I. Definition and Significance of laharsII. Genesis of LaharsIII. Behavior of lahars: Downstream ProcessesIV. lahar DepositsV. Physical Modeling

GLOSSARY

bulking The erosion and incorporation of secondary, exoticdebris by lahars as they move downstream.

debris avalanche A flowing miXture of debris, rock, andmoisture that moves downslope under the influence ofgravity. Debris avalanches differ from debris flows irithat they are not water saturated and in that the load isentirely supported by particle-particle interactions.

debris flow A water-saturated mixture of debris and waterhaving large sediment concentration that moves down-slope under the influence of graviry. Forces related toboth solid and fluid phases act together to drive debrisflows and determine their distinctive behavior. A fairlyuniform mixture of solid and liquid phases in verticalprofiles characterizes debris flows and distinguishes themfrom more water-rich hyperconcentrated flows. The lit-erature suggests a gradational boundary of 50% to as

much as 60% sediment by volume separating debris flowsand hyperconcentrated flows. Sediment concentration isnot definitive because behavior also depends on factorssuch as sediment-size distribution and degree of agi-tation.

debulking A process in which the lahar selectively depositscertain particles, owing to their size or density, as itmoves downstream. Debulking differs from the generaldeposition of bulk sediment in preferentially removingparticles, usually large or dense ones, from the flow.

granular temperature A measure of degree of particle agita-tion, usually defined as the ensemble average of thesquare of the fluctuating components v; of instantaneousgrain velocities, v,: T = (V;2) = «v, - V,)2), where V, is

the mean velocity component. Granular temperaturemay be interpreted as twice the fluctuation kinetic energyper unit mass of granular solids.

hyperconcentrated flow A gravitationally driven, nonuni-form mixture of debris and water having water ~ontentlarger than that of debris flow but less than that of muddystreamflow. Some variation with depth of solids fractioncharacterizes hyperconcentrated flows and distinguishestheir behavior from that of debris flows. The literaturereports depth-integrated solids concentrations of be-tween 20 and 50-60% for hyperconcentrated flows.Hyperconcentrated flows possess fluvial characteristics,yet carry very high sediment loads.

Copyright @ 2000 by Academic PressAll rights of reproduction in any fonn reserved.Enryciopedia of Vokanoes 601

LAHARS

]At\1ES W. VALLANCEMcGill University

LAHARS602

lahar An Indonesian term that most commonly means de-bris flow, transitional flow, or hyperconcentrated floworiginating at a volcano. Though some floods and muddystreamflows are genetically related to lahar events, mostare not, and the term lahar is not generally extended toinclude such flows. Although many workers use laharfor both the process and the deposits mat result fromit, it is better to restrict the term to the process.

muddy streamflow A flow that essentially transports sedi-ment as normal streams do, with fine-grained sedimentin suspension and coarse-grained sediment movingpiecemeal along the bed as bedload. Muddy streamflowsand floods have smaller sediment concentrations thando hyperconcentrated flows.

phase (of flow) The flow type in a lahar wave at some timeand place. Phases associated with lahars include debrisflows, hyperconcentrated flows, and muddy streamflows.

stage (of flow) The height above the channel bottom ofthe flowing lahar. Examples of lahar stages include theinitial rising or waxing stage, the peak-inundation stage,and me final long-duration falling or waning stage.

turbulence Chaotic fluid movement [normal to directionof flow], or deviation of flow from laminar. Turbulentflow characterizes streamflows but not debris flows orsediment-rich hyperconcentrated flows.

mobilized loose debris and generated hundreds of laharsafter Mount Pinatubo in the Philippines erupted cata-clysmically in 1991. Although each of the Pinatubo la-hars was relatively small, within 3 years, they cumula-tively remobilized nearly 3 of the 10 km3 of juvenilepyroclastic debris emplaced during the eruption. Fillingof downstream channels and overbank flow onto sur-rounding fields and villages inundated more than 400km2 and displaced more than 50,000 people.1ri contrast,the 1985 pyroclastic eruption of Nevadc;> del Ruiz inColumbia was relatively small (0.01 km3) but generatedmuch larger lahars (total volume of ~0.1 km3). Theinteraction of hot pyroclastic flows or surges and glacialice and snow at the summit of the volcano caused theselahars. The lahars flowed up to 100 km down four offive drainages that head at the volcano. These werethe deadliest lahars worldwide to have ocCurred duringhistoric time. They destroyed more than 5000 homesand killed more than 23,000 people. in the town ofArmero, 73 km downstream of the volcano, virtually allstructures in the path of a lahar were obliterated andthree-quarters of the irihabitants killed.

Studies of prehistoric lahar deposits indicate the po-tential for even greater disasters. About 5600 years B.P.,the enormous (3.8 km3) prehistoric Osceola Mudflowfrom Mount Rainier in Washington state began with adebris avalanche that transformed to a lahar as it de-formed [and dilated] because of the enormous volumeof water contained in pore spaces and in the volcano'shydrothermal system. It filled valleys from depths of85 m to as much as 200 m, flowed more than 120 kmdown valleys, and continued f6r up to 20 km under theWater of Puget S6und while retaining sufficient c6her-ence to transport large gravel and wood fragments. Itinundated an area in exCess of 350 km2, which is nowpopulated by hundreds of thousands of people. A largepyroclastic flow that mixed with more thati 1 km3 ofglacial ice and snow generated a prehistoric lahar atCotopaxi in ECuador that was as voluminous as the Osce-ola Mudflow. The lahar filled valleys to depths of 80-200 m, flowed more than 300 km to the Pacific Ocean,and covered at least 440 km2.

L AHARS OCCUR DURING volcanic eruptionsor, less predictably, through other processes com-

mon to steep volcanic terrain when large masses of sedi-ment, and water, sweep down and off volcano slopesincorporating additional sediment. Because lahars arewater saturated, both liquid and solid interactions deter-mine their unique behavior and distinguish them fromother related phenomena common to volcanoes such asdebris avalanches and floods. The rock fragments car-ried by lahars make them especially destructive; abun-dant liquid contained in them allows them to flow overgende gradients and inundate areas far away from theirsources. People in such distal areas commonly neitherexpect the danger nor anticipate the destructive powerof lahars.

I. DEFINITION AND SIGNIFICANCE

OF LAHARSB. Purpose

Because the timing of lahar events is unpredictable andworking with active flows can be hazardous, much ofour present knowledge of flow behavior is inferred fromstudy of lahar deposits. Nonetheless, a few keyobserva-tional and experimental studies have improved under-standing of debris-flow processes. The chief purpose of

A. Historic and Prehistoric Examplesof Lahars

Lahars at volcanoes inundate surrounding areas anddamage or destroy nearby communities. Torrential rains

LAHARS 603

this article is to summarize what is known about thenature and behavior of lahars on the basis of observa-tions, experiments, and careful examination of depositsand further to describe carefully the nature of depositsderived from such events.

instead of a lahar. Similarly, when high-discharge flowcontinues for many hours or even days to weeks, as itcan during lake-breakout floods, there is not enoughsediment available along the drainage system to formlahars. Lastly, in catchments where there is little or noeasily erodible clastic debris or soil material, bulking isretarded, and lahars do not occur.

At volcanoes, lahars induced by sudden water releasecan occur by four principal means. (1) Pyroclastic flowsand surges mix with and rapidly melt glacial ice and firn.Generally, such pyroclastic flows come to rest or nearlydo, forming meltwater that then runs off, coalesces, anderodes the pyroclastic debris to form water-rich lahars.The lahars continue to bulk up with volcaniclastic de-bris, glacial drift, alluvium, and colluvium as they flowdownstream so that within a few kilometers to severaltens of kilometers they become debris flows. Laharsof this type are considered primary. (2) Eruptions candisplace large volumes of crater-lake water that formlahars downstream. Crater lakes, caldera lakes, and vol-canically debris dammed lakes can also break out monthsto years after eruptions. Such delayed breakouts occurwhen water levels gradually rise and then overtop andrapidly incise fragile debris dams. (3) Subglacial erup-tions can form subglacial lakes that eventually break outwhen a section of the ice cap becomes buoyant andreleases the trapped water. Small-scale outburst floodsoccur as normal glacial processes during periods of abla-tion. Although huge eruption-driven outbursts causesediment-rich, water floods called jokulhlaups, smallsecondary ones commonly bulk up to form lahars.(4) Lahars owing to intense rainfall often occur afterpyroclastic eruptions deposit abundant loose debris inthe form of pyroclastic-flow or -fall deposits sur-rounding vents of volcanoes. Lahars of this type arecommonly small but frequent in occurrence during rainyperiods. Size and frequency of rain-induced lahars mayincrease in the months or years following the primarypyroclastic eruption and then slowly decrease as drain-age networks and vegetation reestablish themselves (e.g.,Mount Pinatubo after its 1991 pyroclastic eruption).

Because clay-rich sediment is both uncommon on theflanks or aprons of active volcanoes and resistant toerosion, lahars induced by sudden water release gener-ally appear to be clay-poor (less than about 5% clay/(sand + silt + clay).

C. Tenninology

Lahar is an Indonesian term that most commonly meansdebris flow, transitional flow, or hyperconcentrated floworiginating at a volcano. Muddy streamflows and floodshave lower sediment concentrations than do hyper~concentrated flows and essentially transport sedimentas normal streams do, with fine-grained sediment insuspension and coarse-grained sediment moving alongthe bed. Although some floods and muddy streamflowsare genetically related to lahar events, most are not, andthe term lahar is not generally extended to include suchflows. Although many workers use lahar for both theprocess and the deposits that result from it, it is betterto restrict the term to the process.

II. GENESIS OF LAHARS

Lahars may be primary (syneruptive) or secondary (post-eruptive or unrelated to eruptions). Lahar genesis re-quires (1) an adequate water source; (2) abundantuncon-solidated debris that typically includes pyroclastic-flowand -fall deposits, glacial drift, colluvium, soil, etc,;(3) steep slopes and substantial relief at the source; and(4) a triggering mechanism, Water sources include poreor hydrothermal water, rapidly melted snow and ice,subglacially trapped water, crater or other lake water,and rainfall runoff.

A. Lahars Induced by Sudden Melting ofSnow and Ice, Voluminous Floods, orTorrential Rains

Floods of water moving across die easily erodible looseclastic sediments common on die flanks and ap~ons ofvolcanoes easily incorporate diat debris and may quicklybulk up to form lahars (Figs. IA and IB). Bulking iscritical to all lahars diat begin widi sudden water re-leases. Lahar formation depends on die right mix ofreadily available sediment and water discharge. If peak-water discharge is huge, die amount of sediment avail-able for bulking may be inadequate and a flood results

B. Edifice- or Flank-Collapse-Induced Lahars

Although most edifice collapses behave as debris ava-lanches, those with sufficient, widely dispersed, pore and

Il)aJ '"

~ft

E

.- C

.- Q

) .-

I- ",""!U

!U

~

I..a.

"0

"" ~

0 I..

~

..Q~

aJ E

u-ca

.-E

'"

.-.-

a C

0

1--00ij;-

.p"Gi

"=

~""

!II .-

0~

""

E

£ E

c

'" '-

!U

0!II

aJ.-aJ

OJ

1..'"E

c:

... !U

.- E

~

""I-

a >

~

.:: >

"0

~ .g ~

!/) .

a 0

Q)

a. "0>

OJ

1 ..,

aJ ~

(\I ,g

"" Q

)

aJ!IIC6-0

E

~

~

0.-

-t: >

- I..

I- ,\(

Q)

a.

., ~

Q

)...~c

...

Q)

'"~

Q

)~

...~

- Q

) ...

. >

'"

Q)

!U~

E~

- .-

Q) =

I- .D

O

'"

""-C

!U

!U

E""

!UQ

) I..

... ~

!U

.-'p

""'c

Q)

.- ~

Q)I-

I..!U

.",O

""""0

c0

!U

'+=

UI..Q

) C

... .-

~

~~

0

...~.-

'"

~

Q)

"01..C

!U

.- '"

C

Q)

C

'".-

!U

~~

.D

a.

"'~I.. C

!U

0~

.-

!U

.'"-

in

~

C0

~~

...

"0""C

C

.- !U

~

..

OC

O~

C

'" .-

'" C

~

~

g-oI..~"0'"0

'"I..

.-""

Q)

>-'"

~

!U

u~.-

a.

~

~E

0

Q)'+

=~

''"

u .-

V)

I..

.DQ)

-""

w<

t.ci0:

.S ""

:J C

~

~~

ULLo

~

C'"

.-

605,AHARS

hyperconcentrated flows or debris flows. In medial ordistal reaches, debris flows can also transform back towater-rich hyperconcentrated flows or, ultimately,floods. Further, debris avalanches proximal to volcanoescan transform totally to debris flows as they move down-stream.

A. Erosion and Bulking

Lahars cause erosion by undercutting steep slopes andterrace scarps and by scouring their beds. Erosion isstrongest along steep reaches underlain by loose clasticsediment and weakest either along reaches underlain byhighly resistant bedrock or along reaches with very gen-tle gradients. Along any particular reach, water-rich,hyperconcentrated phases are typically more erosivethan sediment-rich debris-flow phases (Fig. 2), but localerosion can occur during any flow phase. The waXingstage of a lahar is likely to coincide with the most wides-pead and voluminous erosion and bulking. The finalwaning stages of a lahar are also commonly erosive andultimately result in the incision of channels in freshlahar deposits.

Erosion at the base of a lahar occurs by piecemealdislodgment of particles, by undercutting at upstreammigrating knickpoints, and by rip up of sediment owingto root heave of falling trees. The presence of undis-turbed, delicate, deposits such as tephra layers at thebase of debris-flow deposits suggests that piecemeal ero-sion of particles occurs chiefly during more water-richhyperconcentrated phases. Piecemeal erosion is proba-bly not an important process outside of active channels.If channel beds comprise erodible sediment, sequencesof upstream migrating knickpoints are common duringhyperconcentrated-flow phases. Typically knickpointsteps are a few tens of centimeters to a meter or so highand spaced on the order of tens to hundreds of metersapart. During more sediment-rich flows, knickpointsbecome higher and more widely spaced. Knickpointsmay disappear during debris-flow phases, which, in fact,commonly aggrade rather than erode their beds. Laharsvoluminous enough to escape channels knock treesdown and incorporate them. Root balls of falling treesdrag considerable sediment into the active flow andloosen even more sediment that is then available forerosion. Voluminous lahars that inundate large areas offorested terrain can incorporate considerable quantitiesof sediment and huge amounts of wood in this way.

Undercutting of steep slopes, fluvial terrace scarps,and active stream banks is probably the most importantway in which lahars erode and incorporate sediment.

hydrothennal water in the precollapse rock liquefy asthe material defonns during collapse. The shallow intru-sion of magma within an edifice is the likeliest causeof collapse-induced lahars larger than about 0.2 kmJ.Smaller collapse-induced lahars may have various trig-gers, including magmatic or phreatic volcanism and vol-canic or tectonic earthquakes.

The process of hydrothennal alteration, especially atglaciated volcanoes, increases the probability of edifice-collapse lahars. Acid-sulfate leaching in hydrothennalsystems removes mobile elements, adds sulfate, andbreaks down framework silicates to fonn silica phases,such as cristobalite and opal, and clay minerals, such askaolinite and smectite. This process weakens the rockso that it more readily disintegrates during defonnationafter collapse. Thus, huge blocks typical of debris ava-lanches are less common in lahars of this type eventhough their origin is the same. Abundant alterationminerals, especially clay minerals, increase porosity anddecrease penneability of the rock and thus, in combina-tion with the hydrothennal system, trap a widely dis-persed reservoir of water within the precollapse mass.Because of its water content and its tendency to disinte-grate, hydrothennally altered rock, unlike fresh rock,easily liquefies as it defonns. Collapse-induced laharsare typically clay-rich, and most are observed to havegreater than 5% clay/(sand + silt + clay).

Clay-rich, collapse-induced lahars appear to be morecommon at ice-clad volcanoes than at volcanoes thatare free of ice. Glacial erosion tends to expose deeper,potentially more altered, portions of volcanoes by moreeffectively incising into an edifice. Further, incisedslopes in altered rock are more susceptible to failure notonly because the rock is weak but also because it isoversteepened. Lastly, melting glacial ice provides aslow-release source of water that is important to theefficient operation of the acid-sulfate leaching process.

At and near some volcanoes, especially tropical ones,tectonic earthquakes cause multiple slope failures thatultimately coalesce and fonn collapse-induced lahars.Such lahars are unrelated to volcanism but, owing tocharacteristics of their deposits (commonly clay-rich,hummocky, and voluminous), can be mistaken for laharsgenerated by major edifice collapse of hydrothennallyaltered sectors of volcanoes.

III. BEHAVIOR OF LAHARS:.

DOWNSTREAM PROCESSES

Lahars can change character downstream. Some floodsincorporate enough sediment proximally to become

-roO;) S;)I;)~~d ~!SU;)P-A\OI 'Apu;)nb;)suo:) "~!SU;)P P!filJ "1U;)W!P;)S ;)!.I~q~I p;)1!sod;)p AIsno!A;).ld ;)S!;)U!;)A!1;);)}J;) ;)q1 ;)S~;).I;)U! 01 (pU~S P;)U!~.I.3'-;)UIJ S;)S~;) ;)WOS AIUOWWO;) S.l~~I}O S;)~~1S ~U~A\ I~U!.!I ";)~~1S ~U!P;);)

U! 1nq 1I!S pU~ A~I;) An~.I;)U;).&) S;)UIj 1U;)!;)9jns pu;)dsns -;).ld ;)q1 mq1 ;)~.I~q;)S!P U! snou!WfiIoA SS;)I 1nq ';)A!SO.l;)s;)s~qd P!filJ ;)soqA\ s.l~q~uo ;);)~J.Ins ;)q1 01 ;)S!.I .I;)1~A\ mq1 ;).lOW pm 'A.I;)1~A\ ;).lOW An~;)!dA1 S! A\OlJ ;)~~1S-~U!mA\

;)SU;)P ;).lOW Apq~!IS S;)I;)~d 'U°!1!PP~ UJ "ApU~Aonq ;)S!.l ~UIJ ;)q.L "1! S;)P°.l;) mq1 .I;)q1~.1 1U;)W!P;)S S1!sod;)pC .I;)1~A\ mq1 ;)SU;)P SS;)I ;).I~ 1~q1 ;);)!Wfid S~ q;)ns S;)I;)~~d AI;)A!1;)~ AIUOWWO;) pm ;)A!SO.l;) SS;)I S! A\OlJ ~~;)d SA\OnO}

"S;).I~ ~U!.I~;)qs .10 ~U!1~.lq!A U! ;)Pli°;) .10 qn.I ~!A~.I.3' 1~q1 A\OlJ ~U!n~} q;)!.I-1U;)W!P;)S ;)q.L "1U;)W!P;)S ;)!1°X;)S~ q;)ns S;);).lO} Apoq Aq p;);)u;)nlJU! S;)I;)!1.1~d I~np!A!pu! ;)1~.l0d.r0;)U! AI!P~;).I 1S0W snq1 pm ;)A!SO.l;) 1S0W ;)q1 ;).I~

. U;)qA\ AlJ;)!q;) S.lfi;);)O SS;);)O.ld ;)q.L "W;)q1 q1~;)U;)q u;)do A\OlJ ;)q1 }O S;)~~1S ~~;)d ~U!A\OnO} pm 1U°.IJ A\OlJ ~U~A\

1~q1 sd~ 01U! p.l~A\UA\Op ;)AOW 'S;);).lO} Apoq I~UO!1~1!A~.I.3' ;)q1 'An~;)!dA.L ":IJ!.lP I~!;)~I~ pu~ 'Wll!Ann°;) 'wn!Ann~}O ;);)u;)nlJU! ;)q1 .I;)pun 'S;)I;)~~d q;)!qA\ U! SS;);)O.ld ~ S! S~ q;)ns '1U;)W!P;)S ;)!1°X;) U! .I;)q;)!.l ;)WO;);)q SA\OlJ S!.lqU°!1~I°;).I;)d "S;)I;)~~d ;)~.I~I }O ~U!YU;)S I~!1U;).I;)};).ld S1;)~ -;)P 'u°!1~.lodsu~.Il p;)nu!1UO;) q1!M. "SA\OlJ S!.lq;)P ;)WO;)

-.l;)1unO;) 1~q1 SS;);)O.ld ~ 'U°!1~I°;).I;)d .lOA~} A~W pu~ S;)SS;);) -;)q AI;)1~W!1In A\OlJ }O S;)~~S ~U!U~A\ pm ~~A\ q1oq-o.ld ;)S;)q1 .I;)PU!q SU°!1;)~.IJ SP!IOS .I;)1~;).lD "S;)SS;);)O.ld 's;)nu!1UO;) SS;);)O.ld ;)q1}1 "(I "~!ill s;)s~qd q;)!.l-1U;)W!P;)S

;)S;)q1 .I;)PU!q 01 S;)I;)~~d .I;)A\;)} ;).I~ ;).I;)q1 ;)sn~;);)q ;)S!.I ;).lOW 01 SA\OlJ P;)1~.IlU;);)UO;).I;)dAq pm SA\OlJ pOOlJ SW.lO}1u~Aonq .10 ~U!yu;)S ;)I;)~~d S.lOA~} A\OlJ ;)1nI!p ;).l°W "UO!S -su~.Il ~u~ynH "S.I~q~I UO .I;)1;)~.I~q;) }O S;)~u~q;) W~;).IlS

-u;)dsns U! S;)I;)~~d P;)U!~.I.3'-;)UIJ }O u°!1.1odo.ld ;)q1 Aq -UA\OP s;)sodw! ~u~ynq W~;).IlSUA\Op ;)A!SS;).I.3'O.ldP;)U!UU;)1;)P ~U!;)q .I;)U~I ;)q1 '~!SU;)P p!filJ pm 'S;)I;)!1.1~d "S1U;)W;)~~.IJ I~np!A!PU! 01U! dn ~~;).lq AI;)1~W

;)S.I~O:J }O uo~odo.ld 'U°!1;)~.IJ SP!IOS Aq P;)1~W;)W ;).I~ -!1yn pu~ W.IO};)P A;)q1 ;).lO};)q W~;).IlSUA\Op S.I;)1;)WOI~S;)SS;);)O.ld UO!1~;).I.3';)S 1u~uodw! 1S0W ;)q11nq ';)Z!S .10 }O SU;)1 ;)AOW A~W S1S~I;)~~;)W ;)!1S~I:) "A~A\ S!q1 U! ~;)O.l

~!SU;)P Aq ;)1~~;).I.3';)S AI;)A!1;);)}J;) m;) s.I~q~I U! S;)I;)~~d -P;)q }O U;)A;) S;)W!1;)WOS pu~ '1U;)W!P;)S P;)1~P!IOSUO;)un

}O (UO!SU;)W!P U! W 01 <) S:>Y;)°Iq~;)W ~U!1~.lod.rO;)U!}O

S~SS~;)O.ld UOp~~~.I.3'~S ;)Iq~d~;) ;).I~ s.I~~I ;)~;re'1 "s;)s~qd A\olJ-P;)1~.IlU;);)uo;).I;)d

A:J!SU~a- pm ~Z!S-~I;)~~d "8 -Aq pm A\olJ-S!.lq;)P q1oq ~U!.rnp ;)A!1;)~ S! ~u!Un;).I;)pufl

.(a-v) UOpJ~S SSOJJ U! s~!sod~p MOI~-S!Jq~p pue -p~~eJ~u~JUOJJ~dAY ~o s~Ju~nb~s p~~J~dx~ ~Y~ SMOYS I~POW ~Y.L .[~od~'d ~ly-u~dO A~AJns leJ!golo~9 .s.n .:s~oueJloA ~e

spJezeH leJ!golOJpAH lep[lJns.. '(6661) UOJS pue UOSJ~!d U! I E"E .g!~ WOJ~ p~~depe] ~! WOJ~ P~A!J~p s~pe~ ~!sod~p pue ~seyd Mol~-p~~eJ~-U~JUOJJ~dAY o~ ~seyd MOI~-S!Jq~p WOJ~ uo!~nl!p we~J~SUMOp gU!ogJ~pun J~A!J e UMOp gU!AOW Jeyel e~o l~poW Jpew~YJS Z 3~ n£>l.:I

SNOI1:>3S SSOI:l:>

c

~

M

S~VHV'1 909

607AHARS

A SIDE VIEW(cross sectionof flow)

Recirculating motionof sm~11 particles

Recirculating motionof large particles

B TOPVIEW (near front of flow) Paths of large particles at surface

-, -,. ~._,. ,._, -,.- ~..'t.., ,r, ~ . '.., i ' '.., .., '.. '..., .

".';.';,""l';:" ':';.';: ';:';.'; ;..;:.;..:.;..;',,:..

":J~ '-:'~ .;;;;~: ;;,;; :;~::: :::;~:;""_. . _." . .; .""

.., :::~::" !~:.;~::

;;.-.:~~~;.~:;:~;:; J

:; :~:;;~:;:~~;

': .;~:: .; ~:: .;~::.; '.';";;";.';'-;.';'; ;.';.';.'-;.~

FIGURE 3 Schematic diagram illustrating how inverse parti-cle-size segregation results in (A) longitudinally and (B) laterallygraded flows.

available to be eroded and incorporated is commonlycoarser grained and better sorted (like alluvium, collu-vium, or bedrock) than debris in the main body of thelahar and will be preferentially incorporated at the flowfront (Fig. 4). This process can enhance or indepen-dently form the coarse-grained, better sorted perimetersthat are commonly observed in moving debris flows andtheir deposits.

C. Downstream Dilutionand Transformation

The gradual incorporation of water at the front of alahar flowing down an active stream channel causes pro-gressive loss of carrying capacity and an eventual changein the nature of the flow with distance downstream (Fig.1). This process is important only in lahars that followactive rivers or other bodies of water. Though dilutionof large lahars can occur, the process has little effect onthe behavior of lahars so huge that their volumes are

monly collect at the surface of lahars and may ultimatelyform rafts of material that appear to move en masse.Friction at the bed retards lahars and causes verticalprofiles in which velocities are smallest at the base andgradually increase upward. Low-density particles willmigrate upward toward the surface where velocities aregreater than average and will then migrate forward to-ward flow margins.

Particles that are more dense than the fluid will settleif the solids fraction is not great. In a viscous liquid,large dense particles will have higher terminal velocitiesand will settle faster than small particles of the samedensity. Through this process, the largest particles col-lect in the lowest horizon of the moving flow and pro-gressively smaller particles will collect in horizons abovethat; thus a normally graded flow develops. If particleconcentrations are great enough, particles collide withand rub one another as they settle. Such particle interac-tions inhibit the settling process.

In shearing or vibrating flows with particles denserthan the flUid and with particle concentrations greaterthan about 40%, percolation predominates rather thansettling. In such flows voids periodically open beneaththe particles. If the size of a particle is comparable toor smaller than the size of a void, the particle falls downinto the void. In a flow with a mixture of grain sizes,the probability is greater that voids of sufficient size willopen beneath small particles than will open beneathlarger ones. Percolation operates only downward; so inorder to preserve vertical mass balance, there must bea process by which particles migrate upward. Force im-balances or particle rotations can push particles fromone layer into another. This process, termed squeezeexpulsion, can operate either up or down and need notbe size preferential. The combination of percolationand squeeze-expulsion processes in shearing or vibratingparticle flows is known as kinetic sieving. The net resultof kinetic sieving is that small particles migrate down-ward and displace large ones, and large ones graduallymigrate upward. (Readers should note that in granularflows such as lahars, increased dispersive stress corre-sponds with increased granular temperature or energydissipation owing to grain collisions and inhibits sizesegregation rather than promotes it as suggested by Bag-nold.) The large particles then migrate forward towardthe margins of the flow because velocities are greatestnear the surface. The kinetic-sieve process thus not onlycan cause the inverse vertical grading that is commonin moving debris flows but can also cause the accretionof large particles at flow perimeters (Fig. 3).

Waxing flow fronts are commonly the most erosivepart of lahars, especially on steep slopes. The debris

LAHARS608

A B

:E0>"Qj.t:Q)0>to

Ci5

\-11/

~cremental11~~position

a &Time 2 Time 3

/!'

tQ.~

Time 1 Time 4

-~'...t.:11 ., ".0"".~;: .

- ~ ---

Time 4

En-massedepositionSchematic deposits from near the valley bottom

(all deposits are vertically exaggerated)~

Time 2

~

Time 3

~

Time 1

Schematic deposits from near the valley bottom (vertically exaggerated)

/""-./

./

Relative topographic positions of A and B.

FIG U R E 4 Schematic hydrographs showing the behavior and downstream changes of lahars that begin as avalanches of water-saturated debris. Note that with distance downstream (A to B), the lahar incorporates secondary exotic particles, especially near itsflow front. The diagram (B) also illustrates how an inverse longitudinally graded flow can accrete incrementally to form a normallygraded deposit [adapted from Vallance and Scott (1997)]. Exotic particles are most common at the base of normally graded deposits

and at inundation limits (B).

significantly greater than that of the water in the riverbeing overrun. The process of downstream dilution oc-curs more readily in clay-poor lahars than in clay-richlahars because (i) clay-poor lahars mix more readily withwater and (ii) clay-rich lahars are typically much larger.

Once off the flanks of the volcano and confined inriver channels, lahars, which typically move faster thannormal stream flow, push river water ahead of them andgradually, with distance downstream, begin to mix withthat water. As the flow front becomes progressively morewatery, it loses its capacity to carry larger gravel parti-cles, and these progressively lag behind the flow front.

With time and distance downstream, a dilution frontprogresses from the front of the lahar to its middle,and eventually the entire lahar becomes more dilute. Inlahars that occurred at Mount St. Helens in 1980 and1982, downstream dilution occurred over the course oftens of kilometers and caused a complete transformationfrom debris flow to hyperconcentrated flow (Fig. 1D).In medial reaches, the hyperconcentrated~flow phasepreceded the debris flow of the lahar because the dilutionprocess began at the flow front and then graduallyworked backward toward the tail as the flow migrateddownstream (Fig. 1 C).

609,AHARS

D. Depositional Processes

Emplacement of lahar deposits can occur en masse, bysteady incremental accretion, or, most likely, by someintermediate process in which accretion begins and thenaccelerates, wanes, or does both alternately. Debris-flowdeposits are poorly sorted, generally massive, and un-stratified. It is common to infer that such massive depos-its are emplaced en masse and that they represent afrozen portion of the debris flow itself. In contrast, hyp-erconcentrated-flow deposits are better sorted, com-monly show faint stratification, and hence are assumedto accrete during significant time intervals.

An increasing body of evidence suggests that depositsof both hyperconcentrated flows and debris flows accreteincrementally. Such evidence includes (1) strong align-ment of elongate clasts parallel to flow directions orimbrication of such clasts in up~tream directions,(2) strong changes in composition of particles with verti-cal position in outcrops, especially those that are graded,(3) inundation-limit deposits with clast compositionssimilar to those at the base of thick valley-bottom depos-its, (4) marks of peak-flow levels in upland valleys thatindicate flow depths 5-10 times greater than depositthicknesses, (5) abundant evidence of cataclasis (break-age of clasts owing to collisions), and (6) stratificationin deposits of transitional Or hyperconcentrated flows.

Although rapid deposition of a vertically size-segre-gated flpw can generate normally and inversely gradeddeposits, incremental accretion of longitudinally si~e-segregated flow can also be responsible. Because lateraland longitudinal variations in both sizes and composi-tions of particles commonly occur in moving lah3:rs,accretion from such laterally graded systems during sig-nificant time intervals can cause vertically graded depos-its. Figure 4B .(Times 1 to 5) illustrates schematicallyhow accretion from a debris-flow wave with a concentra-tion of large particles at its front can generate a normallygraded deposit. Note that accretion occurs for a shorttime only near inundation limits where grading doesnot occur (Time 1-2). The front of a lahar wave thatflows down a river becomes progressively more diluteand less capable of carrying larger particles, which lagbehind (Figs. lC and ID). Therefore, accretion from adilute debri~ flow that coarsens from head to tail pro-duces inversely graded deposits (Fig. 1 C). Farther down-stream, the entire lahar, and especially its flow front,becomes hyperconcentrated so that accretion producesfiner grained beds that may be inversely graded or bothinversely graded and normally graded (Fig. ID). Notethat deposits iri positions higher in the valley may alsobe graded, but often le~s obviously so.

Debris flows that do not undergo downstream dilu-tion commonly form cobble-boulder-rich perimetersowing to the size segregation process described above(Fig. 3). The surface of the flow with its concentrationof large particles moves faster than the rest of the flowso that cobbles and boulders migrate to the flow front.Once the flow peak has passed at any particular crosssection, boulders begin to accrete at flow margins toform coarse levees. Flow fronts, with concentrations oflarge angular particles, that move onto gentle slopesbecome progressively drier because water can more eas-ily escape from more permeable coarse-grained flowperimeters than from fines-rich flow interiors. The netresult isa dry, frictional, resistant flow perimeter thatsurrounds a liquefied interior. When flows of this wereach sufficiently gentle slopes, the frictional perimeterslows to a stop and leaves behind a steep-front~d fines-poor margin and a partly liquefied fines-rich interior.Because resistance to flow is great~r at the margin thanthe interior of the flow, bifurcation of the flow intofingers can result as the more mobile fines-rich interiQrdiverts around the more static perimeter.

Avulsion can occur in lowland areas with gentle slopeswhere sediment-rich portions of lahars accrete and fillthe active channel so that its cross-sectional capacity isdiminished. When channel capacity is sufficiently di-minished, the flow overtops this channel and cuts a newone. In the strict sense, the term applies only to thebreaking away and establishment of the new channel,although a precondition for avulsion is accretion duringdebris-flow phases of lahars.

E. Two Spectra of Downstream Behavior

Edifice-collapse-induced lahars are debris flows mattransfonn directly from avalanching debris mat containenough water to become water satUrated upon defonna-tion. Such lahars commonly contain large megaclastscomposed of relatively fresh volcanic rock from me edi-fice. As mey migrate downstream, mey incorporate ex-otic debris, especially at meir flow fronts (Fig. 4). Be-cause of meir size, avalanche-induced lahars [generally]rarely undergo complete transfonnation to hyper-concentrated flow as mey migrate downstream.

Lahars mat begin wim water floods can bulk up rap-idly to form debris flows or hyperconcentrated flows(Figs. lA and IB). If mey continue down active riverchannels, mey will gradually incorporate water, becomeprogressively dilute, and undergo transfonnations tohyperconcentrated flow and muddy streamflow (Figs.

610 LAHARS

lC and ID). The flow transformation begins at the flowfront and migrates back through the lahar wave as ittravels downstream. If lahars of this type occur in aridregions without perennial streamflow, they will bulk upto become debris flows, not undergo significant down-stream transformation, and remain debris flows to theirtermini. Such debris flows will typically develop rela-tively dry bouldery flow fronts that surround more lique-fied interiors.

IV. LAHAR DEPOSITS

FIGURE 6 Photograph showing hummocks on the surfaceof an avalanche-induced lahar deposit and. in the upper right.Mount Rainier. Hummocks in foreground are up to 40 m in planand 10-15 m high. Mount Rainier, 70 km upstream, is the sourceof the lahar deposit.

A. Characteristics of Deposits

Lahar size, origin, and depositional environment all in-fluence the character of deposits and determine the fa-cies that fonn. Clay-rich lahar deposits at Mount Rainierreflect deposition solely from debris-flow phases. Suchdeposits are massive, extremely poorly sorted, and com-monly nonnally graded (Fig. 5). The Osceola Mudflowat Mount Rainier, for example, has proximal and medialhummocky facies that are megaclast rich (Fig. 6). Proxi-mal and medial valley-side facies fonn thin (0.1-1 m)veneers on steep slopes, and proximal, medial, and distalaxial facies fonn thick (2-20 m) fills with common nor-mal grading in valley bottoms and lowlands. Clay-poorlahar deposits at Mount St. Helens have channel facies,floodplain facies, transition facies, and lahar-runout fa-cies (Figs. 7-10). The last three of these facies reflectdownstream dilution of the lahar. Changes in grain size,sorting, and grading are shown schematically in Fig. 7.

In medial reaches, some channel and floodplain faciesexhibit unusual basal deposits called sole layers (Fig.7). Sole-layer deposits are generally fines deficient andcontain abundant gravel-size particles that are broken bycataclasis. Unlike other deposits, sole layers apparentlyaccrete during the waxing-flow stages of lahars. Clay-poor lahar deposits not affected by downstream dilutionexhibit blunt snouts and prominent lateral levees thatcontain concentrations of large boulder- and cobble-size clasts (Fig. 11). Deposit interiors are more uniform,massive, poorly sorted, and matrix rich. Deposits lefton steep upland slopes typically form thin lags withconcentrations of the coarsest, densest particles and adeficiency of matrix.

Generally, lahar deposits may be massive to crudelystratified and graded to ungraded, depending on theproportion of water that the flows contained and thedegree of their downstream evolution. Inverse and nor-mal grading, or both in the same outcrop, are common.Sorting is generally extremely poor to poor. Althoughlahar deposits formed by debris flows, and the hyper-concentrated flows that commonly evolve from them,have some similarities, they have many differences too,and it is necessary to characterize each type of deposit

separately.Debris-flow deposits are massive and very poorly

sorted to extremely poorly sorted (greater than 2 <I> unitsand typically greater than 4 <I> units on the Wentworthscale). Grain-size distributions are commonly bimodal(Fig. 12). They may be normally or inversely gradedthroughout or, in some cases, can be inversely graded

FIGURE 5 Photograph showing normally graded debris-flowdeposit. Clay-rich normally graded deposits are common to col-lapse-induced lahars. A shovel, 60 cm long, is present for scale.

LAHARS 611

Flood plain facies Transition facies Lahar-runoutfacies

Lahar-relatedstreamflow

deposits

Channel facies

a) b)With lahar-cut With solepavement layer

Whaleback barSilt with pumiceand charred wood. . -. .0,

'" 0 0,,.. '0

I.'..'~.... ..,. t 'J " "."

, , . .. "

'. .:.' . ..' .'... -.

':i:][:~ igb

1:::.:.:::;:f..:.;::.:~:.;:::~11 .

I.:;:~';,:.~,?: ~:::.;.} 19b

~:,..:.~:~.

Qiidj

igbAngularity infine pebble class

igb

igb .1' igb

--::..~>

Underlvina lahar

Allllvi,lm

:.~.;..-II..." .:.. ..

:~'~~~';::;;:;::~i"f:"

;":": :":"::",::; .' "" :

."~.~":'-.::~i.':"- "-: I ;... :.:: ": ;":":: II

Sole 1:~.:.;.;"";""~~.:;i-PoorIY ,",~o.:JI ,!u:l~";" - --aver developed ii4!,..,1:J..,c"

- " sole layer

FIGURE 7 Schematic portrayal of facies types in lahars (clay-poor type) that undergo downstream dilution and transformation to

hyperconcentrated flow [from Scott (1988)].

near their bases and noffilally graded near their tops.Fabrics generally are weakly developed. Deposits areextremely compact or, in some cases, indurated so thatdigging in them is difficult. Particles found within lahardeposits can be monolithologic but are more commonly

heterolithologic; they can be rounded to angular, butprimary particles are usually subangular to angular. De-posits commonly exhibit vesicles in the matrix, whichresult from entrapment of air bubbles. Other commonconstituents include wood fragments, casts of wood frag-ments, and charcoal. Concentrations of coarse particles,especially low-density particles such as pumice, are com-mon at deposit tops.

Thicknesses of debris-flow deposits can vary fromtens of centimeters to tens of meters. Thick fill depositsoccur in valley bottoms and on lowlands. Deposits onhigher terraces and slopes within valleys are thinner

FIGURE 8 Photograph showing clay-poor debris-flow de-posit that is inversely graded at its base and both massive andungraded at its top (dashed line indicates base of deposit). Ashovel. 60 cm lon2. is present for scale.

FIGURE 9 Photograph showing faintly stratified hyper-concentrated-flow deposit and overlying transitional debris-flowdeposit. A lens cap is present for scale.

LAHARS612

B Osceola Mudflow, clay richA South Fork lahar, clay poor

FIGURE 10 Photograph showing hyperconcentrated-flowdeposit that is inversely graded in the lowest two-thirds of theoutcrop and normally graded in the upper third of the outcrop.Note that the deposit comprises silt, sand, granules, and smallpebbles with a single mode of coarse sand and granules. Depositis about I m thick.

than those in valley bottoms, and those on steep slopeswill drape underlying topography as thin veneers. Bothlevees and steep terminal flow fronts are common inthe deposits of debris flows relatively unaffected bydownstream dilution.

Hyperconcentrated-flow deposits have characteristicsintermediate between debris-flow and alluvial deposits(Figs. 7 and 12). They thus have intermediate sortingcoefficients (1-2<1> units) and grain size. They can bemassive (Fig. 10) but commonly have weak stratificationdefined by thin horizontal beds and very low angle crossbed sets composed of fine-grained laminae and thickercoarser grained beds (Fig. 9). Floodplain facies havef[;rain size in the f[;ranule, sand, and silt range, with the

FIGURE I I Photograph showing coarse-grained laterallevee and terminal snout of an experimental debris-flow deposit.Dimension of grid in foreground is I m.

LAHARS 613

FIGURE 14 Photograph showing dish structure in a hyper-concentrated-flow deposit. The structure evolves during or afterdeposition owing to dewatering during compaction. Note thatthe strctures are more strongly inflected toward the top of theunit. Tape measure for scale.

occasional floating pebble, cobble, or boulder (Fig. 13).If pumice was an important constituent of the flows,zones of nearly 100% pumice are commonly present atthe tops of overbank deposits. These pumice concentra-tions result from pumice rafts stranded during falling-stage hyperconcentrated flow. Channel-facies depositscommonly exhibit strong bimodality and clast support,with huge concentrations of cobbles and boulders sur-rounding the granule-sand-silt matrix (Fig. 13). Fabricsare moderately strong. Vesicles are sometimes presentbut less obvious than in debris-flow deposits. Depositsare compact and, except for bouldery channel examples(Fig. 13), have a "chippy" quality when dug with ashovel. Though rare, dewatering features such as dishstructures (Fig. 14) and pillar structures are some-times present.

Hyperconcentrated flow deposits have very flat topsand may vary from a few centimeters to several metersin thickness. Thicker deposits occur in channels or othernearby low areas. Thinner deposits occur on higherground such as floodplains and valley slopes. Flow topsexhibit scattered pebble and larger particles, especiallypumice if present; they also commonly have thin layersof fine sand and silt that form during compaction and de-watering.

Primary particles in lahar deposits derive from con-temporaneous eruptions or, in the case of avalanche-induced lahar deposits, from the original avalanche mass;secondary particles derive from the erosion and incorpo-ration of downstream volcaniclastic debris, alluvium,colluvium glacial drift, bedrock, etc. It is difficult torecognize the provenance of small particles, though itcan be done in some instances. Gravel-sized particlescan be readily divided into three groups. Angular volca-nic rocks derive from the volcano and are generallytaken to be primary. Angular secondary clasts derivefrom exotic colluvium, drift, bedrock, and any othererodible materials along the lahar path. Secondaryrounded clasts derive from exotic alluvium, fluvium, orglaciofluvial deposits. Using these criteria (or sometimesunique compositional or textural features of componentsof the clast population), one can calculate lahar-bulkingfactors. The lahar-bulking factor for a size fraction (orany other component) is defined as the proportion ofsecondary particles to secondary + primary particles inthat size class. The total bulking factor, which is thebulking factor for all size classes, can be difficult todetermine because of the difficulty of measuring lahar-bulking factors for small particles. If there is a size classor another distinctive component that is relativelyuniform in proximal deposits and is not present inbulked material, then apparent bulking factors of smallparticles and total bulking factors can be estimated.The estimated apparent bulking factor is given by the

equation

FIG U REI 3 Photograph showing transitional debris-flow tohyperconcentrated-flow deposits. The basal unit (at the point ofthe ice axe) contains floating. rounded cobbles and boulders andprobably reflects deposition from a hyperconcentrated flow ina channel. ABF = 1 - (Rf/R)(Sj/Sr),

614 LAHARS

deposits do. Mapping will distinguish between landslideand lahar deposits.

V. PHYSICAL MODELING

where R is the proportion of the reference component(or size class) not affected by bulking, S is the proportionof any other component (or size class) of interest, iindicates proximal (initial) value, and f indicates down-stream values. To calculate an ABF total, note that bothSj and Sf have values of 1 and

ABFtotal = 1 - (Rf/RJ.

Debulking is the selective deposition of a compo-nent-generally large or dense particles. If the bulkingfactor is negative, debulking may be indicated and adebulking factor can be calculated. The debulking factoris defined as the ratio of a component lost throughselective deposition to the initial amount of the compo-nent. Assessment of debulking factors of large particlesis straightforward. For small fractions, a debulking fac-tor can be calculated,

ADF = 1 - (Rj/Ru(Sf/SJ,

where R is the proportion of the reference componentnot affected by debulking and S is the proportion of anyother component (f and i are as before).

B. Distinguishing Lahar Deposits fromOther Common Diamictons

No one characteristic serves to distinguish lahar depositsfrom those of unwelded pyroclastic flows, glaciers, de-bris avalanches, and local landslides. Unlike lahar depos-its, unwelded pyroclastic-flow deposits do not have ma-trix vesicles, are not indurated, and contain mainlyjuvenile particles. Carbonized wood and magneticallyoriented clasts help to distinguish them from cold lahardeposits (the vast majority) but not necessarily from hotlahar deposits. Unwelded pyroclastic-flow deposits areusually looser and less compact than lahar deposits. De-bris-avalanche deposits generally have more irregularsurfaces than lahar deposits. Although lahar deposits,especially clay-rich ones, can have hummocks and laterallevees, these features are more prominent in debris-avalanche deposits. Distal and marginal parts of debris-avalanche deposits can be relatively flat and can containmatrix vesicles, but these features are more typical oflahar deposits. Till can be distinguished from lahar de-posits if lateral or terminal moraines can be identified.Unlike lahar deposits, till does not contain matrix vesi-cles, casts of wood fragments, or the wood fragmentsthemselves. Till is also more heterolithologic than mostlahar deposits. Landslide deposits generally have morelocal distribution and more uniform lithology than lahar

Diverse models have been proposed to explain the be-havior of lahars, but there is no convincing evidence thatlahars behave according to the assumptions of traditionalmodels. In the absence of evidence to the contrary, itmight therefore suffice to model these flows in a unifiedmanner as simple cohesionless grain flows moderatedby the influences of diverse grain sizes, sorting processes,varying pore pressure, and varying granular tempera-ture. Any viable model needs to account for significantrecent advances in the understanding of granular fluidmixtures. Such advances include recognition that(1) flows of such mixtures are neither steady nor uniformowing to abrupt initiations, flow instabilities, and parti-cle segregation processes; (2) pore pressures in suchflows are laterally heterogeneous; (3) the mobility ofsuch flows is often governed by highly fluid interiorswith nearly lithostatic pore pressures and resistantcoarse-grained perimeters with little or no pore pres-sure; and (4) there is a gradation from frictional to colli-sional particle interactions, dictated by the granular tem-

perature.The Bingham model was developed to describe the

behavior of clay slurries and is commonly invoked if arigid plug is observed or inferred. In its simplest formthe model assumes that resistance to flow is the resultof viscosity and strength of material:

where T is resisting shear stress, S is the yield strength,.u is viscosity, and u is velocity in the downstream direc-tion. The model assumes a viscous boundary layer andan overriding plug layer but neglects particle interac-tions such as collisions, frictional rubbing, or segrega-tion. Because cataclasis, rounding, and particle segrega-tion are important in those debris flows with significantgravel fractions (i.e., nearly all debris flows), the Bing-ham model is incomplete. A simple alternative hypothe-sis emphasizes particle interactions and explains rigidplugs as the result of particle locking owing to low gran-ular temperature.

A collisional or Bagnold-type model is commonly in-voked if inverse grading is observed and if grain colli-

LAHARS 615

sions are believed important, but Bagnold performedhis analysis for a mixture of uniform, neutrally buoyantparticles and fluid, and the application of his model togravity-driven debris flows therefore requires the relax-ation of his very restrictive assumptions. Further, highgranular temperatures that characterize collisional flowregimes inhibit inverse grading rather than promote it.The resisting stress owing to particle collisions in ashearing granular flow can be simply approximated,

Variation inPresults from the segregation and migra-tion of large particles to flow margins. Concentrationsof large particles at flow margins diminish permeabilitythere and thus preclude even short-duration pressuresgreater than hydrostatic. However, interior parts offlows contain abundant fine-grained particles thatgreatly decrease permeabilities to such small values thatpressures much greater than hydrostatic can be main-tained for times greater than the duration of debris flows.

Because lahars are unsteady and nonuniform, owingto particle segregation processes, and, in particular, ow-ingto probable variation in P and other parameters withposition in each lahar~ no simple mechanical model canbe successfully applied to these flows. Numerical analy-sis can approximate lateral and longitudinal variationsthat are obviously common in most lahars. A two-dimensional, depth-averaged model can include lateralheterogeneities and gives reasonable predictions of flowshape and runout for a wide variety of conditions. In-deed, such a model can qualitatively explain all observedfeatures of large-scale experimental flows described inIverson. In the future, modelers may be able to extendmodels of this type to three dimensions without toomany restrictive assumptions or too many troublesomeanalytical problems.

Tc - Vp,d2 (~dz

where II is the solids fraction, p, is the solids density, dis the characteristic particle diameter, and d2( du/ dz )2is proportion~l to the granular temperature, T. Someauthors combine Bingham and collisional (Bagnoldtype) models to derive a "unified" model. However, themechanics assumed in the two models are incompatiblebecause collisional models (like that of Bagnold) deriveultimately from particle interactions that the Binghammodel does not allow.

A few authors have suggested that lahars, or at leasttheir more dilute phases, can be modeled as gravity orturbidity currents. Models of this type presume verylow solids fractions ~d a consequent lack of particleinteractions. Since observations of deposits and eyewit-ness accounts of lahars refute both of these conditions,gravity-current models are wrongly applied to lahars.

A simple model, in which energy dissipation in theflow includes some combination of Coulomb frictionand collisional losses at higher granular temperatures,mediated by viscous pore fluid, appears adequate to ex-plain the behavior of many natural grain flows, includinglahars. The resisting stress owing to Coulomb frictioncan be given in simple form as

Debris Avalanches. Lahar and ]okulhaup Hazards.Pyroclastic Surges and Blasts. Subglacial Eruptions.Volcanic Lakes

Tf == v(Psgh - P)tan( c/J ),FURTHER READING

where g is the gravitational constant, h is the depth fromthe surface, tan(cf» is the friction coefficient, and P, thefluid pressure, is hydrostatic such that P - pfgh. In large-scale debris-flow experiments, however, Pfgh < P < p.ghis common and the flow is thus partly liquefied. (Notethat if P - Psgh, pressure is lithostatic and the flow is

completely liquefied such that it will move across eventhe gentlest gradient.) These experiments further showthat P varies with position such that the flow perimeteris commonly not liquefied (P = p£gh, or hydrostatic)

and the flow interior is approximately 80% liquefied(P = 0.8 Psgh, or close to lithostatic) (Iverson gives amore detailed discussion of this subject).

Ba~old, R. A. (1954). Experiments on a gravity-free disper-sion of large solid spheres in a Newtonian fluid under shear.Proc. R. Soc. London, SeT. A 225, 49-63.

Crandell, D. R. (1971). Postglacial lahars from Mount Rainiervolcano, Washington. U.S. Geo/. Surv. Prof Pap. 677.

Iverson, R. M. (1997). The physics of debris flows. Rev. Geo-phys. 35(3), 245-296.

Pierson, T. P., Janda, R. J., Thouret, J. C., and Borerro,C. A. (1990). Perturbation and melting of snow and iceby the 13 November 1985 eruption of Nevado del Ruiz,Colombia, and consequent mobilization, flow and deposi-tion of lahars. J. Vokano/. Geotherm. Res. 41, 17-66.

616 LAHARS

Smith, G. A., and Fritz, W.]., (1989). Volcanic influences onterrestrial sedimentation. Geology 17, 375-376.

Vallance,]. W., and Scott K. M. (1997). The Osceola mudflowfrom Mount Rainier: Sedimentology and hazard implica-tions of a huge clay-rich debris flow. Geol. Soc. Am. Bull.109, 143,-163.

Rodolfo, K S. (1989). Origin and early evolution of laharchannel at Mabinit, Mayon volcano, Philippines. Geol. Soc.Am. Bull. 101,414-426.

Scott, K M. (1988). Origins, behavior, and sedimentology oflahars and lahar-runout flows in the Toutle-Cowlitz Riversystem.. U.S. Geol. Surv. Prof Pap. 1447-A.