Walker J. Facies models, Canada, 1992

Подождите немного. Документ загружается.

are termed phreatomagmatic.

Fallout distributions may be uniformly

dispersed around the source, indicating

little influence from wind (Fig.

9B), or

may be asymmetrically distributed due

to prevailing winds (Fig.

9C). In general,

fallout deposits from large magnitude

eruptions tend to be more uniformly dis-

tributed than deposits from smaller

magnitude eruptions (Carey and

Sparks,

1986).

However, fallout pat-

terns can be very complex if there are

different winds at different altitudes or if

wind directions change during the

course of an eruption.

Fallout sheets exhibit systematic

changes in thickness, maximum clast

size, and proportions of pumice, crys-

tals, and

lithic fragments from near-

vent to distal environments. Thick-

nesses generally decrease expo-

nentially away from the vent (Pyle,

1989). Fine-grained fallout deposits

are not necessarily distal, however, as

they may result from low-intensity

eruptions, phreatomagmatic explo-

sions, or co-ignimbrite ash deposition.

In asymmetrically distributed

de~osits

(Fig. 9C), thicknesses decrease more

gradually along the dispersal axis than

on its margins or upwind of the source

volcano. Maximum thicknesses may not

necessarily occur in proximity to the

vent, due to erosion of fallout material

by pyroclastic flows. Secondary thick-

ening of the May

18th, 1980 Mt. St.

Helens fallout ash occurred 300 km

downwind of the volcano

(Fig..9D). This

may have been a result of premature

fallout of fine-grained agglomerates

andlor additions from co-ignimbrite ash

generated by the pyroclastic flows on

the flanks of the volcano. This indicates

that the thicknesses of distal plinian

ashes may have been overestimated by

not recognizing the co-ignimbrite ash

component. The distinction between

plinian fallout ashes and co-ignimbrite

ashes (ash derived from a pumiceous

pyroclastic flow) is indeed difficult in

...

.distal regions. ~0th are fine grained and

Figure

Rhyolite fallout bed showing the transition from massive pumice below the

rich in vitric material. Co-ignimbrite ash

hammer to stratified pumice above.

pyroclastic flow deposit directly overlies the stratified

layers do not thin exponentially and

fallout sequence. Guaje Pumice Bed, Jemez Mountains, New Mexico,

USA.

Figure

Inverse grading in a rhyolite pumice fallout deposit. The thin white beds are

fine-

Figure

Normal grading in a rhyolite

grained units which may represent episodic phreatomagmatic activity. Note the relatively

pumice fallout

unit.

Pre-Bandelier plinian

thick fine-grained unit, marked by the arrow at the base of the inversely graded bed, which

deposit, Jemez Mountains, New Mexico,

may indicate an initial phreatomagmatic phase of the eruption. Scale is graduated in 10-cm

USA.

increments. Cerro Toledo Rhyolite, Jemez Mountains, New Mexico,

USA.

108 LAJOIE. STIX

thus have a much more uniform thick-

ness over a wide area than do

plinian

units. Trace element geochemistry of

the glass shards may be able to dis-

tinguish between distal

plinian and co-

ignimbrite ashes from one eruption. If

the

plinian phase of the eruption is

erupted before the

pyroclastic flow

phase, the

plinian eruption will tap the

upper regions of the magma chamber,

which are most enriched in incompat-

ible trace elements

(e.g., uranium and

niobium). Therefore, the

plinian ash

will have higher concentrations of in-

compatible trace elements than the

co-ignimbrite ash.

The maximum size of pumices and

lithic fragments in outcrops of a fallout

deposit decreases exponentially from

the source

(Pyle, 1989), although

clasts in proximity to the vent may be

anomalously large due to ballistic em-

placement. These maximum size varia-

tions can be mapped as

isopleths of

equivalent clast diameter (Fig. 13). It is

preferable to examine

lithic fragments

because of the problems of breakage

and variable densities associated with

pumices. Systematic lateral changes

in the

isopleths can be related to the

height of the eruption column. The

width of a particular

isopleth is a func-

tion of the height of the column, while

the maximum downwind range of the

isopleth depends upon both the

column height and the wind speed

(Carey and Sparks, 1986). The distri-

bution of maximum

lithic isopleths is

generally less asymmetric than thick-

ness isopachs (compare Figs.

9A and

13B), implying that the isopleths are

less affected by wind.

The grain size and proportions of

the three major components in a

fallout deposit (pumice, crystals, and

lithic fragments) also change from

proximal to distal regions (Fig. 14).

The deposit as a whole becomes finer

grained away from the source. This is

reflected in the distal fining of the

pumice and

lithic populations. The

size of the crystals does not change

laterally, however, because crystal

size is controlled by pre-eruptive con-

ditions in the magma chamber. The

proportion of

lithic fragments remains

approximately constant, while the per-

centage of pumice decreases and

percentage of crystals increases dis-

tally. This reflects the progressive lib-

eration of crystals from pumice frag-

ments as the grain size of the pumices

approaches that of the crystals.

Facies in nuee-ardente deposits

Three different types of nuees ardentes

are recognized by volcanologists,

I) the

Pelee

type where the nuees are pro-

duced by directed explosions gener-

ating high-velocity, low-concentration,

turbulent suspensions, 2) the

St.

Vincent

type which results from column

collapse, with the flow descending the

slopes of the volcano in radial direc-

tions, and

the Merapitype caused by

avalanche of solidified, but hot, blocks

of the lava dome which break up into

smaller pieces during their descent in

high-concentration suspensions

(Escher,

1933; Macdonald, 1972; Williams and

McBirney, 1979; Fig. 15).

There is general agreement among

workers that although the

Pelee nuees

ardentes are initiated by a lateral ex-

plosive thrust, and St. Vincent nuees

ardentes by vertical explosion, once in

motion they behave as gravity-con-

trolled density currents. It follows that

the mechanisms responsible for trans-

porting the debris are similar to those

of other density currents. It is therefore

not surprising that

debris

and

grain

flows

have been invoked to explain the

features found in a number of coarse

pyroclastic deposits, and

turbidity

current

for a limited number of turbu-

lent nuee-ardente deposits.

The

Pelee nuees ardentes

are pro-

duced by strong lateral thrusts leading

to high velocities, roughly 130

mlsec

over St-Pierre at the base of Mount

Pelee, for the May 8th and 20th nuees

ardentes of 1902 (Lacroix, 1904;

Brissette and Lajoie, 1991). Such high

velocities resulted in turbulent low-con-

centration density currents which left

characteristic deposits which have

been well documented at Mount

Pelee.

At their type locality, the Pelee nube-

ardente deposits are characterized by

the common presence of three well-

defined strata (beds I,

II,

and Ill; Fig.

16; Boudon and Lajoie, 1989;

Charland

and Lajoie, 1989). The basal bed

(I)

thinner than the middle bed and con-

tains the coarsest grain size of the

deposit. It is massive, reversely and

more rarely normally graded, and poor

in fines with less than 2 per cent of

grains finer than

4+ (1116 mm). The

middle bed

(II), which may rest either in

sharp or transitional contact with the

underlying bed, represents most of the

deposit. It shows well-developed

normal-population grading (Fig. 16)

with the fines

(4+) increasing to 30 per

cent at the top of the bed. Stratification

is superimposed on the grading in the

upper portion of the bed (Fig. 16). In

more proximal sections these strata are

relatively crude and thick, dipping up-

flow at less than

lo0

(Fig. 16), and are

interpreted as antidunes. In more distal

sections, bedding is better defined, with

thinner laminae dipping downflow, and

interpreted as dune

bedforms (Boudon

and Lajoie, 1989). The upper bed is rel-

atively thin and contains the finest grain

size of the deposit (50 per cent finer

than

449. The coarse fraction increases

at the top of the deposit (Fig. 16) due to

the presence of accretionary

lapilli

(small spheres made up of agglutinated

fine ashes). Generally the upper bed is

normally graded, with rare parallel

laminae and ripple cross bedding at its

very top. The entire deposit of a typical

CHICHON

(MEXICO)

MAX

PUMICE

MAX

LlTHlC

Figure

lsopleths of the largest pumice

clasts

(A)

and lithic fragments (B) in the El

Chichon B fallout deposit, Mexico. From

Carey and Sigurdsson

(1986).

Pelee nuee ardente is therefore nor-

mentary nomenclature would be termed

where bed

represents a thick traction

mally graded. Sorting

(oi)

ranges from

poorly to very poorly sorted. The typical

carpet formed by gravity segregation of

more than

2.5$

at the base, improving

Pelee nuee-ardente deposit is inter-

the coarser fraction within the

nuke.

1.2-1.5$

in the finer-grained part of

preted as having accumulated from a

Here, dispersive pressure played an

the deposit, which by standard sedi-

low-concentration turbulent suspension

important role as the clast supporting

FOG0

SAO MIGUEL

(AZORES)

5.0

-6 -4 -2

4 -6 -4 -2

2 4

-6

-4 -2

-6

-4 -2

PUMICE

FRAGMENTS

SANlDlNE CRYSTALS

LlTHlC

FRAGMENTS

Figure

Proximal to distal variations in proportions of pumice, crystals, and lithic fragments for different size fractions in a fallout deposit.

Fogo

sequence, Azores. Modified from Walker (1971).

Figure

The three different types of nuees ardentes, modified from Escher (1933).

Figure

Depositional sequence in a

typical

Pelbe nube-ardente, at the type

section with

upbed variations of mean grain

size

(M,)

and sorting

(oi).

GRAIN

SIZE

110 LAJOIE, STlX

mechanism. The primary structure se-

quence in the deposit was interpreted

as resulting from a decelerating sub-

aerial particulate suspension where the

fluid was hot gas.

The causes responsible for the May

18th, 1980 Mount St. Helens "blast

surge" are different from those of the

1902 nuees ardentes at Mount Pelee

(Boudon

et a/.,

1990). However, both

eruptions left similar deposits (Fig.

17),

with a massive or normally graded

coarse basal "layer", overlain by a finer

"layer" whose base is massive or nor-

mally graded, passing upward into lami-

nated and cross-bedded ash. In the Mt.

St.

Helens deposit, the coarser basal

bed is channelled in its central portion,

and the thickness of the overlying bed

decreases laterally away from the axis,

where parallel- and cross-laminae be-

come increasingly better developed.

The

Merapi nuee ardentes

are pri-

marily controlled by gravity and lack

the initial lateral thrust of the Pelee

and Mount St.

Helens nuees ardentes.

Their velocities are therefore much

slower (Lacroix, 1904, witnessed ve-

locities ranging from 15 to 50

m/sec

for this type of nuee). These nuees are

sometimes referred to as block-and-

ash flows. At the type section the de-

posits consist of coarse blocks,

lapilli,

and ashes, in which the general bed-

ding and primary structure characteris-

tics do not change

downflow over the

entire

km length. Typically, the de-

posit of the 1984 Merapi eruption con-

sists of a single bed of reversely

graded

lapilli-tuff which lacks stratifica-

tion. The lower contact with the under-

lying deposit is sharp and nonerosive.

The entire population is graded (Fig.

18A), and the deposit is fines-poor,

with less than

per cent fines (4

4).

Generally, the deposit is very poorly

sorted, but grain size characteristics

are not easy to describe due to the

polymodality of the distributions. Some

of these subpopulations fit a Rosin law

better than the more standard log-

normal distribution, which suggests

that fragmentation rather than trans-

port played an important role in the

grain size distributions. The long axes

of the lithic fragments are preferen-

tially oriented parallel to flow. In such

coarse-grained deposits, it is generally

assumed that bed thickness and

maximum grain size decrease down-

flow with distance from the source

vent. However, the 1984 Merapi de-

posit shows a regular and significant

increase in both maximum grain

size and bed thickness,

downflow for

km.

The characteristics of the studied

Merapi deposit have been interpreted

as the result of a concentrated sus-

pension of cohesionless solids ex-

hibiting non-Newtonian behaviour.

Here, dispersive pressure played an

important role in the suspension of the

clasts, such as for an inertial, density-

modified grain flow. The flow travelled

on a

35" slope for the first km, and on

20" slope for the next 2 km. It began

depositing on a

slope, dissipating

all its energy in the next 4 km.

The deposits of the

St. Vincent nuees

ardentes

have not been described in

detail. In their pioneer work, Anderson

and

Flett (1903) depict the original

phenomena as a highly concentrated

mass which descended radially from

the crater for a distance of 4 km at a

relatively low velocity of about 15

mlsec. These authors report that the

most striking feature of the 1902

deposit is its fine grain size, 90 per

cent of the deposit being ash, with

fragments coarser than

cm making

up less than 3 per cent of the mass.

The deposit also contains many scori-

aceous lapilli, and less frequently, large

rounded bombs up to

m in diameter.

The deposit is massive, with no grading

or stratification of any sort, and has a

sharp lower contact. These characteris-

tics are indicative of a highly concen-

trated suspension of cohesionless solids

exhibiting non-Newtonian behaviour.

Facies in hydrovolcanic deposits

Hydrovolcanism is the effect of the in-

teraction of magma with water. It

affects all shallow subaquatic volca-

noes and many subaerial vents, for-

ming tuff cones and tuff rings which are

second in pyroclastic-deposit abun-

dance only to scoria cones. Explosions

which result from the conversion of

groundwater to steam by ascending

magma are called phreatomagma-

tic eruptions. The products are water,

steam, and brecciated country rocks,

and must include juvenile clasts. Hydro-

explosions may generate

ejecta plumes

with an additional horizontal component

known as a

base surge

or a

pyroclastic

AAI

FALL

UNlT

FINE UPPER UNlT

COARSE BASAL

UNlT

Figure

Lateral and vertical variations of bed thickness and primary structures in the May 18, 1980, "surge" deposit of Mount St. Helens.

From Moore and Sisson (1981).

surge that may be dry (superheated

steam media) or wet (condensing steam

media). The general consensus is that,

although initiated by an explosive thrust,

once in motion surges behave largely

as gravity-controlled density currents in

which turbulence is a major grain-sup-

porting mechanism. Cas and Wright

(1987, p. 205-214) argue that surges

can be compared to turbidity currents in

a general way. However, they also

suggest that in the case of wet surges

there is no direct analogy with "normal"

sedimentary processes, because of the

three-phase media (solid-liquid-gas).

The differences between sedimentary

and volcanic processes certainly do

exist but they could have minor effects

on flow rheology as shown by the pro-

posed models of lateral facies varia-

tions in dry- and wet-surge deposits.

Wohletz and

Sheridan (1979) offer a

model for dry-surge deposits in which a

proximal, cross-bedded facies passes

downflow within some 500 m (the scale

is important) to a massive and re-

versely graded bed facies, to be re-

placed further away by what they call a

"planar" facies in the more distal sec-

tions (Fig. 19). These "planar" beds are

not planar or plane beds in the sedi-

mentological sense. They may be 10

cm thick, are reversely graded, and are

believed to have formed from a laminar

GRAIN SIZE

(4)

Figure

Typical Merapi nuee-ardente deposits, at Merapi

(A)

with cumulative distribu-

tions from base to top showing reverse population grading (notebook for scale is

high), and at Mount

Pelbe

(B).

grainflow mechanism. In this model,

bed thickness decreases very rapidly in

the direction of transport in the cross-

bedded facies, but not in the massive

facies where the variation is not sys-

tematic. Sorting is poor throughout the

deposit

(I$

ranging from 1.6 to

1.88),

but

grain size increases

downflow from a

mean of

2.6@ (0.177 mm) in the cross-

bedded facies to

-0.48@ (30 mm) in the

more distal massive beds, a very signifi-

cant increase. Wohletz and

Sheridan

(1979) suggest that on eruption, the

surge is inflated (fluidized) by

volatiles

to such an extent that velocities are ex-

tremely high and viscosities very low.

Therefore, turbulence occurs close to

the vent, resulting in traction that pro-

duces the observed primary structures.

The cloud travels downslope, away

from the vent, aided by gravity. Gases

escape rapidly and the cloud deflates,

resulting in a higher grain concentra-

tion. This leads to laminar flow, which is

responsible for the massive and re-

versely graded distal facies. Cloud de-

flation, however, cannot explain the

large lateral increase in grain size from

fine sand to coarse pebbles in the direc-

tion of transport. Fisher and Schmincke

(1984, p. 254) argue that this model

may provide statistical summations of

many flows through time at a particular

locality. They also suggest that these

facies variations cannot apply to pro-

cesses which occur laterally within a

single event.

An alternative model for the lateral

CROSS BEDDED FACIES

J.:M.j

MASSIVE AND INVERSELY

. .

GRADED FAClES

PLANAR FACIES

Figure

Lateral facies variations in dry

pyroclastic surges according to Wohletz

and

Sheridan

979).

LAJOIE, STIX

facies variations in dry-surge deposits

was proposed by Sohn and Chough

(1989) from the Suwolbong tuff ring on

the island of Chaju in Korea and has

been verified at one other tuff ring

(Chough and Sohn, 1990). In this

model (Fig.

20), the general thickness

of the deposit, as well as the individual

bed thickness and grain size, de-

crease in the direction of transport for

almost three kilometres. The most

proximal facies consists of a majority

of massive beds passing laterally into

a crudely stratified parallel-laminated

bed facies, which is in turn replaced

downflow by a dune cross-bedded fa-

cie~ in the most distal sections. The

model is interpreted in terms of decel-

erating turbulent suspensions with in-

creasing distance travelled and is

therefore very similar in its variations

and process to that observed in many

turbidity-current deposits.

The lateral facies variations in wet-

surge deposits described here were ob-

sewed on the island of Linosa, located

in the Sicilian channel south of Agri-

gento where the deposits are the result

of phreatomagmatic activity (Lanti

a/.,

1988). The presence of vesiculated

tuffs, abundant mud-coated clasts and

accretionary lapilli in these deposits is

considered good evidence that the

surges which transported the clasts

were wet. The lateral variations of

facies in these wet-surge deposits

show a general decrease of bed thick-

ness and grain size downcurrent from

the vent (Fossa Cappellano, Fig. 21) to

the more distal section (Punta

Calcarella), for a distance of 2.5 km.

The general primary-structure se-

quences vary as in turbidites. In the

most proximal section (Section 1, Fig.

21) more than 80 per cent of the

beds are massive. This massive facies

passes

downflow into a sequence

where nearly 50 per cent of the beds

show a parallel-laminated division over-

lying a normally graded or massive divi-

sion (Section 2, Fig. 21). This second

lateral facies is replaced at about

1 km

from the vent by a sequence of thinner

beds with a primary-structure sequence

of normal grading overlain by parallel

laminae. Here, however, dune cross

beds make up an important proportion

of the section (Section

Fig. 21). In

the most distal section (Section

Fig.

21), most beds exhibit dune-type cross

bedding. This lateral variation of facies

MASSIVE

CRllnFI

STRATIFIFn

FACIFS

-..----.

..,.

.--

FACIES

IMASSIVE~

CRUDELY

I-I--STRATIFIED+I- DUNE FACIES

-I

FAClES;

FAC~ES

Figure

Lateral facies variations in dry-surge deposits from Cheju Island, Korea. Modified

from Sohn and Chough (1989).

Table

Differences that may be observed between beds of pyroclastic fall and flow

deposits.

Fall Flow

Sorting Well sorted. Poorly sorted.

Bed thickness Regular and drapes the under-

Irregular; thins over highs,

lying surface (mantle bedding).

thickens in depressions; thins

laterally towards channel

margins.

Grading and Massive beds are rare; normal

Massive beds, reverse grading

laminae grading is rare, but present.

are common in deposits having

Absence of well-defined trac- accumulated from laminar sus-

tion structures such as parallel

pensions (debris and grain

and oblique laminae, but crude

flows). Normal grading is

strata are common.

common in deposits from turbu-

lent suspensions and is com-

monly found underlying or

overlying a laminae division.

Other primary Bomb-sags and accretionary Accretionary

lapilli occur in

structures

lapilli are common in subaerial

upper beds of some subaerial

or shallow water deposits.

nudes ardentes. Rare or absent

Gas-escape pipes are absent.

in subaqueous deposits.

Gas-escape pipes are common.

Primary Absent. Common, and are generally

structure

similar to those

obselved in

sequences other mass-transported sedi-

ments.

in a wet-surge deposit is identical to the

above model proposed by the Koreans

for dry-surge deposits, and in all proba-

bility both sequential variations origi-

nate from similar processes.

The above proposed models of

lateral facies variations in the Korean

and Italian surge deposits are new and

insufficiently tested to be considered

pressure within the magma exceeds the

predictors. However, they are in our

water pressure of the subaqueous envi-

opinion excellent working hypotheses.

ronment. This is the pressure compen-

sation level (PCL; Fisher,

1984),

which

Subaqueous pyroclastic deposits

should normally be at relatively shallow

Pyroclastic fragmentation is explosive

depths. The evidence suggests that the

by definition. Explosions caused by

PCL for most explosive basaltic erup-

magmatic

volatiles can only occur when

tions is less than

200

m, and it seems

PARALLEL LAMINAE

MASSIVE CROSS BEDDED

CALCARELLA

Figure

Lateral facies variations in wet-surge deposits from Linosa Island, Italy. The upper triangle gives the relative proportion of the

primary structures in the stratigraphic sections which are localized in the lower cross section. Grain-size variations are given for the different

grain textures. The small bars on the scale in photos

and

are cm. The notebook in photo

cm in height.

114 LAJOIE, STIX

that acid magma cannot vesiculate sig-

nificantly at depths greater than a few

tens of metres. Consequently most

authors favour a shallow water or a sub-

aerial origin for the pyroclasts found in

subaqueous pyroclastic deposits.

Subaqueous pyroclastic flows can

develop from subaerial nuees ardentes

that moved into water. Because of their

high density, the Merapi and St. Vincent

nuees ardentes should continue to flow

along the bottom, as they reach water

level. Where the slope is sufficiently

steep, these flows may become turbu-

lent, mix with water, and produce cool,

unwelded deposits. They may, how-

ever, enter water without mixing and

retain enough heat to produce welded

deposits. Many workers do not believe

that hot pyroclastic flows can be de-

posited under water, although a few

authors report welded tuffs from marine

sequences. The water interaction

should be different for the low-density,

high-velocity Peke-type

nuees ar-

dentes. As these turbulent nuees reach

water, the ash cloud overlying the

denser lower portion of the flow floats

on the denser water (Anderson and

Flett, 1903; Lacroix, 1904). The denser

lower portion is assumed to continue its

route under water. Here, mixing with

ambient fluid is almost certain because

the flow is turbulent as it enters water.

There are no examples of

Pelee-type

deposits reported from the literature

on subaqueous pyroclastic deposits. It

may be that the three-bed division

which is observed in subaerial de-

posits is lost due to flow separation of

the upper part and mixing with water

of the lower portion of the flow.

Descriptions of subaqueous pyro-

clastic deposits are rare. The evidence

for subaqueous deposition is generally

determined indirectly on stratigraphic

arguments using sequence associations,

interbedded pillow lavas, or fossils, and is

not always convincing. The presence of

normal grading, cross laminae, and

slump structures, for example, is not very

solid evidence for subaqueous transport

and accumulation, as is shown by the

above descriptions of the St.

Helens and

Mount

Pel6e subaerial deposits.

The two examples that will be used

to describe the lateral facies variations

in subaqueous pyroclastic deposits are

taken from the Neogene of Japan, and

from the

Archean of Rouyn-Noranda,

Canada.

The subaqueous deposits of Japan

were erupted just above water level and

deposited in a lake 200 to 500 m deep

(Yamada, 1984). Flows travelled for a

maximum of 8 km. The proximal facies,

observed over the first 2 km (Fig. 22)

consists of massive or chaotic breccia

with cobble- and pebble-sized lithic

fragments, some of the larger blocks

showing radial jointing due to quench-

ing, which suggests that they were hot

at the time of deposition.

The intermediate facies, 2 to 4 km

downcurrent, is characterized by a se-

quence of beds which have erosive

lower contacts and are normally

graded, and in which parallel laminae

gradually develop upwards. Shattered

crystals, the result of rapid cooling, are

common in this facies. In the upper-

most units, cross laminae may be pre-

sent. Locally, accretionary

lapilli are

scattered in the upper tuffs.

The distal facies,

km downflow, is

distinguished by finer grain sizes and

the absence of the massive lower por-

tion of the vertical sequence (Fig.

22).

Yamada (1 984) compared the ver-

tical primary structure sequence in

these subaqueous deposits to that ob-

served in high-density turbidity current

deposits (the R-S-T sequence of

Lowe, 1982).

The deposits of the

Archean

example extend laterally for 17 km,

and range in thickness from some 260

m in the proximal sections to 120 m in

the distal sections (Tasse

etal.,

1978).

Pillow lavas are found both underlying

and overlying the pyroclastic se-

quence. The pyroclasts are andesitic

in composition, consisting of lithic frag-

ments,. crystals, and scoriae. The pre-

sence of abundant and large scoriae

suggests very shallow water or sub-

aerial eruptions. The scoriae are more

deformed than the whole rock, indi-

cating that they were still hot at the

time of deposition. These

Archean py-

roclastic deposits are therefore be-

lieved to be primary. In these deposits

the ratio between lithic fragments and

scoriae decreases with distance from

the vent, that is, the amount of scoriae

increases downflow. This relationship

has also been observed in recent de-

posits, and is opposite to the scoria-

lithic ratio commonly found in pyro-

clastic fall deposits. This is due to the

lower density of the scoria which

settles at lower velocity, and is thus

found further downflow.

In the studied sections of the

Archean deposits, there are two dis-

tinct bed types with different mean

thickness, grain size, and primary

structures. The two bed types were

treated in two different statistical as-

semblages (A and B beds, Fig. 23).

Type A beds (Fig. 24) are thicker than

type B (Fig.

25).

Mean bed thickness

increases

downflow in type A,

whereas it decreases in type

Grain

size varies upsection, but generally it

decreases in the direction of transport

in both bed types. Primary structure

sequences vary systematically away

from the vent, but the sequences are

different in both bed types. In type A

beds, the most abundant structure is

grading, and parallel laminae are rare.

The proximal section has a high pro-

portion of massive or reversely graded

beds, whereas the distal section is

characterized by normal grading with a

greater number of beds where parallel

laminae are present. The two interme-

diate sections show a gradation from

the proximal to the distal facies.

Normal grading is the rule in type B

beds. Traction structures, such as par-

allel and oblique laminae (dunes and

ripples) are more abundant than in

type A, and their proportion increases

ERUPTION

WATER

LEVEL

CENTRE

Figure

Lateral variations of size and primary structures in a subaqueous pyroclastic

deposit of Japan. No vertical scale. Modified from Yamada

(1984).

DIRECTION OF TRANSPORT

MASSIVE

off^

Figure

Lateral variations of mean bed thickness and primary structure sequences in

the direction of transport in an

Archean subaqueous pyroclastic deposit. The variations are

shown for high concentration

(A)

and low-concentration

(B)

suspensions (see text). The fre-

quency distributions of structure sequences, in percentages, total

100

at each locality or

section. Modified from

Tass6

eta/.

(1978).

with distance of transport.

Tasse

et a/.

(1978) interpreted the

above lateral facies variations in sub-

aqueous pyroclastic flow deposits by

analogy with sedimentary density

flows. Type

beds probably result

from the accumulation of decelerating

turbulent suspensions of low density

(turbidity currents). Most of the flows

responsible for type A beds appear to

have been turbulent in distal regions.

Almost half

(45

per cent) of the beds in

the more proximal section have char-

acteristics that are best interpreted as

the result of a concentrated suspen-

sion of cohesionless solids exhibiting

non-Newtonian behaviour. Here, dis-

persive pressure played an important

role in the suspension of the clasts,

(inertial, density-modified, grain flow).

An integrated working hypothesis

Although there has been much work

done on pyroclastic deposits since the

publication of the second edition of

Fao

ies Models,

the data are still spotty.

Most models of vertical and lateral vari-

ations in volcaniclastic rocks are recent,

and have not been tested adequately.

However, the data presently available

may be used to propose an integrated,

idealized working hypothesis for facies

variations from a subaerial vent to a

subaqueous environment. Facies in

pyroclastic rocks reflect variations in

flow density, viscosity and velocity, as

well as transport distance from the vent.

For example, there are Pelde

nude-

ardente deposits on Merapi volcano,

and Merapi deposits on Mount Pelee

(Fig. 18). Most models of lateral and

vertical facies variations that have been

proposed can be explained by analogy

with other sedimentary mass-flow de-

posits. This approach is used to suggest

an idealized scenario of lateral facies

variations in pyroclastic rocks (Fig.

26).

In this model, deposits of Pelee-type

nuees ardentes are restricted to sub-

aerial aprons and lobes. The Merapi

and St. Vincent

nuee-ardente deposits

are channelled and are not restricted to

the subaerial environment. Entering

water, there is a high probability that the

flow mixes with water and that

trans-

forms into lower concentration turbulent

suspensions, resulting in subaqueous

type

beds described above. From the

vent to the most distal sections, bed

thickness and grain size decrease, and

as in other sedimentary mass-flow

de-

LAJOIE. STIX

posits, primary structure sequences

evolve from high flow power sequences

to low flow power ones.

Distinctions between pyroclastic

flow and fall deposits

It is commonly not an easy exercise to

distinguish between pyroclastic fall and

flow, and some of the differences

(Table

may be very subtle. Mantle

bedding is characteristic of fall deposits.

The deposit drapes underlying irregu-

larities with little or no thickness varia-

tions except on relatively steep slopes.

Since the pioneer work of Murai

(1961),

much work has been done with statis-

tical parameters of grain size distribu-

tions to help distinguish the various

types of unconsolidated and consoli-

dated pyroclastic deposits, but with

limited success. Most of the problems

with the techniques originate from the

almost universal use of

intervals in

these analyses, leading to useless

data. Commonly, the distributions are

not log-normal and better fit a Rosin

law. Thus the median, generally used

as a defining parameter of the distribu-

tions, is inappropriate. On the other

hand, primary structures, and particu-

larly primary structure sequences, can

be used successfully to distinguish

among the various types of pyroclastic

deposits.

SUMMARY

In this chapter we have presented what

is known of facies relationships in

auto-

clastic and pyroclastic deposits. Auto-

clastic fragments that are formed in situ

are monogenetic and not welded. They

may be transitional with the parent lava

flow, and have the same composition.

The internal structures in these de-

posits suggest the absence or near

absence of transport.

Pyroclastic fall deposits are crudely

stratified and commonly show a sys-

tematic lateral decrease in grain size

and bed thickness downwind. In these

deposits, sorting is poorest close to

source due to ballistic emplacement

and settling of fragments of different

densities.

fall deposits, the vertical

variations are controlled by eruption in-

tensity and are therefore unpredictable.

nuke ardente deposits, bed thick-

ness and grain size generally decrease

downflow, but close to vent the varia-

tions are not systematic and may even

be reversed. Beds are commonly grad-

ed in these deposits. In subaqueous

deposits the grading of all fragments is

generally normal, but in many subaerial

deposits pumices and scoriae are re-

versely graded. The primary structure

sequences vary systematically

down-

flow in most of the deposits and depict

changing flow conditions and grain

sizes that are

beingtransported.

ACKNOWLEDGEMENTS

Leopold

Gelinas, John Ludden, and

Roger Walker read the earlier editions of

this chapter and offered many helpful

suggestions. We thank Michel Demidoff

for careful work on the illustrations. The

research leading to many of the proposed

models in this chapter was supported by

the Natural Sciences and Engineering

Research Council of Canada.

Figure

Reversely graded subaqueous pyroclastic flow deposit of Archean age, Rouyn-

Noranda, Quebec. Top is left. Notebook is

cm long.

Figure

Prirnary-structure sequence in a type

bed. From base to top: normal grading,

parallel laminae, ripples.

Archean of Rouyn-Noranda.