Bull Volcanol (1998) 59 :233–244 Q Springer-Verlag 1998

ORIGINAL PAPER

P. A. Mothes 7 M. L. Hall 7 R. J. Janda

The enormous Chillos Valley Lahar: an ash-flow-generated debris flow

from Cotopaxi Volcano, Ecuador

Received: 28 April 1997 / Accepted: 19 August 1997

Editorial responsibility: D. A. Swanson

Patricia A. Mothes (Y) 7 Minard L. HallInstituto Geofísico, Escuela Politécnica Nacional, QuitoEcuadorFax: c593 2 567847e-mail: instgeof6uio.satnet.net

Richard J. Janda†U.S. Geological Survey, Vancouver, Washington, USA

Abstract The Chillos Valley Lahar (CVL), the largestHolocene debris flow in area and volume as yet recog-nized in the northern Andes, formed on Cotopaxi vol-cano’s north and northeast slopes and descended riversystems that took it 326 km north–northwest to the Pa-cific Ocean and 130c km east into the Amazon basin.In the Chillos Valley, 40 km downstream from the vol-cano, depths of 80–160 m and valley cross sections upto 337000 m2 are observed, implying peak flow dis-charges of 2.6–6.0 million m3/s. The overall volume ofthe CVL is estimated to be ;3.8 km3. The CVL wasgenerated approximately 4500 years BP by a rhyoliticash flow that followed a small sector collapse on thenorth and northeast sides of Cotopaxi, which meltedpart of the volcano’s icecap and transformed rapidlyinto the debris flow. The ash flow and resulting CVLhave identical components, except for foreign frag-ments picked up along the flow path. Juvenile materi-als, including vitric ash, crystals, and pumice, comprise80–90% of the lahar’s deposit, whereas rhyolitic, dacit-ic, and andesitic lithics make up the remainder. Thesand-size fraction and the 2- to 10-mm fraction togetherdominate the deposit, constituting ;63 and ;15 wt.%of the matrix, respectively, whereas the silt-size fractionaverages less than ;10 wt.% and the clay-size fractionless than 0.5 wt.%. Along the 326-km runout, these par-ticle-size fractions vary little, as does the sorting coeffi-cient (averagep2.6). There is no tendency towardgrading or improved sorting. Limited bulking is recog-nized. The CVL was an enormous non-cohesive debrisflow, notable for its ash-flow origin and immense vol-ume and peak discharge which gave it characteristics

and a behavior akin to large cohesive mudflows. Signif-icantly, then, ash-flow-generated debris flows can alsoachieve large volumes and cover great areas; thus, theycan conceivably affect large populated regions far fromtheir source. Especially dangerous, therefore, are snow-clad volcanoes with recent silicic ash-flow histories suchas those found in the Andes and Alaska.

Key words Debris flow 7 Pyroclastic flow 7 Cotopaxi 7Ecuador 7 North Andes

Introduction

The eruptions of Mount St Helens (MSH) in 1980, Ne-vado del Ruiz (NR) in 1985, Redoubt in 1989–1990 andPinatubo in 1991 resulted in many in-depth studies ofdebris flows (MSH and Redoubt: Janda et al. 1981,Pierson 1985, Scott 1988, Waitt et al. 1983, Trabant etal. 1994; NR, Colombia: Pierson et al. 1990; Pinatubo:Pierson et al. 1992). These flows had matrices whichwere mainly fines-poor and non-cohesive, importantfactors which contributed to their miscibility with riverwater and eventually aided their transformation to hy-perconcentrated streamflows (Scott 1988).

Much less common are debris flows that have a co-hesive matrix with clay-size particle contents greaterthan 3–5% and that travel great distances without sig-nificant textural variation and with little attenuation oftheir peak discharges (Scott et al. 1995). Lahars of thisnature are exemplified by the 3.8 km3 Osceola mud-flow, generated some 5600 years BP by a sector col-lapse of Mount Rainier (Crandell 1971; Vallance andScott 1997), which experienced few textural changesalong its runout. Given the large discharges that wereapparently sustained for great distances by these flows,their generation poses significant hazard for populationcenters downstream.

Debris flow formation requires (a) a significant wa-ter source (ice/snow cover, intense rainfall, crater lake,etc.), (b) abundant unconsolidated material (such astephra, pyroclastic flow deposits, glacial outwash), (c)

234

great relief and steep slopes on the cone, and (d) a trig-gering mechanism (Major and Newhall 1989). The vol-ume of the flow is influenced by the above factors andby the nature of the eruptive activity. For example,fluidized pyroclastic flows of silicic composition,charged with fine-ash particles, are capable of efficientheat transfer as well as vigorous erosion and intermix-ing with ice and snow, thereby accelerating the conver-sion of ice to water (Pierson et al. 1990; Pierson1995).

At Cotopaxi Volcano, the debris flows with thelargest volumes are those that formed by interactionbetween silicic pyroclastic flows and glaciers. This arti-cle describes one of the largest Holocene lahars yet re-ported, formed by the instantaneous melting of Coto-paxi’s glacial cap by a pyroclastic flow of rhyolitic com-position. This – the Chillos Valley Lahar (CVL) – hadan ash-rich matrix which allowed it to travel 326 kmfrom its source without undergoing significant texturaltransformation, a characteristic of cohesive debrisflows. Similar rheological behavior occurred in the Os-ceola mudflow, which flowed 120 km without importanttextural changes (Crandell 1971), attributed mainly toits high clay content. The CVL, however, having fewclay-size particles and yet sharing many of Osceola’scharacteristics, probably owes its behavior to its im-mense volume.

Cotopaxi Volcano

Cotopaxi Volcano (5890 m), a large symmetrical conewith 3000 m of relief, is one of the principal stratovolca-noes of Ecuador’s eastern volcanic row, overlying a me-tamorphic basement of approximately 3000 m elevation(Fig. 1). Its upper glacier-clad flanks have slopes aver-aging 25–307, whereas its lower flanks are incised by can-yons tens of meters deep. Its snow and ice fields coverapproximately 20 km2 with a volume of 1.0 km3 (Jor-dan 1983). Like elsewhere in the Andes, these glaciersare presently in retreat.

Three important drainages head on the volcano: theRios Pita-Esmeraldas to the north, the Rio Cutuchi-Pastaza system to the southwest, and the Rio Tambo-Tamboyacu-Napo system to the east (Fig. 2). The lattertwo drainages ultimately descend the eastern flank ofthe Andes to join the greater Rio Amazon system. TheRio Pita flows northward approximately 90 km throughthe Inter-Andean Valley, then descends the westernflank of the Andes and subsequently enters the PacificOcean at the city of Esmeraldas, after a 326-km jour-ney. To the southwest, the Rio Cutuchi flows south-ward down the Inter-Andean Valley for 130 km andthen eastward through the Cordillera Oriental to be-come an Amazon tributary.

The Rios Pita and Cutuchi both flow through the 50-km-wide Inter-Andean Valley, home to approximately3 million inhabitants and numerous towns and cities,some of which have been severely affected by Coto-

Fig. 1 Cotopaxi’s north flank and possible avalanche scar thatheads at the rock face near its 5890 m summit. Relief is 2100 mover its 3800-m-high base. In middle ground are debris flow chan-nels occupied by historic lahar deposits

Fig. 2 Partial map of Ecuador showing the Inter-Andean Valley,bordered by the Cordillera Occidental and the Cordillera Orien-tal, and occupied by Cotopaxi Volcano and the adjacent Chillosand Latacunga valleys. The Río Pita-San Pedro-Guayllabamba-Esmeraldas river system enters the Pacific Ocean near the city ofEsmeraldas. The Ríos Napo and Pastaza join the Amazon Riverdownstream

paxi’s lahars on several occasions since the arrival ofthe Spanish in 1532 (Mothes 1992).

Cotopaxi has had approximately 30 eruptions since1532 (Hall 1977). These eruptions were typically ac-companied by regional scoria and pumice ash falls,

235

Fig. 3 Map of Chillos Valleyand the Río Pita-San Pedroriver system. Limits of theChillos Valley Lahar (CVL)deposit are shown, as well asrepresentative avalanche hum-mocks and the Canon Col-orado ash-flow deposit. Notethe selected channel cross sec-tions of the CVL unit. Samplelocations refer to granulomet-ric analyses

blocky lava flows, scoria pyroclastic flows, and lahars,all of andesitic character (SiO2p56–58%). However,during the past 10000 years, rhyolitic eruptions haveperiodically taken place in its predominantly andesiticdevelopment. At least four such eruptive periods haveoccurred at intervals of approximately 2000 years, pro-ducing moderate-size ash flows (Hall and Mothes1995).

Around Cotopaxi’s base several cream-colored, ash-rich debris-flow deposits are recognized that are prima-rily comprised of fine-grained rhyolitic ash and are not-able for their large volumes and areas. They can bereadily traced upslope, where they become indistingui-shable from rhyolitic ash flows. The largest of these de-posits in Cotopaxi’s recent past is the Chillos ValleyLahar, formed when the north and northeast sides ofthe cone suffered a sector collapse and moderate-size

ash flows were generated. This debris flow descendedprimarily the Rio Pita passing through the Chillos Val-ley (Fig. 3), but it also gained access to the northern tri-butaries of the Rio Cutuchi and descended that drai-nage for at least 130 km (Fig. 4).

Characteristics of the Chillos Valley Lahar

The CVL’s passage through Ecuador’s Inter-AndeanValley was probably one of the most catastrophic geo-logic events to have occurred during the Holocene. Theextensive area that the lahar covered, its huge volume,its beige color and unique composition, and the vol-canic events associated with its generation distinguish itfrom other Cotopaxi debris flows. In the Chillos Valley,located 20 km southeast and 400 m lower than Quito,

236

Fig. 4 Map of Latacunga Valley and the major streams drainingthe southwest side of Cotopaxi Volcano to form the Río Cutuchi.Limits of the CVL deposit are shown. Neither avalanche nor ash-flow deposits are found on the west side of the volcano. Samplelocations refer to granulometric analyses

the flow’s path was more than 11 km wide and had adepth of 150 m. This prehistoric event also had its hu-man toll, although of unknown proportions, as sherdsof prehistoric pottery are found at the base of the flowor in the underlying soil.

Flow routes and area

The CVL inundated areas substantially greater thanthe narrow valleys occupied by more recent, smaller la-hars generated at Cotopaxi. From historical accountsand stratigraphic studies, it is known that the majorityof the small debris flows since 1532 flowed down thewest and southwest tributaries of the Rio Cutuchi,

whereas few followed the northern drainages. Howev-er, in the case of the CVL, its larger portion descendedCotopaxi’s northern rivers, where it is seen in cuts inthe broad plain at the volcano’s northern base (Limpio-pungo), downvalley along the Rios San Pedro and Pita,and in the Chillos Valley. The CVL also rode up ontothe lower flanks of both Ruminahui and Sincholaguavolcanoes. To the west its route was restricted to theRios San Lorenzo and Cutuchi. All channels have highgradients (4–5%) for the first 25–40 km and lower gra-dients (1–2%) thereafter, until they descend the Andes’steep flanks (Fig. 5). As compared with the heights andtypical stream gradients of other major lahar-generat-ing volcanoes, Cotopaxi is distinguished by its extremerelief (;6000 m) and two steep descents to reach sealevel, that of the cone itself (;3500 m drop) and that ofthe Andes’ western flank (;2500 m drop).

Along the northern flow route, approximately 60%of the discharge was diverted westward to the Rio SanPedro by Pasochoa Volcano (A-A in Fig. 3). The re-maining ;40% of the CVL continued down the RioPita channel, where the flow was constricted and at-tained depths up to 200 m (B-B in Fig. 3). Upon leavingthe confines of the San Pedro and Pita canyons, withreaches of approximately 25 and 15 km, respectively,the debris flow spread laterally into the Chillos Valley.Because of its shorter path, the Rio Pita flow probablyarrived first in the Chillos Valley, possibly creating atemporary barrier for the San Pedro flow which arrivedslightly later.

Once the two flows merged, a formidable cross-sec-tional width of 11 km was attained. Farther downvalleynear Conocoto the width diminished to approximately6 km, and well-defined limits show that depths greaterthan 90 m were achieved (D-D in Fig. 3). For the next12 km downstream, cross-sectional widths first de-creased as the flow was confined to the San Pedro can-yon and then increased as it spread into the Cumbayá-Tumbaco Valley, where depths up to 120 m are ob-served (E-E in Fig. 3). To this point the CVL had cov-ered approximately 270 km2.

Downstream from Cumbayá the CVL followed thenarrow and steep-walled canyon of the Rio Guaylla-bamba for 186 km, along which gradients range from1–6% (Fig. 5). Little deposition occurred along this sec-tion due to the high-energy environment. However, theCVL deposit is found interbedded with fluvial sedi-ments over the last 75 km of the Rio Esmeraldas flood-plain until it reaches the Pacific Ocean (Fig. 2). Thus,this debris flow traveled 326 km and left a deposit, av-eraging 2 m thick, over 440 km2. Its submarine exten-sion onto the offshore platform is unknown.

The CVL deposit is readily identified in outcropswest of Cotopaxi, thanks to its high stratigraphic posi-tion and exceptional lithologic and textural characteris-tics. Upon leaving the high-gradient narrow canyons ofthe Rios Cutuchi and San Lorenzo, the flow spreaddownstream to widths of 3–5 km for the next 25 kmalong the Latacunga Valley (Fig. 4). The flow heights

237

Fig. 5 Longitudinal profiles of the two principal river channelsdraining Cotopaxi Volcano. Note horizontal scale change. Riverprofiles at Nevado del Ruiz, Mount St. Helens, and Redoubt vol-canoes are shown for comparison. (Data from Pierson et al. 1990,Pierson 1985, and Trabant et al. 1994, respectively)

did not exceed 20 m above the valley bottom; thus,more recent lahars often cover the CVL and conse-quently outcrops are few. Where channel gradients are0.5% or less, deposits tend to have thicknesses of2–3 m. Between Latacunga and Panzaleo, a distance of17 km, the flow was constricted to a 2- to 3-km-widepath and was 35–40 m deep. South of Panzaleo the flowentered a steep-walled canyon, in which few depositsare preserved. Approximately 160 km2 were flooded bythe debris flow in the Latacunga Valley downstream toPanzaleo.

Volume and discharge of the CVL

Good preservation of the CVL deposit allows its vol-ume to be estimated. Profiles based upon well-definedstrandlines left by the flow at three sites in the Chillosand Cumbayá valleys give cross-sectional areas of337000 m2 at Conocoto (45 km from the summit),130000 m2 at Guangopolo (50 km from the summit),and 138000 m2 at Cumbayá (62 km from the summit).For the Latacunga Valley, areas of 102000 m2 at Lata-cunga and 71000 m2 at Salcedo are calculated (Figs. 3,4).

Field evidence for the velocity of the CVL is lacking.Velocity estimates for Cotopaxi’s clast-rich 1877 debrisflow obtained from runup calculations and numericalmodeling give minimum speeds of ;5–7 m/s in theChillos Valley (Castro et al. 1992). However, large-vol-ume debris flows are thought to have substantiallygreater velocities. An empirical study of the traveltime–distance relationships for debris flows of varyingsize suggests that flows with peak discharges61!106 m3/s probably travel several orders of magni-

tude faster than medium-size flows (Pierson, in press).The large cohesive Osceola mudflow is estimated tohave had a velocity of ;19 m/s at 40–50 km down-stream from its source (Vallance and Scott 1997). If theCVL had a velocity similar to that of the Osceola Mud-flow, a debris flow comparable in size and behavior tothe CVL, peak flow discharges in the Chillos Valleycould have been of the order of 2.6–6.0!106 m3/s, val-ues far exceeding the daily discharge of the presentAmazon River.

Along the northern drainage the CVL deposit has aminimum total dry volume of approximately 0.90 km3,which is calculated from an average 2-m-thick depositthat mantles 440 km2. Along the southwest drainage, aminimum total dry volume of 0.32 km3 is estimated. Weassume that the CVL was composed of ;30% waterand ;70% sediment by volume – the approximate ra-tio reported for lahars at Mount St. Helens (Piersonand Scott 1985) and other debris flows (Sharp and No-bles 1953; Pierson 1980), as well as by our experimentsthat show that a 28% volume of water is needed to ini-tiate mobility of the dry CVL material. As such, theCVL’s water-saturated volume would have been atleast 1.7 km3, a minimum value which does not consid-er the volume that entered the sea and the Amazon ba-sin.

As discussed later, the CVL is thought to have trans-formed completely from an ash flow, whose volume isestimated at 2.5 km3. This implies a total wet volume ofapproximately 3.8 km3, which does not consider thevolume of material picked up along the runouts.

Textural and compositional characteristics

The CVL deposit is readily identified by its beige-tancolor, its homogeneous appearance, and the fact that itis often the topmost unit and mantles the countryside(Fig. 6). Equally important, in the field it appears toconsist of ;80–90% matrix, composed of ash, pumice,and lithic grains, as well as ;10–20% lithic clasts larger

238

Fig. 6 Typical CVL outcrop, here overlying a blocky-structuredsoil. Note the CVL’s matrix-rich character, homogeneous distri-bution of clasts, and absence of any stratification. Taken at SanRafael, 45 km downstream from Cotopaxi

Table 1 Size distribution of clasts

Sample Location Distancedownstream (km)

Area ofoutcrop (m2)

Clast count of area %)

1128 mm 64–128 mm 32–64 mm 10–32 mm Total

Clast/m2 Matrix/m2

Northern route1A1B6

11121314

PF-campsiteCVL-campsitePedregalSelva AlegreConocotoTumbacoViche

77

20424865

270

1.700.803.903.743.902.704.06

0.0015.096.162.145.138.87

61.08

1.096.96

12.314.933.316.133.63

0.542.607.205.475.962.811.59

6.1622.812.975.26

11.2713.000.28

7.7947.4628.6417.8035.6730.8166.58

92.2152.5471.3682.2074.3369.1933.42

than 1–2 cm. In a few places dense boulders up to 2 min diameter occur. Throughout the Chillos and Tumba-co valleys, and even along the Rio Esmeraldas in thecoastal zone, the deposit has a uniform thickness of ap-proximately 2 m. Thicknesses of 3–4 m are attained inpaleochannels. Lateral pinching out of the deposit oc-curs at the flow margins.

The CVL deposit is distinguished from the ash flowby containing significantly more lithic clasts (up to 30 vs8% by area), carrying rounded pebbles and cobbles of

dense rocks, having a greater variety of polylithic frag-ments many of which are accidental, as well as having aharder, more compacted matrix.

No internal stratification and only rare segregationof larger clasts are observed. Furthermore, there is noevidence of individual flow pulses or basal sole layersoften reported in debris flow deposits. The flow waserosive, as shown by the occasional incorporation ofsoil clots, rounded stream cobbles, and the presence oflinear casts of the CVL’s basal material correspondingto grooves cut into the underlying soil by draggedstones. The CVL matrix contains small microvesicles,remnants of entrapped air and gas in the original ma-trix (Fisher and Schmincke 1984).

Clast and matrix characteristics were determined forrepresentative samples of the deposit along both thenorthern and southern flow paths. Clast parameterswere obtained by simple counting and classification asto size and type of fragments directly on the outcropface. As seen in Table 1 the population of clasts greaterthan 10 mm in diameter accounts for ;18–67% of thesurface area, averaging ;25–30%. Thus, the matrix(^10-mm-size grains) varies from ;33–82%, averag-ing ;70–80% of the surface area. In the ash-flow tuffassociated with the CVL, the clasts (610 mm) and ma-trix comprise 8 and 92% of the outcrop face, respec-tively.

Granulometric analyses of the ^64 mm size fractionof the deposits were carried out by routine sieve andhydrometer methods (Table 2). Sand-size grains(0.063–2 mm) greatly dominate the matrix (47–79 wt.%;Fig. 7). The 2- to 10-mm size fraction varies from8–44 wt.%, whereas the ^0.063-mm size fractionranges from 1–21 wt.%. The clay-size particle contentof five samples as determined by hydrometer analysisremains at only 0.24–0.41% of the matrix.

Noteworthy is the consistent size distribution withdistance, even over the exceptionally long runout of theCVL (Figs. 7, 8). The sand-size fraction of most sam-ples has a limited range of 59–67%, similar to the 61%shown by the ash flow (sample 1; Table 2). The 2- to10-mm fraction also shows a limited variation (8–19%),with few exceptions. Some samples (nos. 2 and 6) have

239

Table 2 Granulometric analysis: Chillos Valley Lahar

Sampleno.

Location Distancefromcrater (km)

Unitthickness(m)

Positionindeposits

Content (wt.%) Mean GrainSize (Mz) Sorting

Gravel64–10mm

2–10mm

Sand Siltandclay

Clay phi mm oG

N. Route, Chillos Valley123456789

1011121314

Campsite PFReservoir AvBulldozerN. Park entrancePedregalPedregal/AvLa CalderaMachachiHushpuppyChillo JijonSelva AlegreConocotoTumbacoViche

79

1114182023303341424865

270

10.05.02.02.52.06.03.02.52.02.51.53.03.02.0

t/ccccbbccccct/cct/c

1.47.82.35.26.9

14.75.4

14.013.717.03.81.87.35.4

17.543.914.826.07.8

35.710.018.619.09.0

19.211.813.519.1

61.046.961.560.278.548.276.661.164.559.063.675.867.565.4

20.11.4

21.48.66.41.47.86.32.8

15.013.010.511.49.8

0.41

0.24

0.40

0.350.31

1.65P0.40

1.550.371.03

P0.831.300.360.030.671.202.001.101.07

0.321.350.340.800.501.800.410.820.990.650.460.250.480.48

2.82.52.62.82.02.62.32.92.83.72.72.42.72.8

S. Route, Latacunga Valley15161718

Cutuchi PfCutuchi CVLLatacungaNagsiche

15154356

3.03.02.04.0

cct/ct/c

4.35.96.52.3

30.113.215.513.8

65.167.166.071.9

0.513.812.012.0

0.000.871.031.37

1.000.550.490.40

2.02.12.72.3

a greater amount of coarser material, reflecting theiravalanche debris contents. The mean grain size for allsamples is 0.70 mm with a range of 0.25–1.80 mm(Fig. 9A). As compared with the cohesive Osceola andElectron debris flows (Fig. 7), the CVL deposit is clear-ly dominated by the sand-size fraction and containsfewer coarse-grained clasts as well as less silt and clay-size material (Fig. 9C). Its sorting coefficients (sG)have a mean value of 2.6 within a variation of 2.0–3.7(Fig. 9B); the CVL is slightly better sorted than the oth-ers. The CVL deposit, like the Osceola deposit, appearsto preserve much of the granulometric character of theoriginal source material, despite its transformationfrom a pyroclastic flow and the great distance trav-eled.

The clay-size fraction of the CVL is negligible com-pared with that of debris flows formed by collapse ofhydrothermally altered edifices, such as at Mount Rain-ier (Scott et al 1995; Vallance and Scott 1997) or at Cit-laltépetl volcano, Mexico (Carrasco-Núnez et al. 1993).Cotopaxi shows no evidence of intense alteration in itsrecent history. In fact, the CVL material is composedalmost exclusively of fresh, juvenile pyroclastic prod-ucts. Even compared with non-cohesive lahars, such asthose of Pine Creek at Mount St. Helens and of the RioAzufrado at Nevado del Ruiz, the CVL deposit con-tains fewer clay-size particles. In Fig. 10 cumulativegrain size frequency curves for the CVL are comparedwith curves for the cohesive Electron and Osceola la-hars and the non-cohesive Molino (NR) lahar; thecurves again emphasize the dominant sand-size frac-tion, the scarce silt-and-finer fraction, and the slightlybetter sorting of the CVL deposit.

Nine major clast types are common in the CVL unit:gray rhyolitic vitrophyre, red- and black-banded rhyol-ite, perlitic obsidian, gray andesitic blocks, white micro-vesicular biotite-bearing pumice, scoria lapilli andbombs, rounded stream cobbles, clods of the underly-ing soil and sediments, and uncommon pods of gray an-desitic and rhyolitic breccia.

The obsidian fragments, gray rhyolite, pumice, andbanded rhyolite are identical to the lithic clasts in thepyroclastic flows and tephra falls of the associated Can-on Colorado ash flow sequence discussed below. Thebreccias and andesitic blocks formed the pre-eruptioncone and were caught up in the lahar. The rhyolite, ob-sidian, and pods of rhyolitic breccia are remnants ofpre-existing domes. Rounded oxidized cobbles are fre-quently found and were incorporated from the streambed. Clods of soil and indurated tuff and pods of loosepumice, ranging from 1 cm to 2 m in diameter, are ob-served in the deposits but amount to ^1% of the vol-ume.

Along the entire 326 km course of the CVL, exceptin a few areas, there is no tendency toward grading orimproved sorting (Fig. 9B). Dense lithic clasts up to 2 min size are found in the deposit in a few places andshow no tendency to concentrate at the base or the topof the flow. This aspect reflects the high-yield strengthof the flow along its entire course, a characteristic ofwell-developed debris flows (T. C. Pierson, pers. com-mun.).

That the CVL unit maintains its integrity with dis-tance is noteworthy, given its long path down high-gra-dient channels, including the Rio Guayllabamba can-yon, where one would expect mixing and dilution with

240

Fig. 7 Grain-size distribution of the CVL matrix (particles^64 mm) along northern runout. Similar graphs are shown forthe Osceola and Electron debris flows from Mount Rainier Vol-cano. Data from Crandell (1971). Note the dominance of fine-grained material in the CVL unit

Fig. 8 Cumulative grain-size frequency curves for CVL samples(^64 mm) along the northern runout for flow margin and flowaxis samples. The curves are almost identical in shape and posi-tion for the entire runout, with the exception of the Conocotosample which is from the extreme western margin of the flow

river water and transformation to a hyperconcentratedflow. However, there is no separation or segregation ofcoarser or denser clasts, and the matrix that arrived atthe Pacific Ocean is seemingly identical to the initial fa-cies; these two characteristics strongly imply that littleinteraction with river water took place. This may havebeen due to the CVL’s limited opportunity to mix withriver water given its presumed viscous nature and rapidpassage. Scott et al. (1995) found that a sand-rich ma-trix tends to retard dilution of a flow by stream water aswell as the settling out of coarse clasts. Additionally,the flow front may not have been greatly influenced bystream and river water encountered along the runout,which presently constitutes a relatively minor quantityas compared with the CVL’s immense volume.

Other aspects of the CVL’s unusual character

Electro-chemical attractions inherent to clay mineralparticles have been cited to explain the cohesiveness offine-grain mudflows (Qian et al. 1980; Rodine 1974;Major and Pierson 1992). The important yield strengthof the CVL during its long journey cannot, however, beattributed to clay minerals, since their quantity is negli-gible (Table 2). Here it is suggested that the high con-centration of irregularly shaped particles of volcanicglass resulted in a relatively tight packing of the CVL’smatrix that would have inhibited both mixing with wa-ter and the winnowing of the fine fraction, as well ascontributed to the CVL’s notable yield strength.

Bulking was a factor in the CVL’s behavior. Asmentioned, pods of soil and pre-existing tuff are scat-tered throughout the deposit. A plot of clast type vsdistance (Fig. 11) shows the relative proportions of for-eign fragments incorporated from the stream bed ascompared with andesite, rhyolite, and pumice clastscarried from Cotopaxi. Andesite and other rock types

241

Fig. 9 Downstream variation in A mean grain size [graphic mean(Mz) in mm], B sorting coefficient [graphic standard deviation(sG)], and C percentage of silt and clay-size particles in matrixalong the northern runout (in wt.%). In Fig. 9A the relativelyconstant trend is perturbed by samples 2 and 6 which carry coars-er avalanche material. Grain-size parameters from Folk (1980)

Fig. 10 Cumulative grain-size frequency curves for three CVLsamples taken 33, 42, and 270 km from source as compared withthe Electron and Osceola debris flows, Mount Rainier (data fromCrandell 1971), and the Molinos lahar, Nevado del Ruiz (datafrom Pierson et al 1990). The CVL unit demonstrates slightly bet-ter sorting than the other flows, reflecting homogeneity of origi-nal ash-flow material

Fig. 11 Percentage of the four principal clast types that comprisethe clast population (Table 1) found in the CVL deposit vs dis-tance from source. All samples from the Río Pita-Esmeraldas ru-nout. Size of clasts 610 mm. Pum/Scor Pumice and scoria; Rhy/Dac rhyolite and dacite

were picked up as the flow descended the steep can-yons leaving the Andes and were then deposited uponthe upper coastal fan. The loss of pumice clasts wasprobably due to abrasion rather than winnowing, giventheir fast attrition rates and the long flow path. Thelarge amount of incorporated rhyolitic material ob-served at the 20-km site is probably related to nearbyavalanche hummocks.

Flow bulking is also implied by a plot of the numberof rounded and subrounded, dense (non-pumice) clastsper square meter of outcrop vs distance, although aclear progressive increase with distance is not indicated.Figure 12 shows a relatively constant number of theseclasts with distance, especially for the 632 mm size.

Apparently, many clasts settled out while crossing thewide coastal plain. The sharp increase of rounded, 8- to32-mm-size clasts observed at the 55-km site (Fig. 12) isapparently related to its close proximity to the mainchannel.

Origin of the Chillos Valley Lahar

The sector collapse and pyroclastic flow events

In Canon Colorado at the northern foot of Cotopaxi,three sequential pumice-rich rhyolitic ash flow units areexposed, which are bounded upward and downward

242

Fig. 12 Number of rounded and subrounded dense clasts persquare meter of CVL outcrop vs distance from source. All sam-ples from the Río Pita-Esmeraldas runout. The large number ofrounded clasts observed at the 55-km site may be related to itsproximity to the main channel

Fig. 13 Interrelations of the CCPF and the overriding avalanchedebris followed by the CVL. Photo taken at the northern base ofthe cone, west of sample site 2 (see Fig. 3)

stratigraphically by typical andesitic deposits of thisvolcano (Hall and Mothes 1995). A few kilometersdownvalley the deposits of the larger ash flows, herecalled the Canon Colorado pyroclastic flows (CCPF Iand II), are associated with clast-rich hummocks andoverlain intimately by the CVL unit (Fig. 13). At theseoutcrops and others around the northeast foot of thecone, the avalanche deposits that make up these hum-mocks are composed primarily of angular clasts of ob-sidian, banded rhyolite, gray aphyric rhyolite, and da-cite in a sandy matrix of the same lithologies, althougholder andesitic clasts are also found. They are attri-buted to a small sector collapse of the upper north andnortheast flank of the cone which included older rhyol-itic domes, remnants of which are presumed to underliethe young tephra cover of the volcano. The avalanchedeposit is found distributed around the north andnortheast foot of the cone and downvalley for 25 km,which suggests an overall volume of approximately2 km3.

Field mapping shows the lower ash flow deposit tobe the most widespread and voluminous of the units.The main trajectory of this flow was northward andnortheastward, where it covered the Limpiopungoplain and the Rio Pita Valley with a deposit tens of me-ters thick (Fig. 3). We estimate that the deposit oncecovered ;170 km2 and had a volume of ;2.5 km3.

Both the CCPF-I and -II units are matrix-rich ashflow tuffs containing ;92% vitric ash, crystals, and pu-mice particles that comprise the matrix (^10 mm), aswell as ;8% clasts (610 mm) of microvesicular pu-mice, obsidian, and black, red, and gray-banded aphyricrhyolite. The pumice typically contains 1–2% gold andblack biotite, 5–10% quartz and plagioclase, 2% mag-netite, and a trace of hyperstene and K-feldspar. Boththe mineralogy of the pumice and the lithic clasts of theCCPF are identical to those found in the CVL deposit.The SiO2 contents of the pumice from these units are71–73% (M. L. Hall and P. Mothes, unpublished data).

Consequently, the overall composition of the CVL’smatrix (i.e., 80–90% vitric ash, pumice, and crystals,and 10–20% of small fragments of dense pumice andrhyolite) accords well with that of the CCPF.

The deposits of the Chillos Valley Lahar lie in inti-mate association with those of the hummocks and theCCPF; the CVL incorporated pods of loose avalanchebreccia and of the lower and middle ash flows. At manyoutcrops varied and reversed stratigraphic relations ex-ist between the CVL, the ash flows, and the avalanchebreccia, suggesting that all three events were approxi-mately simultaneous.

Sources of water

Glacial studies in the Ecuadorian Andes suggest that atapproximately 5000 years BP glacier limits on volca-noes were 300–500 m lower than their present limits of4800 m elevation (Clapperton 1990). Thus, at the timeof the CVL’s formation the glacial ice cap on the north,northeast, and northwest quadrants of Cotopaxi’s conemay have had a total area of approximately 20B5 km2,corresponding to a volume of approximately 1.0–1.3km3 (assuming an ice thickness of 30–50 m). The maxi-mum amount of water available, if all of this ice melted,would have been approximately 0.8–1 km3, sufficient tohave mobilized approximately 2 km3 of dry debris.

Additional sources of water include ground waterfrom the volcano, possible moraine-dammed lakes, andriver water. Undoubtedly, the pre-collapse cone con-tained substantial water, for it was glacier clad. If the

243

Fig. 14 Probable sequence ofevents leading to the forma-tion of the CVL. Downvalleytransit is to the north

volume of the sector collapse was of the order of 2 km3,as postulated previously, and contained approximately5% ground water, an additional 0.1 km3 of water wouldhave been available. Small moraine-dammed lakes arenot uncommon on present-day volcanoes and couldhave contributed water. However, given that morainaldebris is not observed in the proximal CVL deposits,this source is discounted. As discussed previously, theamount of water encountered downstream along therunout was probably not large (estimated at0.0075 km3) as compared with the immense volume ofthe CVL, especially for the first 250 km. Thus, riversprobably contributed little to the overall water re-quired.

Age of the CVL

The lahar’s age is given by several radiocarbon dates.The soil beneath the Canon Colorado ash flows on Co-topaxi’s north flank is dated at 4670B70 14C years BP(Smyth 1991). Tree trunks taken from the CVL depositon the coastal plain give two dates of 4500 14C years BP(N. Banks, pers. commun.). Lastly, pottery and figu-rines of black coral found directly beneath the unit areassigned a mid-Valdivian age, thought to be approxi-mately 4500 years old (P. Norton, pers. commun.).Based on these dates as well as on the logical succes-sion of younger dates for the overlying deposits, weconclude that the CVL was most likely generated be-tween 4500 and 4600 years ago (uncorrected).

Conclusions

The Chillos Valley Lahar, a debris flow of immenseproportions, was generated in a sequence of events thatbegan with the collapse of the north and northeast qua-

drants of Cotopaxi Volcano (Fig. 14), involving parts ofthe andesitic edifice but mainly rhyolitic domes, thelikely remnants of the previous rhyolitic activity thathad occurred at ;6000 years BP (Hall and Mothes1995). The resulting avalanche left a hummocky plainat the north and northeast base of the cone. More im-portantly, the collapse probably thoroughly fragmentedthe existing glacial icecap, a phenomenon previously re-ported at Mount St. Helens and Nevada del Ruiz volca-noes (Waitt et al. 1983; Pierson et al. 1990). Immediate-ly afterward, the CCPF erupted from the site of the col-lapse, rapidly melted much of this fractured ice andsnow (reported elsewhere by Pierson and Janda 1994;Pierson 1995), became saturated with water, and trans-formed quickly into the CVL. Because only a singleflow unit is observed in either the proximal or distalreaches, it appears that once this immense debris flowwas generated, little water remained available for addi-tional debris flow generation. Whereas the ash flow andCVL were mainly directed down the Rio Pita and SanPedro drainages, a significant portion of the CVL wasdiverted westward by the drainage divide at the north-ern foot of the cone and thus gained access to the Lata-cunga Valley. It probably flowed rapidly through theInter-Andean Valley, both northward and southward,in a few hours.

Apparently the ash-rich pyroclastic flow trans-formed immediately and almost entirely into the CVLnear the base of the cone. With few exceptions, bothshare an identical mineralogical and lithological com-position and grain-size distribution.

As such, the CVL was a unique non-cohesive debrisflow, exceptional because of its ash-flow origin andenormous volume, as well as having behavioral charac-teristics similar to those of large cohesive mudflows.The flow’s great volume and discharge played a funda-mental role, since it apparently prevented the potentialdilution of the debris flow by river water, which in turn

244

inhibited significant textural changes with distance, de-spite the very long 326-km runout.

At Cotopaxi, this type of debris flow has a low re-currence rate compared with the more common ande-sitic debris flows generated many times each century.Nevertheless, given that rhyolitic eruptions have takenplace on Cotopaxi every 1500–2000 years during theHolocene, other CVL-like debris flows may happenagain.

Recent studies of large debris flows generated onhydrothermally altered stratovolcanoes (Vallance andScott 1997) emphasize the potential danger of theseevents, given their volumes, speeds, and ability tospread out over large areas. The present study demon-strates that ash-flow-generated lahars can also attainimpressive flow dimensions. Consequently, pristine vol-canoes with silicic ash-flow histories and large water re-sources, such as those in the Andes and Alaska, shouldbe considered potential candidates for generating verylarge debris flows that can cause widespread destruc-tion far from their sources.

Acknowledgements We especially thank our deceased friendand co-author, Dick Janda, who over the years shared his knowl-edge and enthusiasm with all of us who work on lahar problems.We also acknowledge the help of T. Pierson and K. Scott, whoreviewed a previous version of the manuscript, and J. Beget andG. Smith for the final draft. We also thank the on-going supportof the VDAP program of the U.S. Geological Survey and itsmembers.

References

Carrasco-Núnez G, Vallance JW, Rose WI (1993) A voluminousavalanche-induced lahar from Citlaltépetl Volcano, Mexico:implications for hazard assessment. J Volcanol Geotherm Res59 :35–46

Castro M, Mothes P, Hidalgo X, Sánchez A (1992) Uso del Mod-elo HEC-1 para el estudio de los lahares del Volcán Cotopaxi.Politécnica 17 :163–178

Clapperton CM (1990) Quaternary glaciations in the southernhemisphere: an overview. Q Sci Rev 9 :299–304

Crandell DR (1971) Postglacial lahars from Mount Rainier Vol-cano, Washington. Geol Surv Prof Pap 677 :1–73

Fisher RV, Schmincke HU (1984) Pyroclastic rocks. Springer,Berlin Heidelberg New York

Folk RL (1980) Petrology of sedimentary rocks. Hemphill, Aus-tin, Texas

Hall ML (1977) El Volcanismo en el Ecuador. Instituto Panamer-icano de Geografía e Historia, Quito

Hall ML, Mothes P (1995) Bimodal nature of the eruptive historyof Cotopaxi Volcano, Ecuador. Abstract IUGG XXI GeneralAssembly, Boulder, Colorado, p A452

Janda RJ, Scott KM, Nolan KM, Martinson HA (1981) Laharmovement, effects, and deposits. In: Lipmen PW, MullineauxDR (eds) The 1980 eruptions of Mount St. Helens, Washing-ton. US Geol Surv Prof Pap 1250 :461–478

Jordan E (1983) Die Vergletscherung del Cotopaxi, Ecuador.Gletscherkunde und Glaziologie 19 :73–102

Major JJ, Newhall CG (1989) Snow and ice perturbation duringhistorical volcanic eruptions and the formation of lahars andfloods. Bull Volcanol 52 :1–27

Major JJ, Pierson TC (1992) Debris flow rheology: experimentalanalysis of fine-grained slurries. Water Resour Res38 :841–857

Mothes PA (1992) Lahars of Cotopaxi Volcano, Ecuador: hazardand risk evaluation. In: McCall GJH, Laming DJC, Scott SC(eds) Geohazards, natural and man-made. Chapman and Hall,London, pp 53–63

Pierson TC (1980) Erosion and deposition of debris flows atMount Thomas, North Canterbury, New Zealand. Earth SurfProcess 5 : 227–247

Pierson TC (1985) Initiation and flow behavior of the 1980 PineCreek and Muddy River lahars, Mount St. Helens, Washing-ton. Geol Soc Am Bull 96 :1056–1069

Pierson TC (1995) Flow characteristics of large eruption-triggereddebris flows at snow-clad volcanoes: constraints for debrisflow models. J Volcanol Geotherm Res 66 :283–294

Pierson TC (in press) An empirical method for estimating traveltimes for wet volcanic mass flows. Bull Volcanol

Pierson TC, Janda RJ (1994) Volcanic mixed avalanches: a dis-tinct eruption-triggered mass-flow process at snow-clad volca-noes. Geol Soc Am Bull 106 :1351–1358

Pierson TC, Scott KM (1985) Downstream dilution of a lahar:transition from debris flow to hyperconcentrated streamflow.Water Resour Res 21 :1511–1524

Pierson TC, Janda RJ, Thouret JC, Borrero CA (1990) Perturba-tion and melting of snow and ice by the 13 November 1985eruption of Nevado del Ruiz, Colombia and consequent mo-bilization, flow, and deposition of lahars. J Volcanol Geo-therm Res 41 :17–66

Pierson TC, Janda RJ, Umbal JV, Daag AS (1992) Immediateand long-term hazards from lahars and excess sedimentationin rivers draining Mt. Pinatubo, Philippines. US Geol SurvWater Resour Invest Rep 92–4039, 35 pp

Qian Y, Wenhai Y, Wenlin Z, Xiuwen C, Longrong Z, Wengui X(1980) Basic characteristics of flow with hyperconcentration ofsediment. In: Boning L chairperson) Proc Int Symp River Se-dimentation, 24–29 March 1980, Beijing, China. GuanghuaPress, Beijing, pp 175–184

Rodine JD (1974) Analysis of mobilization of debris flows. PhDthesis, Stanford Univ, Stanford, 226 pp

Scott KM (1988) Origins, behaviour and sedimentology of laharsand lahar-runout flows in the Toutle-Cowlitz River system.US Geol Surv Prof Pap 1447-A: 1–74

Scott KM, Pringle PT, Valance J (1995) Sedimentology, behavior,and hazards of debris flows at Mount Rainier, WA. US GeolSurv Prof Pap 1547 :1–56

Scott WE, Iverson RM, Vallance JW, Hildreth W (1995) Volcanohazards in the Mount Adams Region, Washington. US GeolSurv Open-File Rep, pp 95–492

Sharp RP, Nobles LH (1953) Mudflow of 1941 at Wrightwood,southern California. Geol Soc Am Bull 64 :547–560

Smyth MA (1991) Movement and emplacement mechanisms ofthe Río Pita volcanic debris avalanche and its role in the evo-lution of Cotopaxi Volcano. PhD thesis, Univ Aberdeen, Scot-land

Trabant DC, Waitt RB, Major JJ (1994) Disruption of Drift gla-cier and origin of floods during the 1989–1990 eruptions ofRedoubt Volcano, Alaska. J Volcanol Geotherm Res62 :369–385

Vallance JW, Scott KM (1997) The Osceola mudflow from MountRainier: sedimentology and hazard implications of a hughclay-rich debris flow. Geol Soc Am Bull 109 :143–163

Waitt RB Jr, Pierson TC, MacLeod NS, Janda RJ, Voight B, Hol-comb RT (1983) Eruption-triggered avalanche, flood and la-har at Mount St. Helens: effects of winter snowpack. Science221 :1394–1397