Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

Bull Volcanol (1992) 54:504-520

Vol a ology
9 Springer-Verlag 1992

A reappraisal of ignimbrite emplacement:


progressive aggradation and changes from particulate to
non-particulate flow during emplacement of high-grade ignimbrite
Michael J Branney and Peter Kokelaar
Department of Earth Sciences, University of Liverpool, P. O. Box 147, Liverpool L69 3BX, UK

Received March 11, 1991/Accepted November 11, 1991

Abstract. We propose a mechanism by which massive is also influenced by factors that affect flow behaviour,
ignimbrite and layered ignimbrite sequences - the latter such as topography. It may occur at any location later-
liable to have been previously interpreted as multiple ally between a proximal site of deflation (e.g. a foun-
flow units - form by progressive aggradation during tain-fed lava) and a flow's distal limit, but it most com-
sustained passage of a single particulate flow. In the monly occurs throughout a considerable length of the
case of high-temperature eruptive products the mecha- flow path. Up-sequence variations in welding-deforma-
nism simplifies interpretation of problematic deposits tion fabric (between oblate uniaxial to triaxial and pro-
that exhibit pronounced vertical and lateral variations late) reflect evolving characteristics of the depositional
in texture, including between non-welded, eutaxitic, boundary layer (e.g. fluctuations from direct suspen-
rheomorphic (lineated) and lava-like. Agglutination can sion-sedimentation to deposition via traction carpets or
occur within the basal part of a hot density-stratified traction plugs), as well as possible modifications result-
flow. During initial incursion of the flow, agglutinate ing from subsequent, post-depositional hot loading and
chills and freezes against the ground. During sustained slumping. Similar processes can also account for lateral
passage of the flow, agglutination continues so that the lithofacies gradations in conduits and vents filled with
non-particulate (agglutinate) layer thickens (aggrades) welded tuff. Our consideration of high-grade ignim-
and becomes mobile, susceptible to both gravity- brites has implications for ignimbrite emplacement in
induced motion and traction-shear imparted by the general, and draws attention to the limitations of the
overriding particulate part of the flow. The particulate widely accepted models of emplacement involving
to non-particulate (P-NP) transition occurs in and just mainly high-concentration non-turbulent transport and
beneath a depositional boundary layer, where disrup- en masse 'freezing' of high-yield-strength plug flows.
tive collisions of hot viscous droplets give way, via
sticky grain interactions, to fluidal behaviour following
adhesion. Because they have different rheologies, the
particulate and non-particulate flow components travel Introduction
at different velocities and respond to topography in dif-
ferent ways. This may cause detachment and formation in this paper we appraise published and new field ob-
of two independent flows. The P-NP transition is con- servations of ignimbrites, and suggest new ways of
trolled by factors that influence the rheological proper- thinking about deposition from pyroclastic flows. We
ties of individual erupted particles (strain rate, temper- concentrate on high-grade ignimbrites (sensu Walker
ature, and composition including volatiles), by cooling 1983) because the shear fabrics they contain (defined
and volatile exsolution during transport, and by the by agglutinated pumice clasts and matrix) uniquely re-
particle-size population and concentration characteris- cord information about the nature of their depositional
tics of the depositional boundary layer. At any one lo- regime. A major feature of our thesis is that massive
cation along the flow path one or more of these can beds in ignimbrites, and apparently stacked flow units,
change through time (unsteady flow). Thus the P-NP can be formed by continuous incremental deposition;
transition can develop momentarily or repeatedly dur- this is in contrast to en masse deposition. In this scen-
ing the passage of an unsteady flow, or it can occur ario, welding also can occur incrementally throughout
continuously during the passage of a quasi-steady flow the depositional history of a flow, rather than only after
supplied by a sustained explosive eruption. Vertical fa- deposition of an entire ignimbrite. Thus, distinction be-
cies successions developed in the deposit (high-grade tween "primary" and "secondary" welding would be
ignimbrite) reflect temporal changes in flow steadiness inappropriate; it denies the possibility of a process con-
and in material supplied at source. The P-NP transition tinuum.
505
EXTREMELY
EXTREMELY ~ lid LOW-GRADE
HIGH-GRADE continuum

Fountain-fad Lava-like Rheomorphlc Non-rheomorphlc Non-weldedignimbrltes


<Z lava-flows Ignimbriles Ignimbdtes welded ignlmbrltes h~gh aspectra'~o 10~
(e.g. Duflield 1990} {e.g. SranneyetaL 199~,; (e.g. Schmincke& Swanson 1967; (e.g. Ross & Smith 196~ ~
Ekren sial 1984) Chapin & Lowell 1979; Ragan & Sheddan 1972; (e.g.Watker1989)
Orsi & Sheridan 1986) Rlehle 1973)

[ increase in eruption column height and consequent cooling

increase in temperature, decrease in particle viscosity and yield strength I

O3

_5
I--
increase in cnystal fragmentation (explosivity) ~ >

coalescence
173
Z
agglutination
O
O
welding
n
z
< non-particulate flow accompanying emplacement
6O non-particulate flowafter deposition

lateral particulate flow important during emplacement


LLI
0
9 syn- and post-emplacement vesiculation
II
droplets transported laterally

blocky or cuspate shards transported

I increase in accidental lithics, eroded from vent and ground surface


O3
O Fig. 1. The ignimbrite grade
increase in fragmentation of crystals continuum, with associated
rr eruption and emplacement
LU pumice imbrication and grain fabrics processes, conditions and prod-
I-- ucts. All gradations can occur,
(5
,<
non-particulate lineatione, rotated lithics and flow folds even within deposits of a single
rr stratified flow, as processes vary
<
"r" lobate flow margins both through time (i.e. vertically
0 through a deposit) and with dis-
I- widespread upper autobreccia tance from source (laterally
0 within a deposit). Solid lines in-
a Iocelised basal and/or internal autobreccia dicate some characteristic fea-
O
rr
tures; dotted lines indicate pos-
angular or cuspate shards preserved in ignimbdte
Q_ sible features. The diagram is
droplet shapes occasionally preserved necessarily a generalisation;
readers may well know of ex-
ceptions

Some silicic volcanic successions exhibit a wide vari- 1989), southeastern Arizona (du Bray and Pallister
ety of ignimbrite lithofacies ranging from partly 1991), and Pantelleria (Mahood and Hildreth 1986;
welded, through intensely welded and rheomorphic, to Orsi et al. 1991). This paper gives an explanation of
lava-like. Lithofacies can be particularly difficult to in- how such problematic lithofacies associations can be
terpret in terms of depositional process where apparent generated from a single sustained eruption and attend-
lavas and highly welded tufts pass vertically or laterally ant continuous flow.
into less intensely welded tufts, or nonwelded tufts, at
sharp or sharply gradational contacts. Examples of
Terminology
such associations occur at Gran Canaria (Schmincke
1969, 1990), southeast Korea (Park and Kim 1985; Walker (1983) proposed the term 'grade' to refer to the
Reedman et al. 1987a), southwestern Idaho (Ekren et amount of welding-compation exhibited by ignimbrites.
al. 1984), southwestern New Mexico (Duffield 1990), Although this terminology is not yet widely used it is
southwest USSR (references in Cook 1966), Kenya particularly useful because, in addition to reflecting
(McCall 1965; Leat 1985), the English Lake District cooling history, it embraces the concept of a grade con-
(Branney et al. 1992), Trans-Pecos Texas (Henry et al. tinuum (Fig. 1) that reflects variability of particle vis-
506

cosity and yield strength during ignimbrite emplace- clasts, highly attenuated fiamme, and rotated segments
ment, without implying which of many factors may be of broken fiamme, could have been formed by post-
the cause (e.g. eruption temperature, magma chemistry, emplacement loading. But some of the structural fea-
volatile content, syn-emplacement cooling). The contin- tures described by Schmincke and Swanson cbnstitute
uum may be divided into arbitrary categories as fol- classic criteria for rotational shear (see Simpson and
lows: 'extremely high-grade' ignimbrites are intensely Schmid 1983; Vernon 1987) and, as Schmincke (1974)
welded even to their upper surfaces, and can include recognised, the high-grade ignimbrites of Gran Canaria
lithofacies that are 'lava-like' (texturally indistinguisha- are different from lower-grade ignimbrites in that the
ble from lava: Ekren et al. 1984; Branney et al. 1992); viscosity of the magma particles was so low that some
'high-grade' ignimbrites are predominantly welded and particle deformation occurred before, and therefore in-
have intensely welded and rheomorphic zones; 'moder- dependently of, post-emplacement loading and cool-
ate-grade' ignimbrites have both welded and non- ing.
welded zones; 'low-grade' ignimbrites show little or no Ideas contrary to those of Schmincke and Swanson
evidence of welding and may range down to ignim- were presented by Wolff and Wright (1981) who inter-
brites emplaced at ambient temperatures. T h e grade preted the rheomorphic Green Tuff on Pantelleria as an
concept can also be used for tuffs of airfall, pyroclastic airfall deposit. According to this interpretation, rheo-
surge, intrusive, or unknown origin. morphism of the Green Tuff must postdate its initial
In this paper we use the term 'aggradation' for de- deposition because airfall does not involve lateral em-
posit construction by sedimentation. We use 'non-parti- placement. Wolff and Wright (1981) further proposed
culate' to include material Composed of particles that that correspondence between deformation structures in
are stuck together such that they cannot move indepen- the Green Tuff and those in the Gran Canaria ignim-
dently of one another. We describe a particulate flow as brites indicates a common, post-depositional, origin.
'steady' when the rate of deposition at a fixed location Wolff and Wright's definition of 'rheomorphic tuff',
along the flow path remains constant; if the rate and/or viz. a tuff that has undergone deformation following
style of deposition at a fixed location change through emplacement and deposition ("secondary mass flo-
time, the flow is 'unsteady'. We use 'imbrication' to de- wage"), was founded on the premise that non-particu-
scribe clasts that are inclined with respect to the depo- late deformation postdates emplacement and deposi-
sitional surface (Bates and Jackson 1980), even where tion. We believe this should be reconsidered on two
they are supported by matrix and do not overlap. We counts.
refer to inclined fiamme as imbricated even though this Firstly, much of the Green Tuff has since been rein-
fabric may reflect a component of non-particulate de- terpreted as ignimbrite (Mahood and Hildreth 1986;
formation. Orsi and Sheridan 1984, 1986), as it was originally (Vil-
lari 1969, 1974). The deposit locally exhibits imbricate
fiamme (Orsi and Sheridan 1984), a feature which on
Previous ideas Gran Canaria is considered by both Schmincke and
Swanson (1967) and Wolff and Wright (1981) to form
In a seminal paper, Schmincke and Swanson (1967) during initial lateral emplacement. Secondly, Wolff and
presented evidence that flow structures in ignimbrites Wright's hypothesis of post-depositional slumping of
at Gran Canaria were formed by welding and laminar welded tuff is not the only possible explanation for the
lava-like flow during the final stages of emplacement of non-particulate deformation structures common in
formerly turbulent particulate pyroclastic flows. The high-grade tuffs. For example, lineations cited as diag-
idea of syn-emplacement deformation has been slow to nostic of "secondary" flow (Wolff and Wright 1981)
gain acceptance, despite supporting evidence from also form during initial deposition (as clearly demon-
studies of other deposits (e.g. Walker and Swanson strated by Chapin and Lowell 1979; Reedman et al.
1968; Chapin and Lowell 1979; Mahood 1984; Reed- 1987a).
man et al. 1987a; Trigilia and Walker 1986; Buesch and We do not question the possibility of non-particulate
Valentine 1989). A possible factor in this is that, since slumping and gravity sliding (rheomorphism sensu
the 1960s, the welding profiles of several large-volume Wolff and Wright 1981) after emplacement and deposi-
high-silica rhyolitic ignimbrites in the western US have tion; for example, this is demonstrated where a single
been interpreted as reflecting the length of time at fold clearly involves two or more discrete ignimbrites
which loaded parts of that profile remained at high (Hargrove et al. 1984; Branney 1988). However, we pro-
temperature. In other words, the welding profiles were pose that another non-particulate flow mechanism oc-
considered to reflect cooling and loading profiles of in curs during deposition of many high-grade tuffs. Our
situ ignimbrite sheets (Ross and Smith 1961; Smith proposal modifies ideas of Schmincke and Swanson
1960; Riehle 1973; Ragan and Sheridan 1972; Peterson (1967) and Chapin and Lowell (1979), in the light of
1979). An implication of this view was that structures improved understanding of transport (Valentine 1987)
reflecting welding-deformation must have formed after and depositional mechanisms of particulate flows (e.g.
deposition of the entire ignimbrite. Workers apparently Lowe 1982, 1988; Wilson 1985; Fisher 1990; Druitt
regarded this as 'normal'. Ragan and Sheridan (1972) 1992).
even suggested that the structures described by
Schmincke and Swanson (1967), such as rotated lithic
507

Progressive deposition versus en masse deposition: a 1981; Cas and Wright 1987). Wright and Walker (1981)
preliminary discussion suggest that this is incompatible with turbulent em-
placement, but this objection follows only if the mas-
Ignimbrites (especially 'conventional' high-aspect ratio sive part of the ignimbrite (e.g. Wright and Walker's
ignimbrites; Walker 1983) are commonly regarded as "flow unit") formed by en masse deflation. We prefer
high concentration non-turbulent particulate (pyroclas- to interpret the zonation as indicating that ignimbrites,
tic) flows that stopped and deflated en masse (e.g. including massive layers, aggrade progressively at the
Sparks 1976; Wright and Walker 1981; Fisher 1986; base of a flow that is continuously replenished at
Carey 1991). This interpretation arose since Sparks source by an eruption whose magma composition
(1976) attributed poor sorting in ignimbrites to high changes with time. Such progressive aggradation (as en-
particle concentration in poorly expanded flows, rather visaged by Fisher 1966) is supported by the occurrence
than to turbulence in highly expanded flows. Steep- of a widely correlated subtle textural stratification, de-
sided pumiceous levees and distal lobes of small-vol- fined by vertical changes in sorting, within the compo-
ume ignimbrites were cited (Wilson and Head 1981) as sitionally zoned upper "flow unit" (massive layer) of
indications that pyroclastic flows have high yield the Acatlan ignimbrite (Wright and Walker 1981). This
strengths. Hence, a pyroclastic flow body was envis- stratification is difficult to reconcile with plug flow and
aged as undergoing laminar or plug flow (Sparks 1976; en masse deflation, and we interpret it as reflecting
Wright and Walker 1981) before stopping en masse, in slight flow unsteadiness during progressive aggrada-
a way similar to models of debris flows (Johnson 1970), tion.
to produce a massive unit (layer 2 of Sparks 1976) with 2. Instantaneous en masse deposition of a massive
an inverse-graded base (layer 2a). poorly sorted ignimbrite layer from a non-cohesive,
However, ignimbrites are deposits, not flows, and poorly sorted and high-concentration particle disper-
their morphology and sorting characteristics incom- sion is unlikely because the necessary displacement of
pletely reflect the transport regime. We propose that the interparticle fluid (gas and entrained fines) cannot oc-
bedding and sorting characters of an ignimbrite reflect cur instantaneously. Rapid depositional collapse (defla-
mainly the depositionalprocesses in dense basal parts of tion) will cause partial fluidisation (fluid-hindered sedi-
a flow and only to a lesser extent the transport proc- mentation) within the concentrated basal part of the
esses of the entire flow. For example, although a densi- flow, aiding grain support and lowering internal fric-
ty-stratified high-concentration flow might be exten- tion, thus allowing upper parts to flow farther than
sively turbulent, the turbulence would not be directly lower parts.
involved in forming the resultant poorly sorted deposit 3. Without deposition at the flow base there is an unre-
because deposition occurs from basal parts of the flow solved three-dimensional space problem inherent in a
where turbulence is suppressed by high particle con- flow that passes laterally from laminar behaviour to
centration, steep concentration gradients, and lower form a plug flow as depicted by Wright and Walker
velocities due to frictional drag. We envisage that the (1981, figure 12).
lower part of this depositional regime undergoes lami- 4. It is difficult to envisage how a sheet-like plug flow
nar flow and that material is supplied to it from an could travel and then stop almost instantaneously
overriding, relatively turbulent part of the flow (the across its entire extent (in many instances several 100s
transport regime), whose particle concentration proba- of km2), irrespective of topographic irregularities, with-
bly decreases more or less continuously upwards from out undergoing local extension and/or compression.
the high-concentration near the base. Thus, irrespective How can proximal parts and distal parts of a plug flow
of welding, vertical variations in many ignimbrites (and simultaneously know when to stop? Extensive high-
in surge deposits; see Valentine 1987) probably reflect yield-strength plugs will inevitably develop irregular
secular changes in depositional (e.g. boundary layer) upper surfaces, buckles, thrusts (ramps), auto-intru-
processes, rather than the transport processes (e.g. by sions, and pull-apart structures, yet such structures are
instantaneous en masse deposition). Reasons for our not characteristic of (non-rheomorphic) ignimbrite
proposing an alternative model of ignimbrite deposi- sheets (cf complex compressional and extensional
tion are given below. structures of Bingham lava flows, soft-sediment slides
and cohesive debris-flow deposits; Nemec 1991, p 61).
Low aspect-ratio ignimbrites (Walker et al. 1980)
Problems with en masse deposition contain facies known as 'ignimbrite veneer deposits'
whose field relations and stratification have already
1. Vertical chemical zonation in ignimbrites (e.g. Hil- been recognised as incompatible with an origin by en
dreth 1981; Druitt and Bacon 1986) is most simply in- masse 'freezing' of an entire pyroclastic flow body
terpreted as recording incremental deposition. Multiple (Walker et al. 1981; Wilson 1986). Such facies have
flow units might be invoked to reconcile the zonation been interpreted as a "tail" deposit left in the wake of a
with the model of en masse deposition, but vertical highly energetic pyroclastic flow, where a relaxation of
changes in chemistry commonly do not coincide with shear stress favoured deposition of the basal part of the
unequivocal flow unit boundaries, and vertical chemi- flow (Walker et al. 1980; Walker et al. 1981), and as
cal zonation within individual flow units (layer 2) has "the 'skin' of the pyroclastic flow" (Wilson 1986). In
been reported (e.g. Sheridan 1979; Wright and Walker this paper we suggest that incremental deposition from
508

a passing flow (Fisher 1966) is a general case with most ing transport regime, which may surmount slopes, it is
ignimbrites, including the massive layer 2b deposits, the more likely to respond to small topographic irregulari-
'normal ignimbrite' of Walker et al. (1981 p. 416), and ties, for example draining back downslope during ag-
high aspect-ratio ignimbrites. We believe that there is gradation. Furthermore, at any single location, the flow
little evidence that pyroclastic flows stop en masse to direction of a depositional boundary layer may change
form ignimbrites. with time as aggradation modifies topographic irregu-
larities (infills hollows), and as the transport regime's
main stream axis ('talweg') migrates laterally and twists
Progressive aggradation from a sustained current
from side to side during passage of a sustained flow. In
We think an ignimbrite, including massive parts (layer the Upper Bandelier Tuff, New Mexico, flow directions
2), records progressive aggradation from a sustained determined by imbricate AMS ellipsoids at individual
current. The current can be anything from a relatively locations change with height in tuff that shows "no vis-
short-lived, highly unsteady current that rapidly depos- ible lithologic layering or boundaries between layers"
its a thin ignimbrite as it passes by (in some cases prior (MacDonald and Palmer 1990). We interpret the
to lofting), to a more steady current that persists for the changes, evidently occurring within a single flow unit,
duration of a continuous column-collapse eruption as evidence for progressive aggradation of massive ig-
(possibly several hours; e.g. see Sparks et al. 1978) and nimbrite and for deposition from a boundary layer
which deposits a thick ignimbrite incrementally as it whose flow direction changed with time during sus-
passes by. In all cases the ignimbrite is deposited by tained flow. Similarly, AMS ellipsoid orientations vary
progressive aggradation; stratification reflects changes with height in the massive layer 2 of the Peach Springs
in flow steadiness and in the material supplied at Tuff near Kingman, Arizona, (Hillhouse and Wells
source. We view the model in which a pyroclastic flow 1991). Valentine et al. (1989) interpret the layer as a sin-
is a discrete morphological entity (head, body and tail) gle flow unit. Such vertical changes in AMS ellipsoid
that deposits mainly when it stops en masse, producing orientation are akin to grain fabric orientations that
a massive layer 2, as representing only an end-member change with height within individual massive (Bouma
of extreme flow unsteadiness. A) divisions of turbidites (e.g. Parkash and Middleton
1970), which we also think are deposited by progressive
aggradation. Changes in flow direction during progres-
Grain fabric evidence for the progressive aggradation sive aggradation probably contribute to frequently re-
model ported within-site scatter in grain fabric and AMS
Directional fabrics in ignimbrites, such as lineated and orientation in ignimbrites (e.g. Seaman et al. 1991).
imbricated crystals, lithics and pumice, and anisotropy With respect to the plug-flow model of emplace-
of magnetic susceptibility (AMS) ellipsoids, have often ment, it seems unlikely that the massive part of an ig-
been used to infer palaeocurrent directions of pyroclas- nimbrite (layer 2b) would contain directional grain fab-
tic flows (e.g. Knight et al. 1986; MacDonald and rics indicative of flow at the site of deposition if it trav-
Palmer 1990, and references therein). However, direc- elled and stopped as a non-shearing plug. A plug flow
tional fabrics in ignimbrites also provide evidence of is unlikely to revert to late-stage laminar flow because
progressive aggradation (and of P-NP transitions in a an increase in shear stress is required to shear a pre-
depositional boundary layer; see section on 'Direc- viously non-yielding (Bingham) plug, not a decrease as
tional deformation fabrics formed during agglutination' occurs during waning flow. However, directional fab-
below). rics are common in massive (non-rheomorphic) ignim-
Directional fabrics in ignimbrites can be formed by brites (Elston and Smith 1970; Ellwood 1982; Knight et
(1) sedimentary alignment of elongate particles as they al. 1986; Potter and Oberthal 1987; Ui et al. 1989; Mac-
deposit; (2) shear deformation in the depositional Donald and Palmer 1990), even within upper pumice-
boundary layer and non-particulate part of a flow; and rich parts of flow units (e.g. figure 14a of Sparks et al.
(3) post-emplacement slumping or sliding. Sedimentary 1985; Mimura 1984). Central portions of some massive
alignment of particles during deposition is evident from ignimbrite units do have poorly developed or isotropic
grain fabrics in nonwelded and incipiently welded tuffs fabrics (Mimura 1984), as do some massive turbidite
(e.g. Kamata and Mimura 1983; Mimura 1984; Potter divisions (Hiscott and Middleton 1980), and this m a y
and Oberthal 1987). The fabrics are analogous to those indicate reorientation of particles by escaping fluid
in turbidites (Tiara and Scholle 1979; Hiscott and Mid- during static deflation. The intensity of a directional
dleton 1980) and to those produced in experiments in- fabric probably also depends upon the amount of shear
volving sedimenting particulate dispersions (Rees and imparted on a body of particles during their deposition.
Woodall 1975). The amount of shear is controlled by the shear gradient
Kamata and Mimura (1983) interpret directional within the depositional boundary layer, and by the rate
fabrics as recording "late stage laminar flow", but we of aggradation, which affects the length of time any
interpret them as due to laminar flow within the sus- population of particles resides within the depositional
tained depositional boundary layer of a stratified flow. boundary layer; these change during unsteady flow.
A depositional boundary layer need not always or en- Where AMS and grain fabrics are used to infer pal-
tirely flow in the same direction as the (overlying) aeoflow direction, it is important to note that they
transport regime. Unlike the less-dense and faster-mov- probably record only the last increment of shear (depo-
509

sition), and they may not represent the transport direc-


tion of the main flow body. Because flow of the deposi-
tional regime is substantially influenced by gravity, one
Y
can expect the last increment of movement commonly
to be downslope, even though the actual distance of
transport in this direction have be extremely limited
(Suzuki and Ui 1982 p 431). 0
(].)
_r- "0

Grain sorting and grading in the depositional regime i--


time J, time D,
We propose that the sorting character of ignimbrites re-
flects high particle concentrations within near-deposit- A: stepwise aggradation B: gradual aggradation
ing basal parts of a stratified particulate gravity flow
(Fisher 1966; Lowe 1982) and the effects of a velocity Fig. 2. Two types of aggradation at the base of a particulate flow.
Stepwise aggradation (A) results from successiveen masse freez-
gradient where this concentrated basal dispersion is ing of traction carpets or traction plugs. Gradual aggradation (B)
sheared due to drag against substrate. Irrespectlve of results from direct suspension sedimentation. Stepwiseand grad-
whether the higher parts of a high-concentration flow ual aggradation may occur in both hot and cold particulate flows.
are laminar, turbulent, expanded, homogeneous or den- No relative scales are implied
sity-stratified, the basal (depositional) part is likely to
undergo laminar flow. 1. Stepwise aggradation, by successive en masse depo-
In our scenario, ignimbrites are analogous to mas- sition of successively formed traction carpets or trac-
sive or normal-graded poorly sorted sedimentary units tion plugs. Each carpet or plug develops as the basal
with inverse-graded bases deposited from the base of part of the flow acquires yield strength due to friction
high-density turbidity currents ($3 and $2 of Lowe and/or adhesion between particles. Each aggradational
1982). Poor sorting reflects rapid deposition from a step will tend to produce an inverse-graded layer (layer
high-concentration depositional boundary layer. In 2a of Sparks 1976), analogous to the $2 division of
massive parts of ignimbrites, inverse coarse-tail graded high-density turbidity current deposits (Lowe 1982) de-
pumice records pumice buoyancy within the dense dep- spite the fluids being different.
ositional boundary layer. Normal (coarse-tail) grading 2. Gradual aggradation, by rapid suspension sedimen-
of lithics records changing lithic supply to the deposi- tation of individual grains to form a massive deposit
tional boundary layer, due to waning flow (with conse- (layer 2b of Sporks 1976) analogous to the $3 division
quent upstream migration of lithic transport limits) and of high-density turbidity current deposits (Lowe 1982).
possibly also due to decreased lithic supply from a con- Gradual aggradation of a massive layer occurs during
duit in which wall-rock erosion diminishes and from a fairly steady flow in which the base of the flow is suffi-
landscape that is progressively buried by aggrading ciently concentrated to develop a zone of fluid-hin-
tuff. We believe that large and dense lithics can be dered deposition adjacent to the substrate surface. Be-
transported considerable distances by combinations of cause turbulence is suppressed in this zone, grains at
buoyancy-aided saltation, rolling, and partial entrain- the flow base (immediately above the substrate) will de-
ment in the top of the depositional boundary layer, be- posit, there being insufficient underlying fluid (gas)
fore they finally become trapped and stop at the base of source to provide lift or to prevent frictional deposi-
the depositional boundary layer. Inverse-graded ('bas- tion. This mechanism of deposition, gradually up from
al') layers record dispersive forces mostly affecting the base, is similar to that envisaged for deflating lique-
larger clasts within shortlived, poorly sorted carpets of fied and fluidized sediment gravity flows (Lowe 1976
ash and lapilli driven along the substrate surface by p 78).
traction (cf $2 of Lowe 1982). Unsteady flow can pro- Fluctuating input to the depositional regime from
duce a variety of grading patterns, by fluctuation of the overriding part of the flow is likely to cause alterna-
concentration and velocity gradients in the depositional tion between gradual and stepwise progressive aggrada-
boundary layer. In our experience, the majority of ig- tion and produce a deposit comprising several layers.
nimbrite flow units show departures from smooth Each layer might conventionally have been interpreted
coarse-tail grading profiles; stratiform clast concentra- as a layer 2 of Sparks (1976) (for example, see Wilson
tions or grade-trend reversals, which may be localised 1986) but they all accumulated during the passage of a
or laterally persistent, are very commonly present (e.g. single flow (see Fig. 3).
Fisher and Schmincke 1984 figure 8-18). Details of
depositional mechanisms in stratified flows will be dis-
cussed further in another paper (Druitt, Valentine and Particle agglutination and coalescence during
Branney). emplacement of silicic tufts

Stepwise and gradual types of progressive aggradation If particles are sufficiently hot and fluidal they weld al-
most immediately on contact, prior to, and therefore ir-
Deposition within the laminar boundary layer may oc- respective of, post-emplacement cooling and loading
cur in two main ways (Fig. 2): (Mahood 1984). In this paper we use the term aggluti-
510

Fig. 3. Multiple flow units or the prod-


uct of progressive aggradation at the
base of a single (unsteady) flow? Layer-
ing in the Minoan ignimbrite (Bond and
Sparks 1976), south of Acrotiri, Santo-
rini

nation to refer to virtually immediate welding. This is cause(s) for such low viscosity (such as possible high
distinct from post-deposition, post-emplacement weld- eruption temperature, strongly peralkaline chemistry,
ing of higher-viscosity shards under the influence of a high halogen content, eruptive volatile contents far
deposit's in situ cooling rate and load pressure. Our use higher than those preserved in the tuff) may not be
of the term 'agglutination' does not imply any particu- known, the field evidence suggests that when interpret-
lar mode of emplacement (such as by airfall). If par- ing high-grade and extremely high-grade tuffs it is
ticles are even more fluidal (droplets) they may rapidly inappropriate to think in terms of rheologies similar to
form a homogeneous liquid in which remnant particle those of 'typical' obsidian lavas and domes that were
outlines are obliterated; we refer to this as coalescence erupted in an entirely different manner, i.e. at low rates,
(see also Bristow 1962; Ekren et al. 1984). Thus, spat- and after considerable degassing (Jaupart and All~gre
ter-fed lavas form through coalescence. Coalescence, 1991). For example, globule-shaped particles that rep-
agglutination and post-emplacement welding are inter- resent droplets whose viscosity and yield strength were
gradational processes in a continuous spectrum of de- overcome by surface tension effects during a brief pe-
creasing welding rate and effectiveness. The termino- riod of transport occur in high-grade silicic ignimbrites,
logy applies equally well to deposits of airfall, flow, e.g. at Menengai (Bristow 1962; Schmincke 1972; Leat
surge, and intrusive origin. The spectrum occurs be- 1985), Mt Suswa (Johnson 1968; Schmincke 1974; Hay
cause, as temperatures decrease and volatiles exsolve, et al. 1979), Longonot (Scott 1980), Pantelleria and
the viscosity and strength of hot glass change gradually Gran Canaria (Schmincke 1969, 1974), and they occur
rather than sharply (Shaw 1963; Ryan and Blevins in rhyolitic sheets interpreted as high-temperature lava
1987). flows (Bonnichsen 1982, p. 310). Droplet-shaped par-
Agglutination upon initial contact requires low par- ticles are not characteristic of low-grade silicic tuffs or
ticle viscosity and yield strength. Silicic magmas are vent-fed viscous obsidian lavas, but droplets do charac-
generally more viscous than basic magmas, but field terise Hawaiian magmatic fountaining of basalt (e.g.
studies show clearly that many peralkaline silicic mag- MacDonald 1972). On Hawaii, most erupted fluidal
mas (Bonnichsen and Kauffman 1987; Hausback 1987; basalt particles agglutinate to form spatter or coalesce
Mahood 1984) had much lower viscosities during erup- to form fountain-fed lavas, but some chill and solidify
tion and emplacement than is perceived as typical for in the air to be preserved as droplet-shaped particles of
silicic lavas. Furthermore, some non-peralkaline silicic glass ('achneliths' of Walker and Croasdale 1972). The
high-grade and extremely high-grade tuffs similarly ex- similarity of the basaltic droplet particles to those asso-
hibit lithofacies indicative of 'unusually' low magmatic ciated with high-grade silicic ignimbrites indicates simi-
viscosities, e.g. the Wall Mountain Tuff (Chapin and lar viscosities during eruption: i.e. we can infer that ir-
Lowell 1979), the Pitts Head Tuff (Reedman et al. respective of their chemistry, eruptions that form high-
1987b), the Crinkle Tuffs (Branney et al. 1992), the grade and extremely high-grade ignimbrites produce
Weolseong Tuff (Reedman et al. 1987a), the Buckshot particles sufficiently fluidal to agglutinate or coalesce
ignimbrite (Henry et al. 1989), the Moat rhydite (du on impact. This contention is supported by occurrences
Bray and Pallister 1991), and tuffs in the Owyhee of high-grade silicic tuffs that rest upon slopes steeper
Mountains (Ekren et al. 1984). Even though the than 34-38 ~ response angles for loose particles. For ex-
511

ample, the Green Tuff of Pantellefia is plastered onto ity impacts), where particles remain in contact for
near-vertical slopes (Orsi and Sheridan 1984), and the longer intervals, and where there is minimal cooling by
Egoshura Tuff in Kenya locally occurs on 70 ~ topo- ambient fluid. Agglutination at the flow base will be en-
graphic slopes (Leat 1985). The virtually immediate hanced by the insulating effects of the first-deposited
welding that occurs in agglutination is also indicated by hot tuff and the overriding hot particulate flow, and by
rapid and complex vertical variations in degree of the load exerted by the overriding flow. In a low-con-
welding, slope-parallel rather than horizontal welding centration flow (pyroclastic surge), conditions most fa-
zones, welding right to the upper surface of a deposit, vourable for agglutination occur at the aggradation sur-
and occurrences of spiracles (Chapin and Lowell 1979) face, in any traction carpet immediately above it, or just
that resemble gas pipes in flood basalts. These are typ- below the aggradation surface. In a flow of higher con-
ical features of high-grade silicic tuffs as well as of near- centration (pyroclastic flow) agglutination may occur in
vent deposits o f basaltic magmatic explosive eruptions. four situations: (1) within the depositional boundary
Mahood (1984) presented field observations and vis- layer during gradual aggradation of a massive bed; (2)
cosity calculations of high-grade tuffs to suggest that within either a plug-like dense particulate layer (trac-
agglutination can occur in pyroclastic flows as soon as tion plug) separated from the aggradation surface by a
deflation allows significant contact between pyroclasts. shear layer, or within a shearing dense carpet, during
It is perhaps worth noting that the presence of lithic stepwise aggradation; (3) at the aggradation surface;
and crystal-enriched elutriation pipes in a welded ig- and (4) just below the aggradation surface.
nimbrite indicates that welding was mostly post-deposi- Agglutination or coalescence, therefore, may occur
tional at the point in the deposit where they occur. in the presence or in the absence of laminar (simple)
However, high-temperature defluidisation experiments shear. In the presence of laminar shear, sticky grain in-
with sticky particles show that 'gas chimneys' are pro- teractions between hot viscous particles give way to
duced when a fluidised suspension collapses at initial shear within viscous agglutinated particles. Any load
sintering temperatures (Gluckman et al. 1976). Elutria- exerted on the sheafing layer by the overriding flow
tion pipes should not be confused with steam spiracles will aid agglutination of coalescence as sticky p o i n t
(Chapin and Lowell 1979) which are hollow and form contacts of neighbouring particles are driven into one
when agglutinated tuff has assumed the rheological another, increasing contact ares. The resultant deposit,
properties of a lava. including any vesicles that grew during shear, will be
Coalescence of silicic particles is recorded by spat- either lineated (prolate uniaxial fabrics) or, more com-
ter-fed silicic lavas (Baker 1976; Duffield 1990). monly, lineated and foliated (triaxial strain fabrics).
Where agglutination of coalescence occurs in the ab-
sence of laminar shear (as within a traction plug or in
Agglutination at and near the base of some circumstances below the aggradation surface) it
high-concentration particulate flows happens in much the same way as for agglutinated air-
fall deposits (Sparks and Wright 1979; Mahood 1984)
If agglutination occurs in a stratified particulate flow, it and the resultant deposit on a horizontal surface will
is least likely in higher turbulent parts, because high- tend to have a purely compactional (pure shear) flatten-
velocity impacts do not favour agglutination (Freundt ing fabric (oblate uniaxial ellipsoid) and no lineation at
and Schmincke 1990) but cause disruption and/or this stage. Where agglutination or coalescence occurs
bouncing, and because fewer particle collisions occur within a moving particulate plug, the plug may con-
where particle concentration is lower. Statistically, a tinue to move as a non-particulate plug before finally
few collisions will inevitably result in some agglutina- freezing to the substrate (causing the aggradation sur-
tion above the depositional regime, but this will not be face to step upwards).
the dominant agglutination site. Any resultant particle Thus, a lineated or non-lineated basal deposit of
aggregates will have higher settling velocities than indi- welded tuff can form well before the bulk of the flow
vidual particles and thus tend to settle towards the dep- comes to rest; this is evident in basal eutaxitic zones of
ositional regime relatively rapidly. If particles are so lava-like tuffs (e.g. Schmincke 1969; Henry et al. 1989;
hot and fluidal that they do agglutinate or coalesce Branney et al. 1992), sheared basal deposits with imbri-
readily in high parts of a particulate flow, the flow is cate fiamme in rheomorphic ignimbrites (e.g. on Gran
liable to collapse (cf agglomeration in high-temperature Canaria; Schmincke and Swanson 1967), and eutaxitic
fluidisation experiments of Gluckman et al. 1976), pos- vitrophyres with or without lineations or flow folds at
sibly forming a spatter-fed lava. the base of rheomorphic ignimbrites. A thin basal zone
In comparison with cuspate shards of similar size, of lower-grade tuff may result from syn-emplacement
the droplet shape of particles enhances agglutination or cooling within the head of a flow or beneath the front
coalescence, because droplets are both slower to cool part of a flow if cool air is overriden, or from initial
(minimum surface/volume) and more difficult to sup- syn-depositional cooling at the base of a flow due to
port during transport (larger settling velocity) (Wilson cold substrate. Such lower-grade tuff will pass upwards
1980). into agglutinate.
Agglutination is most likely to occur around the base
of a flow, where particle concentration is highest,
where particle collisions are non-disruptive (low-veloc-
512

Directional deformation fabrics formed during


agglutination

For sub-horizontal ignimbrites, sedimentary grain fab-


rics subject to loading-compaction (pure shear) will ro-
tate towards the horizontal. Imbrication of attenuated
fiamme has long been recognised as forming during
emplacement (Schmincke and Swanson 1967, Wolff A
and Wright 1981; Orsi and Sheridan 1984). Loading-
compaction subsequently imposed on the fabric will
lower the angle of imbrication until the fiamme become
sub-parallel to bedding. Continued rheomorphism 9

(simple shear) will also tend to shear out any early


formed imbrication. Therefore,/f welding was entirely
post-emplacement, imbrication would be less obvious
in intensely welded ignimbrites than in non-welded ig-
nimbrites. However, imbrication is generally more ob-
vious with 'flattened' fiamme in welded ignimbrites I -'~ -" ~ " ~ ~-~_~"
(e.g. Figs. 4 and 5) than it is with pumice clasts in non- B $ ~ " ~~- ~--~.,-'~..-',..~'-~-"
welded ignimbrites. Indeed, imbrication in ignimbrites
was first described from thin fiamme in high-grade ig- Fig. 4A, B. Imbrication of attenuated fiamme (A) throughout a
nimbrites (Schmincke and Swanson 1967). Long axes of cooling unit, e.g. units C and D, Mogfin Formation (Schmincke
fiamme clearly dip upflow even where axial ratios (X~/ 1990), Crinkle Tufts (Branney et al. 1992), and (B) persisting
through grain-size layers, e.g. TL, Mogfin Formation. Lower im-
X3) greatly exceed those characteristic of undeformed brication angles towards the base of a record higher shear strain
pumice clasts. In addition to a (loading-modified) sedi- than higher up in the unit; the attitude of fiamme should not be
mentary alignment of solid particles, such imbrication confused with ramp structure defined by flow layering in lava
clearly indicates a component of syn-depositional shear flows
deformation of viscous ellipsoidal particles in a viscous
fluid (non-particulate flow), where long axes of par-
ticles deform and rotate towards the shearing direction cally persist upwards from one massive layer to the
(Gay 1968) and assume an orientation oblique to bed- next, right across inverse-graded fine-grained layers
ding at low strains. Hillhouse and Wells (1991) demon- (Figs. 4B and 5) and other sedimentary layering (e.g.
strate that imbrication angles of AMS ellipsoids in the lithic-rich layers).
Peach Springs Tuff, Arizona, are not significantly re- Our notion of syn-agglutination imbrication during
duced by welding-compaction, and infer that laminar progressive aggradation is very similar to models pro-
flow and compaction were simultaneous. As MacDon- posed for spatter-rich pyroclastic flow deposits at Vul-
ald and Palmer (1990) point out, what has commonly sini (Onano Pyroclastic Formation, Trigilia and Walker
been regarded as a compaction foliation is really a 1986; Marsella et al. 1987) and on Santorini (Mellors
compaction-transposed flow foliation. and Sparks 1991). Like Trigilia and Walker (1986), we
Thus fiamme imbrication must record rather low suppose it to be common. However, we believe that
(momentary?) shear strain in a body of agglutinating fluidal clasts and agglutination are not essential to pro-
particles either within the depositional boundary layer gressive aggradation from a depositional boundary
or just below it. We propose that during hot progressive layer, but that they simply record rather well, via non-
aggradation a zone of shear-deformation passes up- particulate deformation structures, a process that oc-
wards with the aggradation surface (Fig. 6), in some curs during emplacement of most ignimbrites.
cases only affecting an individual layer of agglutinating
particles momentarily. This model predicts that imbri-
cation of attenuated fiamme (as with sedimentary par- Evolution of the non-particulate flow component
ticle alignment) should occur in massive ignimbrite well
above the basal layer. Although apparently not pre- Wolff and Wright (1981, p. 31) considered that pyro-
viously documented, many high-grade ignimbrites do clastic flows stop suddenly en masse, and thus argued
show imbrication of attenuated fiamme well above that non-particulate flow structures would be unlikely
their bases. For example, several high-grade ignimbrites to form during initial ignimbrite emplacement. Our
on Gran Canaraia (e.g. units C and D of the Mogfin model involves continuous agglutination to form a non-
Formation; Schmincke 1990) and in the English Lake particulate flow component that progressively aggrades
District (Crinkle Tuffs at Langdale; Branney et al. beneath a sustained particulate flow (Fig. 6). Boundary
1992) show imbrication throughout most of their thick- layer agglutination and deposition may continue as
ness, even where they lie on sub-horizontal surfaces long as the transport regime of the flow supplies hot
(Fig. 4A). Of course, in massive units imbrication far particulate material to the depositional regime. As the
above the base can be difficult to see. However, within non-particulate layer thickens (aggrades), such that it is
some cooling units, pervasive imbrication fabrics lo- no longer chilled by the ground surface, the zone in
513

~ F

- ~ . . s ...: -,-~.. .__~'- ~ '-~- :'>." ".:.


, 99 _Z_ . . . _ ~ . - - r = - - - - . - ~ . - _~ -_- : .~__.-__--'--- ~

- - " ' " ' ": " ;- " I/Ji/

Fig. 5. Left: A spatter-rich pyroclastic flow deposit, Onano Pyro- rection of the depositional regime. In the Onano case the fiamme
clastic Formation (Marsella et al. 1987); NB: top and base not are clasts transported as poorly inflated flattish rags or ribbons.
shown. Right: Unit D, Moghn Formation (Schmincke 1990). Each Although clast shape generally was modified by shear during dep-
deposit shows imbrication of fiamme, both within massive layers osition, the fiat clasts also occur with a sedimentary imbrication in
(which we do not interpret as individual flow units) and where local non-welded facies of the deposit, showing that fiamme do
stratification (S) occurs. Arrows indicate inferred palaeoflow di- not necessarily reflect in situ compaction or shear

which non-particulate deformation occurs may move in Where the particulate to non-particulate (P-NP)
a laminar (and possibly plug-like) fashion, due to inher- transition involves agglutination as opposed to coales-
itance of momentum, frictional coupling with the over- cence, clast outlines will still be discernible in the non-
riding particulate part of the flow, and gravity-induced particulate flow, although they may be highly atte-
shear. The stratified flow then may be considered to nuated. Many flow-folded, lineated and foliated high-
have developed into two main components: a non-par- grade tufts (e.g. those figured in Schmincke and Swan-
ticulate component and an overriding particulate com- son 1967; Chapin and Lowell 1979; Wolff and Wright
ponent. The proportion of accumulated non-particulate 1981; Branney et al. 1992 figure 5a and 5b) may be o f
material that flows at any one time depends on rheo- this origin. Where the P-NP transition involves coales-
logy and shear stress, and probably varies from a thin cence, the non-particulate part of the flow forms a rock
zone (mm-cm) near the aggradation surface in cases o f identical to that of a lava flow; it will not exhibit rem-
moderate to high-grade ignimbrites, to the entire thick- nant shard textures and it may develop vesicles and
ness of aggraded non-particulate material (several even a trachytic texture. Similarly, fountain-fed basalt
metres) in some extremely high-grade ignimbrites. Our lavas widely do not record their particulate origin in t h e
model is similar to that of Chapin and Lowell (1979), form of vitroclastic textures, and obsidian lavas infer-:
but it differs from that o f Schmincke and Swanson red by some workers to have formed from silicic foam
(1967) in which agglutination occurs throughout the that collapsed during ascent and extrusion (Eichel-
body of an extremely unsteady flow as it deflates and berger et al. 1986; Manley and Fink 1987) generally do
stops en masse. not record origin from a foam in the form of remnant
514

i:

'
,
=
~ '
~
~
o
o DENSTY_GRADI~D.
PARTICULATE '
' o

'
~ ~
" ~
9 ~
*
" -
L
"~
,'.',
~.
'
suggest that the competence of shearing non-particulate
parts of the flow can be affected by preferential volatile
exsolution along shear surfaces. Gas-filled, gas-lubri-
cated shear planes or laminae may form in which the
gas is at a pressure sufficient to support the weight of
the overlying flow (Noble 1968). Competent lenses can
form where depletion of dissolved gas (by shear-in-
m'~ .'~ m'~ ~ i~'** ~ : "~.'o~ "'~ ~" *~ *'~ o'~'. ~176
~ '~176 ~*:~'2-*:
ZONE O F duced exsolution?) and associated formation of fine
..... ~ . . . . . . . ~"~~ ~ ~*"~'*~ . . . . . :Ve~176176176 ~%~AGGLUTINATION
9 ~, ~,~g'~.*,t,,*~'..e~o~,o~e~ ~' o ~ ~ '~.:.~'~. AND/OR foam locally increase viscosity and yield strength.
Where the weight of the overlying flow is great, gas-
AGGRADATION . . . . . NON.PARTIOULATE~'~Z.~ ~ \
lubricated shear laminae may compact to form parallel
SURFACE m'~ ~ " ~ ~" FLOW'COMPONEN'F~..~
~ ~ ' . ~ " ~ - ;~'~_~
t > partings (Branney et al. 1992). In many high-grade tufts
CHILLED SUBSTRATE
it is impossible to determine whether a highly lenticular
BASAL VITROPHYRE fiamme represents a flattened pumice clast, a flattened
Fig. 6. A schematic cross-section showing emplacement and pro- non-vesicular spatter lump, a rag whose pre-deposition
gressive aggradation of an extremely high-grade ignimbrite. No shape was ribbon-like, or a zone in which welded ma-
relative scales are implied. Hot particles (viscous droplets) agglu- trix material has undergone vesiculation (with or with-
tinate and/or coalesce within and just below a depositional out collapse) during shear. For this reason, profiles of
boundary layer that migrates progressively upwards from on the
substrate to a position between non-particulate and particulate fiamme aspect ratio and density profiles in many high-
flow components. The thickness of the non-particulate flow com- grade ignimbrites do not simply record welding intensity,
ponent at any instant may vary from 0-100% of the thickness of and they should be interpreted with caution. Unfortu-
the aggraded non-particulate material. The asterisk indicates the nately, high-grade ignimbrites are particularly difficult
boundary between the non-particulate flow component and un- subjects for strain analysis because the variously col-
derlying stationary tuff at an instant in time; its vertical position lapsing and revesiculating clasts and matrix cause the
will vary with time nature and degree of competence inhomogeneity to
vary irregularly through time and along with the vol-
pumiceous textures. However, because emplacement of ume of any potential strain markers can change in vol-
the non-particulate flow component initially involves ume.
particulate lateral transport, its aspect ratio (thickness/ The non-particulate layer is liable to autobrecciate
flow length) is not purely a function of its rheological like a lava flow during its emplacement. Rigid pumi-
properties, slope and eruption rate, as is the case with a ceous lenses, formed by continued vesiculation in
flow that erupted as a lava directly from the vent. The shearing layers, may crumble during shear to form sec-
non-particulate flow component may derive much of its ond-generation shards (cf Lentia rhyolites, Vulcano;
momentum from the overriding particulate component Pichler 1981). Movement of the non-particulate layer
of the flow via new additions and frictional coupling, stops through ponding or when its yield strength ex-
but as it thickens it will become increasingly indepen- ceeds the applied shear stress. This can arise from topo-
dent of the particulate part of the flow. graphic effects, volatile exsolution (itself possibly en-
hanced by shear), or cooling.

Rheology of the non-particulate flow component


Lithofacies associations resulting from flow detachment
From the conclusion that agglutination and coales-
cence require lower magma viscosities than are charac- Detachment of the particulate and non-particulate flow
teristic of conduit-fed rhyolite lavas and domes erupted components, or stopping of one component before the
slowly in a considerably degassed state (e.g. Fink and other, will inevitably result from developing differences
Manley 1987; Jaupart and All6gre 1991), it follows that in rheology. Some ways in which a stratified flow with
a non-particulate part of a stratified flow will, at least a non-particulate component might evolve are shown in
initially, have viscosity considerably less than such Fig. 7. Actual developments will be influenced by par-
rhyolite lavas. We envisage rheologies closer to those of ticle supply rate to the depositional boundary layer, to-
spatter-fed basalt lava flows. pography, cooling rate, and volatile exsolution rate.
Subspherical vesicles are abundant in flattened pu- Detailed consideration of the roles and interdepen-
mice, spatter and matrix of high-grade ignimbrites dence of such factors is beyond the scope of this paper,
(Gibson 1970; Schmincke 1974, 1990; Henry et al. but it is important to recognise that diverse lithofacies
1989). They indicate that agglutinate was still degassing associations will result from differing and divergent
after non-particulate flow ceased. Sites of vesiculation flow behaviour. Associations that might otherwise have
and the effects of volatile-induced reduction in magma been ascribed to multiple eruptive phenomena may
viscosity are considered further in a later section (see thus be interpreted more simply.
'Sites of volatile exsolution'). Large lenticular gas cavi- Figure 7 depicts only 'snapshot' scenarios. Consider
ties and highly pumiceous shear planes are also com- how each (Figs. 7A-C) would develop during an erup-
mon in high-grade ignimbrites (Schmincke 1974; tion that produces a sustained two-component flow. In
Chapin and Lowell 1979; Reedman et al. 1987a), and the case of Fig. 7A, for example, massive or layered
515

esced, compaction-welded, and non-welded layers. De-


posits formed in this way would resemble c o m p o u n d
cooling units (Smith 1960) with highly irregular weld-
ing-compaction profiles. They can include lava-like
layers, viscous shear layers, erosion surfaces, layers of
autobreccia, poorly welded layers, layers with imbri-
cated fiamme, and layers fused (sensu Christiansen and
A: NON-PARTICULATECOMPONENTSTOPSFIRST
Lipman 1966) by hotter adjacent tufts (Fig. 8) (cf
Walker and Swanson 1968; M a h o o d 1984; Henry et al.
1989). It is wrong to assume that sharp contacts in such
a sequence necessarily represent flow-unit or cooling
:~'.. ,i'~..;~,~!:.':.')'i sedimentatian ~disturbed beds unit boundaries; they may reflect changes in deposi-
i!.iii ~"a.: .... , [.... , ~ ~ autobreccia
tional mechanism due to flow unsteadiness, density in-
terfaces in the parent flow, or plug-bounding shear sur-
faces.
B: PARTICULATECOMPONENTCOLLAPSESBEFORE Rapidly stacked hot layers of different density and
NON-PARTICULATECOMPONENTSTOPS
competence will be particularly prone to soft-state
loading and auto-intrusion, and to development o f
pressure ridges, ramps, attenuation, extensional fractur-
ing or boudinage, lateral spreading, slumping, and
downslope gravity sliding (not indicated on Fig. 8).
-::< ...........::.............::: ~:; ......... .... l"'"al i " These will be especially p r o n o u n c e d if the hot tufts lie
"-1-:;~:~~.~.,...-..... 0 "
-_~-~: -~-- ~ . ...... ." 9 9 .
.- ...~-.;. .... ;,,-_. ..........
. .- " ~ =~-4-~--~-~- -..
" v .
9 " ' on an irregular topography. Even these 'secondary'
processes probably start whilst the ignirnbrite is being de-
posited, as aggradation increases the load on the first-
C: PARTICULATEFLOWCOMPONENTDEPOSITSDENSE deposited tuff. Layered ignimbrite formed in this way
MATERIALAND REMAINDERLOFTS
from a single flow might be mistaken for interstratified
Fig. 7. Three hypothetical 'snapshot' scenarios producing diverse thin lavas and tufts, or interstratified agglutinated fall
lithofacies associations following particulate to non-particulate and flow deposits (see discussion in M a h o o d 1984).
transitions in a stratified flow. No relative scales are implied. A Many ignimbrite eruptions change composition with
Topographic effects (blocking or ponding: Valentine 1987; Fisher time. Such changing inevitably affects viscosity, volatile
1990) or increase in yield strength with distance from source cause content, grain size, and density of particles in the pyro-
the non-particulate component of a flow to stop so that the over- clastic flow, and thus alters their welding behaviour, so
lying particulate component detaches. The detached particulate
flow surmounts topographic highs and leaves a veneer of tuff of that vertical successions of contrasting lithofacies result
lower grade than the non-particulate component. If increasing (e.g. Onano Pyroclastic Formation; Marsella et al.
yield strength is the cause, the non-particulate lithofacies passes 1987; Ignimbrite P1, Gran Canaria; Freundt, in
abruptly downflow into a non-agglutinated tuff across a steep au- Schmincke 1990).
tobrecciated lobate toe. B The particulate component of a flow
becomes unsteady and collapses before the non-particulate com-
ponent comes to rest. This produces a layer of lower-grade (non- Welded pyroclastic surge deposits?
agglutinated) ignimbrite, which becomes deformed as it rides
along on the top of the non-particulate flow. C The overriding It is noticeable to us that there are few literature refer-
flow component becomes sufficiently dilute by expansion of en- ences concerning welded pyroclastic surge deposits re-
trained air and particle sedimentation such that the residual tur- lative to those on welded fall and flow deposits. This
bulent component undergoes wholesale lofting. The high-grade could be taken to indicate that cooling in surges is so
deposit does not pass gradationally (vertically or laterally) into
lower-grade ignimbrite, although it autobrecciates in a manner rapid as to prevent welding. But agglutination or coal-
similar to lava flows, and forms steep lobate margins. The lofted escence in a pyroclastic surge will prevent particulate
ash component forms a widely dispersed, cold co-ignimbrite ash- traction, so typical sedimentary traction structures and
fall deposit, and some falls onto the high-grade tuff forming a thin sorting characteristics will not form, and the beds
(easily eroded) ash mantle with a sharp base. If airfall ash is de- would resemble fall or flow deposits. A corollary o f this
posited before the non-particulate flow stops, the mantle may be is that (like the occurrence of elutriation pipes in
deformed or mixed-in with blocks of autobreccia (cf Branney et
al. 1992; Henry et al. 1990) welded ignimbrites) the presence of typical surge bed-
forms and sorting in a welded deposit implies that
welding was post-depositional at that stratigraphic ho-
rheomorphic or lava-like tuff can continue to accumu- rizon.
late, attaining great thicknesses proximally or in topo-
graphic lows. However, flow is likely to evolve during a
Agglutination and coalescence on steep surfaces and in
protracted eruption, so these scenarios may evolve from
vents
one type to another. As the particle population sup-
plied at source varies, and as the particle supply rate to A close resemblance between the sorting, stratification,
the depositional regime fluctuates (flow unsteadiness), and shape fabrics of subhorizontal ignimbrites and
agglutinate layers can become interstratified with coal- those of tufts deposited onto steep to vertical surfaces
516
co-ignimbrite ash (non-welded fall and surge deposit) During lateral emplacement of the Wall Mountain
Tuff in the Gribbles Run palaeovalley (Chapin and
disrupted non-lineated agglutinate
Lowell 1979), tuff was agglutinated against steep to ver-
tical walls and developed non-particulate lineations
gradational contact (slumped) that generally point along the palaeovalley (alongflow).
lava-like tuff The degree of plunge of the non-particulate lineations
sharp contact fluctuates with increasing distance inwards from a
tuff fused by overlying coalesced tuff
point on the palaeovalley wall (our observation), which
poorly welded tuff
indicates that non-particulate parts of the flow accreted
weeded tuff (oblate fiamme)
successively during passage of the particulate flow. The
variety of (shear) lineation plunge-angles presumably
autobreccia reflects temporally varying resolutions of vectors due to
stream sloshing or surging in the particulate part of the
auto-intrusion of lava-like tulf flow, with those due to gravity acting on the non-parti-
culate part of the flow. Other geometric features of
loaded sharp contact
welding-deformation in the Wall Mountain Tuff that
are incompatible with an origin entirely by post-em-
foad ball (pseudonodule)
placement loading compaction followed by gravity
lava-like tuff slumping include a pervasive foliation that is com-
necking and boudinage monly not compacted around rigid equant lithic clasts
(in contrast to the 'wrapped' lithics that characterise
Ienticular zones of secondary vesiculation
tuffs in which loading compaction was the dominant
prolate fiamme welding deformation), common occurrence of uniaxial
imbrication across a sharp bed contact
prolate (X1 > / > X2 = X3) welding-deformation fabrics,
with inverse graded basal layer and lineations within the rapidly chilled marginal vitro-
lithic-rich layer phyre (Chapin and Lowell 1979).
shear (slide) surface Progressive aggradation involving agglutination
highly attenuated gas cavities from a continuous particulate flow also can account for
imbricate fiamme
occurrences of steeply lineated and/or foliated high-
grade tufts in volcanic conduits (e.g. Almond 1971;
basal vitrophyre (may be ]ineated or non-lineated)
Baker 1976; Reedman et al. 1987a). Thus planar com-
substrate
paction and attenuation fabrics in tuff dykes do not
necessarily indicate compression by closing wall rocks,
Fig. 8. A hypothetical high-grade ignimbrite, showing a vertical as advocated by Wolff (1986). Progressive aggradation
succession of lithofacies that could be produced by a single, long- also explains how high-grade tuff intrusions develop a
lived stratified flow with a variously changing depositional
boundary layer (see text for explanation). In reality, ignimbrites broadly concentric stratigraphy. For example, high-
do not exhibit this entire range of lithofacies at one location. No grade tufts in an elliptical plug at Weolseong, S. Korea
relative scales or ideal vertical sequence are implied. Brittle slid- (Park and Kim 1985; Reedman et al. 1987a) vary from
ing or ductile slumping (not indicated) may subsequently affect lava-like at the margins, through highly lineated (rods)
individual layers or the entire thickness of the ignimbrite. The and eutaxitic, to little-deformed at the centre. Similar
layers could easily be mistaken for the deposits of separate flows, associations between high-grade and poorly welded
surges and falls (and lavas), especially where sharp contacts and
vertical changes in chemical composition occur tuffs occur in dykes in South Kenya (Baker 1976), and
between high-grade, brecciated, and poorly welded
tufts in the 1-km-wide ring dyke along the 15-km-diam-
suggests that similar depositional mechanisms occur in eter ring fault of Hwasan Caldera, Korea (MJB re-
both situations. Intrusive tufts (Almond 1971; Wolff et search in progress).
al. 1989; Davis 1989; Hwasan Caldera ring dyke, Ko- The field evidence cited above suggests that progres-
rea, MJB work in progress) and ignimbrites on steep sive (stepwise and gradual) aggradation and boundary
surfaces (e.g. the Upper Bandelier Tuff at Paliza Ca- layer agglutination can accompany hot particulate flow
nyon and Pueblo Mesa: Self and Lipman 1989; the irrespective of gravitational settling. Gradual progres-
Cape Riva lag breccia: Druitt and Sparks 1982) com- sive aggradation to produce massive layers analogous
monly exhibit thin contact-parallel inverse-graded (2a) to layer 2b of Sparks (1976) and the $3 divison of Lowe
layers between massive poorly sorted (2b) deposits and (1982), on sub-vertical surfaces, suggests the occurrence
the substrate surface. Wolff et al. (1989) describe a of fluid-hindered deposition that is not directly driven
triaxial shape fabric (anisotropy of magnetic suscepti- by gravity. One possible explanation is that flow en-
bility, and locally visible fiamme), that is consistently tirely parallel to steep (and probably irregular) surfaces
imbricated with respect to a dyke wall throughout the is unlikely, so that the particulate flow is locally di-
thickness of an intrusive tuff dyke in Trans-Pecos Tex- rected obliquely onto the surface, inertially driving par-
as. This indicates progressive deposition and agglutina- ticles onto it so they displace less-dense fluid away
tion within a steep laminar depositional boundary from the surface. This model predicts that massive im-
layer. bricated tuff should not occur symmetrically deposited
517

onto both walls of a narrow dyke, because at any single grade of tuffs is the rate at which exsolution of various
location the flow is unlikely to be directed obliquely volatiles occurs relative to eruption ascent. Therefore,
onto both walls simultaneously. The prediction is ful- the nature of the plumbing system and the (possibly
filled in the case of the tuff dyke described by Wolff et changing) geometry of the conduit and vent may be as
al. (1989); intrusive tuff agglutinated only onto one important as eruption temperature.
wall, with asymmetric imbrication.
Conclusions
Sites of volatile exsolution
The continuum of ignimbrite grade reflects a spectrum
Magmatic volatiles play an important role in determin- of emplacement conditions and mechanisms (Fig. 1).
ing the grade of tuffs. Two apparently opposing effects Therefore, when interpreting the origin of an extensive
result from an increased content of magmatic volatiles; lava-like sheet, for example, consideration of the spec-
enhanced magmatic explosivity can promote develop- trum is a better approach to understanding and classifi-
ment of a high convective eruption column, which pro- cation than simplistic drawing of comparisons with li-
motes cooling during emplacement (Sparks et al. 1978), thofacies of low-grade (unequivocal) ignimbrites and of
yet volatiles dissolved in pyroclastic particles can lower stubby silicic lavas. Rare crystal breakage, absence of
their viscosity and yield strength (Schmincke 1974; Ma- remnant cuspate shards, and lack of cooling unit pro-
hood 1984), and thus promote coalescence, agglutina- files have been used to infer that some sheets are lava
tion, or post-depositional welding. flows, but the eruption and emplacement mechanisms
Viscous silicic lava flows form if slowly ascending of lava-like ignimbrites are inherently different to those
magma degasses slowly at depth (Jaupart and Allrgre of lower-grade ignimbrites. It should be recognised that
1991). In contrast, very rapid exsolution of volatiles in a (1) fragmentation of very fluidal magma involves less
conduit may drive a vigorous explosive eruption with crystal breakage than does brittle fragmentation of a
attendant tall column and quenched glass shards. How- high-yield strength explosive magma; (2) cuspate
ever, consider an eruption in which rapid magma as- shards and pumice may not form or may revert during
cent is primarily and largely hydraulic (for example transport to droplet-shapes and non-vesicular spatter
when the collapse of a magma chamber roof is trig- lumps (Gibson 1970; Schmincke 1974; Henry et al.
gered tectonically; Kokelaar and Busby research in 1989); and (3) fluidal particles may coalesce where
progress) as opposed to mainly pneumatic (driven by there is no load, right to the top of the sheet. In addi-
volatile expansion), and in which exsolution is far from tion, abundances of accidental lithics in lava-like ig-
complete by the time magmatic fragmentation occurs. nimbrites should more closely resemble those of foun-
(This may depend on which volatile species are present tain-fed basalt lavas than those of low-grade ignim-
because solubilities of different species vary differently brites. Basal autobreccias have been used to infer lava-
with depth, e.g. Holloway and Jakobsson 1986.) If the flow origin, but if the first-formed non-particulate com-
magmatic particles retained sufficient dissolved vola- ponent of a hot stratified flow is autobrecciated early in
tiles to remain fiuidal during ejection, they would be- the aggradation history, the resultant extremely high-
come droplet-shaped, which would minimise cooling grade ignimbrite will have autobreccia along its base.
during emplacement because low surface-area to vol- Basal autobreccia has been recorded from an unambi-
ume ratios and minimal air resistance do not favour guous ignimbrite on Gran Canaria (TL2 Mog/m Forma-
formation of high convective eruption columns (Wilson tion, Gran Canaria: J Sumner personal communication,
and Huang 1979). Such particles would be liable to ag- 1991) and from rheomorphic ignimbrites in Arizona (du
glutinate or coalesce. That particles do retain volatiles Bray and Pallister 1991).
is widely indicated by occurrences of abundant vesicles Complex vertical (as well as lateral) lithofacies
that formed in both clasts and matrix of extremely changes in ignimbrites can form during single eruptions
high-grade tuffs during and after non-particulate flow and can record successive modes of deposition in a
(Noble 1968; Gibson 1970; Schmincke 1974, 1990; boundary layer that evolves during passage of a single
Chapin and Lowell 1979; Mahood 1984). Schmincke flow. Sharp contacts in the deposit, and inverse-graded
(1974) described juvenile pumice clasts whose primary fine-grained layers, do not necessarily reflect intervals
vesicles collapsed entirely prior to growth of new, between successive flows. Determination of flow-unit
cross-cutting vesicles. However, we do not favour the boundaries requires evidence for a time interval.
term 'secondary vesiculation' because it cannot be as-
sumed that such revesiculation is a single event: a flui- Avoiding straitjacket terms
dal clast (or agglutinated matrix material) might vesicu- The concept of distinction between "primary" and
late and collapse several times during emplacement, on "secondary" welding and flowage (sensu Wolff and
each occasion losing a proportion of its dissolved vola- Wright 1981) is difficult to apply because rarely can it
tiles, before finally freezing in either a vesicular or non- be established that a high-grade ignimbrite formed by
vesicular state. Thus the preserved vesicularity and dis- an ashflow stopping en masse, then deflating, then
solved volatile content of the glass are not representative welding, and then undergoing non-particulate flow, in
of the volatile content during eruption and emplace- distinct stages. It seems that in many cases agglutina-
ment. tion occurs during progressive aggradation, and non-
We infer that an important factor determining the particulate flow can then continue until well after ag-
518

gradation has ceased. Also, the "primary" versus "sec- References


ondary" concept does not allow for the likelihood of
non-particulate flow structures developing whilst dense Almond DC (1971) Ignimbrite vents at Sabaloka Cauldron, Su-
dan. Geol Mag 186:159-176
basal parts of flows locally drain back downslope be- Baker BH (1976) Geology and geochemistry of the O1 Doinyo
fore coming to rest, as occurs within dense parts of cold Nyokie Trachyte Ignimbrite Vent Complex, South Kenya Rift
particulate flows both during and after initial emplace- Valley. Bull Volcanol 39: 420-440
ment (e.g. Fisher 1990). The presence of lineations is Bates RL, Jackson JA (eds) (1980) Glossary of geology. Amer
not diagnostic of post-emplacement flow (hot slump- Geol Inst, Falls Church, Virginia, USA, pp 1-175
ing) as suggested by Wolff and Wright (1981), because Bond A, Sparks RSJ (1976) The Minoan eruption of Santorini,
Greece. J Geol Soc London 131:1-16
lineations f o r m during deposition of non-welded ig- Bonnichsen B (1982) Rhyolite lava-flows in the Bruneau-Jar-
nimbrites, and because all gradations occur between li- bridge eruptive center, southwestern Idaho. In: Bonnichsen B,
neations f o r m e d by sedimentary grain alignment, those Breckenridge RM (eds) Cenezoic geology of Idaho. Idaho Bur
formed during agglutination, and those formed during Mines Geol Bull 26:283-320
hot slumping. Because ignimbrite fabrics reflect the Bonnichsen B, Kauffman DF (1987) Physical features of rhyolite
lava flows in the Snake River Plain volcanic province, south-
depositional regime rather than the transport regime,
western Idaho. In: Fink JH (ed) The emplacement of silicic
both particulate and non-particulate lineations can in- domes and lava flows. Geol Soc Am Spec Pap 212:118-145
dicate depositional b o u n d a r y layer flow either down lo- Branney MJ (1988) Subaerial explosive volcanism, intrusion, sedi-
cal t o p o g r a p h i c slopes or away from source. Thus li- mentation, and collapse in the Borrowdale Volcanic Group,
neation orientation alone cannot be taken to indicate SW Langdale, English Lake District. Unpubl PhD thesis, Shef-
remobilisation after stopping and deflating en masse. field Uni, UK, p 1-235
Branney MJ, Kokelaar BP, McConnell BJ (1992) The Bad Step
We conclude that with high-grade tufts the term "rheo- Tuff: a lava-like rheomorphic ignimbrite in a calc-alkaline pie-
morphic' is best retained for any non-particulate flow cemeal caldera, English Lake District. Bull Volcanol 54:187-199
structure that formed prior to lithification (Bates and Bristow CM (1962) Kenya ignimbrites. Nature 196:364-365
Jackson 1980; Branney et al. 1992). Terms such as hot Buesch DC, Valentine GA (1989) Thickness and flow dynamics as
gravity sliding, slumping, and loading-compaction de- factors controlling welding variations in ignimbrites (abstract).
formation adequately describe various forms of post- IAVCEI General Assembly on Continental Magmatism, Santa
Fe. New Mexico Bur Mines Mineral Resources Bull 131:32
e m p l a c e m e n t r h e o m o r p h i c remobilisation, where these
Cas RAF, Wright JV (1987) Volcanic successions: modern and an-
can be demonstrated specifically. The expressions 'par- cient. Allen & Unwin, London, pp. 1-528
ticulate flow' and 'non-particulate flow' better describe Carey SN (1991) Transport and deposition of tephra by pyroclas-
the nature of flow than " p r i m a r y laminar viscous flow", tic flows and surges. In: Fisher RV, Smith GA (eds) Sedimen-
"viscous mass f l o w a g e ' , and 'secondary viscous flow' tation in volcanic settings SEPM Spec Publ 45:39-57
(Schmincke and Swanson 1967; Wolff and Wright Chapin CE, Lowell GR (1979) Primary and secondary flow struc-
tures in ash-flow tufts of the Gribbles Run Palaeovalley, cen-
1981), because both particulate flow and non-particu-
tral Colorado. In: Chapin CE, Elston WE (eds) Ash flow tufts.
late flow can be laminar and viscous. However, we sug- Geol Soc Amer Spec Pap 180:137-154
gest that the terms 'agglutination' and 'coalescence', as Christiansen RL, Lipman PW (1966) Emplacement and thermal
used in this paper, are useful to distinguish welding history of a rhyolite lava flow near Fortymile Canyon, south-
that occurs rapidly on contact from p o s t - e m p l a c e m e n t ern Nevada. Geol Soc Am Bull 77:671-684
welding in which in situ cooling rate and lithostatic Cook EF (ed) (1966) Tufflavas and ignimbrites: a survey of Soviet
studies. Elsevier Publishing Co, NY, pp 1-212
load are important controls. Davis N (1989) The relationship between ignimbrite eruption and
The terms " p r i m a r y vesiculation" and "secondary caldera collapse in the Central Fells, English Lake District.
vesiculation" also inappropriately imply two stages of Unpubl PhD thesis, Sheffield Uni, UK, p 1-153
vesiculation whereas, in high-grade ignimbrites, vesicu- Druitt TH (1992) Emplacement of 18 May 1980, lateral blast ENE
lation, vesicle collapse, and revesiculation can occur at of Mount St. Helens, Washington. Bull Volcanol (in press)
any point, and possibly m a n y times, throughout the Druitt TH, Bacon CR (1986) Lithic breccia and ignimbrite
erupted during the collapse of Crater Lake caldera, Oregon. J
m a g m a ' s cooling history. Vesiculation m a y occur be-
Volcanol Geotherm Res 25:1-32
fore fragmentation in a conduit, and continue during Druitt TH, Sparks RSJ (1982) A proximal ignimbrite breccia fa-
fragmentation, ejection, particulate flow, agglutona- cies on Santorini, Greece. J Volcanol Geotherm Res 13:147-171
tion, non-particulate flow, and after, until the in situ ig- du Bray EA, Pallister JS (1991) An ash flow caldera in cross sec-
nimbrite finally lithifies. tion: ongoing field and geochemical studies of the Mid-Ter-
tiary Turkey Creek caldera, Chiricahua Mountains, SE Arizona.
Acknowledgements. This work would have been impossible without J Geophys Res 96:13435-13457
the numerous documented case studies of ignimbrites by many Duffield WA (1990) Eruptive fountains of silicic magma and their
different workers over the last 25 years. Encouraging and construc- possible effects on the tin content of fountain-fed lavas, Taylor
tive criticism by Dick Fisher, Phil Leat, Chuck Cbapin, Hans Creek Rhyolite, New Mexico. In: Stein H J, Hannah JL (eds)
Schmincke, Gustav Kobberger, Wendell Duffield, and Greg Val- Ore-bearing granite systems: petrogenesis and mineralizing
entine was greatly appreciated and led to a much improved final processes: Geol Soc Am Spec Pap 246:251-261
version of the manuscript. Thanks are due also to Tony Reedman, Eichelberger JC, Carrigan CR, Westrich HR, Price RH (1986)
Harry Pinkerton, Lionel Wilson and Tim Druitt for discussions on Non-explosive silicic volcanism. Nature 323:598-602
various aspects covered in our paper. Thanks to Janet Sumner for Ekren EB, McIntyre DH, Bennet EH (1984) High-temperature
showing MJB some ignimbrite localities on Gran Canaria, and to large-volume, lavalike ash-flow tufts without calderas in
Donatella De Rita and Danilo Palladino for showing MJB spatter- southwestern Idaho. US Geol Surv Prof Pap 1272:1-76
rich ignimbrites at Vulsini. MJB was funded by a NERC Research Ellwood BB (1982) Estimates of flow directions for calc-alkaline
Fellowship (GT5/F/89/GS/2). welded tufts and palaeomagnetic data reliability from aniso-
519

tropy of magnetic susceptibility measurements: Central San Kamata H, Mimura K (1983) Flow directions inferred from imbri-
Juan Mountains, southwest Colorado. Earth Planet Sci Lett cation in the Handa pyroclastic flow deposit in Japan. Bull
59:303-314 Volcanol 46: 277-282
Elston WE, Smith EI (1970) Determination of flow direction of Knight MD, Walker GPL, Ellwood BB, Diehl JF (1986) Stratigra-
rhyolitic ash-flow tuffs from fluidal textures. Geol Soc Bull phy, palaeomagnetism, and magnetic fabric of the Toba Tufts:
81 :3393-3406 constraints on the sources and eruptive styles. J Geophys Res
Fink JH, Manley CR (1987) Origin of pumiceous and glassy tex- 91:10355-10382
tures in rhyolite flows and domes. In: Fink JH (ed) The em- Leat PT (1985) Facies variations in peralkaline ash-flow tuffs
placement of silicic domes and lava flows. Geol Soc Am Spec from the Kenya Rift Valley. Geol Mag 122:139-150
Pap 212:77-88 Lowe DR (1979) Sediment gravity flows: their classification and
Fisher RV (1966) Mechanism of deposition from pyroclastic some problems of application to natural flows and deposits.
flows. Am J Sci 264:350-363 In: Doyle LJ, Pilkey OH (eds) Geology of continental slopes.
Fisher RV (1986) Systems of transport and deposition within py- SEPM Spec Publ 27:75-82
roclastic surges: evidence from Mount St. Helens, Washing- Lowe DR (1982) Sediment gravity flows: II. Depositional models
ton. EOS Trans Am Geophys Union 67:1246 with special reference to high-density turbidity currents. J Sed
Fisher RV (1990) Transport and deposition of a pyroclastic surge Petrol 52:279-297
across an area of high relief: the 18 May 1980 eruption of Lowe DR (1988) Suspended fallout rate as an independent varia-
Mount St. Helens, Washington. Geol Soc Am Bull 102:1038- ble in the analysis of current structures. Sedimentol 35:765-
1054 776
Fisher RV, Schmincke H-U (t984) Pyroclastic rocks. Springer- MacDonald GA (1972) Volcanoes. Prentice-Hall, Englewood
Verlag, Berlin 472 pp Cliffs, NJ, pp 1-510
Freundt A, Schmincke H-U (1990) The densely welded basaltic MacDonald WD, Palmer HC (1990) Flow directions in ash-flow
ignimbrite P1 on Gran Canada (Abstract). Abstr IAVCEI Int tufts: a comparison of geological and magnetic susceptibility
Volcanol Congr Mainz (FDR), Sept, 1990 measurements, Tshirege member (upper Bandelier Tuff),
Gay NC (1968) Pure shear and simple shear deformation of inho- Valles caldera, New Mexico, USA. Bull Volcanol 53:45-59
mogeneous viscous fluids. 1. Theory. Tectonophys 5:211-234 Mahood GA (1984) Pyroclastic rocks and calderas associated with
Gibson IA (1970) A pantellerite welded ash-flow tuff from the strongly peralkaline volcanic rocks. J Geophys Res 89:8540-
Ethiopian Rift Valley. Contrib Mineral Petrol 28:89-111 8552
Gluckman MJ, Yerushalmi J, Squires AM (1976) Defluidization Mahood GA, Hildreth W (1986) Geology of the peralkaline vol-
characteristics of sticky or agglomerating beds. In: Keairns cano at Pantelleria, Straits of Sicily. Bull Volcanol 48:143-
DL (ed) Fluidization Technology vol 2. McGraw-Hill, NY, pp 172
395-422 Manley CR, Fink JH (1987) Internal textures of rhyolite flows as
Hargrove HR, Denver P-L, Sheridan MF (1984) Welded ruffs de- revealed by research drilling. Geology 15:549-552
formed into megarheomorphic folds during collapse of the Marsella M, Palladino DM, Trigilia R (1987) The Onano Pyro-
McDermitt caldera, Nevada-Oregon. J Geophys Res 89: 8629-8638 clastic Formation (Vulsini Volcanoes): depositional features,
Hausback BP (1987) An extensive, hot, vapour-charged rhyoda- distribution and eruptive mechanisms. Per Mineral 56:225-
cite flow, Baja California, Mexico. In: Fink JH (ed) The em- 240
placement of silicic domes and lava flows. Geol Soc Am Spec McCall GJH (1965) Froth flows in Kenya. Geol Rundsch
Pap 212:111-H8 54:1148-1195
Hay RL, Hildreth W, Lambe R N (1979) Globule ignimbrite of Mellors RA, Sparks RSJ (1991) Spatter-rich pyroclastic flow de-
Mount Suswa, Kenya. In: Chapin CE, Elston WE (eds) Ash posits on Santorini, Greece. Bull Volcanol 53:327-342
flow tuffs. Geol Soc Am Spec Pap 180:167-175 Mimura K (1984) Imbrication, flow direction and possible source
Henry CD, Price GJ, Parker DF, Wolff JA (1989) Mid-Tertiary areas of the pumice flow tuffs near Bend, Oregon, USA. J Vol-
silicic alkalic magmatism of Trans-Pecos Texas: rheomorphic canol Geotherm Res 21:45-60
tufts and extensive silicic lavas. In: Chapin CE, Zidek J (eds) Nemec W (1990) Aspects of sediment movement on steep delta
Field excursions to volcanic terranes in the western United slopes. In: Colella A, Prior DB (eds) Coarse-grained deltas.
States, Vol 1: Southern Rocky Mountain region. New Mexico Spec Publ int Ass Sediment 10:29-73
Bur Mines Mineral Resources Mem 46:202-230 Noble DC (1968) Laminar viscous flowage structures in ash-flow
Henry CD, Price JG, Rubin JN, Laubach SE (1990) Case study of tuffs from Gran Canaria, Canary Islands: a discussion. J Geol
an extensive silicic lava: the Bracks Rhyolite, Trans-Pecos 76:721-723
Texas. J Volcanol Geotherm Res 43:113-132 Orsi G, Sheridan MF (1984) The Green Tuff of Pantelleria:
Hildreth W (1981) Gradients in silicic magma chambers: implica- rheoignimbrite or rheomorphic fall? Bull Volcanol 47-3: 611-
tions for lithospheric magmatism. J Geophys Res 86:10153- 626
10192 Orsi G, Sheridan MF (1986) The Green Tuff of Pantelleria: an
Hillhouse JW, Wells RE (1991) Magnetic fabric, flow directions, example of rheoignimbrite (Abstract). Abstr 1986 IAVCEI Int
and source area of the Lower Miocene Peach Springs Tuff in Volcanol Congr, NZ 67
Arizona, California, and Nevada. J Geophys Res 96:12443- Orsi G, Ruvo L, Scarpati C (1991) The recent explosive volcanism
12 460 at Pantelleria. Geol Rundsch 80:187-200
Hiscott RN, Middleton GV (1980) Fabric of coarse deep-water Park KH, Kim SE (1985) Ash-flow tufts of the Chiselryoung Vol-
sandstones, Tourelle Formation, Quebec, Canada. J Sed Petrol canic Formation and associated welded tuff intrusion, Weol-
50:703-721 seong District, South Korea. J Korea Inst Mineral Geol
Holloway JR, Jakobsson S (1986) Volatile solubility in magmas: 18:357-368
transport of volatiles to planet surfaces. J Geophys Res Parkash B, Middletou GV (1970) Downcurrent textural changes in
91 : 505-508 Ordovician Turbidite Greywackes. Sedimentol 14:259-293
Jaupart C, Allrgre CJ (1991) Gas content, eruption rate and insta- Peterson DW (1979) Significance of flattening of pumice frag-
bilities of eruption regime in silicic volcanoes. Earth Planet Sci ments in ash-flow tufts. In: Chapin CE, Elston WE (eds) Ash
Lett 102:413-429 flow tuffs. Geol Soc Am Spec Pap 180:195-204
Johnson AM (1970) Physical processes in geology. Freeman Co- Pichler H (1981) Italienische Vulkan-Gebiete III: Lipari, Vulcano,
oper, San Francisco, California, pp 1-577 Stromboli, Tyrrenisches Meer. Sammlung Geol F~hrer 69:pp
Johnson RW (1968) Volcanic globule rock from Mount Suswa, 1-233
Kenya. Geol Soc Am Bull 79:647-651 Potter DB, Oberthal CM (1987) Vent sites and flow directions of
520

the Otowi ash-flows (lower Bandelier Tuff), New Mexico. Suzuki K, Ui T (1982) Grain orientation and depositional ramps
Geol Soc Am Bull 98:66-76 as flow direction indicators of large-scale pyroclastic flow de-
Ragan DH, Sheridan MF (1972) Compaction of the Bishop Tuff, posits in Japan. Geology 10:429-432
California. Geol Soc Am Bull 83:95~106 Tiara A, Scholle PA (1979) Deposition of resedimented sandstone
Reedman AJ, Park KH, Merriman RJ, Kim SE (1987a) Welded beds in the Pico Formation, Ventura basin, California, as in-
tuff infilling a volcanic vent at Weolseong, Republic of Korea. terpreted from magnetic fabrics measurements. Geol Soc Am
Bull Volcanol 49: 541-546 Bull 90:952-962
Reedman AJ, Howells MF, Orton G, Campbell SDG (1987b) The Trigilia R, Walker GPL (1986) The Onano Spatter Flow, Italy:
Pitts Head Tuff Formation: a subaerial to submarine welded evidence for a new ignimbrite depositional mechanism. Abstr
ash-flow tuff of Ordivician age, North Wales. Geol J 124:427- Int Volcanol Congr NZ 1986
439 Ui T, Suzuki-Kamata K, Matsusue R, Fujita K, Metsugi H, Araki
Rees AL, Woodall WA (1975) The magnetic fabric of some labo- M (1989) Flow behaviour of large-scale pyroclastic flows - evi-
ratory-deposited sediments. Earth Planet Sci Lett 25:121-130 dence obtained from petrofabric analysis. Bull Volcanol
Riehle JR (1973) Calculated compaction profiles of rhyolitic ash- 51:115-122
flow tuffs, Geol Soc Am Bull 84:2193-2216 Valentine GA (1987) Stratified flow in pyroclastic surges. Bull
Ross CS, Smith RL (1961) Ash-flow tufts, their origin, geological Volcanol 49: 616-630
relations and identification. US Geol Surv Prof Pap 366:1-77 Valentine GA, Buesch DC, Fisher RV (1989) Basal layered depos-
Ryan MP, Blevins JYK (1987) The viscocity of synthetic and nat- its of the Peach Springs Tuff, northwest Arizona, USA. Bull
ural silicate melts and glasses at high temperature and 1 Bar Volcanol 51:395-414
(105 Pascals) pressure and at higher pressures. US Geol Surv Vernon RH (1987) A microstructural indicator of shear sense in
Bull 1764 volcanic rocks and its relationship to porphyroblast rotation in
Schmincke H-U (1969) Ignimbrite sequence on Gran Canaria. metamorphic rocks. J Geol 95:127-134
Bull Volcanol 35:1199-1219 Villari L (1969) On particular ignimbrites of the island of Pantel-
Schmincke H-U (1972) Froth blows and globule flows in Kenya. leria (Channel of Sicily). Bull Volcanol 33:828-839
Naturwissensch 11 : 1-2 Villari L (1974) The island of Pantelleria. Bull Volcanol 38:680-
Schmincke H-U (1974) Volcanological aspects of peralkaline 724
silicic welded ash-flow tuffs. Bull Volcanol 38:594-636 Walker GPL (1983) Ignimbrite types and ignimbrite problems. J
Schmincke H-U, with contributions by Freundt A, Ferriz H, Kob- Volcanol Geotherm Res 17:65-88
berger G, Leat P (1990) Geological field guide Gran Canaria Walker GPL, Croasdale R (1972) Characteristics of some basaltic
1-202. Pluto Press, Witten FRG pyroclastics. Bull Volcanol 35:303-317
Schmincke H-U, Swanson DA (1967) Laminar viscous flowage Walker GPL, Hemming RF, Wilson CJN (1980) Low aspect-ratio
structures in ash-flow tuffs from Gran Canaria, Canary Is- ignimbrites. Nature 283:286-287
lands. J Geol 75:641-664 Walker GPL, Wilson CJN, Froggatt PC (1981) An ignimbrite ven-
Scott SC (1980) The geology of Longenot volcano, Central Kenya: eer deposit: the trail marker of a pyroclastic flow. J Volcanol
a question of volumes. Phil Trans R Soc London Ser A Geotherm Res 8:409-421
296: 437-465 Walker GW, Swanson DA (1968) Laminar flowage in a Pliocene
Seaman SJ, McIntosh WC, Geissman JW, Williams ML, Elston soda rhyolite ashflow tuff, Lake and Harney counties, Oregon.
WE (1991) Magnetic fabrics of the Bloodgood Canyon and US Geol Surv Prof Pap 600-B:37-47
Shelley Peak Tufts, southwestern New Mexico: implications Wilson CJN (1980) The role of fluidization in the emplacement of
for emplacement and alteration process. Bull Volcanol pyroclastic flows: an experimental approach. J Volcanol Geo-
53:460-476 therm Res 8: 231-249
Self S, Lipman PW (1989) Large ignimbrites and caldera-forming Wilson CJN (1985) The Taupo eruption, New Zealand. II The
eruptions. Handbook of the IAVCEI Working Group on Ex- Taupo ignimbrite. Phil Trans R Soc Lond 314:229-310
plosive Volcanism Field Workshop (11WA) in Jemez Moun- Wilson CJN (1986) Pyroclastic flows and ignimbrites. Sci Prog
tains, New Mexico, and San Juan Mountains, Colorado. 134 Oxf 70:172-201
PP Wilson L, Head JW (1981) Morphology and rheology of pyroclas-
Shaw HR (1963) Obsidian-H20 viscosities at 1000 and 2000 bars tic flows and their deposits, and guidelines for future observa-
in the temperature range 700 ~ to 900~ J Geophys Res tions. In: Lipman PW, Mullineaux DR (eds) The 1980 erup-
68: 6337-6343 tions of Mount St. Helens, Washington. US Geol Surv Prof
Sheridan MF (1979) Emplacement of pyroclastic flows: a review. Pap 1250:513-524
In: Chapin CE, Elston WE (eds) Ash-flow turfs. Geol Soc Am Wilson L, Huang TC (1979) The influence of shape on the atmo-
Spec Pap 180:125-136 sphere settling velocity of volcanic ash particles. Earth Planet
Simpson C, Schmid SM (1983) An evaluation of criteria to deduce Sci Lett 44:311-324
the sense of movement in sheared rocks. Geol Soc Am Bull Wolff JA (1986) Welded tuff dykes, conduit closure, and lava
94:1281-1288 dome growth at the end of explosive eruptions. J Volcanol
Smith RL (1960) Zones and zonal variations in welded ash-flows. Geotherm Res 28:379-384
US Geol Surv Prof Pap 354-F: 149-159 Wolff JA, Wright JV (1981) Rheomorphism of welded tuffs. J Vol-
Sparks RSJ (1976) Grain size variations in ignimbrites and impli- canol Geotherm Res 10:13-34
cations for the transport of pyroclastic flows. Sedimentol Wolff JA, Ellwood BB, Sachs SD (1989) Anisotropy of magnetic
23:147-188 susceptibility in welded tuffs: application to a welded-tuff
Sparks RSJ, Wright JV (1979) Welded air-fall tuff. In: Chapin CE, dyke in the Tertiary Trans-Pecos Texas volcanic province,
Elston WE (eds) Ash flow tufts. Geol Soc Am Spec Pap USA. Bull Volcanol 51:299-310
180:155-166 Wright JV, Walker GPL (1981) Eruption, transport and deposition
Sparks RSJ, Francis PW, Hamer RD, Pankhurst RJ, O'Callaghan of ignimbrite: a case study from Mexico. J Volcanol Geotherm
LO, Thorpe RS, Page R (1985) Ignimbrites of the Cerro Galan Res 9:111-131
Caldera, NW Argentina. J Volcano1 Geotherm Res 24:205-
248
Sparks RSJ, Wilson L, Hulme G (1978) Theoretical modelling of
the generation, movement and emplacement of pyroclastic
flows by column collapse. Geophys Res 83:1727-1739 Editorial responsibility: H-U Schmincke

You might also like