Status of glacial Lake Columbia during the last floods from glacial Lake Missoula

QUATERNARY

RESEARCH

27, 182-201

(1987)

Status of Glacial Lake Columbia during the Last Floods from Glacial Lake Missoula BRIAN E ATWATER U.S. Geological Survey at Department of Geological Sciences, University of Washington. AJ-20. Seattle, Washington 98195 Received March 1, 1986 The last floods from glacial Lake Missoula, Montana, probably ran into glacial Lake Columbia, in northeastern Washington. In or near Lake Columbia’s Sanpoil arm, Lake Missoula floods dating from late in the Fraser glaciation produced normally graded silt beds that become thinner upsection and which alternate with intervals of progressively fewer varves. The highest three interflood intervals each contain only one or two varves, and about 200-400 successive varves conformably overlie the highest flood bed. This sequence suggests that jokulhlaup frequency progressively increased until Lake Missoula ended, and that Lake Columbia outlasted Lake Missoula. The upper Grand Coulee, Lake Columbia’s late Fraser-age outlet, contains a section of 13 graded beds, most of them sandy and separated by varves, that may correlate with the highest Missoula-flood beds of the Sanpoil River valley. The upper Grand Coulee also contains probable correlatives of many of the approximately 200-400 succeeding varves. as do nearby parts of the Columbia River valley. This collective evidence casts doubt on a prevailing hypothesis according to which one or more late Fraser-age floods from Lake Missoula descended the Columbia River valley with little or no TJ 1987 University of Washington. interference from Lake Columbia’s Okanogan-lobe dam.

INTRODUCTION

This report concerns the relative timing of floods and ice-lobe movement at the south-central edge of the Cordilleran ice sheet during a late part of the Fraser (late Wisconsin) glaciation. Citing flood-formed features in the Columbia River valley downriver from Foster Creek (Figs. 1 and 2A), some geologists have argued that at least one late Fraser-age flood from glacial Lake Missoula, Montana, descended the Columbia River valley in north-central Washington after the Okanogan lobe had become incapable of blocking great downColumbia floods (Bretz et al., 1956, pp. 993-994; Bretz, 1969, p. 525; Waitt, 1977; Waitt and Thorson, 1983, p. 64). This prevailing view is not necessarily compatible with evidence upriver from Foster Creek, in areas where the lake formed by Okanogan-lobe blockage of the Columbia River valley (glacial Lake Columbia) stood in the low-altitude path of Lake Missoula floods (Fig. 2). There, varve-bounded beds of late

Fraser age suggest progressively smaller jokulhlaups into Lake Columbia or coexisting waters, and a succeeding interval of late Fraser-age varves suggests existence of Lake Columbia during many decades after the last of these floods. This kind of sequence forms the upper part of a 115-m section of alternating flood beds and varves in ,the Sanpoil River valley, as summarized below from Atwater (1986). Similar stratigraphy crops out piecemeal in the upper Grand Coulee and the nearby Columbia River valley, in exposures described herein for the first time. These observations lead to the report’s chief hypothesis-that the last flood from Lake Missoula occurred many decades before the demise of Lake Columbia’s Okanogan-lobe dam-which in turn raises doubt about a Lake Missoula source for certain down-Columbia floods of late Fraser age. PREMISES

(1) Fraser-age Lake Missoula originated after 19,000 yr B.P. This timing is consis182

0033-5894187

$3.00

Copyright 0 1987 by the University of Washington. All rights of reproduction in any form reserved.

LAKE

I

120” I BRITISH

MISSOULA

I

/16~ ’

1

C OLUMBIA m;;,,--------.-.

-q

WASHINGTON

183

FLOODS

<

Path of 3 jbkulhloups of Clogue (197 _..-.

\

114” I

‘< c .

d'illene Lohr

?

\\

FIG. I. Index map. Glacial paleogeography from Waitt and Thorson (1983). Lake Columbia at -500-m level. Lake Missoula at maximum level. Stratigraphic sections: BC, Burlingame Canyon (Waitt. 1980); LC, Latah Creek (Waitt, 1984); MC. Manila Creek (Atwater, 1986): NC. Ninemile Creek (Chambers, 1971): NF, Ninemile Flat: NPC, Nez Perce Creek (Table 1).

tent with the available Canadian chronology of the east-central part of the Cordilleran ice sheet (Clague et al., 1980), with ash-layer and radiocarbon dating of Missoula-flood rhythmites in southern Washington (Waitt, 1985, p. 1284), and with counts and radiocarbon dating of varves among Missoula-flood rhythmites in the Sanpoil River valley (Atwater, 1986, pp. 29, 32). Morainal correlations of Richmond (in press) indicate the possibility that Lake Missoula also existed during a Rocky Mountain alpine-glacial maximum about 22,000 yr B.P. The pertinent correlations being speculative at present, Lake Missoula floods of Fraser age are herein regarded as entirely younger than 19,000 yr B.P. (2) Fraser-age Lake Missoula released a flood each time it rose high against its icesheet dam (the Purcell Trench lobe, Fig. 2). Such self-dumping explains (a) the cyclic tilling and emptying of Lake Missoula that Chambers (1971, 1984) inferred from

the lake’s Fraser-age bottom sediment, and (b) the apparently complementary Missoula-flood rhythmites of southern Washington (Waitt, 1980, 1985). Self-dumping of Fraser-age Lake Missoula further explains the stratigraphy of varve-bounded floodlaid beds in Washington and Idaho (Waitt, 1984, 1985; Atwater, 1984, 1986), and is consistent with mathematical models of the lake’s ice dam (Clarke et al., 1984, p. 296; Waitt, 1985, pp. 1280- 1283). (3) Fraser-age jloods from Lake Missoula ranged greatly in volume. Customarily unqualified use of epithet (catastrophic, cataclysmic, colossal) conceals great variation in the likely Fraser-age magnitude of Lake Missoula floods. The interflood period, as inferred from varve counts in the Sanpoil River valley, ranged from about 1 to 50 yr during Fraser time (Atwater, 1986; summarized below). If Lake Missoula’s minimum level and average inflow rate were fairly constant, then the volume of Fraser-age floods could

184

BRIAN ATWATER

have ranged 50-fold. The range could have been greater if the least frequent floods drained the lake to lower levels than did floods of much greater frequency. Such partial drainage during frequent floods is consistent with the stratigraphic distribution of desiccated, frost-cracked varves in the Lake Missoula section of Chambers (1984, pp. 191, 198). Thus, below I advocate the seeming paradox that the Grand Coulee, greatest of the scabland channels (Bretz, 1932), locally accumulated mere

Mortmum oru rheltsredby +OJanogan loba from lots Fraser-age floods down thp Columbia River volley

IO00 t

z z 2 c 2

beds of sand and silt during some Lake Missoula floods of Fraser age. SANPOIL RIVER VALLEY

Fraser-age Lake Columbia usually drowned the lower Sanpoil River valley to altitude -500 m (Figs. 1 and 2). The lake was not ordinarily self-dumping because subaerial outlets -especially divides near 465 m in the upper Grand Coulee-prevented it from rising high against the Okanogan-lobe dam. The normal level of the

2 = c 5 ,’ ;=

600600 Early Fraser-age landslide dom

400

Uistoulo

flood

F

sronc Coule

fsotures

KCr.

r

MRYorticat 0

cl-u

augg9ration

50

100

X 52110

200km J

FIG. 2. Profiles along drainage between Lake Missoula and Wallula Gap during the Fraser glaciation (Fig. I). Configuration of Purcell Trench lobe from Waitt (1985, p. 1282). Upper limit of Missoulaflood features from Baker (1973, pp. 15-17, Pl. 1) except near Sanpoil River valley, where from Atwater (1986, p. 4). Modern river profiles from District Engineers (1934, PI. 2). Maximum depth of Coeur d’Alene Lake from Ned Homer, Idaho Dept. Fish and Game (oral communication, 1985). (A) Profile via site of Okanogan-lobe dam in Columbia River valley. Dams near Moses Coulee from Waitt (1982). Profile of Great Terrace shows range in altitude of kettled surfaces, some of them alluvial fans graded to main part of terrace. (B) Profile via the Grand Coulee and lower Crab Creek. Boulder diameter in Quincy Basin from Baker (1973, Pl. 1).

LAKE

MISSOULA

185

FLOODS NE

-

Ytghest Lake Missoulo Quincy Basin

n 1000

400

Largest flood -moved boulder (intermediate diameter,m)

Post-glacial

I;0 km NC I+

dam

,

200 A

.$mpare profile

with Columbia Rive!, shorn in Figure 2A

Divide

brtrwn

aoutharn

wPC,I

Stratigraphic to

0

flgura

Columbia

Cbonntl,d

retion

River Seablond

volley lAtrot,r,

-- Acronym

keyed

I or 4A

FIG. 2.-Continued.

lake rose to -715 m while the maximally advanced Okanogan lobe blocked the upper Grand Coulee. Certain anomalously thin Missoula-flood beds mentioned below suggest that this high level was maintained for only about two centuries in Fraser time (Atwater, 1986, pp. 34-36). A 115-m glaciolacustrine section near Manila Creek, a Sanpoil River tributary, contains a bed for each of 89 Lake Missoula floods of Fraser age (Fig. 3). These beds alternate with intervals of varves. A varved interval near the middle of the section has yielded detrital wood with a 14C age of 14,490 f 290 yr B.P. (USGS-1860). Among the lower 45 flood beds, the number of interflood varves decreases upward from a maximum of 50-55 to about 15-25. This trend parallels that observed

by Chambers (1971, 1984, p. 191) in sets of Lake Missoula varves near Ninemile Creek, Montana. Higher in the Manila Creek section the number of interflood varves declines further to just one or two. About 200-400 additional varves overlie the highest flood bed without evident hiatus. These varves, and perhaps the highest several flood beds as well, postdate separation of Manila Creek valley from Lake Columbia by a Sanpoil River sandur whose progradation probably explains the abrupt thickening of varves and flood beds near altitude 490 m in the section (Atwater, 1986, pp. 14, 23). The thickness and maximum grain size of most flood beds in the Manila Creek section tend to vary directly with the number of interflood varves (Fig. 3). Partial excep-

-

-

-

-

-

-

470

465

460

430

425 420

410

0

-

460

503.5

10

NUMBER

--

=--,

-,

10 , , , I-200-400 successive = varves -I

0

=

20

Y

J

4

20 ,

4

30

OF VARVES

-

PER VARVEO

40

unrecovered thlnnlng

60

-L-,1

or structural

INTERVAL

50

core.

30 40 50 50 , , 1 VARVE COUNTS iXPLiNAT;GN ’ ’ 0 Varved rhythmlte of Lake Missoula at Ninemile Creek. Montana ’

rj

0

--

5

10 E

0

0.5

2.9 m thick

3.2 m,?k

thick

2.0

2.6 m -

1.5

I- -

1.0

70 0 5 10 0 0.5 1.0 1.5 2.0 2.5 RANGE OF MEAN THICKNESS THICKNESS OF SUCCEEDING OF VARVES. IN CENTIMETERS. FLOOD BED, IN METERS WITHIN VARVED INTERVAL

-

70

47o-l

465

-

03,5-e--Purcell

i15.

Purcell Trench lobe maximally advanced

Purcell Trench lobe possibly readvances, temporarily

Purcell Trench lobe probably reedvances.

Sanpoil River outwash plain progrades near and eventually dams the mouth of Manila

Trench lobe retreat* from Clark Fork valley (demise of Lake

INTERPRETATION

-

FIG. 3. Vertical trends in ranges of varve counts. in ranges of mean varve thickness. and in flood-bed thickness and grain size near Manila Creek (from Atwater, 1986, p. 22). Column at right shows events inferred from these data. Ordinate divided evenly into 89 increments of largely unequal stratigraphic thickness each corresponding to a known or inferred flood bed and to the known or inferred interval of varves beneath that bed. Varve-count ranges of Lake Missoula rhythmites (from Ninemile Creek section (Figs. 1 and 2) as logged by Chambers. 1971, App. III: 1984, p. 191) are fitted visually to the Manila Creek counts with presumption that nondeposition and erosion prevent the Lake Missoula data from recording all of interflood time (see Chambers, 1971, pp. 30, 59). R, reversal of overall decline in varve counts in upper part of Manila Creek section.

z

-.313

LAKE

MISSOULA

tions to this rule are the aforementioned flood beds near altitude 490 m. These finegrained beds, separated by 15-30 varves, are at least as thick as many of the coarsergrained flood beds separated by 40-55 varves. The other main exceptions are two sets of anomalously thin flood beds low in the section (Fig. 3). I suspect that the floods responsible for these thin beds were attenuated by the maximum extension of the Okanogan lobe (which enlarged ambient Lake Columbia) and Columbia River lobe (which could have deflected floods in the Spokane River valley; Fig. 2); further, I believe that the lack of such anomalous beds in the lowermost part of the section precludes substantial Fraser-age lowering of divides in the upper Grand Coulee (Atwater, 1986, p. 37). The scarcity of gravel in flood beds of the Manila Creek section (Fig. 3) is consistent with their deposition in a deadend arm of a large lake. Given the consequent opportunities for damping of Missoula-flood effects, I infer that certain sandy flood beds represent the largest of the Fraser-age floods from Lake Missoula, hence also the maximum thickening (and advance) of the Purcell Trench lobe (Fig. 3; Atwater, 1986, p. 34). Lacustrine damping of flood effects should have diminished as Lake Columbia shoaled. Flood beds in the Manila Creek section nonetheless become generally thinner and finer upsection-evidence of an overall decrease of Missoula-flood magnitude in late Fraser time. The Manila Creek section thus corroborates Waitt’s (1980, 1985) hypothesis that Lake Missoula floods of late Fraser time became generally smaller (upsection thinning and fining of flood beds) and more frequent (upsection decrease in number of interflood varves) in response to overall late Fraser thinning of the Purcell Trench lobe. The approximately 200-400 varves above the highest flood bed further imply that these trends culminated in the demise of glacial Lake Missoula. But Lake Columbia apparently lived on; scores of varves re-

FLOODS

187

main thick (fed by Sanpoil River outwash) both among and above the highest flood beds, and therefore show no sign that the Sanpoil River regarded to a base level below 500 m either during or shortly after the last Lake Missoula floods recorded in the Manila Creek section. Correlatives of the highest flood beds and intervening varves near Manila Creek are probably lacking in the Ninemile Creek section and in most sections of flood rhythmites in southern Washington. The Ninemile Creek section, at altitude 940 m, stands 200 m above the highest thalweg on the floor of Rathdrum valley (Fig. 2). Lake Missoula had a capacity of -260 km3 in that 200-m vertical interval (Singer and Craig, 1984)-room for roughly five times the annual inflow indicated by the product of maximum volume of Lake Missoula (-2500 km3; Singer and Craig, 1984) and minimum Missoula-flood frequency (V-50 yr; Fig. 3). A Lake Missoula that emptied every year or two was thus unlikely to have surmounted the Ninemile Creek section. As for flood-rhythmite sections in southern Washington, these commonly stand many tens of meters above local base level, as shown in Figure 2 for the Burlingame Canyon section of Waitt (1980). The last Lake Missoula floods probably were too small to overtop such sections-as surmised by Waitt (1980, p. 675) and confirmed by him (Waitt, 1985, p. 1284) and by Baker and Bunker (1985, p. 31) from altitudinal variation in the number of floodlaid rhythmites overlying ash layers of Mount St. Helens set S. But correlatives of the upper 25 m of the Manila Creek section could be present in areas at or below local late-Fraser-age base level along the profile of the glacially dammed and diverted Columbia River (Fig. 2). I believe that they are-that probable correlatives of the highest flood beds near Manila Creek are exposed near Steamboat Rock in the upper Grand Coulee, and that varves younger than the last flood from Lake Missoula are widely exposed in the

188

BRIAN

ATWATER

upper coulee and in nearby segments of the Columbia River valley. UPPER

GRAND

COULEE

A lake or sluggish river continuous with glacial Lake Columbia occupied the upper Grand Coulee during late Fraser time (Bretz, 1932, pp. 76, 78; Flint, 1935, pp. 188-189). Water that was standing or slowly moving is indicated by laminated silt that mantles most of the coulee floor; continuity of this water with Lake Columbia is indicated by a patchy, silty terrace that stands well above the divides on the coulee floor (altitude 465 m; Figs. 2B and 4A) as it descends the coulee’s upper 40 km from altitude 500 m to altitude 480 m. Some of this height and slope probably reflects isostatic rebound (Flint, 1935, p. 189). If equal to the vertical separation between the normal 479-m level of Banks Lake (man-made reservoir dammed just above Dry Falls) and the highest varved silt (deposited in late Fraser-age lake or river whose spillway was Dry Falls), the rebound amounts to 8 m at the Steamboat Rock section and 5 m farther south at the Paynes Gulch section (Fig. 4). Such rebound could have been caused not only by the wasting of the Okanogan lobe but also by the draining of Lake Columbia and consequent incision of its fill. Bretz (1932, p. 78; 1969, p. 525) reported a lack of flood-formed features either within or above the laminated late Fraserage silt of the upper Grand Coulee. Exposures beside Banks Lake suggest an amendment: though the youngest of this silt indeed postdates the last flood through the coulee, older late Fraser-age silt of glaciolacustrine or glaciofluvial origin is interbedded with the deposits of 13 floods that could have come from Lake Missoula. Steamboat

Rock

Section

This section, 0.8 km south of Steamboat Rock, rises 10 m from a bench eroded by Banks Lake (Figs. 4 and 5). The top of the

section is a few meters below the silty terrace, nearby patches of which have an average altitude of about 493 m. The lowermost part of the section rests unconformably on basalt-rich cobble gravel (Fig. 5B) and on a dark gray diamict that is probably lodgment till. The uppermost 0.5 m of section shown in Figures 5A and 5B is mostly structureless sand and silt, probably colluvial or eolian. But tens of successive varves crop out on the Banks Lake bench 50-100 m south of the area shown in Figure 5B. These varves dip south and lap onto a south-dipping unconformity that truncates at least the lowest 4.5 m of the section. The Steamboat Rock section is dominated by 13 graded beds. Overall these beds become much thinner upward; the lowest complete bed is 3.5 m thick, the highest 0.1 m. Grain size also decreases upsection: whereas the 3.5-m bed consists chiefly of medium sand, the 0. l-m bed contains no clast larger than coarse silt. Most beds are reversely graded in their lowest lo-30 cm and normally graded in their highest IO-50 cm. Lamination in the sand, defined by median-grain-size differences estimated to be as much as 1 phi, is planeparallel in the center of most beds, rippled above, and locally rippled below. The uppermost part of most beds is a normally graded layer of yellowish brown (1OYR 5/5) silt and silty clay whose contact with underlying sand is abrupt (Fig. 5C). The lowest 10 graded beds, and perhaps also beds 11 and 12, are separated from one another by a deformed and eroded interval of olive (5Y 5/4) silt-and-clay varves. The base of each varved interval is conformable, the top distinctly disconformable. The interval between graded beds 6 and 7 contains the largest number of countable varves, 3 or 4; the thickness of each of these varves is about 2 cm. Though I neglected to examine carefully the varves above the south-dipping unconformity, I did note that they resemble the varves in the lower half of the Paynes Gulch section

LAKE

MISSOULA

189

FLOODS

old;T;’

:

0

1

‘500 ALTITUDE

1

J -

Rim of the Grand Cwltt

-0

Eartarn lamit of Okonogan Iabe during tht Frarw glacIalinn-Dashed where location opprorimote, dotted rhara location Inferrod End moraine

of tote

and Steamboat

Fr~wr

age

Rock

of it* PG. A ---

J 700

C

Verlital exoggtrati0n Y 100

‘G

-

600 (.I

B

flaw

Crow - coul~~ profile Strotlgraphic swtion Divide Historic

nur

465m

rentsrlina

of

Columbia

River

FIG. 4. Paleogeographic setting of stratigraphic sections in the upper Grand Coulee and adjacent Columbia River valley. Stipple denotes -500-m glacial Lake Columbia and its outflow in the upper coulee. B, Barry; EC, Elmer City; GCD, Grand Coulee Dam (Flint and Irwin, 1939); KC, Kaiser Canyon; MR, Monaghan Rapids; PC, Paynes Gulch; SR, Steamboat Rock. (A) Plan view. Glacial geology chiefly from Hanson (1970, Fig. 19-2). (B) Cross-coulee profiles. Level of Lake Columbia equated with present altitude of highest silty terrace near profile, without adjustment for possible isostatic rebound. (C) Cross-sectional areas along the cross-coulee profiles.

190

BRIAN ATWATER

IN GRADED BEDS

40

T

Biolybaied Load structures 01 burrows / I

Rip-up closts of varves

?

-:*-3-4

“owes preserved locally

”

LT : I

Uppermost port of bed 4 largely contorted

-1.

FIG. 5. Steamboat Rock section; SW/4 NW/4 sect. 11, T.27N. R.?9E, Steamboat Rock SW 7.5-min quadrangle. (A) Generalized stratigraphy, and vertical trends in estimated grain size within flood beds. Contacts shown by solid lines in and left of column; wavy lines left of column denote disconformity. In flood beds (numbered), sand is stippled, silt and clay dashed. Varved intervals (V), unpatterned in column, are mostly too thin to sketch, but disturbed varves are locally prominent within flood beds as flame structures (beds 3 and 4) and rip-up clasts (bed 8). Grain-size trends dashed where commonly

(described below), and that they number at least 30. The Steamboat Rock section suggests a fivefold sequence of glacial-age events. (1) Deposition of the diamict by glacier ice. Probably this occurred during the Okanogan lobe’s last occupation of the Grand Coulee, before which time the upper coulee “had been eroded to present depths” (Bretz, 1932, p. 35). (2) Deposition of the basaltic cobble gravel by subglacial stream or Luke Missoula flood. If deposited by flood, the gravel may belong to a bar that is pendant to Steamboat Rock.

(3) Deposition of most of section by 13 successively smaller floods. At least the first 10 of these found the coulee floor already submerged, and this submergence was probably by waters continuous with Lake Columbia. The interflood period was at least 3 or 4 yr between the 6th and 7th flood. Probably most floods waxed more slowly (gradual upward coarsening) than they waned (abrupt upward change from sand to silt), therein resembling the modern jokulhlaups described by Thorarinsson (1939, p. 227), Mathews (1973, p. 102), and Bjornsson (1974, pp. 5, 7) (see also Waitt, 1984, p. 51, 1985, p. 1285). Many of these

LAKE

MISSOULA

FLOODS

FIG. 5-Continwd interrupted by differences between adjacent laminae. (B) Outcrop viewed from south-southeast: Steamboat Rock in background. Ledges are silt and clay. chiefly the uppermost parts of flood beds. Prominent flood beds numbered. Gravel in foreground exhumed by waves of Banks Lake (normal level 479 ml. (Cl Abrupt contact of silt lz) over fine to medium sand 1s) in two flood beds. Loose dry sand (Is) conceals lowermost parts of flood beds. Lamination: r. ripple drift: p. plane parallel. Shovel 55 cm long.

191

192

BRIAN

ATWATER

floods deposited medium sand as flat beds standing or slowly moving water, in which of the upper flow regime-a combination formed the tens of successive varves above of grain size and bedform that suggests a the south-dipping unconformity. The inuncurrent velocity, averaged through the dation could have been continuous with water column, of at least 1 m/set (Rubin that recorded by the interflood varves, or it and McCulloch, 1980, p. 218). If this ve- could postdate a period when most of the locity applies to the entire cross-sectional coulee floor was subaerially exposed. The area below the terrace altitude of 493 m hypothesis of continued inundation re(Fig. 40, peak flood discharge exceeded quires only that the lack of varves atop 130,000 m3/sec. Although speculative, this beds 10, 12, and 13 be due to nondeposition limiting estimate is probably too low be- on a flood-built shoal, and that the local cause the Steamboat Rock section is lo- erosion low in the section be due to one or cated both in the lee of Steamboat Rock more of the last 10 floods recorded in the (Fig. 4A) and on the point-bar side of the section. channel cross section (Fig. 4B); also floodThe Steamboat Rock section thus conwater may have risen many meters above tains evidence that 13 successively smaller the 493-m terrace. Yet 130,000 m3/sec is jokulhlaups swept through the Grand four times the 24-hr average for maximum Coulee in late Fraser time, and that this Columbia River discharge during the flood series is largely or wholly equivalent to the of 1894, the greatest Columbia River flood increasingly frequent Lake Missoula floods in the historic record (District Engineers, that I infer from the highest flood beds of 1934, p. 1738; discharge determined at The the Manila Creek section (Fig. 3). The first Dallas, downstream of Wallula Gap). More10 of the 13 floods probably occurred while over, 130,000 m3/sec is enough to have the upper coulee held much standing or transmitted 1 yr’s Lake Missoula inflow (at slowly moving water that was continuous -50 km3/yr) in 4 days, and it equals the with Lake Columbia. I cannot tell whether peak discharge calculated for a 36-km3 jo- such ambient water was present during the kulhlaup by means of the Clague and succeeding 3 floods. But the entire series of Mathews (1973) equation as modified by 13 floods had surely ended decades before Beget (1986): this water finally disappeared-as shown further in the Paynes Gulch section. Q = 0.0067v0.69, where Q is maximum instantaneous dis- Paynes Gulch Section A boulderly gravel in the form of a point charge (in m3/sec) and V is volume drained (in m3). (This modified equation describes a bar extends 5 km along the east side of the regression line (r2 = 0.86) through data for upper Grand Coulee’s floor approximately 21 modern jokulhlaups with peak discharge midway between the Columbia River valley in the range 150-50,000 m3/sec and total and Dry Falls (Figs. 4A and 4B). This volume in the range 0.003-7.0 km3.) Thus gravel, overlapped to the west by well-lammany of the sandy beds in the Steamboat inated silt, probably is the silt-draped Rock section connote jokulhlaups of mag- boulder bed of Bretz (1932, p. 78). Lownitude appropriate to those Lake Missoula ering of Banks Lake exposes exhumed tops outbursts recorded high in the Manila of the boulders beside wave-freshened verCreek section. tical outcrops of the silt. Some vertical ex(4) Local erosion of at least the loM)er posures nearly intersect the silty terrace, half of the section. This erosion may have here at altitude 486 m. The Paynes Gulch taken place during any of the last 10 floods. section (Fig. 6) is representative of such or it may totally postdate them. exposures, the stratigraphy of which is sim(5) Continued or renewed inundation of ilar along at least 1 km of the southern part at least the lower half of the section bJ of the bar.

LAKE

MISSOULA

FLOODS

ALTITUDE (meters)

485-

484-

193

SCHEMATIC LlTHOLOGr

VARVE THICKNESS

v- --_- - -- . ---------

h

NUYBER OF VARVES

-111 .---_

5

Fines upward

-210* ,~me4a~ Icm

27-28

Boulder ?,

FIG. 6. Paynes Gulch section; SW/4 NE/4 sect. 7. T.26N. R.29E. Steamboat Rock SW 7.5-min quadrangle. (A) Outcrop viewed from west. Boulders exhumed by 479-m Banks Lake. (B) Estimated grain size of prominent nonvarved strata (patterned). vertical trend in thickness of varves (unpatterned), and paleocurrent directions inferred from cross lamination. Silt dashed, very fine sand stippled: both chiefly structureless. Range in varve thickness shown with wavy line where varves commonly pinch and swell because of ripples (Fig. 7B).

The Paynes Gulch section contains about 180 silty rhythmites, most or all of them varves, and 5-10 beds of silt or very fine sand without any sort of rhythmic&y (Fig. 6B). The varves predominate at altitudes 479-483 m, the silt and sand beds at altitudes 483-485 m. The varves thin upward from an average of 8 cm in the lowest 1.5 m (Fig. 7) to 1-2 cm at altitudes 482-483 m (Fig. S), then thicken slightly to about 3 cm near altitude 484 m (Fig. 6B). Most thick varves contain basal very fine sand with ripple lamination that climbs southward, downcoulee (Fig. 7A). Some of these locally pinch and swell from sand-starved, silt- and clay-draped ripples whose forms climb southward through successive varves (Fig. 7B). Sand interbedded with thin varves in the middle of the section also contains evidence of south-directed currents (Fig. 8). By contrast, most silt and sand beds high in the section lack primary

sedimentary structures other than slight normal grading in the lowest of these beds. The highest 0.5 m is bioturbated and unstratified. No deposits or landforms suggesting flood are superposed atop the section. The Paynes Gulch sequence suggests a twofold sequence of glacial-age events. (1) Formation of the bouldery bar. Though many of the boulders may have been delivered by the Okanogan lobe, the point-bar form indicates that the deposit was at least shaped by flood, as previously inferred by Bretz (1932) and by Hanson et al. (1979). Probably this shaping occurred during the scores of Lake Missoula floods that postdate the Okanogan-lobe maximum and are recorded in the Manila Creek section at altitudes 433-504 m (Fig. 3). (2) Inundation of the west flank of the bar for abordt 180 yr. During some of this time and especially in the first few decades,

BRIAN ATWATER

FIG. 7. Thick varves in lower part of Paynes Gulch section. (A) Tabular varves, altitude 479.5 m. Upward sequence in each of the four complete varves: thin basal layers of very fine sand and silt with gently south-climbing pseudo-lamination; thick, mostly upward-fining layer of yellowish brown (IOYR 514) silt; and thin cap of olive gray (5Y 5/3) clay (c). (B) Wavy varves through which climb draped. sand-starved ripples; altitude 480.0-480.6 m. Sand (light) accumulated chiefly in the lee of the previous year’s clay-draped ripple crest.

LAKE

FIG. 8. Thin varves and a cross-laminated

MISSOULA

FLOODS

195

sand layer (3). altitude 482.3 m in Paynes Gulch section.

downcoulee currents were seasonally strong enough to form ripples of very fine sand but seasonally weak enough to permit the draping of these ripples with clay. Hypothetically, such currents could have been generated by inflow from coulee-rim drainage basins, especially if those basins contained relicts of the Okanogan lobe. But the coulee-rim drainage basins north of the Paynes Gulch section are so small (total area -300 km*) that the more likely cause of the downcoulee currents was outflow from Lake Columbia. Such outflow must have been small or lacking at least seasonally during the last several decades that are represented by the upper 2 m of the section, so that most of the silt and sand there could accumulate without lamination. Much of that part of the section may have been formed by eolian deposition in sluggish or standing water, with Lake Columbia draining chiefly through the remains of its Okanogan-lobe dam in the Columbia River valley. Varves of the Paynes Gulch section probably correlate with the tens of successive varves above the unconformity in the

Steamboat Rock section. This correlation is likely because the varves are the youngest well-laminated deposits at each section and thereby appear to be part of the silt that so widely mantles the floor of the upper coulee. Moreover, the correlation is consistent with the lack, in the Paynes Gulch section, of exposed correlatives of the 13 flood beds of the Steamboat Rock section. This lack befits a threefold difference in cross-sectional area (Figs. 4B and 4C) and the absence of an upstream obstacle comparable to Steamboat Rock. A flood that laid sand and silt in the lee of Steamboat Rock may have chiefly scoured the site of the Paynes Gulch section, and what deposit it left at that site had little chance of surviving any but the last few of the floods represented in the Steamboat Rock section. Varves of the Paynes Gulch section probably correlate also with tall sections of varves in the Columbia River valley. These sections provide further evidence that Lake Missoula jokulhlaups ceased, at least temporarily, before the demise of Lake Columbia.

BRIAN ATWATER

196 COLUMBIA

sandy graded beds at Kaiser Canyon contain basal downvalley-dipping foresets that indicate an easterly source for the floodwater; but foresets in the intervening varves dip exclusively upvalley, away from a westerly (Okanogan lobe) source and toward Lake Columbia’s subaerial outlets. At Grand Coulee Dam and Kaiser Canyon the alternating flood beds and varved intervals underlie lodgment till; at Elmer City they postdate it (Table 1; Fig. 9). Additional sections of alternating flood beds and varved intervals are present farther up-

RIVER VALLEY

Periodic floods from glacial Lake Missoula into glacial Lake Columbia are abundantly recorded in the Columbia River valley within the Fraser-age limit of both the Okanogan and Columbia River lobes. Within the limit of the Okanogan lobe, sandy graded beds alternate rhythmically with sets of varves at Grand Coulee Dam (Flint and Irwin, 1939, p. 666) and farther downvalley at Elmer City and Kaiser Canyon (Table 1; Figs. 2A, 4A, and 9). The TABLE

1.

SIX GLACIOLACUSTRINE

SECTIONS NORTHEASTERN

VALLEY,

Name

WITHIN THE ICE-SHEET LIMIT IN THE COLUMBIA WASHINGTON (FIGS. I, 2, AND 4A)

Location” Sectiorts

Evidence for stratigraphic position (Fig. 9)b

MYthin

the

Kaiser Canyon (KC)

30N/30E-22 and -23 (Alameda Flat)

Monaghan Rapids (MN

30Ni30E-26; NW14 SW/4 (Alameda FlaUc

Barry (B)

29N/30E-I; SW14 (Nespelem) 29N/3 1 E-20; SW/4 SW/4 (Grand Coulee Dam)

Elmer City (EC)

Sections

within

RIVER

the

limit

limit

Ninemile Flat (NF)

29N/35E-9; NE/4 SW14 (Wilmont Creek)

Nez Perce Creek (NW

3 1N/36E-34; SE/4 SE/4 (Hunters)

of the Okunogun

lobe

In SE/4 NW/4 sect. 23, till overlies varves that, in sets of at least 40, alternate with flood beds (sequence first described by Flint and Irwin (1939, p. 676) and Jones et al. (1961, p. 28)) Diamicts, largely tabular and bounded by welllaminated silt and sand, are confined to lower part of section; section butts against and partly overlaps a haystack-bearing till regarded as younger by Flint and Irwin (1939, p. 676) See footnote b Undeformed section set 300 m below and 1.5 km within glacial margin; contains at )tast 6 sandy flood beds separated by sets of about 25 varves of the

Columbia

River

lobe

Thirteen sandy flood beds, 5 of them usually drowned by reservoir, alternate with sets of 40-55 varves (V. L. Hansen, written communication, 1983); beds at altitude 391-398 m resemble the upper set of anomalously thin flood beds in the Manila Creek section (F-10 through F-13, Fig. 3); overlain by landslide deposits but not necessarily by the gravelly sand reported by Jones et al. (1961, p. 20) Top of section forms probable lake-bottom terrace

(1Township/Range section; part of section (USGS IS-min quadrangle). b Chief evidence for correlation among the Monaghan Rapids, Barry, and Nez Perce Creek Sections, and for their correlation with the Paynes Gulch and uppermost Manila Creek sections: lack of evidence of glacial override; apparent lack of Missoula-flood beds; pronounced upsection thinning of varves. c Illustrated by Flint and Irwin (1939, PI. 6, Fig. I). who mistakenly assigned the subject to sect. 23.

LAKE l-COLUMBIA

MISSOULA

197

FLOODS

RIVER VALLEY+-GRAND

COULEE+~SANPOItt,COL

OKANOGAN LOBE-----$Vt?Y

I*

FCOL.

R VALLEY-( R LOBE-l

MC 500 :

LOW-LEVEL LAKE COLUMBIA

Burled thalwag I” bedrock at Grand Coulee Dam I

L

/:

NO HORIZONTAL SCALE

EXPLANATION DEPOSITS

CONTACT ~ Canformable upward

thinning

lllllll

Sandy flood beds and varved intervals

intervening

Diomict xxx*

Probably

lodgement

ears

Probably

subaqueous

....

D~sconformoble

of YO~VBS

till flaw

TIME- STRATIGRAPHIC CORRELATION -Last in main 01 Sole series Of ]dkulhioup’i from take Misrou~o oceuring during the Fraser glociotlo” ~ox\mum Lobe

till

extenslan

of

Qkanogon

Gravel, cobbles, 01 boulders -- Probably floadlold at IC and PG, not show” above

disconfarmltler

of KC,MR,E, and EC

FIG. 9. Correlation of stratigraphic sections in and near the Columbia River valley of northeastern Washington. Arrows at top indicate downvalley and upglacier directions. Figures I and 2 show location of sections; Table 1 gives acronyms for all sections except Manila Creek (MC: Fig. 3), Paynes Gulch (PC; Figs. 6-8). and Steamboat Rock (SR: Fig. 5).

valley in areas vacated by the Columbia River lobe, most notably beside Ninemile Flat (Table 1; Figs. 1, 2, and 9). Also present in valley reaches vacated by the Okanogan and Columbia River lobes, however, are tall varved sections of late Fraser age that contain little or no sign of Missoula flood. Two such sections containing about 200 varves are located near Barry and Monaghan Rapids, between Elmer City and Kaiser Canyon; a third, with about 500 varves, is near the mouth of Nez Perce Creek, 25 km upvalley from Ninemile Flat (Table 1; Figs. 1,2A, 4A, and 9). The varves in all these sections thin upward from I to 5 m near the base (Fig. 10)

to about 10 cm (Barry, Monaghan Rapids) or 2 cm (Nez Perce Creek) at the top. All three sections contain basal cobble gravel: the Monaghan Rapids section also contains basal diamict probably deposited by subaqueous debris flow (Table 1). Terraceforming cobble gravel unconformably overlies the Barry and Monaghan Rapids sections without upward coarsening or thickening of the subjacent varves. None of the three sections contains punctuating strata that resemble the Missoula-flood beds at localities mentioned above. Deposits of relatively small Missoula floods could be mistaken for some of the sections’ thickest varves, the clayey parts of which

198

BRIAN ATWATER

areas may have also produced the Nez Perce Creek section, for the varves of this section likewise extend to altitude 490 m. Additionally or alternatively, the Nez Perce Creek section could have accumulated behind a dam of Missoula-flood gravel located at the mouth of the Spokane River. At least two varved sections in the Columbia River valley thus reveal the same history that most simply explains the FIG. 10. Laminated silt and very fine sand in part of varves in the uppermost part of the Manila an extremely thick varve. lower part of Monaghan Creek section and in the Paynes Gulch secRapids section. Long-wavelength, steeply climbing tion. For many consecutive decades in late ripples propogate through the prominent rhythmites. Fraser time, the Okanogan lobe impounded These rhythmites may represent diurnal fluctuations a -500-m Lake Columbia that received no in meltwater discharge: if so, they are loosely analogous to the modern glaciomarine rhythmites described floods from Lake Missoula. by Cowan and Powell (1986).

SOURCES OF LATE FRASER FLOODS

are commonly injected into one or more succeeding varves (Jones et al., 1961, p. 41; V. L. Hansen, oral communication, 1983). But there is little chance for such confusion higher in the sections, where only upward thinning modifies the rhythmicity of the varves. The varves of the Barry and Monaghan Rapids sections, and perhaps many in the Nez Perce Creek section as well, formed either within or beside Lake Columbia. The Barry and Monaghan Rapids sections belong to the “upper lacustrine unit” that Jones et al. (1961, p. 28, PI. 5) map as high as altitude 490 m. This is about 100 m above the late Fraser-age lake or lakes (Lake Brewster; Fig. 1) into which formed the Great Terrace (Fig. 2A; Waters, 1933, p. 815; Waitt and Thorson, 1983, pp, 62-64) and to which seems graded the gravel atop the Barry and Monaghan Rapids sections (Flint, 1935, pp. 187-188). The varves in these sections thus cannot represent upvalley transgression by Lake Brewster; rather, they suggest downvalley spread of -500-m Lake Columbia-into a canyon dissected during a temporary destruction of the Okanogan-lobe dam (Jones et al., 1961, p. 28), or into giant kettles that opened after mere contraction of that dam. Spread of Lake Columbia into deglaciated

The Columbia River valley between Foster Creek and Crab Creek (Figs. 1 and 2A) contains the published evidence of great floods that might be more recent than the demise of Lake Columbia. Waters (1933, p. 817) cited the kettles on the Great Terrace as evidence of a berg-bearing flood at least 30 m deep with respect to the terrace surface. He attributed that flood to breakup of the Okanogan-lobe dam. Although Flint (1935, p. 191) argued that Waters’s kettles could have been produced without flood, bedforms and boulders found on the terrace reaffirm Waters’s hypothesis (reports cited by Waitt and Thorson, 1983, p. 64). Assuming prior dissection of the Great Terrace, Waitt and Thorson (1983, p. 64) inferred one or more down-Columbia floods as deep as 215 m and called them catastrophic outbursts from Lake Missoula. Additional floodformed features overlooking the Great Terrace that were formerly ascribed to a still deeper flood of late Fraser age (Waitt, 1977) are now believed to predate the Fraser maximum of the Okanogan lobe because they cannot be traced upvalley from the lobe’s southern limit (Waitt, 1982, p. 20). Far downvalley from that limit, a bouldery bar about 30 m high extends most of the way across the valley of lower Crab

LAKE

MISSOULA

Creek. Because much Lake Missoula floodwater followed the course of lower Crab Creek while the Columbia River was diverted by the Okanogan lobe (Fig. 2B), Bretz et al. (1956, pp. 993-994) inferred that the flood responsible for this bar (Beverly bar) descended the Foster Creek-Crab Creek reach of the Columbia River valley. Initially Bretz proposed correlation with Waters’s (1933) torrent from Lake Columbia (Bretz et al., 19.56, p. 994); later he implicated Lake Missoula and assigned the Beverly-bar flood to an interval during which the Okanogan lobe temporarily vacated the Columbia River valley (Bretz, 1969, p. 525). The evidence presented herein-from the Sanpoil River valley, the upper Grand Coulee, and the Columbia River valley above Foster Creek-disallows a Lake Missoula source for these late Fraser-age floods except under certain conditions. (1) Tandem jiikulhlaups from Lakes Missoula and Columbia. As the Okanogan lobe thinned in the Columbia River valley, Lake Columbia may have become susceptible to self-dumping. Perhaps jokulhlaups from Lake Columbia were triggered and fed by some of the floods represented high in the Manila Creek section and in the Steamboat Rock section. Such drainage of Lake Columbia could account for the lack of varves atop flood beds high in the Steamboat Rock section. (2) Temporary retreat of the Okanogan lobe from the Columbia River valley. This could be the withdrawal proposed by Jones et al. (1961, p. 28) and Bretz (1969, p. 525). It could account for the lack of varves atop beds 10, 12, and 13 of the Steamboat Rock section (Fig. 5A); also it could account for the grave1 in the lowermost parts of three of the sections in the Columbia River valley (Fig. 9). Yet as noted above, the Manila Creek section indicates that this kind of retreat, if it occurred at all, must have been brief. (3) Readvance of the Purcell Trench lobe into the Clark Fork valley after final

199

FLOODS

retreat of the Okanogan

lobe from the CoPerhaps the floods from Lake Missoula did not end with the last in the series of jokulhlaups represented in the Manila Creek section (Fig. 3). Rather, perhaps the lake was formed anew by readvance of the Purcell Trench lobe (Lake Missoula’s ice dam; Figs. I and 2) following a period of retreat that is represented in the vicinity of Lake Columbia by the scores of successive varves described above. Precedent for readvance of the Purcell Trench lobe is suggested by certain varve-count trends in the Manila Creek section (labeled R in Fig. 3); these trends reveal one or two temporary reversals in the late Fraser-age decline in interflood period and consequently, by the reasoning of Thorarinsson (1939, p. 223), imply one or two interruptions in the late Fraser-age thinning of Lake Missoula’s ice dam. None of these conditions is required if the Okanogan lobe stably dammed Lake Columbia through all the last floods from Lake Missoula but retreated from the Columbia River valley before later non-Missoula floods, of which Clague’s (1975, p. 235) late Fraser-age jokulhlaups from the Columbia River drainage basin in southeastern British Columbia are likely examples. This scenario best explains the upper part of the Manila Creek section and is compatible with the other sections described herein. It is also compatible with the flood-formed features on the Great Terrace provided that these were formed by non-Missoula flood during or after demise of Lake Columbia, in the spirit of Waters’s (1933, p. 817) proposal; likewise, as stated by Bretz et al. (1956, p. 994). such a nonMissoula flood may have produced Beverly bar. Thus the last outburst from glacial Lake Missoula may have run into a glacial Lake Columbia whose Okanogan-lobe dam passed little or none of the floodwater. lumbia

River

valley.

ACKNOWLEDGMENTS Susan H. Rose assisted with field work Coulee and nearby parts of the Columbia

in the Grand River valley.

BRIAN ATWATER

200

Vicki L. Hansen made available her unpublished work on the Ninemile Flat and Nez Perce Creek sections. James E. Evans supplied Figure 10, and Bernard Hallet pointed out a helpful reference. For careful reviews I thank Victor R. Baker, Eugene P. Kiver. Robert E. Powell. and Richard B. Waitt.

REFERENCES Atwater, B. F. (1984). Periodic floods from glacial Lake Missoula into the Sanpoil arm of glacial Lake Columbia, northeastern Washington. Geology 12, 464-467. Atwater, B. F. (1986). “Pleistocene Glacial-Lake Deposits of the Sanpoil River Valley, Northeastern Washington.” U.S. Geological Survey Bulletin 1661. Baker, V. R. (1973). “Paleohydrology and Sedimentology of Lake Missoula Flooding in Eastern Washington.” Geological Society of America Special Paper 144. Baker, V. R., and Bunker, R. C. (1985). Cataclysmic late Pleistocene flooding from glacial Lake Missoula: A review. Quaternary Science Reviews 4, 1-41. Beget, J. E. (1986). Comment on “Outburst Floods from Glacial Lake Missoula.” by G. K. C. Clarke, W. H. Mathews, and R. T. Pack. Quaternary Research 25, 136-138. Bjomsson, H. (1974). Explanation ofjokulhlaups from Grimsviitn, Vatnajokull, Iceland. Jiikull24, l-26. Bretz, J. H. (1932). “The Grand Coulee.” American Geographical Society Special Publication 15. Bretz, J. H. (1969). The Lake Missoula floods and the Channeled Scabland. Journal of Geology 77, 505-543. Bretz, J. H., Smith, H. T. U., and Neff, G. E. (1956). Channeled Scabland of Washington-New data and interpretations. Geological Society of America Bulletin 67, 957- 1049. Chambers, R. L. (1971). “Sedimentation in Glacial Lake Missoula.” Unpublished M.S. thesis, University of Montant, Missoula. Chambers, R. L. (1984). Sedimentary evidence for multiple glacial Lakes Missoula. Za “Northwestern Montana and Adjacent Canada, 1984 Field Conference and Symposium,” pp. 189- 199. Montana Geological Society, Billings. Clague, J. J. (1975). Sedimentology and paleohydrology of late Wisconsinan outwash, Rocky Mountain Trench, southeastern British Columbia. In “Glaciofluvial and Glaciolacustrine Sedimentation” (A. V. Jopling and B. C. McDonald, Eds.). pp. 223-237. Society of Economic Paleontologists and Mineralogists Special Publication 23. Clague, J. J., Armstrong, J. E., and Mathews, W. H. (1980). Advance of the late Wisconsin Cordilleran Ice Sheet in southern British Columbia since 22,000 yr B.P. Quaternary Research 13, 322-326.

Clague, J. J., and Mathews, W. H. (1973). The magnitude of jbkulhlaups. Journal of Glaciology 13, 501-504. Clarke, G. K. C., Mathews, W. H., and Pack, R. T. (1984). Outburst floods from glacial Lake Missoula. Quaternary

Research

22, 289-299.

Cowan, E. A., and Powell, R. D. (1986). “Deposition of Cyclically Interlaminated Sand-and-Mud in an Ice Proximal Glacimarine Environment,” p. 74. American Quaternary Association (AMQUA), 9th Biennial Conference, Program and Abstracts. District Engineers (1934). “Columbia River and Minor Tributaries.” U.S. House of Representatives, 73rd Congress, 1st Session, House Document 103, Vol. II. Flint, R. F. (1935). Glacial features of the southern Okanogan region. Geological Society of America Bulletin 46, 169-194. Flint, R. F., and Irwin, W. H. (1939). Glacial geology of Grand Coulee Dam, Washington. Geological Society of America Bulletin 50, 661-680. Hanson, L. G. (1970). “The Origin and Development of Moses Coulee and Other Scabland Features on the Waterville Plateau, Washington.” Unpublished Ph.D. dissertation, University of Washington, Seattle. Hanson, L. G., Kiver. E. P., Rigby. J. G.. and Stradling, D. F. (1979). “Surficial Geologic Map of the Ritzville Quad, Washington.” Washington Division of Geology and Earth Resources Open-File Report 79-10, scale 1:250,000. Jones, F. O., Embody, D. R., and Peterson, W. L. (1961). “Landslides along the Columbia River Valley, Northeastern Washington.” U.S. Geological Survey Professional Paper 367. Mathews, W. H. (1973). Record of two jbkulhlaups. In “Symposium on the Hydrology of Glaciers.” International Association of Scientific Hydrology Publication 95, pp. 99- 110. Richmond, G. M. (in press). Tentative correlation of deposits of the Cordilleran ice sheet in the Northern Rocky Mountains. In “Quaternary Glaciations in the United States of America, Part I of Quaternary Glaciations in the Northern Hemisphere.” Pergamon. London. Rubin. D. M., and McCulloch, D. S. (1980). Single and superimposed bedforms: A synthesis of San Francisco Bay and flume observations. Sedimentary Geology 26, 207-231. Singer, M. P., and Craig, R. G. (1984). Volume-areadepth relations in the Lake Missoula basin. Geological Society of America Abstracts u’ith Programs 16, 333. Thorarinsson, S. (1939). The ice-dammed lakes of Iceland with particular reference to their values as indicators of glacier oscillations. Geografiska Annaler 21, 216-242. Waitt. R. B., Jr. (1977). Missoula flood suns Oka-

LAKE

MISSOULA

nogan lobe. Geological Society of America Abstracts with Programs 9, 770. Waitt, R. B., Jr. (1980). About forty last-glacial Lake Missoula jokulhlaups through southern Washington. Journal of Geology 88, 653-679. Waitt, R. B., Jr. (1982). Surlicial deposits and geomorphology. In “Geologic Map of the Wenatchee 1: 100,000 Quadrangle, Central Washington” (R. W. Tabor et al.. Authors). U.S. Geological Survey Miscellaneous Investigations Map I-131 1, pp. 15-21. Waitt, R. B., Jr. (1984). Periodic jokulhlaups from Pleistocene Lake Missoula-New evidence from varved sediment in northern Idaho and Washington. Quaternary Research 22, 46-58.

FLOODS

201

Waitt, R. B. (1985). Case for periodic, colossal jokulhlaups from Pleistocene Lake Missoula. Geological Society of America Bulletin 96, 1271-1286. Waitt, R. B., Jr.. and Thorson, R. M. (1983). The Cordilleran Ice Sheet in Washington, Idaho, and Montana. In “Late Pleistocene Environments” (S. C. Porter. Ed.). In “Late Quatemary Environments of the United States” (H. E. Wright, Jr., general Ed.), pp. 53-70. Univ. of Minnesota Press, Minneapolis. Waters, A. C. (1933). Terraces and coulees along the Columbia River near Lake Chelan, Washington. Geological Society of America Bulletin 44, 783-820.