from lava flows at a wide range of temperatures below the Curie temperature and then cooled completely in the flow-front breccia. Unblocking temperatures for FRs, which range from 100 to 550 7C, place an upper limit on clast emplacement temperatures. Single-component clasts that exhibit magnetization unblocking from low to high temperatures must have cooled to temperatures ~100 7C prior to incorporation in the flow-front breccia. The dispersion of FRs may be a result of settling following deposition or prolonged emplacement history that allowed some clasts to cool to ambient temperature prior to final deposition along with younger lava cores. Unsorted boulder breccias Unsorted boulder breccias sampled in units VI and VII are not intimately associated with lava-flow cores, but most of their clasts are interpreted as autoclasts because they are texturally and mineralogically similar to cores associated with gray breccia described previously and because the poor sorting is inconsistent with a pyroclastic-fall origin. Concentrations of gray clasts without matrix sands are interpreted as products of sliding and transport-related fracturing of segments of cooled block lava cores. Paleomagnetic data from the breccia clasts are also consistent with derivation of the clasts from flowing lava in a two-step process described for the matrix-poor, well-sorted breccia type.

Matrix-rich, well-stratified tuff breccias have clast characteristics similar to matrix-poor varieties, but the abundance of fine-grained material in the matrix is in excess of that in carapace breccias of block lava flows exposed in the study area. The additional presence of scoriaceous fragments is unique to this type of breccia and suggests that the deposits are at least in part pyroclastic, although these fragments may have formed as scoriaceous flow crusts on lava flows not exposed in the study area. The well-sorted hypohyaline to holocrystalline grains of the matrix are also likely of pyroclastic origin, based on grain shape and sorting. Matrix-rich, well-stratified breccias were probably not emplaced by a single mechanism. Massive or weakly graded deposits of ash and coarser fragments of pyroclastic origin, including poorly vesicular smoothsided blocks likely to be of autoclastic origin, are compatible with wet or dry mass flows derived by reworking of slope debris. Internally stratified beds were likely deposited by shallow, poorly channeled stream flow. The presence of clasts emplaced at elevated temperatures in some of these deposits suggests accumulation during, rather than long after, eruptive episodes. Bimodal grain size may be characteristic of debrisflow and flood deposits preserved on composite volcanoes. We have observed strikingly similar, bimodally textured deposits among cone-forming sequences in Tertiary volcanoes (Smith and Grubensky 1993; Lavine et al. 1996). Well-sorted tephra and coarse autoclastic, and lesser pyroclastic, blocks comprise the most readily eroded debris on the volcano slopes. The more wideranging or polymodal grain sizes more typical of debrisflow deposits in distal volcaniclastic aprons may result from bulking of alluvium from streambeds and banks during transport.

Formation of autoclastic breccias The distinction of basal (under the moving flow) and carapace (on top of the moving flow) breccias can be difficult in a sequence of interlayered breccia and coherent lava. We interpret sharp, irregular contacts to separate genetically unrelated breccias. Thus defined, breccia is commonly thicker above, rather than below, lava-flow cores, especially within and near the top of constructional units (Fig. 4). We believe that this is explained by a qualitative appraisal of lava-flow dynamics. The formation of a brittle crust requires cooling and crystallization to rigidus conditions (i.e., crystallization to the point where the material behaves as a solid and can be fractured). Conductive heat loss is greater at the air-lava interface than at the flow base (Borgia and Linneman 1990), especially where lava moves across previously emplaced, but still hot, flows. Flow velocity is also highest in upper levels of the flow because of drag at the base. Brecciation of flowing lava, therefore,

80

takes place preferentially at the top of the flow as the rigidus front migrates downward through the fastest moving part of the flow. The brecciated flow crust, once formed, moves faster than the underlying lava but experiences little internal strain and may exhibit plug flow (Borgia and Linneman 1990). Fragments may not, therefore, experience significant rotations as they cool and acquire part or all of their magnetization while in transit on top of the flow. Because the carapace is moving faster than the fluid interior, these fragments will be transported to, and will avalanche down, the flow front. This process may account for dual-component magnetizations – one acquired during minimal rotation during flow, and the second acquired after the clast came to rest downslope of the flow front. Contrasting magnetizations might be expected for autoclasts originating on domes where continuous, or near-continuous, rotation and displacement of clasts due to endogenous dome growth occurs. These rotations and displacements will ultimately produce autoclasts with complex, multiple-component magnetizations or magnetization directions, identified in demagnetization behavior, that are smoothly varying and unresolvable into discrete components (Grubensky et al. 1993; Grubensky 1996). Basal flow breccia is likely formed primarily by overriding of carapace breccia that has avalanched from the flow front. This interpretation is strongly supported by numerous observations of crude layering in basal breccia that is more steeply dipping than the overlying lava flow. Such crude layers are commonly inversely graded, consistent with formation by grainflow avalanches on steep slopes (Fig. 13). Formation of autoclastic breccia that is not intimately interlayered with lava-flow cores may arise in two ways. Cooling at the top and front of the slowly moving flow may cause wholesale brecciation of a flow front as it reaches the rigidus but continues to be pushed down steep slopes by lava entering at the rear of the flow. In this way, the front of the flow would become a continually advancing mass of breccia that, at the end of the eruption, would form a downslope-tapering rubblestrewn apron in front of interlayered lava and breccia. This mechanism might explain the wedge-shaped geometry of constructional unit VIII and the formation of matrix-poor, well-sorted breccia. A second mechanism is associated with the “collapsing phase” of blocklava-flow advance described by Borgia and Linneman (1990). At Arenal, block-lava flows advance in channels bounded by levees generated by the shouldering aside of collapsing carapace breccia at the flow front. When effusion of lava ends at the vent, the lava drains through the channel to the flow front, leaving a collapsed crust on the channel floor. Because the levees and collapsed crust are both composed of autoclasts, a chaotic breccia results with little or no preserved coherent lava, although such lava should be found in correlative downslope positions. We suggest that unsorted boulder breccia formed in this fashion. Note that these

two mechanisms can explain both downslope increase (Hackett and Houghton 1989) and decrease (Lipman 1968) in the ratio of autobreccia to coherent lava in cone-forming facies.

Comparison to previous studies All of the breccias examined are, with high probability, composed entirely or largely of autoclasts. Most deposits were emplaced concurrent with lava flows. Emplacement of aa and block-lava flows on shallow to moderately steep slopes resulted in deposition of all associated autobreccia clasts in close proximity to cores as flow-front debris aprons. Slopes greater than approximately 307 were present in the paths of many lava flows and, as a result, gravity slides of loose carapace breccia were deposited far from the farthest downslope extent of the parent lava-flow cores. The clast and deposit origins for many of the breccias in the Te Herenga Formation have been classified and interpreted differently in previous studies. Hackett (1985) and Hackett and Houghton (1989) distinguished (and interpreted) monolithologic tuff breccia (autobreccia associated with lava-flow cores), heterolithologic tuff breccia (secondary mass flows of pyroclastic/ autoclastic material), and moderately sorted, bedded sand and gravel (fluvial/colluvial deposits derived by reworking of heterolithologic tuff breccia). Brooker (1992) differentiated autoclastic breccia, matrix-supported breccia (lahars and hyperconcentrated floodflows), clast-to-matrix-supported breccia (lahar deposits), and clast-supported breccia (lahar deposits). We note that similar approaches have been taken at other volcanoes where abundant breccia in the absence, or near absence, of coherent lava has led to the interpretation of cone-forming deposits comprised to a large extent by “laharic” breccia of inferred sedimentary origin (e.g., Lydon 1968; du Bray and Harlan 1998). We suggest that thick autoclastic breccias are more likely components of cone-forming facies than are reworked sedimentary deposits. The past inferred importance of pyroclastic processes to produce breccia fragments, and of aqueous reworking to produce the deposits, relates to three general assumptions: 1. All, or most, sandy matrix material is called ash or tuff, despite the paucity of scoria or shards in most breccia deposits. This interpretation implies that all fine-grained volcaniclastic material is produced explosively. Observations of flow-top breccias gradational with lava-flow cores, however, suggest that autobrecciation of block lava flows is capable of producing a sizeable fraction of fines and is consistent with observations of active flows showing that flowfront breccias can be characterized by an abundance of poorly sorted, fines-rich debris (Rittmann 1962; Linneman and Borgia 1993). We believe that most sub-centimeter grains in Te Herenga Formation breccias originate by autoclastic granulation. Only

81

the abundant well-sorted, largely hypohyaline matrix grains of the matrix-rich, stratified breccias are likely of pyroclastic origin and appropriately referred to as ash and lapilli. 2. Only monolithologic deposits are related to emplacement of lava flows, so that deposits with diverse clast compositions necessarily indicate a pyroclastic component and a likely reworked origin to the deposit. The difficulty here lies in an objective, meaningful definition of monolithologic vs heterolithologic. Pinnacle Ridge breccias are heterolithologic in terms of variations in vesicularity, degree of flow banding, and color. In terms of mineralogy and phenocryst size and abundance, however, the same deposits are clearly monolithologic. Lava flow carapaces in the Te Herenga Formation contain clasts with a range of vesicularity and oxidation coloration, but we view these as monolithologic. Therefore, and consistent with paleomagnetic data, we interpret all breccias, except the matrix-rich, stratified variety, to be monolithologic and of autoclastic origin. 3. Only coherent lava is referred to as lava flows, and most fragmental deposits are of primary pyroclastic or sedimentary origin. This assumption implies that the tops and bottoms of lava-flow cores delimit the margins of the original lava flow, and that autoclastic deposits lacking cores cannot be lava flows. As stated above, it has been shown that most breccias at Pinnacle Ridge were emplaced hot. The interpretations in this study are consistent with historical observations (Finch 1933; Jones 1943; Macdonald 1953, 1967, 1972; Kilburn and Guest 1983; Cigolini et al. 1984) of andesitic and basaltic lava flows that range from thin carapaces of breccia surrounding a predominant, massive core to thoroughly brecciated lava with only minor, or subordinate, massive cores. The lava flows (cores plus associated breccia) at Pinnacle Ridge have been matched to the existing lavaflow classifications using textural characteristics between autoclasts and relative proportions and distributions of lava-core and breccia components. Autoclastic fragments atop, below, or in front of an active flow are part of the flow during emplacement. Some volcanoes (e.g., Summer Coon, Colorado; Smith and Grubensky 1993) may therefore be dominated by autoclastic breccia. Despite efforts to sample deposits anticipated to yield clasts with narrow ranges of emplacement temperatures, and thus a readily interpretable emplacement history, our paleomagnetic data from Pinnacle Ridge suggest that most deposits were thermally heterogeneous upon emplacement. In the absence of consistent thermal demagnetization results, the presence of dualcomponent magnetizations is used as an indicator of an autoclastic origin for clasts. If all clasts in breccias that are not gradational with lava-flow cores had either a single-component magnetization or dual-component magnetizations and a consistent unblocking temperature for the FRs, they could be interpreted as pyro-

clasts. Thorough mixing of clasts with single- or dualcomponent magnetizations requires concurrent formation of both types of clasts within a single flow or thorough mixing of clasts from different pre-existing deposits. Based on the presence of clasts with apparently complex magnetization history in well-sorted, matrixpoor breccias closely associated with the emplacement of lava flow cores, it is more likely that lava flows are capable of producing both types of clasts simultaneously.

Conclusion The implications of new field and paleomagnetic data from breccias and lava-flow cores at Pinnacle Ridge reemphasize the need for reconsidering Macdonald’s (1972) model of composite cone composition, as is implicit in Hackett and Houghton (1989). Unconformitybounded constructional units are highly irregular in geometry. Downslope continuity of layering is precluded by erosion between eruptive periods and lateral transitions between lava-flow cores and autobreccia. Building upon the work of Hackett (1985) and Brooker (1992), our field observations and paleomagnetic data indicate that most breccias in the Te Herenga Formation, which are central to the Hackett and Houghton (1989) model for composite volcano lithofacies assemblages, were emplaced concurrently with lava-flow cores and are derived from fragmentation of core material during flow without any significant pyroclastic component. Lava flows interpreted as aa are composed of one or more flow cores and red, pervasively oxidized carapace breccia. Exposures of block-lava flows, including coherent cores with gradations to sandy, flow-top breccias, indicate that autobrecciation of some lava flows can produce large amounts of fine-grained debris. Block-lava flows moving on steep slopes may also form thick autoclastic breccias that are not intimately associated with lava flows. The alternation of coherent lava and fragmental deposits that “define” composite volcanoes may, in many cases, be better described as an alternation of lava-flow core and associated autoclastic breccia, rather than as interlayered lava flows and pyroclastic deposits. In some cases, the volcaniclastic layers may be of hybrid pyroclastic-autoclastic origin (Grubensky et al. 1998). Paucity of pyroclastic fragments and deposits need not imply, however, a lack of explosive eruptions. Pyroclastic fragments, particularly those that are fine grained and/or highly vesiculated, are likely to have limited preservation on steep slopes unless the deposits are welded or quickly buried by lava flows; otherwise, pyroclastic deposits will likely be readily eroded and preservation, in primary or reworked deposits, will be preferred in lowland areas and aggrading river valleys near the volcano. Matrix-rich, stratified tuff breccia at Pinnacle Ridge likely records such stripping and redistribution of pyroclastic material.

82 Acknowledgements This research was funded by the National Science Foundation (EAR91–17241) with substantial logistical support from the New Zealand Institute of Geological and Nuclear Sciences (IGNS). M.J. Grubensky received additional support from the Geological Society of America. We are indebted to B. Houghton (IGNS) for his substantial assistance and advice, which made the fieldwork possible. B. Macy provided instrumental assistance in the development of the furnace built by M.J. Grubensky for this study, and R. Molina-Garza helped in laboratory maintenance and data interpretation. Comments by D. Swanson and W.R. Hackett improved the manuscript.

References Aramaki S (1966) Geology of Asama Volcano. J Faculty Sci Univ Tokyo, 14 : 229–439 Aramaki S, Akimoto S (1957) Temperature estimation of pyroclastic deposits by natural remanent magnetism. Am J Sci 255 : 619–627 Borgia A, Linneman SR, Spencer D, Morales LD, Brenes JA (1983) Dynamics of lava flow fronts, Arenal volcano, Costa Rica. J Volcanol Geotherm Res 19 : 303–329 Borgia A, Linneman SR (1990) On the mechanism of lava flow emplacement and volcano growth: Arenal, Costa Rica. In: Fink JH (ed) Lava flows and domes, emplacement mechanisms and hazard implications. Springer, Berlin Heidelberg New York, pp 208–243 Borgia A, Poore C, Carr MJ, Melson WG, Alvarado GE (1988) Structural, stratigraphic, and petrologic aspects of the ArenalChato volcanic system, Costa Rica: evolution of a young stratovolcanic complex. Bull Volcanol 50 : 86–105 Brooker J (1992) The geology of northwestern Pinnacle Ridge, Te Herenga Formation, Ruapehu volcano, New Zealand. B.Sc. thesis, Victoria University, Wellington, New Zealand, pp 1–118 Butler RF (1992) Paleomagnetism. Blackwell, Boston, pp 1–319 Cigolini C, Borgia A, Casertano L (1984) Intra-crater activity, aablock lava viscosity and flow dynamics: Arenal Volcano, Costa Rica. J Volcanol Geotherm Res 20 : 155–176 Clement BM, Connor CB, Graper G (1993) Paleomagnetic estimate of the emplacement temperature of the long-runout Nevado de Colima volcanic debris avalanche deposit, Mexico. Earth Planet Sci Lett 120 : 499–510 Cotton CA (1944) Volcanoes as landscape forms. Christchurch, Whitcombe and Tombs, Ltd, pp 1–416 Downey WS, Tarling DH (1991) Reworking characteristics of Quaternary pyroclastics, Thera (Greece), determined using magnetic properties. J Volcanol Geotherm Res 46 : 143–155 du Bray EA, Harlan SS (1998) Geology, age, and tectonic setting of the Cretaceous Sliderock Mountain volcano, Montana. US Geol Surv Prof Pap 1602 : 1–19 Finch RH (1933) Block lava. J Geol 41 : 769–770 Fisher RV (1958) Definition of volcanic breccia. Geol Soc Am Bull 69 : 1071–1073 Fisher RV (1961) Proposed classification of volcaniclastic sediments and rocks. Geol Soc Am Bull 72 : 1409–1414 Fisher RV (1966) Rocks composed of volcanic fragments and their classification Earth Sci Rev 1 : 287–298 Fisher RV, Schmincke HU (1984) Pyroclastic rocks. Springer, Berlin Heidelberg New York, pp 1–473 Grubensky MJ (1996) Volcanic breccias, evaluation of fragment and deposit origins and distribution with composite volcanoes. Dissertation, Univ of New Mexico, Albuquerque, pp 1–260 Grubensky MJ, Smith GA, Geissman JW (1993) Temperature of emplacement studies on block-and-ash flow and debris-flow breccias at Mount Hood, Oregon. Geol Soc Am Abstr Prog 25 (5): 45 Grubensky MJ, Smith GA, Geissman JW (1998) Field classification and paleomagnetic characterization of lithic and scoriaceous breccias, Pleistocene Broken Top volcano, High Cascade Range, Oregon. J Volcanol Geotherm Res 83 : 93–114

Hackett WR (1985) Geology and petrology of Ruapehu volcano and related vents. Dissertation, Victoria University, Wellington, New Zealand, pp 1–407 Hackett WR, Houghton BF (1989) A facies model for a Quaternary andesitic composite volcano: Ruapehu, New Zealand. Bull Volcanol 51 : 51–68 Heiken GM, Wohletz KH (1984) Volcanic ash. Univ California Press, Berkeley, pp 1–246 Heiken GM, Wohletz KH (1991) Fragmentation processes in explosive volcanic eruptions. Soc Econ Paleontol Mineral Spec Publ 45 : 19–26 Hoblitt RP, Kellogg KS (1979) Emplacement temperatures of unsorted and unstratified deposits of volcanic rock debris as determined by paleomagnetic techniques. Geol Soc Am Bull 90 : 633–642 Jones AE (1943) Classification of lava surfaces. Am Geophys Union Trans 24 : 265–268 Kilburn CRJ, Guest JE (1983) Aa lavas of Mount Etna. In: Kilburn CRJ, Luongo G (eds) Active lavas: monitoring and modelling. UCL Press, pp 73–106 Kirschvink JL (1980) The least-squares line and plane analysis of paleomagnetic data. Geophys J R Astron Soc 62 : 699–718 Lavine A, Smith GA, Goff F (1996) Volcaniclastic rocks of the Keres Group: insights into mid-Miocene volcanism and sedimentation in the southeastern Jemez Mountains. In: Goff F, Kues BS, Rogers MA, McFadden LD (eds) The Jemez Mountains region. New Mexico Geological Society Guidebook, 47th Field Conference, pp 211–218 Linneman SR, Borgia A (1993) The blocky andesitic lava flows of Arenal volcano, Costa Rica. In: Kilburn CRJ, Luongo G (eds) Active lavas: monitoring and modelling. UCL Press, pp 25–72 Lipman PW (1968) Geology of the Summer Coon volcanic center, eastern San Juan Mountains, Colorado. Colorado Sch Mines Q 63 : 211–226 Lydon PA (1968) Geology and lahars of the Tuscan Formation, northern California. Geol Soc Am Mem 116 : 441–475 Macdonald GA (1953) Pahoehoe, aa, and block lava. Am J Sci 251 : 169–191 Macdonald GA (1967) Forms and structures of extrusive basaltic rocks. In: Hess HH, Poldervaart A (eds) Basalts: the Poldervaart treatise on rocks of basaltic composition. Interscience, New York, pp 1–61 Macdonald GA (1972) Volcanoes. Prentice-Hall, Englewood Cliffs, New Jersey, pp 1–510 Néel L (1955) Some theoretical aspects of rock magnetism. Adv Phys 4 : 191–242 Perret FA (1950) Volcanological observations. Carnegie Inst Wash Publ 549 : 1–162 Rittmann A (1962) Volcanoes and their activity. Wiley, New York, pp 1–305 Rowland SK, Walker GPL (1990) Pahoehoe and aa in Hawaii: volumetric flow rate controls the lava structure. Bull Volcanol 52 : 615–628 Smith GA, Grubensky MJ (1993) Toward a facies model for breccia-dominated composite volcanoes: Summer Coon Volcano (Oligocene), south-central Colorado. EOS Trans Am Geophys Union 74 (43): 644 Tamura Y, Koyama M, Fiske RS (1991) Paleomagnetic evidence for hot pyroclastic debris flow in the shallow submarine Shirahama Group (Upper Miocene–Pliocene), Japan. J Geophys Res B 96 (21): 779–787 Théllier E (1941) Sur la verification d’une methode permettant de determiner l’intensite du champ terrestre dans la Passe. C R Acad Sci 212 : 281–283 Watson GS (1956) A test for randomness of directions. Mon Not R Astron Soc Geophys Suppl 7 : 153–159 Williams H (1942) The Geology of Crater Lake National Park, Oregon, with a reconnaissance of the Cascade range southward to Mount Shasta. Carnegie Inst Wash Publ 540 : 1–162 Williams H, McBirney AR (1979) Volcanology. Freeman, San Francisco, pp 1–397