THE IMPORTANCE OF PHOEHOEscott/Workshop_reading/Self_etal_199… · P1: NBL/vks P2: NBL/plb QC: NBL/uks T1: NBL March 2, 1998 15:12 Annual Reviews AR055-04 PAHOEHOE 83 small pahoehoe

P1: NBL/vks P2: NBL/plb QC: NBL/uks T1: NBL

March 2, 1998 15:12 Annual Reviews AR055-04

Annu. Rev. Earth Planet. Sci. 1998. 26:81–110Copyright c© 1998 by Annual Reviews. All rights reserved

THE IMPORTANCE OF PAHOEHOE

S. SelfHawaii Center for Volcanology, School of Earth and Ocean Science and Technology,University of Hawaii, Honolulu, Hawaii 96822; e-mail: [email protected]

L. KeszthelyiHawaii Center for Volcanology, School of Earth and Ocean Science and Technology,University of Hawaii, Honolulu, Hawaii 96822; Lunar and Planetary Laboratory,Tucson, Arizona 85721; e-mail: [email protected]

Th. ThordarsonCSIRO, Exploration and Mining, Wembley WA 6014 Australia;e-mail: [email protected]

KEY WORDS: lava, basalt, pahoehoe, flood basalts, environmental impact of volcanism

ABSTRACT

Pahoehoe lava flows are common in every basaltic province, and their submarinevariants, pillow lavas and sheet flows, cover the bulk of the Earth. Pahoehoeflows are emplaced by inflation—the injection of molten lava underneath a so-lidified crust. Only in the past few years has an understanding of the inflationprocess and the ability to recognize ancient inflated lava flows been achieved.All large terrestrial basaltic flow fields studied to date, including flood basalts,were emplaced as thermally efficient, inflated, compound pahoehoe sheet flows.This leads us to propose that this is the standard way of emplacing large lavas(the SWELL hypothesis). The atmospheric impact of such flood basalt eruptionscould have been protracted and severe, providing a plausible link between floodbasalt eruptions and mass extinctions.

INTRODUCTION

The Importance of PahoehoeThe most common rock type at the surface of the Earth, and on the other terres-trial planets, is basalt. Basaltic lavas come in two forms: aa and pahoehoe (from

810084-6597/98/0515-0081$08.00

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



82 SELF ET AL

the Hawaiian‘a‘ a andpahoehoe). Pahoehoe flows have often been thought ofas small, slow-moving, inconsequential lavas. It is thus not surprising that theprocesses involved in the emplacement of large, fast-moving, channelized aaflows have received greater attention (see Kilburn & Luongo 1993, Crisp &Baloga 1994, Pinkerton & Wilson 1994, and references therein). However, asin the fable of the tortoise and the hare, it is the slow but unrelenting pahoe-hoe lava flows that ultimately grow larger and longer than the spectacular butshort-lived channelized rivers of lava that produce aa flows.

In terms of both areal coverage and total volume, pahoehoe flows dominatebasaltic lavas in the subaerial and submarine environments on Earth. The mostabundant type of lava, submarine pillows, is closely related to pahoehoe in theirstyle of emplacement (e.g. Macdonald 1953, Williams & McBirney 1979). Acompilation of the rather sparse information on intermediate length (50–100km) and long (>100 km) lava flows on the Earth (Table 1) shows that pahoehoeis far more common in these larger flows. Several large extraterrestrial flowsalso seem to be pahoehoe (e.g. Theilig & Greeley 1986, Bruno et al 1992,Campbell & Campbell 1992). The emplacement of pahoehoe flows is thereforea fundamental process in crustal formation on the Earth and the other terrestrialplanetary bodies.

Recent AdvancesUntil the groundbreaking work of Hon et al (1994), understanding of pahoehoeflows was largely qualitative, and there seemed to be no connection between the

Table 1 Examples of large basaltic lava flow fields

Length VolumeNamea Age (km)b (km3)b Dominant surface

Mauna Loa, Hawaii 1859 >47 >0.11 PahoehoeMauna Loa, Hawaii 1859 >51 >0.27 AaCarrizozo, New Mexico ∼1 Ka 75 4.5 PahoehoeToomba, Australia 13 Ka 120 12 PahoehoeLaki, Iceland 1783 65 14.7 Pahoehoe8◦S, East Pacific Rise ∼1960 18 15 UnknownEldgja, Iceland 934 >70 >15 PahoehoeThjorsa, Iceland 8.5 Ka >140 >21 PahoehoeUndara, Australia 190 Ka 160 ∼25 PahoehoeNorth Arch, Hawaii <2 Ma 100 >25 “Sheet”Pomona, CRBGc 12 Ma >600 >760 UnknownRoza, CRBG ∼15 Ma 350 1300 Pahoehoe

aListed by increasing volume.b> indicates that only subaerial portion of flows reaching the sea are given.cCRBG = Columbia River Basalts Group.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 83

small pahoehoe lobes in Hawaii and the enormous pahoehoe sheets that makeup large lava flows such as flood basalts. The understanding of this relationshipcame from recognizing the inflation process, which can convert 20- to 50-cm–thick lobes into sheets many meters thick over a period of weeks. Inflation(also known as “endogenous growth” or the “lava rise” mechanism; Walker1991) is the lifting of the upper surface and crust of a flow by the injection ofadditional liquid lava into the molten core of the flow. The word “inflation”was applied to pahoehoe flows as early as the 1950s (Macdonald 1953) anddoes not imply that the lifting is done by the gas phase. The formation ofinflation features was described by Theilig (1986) and Walker (1991), but thecomplete picture emerged only after the analysis of field measurements fromactively inflating flows in 1988–1990 on Kilauea Volcano (Hon et al 1994).This comprehensive understanding of how pahoehoe lava flows work providedby the active Kilauea lava flows has allowed decades of observations fromaround the world to be tied together into a single, self-consistent, field-testedtheory.

Since the initial work of Hon and colleagues, other studies have expandedand refined their Kilauea results (e.g. Keszthelyi & Pieri 1993, Chitwood 1994,Thordarson 1995, Self et al 1996, 1997, Keszthelyi et al 1997). In this paperwe review the results of the last few years of investigations on the emplacementof pahoehoe flows. The smooth, billowy, or ropey surfaces are only one of theintrinsic features of inflated pahoehoe flows; the internal distribution of vesicles,crystallinity, and joints are also diagnostic of the inflation process (Thordarson1995, Self et al 1996, 1997, Cashman & Kauahikaua 1997). In fact, morequantitative constraints have been placed on the emplacement of pahoehoeflows using their internal structure than surface morphology. Perhaps the moststartling result has been the idea that enormous continental flood basalt (CFB)lava flows were probably emplaced over one or more decades (Thordarson1995, Thordarson & Self 1996, Self et al 1996, 1997) instead of the few daysor weeks suggested by earlier models (Shaw & Swanson 1970).

OverviewIn the following, we first describe the formation of typical pahoehoe lava flows,starting with small features and ending with the product of an entire extendederuption. By recognizing and understanding the formation of different featureswithin ancient pahoehoe flows, we can use those features as indicators thatspecific processes took place in past eruptions. We then try to demonstrate thatinflated pahoehoe flows are common in every type of basaltic volcanic terrane.We end by showing that there is mounting evidence that large basaltic eruptionscan cause major atmospheric perturbations and discuss their relationship tofaunal mass extinctions.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



84 SELF ET AL

THE FORMATION OF PAHOEHOE

Qualitative descriptions of the formation of pahoehoe lava flows are well sum-marized in earlier works such as Macdonald (1953) and Swanson (1973). Inthis paper we concentrate on the more recent and more quantitative studies ofthe emplacement of pahoehoe flows via inflation. Many of the ideas we discusswere first introduced by Hon et al (1994), though several papers using theirconcepts were published earlier (e.g. Mattox et al 1993, Keszthelyi & Pieri1993, Chitwood 1994). This section consists of broad generalizations fromobservations of pahoehoe flows from across the globe. Individual flows arediscussed in the next section.

Small-Scale Surface FeaturesPahoehoe flows are most often recognized by their distinctive millimeter- todecimeter-scale surface textures and features. While flowing, pahoehoe flowsare characterized by a smooth continuous flexible skin of cooler lava. In theoutermost millimeters, rapid quenching of the lava preserves stretched bubbles.Typically, 3–5 mm into the rind of a pahoehoe flow, microlites begin to crystal-lize and the bubbles are able to return to a spherical shape before solidification(Hon et al 1994). Ropes and other small-scale features form when the flexibleskin is deformed by the motion of the lava (Fink & Fletcher 1978). The variousforms of pahoehoe indicate different lava flow facies. Shelly pahoehoe is almostexclusively a near-vent phenomenon, and spiny, slabby, toothpaste, and shark-skin pahoehoe are lava types transitional to aa (e.g. Rowland & Walker 1987).Although much qualitative information can be derived from these surface tex-tures, more research is needed before we can extract quantitative constraints oneruption parameters.

LobesWe call the smallest coherent package of lava a lobe. Different types of pahoe-hoe lobes have been distinguished on the basis of vesicle and crystal textures(Figure 1). Walker (1989) used the term S-type pahoehoe to describe lobes thatwere “spongy” with spherical vesicles distributed throughout but with higherconcentrations toward the center. Wilmoth & Walker (1993) used the termP-type pahoehoe to designate lobes containing pipe vesicles. P-type flows typ-ically have a dense interior and more vesicular exteriors, the exact opposite ofS-type lobes. S-type lobes form with minimal inflation, whereas inflated lobesinvariably have the basic characteristics of P-type lobes, even when they lackpipe vesicles.

Another informal terminology in Hawaii uses “silvery” and “blue-glassy” toseparate two major classes of pahoehoe surfaces. The highly reflective surfacesof silvery lobes are caused by numerous stretched vesicles trapping a thin layerof air under a<1-mm–thick glass layer. Silvery lobes often correspond to the

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 85

(a)

Figure 1 Photographs of pahoehoe lobes on Kilauea, Hawaii. (a) Active pahoehoe toes on theKupaianaha flow field in 1991. The 30- to 70-cm–wide lobes in the foreground are only 10–15cm thick, but the surrounding few-minute-old lobes have already inflated to a>30-cm thickness.These lobes would be classified as “silvery” S-type lobes, though continued inflation could convertthem to P-type lobes. This stage of inflation corresponds to Figure 2a. (b) Cross section througha 1996 blue-glassy lobe on the Puu Oo flow field. Note the∼2-cm–thick, dark blue, dense, glassyrind and the large vesicles in the interior. This lobe fits the definitions of both P-type and S-typelobes: There are incipient pipe vesicles at the base, but the vesicles are distributed evenly, withthe largest concentrated in the center of the lobe. Most S-type lobes have a more continuousdistribution of vesicle sizes. (c) Cross section of the lower part of a 1-m–thick inflated P-type lobein the∼500-year-old Kane-nui-o-Hamo flow field. Note the 20-cm–thick vesicular lower crustwith pipe vesicles overlain by the dense core of the flow (cf Figure 3).

vesicular, S-type lobes, but there are also flows with silvery vesicular flow topsand pipe vesicles at their bases.

Blue-glassy lobes (Figure 1b) are characterized by a dense glassy surface thatoften has a gun-metal blue sheen. While the outer 0.5–2 cm of blue-glassy flowsare nearly devoid of vesicles, numerous large (1–10 cm) vesicles occur belowthis dense rind. In some cases, these large bubbles are able to rise through theductile skin and blow delicate glass domes (Oze 1997). Overall, there may belittle difference in the total porosity of blue-glassy flows and the more commonsilvery lobes. Blue-glassy lobes emanate from the base of inflation features andfirst appear after 10–14 days of inflation at a temperature 5◦–10◦C cooler thanadjacent silvery lobes (Keszthelyi 1996). We hypothesize that blue-glassy lavaspends about a week or two inside inflating lobes before breaking out onto thesurface. During this time, the lava cools and the bubbles coalesce. This hypoth-esis is corroborated by a comparison of the petrographic textures of silvery and

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



86 SELF ET AL

(c)

(b)

Figure 1 (Continued)

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 87

blue-glassy pahoehoe (Friedman et al 1996) but differs from the explanationsuggested by Hon et al (1994). Hon and colleagues suggested that the differencesbetween blue-glassy and silvery lobes are caused by the∼3 bars pressure withininflation features forcing the gas back into solution. The recognition of blue-glassy surfaces in ancient terranes is important because it is a strong indicator ofinflation occurring for at least two weeks during the emplacement of that flow.

Typical advancing pahoehoe lobes in Hawaii and elsewhere are 20–50 cmthick, 20–300 cm wide, and 0.5–5 m long. This minimum thickness seems tobe controlled by the stiffness of the bubble-laden lava. Typical S-type or silverylobes are∼50% bubbles, making them stiff foams, not Newtonian fluids. Thebubble-poor part of blue-glassy flows can form lobes as thin as 2 cm.

Meter-scale pahoehoe lobes can coalesce laterally during inflation (Hon et al1994), producing sheets hundreds or even thousands of meters wide. Suchimmense sheets are also meters to tens of meters thick and range in length fromhundreds of meters to perhaps as much as several tens of kilometers (Self et al1996, 1997). The inflation of a lobe is a continuous process that can last days,months, or even years but can be divided into stages (Figure 2).

The duration of active inflation of an ancient inflated lobe can be estimatedfrom the thickness of its vesicular upper crust. During inflation, a flux of fresh,bubble-laden lava is continually brought into the lobe (Figure 2b,c). Afterstagnation, for typical basalt viscosities, the remaining bubbles should rise tothe surface within days to weeks (Aubele et al 1988, McMillan et al 1989,Manga 1996), leaving a dense core (Figures 2d and 3). However, during thecrystallization of this dense core, incompatible elements, including volatiles,concentrate in the residuum. Secondary vesiculation can cause this residuum torise buoyantly as diapirs (Goff 1996, Rogan et al 1996). The residuum is oftenhighly silicic, sometimes reaching rhyolitic compositions, and is more coarselycrystalline than the surrounding lava (Greenough & Dostal 1992, Puffer &Horter 1993). When these rising diapirs hit the base of the upper crust, theyspread to form subhorizontal vesicle sheets (VSs) (Figure 2d ). These sheetsare typically 1–5 cm thick, have knife-sharp boundaries, and have 10-cm scaleirregularities reflecting the uneven topography of the base of the upper crust.The bubbles in these sheets of residuum have relatively uniform size, thoughthey occasionally coalesce to form small diapirs rising into the mushy base ofthe upper crust. These diapirs are preserved as bell-jar– or half-moon–shapedmega-vesicles (Thordarson 1995, Self et al 1997).

Vesicle sheets (VSs) are readily distinguished from the horizontal vesicularzones (VZs) that are often found within the upper crust (Figures 2 and 3). UnlikeVSs, VZs are usually tens of centimeters to many meters thick and have grada-tional contacts with bubble sizes often grading downward toward larger bubblesand then back to small bubbles at the base (Thordarson 1995, Self et al 1997).VZs form because of fluctuations of pressure and/or flux during active inflation

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



88 SELF ET AL

(Hon et al 1994, Cashman & Kauahikaua 1997), whereas VSs form only afterthe flow has stagnated and is no longer inflating. As shown in Figure 3, thedepth between the lowermost VZ and the uppermost VS marks the base ofthe upper crust when inflation ended (Thordarson 1995, Self et al 1997). Theempirical cooling model of Hon et al (1994) permits estimation of the timerequired to grow this crust via the equation

t = 164.8H2, (1)

wheret is the time in hours andH is the thickness of the upper crust in meters.

(b)

(c)

(d)

(a)

CROSS SECTIONSMAP VIEWS

800 °C1070 °C

Britt le Crust Visco-elastic Crust Liquid Lava

RP

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 89

The Hon et al cooling model can marginally overestimate cooling rates andthus underestimate the emplacement duration. Three factors cause excessivecooling: 1. Their lava flow data rely heavily on the depth of cracks, but crackscause locally enhanced cooling; 2. the fit is largely controlled by the data fromthe stagnant Makaopuhi lava lake, but the flux of fresh, hot lava into an inflatingsheet should reduce the rate of crust growth; and 3. all the data come from theEast Rift Zone of Kilauea, which is subjected to as much as 7.5 m of rain per year,something few lava flows elsewhere in the world have to contend with. Thesethree factors probably explain why using the Hon et al (1994) cooling modelunderestimates the inflation duration of the Laki eruption by up to a factor of two(Thordarson & Self 1997a). While this technique can slightly underestimate theemplacement duration of an inflated lobe, currently it is the only quantitativemethod of estimating the duration of active emplacement of ancient lava flows.

The formation of the lower crust is less well understood. Based on simpleconductive cooling models, Hon et al (1994) suggested that the lower crustgrows at two thirds the rate of the upper crust and that the thickness of the liquidcore remains constant during inflation. However, direct measurements of thecooling at the base of pahoehoe lobes show that the cooling is much slower thanpredicted by any simple conductive cooling model (Keszthelyi 1995a). Also,vesicle cylinders form near the base of flows that have thick upper vesicularcrusts. Like horizontal vesicle sheets, vesicle cylinders are formed by diapiricrise of low density residuum (Goff 1996) and should only be preserved after theflow has stagnated (Figure 3). Thus we believe that the lowest vesicle cylinders

←−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−−Figure 2 Cartoon of the development of an inflated pahoehoe sheet flow.On the leftare crosssections at a fixed location, andon the rightare concurrent map views of the development of theflow field. The map location of the cross section is indicated by thedot. This cartoon appliesequally to inflated pahoehoe sheet lobes that range from 1–100 m in thickness and from<100 mto >10 km in lateral dimensions. The emplacement of a sheet lobe can take anywhere betweendays and years. In the cross sections theshadingdarkens with cooling, and in the map view itdarkens with age. (a) A new lobe advances from right to left. Incandescent lava is exposed only atthe very tip of the flow. (b) The lobe thickens by inflation as it extends. Bubbles from the movinglava are trapped in the crust, forming vesicles. The lower crust grows much more slowly thanthe upper crust. (c) Inflation continues. Depressurization from the formation of a new breakoutleads to a pulse of vesiculation within the liquid lava (Hon et al 1994). These bubbles are trappedto form a horizontal vesicular zone. Buoyant, vesicular silicic residuum (R) that rises from thelower crystallization front is disrupted and mixed into the flowing lava. Pipe vesicles (P) grow inthe lower crystallization front. Some cracks in the upper crust become major clefts that reach thevisco-elastic layer. (d ) Flow stagnates. The remaining primary bubbles rise to the top of the flow ina few days to weeks. The vesicular residuum (R) is able to rise through the stagnant lava, forminghorizontal vesicular sheets at the base of the upper crust. Cooling is enhanced around deep clefts.Note the complex planform of a typical compound pahoehoe sheet flow.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



90 SELF ET AL

��

��

��

��

��

UPPER CRUST

CORE

LOWER CRUST

JOINTINGVESICLE STRUCTURE

VESICULARITY CRYSTALLINITY

VS's

VZ

VZ

VZ

VC's

P

Figure 3 Idealized cartoon of the cross section through an inflated pahoehoe lobe. The lobeis divided into three sections on the basis of vesicle structures, jointing, and crystal texture. Theupper crust makes up 40–60% of the lobe and the lower crust is 20–100 cm thick, irrespective of thetotal lobe thickness.Upper crust: Vesicular, often with discrete horizontal vesicular zones (VZs)that form during active inflation (see Figure 2c). Bubble size increases with depth. Prismatic orirregular jointing, sometimes equivalent to the entablature in thick lava flows. Petrographic textureranges from hypohyaline to hypocrystalline (90–10% glass).Core: Very few vesicles. Porosity isdominated by diktytaxitic voids. Vesicles are mostly in the silicic residuum, which forms vesiclecylinders (VCs) and vesicle sheets (VSs). Holocrystalline (<10% glass).Lower crust: Nearly asvesicular as the upper crust, few joints, and 50–90% glass.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 91

record the upper limit of the lower crust at the time inflation stopped (Figure 3).In lobes>10 m thick, our field observations suggest that the lower crust grows5–10 times slower than the upper crust. More detailed thermal models are beingconstructed to test if the lower crust in these large flows reaches an equilibriumbetween crustal growth and thermal erosion.

Flows and Flow FieldsWhile the detailed examination of a single lobe provides some useful informa-tion, one needs to examine entire flows and flow fields in order to understand theproducts of an entire eruption. “Flows” are the product of a single outpouring oflava, but most eruptions that cover a large area with pahoehoe have many sep-arate outpourings and form “flow fields.” Distinguishing lobes, flows, and flowfields is often very difficult in the field, but it is essential if one is determined tounderstand the eruption as a whole. The distinction between lobes and flows canbe blurred in large sheet lobes. Such lobes have sometimes been called “simpleflows,” in contrast to “compound flows” composed of many similar-sized lobes(Walker 1971). However, every “simple” flow we have examined is technicallycompound, though the other lobes within the flow may be much smaller thanthe main sheet or may be juxtaposed laterally rather than vertically.

The morphology of individual flows is controlled by many parameters andis still not understood quantitatively. However, our observations suggest thatrelatively rapid, continuous emplacement over smooth surfaces with shallowslopes favors the production of flows dominated by sheet-like lobes (i.e. “sheetflows”). Conversely, relatively slow, discontinuous emplacement over a roughsurface on steeper slopes favors the production of “hummocky” flows with manydiscrete tumuli (Figures 4 and 5). During inflation, the surface topography oftenbecomes inverted; high areas in the underlying topography become inflation pitswhile low areas become tumuli (Figure 6). Distinguishing inflation pits fromcollapse pits or skylights can be challenging but is important in decipheringthe emplacement history of a flow. This is most reliably done by searching forhorizontal clefts that form during inflation (Figure 6) but not during collapse.

The distribution and morphology of the pathways feeding molten lava througha sheet flow and a hummocky flow must differ significantly (Figure 7). A sheetflow by definition has a continuous body of liquid lava across the entire widthof the flow. In contrast, the pathways within a hummocky flow should havemany dead ends (ending in tumuli) and areas in which the lava cannot flow(inflation pits). As the flow cools and solidifies with time, discrete lava tubesform in a hummocky flow. It is common for sheet flows to stagnate long beforecooling leads to the formation of cylindrical lava tubes. However, if a sheetflow remains active for a long time (i.e. long enough for parts of the sheet tofreeze shut), irregularities in the thickness of the crust will lead to concentration

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



92 SELF ET AL

(a)

Figure 4 Photographs of hummocky and sheet flows. (a) Large tumulus that formed in 1995within a hummocky portion of the Puu Oo flow field. Note people for scale. The flow surfacewas originally a flat sheet, but later inflation was localized to smaller areas as the fluid core of theflow froze shut. (b) Compound sheet flows within the 1969–1974 Mauna Ulu flow field. Despitetopographic roughness at the 10-cm to 1-m scale and a 2◦–3◦ slope, the surface of the flow field isremarkably flat. Note patches of trees surrounded by lava in front of the Mauna Ulu shield on theskyline. These trees are on kipukas (islands) of older lava that were high points in the pre-eruptiontopography but are now low areas because of inflation (See Figure 6). (c) Large compound pahoehoesheets in the Columbia River Basalts Group (CRBG) at Palouse Falls, Washington. Waterfall isabout 50 m tall and cuts through the uppermost flows of the Grande Ronde Formation. While notvisible at this scale, these large sheets have 10-cm– to 1-m–scale surface features identical to thosein b, and similar sheets have identical internal structures as Hawaiian sheets (see Figure 8).

of the flow into narrow pathways. This can lead either to the production of asingle long tumulus over a lava tube or to a hummocky flow, depending on howirregular the cooling is. Thus larger, thicker pahoehoe flows are more likely tobe sheet-like, whereas thinner flows are more likely to become hummocky. Thisis illustrated by the prevalence of thick sheet lobes in flood basalt provinces,while Kilauea has many hummocky pahoehoe flows.

In a simplistic model of basaltic volcanism, each eruption should feed a singleflow. However, long-lived eruptions are usually broken into distinct episodesby pauses and migrations of the vent (e.g. Heliker & Wright 1991, Thorarsdon& Self 1993). Even if effusion resumes at the same vent as before, the lavatransport system can collapse during pauses and the new lava can be forced tobuild a new flow (Mattox et al 1993). The new pahoehoe flow front will onlybe 20–50 cm thick and can be initially diverted around earlier inflated flows

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 93

(c)

(b)

Figure 4 (Continued)

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



94 SELF ET AL

(+t ime)

(+t ime)

Loweref fusionrate

Moderate effusion rateHigher effusion

rate

"Hummocky" Flow

"Sheet" Flow

(a)

(b)

(c)(d)

(e)

( f )

Figure 5 Cartoon cross sections of the formation of hummocky and sheet flows. The morphologyof the flow is controlled by a number of parameters in addition to effusion rate. More continuouseffusion, lower slopes, and smoother topography have the same effect as increasing effusion rate.(a) The initial pahoehoe lobe arrives. (b) If effusion rates are very low, the lobe will freeze quicklyand the eruption forms a stack of small lobes. (c) At somewhat higher effusion rates, inflation andcrust growth proceed at similar rates. Inflation is confined to discrete tumuli, and the flow advancesvia smaller breakouts. (d ) At higher effusion rates the flow is able to inflate evenly, forming a flatsheet. However, the sheet may become hummocky if the effusion rate decreases later. (e) Withtime, the hummocky flow builds up a field of tumuli and breakouts of various ages. The path thelava must travel to the flow front is very tortuous. (f ) Given sustained high effusion rates, the entiresheet remains an open pathway for the fluid lava. Horizontal inflation clefts or large monoclinalfolds can be found on the margins of thick sheets.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 95

(a)

(b)

(c)

(d)

(a)

(b)

(c)

(d)

Figure 6 Cartoon cross sections of the formation of tumuli and inflation pits and the inversion oftopography during inflation. (a) The initial, small pahoehoe lobes are confined to the lowest areas.(b, c) As the flows continue to advance, they also inflate and spread laterally, filling the depressions.Because the thickest crust is on the oldest part of the sheet, the highest inflation is over what wasoriginally the lowest ground. Breakouts from the thick inflation features can eventually cover theearlier high ground. (d ) Depending on the rate of inflation relative to the rate of crust growth, eithera continuous sheet or an inflation pit forms over the previous high point(s). Most inflation pitswiden with depth as the lobe cools inward with time. Other inflation pits have steep monoclinalfolds like those at the margins of sheet lobes.

(Figure 7). For the largest pahoehoe flows, there is time for some erosion orsediment deposition to occur before a new flow is able to overtop the older lava.These processes, and chemical changes in the erupted lava during a long-livederuption, can make mapping out individual lobes, flows, and/or flow fields verydifficult. It is a remarkable testimony to decades of work that this has beenaccomplished in at least parts of the Columbia River Basalts Group (CRBG)(Tolan et al 1989; Mangan et al 1986 and references therein).

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



96 SELF ET AL

Figure 7 Cartoon map view of a compound inflated pahoehoe flow field. Flow fields are builtup by a number of separate flows, sometimes from distinct vents. The lava is transported throughan anastomosing lava tube system to the hummocky flow while essentially the entire width of asheet flow is able to transport lava. Earlier flows often act as barriers to later flows. The featuresshown in this cartoon are distilled from our observations of flow fields ranging in size from a fewkilometers up to a few hundred kilometers in length.

The vents for the pahoehoe flows we have seen are rather unremarkableconstructs. This is largely because the lava transport system from the vent tothe flow front is very efficient and little lava accumulates near the vent. Thevent deposits associated with pahoehoe flows can be generally divided intothose formed during short-lived fissures and those formed from longer-livedlava ponds. Most basaltic eruptions begin with the opening of a fissure andfountaining along its length, which builds up spatter ramparts and scoria fields.Long fissures are thermally inefficient and historically have broken down intopoint sources in a matter of hours to days. Scoria and spatter cones and lavaponds commonly form over these point sources. If the eruption continues, anefficient connection between the subsurface plumbing and the lava transportsystem forms, and lava can be erupted without ever being exposed directly tothe surface. Long-lived pahoehoe flows are typically fed from lava ponds onbroad shields built up by pond overflows and short flows.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 97

During eruptions that last months or years, it is also quite common for thevent to migrate by opening up new fissures and building new ponds. Attemptingto decipher whether a series of vents is the product of a single eruption, a multi-episode eruption, or the products of totally different eruptions can be verydifficult.

EXAMPLES

In this section we present the evidence that inflated pahoehoe flows, formed inthe manner described in the previous section, are ubiquitous and constitute thelargest lava flows on Earth.

HawaiiThe mid-plate hot spot volcanism in Hawaii is the birthplace of the term pa-hoehoe. It covers 86% of Kilauea Volcano (Holcomb 1987). Mauna Loa, withits steeper slopes and higher average eruption rates, has roughly equal amountsof pahoehoe and aa (Lockwood & Lipman 1987). The current eruption ofKilauea has produced mostly pahoehoe for the last 11 years from two separatevents: Kupaianaha from 1986–1992 and Pu’u’O’o from 1992 until this writ-ing (Heliker & Wright 1991, Heliker et al 1997). Such long-lived rift-zoneeruptions may not be exceptional: Mauna Ulu produced mostly pahoehoe from1969–1974 (Tilling et al 1987), and the Kane-nui-o-Hamo shield may have fedpahoehoe flows for 50 years (JP Lockwood, personal communication). Oneof the longest (in terms of both length and duration) historical lava flows onMauna Loa is the pahoehoe phase of the 51-km- and∼300-day-long 1859 flow(Rowland & Walker 1990). As noted before, many pahoehoe flow fields onHawaii are hummocky, with sheet flows largely confined to the very shallowslopes (<1◦) of the coastal flats (Hon et al 1994).

IcelandPahoehoe flows are also common at the confluence of hot spot and mid-oceanridge volcanism in Iceland. Many examples of excellent hummocky flowswith fields of tumuli occur (e.g. Rossi & Gudmundsson 1996) as well as largerinflation ridges (also called “festoon ridges”) (Theilig & Greeley 1986). Icelandpossesses several large recent flow fields such as the 65-km-long 1783–1784Laki flow field (Thordarson & Self 1993), the>70-km-long 934 AD Eldgjaflows (Miller et al 1996), and the>140-km-long, 8500-year-old Thjorsa flows(Hjartarson 1988). These three flow fields each have volumes of 15–20 km3

and are dominantly pahoehoe, with significant fractions of slabby pahoehoe andaa (∼30% in the case of Laki). The proportion of aa diminishes in the longerflows. The internal structures within lobes of these flow fields are identical tothose of other inflated pahoehoe lobes (Figure 7).

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



98 SELF ET AL

While details of the emplacement of the Eldja and Thjorsa flow fields arenot recorded, the Laki eruption was well documented (Thordarson & Self 1993and references therein). The eruption lasted from the 8th of June, 1783, untilFebruary 7th, 1784; however, 90% of the lava volume was erupted in the first5 months, and peak eruption rates reached 4000 m3/s. The 27-km-long ventsystem consisted of 10 en-echelon fissure segments ranging in length from1.6–5.1 km. For a few days, parts of 3 fissure segments, with a total lengthof ∼10 km, were active simultaneously, but the segments were mostly activeseparately. The Eldja flow is also thought to have erupted over about a year,but its fissure system was 75 km long.

Western United StatesPahoehoe flows are also common in areas of crustal extension. The Basin andRange province is littered with small basaltic lava flows, most of which areinflated hummocky pahoehoe flows fed from scoria cones (e.g. the Amboy andPisgah flow fields in the Mojave Desert, California). Basaltic flow fields withabundant inflated pahoehoe can be found across the Colorado Plateau (e.g. theMcCarthy’s flow field, New Mexico). The Taos volcanic field and the Potrillomaar field (which includes the famous Kilburn Hole) in the Rio Grande Riftalso contain many pahoehoe sheet flows. The longest preserved Quaternaryflow field in the United States is the 75-km-long,∼2000-year-old Carrizozoflow field, which is also within the Rio Grande Rift. It is exclusively pahoehoe,with some patches of slabby pahoehoe. The flow is a cross between hummockyand sheet flows; the surface is covered with flat-topped tumuli many meters talland hundreds of meters in lateral dimensions, punctured by inflation pits tensof meters in diameter. While no drained lava tube has been found, a series oflong tumuli appear to mark the tube’s location for part of the flow field’s length.Initial estimates of the emplacement duration, based on the pahoehoe nature ofthe lava surface, are on the order of 20 years (Keszthelyi & Pieri 1993).

Classic inflation features are also present throughout the Snake River Plainbasalts of Idaho. This area is the type example of plains volcanism, a style ofvolcanism intermediate between shield volcanoes and flood basalts (Greeley &King 1977). Lava tubes, tumuli, inflation ridges, and inflation pits are exhibitedin the Craters of the Moon National Park.

AustraliaThe Cenozoic volcanic region of Eastern Australia is dotted with large vol-canic provinces, each of which is dominated by pahoehoe flow fields. The160-km-long,∼200,000-year-old Undara flow of the McBride province and the120-km-long,∼15,000-year-old Toomba flow of the Nulla province in NorthQueensland challenge Iceland for the title of the longest Holocene lava flows

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 99

on Earth (Stephenson & Griffin 1976). These flows average 10–15 m thickand have volumes of 10–30 km3. Their surfaces are dominantly pahoehoe,with minor amounts of slabby pahoehoe and no reported patches of aa. Theinitial flow surfaces have been inflated, forming spectacular tumuli, inflationridges, inflation pits, and sheet flows (Stephenson et al 1996). The lava tubes inthe Undara flow field in particular are world renowned (Atkinson & Atkinson1995). These flow fields are fed from small (∼100 m tall) scoria cones. Similarflows exist throughout the Victoria basaltic province in southeastern Australia,with inflated, compound tube-fed pahoehoe flows up to 60 km long (Joyce &Sutalo 1996).

Continental Flood BasaltsThe largest terrestrial lava flows are within continental flood basalt (CFB)provinces. The 17.5- to 6.0-Ma,∼174,000-km3 Columbia River Basalts Group(CRBG) are the most recent and best studied of the flood basalt provinces (e.g.Waters 1961, Hooper 1982, Mangan et al 1986, Tolan et al 1989, and refer-ences therein). While the emplacement of the CRBG took over 11 millionyears,∼85% of its volume (the Grande Ronde Formation) was erupted in 1 Maor less (Tolan et al 1989) and∼50% of the total volume was erupted in about300,000 years around 15.8 Ma (Baksi 1989). The province as a whole is builtup of about 300 geochemically separate lava flows, most ranging in volumefrom a few hundred to 3,000 km3, with lengths up to 600 km (Tolan et al 1989).

Pahoehoe surfaces are common throughout the CRBG, whereas aa flows aremostly confined to what is thought to be the near-vent regions of the province(Swanson & Wright 1980). Inflation features in the CRBG range from tumuliof the same scale as those common in Hawaii up to sheet lobes many tens ofmeters thick and kilometers in lateral dimensions (Thordarson 1995, Self et al1996, 1997). No cylindrical lava tubes have been found in the CRBG, but thisis to be expected because on such shallow slopes the tubes are sheet-like anddo not drain. Although the scale of many CRBG sheet lobes is huge whencompared to Hawaiian examples, the similarity of the internal structures andsurface features clearly indicates that the same process, inflation, created both(Figure 7). The CRBG also exhibits examples of topographic inversion andinflation pits. Some of the large inflation pits have been mistaken for spiracles(features formed by steam from heated groundwater blasting through a flow).

While ample evidence exists that most CRBG lava flows were emplaced asinflated pahoehoe sheet flows, few outcrops look exactly like the cartoon inFigure 3. In many areas, flow lobes interleave, which greatly complicates thepicture. Some large flows have slabby pahoehoe to rubbly flow tops, and afew thick aa flows occur far from any known vents. Pillow basalts and relatedhyaloclastites are also common throughout the province.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



100 SELF ET AL

Many dozens of separate outcrops must be carefully examined before confi-dent conclusions can be drawn about any eruption. To date, this has been doneadequately for only the∼14.7-Ma, 1300-km3 Roza Member of the CRBG, oneof the best-exposed flow fields of the entire province (e.g. Swanson et al 1975,Martin 1989, Thordarson 1995, Thordarson & Self 1996). An exhaustive ex-amination of the available outcrops showed that the flow field consists of fiveseparate flows (Martin 1989) and is completely dominated by inflated pahoehoeflows, which are hummocky on a scale of a few tens of meters in some areasand in other areas are covered with sheet lobes that are kilometers in lateralextent (Thordarson 1995, Self et al 1996, 1997).

The duration of active inflation of each lobe at each outcrop can be estimatedfrom the thickness of its upper vesicular crust. While it is possible that morethan one of the five distinct flows within the Roza flow field were active at anygiven time, to us it seems implausible for two thick sheet flows to be inflatingsimultaneously at one location. Thus the durations at any one outcrop can besummed, assuming that no time passed between the stagnation of one lobe andthe initial emplacement of the second. For example, an outcrop at SummerFalls, Washington, has three Roza sheet lobes that we estimate to have beenactive for 6.8, 11, and>9 months (oldest to youngest). Thus this one outcropshows that the Roza eruption fed this location for well over 2 years. An outcropat Horton Grade, Washington, preserves evidence for 6.4 years of activity duringthe emplacement of two of the five Roza flows. Thus we conservatively estimatethat the entire Roza flow field took on the order of a decade to be emplaced(Thordarson 1995, Self et al 1996, 1997).

Dividing the total volume of the Roza Member by 10 years leads to anaverage volumetric effusion rate of∼4000 m3/s, which is similar to the peakLaki eruption rate. Although we believe that the Roza eruption lasted abouta decade, it is important to note that individual packages of lava do not takethat long to reach the flow front. Calculated flow velocities in the Roza sheetsare approximately 0.2–0.4 m/s, allowing lava to travel 300 km from the ventin less than 17 days (Keszthelyi & Self 1997). Some previous workers haveconfused the time for lava to travel from the vent to the distal end of the flowwith the time required to emplace the entire flow field. The compound natureof pahoehoe flow fields and the inflation process confound such simplisticestimates of eruption duration from flow velocity.

The detailed stratigraphy and mapping of other flood basalt provinces isincomplete. However, the limited data do suggest that they are dominated byinflated pahoehoe sheet flows. The Deccan Traps in India are largely pahoehoewith minor amounts of aa; small pahoehoe lobes, ropy surfaces, tumuli, andinflated sheets are common (Agashe & Gupte 1971, Phadke & Sukhtankar1971, Walker 1971, Kale & Kulkarni 1992, Keszthelyi et al 1997). Even less is

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 101

Figure 8 Cross sections of inflated pahoehoe flows from Hawaii, Iceland, and the Columbia RiverBasalts. Despite the differences in scale, the similarities are striking. The features in these flowsare diagnostic of inflated pahoehoe flows (compare to Figures 2 and 3).

known about the morphology of the lavas in more ancient flood basalt provinces,but the Parana, Karoo, and Keweenawan provinces also exhibit features ofclassic inflated pahoehoe flows (A Duncan, personal communication, 1995;M Rampino, personal communication, 1997). (See Figure 8.)

The vast tracts of Archean komatiite flows might also be considered flood“basalts.” Komatiites were originally thought to typify rapidly emplaced tur-bulent lava flows (e.g. Huppert et al 1984), but more recent studies suggest thatthey may have been emplaced as inflated tube-fed pahoehoe sheet flows thatreached thicknesses up to 500 m (Hill & Perring 1997). Identifying inflationfeatures is difficult in these ancient altered and deformed terranes, so much hasto be inferred from the petrographic textures within the flows.

Submarine LavasThe ocean floor is mostly covered with pillow basalts, which are closely relatedto pahoehoe flows in both their morphology and style of emplacement. Detailedknowledge of seafloor lava flows is in its infancy, but evidence is steadilyemerging for the same inflation process common in subaerial pahoehoe actingin the submarine environment (e.g. Ballard et al 1979, Applegate & Embley1992, Gregg & Chadwick 1996). Most documented submarine eruptions formsmall flow fields that extend only a few kilometers from the vents, but a fewlarger flow fields are known (Davies 1982, Macdonald et al 1989). Perhaps

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



102 SELF ET AL

the largest submarine flow field known is on the Hawaiian arch where∼25,000km2 is covered with thick, sheet-like flows reaching 100 km to either side ofthe likely vent areas (Clague et al 1990).

Other LocalitiesMount Etna is dominantly aa but has several large pahoehoe flow fields. Infact, the longest lived historic eruption of Mount Etna (the 1614–1624 eruption)produced 36 km2 of tube-fed hummocky pahoehoe. Some of the tumuli on thisflow field are about a kilometer in diameter, large enough to be named “mounts”of their own (Chester et al 1985).

Pahoehoe flows are also common on composite volcanoes such as Mount St.Helens, which contains the well-visited Ape Cave lava tube within an inflatedpahoehoe flow. Other inflated pahoehoe flows in the Cascades were describedby Chitwood (1994).

EFFECTS OF LARGE VOLUMEPAHOEHOE ERUPTIONS

Atmospheric Impact of Pahoehoe Lava EruptionsOne of the major reasons large basaltic eruptions are worthy of study is theirimpact on our atmosphere. Among common magmas, basalts have the greatestpotential for dissolved sulfur concentration (Haughton et al 1974) and emissionof SO2 or H2S during eruption (Thordarson et al 1996). Atmospheric emissionshave been estimated by various methods for two historic pahoehoe lava erup-tions. The first is the ongoing Kilauea eruption, presently into its fourteenthyear of quasi-continuous emission, on average emitting∼2000 tons of SO2 perday (Elias et al 1993). This SO2 forms sulfate aerosols in the lower 2–5 km ofthe troposphere, and an aerosol plume is blown around the island of Hawaii,occasionally reaching other islands in the Hawaiian chain. This “vog” (vol-canic fog) causes significant eye and respiratory irritation in the population ofthe state of Hawaii but does not have more far-reaching effects.

The other example is the 1783 Laki eruption, which emitted as much as1.7 megatons (Mt) of SO2 per day in the early weeks of this eight-month-long eruption (Thordarson et al 1996). For each of its first three months, thisspectacular eruption released as much SO2 as in the entire 1991 Mount Pinatuboeruption, the second largest eruption of the twentieth century. About 70% ofthe amount of degassed volatiles was released at the Laki vents in fire fountains;the other 30% issued from the lava during flowage. This led to two acid aerosolclouds, one near the surface and the other at 5- to 10-km altitudes. Very largeamounts of HCl and HF were also generated by the eruption plume, the lattercausing the most damage to livestock and triggering the ensuing deadly famine

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 103

in Iceland. The Laki aerosols created a thick haze (or dry fog) that persistedover Europe and into Asia for about three to four months (Stothers 1996).

The Laki eruption column could reach the stratosphere and affect widespreadareas because the tropopause over Iceland is low (∼10 km) and the eruption col-umn exceeded this height during much of the first few months of activity. Witha total SO2 release of>120 Mt, Laki generated an estimated stratosphericaerosol load of 250 Mt, resulting in a short-lived total column optical depth(at 0.55 microns) in the high latitudes of the northern hemisphere of up to 2.0(Stothers 1996), compared to a maximum of 0.2 for Mount Pinatubo aerosols(McCormick et al 1995). Sulfate aerosols remaining in the stratosphere perhapsled to the two years of abnormally low temperatures and poor weather in theNorthern Hemisphere (e.g. Wood 1992, Fiacco et al 1994, Thordarson & Self1997b).

Considering the effect of the Laki eruption, what effect could flood basalteruptions have had on the atmosphere and climate? The Roza flow of theColumbia River Basalts Group (CRBG), with its well-preserved glassy lavaand dike selvages, was our first target for detailed study. Samples were col-lected from the dike, near vent pyroclastics, and various distances along thelava flow field. Using standard petrologic techniques to estimate the pre- andpost-eruption volatile (S, Cl, F) contents of the glass in inclusions trapped inphenocrysts and the degassed glassy lava matrix (Thordarson & Self 1996 andreferences therein), we estimate that, similar to Laki,∼77% of the volatilerelease of the Roza eruption took place in fire fountains at the vents and theremaining∼23% escaped from the flowing lava. This should have formed atwo-level aerosol cloud. The estimated total release of>12,000 Mt SO2 (andsmaller amounts of HCl and HF) is stupendous. However, the atmospheric andclimatic impact also depends on the duration, altitude, and variability of thisSO2 release. If the Roza eruption lasted for about 10 years, the annual H2SO4aerosol release would have been on the order of 900 Mt, roughly four times thetotal Laki release. One can only conjecture at the environmental devastationcaused by the acid haze and rain and at other climatic effects of aerosols fromsuch an immense, sustained volatile release.

Relationship Between Flood Basalts and Mass Extinctions?While there now seems to be incontrovertible evidence for a large 64.5-Maimpact with devastating effects on global biota at Chixulub at the Cretaceous-Tertiary (K/T) boundary (e.g. Sharpton et al 1996, Pierazzo 1997, Toon et al1997, and references therein), there is also an apparent coincidence between thegeological ages of mass extinctions and those of the formation of flood basaltprovinces (Rampino & Stothers 1988, Rampino & Caldeira 1992, Stothers1993, Courtillot 1994, Rampino & Haggerty 1997). Apparently, for at least the

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



104 SELF ET AL

past 300 Ma, long periods of geological time have passed without the occur-rence of either phenomenon and then, simultaneously to within the resolution ofdating, flood basalt volcanism flared up and a major extinction occurred. Whilethere are both extinctions unaccompanied by flood basalt episodes and viceversa in the past 300 Ma, the fact that at least 60% of the major extinctions isreported to have synchronous continental flood basalt (CFB) eruptions cannotbe ignored.

The Deccan Traps flood basalt province contains some 2× 106km3of basalticlava and is one of the great large igneous provinces (LIPs; Coffin & Eldholm1994). The consensus of age dating shows that the Deccan lavas were eruptedbetween 68 and 65 Ma with peak volume production lasting possibly<1 Ma,somewhere between 67 and 65 Ma (Vandamme et al 1991, Venkatesan et al1993, Baksi 1994), and that the Chixulub impact occurred near the end ofDeccan volcanism (McWilliams et al 1992). This raises the unresolved ques-tions of when species died off relative to the ages of the Deccan lava flows andthe Chixulub impact (Jablonski 1997) and whether flood basalt volcanism canbe linked with mass extinctions.

We can try to address the latter question. The only plausible mechanism bywhich flood basalt eruptions can cause widespread, deleterious impact on biotais by release of volatiles into the atmosphere. It seems reasonable to assume thatmany flood basalt eruptions could have degassed similar amounts of noxiousvolatiles as the Roza eruption. The climatic effects of maintaining elevatedlevels of sulfate aerosols in the middle troposphere and lower stratosphere fora decade or more are currently not quantified (Bekki et al 1996) but must havebeen severe.

Even if the effects were global and severe, it is unclear whether flood basaltvolcanism alone could cause mass extinctions. Certainly, individual flood lavaevents lasting even decades are unlikely to have had such an extreme effect,but, as typified by the CRBG, flood basalt provinces are composed of hundredsof such eruptions. However, even during the period of peak output, individ-ual eruptions would have average recurrence intervals of 5,000–10,000 years,which may have given the environment sufficient time to recover between erup-tions. The most reasonable statement, given our current knowledge, is that aCFB eruption probably could not cause a mass extinction alone, but a series ofthem during the development of a whole CFB province would have been ableto stress the environment to such an extent that any other major perturbationwould have had a more extreme effect.

Also noteworthy of consideration are the possible effects of effusion of hugevolumes of lava under the oceans, such as in the Ontong Java plateau, and theirpotential effect on the marine environment (Coffin & Eldholm 1994). Beforewe can go any further with such speculations on this important and tantalizingtopic, we need to know the following about major flood basalt eruptions and

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 105

provinces: (a) the amounts of volatile species liberated in individual eruptionsand (b) the style and duration of effusion. If the flows are inflated pahoehoelobes, we should be able to estimate the latter. Obtaining data for the formerwill be difficult owing to poor preservation of the glassy portions of the lava inthe older, larger flood basalt provinces.

THE SWELL HYPOTHESIS

Long, large volume basaltic lava flows share a remarkable commonality irre-spective of geography, geologic setting, and age: They are all pahoehoe sheetflows. This leads us to propose the SWELL hypothesis, which states that theStandard Way of Emplacing Large Lavas is as inflated, compound pahoehoesheet flows. While we anticipate evidence for large volume flows that do notfit the SWELL hypothesis, the available data from Earth are entirely consistentwith this suggestion.

The identification of pahoehoe sheets in most ancient long lava flows wouldnot be possible without recent advances that allow identification from partialoutcrops. The recognition that a given lava lobe consists of a vesicular uppercrust with a moderately orderly distribution of spherical vesicles, dense core,and a thin vesicular basal crust is sufficient to confidently identify the lava asan inflated pahoehoe lobe. The reasoning is so simple that it is sometimes metwith skepticism: The presence of an upper zone with spherical vesicles and theabsence of rubbly flow tops and/or flow bases indicate that the flow is pahoehoe,not aa (Macdonald 1953). The formation of a dense core (i.e. a P-type lobe asopposed to an S-type lobe) is clear evidence for inflation. Thus the lobe mustbe inflated pahoehoe, emplaced in the manner described above.

One obvious complication is the need to distinguish rapidly emplaced pondedflows from inflated flows. This can be done by comparing the thickness of theupper vesicular crust to the total thickness of the flow. For inflated pahoehoeflows, the upper vesicular crust almost invariably comprises 40–60% of the totalflow thickness (Self et al 1997). In contrast, the upper vesicular crust of lavalakes, which is analogous to that of ponded flows, is much thinner:<30% ofthe 14-m–thick Alae lava lake (Peck 1978),<6% of the 83-m–thick Makaopuhilava lake (Wright & Okamura 1977), and<8% of the 140-m–thick Kilauea Ikilava lake (Mangan & Helz 1986). This is because it takes only a few weeksfor the bubbles to rise to the top of a stagnant lava lake or ponded flow (e.g.McMillan et al 1989), and the upper crust of such a ponded flow will havegrown to no more than a few meters thick at that point. Thus, in the case oflarge sheet lobes (which are typically 15–30 m thick), a vesicular upper crustthicker than 5–6 m is a strong indicator that the lobe was inflated, not ponded.

Inflated pahoehoe flows can attain their great size because they are extremelythermally efficient. The thick, stationary crust on top of inflated pahoehoe flows

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



106 SELF ET AL

is a remarkable insulator. Active Kilauea lava tubes cool roughly 1◦C/km(Peterson et al 1994, Cashman et al 1994), but Kilauea is an unfavorable envi-ronment for thermally efficient lava tubes (Keszthelyi 1995b). Temperatures de-rived by glass geothermometry allow estimates of cooling rates down the lengthof an emplaced flow. The cooling rates along large lava flows are remarkablylow: about 0.5◦C/km for Laki (Thordarson & Self 1996), 0.04◦–0.05◦C/km forthe Roza flow field (Self et al 1997), and∼0.04◦C/km in the Ginkgo Flow of theFrenchman Springs Formation of the CRBG (Ho & Cashman 1997). This de-gree of thermal efficiency is predicted for lava tubes or sheet flows in the CRBGand other large pahoehoe flows (Keszthelyi 1995b, Keszthelyi & Self 1997).

SUMMARY

1. Pahoehoe is everywhere that there is basalt: on land, on the seafloor, and onthe surfaces of other terrestrial planets.

2. The largest well-documented outpourings of lava on the Earth are pahoehoeflow fields.

3. We hypothesize that all large lava flows are formed as inflated pahoehoesheets—i.e. the SWELL hypothesis.

4. The eruption of large pahoehoe lava flows can have far-reaching conse-quences to the global environment. This might provide a link betweencontinental flood basalt (CFB) eruptions and mass extinctions.

ACKNOWLEDGMENTS

We would like to thank our colleagues at the University of Hawaii, the USGeological Survey Hawaiian Volcano Observatory, and the Jet PropulsionLaboratory for sharing their many invaluable observations, discussions, andcomments. The authors’ research was funded by several grants from the NationalScience Foundation and the National Aeronautics and Space Administration.

Visit the Annual Reviews home pageathttp://www.AnnualReviews.org.

Literature Cited

Agashe LV, Gupte RB. 1971. Mode of eruptionof the Deccan Traps basalts.Bull. Volcanol.35:591–601

Applegate B, Embley RW. 1992. Submarine tu-muli and inflated tube-fed lava flows on AxialVolcano, Juan de Fuca Ridge.Bull. Volcanol.54:477–58

Atkinson A, Atkinson V. 1995.Undara Vol-cano and Its Lava Tubes.Brisbane: Vernon& Anne Atkinson. 86 pp.

Aubele JC, Crumpler LS, Elston WE. 1988.Vesicle zonation and vertical structure ofbasalt flows.J. Volcanol. Geotherm. Res.35:349–74

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 107

Baksi AK. 1989. Reevaluation of the timing andduration of extrusion of the Imnaha, PictureGorge, and Grande Ronde Basalts, ColumbiaRiver Basalt Group. See Reidel & Hooper1989, pp. 105–12

Baksi AK. 1994. Geochronological studies onwhole-rock basalts, Deccan Traps, India:evaluation of the timing of volcanism rela-tive to the K-T boundary.Earth Planet. Sci.Lett.121:43–56

Ballard RD, Holcomb RT, Andel TH. 1979. TheGalapagos Rift at 86◦W: sheet flows, collapsepits, and lava lakes of the rift valley.J. Geo-phys. Res.84:5407–22

Bekki S, Pyle JA, Zhong W, Toumi R, HaighJD, Pyle DM. 1996. The role of microphysi-cal and chemical processes in prolonging theclimate forcing of the Toba eruption.Geo-phys. Res. Lett.23:2669–72

Bruno BC, Taylor GJ, Rowland SK, Lucey PG,Self S. 1992. Lava flows are fractals.Geo-phys. Res. Lett.19:305–8

Campbell BA, Campbell DB. 1992. Analysis ofvolcanic surface morphology on Venus fromcomparison of Arecibo, Magellan and ter-restrial airborne radar data.J. Geophys. Res.97:16293–314

Cashman KV, Kauahikaua JP. 1997. Reevalu-ation of vesicle distributions in basaltic lavaflows.Geology25:419–22

Chester DK, Duncan AM, Guest JE, KilburnCRJ. 1985.Mount Etna: The Anatomy of aVolcano.Stanford: Stanford Univ. Press. 404pp.

Chitwood LA. 1994. Inflated basaltic lava: ex-amples of processes and landforms from cen-tral and southeast Oregon.Or. Geol.56:11–21

Clague DA, Holcomb RT, Sinton SM, DetrickRS, Torresan ME. 1990.Earth Planet. Sci.Lett.98:175–91

Coffin MF, Eldholm OL. 1994. Large igneousprovinces: crustal structure, dimensions, andexternal consequences.Rev. Geophys.32:1–36

Courtillot VE. 1994. Mass extinctions in thelast 300 million years: one impact and sevenflood basalts? Isr. J. Earth Sci.43:259–66

Crisp J, Baloga SM. 1994. Influence of crystal-lization and entrainment of cooler materialon the emplacement of basaltic aa flows.J.Geophys. Res.99:11819–31

Davis EE. 1982. Evidence for extensive basaltflows on the sea floor.Geol. Soc. Am. Bull.93:1023–29

Decker RW, Wright TL, Stauffer PH, eds. 1987.Volcanism in Hawaii. US Geol. Surv. Prof.Pap. 1350.1667 pp.

Elias T, Sutton AJ, Stokes JB. 1993. CurrentSO2 emissions at Kilauea Volcano: quantify-

ing scattered degassing sources.Eos74:670–71

Fiacco RJ, Thordarson Th, Germani MS, Self S,Palais JM, et al. 1994. Atmospheric aerosolloading and transport interpreted from ashparticles and acidity due to the 1783–1784Laki eruption in the GISP2 ice core.Quat.Res.42:231–40

Fink JH, Fletcher RC. 1978. Ropy pahoehoe:surface folding of a viscous fluid.J. Volcanol.Geotherm. Res.4:151–70

Friedman RC, Taylor GJ, Keszthelyi L, Thorn-ber CR. 1996. Petrographic textures as in-dicators of Kilauea lava flow emplacementhistory.Eos77:F807 (Abstr.)

Goff F. 1996. Vesicle cylinders in vapor-differentiated basalt flows.J. Volcanol.Geotherm. Res.97:19729–37

Greeley R, King JS, eds. 1977.Volcanism ofthe Eastern Snake River Plain, Idaho: AComparative Planetary Geology Guidebook.Washington, DC: NASA. 308 pp.

Greenough JD, Dostal J. 1992. Cooling historyand differentiation of a thick North Moun-tain Basalt flow (Nova Scotia, Canada).Bull.Volcanol.55:63–73

Gregg TKP, Chadwick WW. 1996. Inflated sub-marine lava flows: a model for the formationof lava pillars.Geology24:981–84

Haughton DR, Roeder PL, Kinner BJ. 1974.Solubility of sulfur in mafic magmas.Econ.Geol.69:451–67

Heiker CC, Mangan MT, Mattox TN,Kauahikaua JO, Helz RT. 1997. Episodes50-53 of the Pu’u’O’o-Kuapaianaha eruptionof Kilauea Volcano, February 1992-February1994.Bull. Volcanol.In press

Heliker CC, Wright TL. 1991. The Pu’u ‘O’o-Kupaianaha eruption of Kilauea.Eos72:521,526, 530

Hill RET, Perring CS. 1997. The evolution ofArchean komatiite flow fields: Are they in-flationary sheet flows?J. Geophys. Res.Inreview

Hjartarson A. 1988. The great Thjorsa lava:Earth’s largest Holocene lava flow.Nat-turufraedingurinn58:1–16

Ho AM, Cashman KV. 1997. Temperature con-straints on the Ginkgo flow of the ColumbiaRiver Basalt Group.Geology25:403–6

Holcomb RT. 1987. Eruptive history and long-term behavior of Kilauea Volcano. SeeDecker et al 1987, pp. 261–350

Hon K, Kauahikaua J, Denlinger R, Mackay K.1994. Emplacement and inflation of pahoe-hoe sheet flows: observations and measure-ments of active lava flows on Kilauea Vol-cano, Hawaii.Geol. Soc. Am. Bull.106:351–70

Hooper P. 1982. The Columbia River Basalts.Science215:1463–68

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



108 SELF ET AL

Huppert HE, Sparks RSJ, Turner JS, Arndt NT.1984. Emplacement and cooling of komatiitelavas.Nature309:19–22

Jablonski D. 1997. Progress at the K-T bound-ary.Nature387:354–55

Joyce B, Sutalo F. 1996. Long basaltic lava flowsin southeastern Australia: Mt. Rouse andother late-Cenozoic flows of the Newer Vol-canic Province. InChapman Conf. Long LavaFlows, ed. PJ Stephenson, PW Whitehead,pp. 30–31. Townsville, Australia: JamesCook Univ. Qld. (Abstr.)

Kale VS, Kulkarni H. 1992. IRS 1A and LAND-SAT data in mapping Deccan Trap flowsaround Pune, India: implications on hydro-geological modeling.Int. Arch. Photogramm.Remote Sens.29:429–35

Keszthelyi L. 1995a. Measurements of the cool-ing at the base of pahoehoe flows.Geophys.Res. Lett.22:2195–98

Keszthelyi L. 1995b. A preliminary thermalbudget for lava tubes on the Earth and planets.J. Geophys. Res.100:20411–20

Keszthelyi L. 1996. Measurements of the cool-ing of pahoehoe lava flows.Eos 77:F807(Abstr.)

Keszthelyi L, Pieri DC. 1993. Emplacementof the 75-km-long Carrizozo lava flow field,south-central New Mexico.J. Volcanol. Geo-therm. Res.59:59–75

Keszthelyi L, Self S. 1997. Some physical re-quirements for the emplacement of longbasaltic lava flows.J. Geophys. Res.In press

Keszthelyi L, Self S, Thordarson Th. 1997. Ap-plication of recent studies on the emplace-ment of basaltic lava flows to the DeccanTraps.Mem. Geol. Soc. India.In press

Kilburn CRJ, Luongo G. 1993.Active Lavas:Monitoring and Modelling.London: UCLPress. 287 pp.

Lockwood JP, Lipman PW. 1987. Holoceneeruptive history of Mauna Loa Volcano. SeeDecker et al 1987, pp. 506–36

Macdonald GA. 1953. Pahoehoe, aa, and blocklava.Am. J. Sci.251:169–91

Macdonald KC, Haymon R, Shor A. 1989. A220 km2 recently erupted lava field on theEast Pacific Rise near latitude 8◦S. Geology17:212–16

Manga M. 1996. Waves of bubbles inbasaltic magmas and lavas.J. Geophys. Res.101:17457–65

Mangan MT, Helz RT. 1986. The distributionof vesicles and olivine phenocrysts in sam-ples from drill hole KI 79-3, Kilauea Iki lavalake, Hawaii.US Geol. Surv. Open-File Rep.86-424.25 pp.

Mangan MT, Wright TL, Swanson DA, ByerlyGR. 1986. Regional correlation of GrandeRonde Basalt flows, Columbia River BasaltGroup, Washington, Oregon, and Idaho.

Geol. Soc. Am. Bull.97:1300–18Martin BS. 1989. The Roza Member, Columbia

River Basalt Group: chemical stratigraphyand flow distribution. See Reidel & Hooper1989, pp. 85–104

Mattox TN, Heliker C, Kauahikaua J, Hon K.1993. Development of the 1990 KalapanaFlow Field, Kilauea Volcano, Hawaii.Bull.Volcanol.55:407–13

McCormick MP, Thomason LW, Trepte CP.1995. Atmospheric impact of the Mt.Pinatubo eruption.Nature373:399–404

McMillan K, Long PE, Cross RW. 1989. Vesic-ulation in Columbia River basalts. See Reidel& Hooper 1989, pp. 157–67

McWilliams MO, Baksi AK, Bohor BF, IzettGA, Muali AV. 1992. High precision relativeage of K/T boundary events in North Amer-ica and Deccan Trap volcanism in India.Eos73:363

Miller J, Thordarson Th, Larsen G. 1996. Sul-fur degassing and nature of eruptive activityduring the 935AD Eldgja eruption, S.Iceland.Eos77:F803 (Abstr.)

Oze C. 1997.A petrological analysis of blue-glassy pahoehoe, Kiluaea, Hawaii.BS the-sis. Whitman College, Walla Walla, WA. 41pp.

Peck DL. 1978.Cooling and vesiculation ofAlae lava lake, Hawaii. US Geol. Surv. Prof.Pap. 935B, pp. B148–51

Peterson DW, Holcomb RT, Tilling RI, Chris-tiansen RL. 1994. Development of lava tubesin the light of observations at Mauna Ulu,Kilauea Volcano, Hawaii.Bull. Volcanol.56:343–60

Phadke AV, Sukhtankar RK. 1971. Topographicstudies of Deccan Traps hills around Poona,India.Bull. Volcanol.35:709–18

Pierazzo E. 1997.The Chicxulub impact and theenvironmental catastrophe at the end of theCretaceous period.PhD thesis. Univ. Ariz.,Tucson. 208 pp.

Pinkerton H, Wilson L. 1994. Factors control-ling the lengths of channel-fed lava flows.Bull. Volcanol.56:109–20

Puffer JH, Horter DL. 1993. Origin of peg-matitic segregation veins within flood basalts.Geol. Soc. Am. Bull.105:738–48

Rampino MR, Caldeira K. 1992. Episodes ofterrestrial geological activity during the past260 million years: a quantitative approach.Celest. Mech. Dyn. Astron.51:1–18

Rampino MR, Haggerty BM. 1997.40Ar/39Arages of continental flood basalts: correla-tions with mass extinctions and other geo-logic events.Earth. Planet. Sci. Lett.In revi-sion

Rampino MR, Stothers RS. 1988. Flood basaltvolcanism during the past 250 million years.Science241:663–68

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



PAHOEHOE 109

Reidel SP, Hooper PR, eds. 1989.Volcanismand Tectonism in the Columbia River FloodBasalt Province. Geol. Soc. Am. Spec. Pap.239.386 pp.

Rogan W, Blake S, Smith I. 1996. In situ chem-ical fractionation in thin basaltic lava flows:examples from the Auckland volcanic field,New Zealand, and a general physical model.J. Volcanol. Geotherm. Res.74:89–100

Rossi MJ, Gudmundsson A. 1996. The mor-phology and formation of flow-lobe tumulion Icelandic shield volcanoes.J. Volcanol.Geotherm. Res.72:291–308

Rowland SK, Walker GPL. 1987. Toothpastelava: characteristics and origin of a lava struc-tural type transitional between pahoehoe andaa.Bull. Volcanol.52:631–41

Ryder G, Fastovsky D, Gartner S, eds. 1996.TheCretaceous-Tertiary event and other catas-trophes in Earth history. Geol. Soc. Am. Spec.Pap. 307.569 pp.

Self S, Thordarson Th, Keszthelyi L. 1997. Em-placement of continental flood basalt lavaflows. In AGU Geophys. Monogr. Large Ig-neous Processes, ed. JJ Mahoney, M Coffin.In press

Self S, Thordarson Th, Keszthelyi L, WalkerGPL, Hon K, et al. 1996. A new model for theemplacement of the Columbia River Basaltsas large inflated pahoehoe sheet lava flowfields.Geophys. Res. Lett.23:2689–92

Sharpton VL, Marin LE, Carney JL, Lee S, Ry-der G, et al. 1996. A model of the Chicxulubimpact basin based on evaluation of geophys-ical data, well logs, and drill core samples.See Ryder et al 1996, pp. 55–74

Shaw H, Swanson D. 1970. Eruption and flowrates of flood basalts.Proc. Second ColumbiaRiver Basalt Symp., pp. 271–99. Cheney:East. Wash. State College Press

Stephenson PJ, Burch-Johnson AT, Stanton D.1996. Long lava flows in North Queens-land: context, characteristics, emplacement.In Chapman Conf. Long Lava Flows, pp. 86–87. Townsville, Australia (Abstr.)

Stephenson PJ, Griffin TJ. 1976. Some longbasaltic lava flows in North Queensland. InVolcanism in Australia, ed. RW Johnson, pp.41–51. Amsterdam: Elsevier

Stothers RB. 1993. Flood basalts and extinctionevents.Geophys. Res. Lett.20:1399–402

Stothers RB. 1996. The great dry fog of 1783.Clim. Change32:79–89

Swanson DA. 1973. Pahoehoe flows from the1969-1971 Mauna Ulu eruption, Kilauea Vol-cano, Hawaii.Geol. Soc. Am. Bull.84:615–26

Swanson DA, Wright TL. 1980. The regionalapproach to studying the Columbia RiverBasalt Group.Mem. Geol. Soc. India3:58–80

Swanson DA, Wright TL, Helz RT. 1975. Linearvent systems and estimated rates of magmaproduction and eruption for the Yakimabasalt on the Columbia Plateau.Am. J. Sci.275:877–905

Theilig E. 1986.Formation of pressure ridgesand emplacement of compound basaltic lavaflows.PhD thesis. Ariz. State Univ., Tempe.212 pp.

Theilig E, Greeley R. 1986. Lava flows on Mars:analysis of small surface features and com-parisons with terrestrial analogs.J. Geophys.Res.92:E193–206

Thordarson Th. 1995.Volatile release and at-mospheric effects of basaltic fissure erup-tions.PhD thesis. Univ. Hawaii, Manoa, Hon-olulu. 580 pp.

Thordarson Th, Self S. 1993. The Laki (Skaf-tar Fires) and Grimsvotn eruptions in 1783-1785.Bull. Volcanol.55:233–63

Thordarson Th, Self S. 1996. Sulfur, chlorineand fluorine degassing and atmospheric load-ing by the Roza eruption, Columbia RiverBasalt Group, Washington, USA.J. Volcanol.Geotherm. Res.74:49–74

Thordarson Th, Self S. 1997a. The Roza Mem-ber, Columbia River Basalt Group, a giganticpahoehoe lava flow field formed by endoge-nous processes.J. Geophys. Res.In review

Thordarson Th, Self S. 1997b. Atmosphericand environmental effects of the 1783-84 ADLaki eruption.Glob. Planet. Change.In revi-sion

Thordarson Th, Self S, Oskarsson N, Hulse-bosch T. 1996. Sulfur, chlorine, and fluo-rine degassing and atmospheric loading bythe 1783-1784 AD Laki (Skaftar Fires) erup-tion in Iceland.Bull. Volcanol.58:205–25

Tilling RI, Christiansen RL, Duffield WA, EndoET, Holcomb RT, et al. 1987. The 1972-1974Mauna Ulu eruption, Kilauea Volcano: an ex-ample of quasi-steady-state magma transfer.See Decker et al 1987, pp. 405–70

Tolan TL, Reidel SP, Beeson MH, AndersonJL, Fecht KR, Swanson DA. 1989. Revisionto the estimates of the areal extent and vol-ume of the Columbia River Basalt Group. SeeReidel & Hooper 1989, pp. 1–20

Toon OB, Zahnle K, Morrison D, Turco RP,Covey C. 1997. Environmental perturba-tions caused by the impacts of asteroids andcomets.Rev. Geophys.35:41–78

Vandamme D, Courtillot V, Besse J, MoutignyR. 1991. Paleomagnetism and age determi-nations of the Deccan Traps (India): resultsof a Nagpur-Bombay traverse and reviews ofearlier work.Rev. Geophys.29:159–90

Venkatesan TR, Pande K, Gopalan K. 1993.Did Deccan volcanism pre-date the Creta-ceous/Tertiary transition?Earth Planet. Sci.Lett.119:181–89

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.



110 SELF ET AL

Walker GPL. 1971. Compound and simplelava flows and flood basalts.Bull. Volcanol.35:579–90

Walker GPL. 1989. Spongy pahoehoe inHawaii: a study of vesicle-distribution pat-terns in basalt and their significance.Bull.Volcanol.51:199–209

Walker GPL. 1991. Structure, and origin by in-jection under surface crust, of tumuli, “lavarises,” “lava-rise pits,” and “lava inflationclefts” in Hawaii. Bull. Volcanol. 53:546–58

Waters AC. 1961. Stratigraphic and lithologicvariations in the Columbia River Basalt.Am.J. Sci.259:583–611

Williams H, McBirney AR. 1979.Volcanol-ogy. San Francisco: Freeman Cooper. 397pp.

Wilmoth RA, Walker GPL. 1993. P-type and S-type pahoehoe: a study of vesicle distributionpatterns in Hawaiian lava flows.J. Volcanol.Geotherm. Res.55:129–42

Wood CA. 1992. Climatic effects of the 1783Laki eruption. InThe Year Without Summer?,ed. CR Harrington, pp. 58–77. Ottawa: Can.Mus. Nat.

Wright TL, Okamura RT. 1977.Cooling andcrystallization of tholeiitic basalt, 1965Makaopuhi lava lake, Hawaii. US Geol. Surv.Prof. Pap. 1004.78 pp.

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.

Annual Review of Earth and Planetary Science Volume 26, 1998

CONTENTSContemplation of Things Past, George W. Wetherill 1

Volcanism and Tectonics on Venus, F. Nimmo, D. McKenzie 23

TEMPERATURES IN PROTOPLANETARY DISKS, Alan P. Boss 53

THE IMPORTANCE OF PAHOEHOE, S. Self, L. Keszthelyi, Th. Thordarson 81

CHINESE LOESS AND THE PALEOMONSOON, Tungsheng Liu, Zhongli Ding 111

STELLAR NUCLEOSYNTHESIS AND THE ISOTOPIC COMPOSITION OF PRESOLAR GRAINS FROM PRIMITIVE METEORITES, Ernst Zinner

147

NOBLE GASES IN THE EARTH'S MANTLE, K. A. Farley, E. Neroda 189

SATELLITE ALTIMETRY, THE MARINE GEOID, AND THE OCEANIC GENERAL CIRCULATION, Carl Wunsch, Detlef Stammer 219

CHEMICALLY REACTIVE FLUID FLOW DURING METAMORPHISM, John M. Ferry, Martha L. Gerdes 255

CHANNEL NETWORKS, Andrea Rinaldo, Ignacio Rodriguez-Iturbe, Riccardo Rigon 289

EARLY HISTORY OF ARTHROPOD AND VASCULAR PLANT ASSOCIATIONS, Conrad C. Labandeira 329

Ecological Aspects of the Cretaceous Flowering Plant Radiation, Scott L. Wing, Lisa D. Boucher 379

The Re-Os Isotope System in Cosmochemistry and High-Temperature Geochemistry, Steven B. Shirey, Richard J. Walker 423

DYNAMICS OF ANGULAR MOMENTUM IN THE EARTH'S CORE, Jeremy Bloxham 501

FISSION TRACK ANALYSIS AND ITS APPLICATIONS TO GEOLOGICAL PROBLEMS, Kerry Gallagher, Roderick Brown, Christopher Johnson

519

Isotopic Reconstruction of the Past Continental Environments, Paul L. Koch 573

The Plate Tectonic Approximation: Plate Nonrigidity, Diffuse Plate Boundaries, and Global Plate Reconstructions, Richard G. Gordon 615

LABORATORY-DERIVED FRICTION LAWS AND THEIR APPLICATION TO SEISMIC FAULTING, Chris Marone 643

Seafloor Tectonic Fabric by Satellite Altimetry, Walter H. F. Smith 697

Ann

u. R

ev. E

arth

Pla

net.

Sci.

1998

.26:

81-1

10. D

ownl

oade

d fr

om a

rjou

rnal

s.an

nual

revi

ews.

org

by S

choo

l of

Oce

an &

Ear

th S

cien

ce &

Tec

hnol

ogy

on 0

1/02

/08.

For

per

sona

l use

onl

y.

Documents

THE IMPORTANCE OF PHOEHOEscott/Workshop_reading/Self_etal_199… · P1: NBL/vks P2: NBL/plb QC: NBL/uks T1: NBL March 2, 1998 15:12 Annual Reviews AR055-04 PAHOEHOE 83 small pahoehoe