CHAPTER F O U R

InItid statement
Lavas are flows of coherent magma that are enipted The principal requirernent for rhe effusive emption
at rhe Earth's surface during effusive volcanic of magma as coherent lava is that the eruption not
activity which is essenrialb non-explosive, or, for be explosive or that it be a relativelv small fire
some basaltic lavas, that are fed by lava fire fountain of lava. This requires thas the exsoked
fountains. Lava flows show great variations in size volatile content (Ch. 3) of the magma chamber
and shape, and in their surface and interna1 immediately before eruption, and of the magma
features. In this chapter we describe these variations during eruption, be sufficiently low to prevent the
and discuss ñhe different controls on lava mor- build-up of a gas pressure which c w l d cause fully
phology, such as effusion rate, physical propenies dcveloped explosive frñgmentation (Ch. 3). With
(particularly viscosity and yield strength) and the possible exception of some basic and ultrabasic
environmental factors. Many of the differences magmas, most subaerially erupsed magmas could
between lava flows can be ireated by initiallv potentially contain sufficient magrnatic water to
dividing them into low viscosity (low siBca) and cause explosive eruptions, given the initial aval-
high viscosiry (high silica) types. The characteristics ability of sufficient water in the magma source
of the flows are thereaftcr controlled bv rhe nature regions. Therefore, for most ilolcanoes, and excepr-
of the environment into which they are extruded, or ing hasaltic volcanoes, he volrrme of volcaniclastics
the nature of the environment into which they then erupted far exceeds the volume of coherent lavas.
flow, or both. However. basaltic volcanic centres are usually
exceptions ro this, because basic magrnas usually
contain lower magrnaric volatile contents.
For coherent magmas ta bc erupted from magma
sources with high volatile contents the magma has
to degas ro prevent build-up of m a p a t i c gas
pressure. This can be done bv the following
rnethods.
(a) Disect escape of exsolving volatiles, either
gradually through the vent, hydrothermal
springs and fumaroles or, more rapidly, by
episodic hydrorhermal (or sream) explosions
(Ch. 3 ) .
(h) A previous or contemporaneous phase of
explosive activiry dusing which most o£ thc
volatiles of the magma and magma chamber
have been removed. L'his suggcsts that lavas
could be a terminal event of many explosive
eruptions and, indeed, this seems to be the
case (e.g. the 1980 Mt Sr Helens eruption).
The exsolution and degassing of volatiles from a
magma wili increase its viscosity and yield strength,
which will affecr the rnobility, flow disrance and
thickness of any fava flow (Ch. 23. However,
cornpared with more-silicic lavas, most basic Iavas
are fluida1 and mobile, irrespective of their water
content. This is because basic magrnas arc less
polymerised than silicic magmas, since they contain
fewer intermolecular Si-O bonds a r hridges (Burn-
ham 1979; Ch. S). In general, the vnlume and
Iength of lava Anws decrease as the content of SiO,
increases. Thus, basalt flows are usually much
more voluminous and longer than rhyolite flows.

4.2 Size and form of subaerial lava flows

The Iargest known lava flows are flood lavas. These
include both continental Aood basalts and those of
mid-ocean ridges (e.g. Iceland). Rarer Archaean
ultramafic Aows (komatiites; Section 4.12) and Rozo rnernbcr
hver
\ !ixi
extensive phonolite and trachyte flows thar have Figure 4.7 (a) The 1783 Laki basalt lava flow in southern
been described from the East African Rift can also Iceland erupted from 115 craters along the 25 km fissure
(after Thorarinscon 19695. (bl The Columbia River flmd
be considered as flood lavas. Cuch lavas are eruptcd basalt plateau with the distribution nf the Roza Milernber and
from large fissures, which are laterally conrinuaus its fissure vent systcm (see also Frg 13 7 ) (after Swanson et
fractures that can he called line sources (see below). al 1975).
 The largest historic lava of this type is the 1783 covered a minimurn area of 40 000 km' (Fig. 4.1b).
Laki basalt flow (Fig. 4. la) which travelled about Al1 of the flows were probably erupred over a
40 km and has a voIume of 12.3 km3(Thorarinsson priod of a few hundred ofears or less, but individual
1969). Very much larger flood basalts over 20CI km flo~vsare thought to have been esupted in a matter
long are known in the geological record. In rhe of days and are themselves (except the thin near-
mid-Miocene Colurnhia River Plateau nf zhe w.stern vrnr flows) srilI extremely large. The largest indi-
USA, flows of the Roza Member (Fig. 4. Ih) i-idual lava flows must have vvolurnes up ?o700 km3
travef Ied 300 km and its volurne exceeded 1 5 0 km3 [Sxvanson 81 Wright 1481).
{Swanson et al. 1975). This and other lavas in the The average rhickness of single lava flows in
Columbia River Plateau (e.g. Srvanson & N-right mrasured sections of the Columbia River Plateau
1981) are volumerricaIlv three to five times larger tiasalts is between 15 and 35 m (Warers 1961).
[han many of the world's largest stmta~~olcanoes Comparable lava flow thicknesses are found in the
(Ch. 13). For a general description of flood basalt flood basalts of tthe Deccan Traps in Central India
'volcanoes' and provinces, see also Chapter 13. I Subbarao & Sukheswala 1481); Chaubey (1973)
The largesr lava unit thar has been descrikd is, found average thicknesses of between 15 and 30 m.
indeed, that of the Roza Memher in the Columbia In the Aood basaIts of eastesn Iceland, G. P. L.
&ver Plateau. The Roza Member actualIy consists ITalker (1963) found that the average flow thickness
of a small number of lava flows and, tl-pically, \\-as 17 m.
stratigraphic secrions show one or two thick flows, Other basaltic lavas can be considered to be
but sometimes three or more. Also, a number of empted from central volcanoes, point-source vents
thinner flow (or flow units; see Section 4.3 occur or small restricted fissures. These lava flows are
near vent. Cumalatively these flows originally much less voluminous and cover much smaller

O 5
DiameterH(krn1 "
O 1 2 5 10 20 50
! 1 I I I I I

m - Area covered (km2)

1.'.1
rhyolirer ln 176)

trachyfes, andcritcs, docites (n- 2391

baralts (n -479)
1 . '
3
high vncasity

IOW v~~arity

Figure 4.2 ~im>'qsionsof lavas of different cornposit ons. The two scales along the x-axis give the
area covered by thd extrusion and the diarneter of a circ!e having this areif. The broken Iines give the
aspect ratio VI# The dotted Iinesqive t% volurnes af circular drsc-like bodies of the dimensions
shown as a rough guide to the votumes of the lava extrvsions (After G P L Walker 7973a 1
( a ) TWOsimple lava flows thousands of flow units) and how new boccas Csmall
cipenjngs) feeding new flow units carnrnonly
formed at the fronts of mature flow units which had
othenvise ceased to flow.
Large single Aow unit flood basalts found in the
geological record are believed to be erupted at very
high effusion rates. Order of magnitude caiculations
by Swanson el al. (1475) suggest that the effusion
rate along the Roza fissure vent sysrem was 1 km3
d-' km-' (d = day) for one flood basalt flow.
(b) Compound lava flow
This calculation uses an individual-lavaflow volume
of 700 k m ~ a p p r o ~ i m a thaIf
e l ~the total tiolurne of
the Roza Member} for each of what appears to be
two main lava flows (see Section 4.2), an eruption
duration of seven davs ancf a length of 100 km for
the fissure vent svstem. This leads to an effusion
rate of 1 x 10h rn3 S-' for rhe whoIe vent system or
1 x lo4 m3 S-' km-'. These estimates of discharge
rate for the Roza flood basalt flows are comparable
with those estimared for highly explosive ignirn-
block
brire-forming eruptions (10-10' m3 S-', Ch. 8).
A sunreyof hictoric more silicic, higher viscosity
lavas (Table 4.2) shows that average effusion rates
aa are between 0.05 and 11.6 m3 S - ' , generally a few
orders of magnitude lower than those far basic
pahoehoe lavas. hiore-viscaus extrusions might also be ex-
pected ro form compound lavas at low exirusion
rates. For example, the Santiaguito dacite dome,
Santá PvIaria volcano, Guatemala, is a compound
Figure 4.3 Simple and cornpound lava flows. (After G. P. L.
Walker 1971.) Table 4.2 Effuston rates of some andesitic and dacitic lavas
{after Newhalt and Melson 1983)

extruded, at low effusion rates produce Rows com- Eruption Average vofumetric
posed of srnall flow units which pile up close to the effusion rate
vent and produce compound lavas. The Laki 1783 (m3 S-')
Aow in Iceland was extruded at a relatively high Santorini. Greece 188670 0.7
effusion rate (Table 4.11, and is compased of only a Santiaguito. Santá Maria volcano 04
few flow units, so ir can be considered a simple 1922-present
lava. The hisioric lava flows of Mr Etna have been ~t Lamington 1951-6 58
erupted at much Iower effusion rates (Table 4.1), Bezymianny 1955-present 7 8
Lolima 1 9 7 5 6 O 05
and form cornpound lava flows or Aow-fields Augustine 1976 11.6
(Wadge 1978, Pinkerton & Sparks 1976). IVadge Mt St Helens 198&present 0.5
also demonstrated that the maximurn distance Ucu 1910 IMeiliSRinzan cryptodome*) 3.5
attained by these lava flows increases linearlv with Usu 1 9 4 3 5 (Showa-Shinzan cryptodome) 1.2
increasing effusion rate for flows greater than I km Usu 1977-present (Usu-Shinzan 06
cryptodome
long. Pinkerton and Sparks (1976) describe the
formation of rhe 1975 flow-field (composed of many Cryptodomes are explained in Section 4 8.
64 LAVA FLOWS

silicic lava which began in 1922 and continues to flowtng Iava at anv given point, and rhis wiIl have a
grow to date; it now consists of at leasr 14 rnarked effecr on the yield strength. An increased
recognisable Aolv units (Rose 1972a). effusion rate wouId increase the temperature, and
rherefore reduce the yield srrength, at any particdar
point in the lava.
4.3.2 PHYSICAL PROPERTIES
Hulme (19741,who modelIed lavas as Bingham
4.3.3 SLOPE
substances (Ch. 2), indicattd that the principal
factor governing their shape was rheir non-New- Flow width varies inverseIy with ground slope
tonian rheology. His theoretical analysis and exper- (Hulme 1974). Hawever, the effect of slope on lava
iments with kaolin suspensions, which are cIose to length has been shown to be small compared with
Bingham substances In rheology, showed that other factors (G. P. L. Walker 1973a).
aspect ratio was mainly dependenr on yieId stress.
For a Bingham substance to flow downhill, it
rnust form a layer deep enough for the shear stress 4.4 Eruption of subaerial basaitic lavas
at the base to exceed the yield strength. Close to the
lateral rnargins the deprh is not great enough for Basaltic lavas are empted from either fissures or
downhill flow to occur, and dead zones of stationary central (also calIed point source) vents. Fissure
fluid form levées along margins. The depth and systems ihat feed Iarge flood basalts rnay be very
width of a flow, and the width of each dead zone, large (e.g. >PO0 km in length, Fig. 4. lb). Centra1
are related to five independent initial parameters: vents are typical of larger basaltic voIcanoes, scoria
effusion rate (F), the slope (a}and three praperties conec and other types of srnaller basaltic volcanoes.
of the fluid - Bingham viscosity (q}, vield srrength However, these smaller centres ase commonly
(oo) and specific weight @er where g is the associated with fissures, and even on the large
acceleration due ro gravity and Q is the densiry of volcanoes fissures may control flank eruptions.
the fluid). The critica1 depth (d,) which rnust be Many eruptions of basaltic lava rnay begin along a
exceded for any flow io occur is given by large length of a fissure, but activiry quickiy
localises to a few point sources or 'nodesYL.
dc = .crd~ea (4.1) Wilson & Head 1981, Delaney & PoIlard 1982).
For lavas with highes yield srrengths, d, is therefore Even for the large flood basalts this aIso seems to be
larger, and the thickness of the lava flow is greates. true (Ch. 13).
The aspect ratio of a Iava flow can be predicted Basaltic lavas can issue from vents as:
from
a coherent flows frorn small boccas (openings),
aspect rario = ~ d ( l . r (gQ)0.7í
) ~ . ~ ~(4.2) or fram the overspill or breaching of a lava
lake ponded in a crater or
Aspect ratio therefore dcpends mainly on yield
(h) fire fountains of Iava that reconstitute around
strength. Equation 4.2 predicts that iavas with low
the vent and then flow away.
yield strengths, such as basalts, give rise to flows
with Iowes aspect ratios, and more-silicic lavas with Many eruptions of basalt lava flows begin with a
higher yield strengths will occur as h i ~ h e raspect phase of fire-fountaining of gas-rich magma, suc-
ratio flows, which is in general agreemenr with ficld ceeded by the exthsian of coherent flows of
observations (Fig. 4.2). relatively gas-poor magma. There would also be
From Equation 4.2, aspect ratio would seem to periods when lava is issuing as coherent flows and
be insensitive EO changes in effusion rate, but in fountains at the same time, either from the same
reality this is more complicared becausc of the effecr vent or separateIy along a fissure. Flows forrned
of temperature variations. A change in cffusion rate from agglutinated lava spatter are associated with
will lead to a change in the temperature within spatret cones and spatter rarnparts (Fig. 4. l a , Chs 6
 FEATURES OF SUBAERIAI. BASRLTIC LAVA FLOWS 65

& 13). Lavas in which obvious spatter fragments (a) Pahoehoe

are observed can be called clastogenic Iavas; frag-
ments wiIl be httened, stretched and defomed as
in some welded mffs, and in many ways they form
by an analogous mechanism to welded air-fa11 mffs
(Ch. 6). Irn
Sparks and Pinkerton (1978) suggested that the
loss of volatiles during ,lava fountaining has an
important effect on rhe rheology of rhe lava.
Degassing of the lava leads to considerable under-
cooling, rapid growth of quench crystallites, a rapid
increase in the viscosity and the development of a stecp flow imnt
I
-
,rough clinker top
high yield srrengrh. Thus, highly gas-charged
magrnas giving rise to intense lava fountaining are
likely to generate higher viscosity basaltic flows
with higher yield strengths. Magrnas with lower
initial gas contents shonld therefore f o m more-
fluida1 lavas from less-vigorous fire fountains or
lava lakes. The lavas empted in 1961 at Askja in \ massive lava
wlth bZodry joints
Iceland changed from higher viscosity and higher
Figure 4.4 tongitudinal sections through the two main
yield aa pahoehoe flows lSection4'5'1) typec af suberialbsa[tlc lava flaw (Aher L r i c h 0 d &
later in tlie eruption as the intensity of fire- Lipman loi8Dj
fountaining waned (Sparks & Pinkerton 1978).

characterised by smooth, billowy and sometimes

4.5 Features of subaerial basdtic lava ropy and "oe' surfaces. I n contrast, aa lavas have
flows exceedingly rough spinose and fragmented surfaces.
These are both end-mernber types with al1 tran-
Many of the feaiures of basalt lava flom have been sirions berween them; slabby and block lavas
well documented from studies in Hawaii, and we resemble aa, but are less spinose, ~vithfragments
refer the reader to the descriptions and illustrations that ase more regular in form. Pahoehoe and aa
in MacDonald (1967, 1972). Basalric lava flows commonly f o m in the same lava flow. Pahoehoe
contain a large array of surface features, but the may change downslope to aa, but the opposite has
preservation potential of many o t these in the never been observed.
geological record is very Iimited. We shall split our The early character of most lavas erupted on
description of some of the features of subaerial Hawaii are alrnast always pahoehoe. Puhoehoe is
basaltic lavas into the following: generally a very fluid, fast flowing Iava but it can
also form from viscous magma at low effusion rates.
pahoehoe and aa lavas Generallv small, highly mobile flows advance as a
flood basalts coherent unit with a smooth rolling motion. Larger,
plains basalts %SS mobile flows advance by protrusion of b u l h u s
'toes' of lava. On Hawaii, pahoehoe is common on
smooth, gentle slopes (see below), and tends to
4.5.1 PAHOEHOE AND AA LAVAS form rarher thin flows (often less than 1-2 m; Figs
These are the Hawaiian names given to the two 4.4 & 5a). Internally, pahoehoe lavas are character-
main types of basaltic lava Aow that have been ised by Iarge numbers of smooth, reguiar spher-
disringuished (Figs 4 . 4 6 ) . Pahoehoe lavas are oidal vesicles. Manv fio~vscontain more than 20%
Fgura 4.5 Ifacing page and above) Pahoehoe lava flwus. (a) Succession of five thin flow units exposed within the crater rim
of Mt Hamilton. Victoria, Australia. These flow unrts have non-vesicular bases with narrow oxidised margins. which grade into
highíy ~esicularupper and middle portions. White inclustons within the base of some flow unitc are locally dertved vetn quartz
xenoliths (61 Smooth. brllowy pahoehoe curface of the 1975 flow in Kilauea caldera, Hawaii. (cl Shelty pahoehoe. Mauna Iki. (d)
Crust of shelly pahoehoe. Mauna Ikr. (e) and ( f ) Ropy pahoehoe. Mauna Ulu 1969-74 flows (near Mauna Ulu). (g) Pahoehoe
toes in a Mauna Ulu 1969-74 flow fed by lava tubes dawn the Hilina fault cystem (about 8 km from the vcnt) Figure (circled)
indicates scale. Ihl Sectron through a pahoehoe toe buried within a compound lava. Mt Eccles, Victoria, Australia. (1)
Weathered ropy pahoehoe surface on the 500&6000-year-old Harman Valley flow. Wallacedale. Victoria. Australia.

vesicles, though ir is not uncommon ro find parts of low preservation porential in the geological E'ecard
Aows with 50% vesicles. (e.g. Fig. 4.5i). Ropy surfaces and toes may'be
Swanson (1973) described severa1 different types preserved, especially if quickly covered _over by
of pahoehoe lava flow formed during the 1969-71 another lava flow or flow unit (e.g. Fig. 4,5h): If
activiv from Mauna Ulu in Hawaii. A very found, mnvex trains of pahoehie topes can be uSed
vesicular, cavernous type, called shelly pahoehoe as palaeoflow direction indicators, although caution
(Figs 4 . 5 ~& d), formed when gas-charged lava is required in determining flow direction based on
welled out of a fissure with littIe or no accompanying only one or two occurrences, since some pahoehoe
fountaining. A relatively smooth and denser type ropes rnay be a response to local eddies on the flow
formed from the fall-out of fire fountains >300 m surface (MacDonald 1972). . . .
in height. The third type, characterised by over- When a thickened -trust foms on a flow, lava
lapping, denser (<2OC1/0 vesjcfes), pahoehoe toes tubes commonly form internally (Fig 4.7). Lava
and lobec (Fig. 4.5g), formed when largely degassed tuhes are alrnast exclusively restricted to pahoehoe
magma issued from tubes several kilometres from Aows. They can range in size from less tha' 1 m in
the vent. diameter ro large caves >30 m wide and P 5 m high,
The ropy íype of pahoehoe (Figs 4.52, f & i), and can form large distributary networks which can
although perhaps the most distinctive, is acrually carry lava below the nearly starjonary lava surface
quantitarively limited in extent (MacDonaId 1972). for distances of many kilometres. Some of the best
The ropes consist of a regular train of corrugations examples that have been described are from the
a few centimetres in height, their long axes being Quarernary basaktic provinces in Australia (e.g.
perpendicular to or convex to the local direction in Ollier & Brirwn 1965, Ollier 1969, Atkinson e1 al.
which the flow is moving. Fink and Fletcher (1978) 1975; Fig. 4.7), and Atkinson et al. (1975) report a
have done a structuraI analysis of these feanires. sysrem of lava tubes which may have extended for
They can be interpreted as folds which develop at "more than 100 km in north Queensland. Trrbes
the surface of a fluid whose viscosity decreases with may later collapse to produce large open channels
depth. The braided appearance and more-complex and depressions on the surface of older flows (Figs
strucnires found in many pahoehoe flows can be 4.7b & c),
explained by the superposition of two or more Lava tubes are important because they inhibir
episdes of folding . radiative heat Iosses from the surface of a flow, nnd
These pahoehne surface features generally have a enable the flow to travel long distances. Tube-fed
 FEATURES OF SURAERIAL BASAI-TIC LAVA FLOWS 69

Figure 4.7 (a) Cave formed by lava tube on Mauna Iki. (bl
Collapsed lava tube on Mauna Iki. (c) Cave exposed by
' collapse . of lava tunnef rmf. Byaduk
. .
Caves. Victoria.
Australia.

. .
, .

pahoehoe flows can achieve lengths much greatet f l o ~ are

~ s hornitos, pressure ridges and tumuli (lava
than aa flows of equivalen1 effusion Late. blisters; Fig. 4.8). Hornitos are small, rootless
Peterson and Swanson (1974) observed lava spatter cones up to several metres high formed by
tubes foming during the 1970-1 acrivity of híauna explosions due ro, for instance, trapped ground
Ulu, Hawaii. They were observed zo form by: water. Pressure ridges are elongate uplifrs of the
lava cmst, occurring subparallel to the flow direction
(a) gradual roofmg-over of a Iava srream from its at flow margins, but perpendicular in central
levées by the accretion of lava spatter along the
portions. They are thought to be due to vpward
edges and pressure from still-Iiquid lava flowing beneath the
(b) cooling of a lava surface to produce a crust, solidifying surface. Tumuli are srnall mounds or
beginning at the levées and growing in~rtard
dome-like blisters up to 20 m or more in diarnerer
and downstrearn.
on the crust of a lava flow, again caused by pressure
Ollier and Brown (1965) previously suggested that from underfiowing lava, or pressure associated.with
thick flows would develop shear planes, and that volatilisation of groundwater.
only the hottest, thickest layers would keep flowing, A a f í m s are generally thicker (frorn 2 to 3 m, up
leaving voids or tubes. However, Peterson and "0 about 20 m) than pahoehoe flows and advance
Swanson (1974) found no evidence for t h s in much more slowly. ?'he jagged flow-front (Eig.
Hawaii. 4.4b)creeps forward and steepens until a section
Other surface features that occur on pahoehoe becomes unsrable and breaks off. Collapse is

4 Figure 4.6 Aa lava flows on Hawaii (a) 1868 lava flaw on Mauna Loa (b) Detall of fragrnented clinker top to the 1868
Mauna Loa flow (c) Spinose top of pre-Missionan/ flow from Mauna Loa
70 LAVA FLOWS

Figure 4.8 (a) Pressure ridge in a thick colurnnar jornted

cooling crust. Wallacedale, Victoria. Australra Uplifted
columns have separated along the axis of the pressure ndge
to produce radial V-shaped fractures. Tilted vesicle layers tn
foreground are parallel to the forrner flow surface. (6)
Turnulus with large radial fractures f o m e d on the 1919
pahoehoe lava flow tn Kilauea caldera, Hawaii. (cj Turnulus in
columnar jointed flow surface, Wallacedale. Vrctoria.
Australia.

repeared as the flow slowlv advances in carerpillar- ments of aggregate gradually complete the transiuon
track fashion over an au~obrecciated layer of 10 aa.
fragmented lava. Internallv, aa lava is characterised However, aa lavas also form at vent. When lavas
by irregular elongate vesicles thar are drawn out in have a moderare to high viscosiry, pahoehoe lai-as
response ro interna1 flow, and a stratification will only form at low effusion rates, ivheseas aa \vil1
consisting of a solid massive Iava bodv sandwiched form when effusion rates have exceeded a critiml
between Iayers of fragmented clinker that may be value (this was 2 x lo-"' S - ' on Etna in 1975:
welded together (Figs 4.4b & 6). Sparks & Pinkerton 1978). O n the other hand,
The transition fmm pahoehoe tu aa is generally lavas with low initial víscosities will form pahoehoe
regarded to result from the increase in viscosity even at high effusion rates (rhere is no ljmiting
caused by cooling, gas loss and greater crystallinity effusion rate).
with time. Petercon and Tilling (1980) made a Roth pahoehoe and aa lavas form levées. In a
detaiIed studv of rhe transition, which occurs ar study oí Ievées formed by lavas of klt Etna, Spasks
some critica1 paint in the reIationship between et al. (1976) found four principal n p e s of levées
viscosity and rate of shear srrain. Ef the viscosity is (Fig. 4.9). Initial levées are formed because of the
low,.then the transition only occurs if rhere 1s a high yield strength of the lava, as indicated by the
rate of shear; for example, as caused bv fiow over a srudies of HuIrne (1974) (see Secrion 4.3.2 abovel,
steep slope. If viscosity is high, only a low rate of These form in both pahoehoe and aa flon-s.
shear is required. At the transition, srifT cIots form Accretionary levtes were observed near boccas, and
in parts of the flowing lava where the shear rate is consisted of piIes of clinker accreted to smooth
greatest, and rernaining fluld adheres to these. pahoehoe lava channels. The clinkers weld them-
Also, fragments of solidified pahoehoe crust are selves rogerher to form a sreep, solid lcvée. In flows
incarporated into the floxv, and masses and frag- where aa has developed fully, the flow front
 tensional stresses ser up during contraction may
produce regular joint sets perpendicular to the
cooling surfaces, and usuaIly vertical to sub-vertical
in oriernation. WeIl dcfined intersecting joint sets
may produce regular polygonal colurnns. The joint
faces (and columns) propagate inwards from she
initiol accrttionary
cooling surfaces as rhe 'cooling front' adtrances
inwards. This progessive propagation may be
reflected complernentary suh-horizontal joints
within colurnns, or by a segmentation pattern en
the vertical ioints, reflecting successive propagation
stages (Fig. 1.10b). Columnar jointing can exhibit a
nvo- or three-tiered arrangement (Spry 1962,
rubble ovcrflow
MacDonaId 1967, 1972; Figs 4.10b & d). The
Figure 4.9 Cross sections thmugh four different V P B Sof
lava levée observed on Mt Etna. Heavier stipple is mascive
bottom consists of rhick, usually well formed
lava; sparsely stippled areas represent flowing lava. {Alter coIumns noma1 ro the base of the flow. Above this
Sparks et al. 1976.) coIonnade, a Iayer of thinner, less seglar, of~en
chaotic coIumns essentially normal to thc flúw rop,
advances and the sides aIso expand by avalanching but highly irregular in,structnre, is found. This
of aa debris. These rubble levks are at angles o£ layer is caIled the entabhnzle. .There rnay be an
repose of 30-35". The fourth type, overffow levées upper colonnade abwe this. Twe-tiered colurnnar
form when lava repeatedly floods over existing jointing is common in [he Columbia ,%ver basalts
rubble lwées. Most levées on Mt Etna are h~brids (e.g. Swanson & Wright 1981).
of two or more of the four types. Thus, although Recently, Kanrha (1980, 1981) proposed that
Kulme's (1974) theory of levée formation was columnar joinring in basalts results from a fluid
confkmed by these observations, accretionary , dynamic process opmting in the lava during
rubble and overflo~vlevEes nucleate and modify cooling. Double-diffusive convective processes,
initial levées. due to temperature and chernical differences be-
hveen the top and-bottom of a stagnant meIt, are
rhought to drive columnar 'finger' motions in the
melt. When solidificauon eventually occurs, con-
Flood basalts form extensive sheets af lava with traction cracks would have preferred propagation
very low aspect razios (l'late 4, Fig. 4.10a). Com- paths along the bomdaries of adjacent 'basalt
positionally, these lavas are dominantly tholeiitic .fingers', dging rise to columnar jointing. Similar
(e.g. Swanson & Wright 1981), aithough cornrnonly 'sal1 fmgers' can be produced experimentaIly, and
they can be alkli basalts, e.g. in the Ethiopian also occur in nature in the 0ceans:Kantha (1980,
province (Mohr 1983; Ch. 13) and in the Deccan 1981) pointed out the striking similarities 8f-basalt
Traps of India (Krishnarnurthy & Udas 1981). columns to these. Although Kantha's ideas are very
They are pahoehoe flows, and sometimes raW interesting, not al1 columnar jointing can be attri-
surface features are preserved. Many of the larger buted to 'finger motions'. Welded tuffs, for ex-
flows of this type must have ponded as wst lava "ampie, often display very weEI developed columnar
lakes, taking years to tens of years to solidify, as jointing (Ch. 8) which cannot be explained by this
indicated by the welI developed massive mlumnar process .
jointing (Fig . 4.10). Idava tubes, lava channeIs, and other large scale
Cooling is accompanied by contraction, and flow features are generallv lacking in ilood basaIts.
takes place from the cooling surfaces (psincipally This may he because they did not form, or because
zhe top and bottom of the flow) inwards. The rhev were destroyed by later movements within the
Figure 4.10 (a) Flood tiasalt lava flow of the Picture Garge Basalt in the Culurnbia Rzvcr basalt pla:~au, 0:egori (ti) Icclandic
flood basalt with tower colurnnar jointed colonnade and upper entablature Note horizontal segmentation pattern on vertical
lotnts (see textl. tcl Top of colurnnar lointed flood basalt lava in lceland chowing pofygonal form of columns. (d) Two-trered
columnar iointing, Campaspe River. Victoria. Australia (e) Large uniforrn columns in a thtck. rnassrve flow. Organ Pipes,
Victoria. Australia.'
 SURMARKNE BASALTIC LALTAS 73

ponded iakes of lava, perhaps by convective circu- la)

lations. Large, circuEar down-sag structures have Contacts and Flow type
interno1
been described, which may result from magma characteristics
withdrawal. surfacc f e o t u r e
tfolds and whortsl'
7 l- 5implc sheet
Palaeofiow direcuon in ffood basalts can be
determined if spiracles or pipe vesicles are present.
These are concentrations of vesides in small,
curved pipe-like suuctures found at the base of
Aows. They form when bubbles of steam from
heated p u n d water rice into rhe lava, and are then
stretched in the direcrion of flow as it continues ro t compltx
shcct flow

move (Wattrs 1960, MacDonald 1967).

4.5.3 PLAINS BASXLTS

complca
sheet flow
Althongh these are large basattic flows, Greeley
(1977b, 1982) grouped them as a separate type from
fload basalts. Plains basalts have characteristics of
both flood basalts and the smalIer shield-building
pahoehoe lavas, such as those in Hawaii that were
[ undulaflng

pillows-

discussed above land in Ch. 13). They are typified

by the Snake Kver Plain, in the western USA
(GreeIey 1977b, 1982). Lavas have been erupted
from central vents to produce low coalescing
shields, or from fissures to produce sheets. Lavas
are compound, and flow units are up to about 10 m
thick. Lava tubes and lava channels are an import-
ant rneans of flow propagation.
coollng joints
quenched qinss '
(palagoniti 1
4.6 Submarine basaltic lavas fitling intarstltial covity

figure 4.11 [a) Succession of subrnarine prllow lavas and

The formation of pillo\vs or pi1Iow lava (Figs 4.11 & sheet flows (after Hargreaves & Ayres 1979).(b) D e t a l l of
12) wouId generally be regarded the most distinctive pillows. In cross section p i l l o w s can vary from 10cm
feature of basaltic lavas erupted under water. From spheres to large forms several metrec across They are
usually at least ceveral tens of centimetres in diameter.
studies of the present ocean floor and of ancient
successions, submarine pillow lavas are also known
to be intimately associated with massive or sheet discrete pillows represent cross sections through
flows (Fig. 4.1 1). interconnected lava tubes (Fig. 4.13). Although
There has been considerable debate about the erupted subaerial-, submarine observations of the
fonnation of pillow lavas (e.g. J. G. Jones 1968, v Mauna Ulu flows (J. G . Moore er
S ~ a r c h - ~ a1471
J . G. Moore 1975, Vuagnat 1975, de Wit & S t e m al. 1973) indicared that pillows formed and the lava
1978). In many two-dimensional outcrops, most advanced bv the budding of subaqueous Iava tubes.
pillows appear to be discrete entities, although This process is rherefore quite analogous to the
careful obsenrarion cornmonly reveals some inier- digital advance of subaerial pahoehoe lava and the
connected pillows. However, good three-dimen- formaiion of pahoehoe toes, as first suggested by
sianal obserwtiona1 dara show thai many apparently Lewis (1914).
74 LAVA FLOWS

Figure 4.12 (a) Prllows with well developed radial cracks

and thin quenched margins. Bnatmans Harbour. Oamaru.
South Island. New Zealand Inter-pillow spaces are filled
with pelagic and skeletal carbonate sedirnent. (b) Steeply
dipptng (tectonicl pillow lavas in the Franctscan Formation.
Califomia. Way up 1s f r ~ mright to left. (c) Troprcally
weathered pillows in the Rio Orcovis Formation, Puerto
Rico

Palaeoflow direcrions in pillow lavas can be

determined by measuring the direction of budding
from re-entrant selvedges (Fjg. 4.14). The shapes
of pil1o.i~~aIso allow the determínation of younging
directions in ancient deposits (e.g. Fig. 4.12h).
Massive flows of basaltic Iava have frequently
been encountered during sea-floor drilling, and
have been described as sheet flows from the
Galapagos rift valley (Ballard et al. 1979). Thcse
may have a variety of surface features, including
folds and whorls like subaerial pahoehoe, or they
may be flat or hroken. The transition from pillowed
to massive morphology, within a single flow or
between ffosvs (Fig. 4.11), could reflect an increased
discharge rate. Ralla$ et aI. (1974) interpreted
sheet flows as analogous to rnodern subaerial
unchannelled pahoehoe finws empred at high dis-
charge m e s , and pillow basalts as analogous to
c-sa-cl tube-fed pahoehoe lavas erupted at rnuch Iower
Figure 4.13 Plan view of. and three orientated cross
discharge rates.
sections through. pillnw lavas. (After Hargreaves & Ayres Submarine basaItic lavas are erupted either along
1979.1 fissures at mid-ocean ridges or from centra? vents al
seamounrs (Ch. 13). ,Mid-ocean ridge (MOR)
volcanic activity produces quiet effusion of pilIow
and sheet lava flows (Bonarti 1967). There is lirtle
physical interaction benveen lava and sea water,
and this is generally resuicted ro the fomation O£
thin glassy crusts. These may be minor quench
shattering and autobrecciation, or collapse pits with
breccias in sheet flows (Ballard cr al. 1979j, but
there is generally lirtle fragmenta1 volcaniclastic
material produced (Fig. 4.11). Seamounts have
been obsented ro have both pillow and sheet flows
at their summits (Lonsdale & Batiza 1980). They
also have estensive amounts of hyaloclasrite (Fig.
4.15). These may f o m debris flows (hyaloclastite Figure 4.15 Sketch of summtt area of a ceamount near the
stone streams of Lonsdale & Batiza (1980)) down East Pacific R ~ s e(After Lonsdale & Batita 1980 )

(a)
ruprvre ~n sbm uf
the sides of seamounts. Some of the largest may
oider pi!iow have debouched into the ocean basin and account
for the thick sequences of rhis type of deposit in the
ocean cmst found in off-axis drilling of mid-ocean
ridges (Schmuicke ct al. 1979). Lonsdale and Batiza
(1980) traced debris flows back into broken pillows
r e - eniront / and pillow lavas at the seamounr summit, and
/
,,
------- 'x.--<--- suggested that they were formed, in part, bv
hydrovalcanic explosions. Quench-fragmentation
and simple gravitational collapse are probablv also
important proccsses.
avwjng o f new
,+---- /--

4.7 Subaerial basaltic lavas flowing into

water
When basaItic lavas flow frorn land into water (e.g.
a lake, ñhe sea or glacial meltwater ponds formed
during the eruption of intraglaciail volcanoes), Iava
deltas are often built out from ahe shore (J. 6.
Jones & Nelson 1970, J . G. Moore el al. 1973). In
general, such deltas consist of a lower parí of
palagonitised hyaloclastite breccias and pillow lavas
characterised by steep foreset beds (up to 40")
I swhich have been termed flow-fmt breccias (J. G.
\ i
mclpreni 1 \ 'eventrant
re-entrant '- Jones & Nelson 1970; Fig. 4.16). 'These are capped
\,-cupture m \----" x---- /R+'x---.----d
with near flat-Iying massive lavas. A passage zane
sbin of
ywngesf pllow -
d
drrectron uf flow ond
pil/ow propogohoo
between these marks the approximate water leve1 at
Figure 4.74 Cross section showrng the development of re-
that time. J. G. Jones and NeIson (1970) showed
entrant selvedgec by budding ot a new pillow. (After how relative movements of water leve1 and a
Hargreaves & Ayres 1979 1 volcanic pile or rerraín over a period could be
 SUBAERIAL ANDI<SITICAND DACTIC LAYAS 77

I
'Figure 4.18 Andesitic 2nd dacitic lavas bi k i d e s i ! : ~ block lava eruptcd 'riigti on Colirria voltario. Mexico tn 1975 Note thc
well developed levees. tb) Uacitic block lava on Nea Kalmsni. SaMorini (c) Ths 1981 Mt St Helens dacile dome lafter M & K
Kraft tn Christiansen & Peterson 19811. (d) The 1902 dacitic spine of M t Pelee (aiter Bullard 1976) ( e )Gros Prton (daciric), St
Lucia. West Indies
4.9 Emption of subaerial rhyolite lava Mexico (Fig. 4.2 la). R. L. Srnith and Bailey (1968)
fiows suggested t hat extrusion of rhyoIite lavas comrnonly
follows resurgence of magma afcer climactic ignim-
As far as ute are aware, there has only been one brite eruptions which resuit in caldera subsidence
observed historic eruption of rhyelite lava. This (Ch. 8). In many of these situations it seems that the
was during the 1953-7 activity which formed the lavas haw been extruded around the ring fanlt or
Tuluman Islands, two new islands in the St Andrew fracture on which caldera coilapse took place. At
Strait, norrhern Papua New Guinea ( A l . A. Rey- Mono Craters an arcuate Iine of rhyolite lavas is
nolds & Best 1976, M. A. Reynolds et al. 1980). thought to represent activiq 01-er part of an actively
The final phase of the esuption, beginning in developing ring fracture system around the foun-
November 1956, produced cubaerial lava flows. dering roof of a large crustaE magma body in a pre-
Earlier phases were characrerised by dominantly caldera stage of evolution (Hemance 1983, Rundle
submarine activity, and pruduced fields af Roating, & Eichelherger 1983). In some examples (e-g. La
vesicuIas lava blocks. iMany rhyolite Iavas and Primavera, Fig. 4.21a) the caldera may become
domes often occur in arcuate distributions about frlled with a lake, and these post-caIdera rhyolites
central calderas or volcanic depressions, as seen, for zrre emplaced in association with lacustrine sedi-
example, in the Taupo Zone of New Zealand. the ments of the caldera. Hoiverer, the geology of
Va!les and Long Valley calderas and Mono Craters rhvolitic volcanic centres will be expanded upon
in rhe western USA and La Primavera volcano in later, in Chapter 13. The Tuluman Islands also

(a 3 La Primavera volcano (bl Southern Cipari

m rhyolrte laros
with ogires
m coidem lake sedirnenti

""~"C"'
m ernbryc caldera morgin
-100-contours in metrep

Figure 4.21 (al Map of the distaibution and surface features of the rhyolite lava flows of La Primavera volcano. Mexico Sorne
oi the linear features are faults. curved features are oqrves (see Plate 2 and text) (after Clough 1981i (b) Map of t h e rhyolite
domes of southern Lipari. Aeolian Islands (after Richardson 1978)
80 LAVA FLOWS

seem to occnr on an arcuate line of thyolite lavas, Many rhvoIitic lavas are associated with pyro-
but here M. A. Reynolds et al. (1980) have clastic deposirs, each lava being aImost invariably
speculated that rhese lie above a ring fracture associated with precedirig phases of explosive pyro-
developing above a mantle hot spot. clastic activity. The style of explosive acrivizy can
In other areas, rhyolite lavas do not seem to be vary from mainly phreatic eruptions, producing
obviously associated with a caldera, e.g. the spec- rings of lithic breccias surrounding -the lava, e-g.
tacular wncentration of ~hyolitedomes in the Panum Crater (one of the Mono Ctaters), to highly
southern part of Lipari in the Aeolian Islands, Italy explosive piinian and ignimbrite-forming eruptions.
p i g . 4.2lb), and in Papva New Guinea on the During such explosive phases a pumice cone or tuff
DJEntrecastreauxIslands and at Talasea in New ring can be built-up around the vent (e.g. Fig.
Britain (1. E. M. Smith 1976, Smith & Johnson 4.22e). Even while the rhyolite lava is growing,
198 1, Lowder & Carmichael 1970). explosive eruptions may continue, and evidence for

Figure 4.22 Riyo'rtic Idvac. ( a ) Mt Gunrdia d o ~ c Lipari

. Iphotoqrdpti by S tJnl) (hl Crrro kI CPdto uoriin. a Pr r~i;vt!r,i
volcano {ci Cerro El Colli mesa lava. the youngest of the La Prrrnavera lavas (after Clough 1981) (d) Mesa FI Malahuate m r i a
lava. l a Primavera volcano (e) Coulée of La Primavera volcano which has flowed to left from a vent at the summit of the
pumice cone seen to rnght (after Clough 1981)
 SUBAERIAL RIIYOI.I?'TC L.iVii FI-OW: FEATURES 81

this would be unusually large amounts of obsidian

4.20.1 SHRPE
ejecra amongst rhe pyroclastic deposits. Also, .risible
craters may be present on some of rhe rhyolice Rhyolite lavas (Fig. 4.22) can be subdivided
domes, and this can be seen in sorne of the domes of according ro lheir shape into:
Lipari (Fig. 4.21b). Formation of cratered domes is
(a) donaes (or tholoids), which are circular in plan
attribured to rhe lare stage builil-up of gas beEow the
with a small surface area (Figs 4.22a & b),
viscous magma, which is released in the form of an
explosion with no fresh magma effusion. (b) mesa h a s , which are lavas with an approxi-
rnateiy circular plan forming biscuit-shaped
iMost rhyolite Iavas would seem to reach the
b d i e s (Figs 4 . 2 2 ~& d) and
surface through a circular conduit, which presents a
much smaIler cooling surface to the country mcks (c) cmlées, which are lavas which f o m when fiow
is asymrnezric and concentrated to one side of
than a fissure vent. Howwer, some of the dome
the vent producing an extnision which is
lavas of La Primavera are elongare, and surface
elongate in plan (Fig. 4.22e).
ridges are parallel to the caldera ring-fault (Fig.
4.21a), suggesting that rhyolite lavas rnay also be Alrhough these terms most commonly apply to
extnided along fissures (Clough 1981). The Cirgle rhyolitic lavas, they can alse be used to describe rhe
Creek rhyolite in Nevada is thought to have been form of sume dacite, and even andesite, lavas.
entpted zhrough a Iarge fissure sysrem (Coates These three lava types develop in respense to the
1968). This is a rhyolite lava flow covering 130 km2, varying conuoIling factors discussed previciusly..
uith multiple venzs aligned on linear trends. These Rhyolite lavas have a wide range in thickness fiom
trends are rhoughr ro represent fissures which 4 0 m to >500 m (Fig. 4.2). However, the
closed to a series of subcylindrical vents as lava was average thickness is probably a b w t lOOm (in
extmded, in a similar manner zo that observed in descriptions in the literature and Fig. 4.2, rhyolíte
basaltic fissure eruptions. It w a s suggested that the domes tend to be over represented compared with
motive force for this eamsion was the weight of tht thinner coulées, because domes tend to survive
fissured cmst that downsagged into a magma much longer as topographic features).
charnber, thus forming a sag-basin as opposed to a Some of [he thinnest rhyolite lavas known are
caldera. In eastern IceIand, Gibson and Walker aphyric, and have bem ascribed unusuaily low
(1963) have traced composire or rnixed rhyolite and viscosities. For emmple, the early rhyohtes of Long
basalt Iavas to composite dykes which would be Valley caldera (Bailey et. al. 1976) contain some flow
feeder fissures. The Tarawera Volcanic Complex of units onIy S0 m thick which have flowed up t o
New Zealand (Cole 1970) consists of a cluster of 6 km. The aphyric condition may suggest an
rhyolite domes and associated pyrocIastics, aIong a extremely high magma temperarure at the time of
NNE trend which reflecrs a crustal fissure-fracnire eruption as the cause of the increased fluidity.
zone within the Okaraina Volcanic Centre. Other thin rhyolite lavas recorded are the acid-basic
mixed lavas of Iceland (Gibson & Walker 1963),
having an average thickness of 60 m. These lavas
4. PO Features af subaerial rhyolite lava may have had reduced yield strengths and viscosity
fiows due to superheating on contact with basic magma.
One of the factors which couíd determine the
Our description of the features of subaerial rhyolite shape of rhyoIite lava flows is the presence of a
lava flowc can be subdivided inro the following: confining crater built by earlier pyroclastic ex-
plosions. This is prohablv true for smaller rhyolite
shape domes, but larger ones rnay exceed the critica1
lithology crarer volume and flow away from the crater arca
surface features laterallv. Manp domes, on rhc other hand, do not
growth and interna1 srructure appear ro be associated with an easlier construc-
82 LAVA FLOWS

tional crater breceding pyroclastic material being which are generally termed 'inm'zie rhyolices' by
More widely dispersed from the vent), bur it is workers in ancient successions, based on demon-
possible that such a feature could have ibeen strable intrusive contacts. The Devonian Boyd
complerely submerged beneath the succeeding Volcanic CompIex in Eastern Australia shows
dome. excellent exarnples of such intrusive rhyoIite lavas
Rhyolite lavas also ctimmonly form cryptodomes, (e.g.Fig. 3.24).Some of the rhyolite domes of La
 SUBAERIAI, RNYOLITIC LAVA FLOKJ: FE'A'I'URES 83

Figure 4.23 (facing page and above) Lithologies of subaerial rhyolite lava flows. (a) Flow-banded obsidian from a glass flow.
Newberry Crater. Oregan. (bl Platy lointed obsidian dome, Okataina Complex, New Zealand (c) lnterbanded obsidian and
spherulitic layerc. Rocche Rosse flow, northern Lipari. id) Flow-folded obsidian (now partly perlrtised), La Primavera volcano. {e)
Flow-folded Upper Devonian rhyolite lava at Tathra. New South Wales, Australia (photograph by S. Ralser) ( f ) Stony rhvolite
lava with basaltic inclusions. southern Lipari (photograph by S. Hall).

Primavera volcano which are found in conract with Black, vitrmus obsidian sometimes occurs as
'

caIdera Eake sediments (Fig. 4.21a) are thoughr to thick foliated Iayers, often interbanded or as lenses,
have formed as cryptodomes (Clough el al. 1981, within layers of the other lithologies (Fig. 4.23).
1982; Ch. 13). Caldera lake sediments are locally This layering, or flow foiiation, is frequently
folded and faulted, and invariably dip away from folded. Obsidian usudly foms a chilled glassy
the rhyolite lavas. carapace around rhyolite lavas, commonly about
10 m thick over the top and around the flow front,
$i.ith a thinner layer along the base (Fig. 4.24). The
4.10.2 LITHOLOGY
cores of many lavas usualIy consist of stony
In rhyolite Iava flows a variety of lithologies and rhyolire. Some of the thinner 'obsidian flows' and
textura1 features can be found: obsidian, layers 'glass flows' may be obsidian throughout their
, containing spherulites, pumiceous kayers, horizons
of stony rhyolite (lithic rhyolite), and in lavas where
interiors.
Spheruiites are radiating aggregates of alkali
hydration of ohsidian has occurred, periite. feldspar, with or withou~cristobalite and tridymite,
84 LAVA FLQWS

Obsiplan pmiceous block many young rhyolite lavas, however, llttle nf ~ h i s

rock is seen because erosion will have had insuf-
ficient time ro cut through and expose the interiors.
c4sidian -e- Rands and Ienses of Iight grey perIite are formed
by the hydration of obsidian. Obsidian adsorbs
water from the armosphere, forming an hydrated
layer which rhickens wirh time as the water diffuses
into the glass. From measurements of the thickness
of the hydration rind on artifacts collected from
co-cruptive
purnice fall dtposit ~,,,,
archaeologicaf sites and experimental studies (1.
Figure 4.24 Schematic section showing distribution of ~ ~ & L~~~ i 1976),~ it is known
d {hat~ rhe ~ ~
Iirholcgies in a rhyolite lava flow
square of rhe rind thickness is approximately
-. proportional to time, and varies from Iess rhan
which are cornrnonly four~din the glassy carapace . 0.4 (prnj2 per 1000 years to as much as 30 (~m)'
(Fig. 4.25a; Ch. 14). They commonly have dia- per 1000 years. This variarion is p a r t l due to
meten of 0.1-2 m, but can be much Iarger and
occasianally grow up re nearly 10 cm. They often la1
camprise specific flow layers (Fig. 4.23~).However ,
-0
rhey are usually superimposed on flow structures,
and the flow foliation can-betraced chrough, and is
not .deflected by- the sphemlites, showing that
crysralIicati0n.-tmk place after flowage of the lava
had riearly ceased. Facters governing the develop-
ment of spherulires are discussed by Lofgren -scm
(1971a; Ch. 14). Higher water contents in some
layers could promote gmmh rates of spherulite
fibres -1ocally.
Some cphemliuc .growths are, in fact, litho-
physae, which are radiating aggregates of fibrous
crystals which have formed around an expanding
vesicle (Fig. 4.25b). These vesicles have formed in
a rnelt while it was s¿11 capabIe of flowing. Crusred
and broken lithophysae often testify ro Iarer flowage.
More-vesiculated pumiceous layers rnay occur
interbanded with absidian and spherulitic layers.
MaRp rhyolite lavas are capped by blocks of pumice
or more purnicmus Iava (Fig. 4.24). In older flows
these are unlikely to be preserved. Pumiceous
breccias forrned during flow can.alsobe found at
their bases (Fig. 4 . 2 0 , as- well as co-emptive '-
pumiceovs pyroclastic deposits.
The principal 1ithoIogical componen1 of most
rhyolite lavas, especially domes, is foliated srony /
Iithophysoe \vold

rhvolire 'Fig- 4.231)- This is formed post- Rgure 4.25 (a) White spherulites in blsd. glaszy obnidian.
eruption crysta'lisation of the me't to a finely this attractive rock rc sometimes called 'snowflake' obsidian
c ~ s t a l l i n erock. This maY OcCUr during emplace- rb} lithophycae in perlitrsed flow banded obsrdrari. Boih
ment, as well as during snbsequent cooling. With exarnpies from Mexico
 SURAERI'AL RHYOLITIC LAVA FI,OW: FFATURES 85

differences in composition of rhe obsidian (trachytic have spacings of rens of metres. Ridges on rhyolite
obsidian hydrates much more rapidly h a n rhyo- lavas have been variously inrerpreted as ramp
litic), and partIy to differences in dimare. structurex (MacDonaId 19721, folds on the surface
In older rhyolites the obsidian is divided into of the Iavas ( h n e y ' 1968, -%k 1980a) or as
smaIl rounded kernels of stntcrureless obsidian 'squeeze-up' extrusions through cricks during
about 1 cm across, surrounded by concentric crack5 stretching of the ,flow surface..
in light grey friable perlite {this can be seen in the Loney (1968)described the .~outh& Coulée of
flow-foIded obsidian in Fig. 4.23d). The cracks Mono Craters, in which ogives there were open
(called perlitic macks) which subdivide the glass are anticlines within the flow foliation caused by
probably created by thermal stresses set up during longitudinal compression of the lava stream due to
cooling of rhe lava. Water then diffuses from these externa1 resistance in their advance, and the analogy
cracks into rhe bordering glass, and an hydration was rnade between these features and the compres-
fronr advances on the residual kernel of non- sion waves in glaciers. Fink (1980a) suggesred xhat
hydrated obsidian. The front is visible in thin some - surface ridges have a similar origin : to
section because the refractive index of the hydrated pahoehoe ropes, and presented results of foIding
glass is lower rhan rhat of the non-hydrated glass. analysis ro show them to be compatible. The
Ir is also becoming increasinglv apparent that wavelength and amplitude of the folds are depen-
many rhyolite lavas contain a very srnall. but dent on the íemperature gradient, the .contrast
significant, proportion of basic inclusions. Basaltic
inclusions are found in severa1 of the lavas of La
Primavera volcano, in some of the rhyolites in
dome
southern Lipari (Fig. 4.23d) and jn some from the
Late Devonian Boyd Volcanic Complex, sourh-
eastern Australia. These basic inclusions e h i b i t
varying degrees of origina1 fluidi-. Some show
embayed and crenulate margins, indicaung rhat the
inclusion was still Auid while the h o s ~obsidian uas I b ) Mt Guardia 1
Iiqüid. In orher examples the basic inclusions show
angular contacrs, showing thar they had cooled
sufficiently ro solidifv while the lower melting
temperature obsidian was still capable of fiotving.
The inclusions within La Primavera lavas are very
similar to the basaltic-andesite lavas whjch have
erupted around the volcano. Alrhough these basic
inclusions, with their fluida1 characteristics, form (cl Growth of
only a voIumetrically small proporlion of their host
lavas, their presence indicates that these are rnixed
lavas resulting fsom the coexistente of rhyolitic and
basaltic magma (Ch. 3). This bimodal association is
characteristic of rhyoIite volcanoes (Ch. 13). S

4.10.3 SURFACE FEATURES injecrion o t rhyolirr l o v a

The uppes surfaces of young rhyolite lavas are Figure 4.26 (al and Ib) Crosc sections through two of the
rhyolite domes in southern Lipari (located on Fig 4 21 bl
typically bIocky and rough. They have cunred, Horizontal scale same as vertical Mt Guardra is also pictured
concentric ridges called ogives (Fig. 4.21a 8: Plare in Figure 4 22 (c)3-dimensional vrew of the growth of a
2). These are concave in the upflow dircciion, and rhyolite dome (after Richardson 1978)
86 LAVA FLO\XrS

- +>

Figure 4.27 Ramp stcuciure rn the Rocche Rosse obsidian coulée, northern Liparr (phorograph by S. Hall)

between [he surface and interior viscosities and the (see below) rather than surface folds. Fink's analy-
ratio of the compressive stress (due to flow) to the sis can only applv to a limited period during the
gravitational stress (due to the weigh~of lava). emplacement hisrory of a shyolite lava flow. Once
Fink's analysis also allows calculation of the rnini- the surface is cold, it will behave in a bnttle
mum viscosity of many flows, for which such data manner, and fracruring and ramping could then be
are unavailable. irnportant and superimposed on these earlier
The predominan1 Row foljation orientation on features.
the surface of La Primavera rhyolite lavas (Fig. Other surface fearuses of rhyolite lavas include
4.21a) is subvertical (Clough 1981). The same is pumice diapirs. Fink (E9XOb, 1983) described in
true of the rhyolite lavas of Lipari (Figs 4.2628). detail pumiceous diapirs in rhe Rig and Litrle Glass
This suggests that in these cases the ogives are an Mountain obsidian flows, northern California
outer rnanifestation of an interna1 ramp structure (Plate 2). These rose from the basal pumiceous

Figure 4-28 Cross snction through t h e length of the Rocche Rosse obsidian coulke, with generalised flow foiialion patterns.
(After Hall 1978.1
 KOMATIITES 89

and flow volumes are two ordws of rnagnitude

Iarges than for most siiticic lavas. Cas (1978a)
suggested that, although subaerial siIicic flows are
demonstrably viscous and immobile, subaqueously
erupted and emplaced ones may behave fluidly md
be highly mobile. This would be because of the
inability of volatiles to escape under the high
h y d ~ s t a t i cpressure in a deep-water environment
(Ch. 3). Such an explanation is needed to account
for the wide areal extent and voluminous nature of
the silicic submarine Merrions Tuff lavas. A
consequence of this is that the products sif deep-
water silicic volcanisrn may be markedly different
from those of subaerial, or shallow-water silicic
volcanism. However, we know of no similar des-
criptions in the Eteramre. Some accounts in earlier
volcanic literanire have described exrensive sub-
marine rhyolite lavas from North WaIes, bur these
unirs are now known to be welded ignimbrites.
Investigation of modern oceans has not found
Iarge silicic flows, although some oceanic ishndc
are known to have erupted small. voIurnes. ,This
suggests rhat therc may also be somc tecronic
conrrol on the nature of this - p e of volcanisrn in
the past, and. we will discuss large ensialic sub-
marine basins in Chapter 15.

Figure 4.30 (a) fsopach maps of the three extensive silicic 4.12 Komatiites - peculiarities of the
lavas within the Merrions Tuff, Australia These are con- Archaean
ctructed relative t o the unfolded configuration of the
formation, isopachs in metres (b) lnferred sequence of The discussion about &e physical properties of
emplacement and spatiar relatronshrps of the lavas. Not
magmas in Chapzers 1 and 2, and of the features of
shown are the deep-water marine sediment horizons
separating each unit. (After Cas 1978a.l lavas and their emptions discussed in this chapter
so far, have been reIwant to 'normaI1 or penecon-
temporary magmas and lavas of the Phanerozoic,
exosion accompanying the emplacement of sedi- and even rhe bulk o l the Proterozoic. However,
ment overlying one oi' the anits, and their presence during the Archaean there was a unique, but
has influenced later patterns of sedimenration. comrnon group of magrnas and lavas called korna-
TmuralIy, the porphyries show no evidence of tiites. Their unique character appears to have been
vitriclastic texture, broken crystais, pumice or %ot just compositional, but also in t e m s of their
foreign lithic fragments, which may suggest they physical properties and behaviour as lavas.
were deep-water welded ash flows or ignimbrites Komatiites are ultrabasic volcanic rocks - either
(see Chs 8 & 9). In places the units have fragrnented lavas, tuffs, hyaloclastites or autobreccias (Arndt &
margins, thought to be partIy quenched and partly Nisbet 1982). Compositionally, komatiites are
autobrecciated. unique because they are extzemely high rnagnesian
Aspect ratios of the two larger units are ve- low, rocks, with MgO conrents greater than 18% (Arndt
90 LAVA ELOWS

Komariites are thought to have been erupted at

ternperatures benveen 1400 and 1700"C, and vis-
cosities have been estimated at 10-100 poises,
compared with recent basalts with ternperatures of
1200°C and viscosities of >5000 poises (Huppert et
al. 1984; also see Tables 2 . 2 4 ) . Huppert et al.
(1984) calculated that the Reynotds Numbers for
komatiitic magrnas being dischrged at varying
velocities from fissures of varying width, may have
been in the range 5.2 x lo3 to 1.0 x lo6, well
above the critica1 boundary of 2000 distinguishing
laminar flow from turbulent flow conditions (Ch.
2). Komatiires therefore probably flowed turbu-
lently, and their rheology, unlike most lavas, would
not have been like a Bingham substance, but more
likeIy pseudoplastic or even Newv~onianin eharactes
(Ch. 2). For a fissure only 10 m wide and 100 m
long, magma discharge rates rnay have been
>10 000 rn3 S-' (cf. Table 4.1).
Fluid turbulence in lavas facilitates convective
heat loss- and this was probably responsible for rhe
abnormaIly high cooling rates in komatiites, and
he consequence of this, spinifex texture (Huppert
et al. 1981).
The extremely hgh temperatures of komatiites
may have aI1owed them to meEt and thermally erode
Figure 4.31 Spinifex texture charactcristic of komatiitic
lavas. (After Donaldson 7982 )
the substrare over which they flowed, producing
lava channels incised into, and bounded by, the
& Nisber 1982). This is still much higher than substrate (Huppert et al. 1984).
normal basalts (<lo% MgO), and there is an Komatiite lavas have rhicknesses ranging from
intermediate group of rocks, called komatiitic <1 m to >IO m. They frequently form compound
basalts, with MgO contents between 18% and 10%. lava units (Fig. 4-31, and single cooling units
Mineralogically, rhe rocks are ultramafic, the consisting of multiple flows that essentially cooled
only crystalline phases in nearly al1 cases being as one (cf. ignimbritec, Ch'. S). Many normal
olivine, pyroxene and chromite, which have usually fearures of basaltic lavas have been described in
been pseudomorphed by rneramorphic assemblages. komatiites: pillows (rare), polygonal cooIing j oint
TexturaEly, komatiites are distinguished by the sets, vesicles, and brecciated rnargins of both
presence of needle-shaped crystaIs or crysrallites of hyalocIastite and autobreccia. Komariites therefore
oIivine which ciiss-cross each other and/or are appear to have flowed in subagueous, and probably
arranged in radiating sheaths or cIusters known as in subaeriaI environments as well.
spinifex texture (Fig. 4.31). This texture gives Frorn h e snzdies 6f Pyke et al. (1973) and Arndt
many komatiites a distinctive appearance in hand et al, (1977), a general facies model highlighting the
specimen or thin section. Komatiitic basalts can interna1 textura1 zonation of kornatiitic flows (Fig.
have pyroxene spinifex textures, rather than olivine 4.32) has been wideIy accepred and used as a darurn
spinifex textures. Spinifex textures are cornmonIy for cornparison in other studies of komatiitic rocks
accepted as resulring from the exrremely rapid (e.g. Arndt & Nisber 1982).
cooling (also called quenching) of the lavas.
 FURTrIER REAVING 91

Overlying
flow unit

Underlying
flow unit olivine cumutote
O 1 Pmetres
1

Upper w r t ot flor unit

Chilled and thifled flow lüp Chilled f l m top with
4, fractured flw Al wLth fine poly- A fine polyhedral
hP hedrel jointtng pinting
A 2 Sprnifex AL Spinifex Main port of tlow ,
L w e r part of flow unlt Mediurn- ta fine- -
medium Io fine-
Foliated sRleMZ grained perrdotite B grarned pcridotite
" ctivine -
82 - -
Mndium to Rne
B4 grained ptridofttt
- uith conrse p o ~ y
hedrol lointing

B 3 Knobby peridotite

Figure 4.32 Facies rnodels for (a) komatlite lavas and íb) komatiitic basalt lavas, highlighting t h e interna1 textura1
4After Arndt et al. 1977 )

4.13 Further reading

For a quan~tarive analysis of the eruption of siIicic lavas and Cousine~uand-Dirnroth (1982) for
basaltic lava, the most rigerous study is that by L. an andesitic subrnarine pile. F i k (1983)'is the best
Wilson and Head (1 981); we shall refer to rhis work published account documenting the strucmre and
again in Chapter 6 . Some of the best descriptions emplacement of a rhyolire flow, and is a good guide
and interpretative analyses af submarine lavas come ro the Vpe of strucniral mapping needed tu study
from detailed examination of Archaean successions these lavas, The geology of Amhaean kornatiite
in Canada. We recommend to the reader the lavas is summarised in Arndt and Nisbet (1982) and
excellent studies by Dimroth et al. (1978) and Nesbitt e¿ al. (1982); however, the main approach
Hargreaves and Ayres (1979) for submarbe basaltic bns been the study of their geochemisrry and
lavas, de Rosen-Spence et al. (1980) for submarine petragenesis.
Plate 5 Pit dug to study a section through a young scoria-fall deposit erupted from Sunset Crater in about AD 1065. San
Francisco Volcantc Field. Arizona. USA Most pyroclastic deposits when freshly erupted are loose fragmenta1 aggregates - so
use a spade and leave your hammer a! home?