23.07.2013 Views

Distinguishing strongly rheomorphic tuffs from extensive silicic lavas

Distinguishing strongly rheomorphic tuffs from extensive silicic lavas

Distinguishing strongly rheomorphic tuffs from extensive silicic lavas

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

Bull Volcanol (1992) 54:171-186<br />

<strong>Distinguishing</strong> <strong>strongly</strong> <strong>rheomorphic</strong> tufts<br />

<strong>from</strong> <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong><br />

Christopher D Henry 1 and John A Wolff /<br />

1 Bureau of Economic Geology, University of Texas at Austin, Austin, TX 78713, USA<br />

2 Department of Geology, University of Texas at Arlington, UTA Box 19049, Arlington, TX 76019, USA<br />

Received August 23, 1990/Accepted October 3, 1991<br />

Abstract. High-temperature <strong>silicic</strong> volcanic rocks, in-<br />

cluding <strong>strongly</strong> <strong>rheomorphic</strong> <strong>tuffs</strong> and <strong>extensive</strong> <strong>silicic</strong><br />

<strong>lavas</strong>, have recently been recognized to be abundant in<br />

the geologic record. However, their mechanisms of<br />

eruption and emplacement are still controversial, and<br />

traditional criteria used to distinguish conventional<br />

ash-flow <strong>tuffs</strong> <strong>from</strong> <strong>silicic</strong> <strong>lavas</strong> largely fail to distin-<br />

guish the high-temperature versions. We suggest the<br />

following criteria, ordered in decreasing ease of identif-<br />

ication, to distinguish <strong>strongly</strong> <strong>rheomorphic</strong> <strong>tuffs</strong> <strong>from</strong><br />

<strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>: (1) the character of basal depo-<br />

sits; (2) the nature of distal parts of flows; (3) the rela-<br />

tionship of units to pre-existing topography; and (4) the<br />

type of source. As a result of quenching against the<br />

ground, basal deposits best preserve primary features,<br />

can be observed in single outcrops, and do not require<br />

knowing the full extent of a unit. Lavas commonly de-<br />

velop basal breccias composed of a variety of textural<br />

types of the flow in a finer clastic matrix; such deposits<br />

are unique to <strong>lavas</strong>. Because the chilled base of an ash-<br />

flow tuff generally does not participate in secondary<br />

flow, primary pyroclastic features are best preserved<br />

there. Massive, flow-banded bases are more consistent<br />

with a lava than a pyroclastic origin. Lavas are thick to<br />

their margins and have steep, abrupt flow fronts. Ash-<br />

flow <strong>tuffs</strong> thin to no more than a few meters at their<br />

distal ends, where they generally do not show any sec-<br />

ondary flow features. Lavas are stopped by topographic<br />

barriers unless the flow is much thicker than the bar-<br />

rier. Ash-flow <strong>tuffs</strong> moving at even relatively slow velo-<br />

cities can climb over barriers much higher than the re-<br />

sulting deposit. Lavas dominantly erupt <strong>from</strong> fissures<br />

and maintain fairly uniform thicknesses throughout<br />

their extents. Tufts commonly erupt <strong>from</strong> calderas<br />

where they can pond to thicknesses many times those<br />

of their outflow deposits. These criteria may also prove<br />

effective in distinguishing <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> <strong>from</strong> a<br />

postulated rock type termed lava-like ignimbrite. The<br />

latter have characteristics of <strong>lavas</strong> except for great areal<br />

Offprint requests to: CD Henry<br />

Volc a' ology<br />

9 Springer-Verlag 1992<br />

extents, up to many tens of kilometers. These rocks<br />

have been interpreted as ash-flow <strong>tuffs</strong> that formed<br />

<strong>from</strong> low, boiling-over eruption columns, based almost<br />

entirely on their great extents and the belief that <strong>silicic</strong><br />

<strong>lavas</strong> could not flow such distances. However, we inter-<br />

pret the best known examples of lava-like ignimbrites<br />

to be <strong>lavas</strong>. This interpretation should be tested<br />

through additional documentation of their characteris-<br />

tics and research on the boiling-over eruption mecha-<br />

nism and the kinds of deposits it can produce. Flow<br />

bands, flow folds, ramps, elongate vesicles, and proba-<br />

bly upper breccias occur in both <strong>lavas</strong> and <strong>strongly</strong><br />

<strong>rheomorphic</strong> tufts and are therefore not diagnostic. Pu-<br />

mice and shards also occur in both <strong>tuffs</strong> and <strong>lavas</strong>, al-<br />

though they occur throughout ash-flow tufts and gener-<br />

ally only in marginal breccias of <strong>lavas</strong>. Dense welding,<br />

secondary flow, and intense alteration accompanying<br />

crystallization at high temperature commonly obliterate<br />

primary textures in both thick, <strong>rheomorphic</strong> tufts and<br />

thick <strong>lavas</strong>. High-temperature <strong>silicic</strong> volcanic rocks are<br />

dominantly associated with tholeiitic flood basalts. Ex-<br />

tensive <strong>silicic</strong> <strong>lavas</strong> could be appropriately termed<br />

flood rhyolites.<br />

Introduction<br />

The textures and structures developed in <strong>silicic</strong> ash-<br />

flow <strong>tuffs</strong> (ignimbrites), including densely welded <strong>tuffs</strong>,<br />

are sufficiently different <strong>from</strong> those of <strong>lavas</strong> that the<br />

two can be distinguished even in ancient deformed and<br />

metamorphosed terranes (Table 1). These contrasts ar-<br />

ise <strong>from</strong> the very different eruption and emplacement<br />

mechanisms of <strong>lavas</strong> and tufts. In contrast, high-tem-<br />

perature, <strong>strongly</strong> <strong>rheomorphic</strong> <strong>tuffs</strong> and <strong>extensive</strong><br />

<strong>silicic</strong> <strong>lavas</strong>, recently recognized to be abundant in the<br />

geologic record (Table 2), undergo different and/or<br />

more complex eruption and emplacement processes<br />

that blur these distinctions. High-temperature, large-<br />

volume <strong>silicic</strong> <strong>lavas</strong> have areal extents and aspect ratios<br />

(mean thickness divided by the characteristic horizontal


172<br />

Table 1. Characteristics commonly used to distinguish rhyolite<br />

<strong>lavas</strong> <strong>from</strong> ash-flow tufts ~<br />

Common in conventional Common in rhyolite <strong>lavas</strong> and<br />

ash-flow tufts in <strong>rheomorphic</strong> tufts<br />

Fiamme<br />

Eutaxitic texture<br />

Abundant lithic fragments<br />

Nonwelded tops, bottoms,<br />

sides<br />

Gradual thinning at edges<br />

Wide areal extent<br />

Large aspect ratio<br />

Glass shards<br />

Broken phenocrysts ranging<br />

widely in size<br />

Gas elutriation pipes<br />

Flow bands<br />

Ramp structures<br />

Elongate vesicles<br />

Autobreccias<br />

Vitrophyres at or near tops<br />

Areally restricted b<br />

Small aspect ratio b<br />

Similar discrimination characteristics were discussed by Bristow<br />

and Cleverly (1979), Cleverly and Bristow (1982), Bonnichsen and<br />

Kauffman (1987), Henry and others (1988, 1989), and Bristow<br />

(1989)<br />

Rheomorphic tufts may have dimensions similar to those of<br />

other ash-flow tufts<br />

dimension of the rock body; Walker et al. 1980) com-<br />

parable to those of tufts (Cas 1978; Bonnichsen and<br />

Kauffman 1987; Henry et al. 1989, 1990; Christiansen<br />

and Hildreth 1989). Strongly <strong>rheomorphic</strong> tufts un-<br />

dergo such intense episodes of coherent viscous flow<br />

that they develop flow bands, flow folds, breccia zones,<br />

Table 2. Occurrences of <strong>extensive</strong> <strong>silicic</strong> volcanic rocks of controversial origin<br />

and other features of <strong>lavas</strong> (Schmincke and Swanson<br />

1967; Chapin and Lowell 1979; Wolff and Wright<br />

1981; Orsi and Sheridan 1984; Bonnichsen and Kauff-<br />

man 1987; Henry et al. 1989). Because of this conver-<br />

gence of characteristics, the traditional criteria of Table<br />

1 largely fail to distinguish <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> <strong>from</strong><br />

<strong>strongly</strong> <strong>rheomorphic</strong> tufts. Both rock types may be<br />

found in some volcanic provinces (e.g. SW Idaho, Bon-<br />

nichsen and Kauffman 1987; Trans-Pecos Texas, Henry<br />

et al. 1989), which can further hinder interpretation of<br />

individual units, inasmuch as investigators have tended<br />

to interpret all widespread <strong>silicic</strong> units within a particu-<br />

lar province as being of either one or the other type.<br />

Until recently, the existence of <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong><br />

was controversial. With rare exceptions, any <strong>extensive</strong><br />

<strong>silicic</strong> volcanic unit was assumed to be an ignimbrite.<br />

The main purpose of this paper is to propose diagnostic<br />

criteria to distinguish between <strong>strongly</strong> <strong>rheomorphic</strong><br />

tufts and widespread <strong>silicic</strong> <strong>lavas</strong>.<br />

High-temperature <strong>silicic</strong> magmas, emplaced as pyro-<br />

clastic deposits or as widespread <strong>lavas</strong>, are significant<br />

throughout the geologic record (Table 2). Recognition<br />

of these rocks is especially significant for interpretation<br />

of old, eroded, and deformed volcanic terranes; for ex-<br />

ample, facies models for ancient terrains take <strong>silicic</strong><br />

<strong>lavas</strong> to be near-source rock bodies (Cas and Wright<br />

1987). The continuously exposed extent of the Oligo-<br />

cene, single-flow-unit Bracks Rhyolite of Trans-Pecos<br />

Texas is 55 km, with a probable source area near one<br />

end of that extent (Henry et al. 1988, 1989, 1990). Clear-<br />

Volcanic province Characteristics of <strong>silicic</strong> volcanic rocks References<br />

Basal breccia Upper breccia Distal thinning Temperature (~ C)<br />

Snake River, Idaho Yes Yes No a 850-1050 Ekren et al. 1984; Bon-<br />

nichsen and Kauffman<br />

1987; Manley 1990; Honjo<br />

et al. 1992<br />

Yellowstone Park Yes Yes No > 850 Christiansen and Hildreth<br />

1989<br />

Trans-Pecos Texas Yes Yes No a > 900 Parker and McDowell 1979;<br />

Franklin et al. 1987, 1989;<br />

Henry et al. 1988, 1989,<br />

1990<br />

Keweenawan, Minnesota Minor Yes No > 1000 Green 1989; Green and Fitz<br />

1989<br />

Parana, Brazil Minor Yes No 1000-1150 Bellieni et al. 1984; Petrini et<br />

al. 1989; Whittingham 1989<br />

Etendeka, Namibia Minor Yes No 1000-1100 Milner 1986; Milner et al.<br />

1992<br />

Lebombo (Karoo) Minor Yes No 900-1100 Bristow and Cleverly 1979;<br />

South Africa Cleverly and Bristow 1982;<br />

Bristow 1989<br />

Rooiberg (Bushveld) Minor Yes No 1100 Twist and French 1983; Twist<br />

South Africa et al. 1989; Twist and EI-<br />

ston 1989<br />

Australia Yes Yes No 1100 Cas 1978; Creaser and White<br />

1991<br />

Some definite <strong>rheomorphic</strong> tufts thin to a few meters at margins, where they show little or no secondary flow. Lavas in Idaho and Texas<br />

have steep, abrupt flow fronts


ly, a conventional facies interpretation of a sequence<br />

containing such units could be seriously in error.<br />

Emplacement mechanisms and magma rheology<br />

To identify criteria useful to distinguish among the var-<br />

ious types of <strong>extensive</strong> <strong>silicic</strong> volcanic rocks, it is neces-<br />

sary to consider eruption and emplacement mecha-<br />

nisms and what they imply about resulting deposits.<br />

Thus, we envisage three possible rock types.<br />

Rheomorphic welded tuff<br />

The unit is distributed over most or all of its extent as a<br />

gas-supported particulate flow and undergoes welding<br />

and secondary coherent viscous flow after depositon<br />

(Wolff and Wright 1981). Only pyroclastic flow (and<br />

not fall) deposits are likely to undergo rheomorphism<br />

over a wide area, and hence the term rheoignimbrite is<br />

synonymous in this context.<br />

Lava-like tuff or lava-like ignimbrite<br />

This postulated rock type may form <strong>from</strong> a weakly ex-<br />

plosive eruption and low column. Deformation of con-<br />

stituent clasts occurs during primary particulate flow<br />

(i.e. within-flow agglutionation; see Henry et al. 1989,<br />

for a fuller discussion of terminology). This is similar to<br />

the primary laminar viscous flow concept of Schmincke<br />

and Swanson (1967) or Chapin and Lowell (1979). We<br />

recognise this as a separate rock type because clast<br />

properties influence deposition of material <strong>from</strong> the<br />

primary flow, and because agglutination requires lower<br />

clast viscosities than welding and rheomorphism (Wolff<br />

1983; Wolff and Horn 1991). Rheomorphism is there-<br />

fore likely to be more intense in such a deposit (and<br />

recognition of a pyroclastic origin correspondingly<br />

more difficult) than in one where coherence was only<br />

achieved after deposition, i.e. during welding.<br />

Although agglutination is known to occur in the for-<br />

mation and deposition of some pyroclastic flows (Tri-<br />

gila and Walker 1986), we point out that the concept of<br />

lava-like <strong>tuffs</strong> arose following identification of rock<br />

bodies that have the outcrop and textural characteris-<br />

tics of <strong>lavas</strong> but the low-aspect ratios of ignimbrites<br />

(Bristow and Cleverly 1979; Cleverly and Bristow 1982;<br />

Ekren et al. 1984; Milner 1986; Milner et al. 1992), and<br />

not through recognition of a distinct depositional<br />

mechanism. Having examined the major examples of<br />

lava-like <strong>tuffs</strong>, we conclude that they are <strong>lavas</strong> (see<br />

Concluding Remarks). Also, the units described by<br />

Schmincke and Swanson (1967) and Chapin and Low-<br />

ell (1979) are unequivocally pyroclastic and, in the<br />

former case, the recognition of 'primary welding' struc-<br />

tures is questionable (Wolff and Wright 1981; Leat and<br />

Schmincke 1990). Our experience indicates that, with<br />

sufficient information, individual units can be unequi-<br />

vocally identified as either <strong>lavas</strong> or <strong>rheomorphic</strong> tufts,<br />

173<br />

and there may ultimately be no need for a category of<br />

lava-like <strong>tuffs</strong>. We recognize it in this discussion be-<br />

cause the possibility of a distinct depositional mecha-<br />

nism, involving agglutination, has some implications<br />

for the diagnostic criteria we wish to evaluate.<br />

Extensive <strong>silicic</strong> <strong>lavas</strong><br />

These are emplaced over their great extents as coherent<br />

flows, even though eruptions may vary <strong>from</strong> continuous<br />

effusion to some mildly explosive 'fire fountaining'<br />

(e.g. Duffield 1990).<br />

The above definitions focus on how the magma is<br />

emplaced over its observed extent, rather than the form<br />

in which it emerges <strong>from</strong> the vent. The need for this<br />

approach becomes clear when one considers that a low<br />

pyroclastic fountain playing over a vent could, depend-<br />

ing on the grainsize and 'stickiness' of the clasts, pro-<br />

duce either a partly fluidized particulate flow or a spat-<br />

ter-fed lava. In general, we envisage that intensely rheo-<br />

morphic ignimbrites are erupted <strong>from</strong> thermally conser-<br />

vative, low pyroclastic fountains (collapsing eruption<br />

columns), whereas widespread <strong>silicic</strong> <strong>lavas</strong> are spatter<br />

fed and/or the products of direct effusion.<br />

The fundamental physical property controlling be-<br />

havior of magmas during surface emplacement is vis-<br />

cosity (and allied rheological parameters), itself de-<br />

pendent on temperature and chemical composition.<br />

Welding and rheomorphism of ignimbrites is controlled<br />

by clast viscosity and deposit thickness, which in turn<br />

controls internal load pressure and cooling rate. Many<br />

intensely <strong>rheomorphic</strong> ignimbrites are, however, only a<br />

few meters thick and must have cooled rapidly after<br />

emplacement. It has been recognized for some time<br />

(e.g. Schmincke 1974) that intense welding and rheo-<br />

morphism, and possibly also the effects of within-flow<br />

clast agglutination, are common in pantelleritic and<br />

comenditic ignimbrites. These magmas produce pyro-<br />

clasts with lower viscosities at emplacement than most<br />

<strong>silicic</strong> magma types, by virtue of higher temperatures<br />

and Na-, K-, Fe-, and F-rich compositions. It is clear<br />

<strong>from</strong> several recent studies (Ekren et al. 1984; Bon-<br />

nichsen and Kaufman 1987; Henry et al. 1988; Honjo<br />

et al. 1992) that many intensely <strong>rheomorphic</strong> ignim-<br />

brites in continental settings are not peralkaline, but<br />

had magmatic temperatures (850-1100~ higher, in<br />

many cases much higher, than common <strong>silicic</strong> rocks<br />

(Table 2). They are typically porphyritic and lack evi-<br />

dence for superheating; the high temperatures there-<br />

fore require that they had unusually low water contents<br />

prior to eruption. Calculated magmatic viscosities, us-<br />

ing the method of Shaw (1972), are in the range 103-106<br />

Pa. s for compositions reported in the literature, using<br />

temperatures (900-1100 ~ C) given by the respective au-<br />

thors, and allowing 0-2 wt% water. These values are<br />

several orders of magnitude less than the eruptive vis-<br />

cosities of 'conventional' volatile-rich <strong>silicic</strong> magmas,<br />

which have lower initial magmatic temperatures, degas<br />

during eruption, and may undergo up to a few hundred<br />

degrees of eruptive cooling (Sparks et al. 1978).


174<br />

Formation of <strong>extensive</strong> <strong>silicic</strong> lava flows is also fa-<br />

vored by low magmatic viscosity and therefore high<br />

temperature. Although extrusion rate is probably the<br />

most important control on lava-flow length (Walker<br />

1973; Pieri and Baloga 1986), maximum magma rise<br />

rates in the conduit (and hence extrusion rates at con-<br />

stant conduit width) are controlled by viscosity or yield<br />

strength (Wilson and Head 1981). Thus a low-viscosity<br />

magma is inherently more capable of forming <strong>extensive</strong><br />

flows than a high-viscosity equivalent.<br />

As pointed out by Mahood (1984), unusually fluid<br />

felsic magmas only exhibit 'anomalous' behavior when<br />

compared with common magmas of similar silica con-<br />

tent. When compared instead with rheologically similar<br />

types (e.g. andesites), closely similar behavior is seen.<br />

Andesites exhibit the whole range of eruptive behavior<br />

seen among <strong>silicic</strong> magmas (both 'hot' and 'normal');<br />

they may form <strong>extensive</strong> <strong>lavas</strong> or domes and coulees,<br />

and may erupt explosively to yield plinian fallout and<br />

pyroclastic flow deposits that show agglutination, weld-<br />

ing, and rheomorphism.<br />

The difference between effusive and explosive erup-<br />

tions are relatively well understood, allowing straight-<br />

forward analysis of most of the deposits they can pro-<br />

duce. In contrast, the eruptive mechanism (and even<br />

existence) of lava-like <strong>tuffs</strong> is not established. They may<br />

form <strong>from</strong> low, boiling-over type eruptive columns in<br />

which heat loss to the atmosphere is minimal (Sparks et<br />

al. 1978; Bristow and Cleverly 1979; Ekren et al. 1984;<br />

Wolff 1989). Such an eruptive style has been recognized<br />

for Cotopaxi, Ecuador (Wolf 1878), and Mount St. Hel-<br />

ens, USA (Rowley et al. 1981). However, these erup-<br />

tions produced small, nonwelded <strong>tuffs</strong>, decidedly un-<br />

like the postulated lava-like ruffs. Our lack of under-<br />

standing of the mechanisms involved in producing<br />

lava-like tufts greatly restricts our ability to propose<br />

distinguishing criteria for them.<br />

<strong>Distinguishing</strong> criteria<br />

The extent to which a <strong>rheomorphic</strong> tuff or lava-like tuff<br />

resembles a lava depends on the extent to which vis-<br />

cous flowage has obscured features diagnostic of pyro-<br />

elastic origin. Rheomorphic <strong>tuffs</strong> commonly preserve<br />

abundant evidence of their pyroclastic origin, and sec-<br />

ondary flow generally obscures such evidence only lo-<br />

cally. Tuffs that have undergone <strong>extensive</strong> secondary<br />

viscous flow closely resemble <strong>lavas</strong>, and lava-like <strong>tuffs</strong><br />

may be indistinguishable. Hence, much of this discus-<br />

sion emphasizes features diagnostic of <strong>extensive</strong> <strong>silicic</strong><br />

<strong>lavas</strong>.<br />

The following discussion is based largely on our<br />

work in Trans-Pecos Texas (Franklin et al. 1987, 1989;<br />

Henry et al. 1988, 1989, 1990) but also upon field exam-<br />

ination and published reports of <strong>extensive</strong>, high-tem-<br />

perature <strong>silicic</strong> volcanic rocks in Idaho (Bonnichsen<br />

and Kauffman 1987; Ekren et al. 1984), Yellowstone<br />

National Park (Christiansen and Hildreth 1989), Nami-<br />

bia (Milner 1986; Milner et al. 1992), and South Africa<br />

(Bristow and Cleverly 1979; Twist and French 1983),<br />

and published reports of many others. We are particu-<br />

larly indebted to earlier attempts by the above authors<br />

to provide distinguishing criteria. Field criteria, espe-<br />

cially those features most likely to be preserved in an-<br />

cient sequences, are emphasized.<br />

Basal deposits<br />

Basal deposits potentially provide the most diagnostic<br />

and most easily recognized characteristics with which<br />

to distinguish <strong>rheomorphic</strong> tufts <strong>from</strong> <strong>extensive</strong> <strong>silicic</strong><br />

<strong>lavas</strong>. Basal deposits of <strong>tuffs</strong> and <strong>lavas</strong> are commonly<br />

markedly different and are well preserved due to<br />

quenching against the ground and immediate burial by<br />

the flow of which they are part. Also, they can be ob-<br />

served in single outcrops and do not require knowing<br />

the full extent of a unit. Unfortunately in many cases,<br />

such as much of the mid-Tertiary of the western USA,<br />

basal sections are poorly exposed or covered.<br />

The bases of <strong>lavas</strong> are characterized by breccias that<br />

develop as the chilled, plastic to brittle fragments of<br />

broken crust on the top of a flow are brought to the<br />

flow front and overridden by it (cf. 'caterpillar-track'<br />

motion of aa <strong>lavas</strong>). Under some conditions, a large<br />

proportion of fragments may also spall directly <strong>from</strong><br />

the flow front (Cas and Wright 1987; Manley and Fink<br />

1987). Such breccias consist of clasts of a mix of lava<br />

textural types, including massive, vesicular, flow-<br />

banded and flow-folded, glassy, pumiceous, and devi-<br />

trifled (Fig. 1), all of which are exposed in flow tops<br />

and fronts (Christiansen and Lipman 1966; Fink 1983);<br />

large clasts reside in a matrix of more finely commin-<br />

uted material, which can include pumice and shards.<br />

Basal flow breccias can occur throughout the areal ex-<br />

tent of a flow, exhibit no systematic grain-size varia-<br />

tions (although sorting chracteristics may vary), may<br />

vary markedly in thickness over short lateral distances,<br />

and may locally be absent. Thickness varies because<br />

different amounts of material spall off the front as the<br />

flow progresses. Felsic lava breccias tend to thicken in<br />

valleys due to a tendency for increased breccia forma-<br />

tion when the lava is poised on a slope (C. R. Manley,<br />

personal communication, 1988). Because the breccia is<br />

a product of the particular emplacement mechanism,<br />

the occurrence of basal breceias throughout most of the<br />

extent of a unit argues that it flowed as a lava over its<br />

entire extent.<br />

Because clast deformation in a <strong>rheomorphic</strong> tuff (as<br />

defined above) only occurs after deposition, chilling of<br />

the base of a tuff should prevent it <strong>from</strong> participating in<br />

secondary flow (and to some extent in welding) and<br />

preserve its pyroclastic character. The basal flow unit in<br />

the Barrel Springs Formation in Texas, the most in-<br />

tensely and pervasively <strong>rheomorphic</strong> tuff known to us<br />

(Henry et al. 1989), provides a good example (Fig. 2).<br />

The upper 95 m of this 100-m-thick tuff is foliated,<br />

flow-banded, flow-folded, ramped, and even brec-<br />

ciated; the only indication of a pyroclastic origin is the<br />

presence of lithic fragments in low to moderate abun-<br />

dance. However, the 5-m-thick basal facies contains un-


equivocal evidence of the pyroclastic origin of the unit;<br />

it grades upward <strong>from</strong> poorly through densely welded<br />

to <strong>rheomorphic</strong> tuff, contains easily recognizable pu-<br />

mice (fiamme where densely welded), and has abun-<br />

dant crystal- and lithic-enriched gas elutriation pipes.<br />

Distinction between the base of a lava and a lava-<br />

like tuff may be more difficult. Within-flow agglutina-<br />

tion prior to deposition, and, presumably development<br />

of a significant temperature profile through the deposit,<br />

could obliterate most evidence of pyroclastic origin,<br />

even at the basal contact. Ekren and others (1984) sug-<br />

gested that subsequent <strong>rheomorphic</strong> flow could brec-<br />

ciate the base and that this process generated basal<br />

breccia in what they interpreted to be lava-like <strong>tuffs</strong> in<br />

southwestern Idaho (but interpreted by us as <strong>lavas</strong>).<br />

The process is essentially the same as brittle failure and<br />

brecciation due to intense shearing within the interior<br />

of a lava flow (Bonnichsen and Kauffman 1987; Man-<br />

ley and Fink 1987). However, such internal breccias are<br />

composed only of clasts <strong>from</strong> the immediately adjacent<br />

parts of the flow. Breccias resulting <strong>from</strong> this process<br />

acting on the base of a lava-like tuff should be distin-<br />

guishable <strong>from</strong> those of a lava flow. Clasts in the brec-<br />

cia could consist only of basal agglutinated and densely<br />

welded tuff and the lowest part of the <strong>rheomorphic</strong><br />

portion. The mix of lithologies that can occur in a lava-<br />

flow breccia would not occur. A/so, such breccias, by<br />

analogy with those developed in the interior of lava-<br />

flow units, should be limited in extent and of sporadic<br />

occurrence. It is also possible that secondary flow of<br />

tuff beyond the limits of the initial pyroclastic flow<br />

deposition could generate a 'lava-flow' breccia, but<br />

such a breccia would be confined to the outer margins<br />

of the body. Not only would the distribution of such a<br />

breccia be unlike that of a true lava-flow breccia, but<br />

secondary flow is least likely in the thin, distal parts of<br />

an ignimbrite, which is also the part of the sheet with<br />

the least preservation potential.<br />

175<br />

Fig. 1. Basal breccia of <strong>extensive</strong> <strong>silicic</strong> lava,<br />

southwestern Idaho (Bonnichsen and Kauff-<br />

man 1987), formed by collapse of chilled flow<br />

front. Breccia clasts include a variety of lava<br />

textural types (massive, flow-banded and<br />

folded, vesicular, glassy, and devitrified) in a<br />

finely clastic matrix. Such a deposit is diagnos-<br />

tic of a lava-flow origin<br />

A basal coarse-clast accumulation layer of an ash-<br />

flow tuff (such as the ground layer of Walker et al.<br />

198I) might be difficult to distinguish <strong>from</strong> a basal<br />

breccia of a lava if the dominant 'lithics' were poorly<br />

vesiculated magma lumps rather than accidental frag-<br />

ments. However, in most cases, clast variety and shape<br />

serve to distinguish the two breccia types. Additionally,<br />

clasts in a basal accumulation layer should show a sys-<br />

tematic size variation related to distance <strong>from</strong> source<br />

and to the size of clasts (if discernible) in the overlying<br />

part of the unit, which would not be the case in a basal<br />

lava breccia.<br />

Basal breccias are locally absent in some <strong>silicic</strong> lav-<br />

as, particularly near source (e.g. Figure 9 of Bonnichsen<br />

and Kauffman 1987), and are rare and, where present,<br />

thin and discontinuous in many of the postulated lava-<br />

like <strong>tuffs</strong> (Table 2; Fig. 3; Milner et al. 1992; CDH, per-<br />

sonal inspection). The bases of these lava-like units are<br />

mostly flow-banded, flow-folded, and slightly vesicu-<br />

lar; they resemble bases of many rapidly emplaced,<br />

low-viscosity basaltic <strong>lavas</strong> (e.g. Figure 9b of Green<br />

1989). Therefore, absence of basal breccia is not evi-<br />

dence for either origin. We note that basal breccia is<br />

least common in units <strong>from</strong> those provinces with the<br />

highest reported magmatic temperatures (Table 2) and<br />

therefore the lowest probable viscosities. Whether or<br />

not flow-banded and folded, unbrecciated bases can<br />

also form <strong>from</strong> pyroclastic flows is not established.<br />

Also, prolonged heating by an overlying thick and hot<br />

flow interior may homogenize initial basal breccia (C.<br />

R. Manley, personal communication, 1990).<br />

The presence of some pyroclastic material at the<br />

base of a unit is not proof of either origin. Pyroclastic<br />

eruptions commonly precede lava emplacement in<br />

small <strong>silicic</strong> domes and flows (Fink 1983; Heiken and<br />

Wohletz 1987). The eruptions produce air-fall or surge<br />

deposits, which may contain large clasts of volatile-<br />

poor lava. Bonnichsen and Kauffman (1987) suggested


176<br />

~,. .:~<br />

,.,'. On'.'~..~<br />

3_: W"5 if.O:<br />

-~a W, .'t~.. o' ,~,<br />

"" "- b.*<br />

z~o~o<br />

~. :o.:r ~.<br />

r.O<br />

t,..<br />

N<br />

Irregular top~ either eroded or primary top<br />

of <strong>rheomorphic</strong> tuff<br />

Glossy to devitrified <strong>rheomorphic</strong> flow breccia;<br />

blocks to 4m, commonly flow-bonded glass is<br />

hydrated portly to enlirely spherulitically de-<br />

vitrified matrix consisls of finer closts hy-<br />

drated 1o devtrified and commonly s lcified,<br />

alteration increases upward<br />

Massive, flow-bonded, devitrified zone within<br />

breccia<br />

Not exposed<br />

Massive, devitrified, flow-banded <strong>rheomorphic</strong><br />

tuff containing lithics<br />

fleginning of brecciafion<br />

Beginning of flow folding<br />

Massive, devitrified, folioted <strong>rheomorphic</strong><br />

tuff containing Iithics<br />

Basal ash-flow tuff; phenocrysts of alkali<br />

feldspar and clinopyroxene (5-10%); sparse<br />

dark pumice and lithics; gas-escape pipes;<br />

welding increases upward<br />

Fig. 2. Section through <strong>rheomorphic</strong> tuff of the Barrel Springs<br />

Formation, Trans-Pecos Texas, the most intensely <strong>rheomorphic</strong><br />

tuff known to us. Basal 5 m of unit preserves abundant ash-flow<br />

features, including shards, pumice, lithic fragments, and gas elu-<br />

triation pipes. Above a narrow transition interval, the rock is fol-<br />

iated. Pumice and shards are obliterated. However, the rock is<br />

chemically and mineralogically identical to the tuff, and lithic<br />

fragments, identical to those in the tuff, are common. Upward, the<br />

rock is flow-banded, folded and brecciated. The upper half of the<br />

unit is a breccia identical to those of lava flows; clasts include a<br />

variety of textural types of the unit in a matrix of finer clasts<br />

that volcanic ash beneath two large-volume <strong>silicic</strong> <strong>lavas</strong><br />

in Idaho represents near-vent fall deposits. Fusion of<br />

initial pyroclastic deposits by later lava (e.g. Chris-<br />

tiansen and Lipman 1966), particularly if the latter is<br />

massive at its base, could result in a very misleading<br />

apparent increase in degree of 'welding' upward to co-<br />

herent rock. Similar complications result when second-<br />

ary pyroclastic flows are generated at the front of an<br />

advancing lava, as described by Fink and Manley<br />

(1989). In this process, sudden collapse of the lava-flow<br />

front exposes the volatile-overpressured interior of the<br />

flow, with a resultant explosion producing a pyroclastic<br />

flow. Deposits <strong>from</strong> the pyroclastic flow may be over-<br />

ridden and fused by the moving lava (Fig. 4), resulting<br />

in an apparent welding/rheomorphism zonation.<br />

In turn, the absence of air-faU or surge deposits at<br />

the base of a unit is not evidence for a lava origin. Py-<br />

m<br />

5O<br />

40<br />

30<br />

20<br />

I0<br />

roclastic flows may move beyond the extent of any pre-<br />

cursor deposits, and in any case explosive eruptions of<br />

volatile-poor magmas commonly do not involve an ini-<br />

tial plinian phase (Sparks and Wilson 1976).<br />

Interiors and tops of units<br />

Interiors and tops of units provide the least useful in-<br />

formation to distinguish <strong>rheomorphic</strong> <strong>tuffs</strong> <strong>from</strong> exten-<br />

sive <strong>silicic</strong> <strong>lavas</strong>. The interior of a <strong>strongly</strong> <strong>rheomorphic</strong><br />

tuff will appear most lava-like because secondary flow<br />

is most intense there. Flow bands, flow folds, ramp<br />

structures, and elongate vesicles are common in the up-<br />

per parts of many easily recognizable <strong>rheomorphic</strong><br />

<strong>tuffs</strong> (Fig. 2; also, Chapin and Lowell 1979; Wolff and<br />

Wright 1981; Orsi and Sheridan 1984; Hargrove and<br />

Sheridan 1984; Parker 1986). Additionally, intense re-<br />

crystallization within the interior of a flow commonly<br />

disguises primary textures.<br />

Upper breccias, in contrast to basal breccias, are not<br />

diagnostic of emplacement mechanism, although they<br />

are undoubtedly less common in <strong>rheomorphic</strong> <strong>tuffs</strong><br />

than in <strong>lavas</strong>. The only example of an upper breccia in<br />

a <strong>rheomorphic</strong> tuff known to us is in the <strong>strongly</strong> rheo-<br />

morphic flow unit in the Barrel Springs Formation (Fig.<br />

2; see also Henry et al. 1989). This unit contains an up-<br />

per breccia as much as 40 m thick that developed dur-<br />

ing extreme secondary flow. Upper breccias are also<br />

common in rocks interpreted to be lava-like <strong>tuffs</strong> (Table<br />

2; Ekren et al. 1984; Milner et al. 1992).<br />

Evidence of welding zonation, particularly including<br />

cooling breaks, or of upper nonwelded or poorly<br />

welded tuff, would be diagnostic of a pyroclastic origin.<br />

The only reported examples of cooling breaks in lava-<br />

like <strong>tuffs</strong> are <strong>from</strong> southwestern Idaho (Ekren et al.<br />

1984). However, we interpret most of those examples to<br />

be contacts between lava flows. A few others (see Fig-<br />

ure 20 of Ekren et al.) appear to represent discontinui-<br />

ties where cooling joints that propagated <strong>from</strong> the bot-<br />

tom and top of a flow meet in the middle (Bonnichsen<br />

and Kauffman 1987). One <strong>rheomorphic</strong> tuff of the Ke-<br />

weenawan province has a poorly welded top (J. C.<br />

Green, personal communication, 1990). Generally, up-<br />

per nonwelded material is unlikely to be preserved and<br />

could be confused with highly vesicular upper breccia<br />

produced by lava flows.<br />

Distal thinning and unit margins<br />

Strongly <strong>rheomorphic</strong> ruffs and <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong><br />

should be most easily distinguished in distal deposits.<br />

Ash-flow <strong>tuffs</strong> emplaced on flat topography thin grad-<br />

ually toward their margins unless they encounter a to-<br />

pographic barrier (Cas and Wright 1987), although lev-<br />

ees may be present at the margin. The most <strong>strongly</strong><br />

<strong>rheomorphic</strong> tufts of Texas thin to as little as 2 m at<br />

their distal edges, where they show few or no secondary<br />

flow features (Figs. 5 and 6). In contrast, <strong>silicic</strong> <strong>lavas</strong><br />

generally maintain considerable thicknesses throughout


(a)<br />

' "~ 9 "" _~__ " "' ~ / ~ ~'~lj<br />

(b) Proximal Distal<br />

~ Massive lava<br />

~"~""~.~n Basal breccia to lava<br />

" 9 _'7".'" 9 ~"." -- -<br />

~ '."~'-" "~" :-: f.'D. " :'" ::" ~"~ i'." ~ Pyroclastic flow deposit . ii..'~. ('!'.~" .~'~-I i'.:<br />

.~ :.':c<br />

r-~ ,* 9 Lava front breccia -. ~ .~. 6t::2 :i<br />

Fig. 4. a Diagram showing collapse of lava-flow front and explo-<br />

sion of exposed, volatile-oversaturated lava interior, to produce a<br />

secondary pyroclastic flow (<strong>from</strong> Fink and Manley 1989). b Two<br />

schematic sections depicting resulting relations, after the pyro-<br />

clastic flow deposit has been overridden by the advancing lava:<br />

'proximal' section applies to the vicinity of the lava-flow front<br />

where the secondary explosion occurred; 'distal' section applies<br />

to any location further along the lava path. Note that heat <strong>from</strong><br />

the overriding lava may fuse and flatten underlying glassy clasts<br />

their distribution and have steep, abrupt flow fronts<br />

(Figs. 5 and 6; Table 2; see also Figure 23 of Bon-<br />

nichsen and Kauffman 1987). Well-preserved flow<br />

fronts in southwestern Idaho (Bonnichsen and Kauff-<br />

1<br />

177<br />

Fig. 3. Rare occurrence of basal breccia in unit<br />

3, rhyolitic lava-like tuff of the Jozini Forma-<br />

tion, Lebombo Monocline, South Africa (Bris-<br />

tow and Cleverly 1979). Base is massive on left<br />

half of photo. On right is thin breccia, 10 to<br />

40 cm thick, composed of vesicular clasts.<br />

Rhyolite overlies basalt<br />

man 1987), Yellowstone National Park (Christiansen<br />

and Hildreth 1989), and Texas (Henry et al. 1989) range<br />

<strong>from</strong> a minimum of 25 m to as much as 200 m high.<br />

Most lava-like ruffs remain thick throughout their dis-<br />

tribution (Table 2; Milner et al. 1992; CDH, personal<br />

inspection), which we consider to be consistent with a<br />

lava-flow origin. Erosion of distal parts of a rheomor-<br />

phic welded tuff could complicate interpretation if, for<br />

example, an abrupt transition existed between rheo-<br />

morphic and non-<strong>rheomorphic</strong> parts of a tuff. Both<br />

<strong>lavas</strong> and pyroclastic flow deposits may split into indi-<br />

vidual flow lobes at their margins. Although we know<br />

of no examples, it is also possible that the <strong>rheomorphic</strong><br />

flow of a welded tuff beyond its original depositional<br />

extent could develop flow lobes and steep flow fronts.<br />

Aspect ratios<br />

The sheetlike geometry and low-aspect ratio of these<br />

<strong>silicic</strong> rock bodies have, in many cases, been the princi-<br />

pal features used to infer a pyroclastic origin (Bristow<br />

and Cleverly 1979; Cleverly and Bristow 1982; Ekren et<br />

al. 1984; Milner 1986; Henry et al. 1988; Twist et al.<br />

1989; Milner et al. 1992) and the root of controversy<br />

about emplacement mechanism (i.e. lava versus pyro-<br />

clastic flow). On this basis, aspect ratio has no value as<br />

a genetic indicator. Nevertheless, aspect ratios of well-<br />

documented <strong>lavas</strong> are larger than those of many pyro-<br />

clastic flow deposits, and, in particular, averaged thick-<br />

nesses of <strong>lavas</strong> are greater (Fig. 7). The range of thick-<br />

nesses of widespread <strong>silicic</strong> <strong>lavas</strong> is similar to that of<br />

conventional flows and domes (Fig. 7). Insufficient<br />

data exist to calculate aspect ratios for other possible<br />

<strong>lavas</strong>, but their considerable thicknesses suggest a simi-<br />

lar pattern. Comparison of the <strong>rheomorphic</strong> Buckshot<br />

Ignimbrite with the Bracks Rhyolite, an <strong>extensive</strong> <strong>silicic</strong><br />

lava, illustrates these differences (Fig. 7). Both crop out<br />

over similar areas in Texas and were emplaced on fairly


178<br />

2~ ] ~ _ ~ Mitchell Mesa Rhyolite - conventional ignimbrite<br />

2~] Gomez Tuff - <strong>rheomorphic</strong> ignimbrite<br />

OJ~<br />

0 J<br />

Silicic Lavas<br />

Bracks R h y o l i ~<br />

/ Dorsey Creek Rhyolite Covered<br />

0 10 20 30 km 20X Vertical Exaggeration<br />

Fig. 6. Longitudinal sections through a conventional ignimbrite<br />

(the Mitchell Mesa Rhyolite), a <strong>strongly</strong> <strong>rheomorphic</strong> tuff (the<br />

Gomez Tuff), and two <strong>extensive</strong> siticic <strong>lavas</strong> (the Bracks Rhyolite<br />

and Dorsey Creek Rhyolite), illustrating marked differences in<br />

distal thinning and flow margins. The first three are in Trans-<br />

Pecos Texas (our data plus Parker 1986); the Dorsey Creek<br />

Rhyolite is in southwestern Idaho (data <strong>from</strong> Bonnichsen and<br />

Kauffman 1987 and B. Bonnichsen, personal communication<br />

1989). Sections of the two ignimbrites start <strong>from</strong> near caldera<br />

sources (where they are ponded to much greater thicknesses) at<br />

the left edge of diagram. Sections of the two <strong>lavas</strong> extend through<br />

entire exposures; source of the Bracks Rhyolite is probably be-<br />

neath the thick part (Henry et al. 1990)<br />

flal topography. The Buckshot Ignimbrite has an out-<br />

crop area of about 2000 km 2 and a maximum thickness<br />

of about 20 m. The Bracks Rhyolites covers 1000 km 2,<br />

has a maximum thickness of 120 m, and is nowhere less<br />

than 25 m thick (Henry et al. 1990).<br />

The only comprehensive data for lava-like tufts are<br />

for the Etendeka sequence of Namibia (Fig. 7; Milner<br />

et al. 1992). Exceptionally low aspect ratios (1:2100)<br />

calculated for these rocks are based on lateral extents<br />

of 130 km. However, this large extent assumes correla-<br />

tion of chemically similar rocks in two areas of outcrop<br />

50 km apart. If units in these areas are considered sepa-<br />

rately, their aspects ratios would be similar to those of<br />

the <strong>silicic</strong> <strong>lavas</strong> plotted in Fig. 7. Interpreting aspect ra-<br />

tios of older and highly eroded deposits is difficult<br />

I000,<br />

Fig. 5. Contrast in thickness of distal parts of<br />

the <strong>strongly</strong> <strong>rheomorphic</strong> Gomez Tuff (area<br />

outlined by trees on plateau) and an <strong>extensive</strong><br />

<strong>silicic</strong> lava (massive cliffs) of the Star Moun-<br />

tain Formation in Trans-Pecos Texas. Tuff is<br />

approximately 2 m thick here and not rheo-<br />

morphie. Lava is about 70 m thick<br />

0a i ~b ~do 100o<br />

Diameter of a circle with an area equal to that of unit (km)<br />

Fig. 7. Aspect ratio plot of <strong>extensive</strong> silieic iavas (T, Texas, Sleep-<br />

ing Lion, Adobe Canyon, and Star Mountain flows, Henry et al.<br />

1989;Y, Yellowstone National Park, Pitehstone Plateau flow,<br />

Christiansen and Hildreth 1989; I, southwestern Idaho, Dorsey<br />

Creek and Sheep Creek flows, Bonnichsen and Kauffman 1987;<br />

A, Australia, unit B1, Cas 1978). E, the Lower Springbok unit of<br />

the Etendeka, Namibia, interpreted by Milner et al. (1992) as a<br />

rheoignimbrite. Extensive <strong>silicic</strong> <strong>lavas</strong> generally plot above or just<br />

within the field of high-aspect-ratio ignimbrites (i.e. the <strong>lavas</strong> have<br />

similar or higher ratios of thickness to extent). The Bracks Rhyol-<br />

ite (Br), an <strong>extensive</strong> silicie lava, and Buckshot Ignimbrite (Bu), a<br />

<strong>rheomorphic</strong> tuff, crop out over similar areas of Texas but have<br />

markedly different aspect ratios. Fields are FL, felsic lava; ML,<br />

marie lava; HI, large-aspect-ratio ignimbrite; and LI, small-as-<br />

pect-ratio ignimbrite<br />

(Christiansen and Hildreth 1989; Henry et al. 1989).<br />

There are several known examples of clusters of dis-<br />

crete but compositionally similar rhyolitic lava domes<br />

(Christiansen and Hildreht 1989; Henry et al. 1989;<br />

Duffield and Dalrymple 1990) that could easily be mis-<br />

identified as a single flow, particularly if flow margins<br />

have been removed by erosion.


Response to topography<br />

Ingnimbrites are produced <strong>from</strong> collapsing eruption<br />

columns (pyroclastic fountains) up to a few km in<br />

height (Smith 1960; Sparks and Wilson 1976; Sparks et<br />

al. 1978). The more violently emplaced flows (for pres-<br />

ent purposes, 'conventional' ignimbrites) have flow vel-<br />

ocities up to several hundred meters per second, high<br />

enough for them to surmount considerable topographic<br />

barriers (Sparks et al. 1978). High-temperature, rheo-<br />

morphic <strong>tuffs</strong> are inferred tO form <strong>from</strong> low, thermally<br />

conservative fountains and may flow much more slow-<br />

ly. However, even at flow velocities of 50 m/s, a pyro-<br />

clastic flow can rise more than 100 m (Sparks et al.<br />

1978) and thus should still be able to overtop modest<br />

barriers. Inflation of pyroclastic flows also allows them<br />

to be deposited on top of barriers higher than the final<br />

deposit thickness. In contrast, <strong>lavas</strong> can only flood bar-<br />

riers lower than the flow thickness; otherwise they must<br />

pond against them. Two examples <strong>from</strong> Texas illustrate<br />

this behavior. In Fig. 8, the <strong>rheomorphic</strong> tuff thins but<br />

continues over a 40-m-high lava flow front. In the sec-<br />

ond example, the <strong>rheomorphic</strong> Gomez Tuff overlies<br />

and thins on top of a rhyolite dome that is approxi-<br />

mately 100 m higher than the base of the tuff. A lava of<br />

the Adobe Canyon Formation terminated against the<br />

dome, even though the mantling Gomez Tuff had re-<br />

duced the topographic relief.<br />

Sources<br />

Large-volume ash-flow <strong>tuffs</strong>, including rheoignim-<br />

brites, are commonly associated with calderas inferred<br />

to contain their source vents. Rheomorphic <strong>tuffs</strong> in<br />

Texas can be traced back to source calderas where they<br />

pond to thicknesses many times greater than in outflow<br />

sheets. The <strong>strongly</strong> <strong>rheomorphic</strong> Barrel Springs tuff is<br />

no more than 100 m thick in outflow deposits, yet<br />

reaches a thickness of 800 m within its caldera (Hender-<br />

179<br />

Fig. 8. Strongly <strong>rheomorphic</strong> tuff of the Barrel<br />

Springs Formation in Texas illustrating re-<br />

sponse to topography. Tuff is 50 m thick at left,<br />

has a preserved pyroclastic base, and is flow<br />

banded and folded in the middle and upper<br />

parts. It flowed <strong>from</strong> left to fight over the 40-<br />

m-high flow front of a <strong>silicic</strong> lava. Above the<br />

lava, it is approximately 10 m thick, shows no<br />

rheomorphie features, and is obviously an ig-<br />

nimbrite. A <strong>silicic</strong> lava of comparable thickness<br />

encountering a similar topographic barrier<br />

would have ponded against it<br />

son 1987). The Gomez Tuff is as much as 450 m thick in<br />

the Buckhorn caldera (Parker 1986) but no more than<br />

100 m thick elsewhere. It may, however, be difficult to<br />

identify source calderas in regions of heavily dissected<br />

paleotopography, where ignimbrites may have ponded<br />

in large, steep-walled canyons.<br />

Few sources for large-volume <strong>silicic</strong> <strong>lavas</strong> have been<br />

identified, but, as with many conventional <strong>lavas</strong>, at<br />

least some appear to have erupted <strong>from</strong> fissures (Bon-<br />

nichsen and Kauffman 1987; Christiansen and Hildreth<br />

1989; Henry et al. 1989; Manley 1990). A feeder dike to<br />

a widespread <strong>silicic</strong> lava flow of the Star Mountain<br />

Formation in Texas is the best example of probable fis-<br />

sure eruption known to us (Fig. 9). The northern end of<br />

the 3-km-long dike disappears beneath and passes<br />

without interruption into the overlying flow; the dike<br />

and flow are petrographically and chemically indistin-<br />

guishable. Manley (1990) interpreted coarse tephra de-<br />

posits to flank a fissure vent for a lava of southwestern<br />

Idaho that had previously been interpreted as a lava-<br />

like tuff (Ekren et al. 1984).<br />

The sources of many other large-volume <strong>silicic</strong> <strong>lavas</strong><br />

have not been identified (Cas 1978; Twist and French<br />

1983; Bonnichsen and Kauffman 1987; Green 1989;<br />

Green and Fitz 1989). These flows maintain nearly uni-<br />

form thicknesses throughout their outcrop and clearly<br />

do not pond within calderas. If fissure-fed, the sources<br />

are likely to be buried beneath the flows, as is the case<br />

for many small rhyolite domes (Fink and Pollard 1983)<br />

and flood basalts (Cas and Wright 1987). In the ab-<br />

sence of topographic control, elongation of outcrop<br />

may suggest eruption <strong>from</strong> a fissure. Linear joints that<br />

are associated with vents for some <strong>silicic</strong> domes (Fink<br />

and Pollar 1983) may also indicate an underlying fis-<br />

sure system.<br />

The sources of lava-like <strong>tuffs</strong> are also equivocal.<br />

Ekren et al. (1984) suggested the lack of calderas asso-<br />

ciated with lava-like <strong>tuffs</strong> of southwestern Idaho indi-<br />

cated the <strong>tuffs</strong> erupted <strong>from</strong> great depths. In contrast,<br />

Milner et al. (1992) suggest a possible caldera source


180<br />

for the Etendeka rocks. No sources have been identif-<br />

ied for lava-like tufts of the Jozini Formation (Lebom-<br />

bo) in South Africa (Bristow and Cleverly 1979). How-<br />

ever, a 10-m-thick dike, petrographically similar to the<br />

Jozini rhyolites, cuts one flow (J. W. Bristow, personal<br />

communication, 1990). Its relation to the rhyolites is<br />

uncertain.<br />

Recognition of distinctive sources seems, empirical-<br />

ly, to be a good indicator of origin. However, there is<br />

no clear reason why a voluminous, <strong>extensive</strong> lava<br />

should not be erupted <strong>from</strong> and pond within a ealdera,<br />

nor why large volumes of similarly volatile-poor <strong>silicic</strong><br />

magma should not explosively erupt <strong>from</strong> fissures unre-<br />

lated to calderas. Indeed, many of the Yellowstone<br />

<strong>lavas</strong> are ponded whitin an existing caldera (Chris-<br />

tiansen and Hildreth 1989). The eruptive style of a hot<br />

pyroelastic flow fed <strong>from</strong> a low fountain is little differ-<br />

ent <strong>from</strong> that a spatter-fed lava; which type of rock<br />

body results may depend only on the proportion and<br />

retention of fine particles (enabling partial fluidisation<br />

of a pyroclastic flow) produced in the fountain (Wolff<br />

and Horn 1991). Much more work is required to prop-<br />

erly understand the eruptive mechanisms for these<br />

rocks and the nature and location within the crust of<br />

their source magma bodies.<br />

Petrooraphic criteria<br />

It is often assumed that petrographic features such as<br />

shards, pumice, lithic fragments, and broken pheno-<br />

crysts provide unequivocal evidence of pyroclastic ori-<br />

gin. The presence of obscure shard- and pumice-like<br />

features at the base or top of some rock units has been<br />

interpreted as evidence of pyroclastic origin for several<br />

lava-like tufts (Fig. 10; Bristow and Cleverly 1979;<br />

Ekren et al. 1984; Milner 1986; Milner et al. 1992).<br />

However, these features are poorly preserved or even<br />

lacking in many <strong>rheomorphic</strong> tufts due to secondary<br />

flow, recrystallization, and erosion. Also, <strong>lavas</strong> com-<br />

monly develop textures that can be confused with those<br />

Fig. 9. Feeder dike to <strong>extensive</strong> <strong>silicic</strong> lava of<br />

the Star Mountain Formation (background) in<br />

Texas. Dike is petrographically and chemically<br />

identical to lava. Flow bands in dike are verti-<br />

cal and turn abruptly horizontal at level of<br />

flow<br />

of pyroclastic flows (e.g. Manley and Fink 1987; Allen<br />

1988). Thus it is imperative to understand exactly how<br />

these features form and where they can occur and be<br />

preserved in high-temperature <strong>tuffs</strong> and <strong>lavas</strong>.<br />

Magma vesiculation and fragmentation are integral<br />

parts of the eruption of pyroclastic flows. Therefore,<br />

pumice and shards can occur throughout an ash-flow<br />

tuff. However, several factors work to minimize their<br />

abundance and preservation in high-temperature tufts.<br />

First, the inferred low volatile content and low explo-<br />

sivity of eruptions that produce high-temperature pyro-<br />

elastic flows should minimize the amount of vesicula-<br />

tion and breakage. The abundance of shards and pu-<br />

mice relative to dense magma lumps is probably much<br />

less than in low-temperatue, more volatile-rich, pyro-<br />

elastic flows. Additionally, the dense welding that com-<br />

monly occurs throughout high-temperature ash-flow<br />

<strong>tuffs</strong>, even those that do not undergo sceondary flow,<br />

tends to disguise primary pyroclastie texture. Finally,<br />

secondary flow and intense crystallization during devi-<br />

trification can completely obliterate pumice and<br />

shards.<br />

In the strictest sense, pumice and shards indicate<br />

only that a magma vesiculated and fragmented. Lavas<br />

vesiculate, in some examples throughout a flow, and<br />

brecciate at their margins. Manley and Fink (1987) de-<br />

scribe coarsely vesicular and finely vesicular pumice in<br />

unequivocal <strong>silicic</strong> <strong>lavas</strong> that are virtually indistinguish-<br />

able <strong>from</strong> pumice of pyroclastic flows. Highly vesicular,<br />

pumiceous clasts are common in upper and lower brec-<br />

cias of all <strong>silicic</strong> <strong>lavas</strong>; Fink (1983), Fink and Manley<br />

(1987), Manley and Fink (1987), and Sampson (1987)<br />

provide excellent photographs of pumiceous textures<br />

<strong>from</strong> conventional, small-volume <strong>silicic</strong> <strong>lavas</strong>. Attrition<br />

of this pumice can yield a matrix composed of glass<br />

shards. However, these breccias commonly contain a<br />

spectrum of textural types that additionally include<br />

massive, flow-banded, devitrified, or vitrophyric clasts<br />

(Fig. 1). This assemblage of lithologies is, in fact, diag-<br />

nostic of a lava flow despite the presence of pumice.


Lensoid features resembling flattened pumice<br />

(fiamme) can develop in <strong>silicic</strong> <strong>lavas</strong>, although the<br />

mechanisms by which they form are not completely un-<br />

derstood. One mechanism is plastic deformation and<br />

stretching, presumably at low strain rates, of clasts in<br />

internal breccia zones originally produced by brittle<br />

failure at high strain rates. Zones of 'pseudo-fiamme'<br />

formed in this way are not systematically distributed<br />

with respect to flow margins, unlike true fiamme in<br />

<strong>rheomorphic</strong> and lava-like <strong>tuffs</strong>, which should be most<br />

abundant around flow tops, bases, and margins. Incom-<br />

plete magma mixing can also produce fiamme-like tex-<br />

tures (see Fig. 6b in Sampson 1987).<br />

Alteration processes may also produce pseudo-pyro-<br />

clastic textures in <strong>lavas</strong>. Allen (1988) describes <strong>silicic</strong><br />

<strong>lavas</strong> in Australia in which hydrothermal alteration of<br />

initially massive, glassy material with abundant perlitic<br />

cracks produced textures that closely mimicked shards.<br />

He also noted that granulation, including dismember-<br />

ment of phenocrysts, by tectonic processes accentuated<br />

this false pyroclastic appearance. Fiamme-like struc-<br />

tures also develop in weakly metamorphosed and tec-<br />

tonized <strong>silicic</strong> <strong>lavas</strong> (Cas and Wright 1987; Allen 1988).<br />

Although Allen's examples were exclusively in areas of<br />

significant hydrothermal alteration, similar textures<br />

were produced in Jozini Rhyolites by normal deuteric<br />

alteration (J. W. Bristow, personal cummunication,<br />

1990).<br />

We conclude <strong>from</strong> this discussion that the presence<br />

of pumice and shards throughout a flow is proof of a<br />

pyroclastic origin, but such preservation should be un-<br />

common in high-temperature, <strong>rheomorphic</strong> <strong>tuffs</strong>. The<br />

presence of pumice and shards only in breccias at the<br />

base or top of a flow is at least equally consistent with<br />

origin as a lava flow. It is precisely the lack of preserva-<br />

tion of these textures in either <strong>rheomorphic</strong> <strong>tuffs</strong> or ex-<br />

tensive <strong>silicic</strong> <strong>lavas</strong> that complicates interpetration. In<br />

fact, preservation of recognizable pumice and shards in<br />

a <strong>rheomorphic</strong> or lava-like tuff suggests that primary<br />

181<br />

Fig. 10. 'Streaky' texture near base of lava-like<br />

tuff, unit 6 of Jozini Formation, Lebombo<br />

Monocline, South Africa (Bristow and Cleverly<br />

1979). This texture may represent highly<br />

stretched pumice (fiamme) as a result of pri-<br />

mary laminar or secondary flow of an ash-flow<br />

tuff. See text for discussion<br />

laminar or secondary flow was, at most, minor. The py-<br />

roclastic origin of such a deposit should be obvious.<br />

The presence of lithic fragments, characteristic of<br />

conventional ash-flow tufts, is not diagnostic but has<br />

some utility in distinguishing <strong>rheomorphic</strong> tufts <strong>from</strong><br />

<strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>. Lithic fragments in pyroclastic<br />

flow deposits are derived <strong>from</strong> erosion of event walls<br />

during eruption (Wilson et al. 1980) or are picked up<br />

<strong>from</strong> the ground across which the flow travels. In the<br />

whole spectrum of explosive <strong>silicic</strong> eruptions, vent ero-<br />

sion is probably at a minimum for the weakly explosive<br />

eruptions of gas-poor magmas envisaged to generate<br />

<strong>strongly</strong> <strong>rheomorphic</strong> and lava-like <strong>tuffs</strong>. Lithic abun-<br />

dances should generally be, and indeed are, low in such<br />

<strong>tuffs</strong> (authors' observations). Most examples that we<br />

have studied have no more than 1-2% lithics, but one<br />

peralkaline rhyolitic <strong>rheomorphic</strong> tuff on Gran Canar-<br />

ia, Canary Islands, locally contains 5-10% lithics (JAW,<br />

personal inspection). Silicic <strong>lavas</strong> may also contain lit-<br />

hic fragments, for example the Quaternary Banco Bon-<br />

ito flow of the Jemez Mountains, New Mexico (Self et<br />

al. 1991). If, as suggested above, some widespread<br />

<strong>silicic</strong> <strong>lavas</strong> are spatter-fed <strong>from</strong> pyroclastic fountains<br />

similar to those that feed hot, intensely <strong>rheomorphic</strong> ig-<br />

nimbrites, then a similar lithic content might be ex-<br />

pected. Also, autoliths or magma inclusions, both of<br />

which can be common in <strong>lavas</strong> (Bacon 1986), must be<br />

distinguished <strong>from</strong> true foreign fragments.<br />

The location of lithic fragments or inclusion within a<br />

unit may also be helpful. Large lithics are typically con-<br />

centrated near the base of a pyroclastic-flow deposit,<br />

including individual flows within compound ignim-<br />

brites. Secondary flow may redistribute them slightly<br />

but should not disperse them. Distinct lenses of lithics<br />

within the <strong>rheomorphic</strong> Gomez Tuff parallel secondary<br />

flow foliation and probably mark internal flow boun-<br />

daries. In contrast, inclusions transported by a lava<br />

may be more randomly distributed through the interior<br />

of a flow. A quartz latite of the Etendeka province in


182<br />

Namibia (interpreted by Milner et al. 1992, as a rheoig-<br />

nimbrite) contains quartzite inclusions as much as<br />

50 cm in diameter scattered apparently at random<br />

throughout it. Margins of the inclusions are delicately<br />

cuspate (Fig. 11), indicating significant dissolution. The<br />

distribution and texture of these inclusions are better<br />

accounted for by transport in a lava rather than in a<br />

pyroclastic flow.<br />

Phenocryst morphology also provides equivocal evi-<br />

dence of origin. Broken phenocrysts are typical of<br />

many 'conventinal' <strong>tuffs</strong>, whereas euhedral phenocrysts<br />

and glomerocrysts are more common in <strong>lavas</strong>. Howev-<br />

er, phenocryst breakage during explosive eruption is<br />

presumably controlled by magmatic viscosity and vola-<br />

tile content; broken phenocrysts will be uncommon in<br />

eruptives produced <strong>from</strong> relatively hot, fluid, and gas-<br />

poor magmas. For example, liberated phenocrysts in<br />

mafic scoria accumulations are rarely broken. Observa-<br />

tions of <strong>rheomorphic</strong> <strong>silicic</strong> <strong>tuffs</strong> are also consistent<br />

with this conclusion. Broken phenocrysts are sparse in<br />

the rheomorhic Green Tuff of Pantelleria (JAW, per-<br />

sonal inspection); the <strong>strongly</strong> <strong>rheomorphic</strong> Barrel<br />

Springs tuff contains many complex glomerocrysts that<br />

could not have survived <strong>strongly</strong> explosive eruption<br />

(Henry et al. 1989).<br />

Contrasting crystallization textures that result <strong>from</strong><br />

devitrification of glass versus primary crystallization of<br />

homogeneous liquid constitute promising but still<br />

poorly understood criteria (Twist and Elston 1989).<br />

Dendritic or swallowtail crystallites indicate rapid crys-<br />

tallization at high degrees of undercooling <strong>from</strong> super-<br />

heated melts (Lofgren 1980; Twist and Elston 1989).<br />

Such texture can only form <strong>from</strong> liquids, not by devi-<br />

trification, and thus may be more indicative of a lava-<br />

flow origin. Examples have been found in very old (2-<br />

Ga Bushveld Complex; Twist and French 1983) and<br />

young (37-Ma Bracks Rhyolite, Texas; Henry et al.<br />

1990) <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>.<br />

Compositional homogeneity<br />

Fig. 11. Quartzite inclusion in the Upper<br />

Springbok unit, Etendeka Formation, Namibia,<br />

interpreted as a rheoignimbrite (Milner et ai.<br />

1992). Inclusions up to 50 cm in diameter are<br />

widely dispersed through the interior of the<br />

flow and commonly have delicately cuspate<br />

margins, indicating considerable dissolution.<br />

See text for discussion<br />

Given our current understanding, chemical composi-<br />

tion is not a diagnostic feature. Nevertheless, many<br />

studies have established that pyroclastic deposits are<br />

commonly compositionally zoned, whereas <strong>lavas</strong> tend<br />

to be homogeneous. However, there are numerous ex-<br />

ceptions to both generalizations. Preservation of any<br />

compositional heterogeneity <strong>from</strong> magma chamber to<br />

surface eruptive unit depends on many factors, espe-<br />

cially the nature of the pre-eruptive magma body and<br />

the mechanics of extraction (Wolff et al. 1990, and ref-<br />

erences therein). Without wishing to propose composi-<br />

tional variability as a diagnostic criterion, we can state<br />

that individual <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> and units de-<br />

scribed as lava-like ignimbrites are strikingly homoge-<br />

neous (Ekren et al. 1984; Bonnichsen and Kauffman<br />

1987; Christiansen und Hildreth 1989; Henry et al.<br />

1989, 1990; Milner et al. 1992). These rocks include tra-<br />

chytes, trachydacites, dacites, low-silica rhyolites, and,<br />

more rarely, high-silica rhyolites, or a silica range of<br />

about 65 to 77%. Most commonly, they contain about<br />

68 to 70% SiO2 and can be mildly peraluminous to<br />

mildly peralkaline. The bimodal association of most ex-<br />

tensive <strong>silicic</strong> <strong>lavas</strong> or lava-like ignimbrites with tho-<br />

leiitic flood basalts suggests the <strong>silicic</strong> rocks represent<br />

deep crustal melts (Bellieni et al. 1984; Bonnichsen and<br />

Kauffman 1987; Green 1989; Wolff 1989). Large de-<br />

grees of partial melting potentially could generate large<br />

volumes of homogeneous magma. Examples of <strong>strongly</strong><br />

compositionally zoned <strong>rheomorphic</strong> <strong>tuffs</strong> are the peral-<br />

kaline units of Pantelleria and Gran Canaria (Mahood<br />

and Hildreth 1986; Crisp and Spera 1987).<br />

The Bracks Rhyolite is a striking example of a ho-<br />

mogeneous lava (Henry et al. 1988, 1990). Fifteen sam-<br />

ples representing the complete lateral and vertical ex-<br />

tent of the flow contain about 69% SiO2 and are chemi-<br />

cally and mineralogically indistinguishable. Although<br />

concentrations of some of the more mobile major ox-<br />

ides, such as the alkalis, vary slightly as a result of mi-


nor post-emplacement alteration, concentrations of<br />

more immobile elements, including the rare-earth ele-<br />

ments, vary no more than analytical uncertainty. The<br />

abundance and compositions of phenocrysts (alkali<br />

feldspar, clinopyroxene, fayalite, and magnetite) are<br />

identical throughout the flow.<br />

Discussion<br />

Table 3 presents a revision of criteria for distinguishing<br />

<strong>rheomorphic</strong> <strong>tuffs</strong> <strong>from</strong> <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>. Several<br />

points must be emphasized. Many individual character-<br />

istics listed in Table 3, commonly thought of as diag-<br />

nostic of one rock type or the other, are found in both<br />

<strong>rheomorphic</strong> <strong>tuffs</strong> and <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>. They can-<br />

not be used to distinguish the two. Some characteristics<br />

require knowledge of the complete distribution of a<br />

volcanic unit, which can greatly complicate interpreta-<br />

tion where correlation is uncertain for any reason. Few<br />

if any features alone are truly diagnostic. All must be<br />

carefully evaluated in the overall context of the rock<br />

being examined.<br />

The features that seem most useful in distinguishing<br />

<strong>rheomorphic</strong> tufts and <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> are basal<br />

deposits, flow margins, response to topography, and,<br />

183<br />

where distribution is fully determined, source. The<br />

characteristics of basal deposits are probably most di-<br />

agnostic and most easily determined for many units.<br />

Basal deposits quench against the ground, are most<br />

likely to preserve primary features, and can be observed<br />

in single outcrops. Flow margins are also revealing.<br />

Silicic <strong>lavas</strong> typically maintain thicknesses of several<br />

tens of meters to their flow fronts. Ash-flow <strong>tuffs</strong> thin<br />

to at most a few meters and generally do not show see-<br />

ondary flow features at their margins. Tufts, even those<br />

generated <strong>from</strong> low eruption columns, can be deposited<br />

across topographic barriers. Lavas pond against bar-<br />

riers unless the flow is much thicker than the height of<br />

the barrier. If an ash-flow tuff tuff can be traced to a<br />

source caldera, a thick correlative intracaldera fill<br />

should be present.<br />

<strong>Distinguishing</strong> lava-like <strong>tuffs</strong> <strong>from</strong> other widespread<br />

<strong>silicic</strong> volcanic rocks is hampered by the lack of agreed-<br />

upon examples on which to base criteria. Based on our<br />

criteria, we suggest that most of the described examples<br />

in the literature are in fact <strong>lavas</strong>, and that the category<br />

of 'lava-like tuff' may not be valid. However, much<br />

more research is needed on possible eruption and em-<br />

placement mechanisms, especially the mechanics of<br />

low pyroclastic fountains, agglutination, and particle<br />

collisions within moving pyroclastic flows (Freundt and<br />

Schmincke 1990; Wolff and Horn 1991).<br />

Table 3. Proposed distinguishing characteristics of <strong>rheomorphic</strong> <strong>tuffs</strong>, lava-like <strong>tuffs</strong>, and <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong><br />

Characteristic Rheomorphic tuff Lava-like tuff Extensive <strong>silicic</strong> lava<br />

Wide areal extent Yes Yes<br />

Large aspect ratio Yes Yes<br />

Flow margins Distal thinning to poorly or nonwelded<br />

deposits, but dependent on topography<br />

Same as <strong>rheomorphic</strong> tuff?<br />

Response to topography Can climb over many barriers, dependent<br />

on gas content and height of eruption<br />

column<br />

Same as <strong>rheomorphic</strong> tuff?<br />

Source<br />

Commonly caldera; tuff can pond within<br />

caldera<br />

Same as <strong>rheomorphic</strong> tuff<br />

Welding zonation Yes, but can be obscured by rheomor- ?<br />

phism; high-T or peralkaline <strong>tuffs</strong> can<br />

be welded throughout<br />

Basal breccia No, but basal lithic concentration zone Same as <strong>rheomorphic</strong> tuff<br />

could appear similar<br />

Upper breccia Yes, in rare cases Same as <strong>rheomorphic</strong> tuff<br />

Flow bands Yes Yes<br />

Ramp structures Yes Yes<br />

Elongate vesicles Yes Yes<br />

Pumice Yes, but diagnostic only if throughout Unlikely to be perserved<br />

flow; commonly obscured in interior except possibly at top<br />

of flow<br />

Shards Yes, same as pumice Same as <strong>rheomorphic</strong> tuff<br />

Gas escape pipes Yes No?<br />

Broken phenocrysts with Can be present but breakage is probably Same as <strong>rheomorphic</strong> tuff<br />

wide size range minor in low-explosivity <strong>tuffs</strong><br />

Lithic fragments Yes, but can be sparse in low-explosivity Same as <strong>rheomorphic</strong> tuff<br />

tufts<br />

Composition Commonly vertically zoned Same as <strong>rheomorphic</strong> tuff?<br />

Yes<br />

Yes<br />

Remain thick and massive or<br />

brecciated to flow margin<br />

Stopped by most barriers un-<br />

less much less than flow<br />

thickness<br />

Commonly dike/fissure; lava<br />

maintains uniform thickness<br />

throughout distribution<br />

No<br />

Yes, composed of mixed tex-<br />

tural types of the lava<br />

Yes<br />

Yes<br />

Yes<br />

Yes<br />

Yes, particularly in marginal<br />

breccias<br />

Yes, in marginal breccias<br />

No<br />

No, but could develop in<br />

breccia or in tectonically<br />

disrupted <strong>lavas</strong><br />

Can contain inclusions<br />

Commonly homogeneous


184<br />

Recognition of <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> is new, and<br />

their distinction <strong>from</strong> <strong>rheomorphic</strong> <strong>tuffs</strong> is still to some<br />

extent controversial. The criteria presented here should<br />

be considered our best current understanding, which<br />

will certainly evolve as the rocks are more thoroughly<br />

studied.<br />

Concluding remarks<br />

High-temperature <strong>silicic</strong> volcanic rocks occur in many<br />

of the world's major volcanic provinces (Table 2). They<br />

include low-silica rhyolites, dacites, trachydacites, or<br />

traehytes that typically contain a low percentage of an-<br />

hydrous phenoerysts and range <strong>from</strong> mildly peralumi-<br />

nous to mildly peralkaline. Most are temporally and<br />

spatially associated with flood basalts (Table 2). Trans-<br />

Peeos Texas is exceptional, inasmuch as the surface ex-<br />

pression of marie magmatism is minor. However, pre-<br />

liminary geochemical data <strong>from</strong> this province suggest<br />

that the <strong>silicic</strong> magmas were produced by melting of<br />

anhydrus continental crust, on a regional scale, by mas-<br />

sive intrusion of marie magma into the lower crust<br />

(Wolff and Davidson 1990). This model has obvious<br />

appeal for other high-temperature <strong>silicic</strong>-magma prov-<br />

inces (Wolff 1989). Many more geochemical data are<br />

required on this class of rocks before this suggestion<br />

can be properly tested.<br />

The emplacement mechanisms for the rocks of Table<br />

2, whether turfs or <strong>lavas</strong>, are controversial. All have<br />

wide areal extents and low aspect ratios, and many<br />

have pumice- or shad-like structures in basal zones.<br />

However, we do not consider these features to be diag-<br />

nostic. Nevertheless, some units in each province have<br />

characteristics that we consider diagnostic of emplace-<br />

ment as <strong>lavas</strong>, particularly the occurrence, albeit minor<br />

in some cases, of basal breccia. Additionally, units in<br />

all these areas maintain large and relatively uniform<br />

thicknesses throughout their known distribution, al-<br />

though steep, abrupt flow margins are known only in<br />

the first three examples. Finally, rocks initially inter-<br />

preted as lava-like <strong>tuffs</strong> of the Lebombo and Rooiberg<br />

regions of South Africa, two provinces in which basal<br />

breceias are minor, have recently been reinterpreted as<br />

<strong>lavas</strong> (Twist et al. 1989). We emphasize that some of the<br />

original workers <strong>strongly</strong> disagree with our interpreta-<br />

tion that many of these are <strong>lavas</strong>. Both <strong>lavas</strong> and<br />

<strong>strongly</strong> <strong>rheomorphic</strong> <strong>tuffs</strong> may occur within individual<br />

provinces. Thus, the presence of one does not preclude<br />

the presence of the other. Nevertheless, we suggest that<br />

<strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong> are far more common than pre-<br />

viously recognized and that they are an integral part of<br />

most flood basalt provinces. The term "flood rhyolite'<br />

seems appropriate.<br />

Acknowledgements. We acknowledge the contributions of many<br />

geologists who generously gave us the opportunity to compare<br />

rocks in other areas with those in Texas, especially Bill Bon-<br />

nichsen (southwestern Idaho), Simon Milner, Andy Duncan, and<br />

Tony Ewart (Etendeka, Namibia), John Bristow (Lebombo, South<br />

Africa), and David Twist (Rooiberg Felsites, Bushveld Complex).<br />

Collaboration with Steve Self, Richard Franklin, Don Parker, and<br />

Jon Price greatly aided our research in Texas. The views of these<br />

people do not match ours in all cases. Discussions with Curtis<br />

Manley, and thorough reviews by A. Ewart, J. McPhie, S. C. Miln-<br />

er, and A. J. R. White are greatly appreciated. Research was sup-<br />

ported (CDH) by the COGEOMAP program of the US Geologi-<br />

cal Survey through Agreement No. 14-08-0001-A0516 and by<br />

Grant Gl144148 <strong>from</strong> the US Bureau of Mines to the Texas Min-<br />

ing and Mineral Resources Research Institute.<br />

References<br />

Allen RL (1988) False pyroclastic textures in altered <strong>silicic</strong> <strong>lavas</strong>,<br />

with implications for volcanic-associated mineralization. Econ<br />

Geol 83:1424-1446<br />

Bacon CR (1986) Magmatic inclusions in <strong>silicic</strong> and intermediate<br />

volcanic rocks. J Geophys Res 91:6091-6112<br />

Bellieni G, Brotzu P, Comin-Chiaramonti P, Ernesto M, Melfi A,<br />

Pacca IG, Piccirillo EM (1984) Flood basalt to rhyolite suites<br />

in the southern Parana Plateau (Brazil): palaeomagnetism, pe-<br />

trogenesis and geodynamic implications. J Petrol 25: 579-618<br />

Bonnichsen B, Kauffman DF (1987) Physical features of rhyolite<br />

lava flows in the Snake River Plain volcanic province, south-<br />

western Idaho. In: Fink JH (ed) The emplacement of <strong>silicic</strong><br />

domes and lava flows. Geol Soc Amer Sp Pap 212:119-145<br />

Bristow JW (1989) Retracing Vulcan's fiery footsteps. Nuclear Ac-<br />

tive 41:30-37<br />

Bristow JW, Cleverly FW (1979) Volcanology of the Lebombo<br />

Rhyolites. Geol Soc South Africa Geocongress 79:60-63<br />

Cas RAF (1978) Silicic <strong>lavas</strong> in Paleozoic flyschlike deposits in<br />

New South Wales, Australia: behavior of deep subaqueous<br />

<strong>silicic</strong> flows. Geol Soc Amer Bull 89:1708-1714<br />

Cas RAF, Wright JV (1987) Volcanic successions; modern and an-<br />

cient. Allen & Unwin, London, 528 pp<br />

Chapin CE, Lowell GR (1979) Primary and secondary flow struc-<br />

tures in ash-flow tufts of the Gribbles run paleovaIIey, centraI<br />

Colorado. In: Chapin CE, Elston WE (eds) Ash-flow tufts.<br />

Geol Soc Amer Sp Pap 180:137-154<br />

Christiansen RL, Hildreth W (1989) Voluminous rhyolitic <strong>lavas</strong> of<br />

broad extent on the Yellowstone Plateau. Continental Magma-<br />

tism Abstracts, New Mex Bur Mines Res Bull 131:52<br />

Christiansen RL, Lipman PW (1966) Emplacement and thermal<br />

history of a rhyolite lava flow near Fortymile Canyon, south-<br />

ern Nevada. Geol Soc Amer Bull 77:671-684<br />

Cleverly RW, Bristow JW (1982) Flow-banded and contorted<br />

tufts: examples <strong>from</strong> southern Africa (Lebombo Province) and<br />

western North America. Geol Soc Amer Abs w Prog 14:464<br />

Creaser R_A, White AJR (1991) Yardea Dacite - Large-volume,<br />

high-temperature felsic volcanism <strong>from</strong> the Middle Protero-<br />

zoic of South Australia. Geol 19:48-51<br />

Crisp JA, Spera FJ (1987) Pyroclastic flows and <strong>lavas</strong> of the Mo-<br />

gan and Fataga Formations, Tejeda Volcano, Gran Canaria,<br />

Canary Islands: mineral chemistry, intensive parameters, and<br />

magma chamber evolution. Contrib Mineral Petrol 96:503-<br />

518<br />

Duffield WA (1990) Eruptive fountains of <strong>silicic</strong> magma and their<br />

possible effects on the tin content of fountain-fed <strong>lavas</strong>, Taylor<br />

Creek Ryholite, New Mexico. Geol Soc Amer Sp Pap<br />

246: 251-262<br />

Duffield WA, Datrymple GB (1990) The Taylor Creek Rhyolite of<br />

New Mexico: a rapidly emplaced field of lava domes and<br />

flows. Bull Vole 52:475-487<br />

Ekren EB, Mclntyre DH, Bennett EH (1984) High-temperature,<br />

large-volume, lavalike ash-flow tufts without calderas in<br />

southwestern Idaho. US Geol Surv Prof Pap 1272<br />

Fink JH (1983) Structure and emplacement of a rhyolitic obsidian<br />

flow: Little Glass Mountain, Medicine Lake Highland, north-<br />

ern California. Geol Soc Amer Bull 94:362-280<br />

Fink JH, Manley CR (1987) Origin of pumiceous and glassy tex-<br />

tures in rhyolite flows and domes. In: Fink JH (ed) The em-


placement of <strong>silicic</strong> domes and lava flows. Geol Soc Amer Sp<br />

Pap 212:77-88<br />

Fink JH, Manley CR (1989) Explosive volcanic activity generated<br />

<strong>from</strong> within advancing <strong>silicic</strong> lava flows. In: Latter JH (ed)<br />

Voicanologic hazards: assessment and monitoring. IAVCEI<br />

Proc. in Volcanol. 1 : 169-179<br />

Fink JH, Pollard DD (1983) Structural evidence for dikes beneath<br />

<strong>silicic</strong> domes, Medicine Lake Highland, California. Geol<br />

11:458-461<br />

Franklin RR, Self S, Wolff JA (1987) Lava-like tufts of Trans-<br />

Pecos Texas: the Star Mountain Formation. EOS 68:1544<br />

Franklin RR, Wolff JA, Self S (1989) Recognition and origin of<br />

widespread <strong>silicic</strong> <strong>lavas</strong>. Geol Soc Amer Abs w Prog 21:11<br />

Freundt A, Schmincke H-U (1990) The densely welded basaltic<br />

ignimbrite P1 on Gran Canaria. Int Volc Congr abs Mainz<br />

(FRO)<br />

Green JC (1989) Physical volcanology of mid-Proterozoic plateau<br />

<strong>lavas</strong>: the Keweenawan North Shore Volcanic Group, Minne-<br />

sota. Geol Soc Amer Bull 101:486-500<br />

Green JC, Fitz TJ (1989) Large rhyolites in the Keweenawan mid-<br />

continent rift plateau volcanics of Minnesota - <strong>lavas</strong> of rheoig-<br />

nimbrites. New Mex Bur Mines Min Res Bull 131:113<br />

Hargrove HR, Sheridan MF (1984) Welded tufts deformed into<br />

mega<strong>rheomorphic</strong> folds during collapse of the McDermitt cal-<br />

dera: Nevada-Oregon. J Geophys Res 89:8629-8638<br />

Heiken G, Wohletz K (1987) Tephra deposits associated with<br />

<strong>silicic</strong> domes and lava flows. In: Fink JH (ed) The emplace-<br />

ment of <strong>silicic</strong> domes and lava flows. Geol Soc Amer Sp Pap<br />

212:55-76<br />

Henderson GD (1987) Intracaldera kaolinization and silicification<br />

of ash-flow tuff, Medley area, Davis Mountains, Texas. Geol<br />

Soc Amer Abs w Prog 19:170<br />

Henry CD, Price JG, Rubin JN, Parker DF, Woff JA, Self S,<br />

Franklin R, Barker DS (1988) Widespread, lava-like volcanic<br />

rocks of Trans-Pecos Texas. Geol 16:509-512<br />

Henry CD, Price JG, Parker DF, Wolff JA (1989) Mid-Tertiary<br />

<strong>silicic</strong> alkalic magmatism of Trans-Pecos Texas: <strong>rheomorphic</strong><br />

tufts and <strong>extensive</strong> <strong>silicic</strong> <strong>lavas</strong>. In: Chapin CE, Zidek J (eds)<br />

Field excursions to volcanic terrains in the western United<br />

States, I: Southern Rocky Mountain region. New Mex Bur<br />

Mines Min Res Mere 46:231-274<br />

Henry CD, Price JG, Rubin JN, Laubach SE (1990) Case study of<br />

an <strong>extensive</strong> <strong>silicic</strong> lava: the Bracks Rhyolite, Trans-Pecos<br />

Texas. J Volcanol Geotherm Res 43:113-132<br />

Honjo N, Bonnichsen B, Leeman WP, Stormer JC (1992) Minera-<br />

logy and geothermometry of high-temperature rhyolites <strong>from</strong><br />

the central and western Snake River Plain. Bull Volcanol<br />

54:220-237<br />

Leat PT, Schmincke H-U (1990) Emplacement of <strong>strongly</strong> welded<br />

ignimbrite E, Mogan Formation, Gran Canaria. Int Volc<br />

Congr abs Mainz (FRG)<br />

Lofgren GE (1980) Experimental studies on the dynamic crystalli-<br />

zation of silicate melts. In: Hargraves RB (ed) Physics of mag-<br />

matic processes. Princeton, Princeton Univ. Press, pp 487-<br />

551<br />

Mahood GA (1984) Pyroclastic rocks and calderas associated with<br />

<strong>strongly</strong> peralkaline magmatism. J Geophys Res 89:8540-<br />

8552<br />

Mahood GA, Hildreth W (1986) Geology of the peralkaline vol-<br />

cano at Pantelleria, Strait of Sicily. Bull Volcanol 48:143-172<br />

Manley CR (1989) Cooling, devitrification, and flow or large, hot<br />

rhyolite lava flows: numerical modeling results. New Mex Bur<br />

Mines Min Res Bull 131:174<br />

Manley CR (1990) Exposed vent area for a voluminous rhyolite<br />

lava flow: the 'tuff' of the Badlands, Owyhee Plateau, south-<br />

west Idaho. Geol Soc Amer Abs w prog 22:A290<br />

Manley CR, Fink JH (1987) Internal textures of rhyolite flows as<br />

revealed by research drilling. Geol 15:549-551<br />

Milner SC (1986) The geological and volcanological features of<br />

the quarz latites of the Etendeka Formation. Comm Geol Surv<br />

Southwest Africa/Namibia 2:109-116<br />

185<br />

Milner SC, Duncan AR, Ewart A (1992) Quartz latite rheignim-<br />

brite flows of the Etendeka Formation, north-western Nami-<br />

bia. Bull Volcanol 54:200-219<br />

Orsi G, Sheridan MF (1984) The Green Tuff of Pantelleria:<br />

rheoignimbrite or <strong>rheomorphic</strong> fall? Bull Volcanol 47:611-<br />

626<br />

Parker DF (1986) Stratigraphic, structural, and petrologic devel-<br />

opment of the Buckhorn caldera, northern Davis Mountains,<br />

Trans-Pecos Texas. In: Price JG, Henry CD, Parker DF,<br />

Barker DS (eds) Igneous geology of Trans-Pecos Texas. Univ<br />

Texas Austin Bur Econ Geol Guideb 23:286-302<br />

Parker DF, McDowell FW (1979) K-Ar geochronology of Oligo-<br />

cene volcanic rocks, Davis and Barrilla Mountains, Texas.<br />

Geol Soc Amer Bull 90:1100-1110<br />

Petrini R, Civetta L, Iacumin P, Longinelli A, Bellieni G, Comin-<br />

Chiarmonte P, Ernesto N, Marques LS, Melfi A, Pacca I, Pie-<br />

cirilo EM (1989) High temperature flood <strong>silicic</strong> <strong>lavas</strong> (.9) <strong>from</strong><br />

the Parana Basin (Brasil). New Mex Bur Mines Min Res Bull<br />

131:213<br />

Pieri DC, Baloga SM (1986) Eruption rate, area, and length rela-<br />

tionships for some Hawaiian lava flows. J Volcanol Geotherm<br />

Res 30:29-45<br />

Rowley PD, Kuntz MA, MacLeod NS (1981) Pyroclastic flow de-<br />

posits. US Geol Surv Prof Pap 1250:489-512<br />

Sampson DE (1987) Textural heterogeneities and vent area struc-<br />

tures in the 600-year-old <strong>lavas</strong> of the lnyo volcanic chain, east-<br />

ern California. In: Fink JH (ed) The emplacement of <strong>silicic</strong><br />

domes and lava flows. Geol Soc Amer Sp Pap 212:89-101<br />

Schmincke H-U (1974) Volcanological aspects of peralkaline<br />

<strong>silicic</strong> welded ash-flow <strong>tuffs</strong>. Bull Volcanol 38:594-636<br />

Schmincke H-U, Swanson DA (1967) Laminar viscous flowage<br />

structures in ash-flow tufts <strong>from</strong> Gran Canada, Canary Is-<br />

lands. J Geol 75:641-664<br />

Self S, Wolff JA, Spell TL, Skuba CE, Morrissey MM (1991) Re-<br />

visions to the stratigraphy and volcanology of the post-0.5 Ma<br />

units and the volcanic section of VC-1 corehole, Valles Cal-<br />

dera, New Mexico. J Geophys Res 96: 4107-4116<br />

Shaw HR (1972) Viscosities of magmatic silicate liquids: an em-<br />

pirical method of predicton. Amer J Sci 272:870-893<br />

Smith RL (1960) Ash flows. Geol Soc Amer Bull 71:795-842<br />

Sparks RSJ, Wilson L (1976) A model for the formation of ignim-<br />

brite by gravitational column collapse. J Geol Soc London<br />

132: 441-452<br />

Sparks RSJ, Wilson L, Hulme G (1978) Theoretical modeling of<br />

the generation, movement, and emplacement of pyroclastic<br />

flows by column collapse. J Geophys Res 83:1727-1739<br />

Trigila R, Walker GPL (1986) The Onano spatter flow, Italy: evi-<br />

dence for a new ignimbrite depositional mechanism. Int Volc<br />

Congr New Zealand abs 81<br />

Twist D, Bristow JW, Van der Westhuizen W (1989) High-temper-<br />

ature flow banded ash-flow tufts: a review of occurrences in<br />

southern Africa. New Mex Bur Mines Min Res Bull 131:273<br />

Twist D, Elston WE (1989) The Rooiberg Felsite (Bushveld Com-<br />

plex): textural evidence pertaining to emplacement mecha-<br />

nisms for high-temperature siliceous flows. New Mex Bur<br />

Mines Min Res Bull 131:274<br />

Twist D, French BM (1983) Voluminous acid volcanism in the<br />

Bushveld Complex: a review of the Rooiberg Felsite. Bull Vo-<br />

canol 46:225-242<br />

Walker GPL (1973) Lengths of lava flows. Phil Trans R Soc Lon-<br />

don 274:107-118<br />

Walker GPL, Heming RF, Wilson CJN (1980) Low-aspect ratio<br />

ignimbrites. Nature 283:286-287<br />

Walker GPL, Self S, Frogatt PC (1981) The ground layer of the<br />

Taupo ignimbrite: a striking example of sedimentation <strong>from</strong> a<br />

pyroclastic flow. J Volcanol Geotherm Res 10:1-11<br />

Whittingham AM (1989) Geological features and geochemistry of<br />

the acidic units of the Serra Geral volcanic formation, south<br />

Brazil. New Mex Bur Mines Min Res Bull 131:293<br />

Wilson L, Head JW (1981) Ascent and eruption of basaltic magma<br />

on the Earth and Moon. J Geophys Res 86:2971-3001


186<br />

Wilson L, Sparks RSJ, Walker GPL (1980) Explosive volcanic<br />

eruptions, IV. The control of magma properties and conduit<br />

geometry of eruption column behavior. Geophys J R Astr Soc<br />

63:117-148<br />

Wolf T (1878) Der Cotopaxi und sein letzte eruption am 26 Juni<br />

1877. Neues Jahr Mineral Geol Palantol 113-167<br />

Wolff JA (1983) Petrology of Quaternary pyroclastic deposits<br />

<strong>from</strong> Tenerife, Canary Islands. PhD thesis London Univ 542<br />

PP<br />

Wolff A (1989) Origin of widespread <strong>silicic</strong> <strong>lavas</strong> and lava-like<br />

tufts. New Mex Bur Mines Min Res Bull 131:297<br />

Wolff JA, Davidson JP (1990) Intraplate geochemical signature in<br />

pre-extensional Cenozoic <strong>lavas</strong> of Trans-Pecos Texas. Geol<br />

Soc Amer Abs w Progs 22:A166<br />

Wolff JA, Horn M (1991) Modelling volcanic agglutinates. Geol<br />

Soc Amer Abs w Progs 23:107<br />

Wolff JA, Wright JV (1981) Rheomorphism of welded turfs. J Vol-<br />

canol Geotherm Res 10:13-34<br />

Wolff JA, Warner G, Blake S (1990) Gradients in physical param-<br />

eters in zoned felsic magma chambers: implications for evolu-<br />

tion and eruptive withdrawal. J Volcanol Geotherm Res<br />

43:37-55<br />

Editorial responsibility: H.-U. Schmincke

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!