Depositional Record of a Glacial-Lake Outburst: Glacial Lake, Souris, North

Depositional record of a glacial-lake outburst: Glacial Lake Souris, North Dakota

MARK L. LORD Department of Geology and Department of Environmental Science, Allegheny College, Meadville, Pennsylvania 16335

ABSTRACT

Glacial Lake Souris, located in what is now North Dakota, was Continuous, low-density turbidity currents caused by the influx of inundated by about 74 km3 of sediment-laden water from the outburst sediment-laden flows and by residual currents from high-density flows of Glacial Lake Regina (Saskatchewan). Glacial-lake outbursts were characterized the outburst event in Lake Souris. Most high-density common during the late Pleistocene, but detailed studies of outburst- flows resulted from the initial influx of the outburst or from continu- deposited sediments in lake environments have not been done. The ous avalanching as a result of rapid deposition. purpose of this paper is to document the sedimentologic effects of the Thick, homogenous sections of outburst sediments; a gradual, outburst on Lake Souris and to develop a depositional model. systematic fining of sediments distally from the Souris spillway; and The outburst-deposited sediments in Lake Souris are grouped the scarcity of silt-sized and clay-sized sediment indicate that there into three lithofacies on the basis of textural analyses of subsurface was little decrease in flow energy during the outburst. Nonrepetitive samples and surficial mapping: (1) sand, (2) matrix-rich gravel, and vertical sequences, showing little textural variation, indicate that there (3) matrix-deficient gravel. Sand is the most common sediment and was little migration of individual depositional processes. The de- generally overlies other lithofacies. Lignite particles are present in scribed characteristics may be used to identify and interpret other most sands and, at depth, are significantly larger than other particles. outburst events in similar settings and to help evaluate the Pleistocene Matrix-rich, lignite-bearing gravel generally occurs at the base of the sedimentary record in the marine offshore zone. outburst sections. Matrix-deficient gravel occurs near the ground surface, overlying other outburst lithofacies, in the inlet area. INTRODUCTION Deposition of the outburst sediment is attributed to three main processes: low-density turbidity currents (the predominant process), Most major stream valleys in the glaciated midcontinent region origi- high-density turbidity currents, and aggradation in braided rivers. nated from or were significantly altered by glacial-lake outburst flows

MOOSE LAKE HIND MOUNTAIN

SOURIS

Figure 1. Regional glacial lakes and glacial-lake spillways (names in italics) related to Glacial Lake Souris.

- 4 7

104 100° 96°

Geological Society of America Bulletin, v. 103, p. 290-299, 13 figs., 1 table, February 1991.

290

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021 DEPOSITIONAL RECORD OF GLACIAL-LAKE OUTBURST, NORTH DAKOTA 291

(Kehew and Lord, 1986, 1987). Most dams impounding proglacial lakes along the southern margin of the Laurentide Ice Sheet were highly unsta- ble and often failed, causing sudden, catastrophic discharges of water. Spillways and spillway deposits that formed by glacial-lake outbursts have been extensively studied. Detailed studies of the sedimentology of outburst-deposited lake sediment have not been done, however. These deposits are the topic of this paper. The information presented provides further evidence in support of the hypothesis that glacial-lake outbursts were an important and, in many places, the predominant fluvioglacial process during the late Pleistocene. Specifically, this study reports the depositional effects of the Glacial Lake Regina outburst on the downstream Glacial Lake Souris. The objec- tives of this paper are (1) to document the pattern of sediments deposited by outbursts into Glacial Lake Souris and (2) to present a depositional model for these outburst flows. Textural and stratigraphic data presented on these sediments provide detailed information about areal sediment patterns and vertical sequences. These data are used to develop a depositional model for the outburst sequences and to support an interpretation of deposition dominated by low-density and high-density flows during a short-lived event.

GEOLOGIC SETTING

Regional Geology

The Glacial Lake Souris plain is located in north-central North Da- kota and extends slightly into southwest Manitoba (Fig. 1). Glacial Lake Souris, which covered about 6,000 km2, was one of several proglacial lakes along the southwestern margin of the Laurentide Ice Sheet during the Wisconsinan glaciation. Pertinent to this study are the lakes and spillways that were connected to Lake Souris. Glacial Lake Regina (Fig. 1) drained Figure 2. General geologic map of the Glacial Lake Souris plain; through the Souris and Des Lacs spillways (Christiansen, 1956); this sys- see Figure 1 (modified from Lord, 1988). Outer line on map marks tem terminates at Lake Souris. The final drainage of Lake Souris, triggered limit of Lake Souris. by the inundation of flows from the Glacial Lake Regina outburst, occurred northward, through the Souris-Hind spillway into Glacial Lake Hind (Kehew and Clayton, 1983). Lake Hind subsequently drained through the Pembina spillway into Lake Agassiz. Glacial drift covers the region and ranges in thickness from a few meters to about 60 m. Directly beneath the drift are lignite-bearing, poorly

Figure 3. Subsurface contour map of the top of the diamicton underlying the southern half of the Glacial Lake Souris plain; sea level is datum. The outlines of the southern half of Lake Souris and the Souris spillway are marked for reference; the same lake outline is used in subsequent related figures.

100"

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021 292 M. L. LORD

TABLE 1. COMPARISON OF TEXTURES AND VOLUMES OF MATERIAL ERODED VERSUS REDEPOSITED, WITHIN THE SOURIS AND DES LACS SPILLWAYS, MEAN AND MEDIAN BY THE GLACIAL LAKE REGINA OUTBURST (KEHEW AND LORD, 1987)

Sediment eroded Outburst Percent from spillways sediment deposited material within spillways redeposited Volume in spillway (km ) Volume (km )

Clay & silt 67 17.4 2 0.01 0.1 Sand 28 7.3 17 0.13 1.8 Gravel 5 1.3 81 0.61 46.1

-1-0 1-2 2-3 3-4 4-5 HI INTERVAL IMEAN MEDIAN indurated, fine-grained Paleocene and Cretaceous bedrock formations (Clayton, 1980). The sediments of the Lake Souris basin (Fig. 2) consist of diamicton (mostly till), overlain by laminated silt and clay, overlain in turn SORTING by sand or gravel (Bluemle, 1982, 1985). Thè surface of the diamictón underlying the Lake Souris plain forms two shallow basins with 15 to 20 m of relief, separated by a linear divide that trends northwest-southeast (Fig. 3). The contact between the clay and silt and the overlying sand and gravel, of outburst origin, is sharp and nongradational.

Glacial Lake Regina Outburst

The hydraulic characteristics of the Lake Regina outburst flows,

summarized below, strongly influenced sediment transport and deposition <.35 S5-.5 .5-.71 .71-1 1-2 2-4 PHI INTERVAL

Figure 5. Histograms of selected textural parameters for all samples (n = 169); (A) mean and median, (B) sorting. Sorting is shown Using the Inclusive Graphic Standard Deviation classification scale (Folk, 1980).

in Glacial Lake Souris. Discharges from the catastrophic drainage of Lake Regina flowed down the Souris and Des Lacs spillways (Fig. 1) (Kehew, 1982). Ah estimated 7.4 x 1010 m3 of water was released within a few weeks from Lake Regina (Kehew and Clayton, 1983). The outburst flows carved out the entire Souris spillway upstream from the confluence with the Moose Mountain spillway and markedly enlarged the Des Lacs and Souris spillways downstream from this confluence (Kehew, 1982; Lord and Kehew, 1987) (Fig. 1). Palebhydraulic estimates for the Regina outburst indicate discharges of 5.8 x 104 to 8.2 x 10s m3s and velocities of 2.9 to 11.7 ms"1 were achieved (Lord and Kehew, 1987). Because the sediments into which the spillways are eroded are generally poorly consolidated and fine grained, and because of the erosive power of the outburst flows, the outburst flows probably were hyperconcentrated (flow transitional between clear water and debris flow) (Lord arid Kehew, 1987). Volumetric comparisons between the sediment eroded by the Regina outburst and that redeposited (Table 1) show that only a minor percentage was redeposited within the spillways (Kehew and Lord, 1987). These data have important implications for Lake Souris because it received the sediment-charged water; an estimated 25 km3 of sediment, consisting of about equal parts sand, silt, and clay, was delivered to Lake Souris.

SAMPLE COLLECTION AND ANALYSIS

Figure 4. Locations and identification numbers of sample holes Most samples were collected from 61 holes drilled by the North drilled for study. Labeled lines show locations of cross sections in Dakota Geological Survey's truck-mounted hollow-stem auger (Fig. 4). Figure 8. Sediment samples were collected directly from auger flights at about 0.75-

m intervals and at noticeable lithologic changes. Drilling continued until A. TEST HOLE 12, DEPTH • 3.8 m either clay or till was reached (from about 2 to 28 m, average about 10 m). Although the sedimentary structures of most samples were destroyed by drilling, careful drilling technique permitted pure samples to be collected with excellent depth control. Samples from test holes were chosen for textural analyses at about 1.5-m intervals and at noticeable textural changes. The textures of 169 samples were determined using standard settling-tube techniques (Gibbs, 1974). Coarse-grained samples were analyzed using standard sieve techniques (Folk, 1980) with screens at quarter-phi intervals. Sample textural statistics were calculated using moment measures (LeFever, 1986).

DESCRIPTION OF OUTBURST SEDIMENTS

General -3.0 -2.0 -1.0 0.0 1.0 2.0 4.0 GRAIN SIZE IN PHI The outburst sediments in the Lake Souris basin comprise two dis- tinct facies: sand and gravel. The gravel facies may be divided into two subfacies: matrix-rich gravel and matrix-deficient gravel. Exposures of the outburst-deposited sediments in Lake Souris are very scarce except for gravels near the lake basin inlet. On the whole, the outburst sediments are B. TEST HOLE 30, DEPTH - 13.0 m composed of fine- to medium-grained sand (Fig. 5A) and are generally well sorted to moderately well sorted (Fig. 5B). I 35 N Sand Facies D 30 I V The sand facies is the most abundant and comprises well-sorted to I 25 very well-sorted, lignite-bearing sands. The lignite content typically ranges D U 20 from 2% to 10%. Outsized lignite clasts (as large as 3 cm) occur in other- A L wise well-sorted sands in about a third of the holes augered (Fig. 6). The 15 sands typically have a unimodal size distribution (Fig. 7A), but generally W E are bimodal where outsized lignite is present. I 10 G Most sand facies samples appear massive, but some exhibit indistinct H layering from 2 to 30 mm thick. Where exposed in the southwestern part T of the Lake Souris plain, the sediment consists of fine and very fine sand and exhibits climbing ripple cross-laminations interbedded with planar •2.0 -1.0 0.0 1.0 2.0 beds about 1 cm thick. GRAIN SIZE IN PHI

C. TEST HOLE 5, DEPTH - 0.8 m

4.0 -3.0 -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 GRAIN SIZE IN PHI

Figure 7. Histograms of individual frequency percent grain size Figure 6. Sediment sample of sand lithofacies with abundant out- for representative samples of lithofacies (A) sand, (B) matrix-rich sized lignite (darker clasts); sample from hole 9 at depth 4.5 m. gravel, (C) matrix-deficient gravel.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021 294 M. L. LORD

H-3 2 H 31 " 460 m H~1 H--m 439 m 455 m H 11 H-49 Figure 8. Geologie cross 448 m sections of the Lake Souris basin sediments oriented (A-A') parallel to flow direction, (B-B') perpendicular to flow direction. The diamicton and the laminated silt and clay were deposited 5 m prior to inundation of outburst waters. See Figure 2 for specific locations of 10 km B' sections. H-32 H-50 H-33 H-56 H-13 H-57 460 m 461 m 461 m 457 m 458 m 458 m

LAKE / SOURIS /

/ / /

LITHOFACIES

SAND ill DIAMICTON 5 m MATRIX-RICH GRAVEL [//] INDISTINCT BEDDING

5 km MATRIX-DEFICIENT GRAVEL EZ] OUTSIZE LIGNITE CLASTS

LAMINATED SILT & CLAY H2 SANDY

Gravel Facies: Matrix-Rich Gravel STRATIGRAPHIC RELATIONSHIPS

Matrix-rich gravels account for the majority of the gravel facies. The stratigraphic relationships in the Lake Souris basin are shown in These gravels contain an abundance of fine-grained material, are poorly two geologic cross sections, one parallel and one perpendicular to the sorted, and commonly contain lignite clasts. Particles in the matrix-rich Souris spillway (Fig. 8). The most common sequence is sand overlying gravels range from silt to cobble size. These gravels typically exhibit a laminated silt and clay or till. Matrix-rich gravel occurs almost exclusively polymodal distribution with no strong modal group (Fig. 7B). at the bottom of the outburst sediments and in most places is overlain by Lignite content typically ranges from 3% to 10% of the clasts, but, in sand. The matrix-deficient gravels are relatively scarce and generally occur some samples, is as much as 90% of the gravel-sized particles. Clasts of near the Souris spillway at or near the land surface. The sediments show an other easily erodible lithologies, such as till and laminated clay, also occur overall decrease in grain size away from the lake inlet (Fig. 2). within the matrix-rich gravels but are rare. No sedimentary structures Detailed vertical sequences in the outburst sediments are shown in could be recognized in the matrix-rich gravels. plots of textural parameters versus depth (Fig. 9). About half of the sample holes show little textural variation with depth (Fig. 9A); that is, the mean Gravel Facies: Matrix-Deficient Gravel grain size varies less than bout 1.5 phi over the depth of the hole sampled. Sediments in most sections show a gradual coarsening-upward trend. Matrix-deficient gravels generally lack silt and very fine sand-sized Some sections, such as at hole 13 (Figs. 4, 8B, and 9A), are remarkably particles and contain clasts as large as cobbles. Lignite clasts are scarce. homogeneous over their entire depth. About 20% of the holes have coarse- Most size distributions of these sediments are bimodal (Fig. 7C). Poor grained sediment at the base of the outburst sediments overlain by homo- exposures of this facies, along terraces adjacent to the Souris spillway near geneous sections (Fig. 9B). Two holes near the Lake Souris inlet have its inlet to Glacial Lake Souris, show tabular cross-bedding with coset coarsening-upward trends (Fig. 9C). The remaining holes have irregular thicknesses of 0.5 to 1 m. grain size trends.

TEST HOLE

Figure 9. Selected plots of depth versus textural-parameter values for (A) a homogeneous section, (B) a homogeneous section with a basal coarse-grained segment, (C) a coarsening- upward section.

- 5-4-3- 2- 1 0 1 2 3 4 S PHI SIZE PHI SIZE COARSEST 1% MEAN SORTING

To show areal variations of sediments in the Lake Souris basin, selective deposition did not occur and that sediments deposited were ap- textural parameters (mean, coarsest 1%, sorting, and skewness) were aver- proximately equally transportable. This is consistent with particle transport aged for each hole and then contoured. The maps of averaged mean in a concentrated basal dispersion followed by en masse deposition (His- (Fig. 10) and coarsest 1% exhibit similar trends: two coarse-grained areas cott and Middleton, 1979; Lowe, 1982). that coincide with the northern and southern fan locations and sediment Much more sand than gravel was delivered to Lake Souris by the sizes that gradually decrease away from the fans. The averaged sorting and outburst (Table 1); therefore, it should be expected that much of the sand skewness data show no significant trends. was deposited by high-density flows in a manner similar to that for the matrix-rich gravels. The strongest physical evidence for deposition of some INTERPRETATION OF DEPOSITIONAL PROCESSES sands from concentrated basal dispersion is the presence of outsized clasts, mostly lignite, at depth. Outsized clasts are an indicator of deposition from The analysis of the depositional processes active in the development grain flows (Stauffer, 1967). In a representative sample (hole 9: 6.9 m of the outburst-deposited sediment fan in Lake Souris includes lithologic, depth), the settling velocity of the 7 largest lignite clasts averaged about 3 sedimentologic, and stratigraphic evidence. Furthermore, depositional times faster than the fastest 1% of the remainder of the sample (about 17 processes are evaluated within the framework established by the probable versus 5 cm s~'); hence, the outsized lignite clasts were not in hydraulic conditions of the Lake Regina outburst, the geologic setting of Lake equivalence with the sand. Instead, the lignite probably "floated" up from Souris, and processes known to be active in analogous geologic settings. In the high-shear-stress zone at the base of the flow by dispersive pressure general, the Lake Souris fan is interpreted to have formed by deposition (Lowe, 1982; Eyles and others, 1987). from low- and high-density, sediment gravity flows and by deposition from braided rivers.

Sediment Gravity Flows

The abundance of thick, homogeneous sections of sand throughout much of the Lake Souris basin suggests deposition by turbidity currents (Gilbert, 1975; Lowe, 1982; Eyles, 1987; Liverman, 1987). The Lake Regina outburst was strongly erosive and consequently transported great amounts of sediment; the majority of sediment carried into Lake Souris consisted of silt and montmorillonite-rich clay (Table 1) (Kehew and Lord, 1987; Lord and Kehew, 1987). It is highly probable that the outburst flows entering Lake Souris would have had sufficient sediment concentrations to be low-density turbidity currents for their entire duration and, possibly at times, high-density flows. In addition to the continuous turbidity currents, conditions on the fan of sediment prograding into Lake Souris would have been conducive to the formation of sediment gravity flows. Rapid sedimentation, especially of coarse-grained clasts, would have occurred near the inlet. Moreover, deposited sediments must have been subjected to continual shear stresses applied by the high-discharge flow, especially in areas of channelized flow. Such shear stresses facilitate the development of high dispersive pressures in the sediments (Hein, 1982). The origin of the matrix-rich gravel lithofacies is most easily explained by deposition from either high-density turbidity currents or modified grain flows. The long transport distance for the matrix-rich gravels from the inlet, the large depth at which this fades occurs, and the presence of fragile clasts support a density-flow origin (Houbolt and Jonker, 1968; Figure 10. Contour map of sample means averaged for each hole; Smith and Ashley, 1985; Weirich, 1989). The flatness of the individual values of basal coarse-grained segments are not included in the aver- frequency grain size curves for these sediments (Fig. 7B) indicates that aged values.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021 296 M. L. LORD

Braided Rivers turbulence and dispersive pressure. Probably matrix support also was important; waters with clay concentrations as little as 2% can support fine The gravel exposed at and near the surface of the Lake Souris basin, sand in suspension (Hampton, 1975). The relative predominance of low- the matrix-deficient gravel, is interpreted to have been deposited by density flows versus high-density flows, based on comparison of the total braided rivers because of (1) the braided-channel scars on the surface, thickness of sand without outsized lignite clasts with the total thickness of (2) the presence of tabular cross-bedding, and (3) the poor sorting and the matrix-rich gravel and the sand containing outsized lignite clasts, is bimodal distribution of the sediments (Fahnestock, 1963). Other informa- about 4 to 1. Low-density turbidity currents, therefore, are interpreted to tion also indicates that braided rivers should have been present. The water have been the predominant depositional process in Lake Souris. level in Lake Souris probably was not higher than 457 m at the time of the outburst (Fig. 11, profile B) (Lord, 1988), but near the inlet, braided- DISCUSSION OF FAN DEVELOPMENT channel scars underlain by gravel occur at elevations up to about 468 m. The outburst flows probably emerged from confined conditions within the Indirect evidence may be used to establish a general basis for the Souris spillway (Fig. 11, profile A) to unconfined, but not subaqueous, sequence of fan development. Among the sediments deposited by sediment conditions within the Lake Souris basin (Fig. 11, profiles B-H). In the gravity flows, the coarse-grained sediments occur predominantly at the unconfined reach, upslope from the 457 m lake level, braided-river depos- bottom of vertical sections. Furthermore, the mostly coarse-grained sedi- its accumulated as a far-shaped deposit. Because of the large differences in ments adjacent to the Souris spillway must have been deposited before water levels between the spillway and lake, it is probable that this interme- outburst discharges had declined substantially, because they were subse- diate area was entirely inundated during the early stages of the outburst quently deeply incised by the outburst flows. This relationship is depicted before braided rivers developed. In an analogous setting, the Charlotte in the topographic profiles across the Souris spillway (Fig. 11), especially Lobe delta, Alaska, was inundated by flows from the sudden release of in the vicinity of the inlet and outlet where the channel bottom was incised water dammed within the Martin River Glacier (Erickson, 1967). far beneath the top of the outburst sediments.

Summary Vertical Sequences

The majority of the coarse-grained sediments in Lake Souris are Vertical sections in the outburst deposits may be used to interpret the interpreted to have been deposited by sediment gravity flows; the excep- predominance of different sedimentation patterns present in the Lake tions are the braided-river deposits. Sediment was transported predomi- Souris basin. The prevalent vertical trend is a simple sequence that exhibits nantly in suspension. Sediment support mechanisms included fluid little textural variation (Fig. 9A). This trend is most common in areas distal from the Souris spillway. The homogeneity of these sequences indicates little variation in the character of the depositional flows. The simplest ELEVATION IN METERS WEST EAST 490-1 450- 440- 430

460-H 450- 440- 430-

450- LAKE SOURIS WATER 440- LEVEL "457 m 430 HIGH-WATER LEVELS 460-1 450 IN SPILLWAYS 440- 430J

Figure 11. Topographic profiles, perpendicular to the Souris spillway, from Lake Souris inlet to outlet area; probable water levels are indicated (from Lord, 1988).

Figure 12. Generalized distribution of deposits by three depositional processes listed. The boundaries are drawn based on lithofacies present in sample holes and the geologic map of Lake Souris (Fig. 2).

LOW-DENSITY TURBIDITY FLOW DEPOSITS

X/\ HIGH-DENSITY FLOW DEPOSITS

BRAIDED RIVER DEPOSITS

explanation for these sediments is deposition dominated by a single process, probably low-density turbidity currents. The second most common vertical sequence in the outburst-deposited sediments shows a coarse-grained segment at the base of the section, deficient gravels. These sections are interpreted to have resulted from the overlain by finer-grained sediments that exhibit little textural variation progradation of braided-river deposits into Lake Souris. (Fig. 9B). The upper parts of these sections are similar to the simple, homogeneous sequences described above and, therefore, are interpreted to Depositional Model be similar in origin. The sediments composing the basal coarse-grained segments are matrix-rich, lignite-bearing gravels that as discussed pre- A model of deposition for the Glacial Lake Souris sediment fan viously, are thought to have been deposited by high-density flows. These constructed by the outburst flows from Glacial Lake Regina must explain vertical sections, then, represent two modes of deposition, initial outburst three major observations. First, there is little evidence that flow energy sedimentation from high-density flows succeeded by sedimentation from decreased with time. Only a small amount of silt-sized sediment was low-density turbidity currents. deposited by the outburst in Lake Souris and virtually no clay-sized sedi- Vertical sections that show irregular patterns of sedimentation are ment, yet most of the sediment delivered to Lake Souris by the outburst uncommon; they presumably represent deposition by two or more proc- flows was of silt and clay size (Table 1). Second, the majority of sections esses. The two vertical sequences (holes 5 and 35) that show a marked sampled have simple textural patterns, indicating little migration of speci- coarsening-upward trend (Fig. 9C) are located in the vicinity of matrix- fied depositional processes. Third, the depositional processes interpreted to

N GLACIAL LAKE SOURIS

DIAGRAMMATIC LONGITUDINAL SECTION FLOW

Figure 13. Diagrammatic longitudinal section through the Lake Souris basin, depicting re- DIAMICTON (D) SAND (S) lationships of sediments deposited by different GLACIAL DEPOSITS • LOW-DENSITY TURBIDITY CURRENT processes. ^ISILT a CLAY [=>~| SAND (S) WITH OUTSIZE LIGNITE ZzJ RYTH MITES (SCI) HIGH-DENSITY FLOW DEPOSITS

SAND 8 GRAVEL(Gmd) SAND DUNES BRAIDED RIVER DEPOSITS ^o] SAND a GRAVEL (Gmr) HIGH-DENSITY FLOW DEPOSITS

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021 298 M. L. LORD

have been active in the formation of the fan are braided rivers, low-density FURTHER STUDY AND IMPLICATIONS turbidity currents, and high-density turbidity currents or modified grain flows. A map showing the distribution of sediments deposited by the During the Pleistocene, hundreds of channels were deeply incised by different processes (Fig. 12), constructed by keying lithofacies to their glacial-lake outbursts along the midcontinent margins of the Laurentide interpreted origin, is consistent with the overall interpretation of the differ- Ice Sheet (Kehew and Lord, 1987). Whereas flood waters initially were ent depositional processes. Braided-river deposits occur along the Souris dumped into downstream glacial lakes, such as Lake Souris, the ultimate spillway, especially near the inlet, and at the mouths of the overflow destination of outburst flows was the ocean. Because many of the dis- channels (Kehew, 1982) along the western edge of Lake Souris (Fig. 12). charges of major outbursts funneled into the Mississippi Valley, sedimenta- The high-density flow deposits occur in the topographic lows and away tion by outburst flows probably dominated deposition on the Mississippi from the margins of the lake basin, and the low-density flow deposits are fan and Louisiana offshore zone during glacial intervals (Kehew and oth- widespread throughout the basin (Fig. 12). ers, 1986; Kehew and Lord, 1987). Outburst deposits may be recognized The coarse-grained sediments of Lake Souris may be explained by by their most striking characteristics: (1) the constancy of vertical sections, various types of depositional processes resulting from rapid sedimentation commonly as much as 10 m thick; (2) the pervasiveness of anomalously by sediment-laden outburst flows that catastrophically inundated the Lake coarse-grained sediments over most of the lake basin; and (3) the abun- Souris basin. The three depositional processes responsible for the Lake dance of sediments interpreted to have been deposited by high-density Souris fan probably were active concurrently for at least a part of the flows caused by continual sedimentation rather than by discrete slumps. outburst event. Sedimentation at the inlet was primarily by a braided river; Fans formed by outburst flows have been recognized in Glacial Lakes much of the deposition occurred above the pre-outburst water level of Agassiz, Dakota, and Hind (Fig. 1) because they occur at the mouths of Lake Souris (Figs. 8A and 11). This part of the fan is envisioned to have glacial-lake spillways and because their surficial geology is similar to that been similar to a "supra-aquatic delta," a term used by Hjulstrom (1952) of Lake Souris (Kehew and Clayton, 1983; Kehew and Lord, 1987). The to describe glaciofluvial sequences he studied in Scandinavia and Iceland. recognition of outburst deposits on the marine offshore zone is considera- Supra-aquatic deltas accumulate above lake level and result from rapid bly more complicated. Nonetheless, on the Mississippi fan, discrete, large sedimentation and high competence (Church and Gilbert, 1975). fresh-water pulses indicated by negative oxygen-isotope anomalies (Emili- The outburst flows, upon entering Lake Souris, probably caused ani and others, 1978; Leventer and others, 1982), anomalously coarse- continuous, rapid sedimentation from both high- and low-density currents. grained sediments, and channelized density-flow deposits (Feeley and The constancy of the processes is attested to by the simple, nonrepetitive others, 1985) suggest that sedimentation from glacial-lake outbursts in the sediment sections. For example, most of the high-density flow deposits midcontinent region did affect the offshore zone (Kehew and others, occur at the bottom of the outburst-deposited sections and are not repeated 1986). Also, deep-sea gravels in the Cascadia Channel, off the northwest above (Fig. 9B). The majority of the density current deposits, therefore, coast, are interpreted to have been deposited by high-density turbidity were probably not generated by discrete slumps as is most common in currents that were transported hundreds of kilometers offshore by outburst glacial lakes and on submarine fans (Smith and Ashley, 1985; Shanmu- flows from Glacial Lake Missoula (Griggs and others, 1970). gam and Moiola, 1988). Whether the high-density flows resulted directly If detailed studies of offshore fans in North America, Europe, and from the initial inflow or from continuous avalanching due to rapid sedi- Asia can differentiate glacial-lake outburst deposits from other glacial- mentation (Allen, 1968) is uncertain. Some of the low-density turbidity meltwater deposits, the relative roles of each as a glaciofluvial process may currents may have been a residual stage of a high-density flow or a grain be assessed. Furthermore, if glacial-lake outbursts are found to be a signifi- flow (Middleton and Hampton, 1976; Gilbert, 1975; Lowe, 1979, 1982; cant process on a global scale, it may assist the interpretation of ice Hein, 1982; Weirich, 1989). volumes and climatic variations based on oxygen-isotope data. Glacial Low-density turbidity currents became more predominant through- lakes, by temporarily storing glacial meltwater (as long as hundreds of out the duration of the outburst flows, as evidenced by sedimentary record years), act as a buffer between glacier melting and delivery of that melt- and inferences based on the sediment-concentration hydrograph (Lord and water to the ocean. These implications have been discussed with respect to Kehew, 1987) for outbursts and geology of the spillways. High-density subglacial floods (Shaw, 1989) and glacial-lake outbursts (Kehew and flow deposits still occurred, but were less common and probably were Lord, 1990; Shaw, 1990). slump-generated. A diagrammatic representation of the locations of the different types of deposits, which summarizes what has been discussed ACKNOWLEDGMENTS above, is given in Figure 13. The duration of the Lake Regina outburst is estimated to have been Several organizations have contributed valuable support to this study. very short, no more than a few weeks (Kehew and Clayton, 1983). The Most important has been the North Dakota Geological Survey, having brevity of this event is supported by the deficiency of silt-sized and provided research support, a drilling rig, and drillers. Arden Mathison, of clay-sized particles among the outburst sediments despite their abundance the U.S. Bureau of Reclamation, provided valuable data on sediments in material eroded from the spillways. From this, it may be concluded that from the Lake Souris plain. The University of North Dakota Graduate the retention time of outburst flows in Lake Souris must have been short School provided research support and two summer doctoral fellowships. and, accordingly, the catastrophic drainage of Lake Souris must have been Alan E. Kehew has continually provided valuable insights, encour- triggered before the majority of the silt- and clay-sized sediment had agement, and editorial comments that have contributed greatly to this sufficient time to settle out. Once the drainage of Lake Souris began, work. In addition, John R. Reid, Richard D. Lefever, Kenneth L. Harris, conditions would have changed from ponding to through-flow, thus per- and John P. Bluemle all contributed valuable ideas and editorial com- mitting the passage of the silt and clay through the basin. ments that have improved this paper.

REFERENCES CITED 1987, Glacial-lake outbursts along the mid-continent margins of the Lauremide Ice Sheet, in Mayer, L., and Nash, D., eds., Catastrophic flooding: Binghamton Symposia in Geomorphology (International Series, Number Allen, J.R.L., 1968, The diffusion of grains in the lee of ripples, dunes, and sand deltas: Journal of Sedimentary Petrology, 18): Boston, Massachusetts, Allen and Unwin, p. 95-120. v. 38. p. 621-633. 1990, Comment "Drumlins, subglacial meltwater floods, and oceanic responses" (Geology, v. 17, p. 853-856]: Bluemle, J. P., 1982, Geology of McHenry County, North Dakota: North Dakota Geological Survey Bulletin 74, Part I, Geology, v. 18, p. 479-480. and North Dakota State Water Commission County Groundwater Studies 33, Part 1,49 p. Kehew, A. K., Lord, M. L., and Lowrie, Allen, 1986, Tracing catastrophic glacial-lake outbursts from the southern margin — 1985, Geology of Bottineau County, North Dakota: North Dakota Geological Survey Bulletin 78, Part I, and of the Laurentide Ice Sheet to deposition along the Louisiana offshore: Geological Society of America Abstracts North Dakota State Water Commission County Groundwater Studies 35, Part 1,57 p. with Programs, v. 18, p. 653. Christiansen, E. A., 1956, Glacial geology of the Moose Mountain area, Saskatchewan: Saskatchewan Department of LeFever, R. D., 1986, Texture [unpub. computer program]: Grand Forks, North Dakota, University of North Dakota. Mineral Resources Report 21, 35 p. Leventer, Amy, Douglas, F. W., and Kennett, J. P., 1982, Dynamics of the Laurentide Ice Sheet during the last Church, Michael, and Gilbert, Robert, 1975, Proglacial fluvial and lacustrine environments, in Jopfing, A. V., and deglariation: Evidence from the Gulf of Mexico: Earth and Planetary Science Letters, v. 59, p. 11-17. McDonald, B. C., eds., Glariofluvia} and giaciolacustrme sedimentation: Society of Economic Paleontologists aod Liverman, D.G.E., 1987, Sedimentation in ice-dammed Hazard Lake, Yukon: Canadian Journal of Earth Sciences, v. 24, Mineralogists Special Publication 23, p. 22-100. p.1797-1806. Clayton, Lee, 1980, Geologic map of North Dakota; U.S. Geological Survey, scale 1:500,000. Lord, M. L., 1988, Surface geology of the Souris River map area. North Dakota Geological Survey Atlas Series, Map 4, Emiliani, Cesare, Rooth, Claes, and Stipp, J. J., 1978, The late Wisconsin flood into the Gulf of Mexico: Earth and Sheet 1, scale 1:250,000. Planetary Science Letters, v.41,p. 159-162. Lord, M. L., and Kehew, A. E., 1987, Sedimentology and paleohydrology of glacial-lake deposits in southeastern Erickson, Kirtb, 1967, Hydrology of the Charlotle Lobe delta: North Dakota Academy of Science Proceedings, v. 21, Saskatchewan and northwestern North Dakota: Geological Society of America Bulletin, v. 99, p. 663-673. p. 9-16. Lowe, D. R., 1979, Sediment gravity flows: Their classification and some problems of application to natural flows and Eyles, C. H., 1987, Glacially influenced submarine-channel sedimentation in the Yakataga Formation, Middleton Island. deposits, in Doyle, L. J., and Pilkey, O. H., Geology of continental slopes: Society of Economic Paleontologists and Alaska-. Journal of Sedimentary Petrology, v. 57, p. 1004-1017. Mineralogists Special Publication 27, p. 75-82. Eyles. Nicholas, Clark, C. M„ and Clague, J. J., 1987, Coarse-grained sediment gravity flow facies in a large supraglacial 1982, Sediment gravity flows: II. Depositional models with special reference to the deposits of high-density lake- Sedimentology, v. 34, p. 193-216. turbidity currents: Journal of Sedimentary Petrology, v. 52, p. 279-297. Fahnestock, R. K„ 1963, Morphology and hydrology of a glacial stream: White River, Mount Rainier, Washington: U.S. Middleton, G. V., and Hampton, M. A., 1976, Subaqueous sediment transport and deposition by sediment gravity flows, Geological Survey Professional Paper 422-A. 70 p. in Stanley, D. J., and Swift, D.J.P., eds.. Marine sediment transport and environmental management: New York, Feeley. M. H., Buffer, R. T., and Bryant, W. R„ 1985, Deposition units and growth pattern of the Mississippi fan, in John Wiley and Sons, p. 197-218. Bouma, A. H., Normark. W. R., and Barnes, N. E., Submarine fans and related turbidite systems: New York, Shanmugam, G., and Moiola, R. J., 1988, Submarine fans: Characteristics, models, classification, and reservoir potential: Springer-Verlag, p. 253-257. Earth-Science Reviews, v. 24, p. 383-428. Folk, R. L., 1980, Petrology of sedimentary rocks: Austin, Texas. Hemphill Publishing Co., 185 p. Shaw, John, 1989, Drumlins, subglacial meltwater floods, and oceanic responses: Geology, v. 17, p. 853-865. Gibbs, R. J., 1974, A settling tube system for sand-size analysis: Journal of Sedimentary Petrology, v. 44, p. 583-588. 1990, Reply "Drumlins, subglacial meltwater floods, and oceanic responses'* [Geology, v. 17, p. 853-856]: Gilbert, Robert, 1975. Sedimentation in Ullooet Lake, British Columbia: Canadian Journal of Earth Sciences, v. 12, Geology, v. 18, p. 479-480. p. 1697-1711. Smith, N. D., and Ashley, G. M., 1985, Proglacial lacustrine environment: Physical process, in Ashley, G. M., Shaw, John, Griggs, G. B., Kulm, L. D., Waters, A. C., and Fowler, G. A., 1970, Deep-sea gravel from Cascadia Channel: Journal of and Smith, N. D., eds.. Glacial sedimentary environments: Society of Economic Paleontologists and Mineralogists Geology, v. 78. p. 611-619. Short Course 16, p. 135-215. Hampton. M. A., 1975, Competence of fine-grained debris flows: Journal of Sedimentary Petrology, v. 45, p. 834-844. Staviffer, P. H., 1967, Grain-flow deposits and their implications, Santa Ynez Mountains, California: Journal of Sedimen- Hein, F. J., 1982, Depositional mechanisms of deep-sea coarse clastic sediments. Cap Enrage Formation, Quebec: tary Petrology, v. 37, p. 487-508. Canadian Journal of Earth Sciences, v. 19, p. 267-287. Weirich, F. R., 1989, The generation of turbidity currents by subaerial debris flows, California: Geological Society of H/scofi, R. N., and Middleton, G. V., 1979, Depositional mechanics of thick-bedded sandstones at the base of a submarine America Bulletin, v. 101, p. 278-29). slope, Tourelle Formation (Lower Ordovician), Quebec, Canada, in Doyle, L. J., and Pilkey, O. H.. Geology of continental slopes: Society of Economic Paleontologists and Mineralogists Special Publication 27, p. 307-326. Hjulsirom, Filip, 1952, The geomorphotogy of the alluvial outwash plains (sandurs) of Iceland, and the mechanics of braided rivers: International Geographical Union Congress, 17th, Washington, D.C., Proceedings, p. 337-342. Houboult, J.J.H.C., and Jonker, J.B.M., 1968, Recent sediments in the eastern part of the Lake of Geneva (Lac Leman): Geologie en Mijnbouw, v. 47, p. 131-148. Kehew, A. K„ 1982, Catastrophic flood hypothesis for the origin of the Souris spillway, Saskatchewan and North Dakota: Geological Society of America Bulletin, v. 93, p. 1051-1058. Kehew, A. K., and Clayton, Lee, (983, Late Wisconsinan catastrophic floods and development of the Souris-Pembina spillway system, in Teller, J. T., and Clayton, Lee. eds., Glacial Lake Agassiz: Geological Association of Canada Special Paper 26, p. 187-209. MANUSCRIPT RECEIVED BY THE SOCIETY AUGUST 7,1989 Kehew, A. K.. and Lord, M. L., 1986, Origin and large-scale erosional features of glacial-lake spillways in the northern REVISED MANUSCRIPT RECEIVED JULY 3,1990 Great Plains: Geological Society of America Bulletin, v. 97, p. 162-177. MANUSCRIPT ACCEPTED JULY 12,1990

Printed in U.S.A.

Downloaded from http://pubs.geoscienceworld.org/gsa/gsabulletin/article-pdf/103/2/290/3381219/i0016-7606-103-2-290.pdf by guest on 01 October 2021