Catastrophic Ice Age Floods in Colorado

Title: Catastrophic Ice Age Floods in Colorado: Glacial Outburst Floods Scoured the Upper Arkansas Valley

Author: Keenan Lee

Publication: The Outcrop, June 2012, p. 6-12

During the Pleistocene, three glaciers from the Sawatch Range flowed down contiguous tributary valleys to the Arkansas River near Granite, Colorado. The Lake Creek glacier probably pushed the river out of its channel, and the Clear Creek and Pine Creek glaciers
crossed the river and rammed into granite walls on the far side of the valley. These glaciers formed an ice dam about 670 ft high that blocked the Arkansas River and created a large lake about 600 ft deep that extended 12 miles upstream.

When the ice dam broke, the lake drained catastrophically. The outburst flood tore out the ends of the moraines and carried the detritus down the valley in a torrent of dirty water that deposited a sheet of flood boulders 60 ft thick in a matter of hours. Many flood boulders are tens of feet in diameter, and some can be found 150 ft up the valley wall.

At least four catastrophic floods swept the Upper Arkansas Valley, together called the  Three Glaciers Floods. Field evidence provides a pretty clear picture of the last two such floods, but only patches of older flood boulders attest to two earlier floods. The most recent, or fourth, flood is from Pinedale glaciers that reached their peak 16-22 ka (Young et al., 2011). The first flood was older than 760 ka, and the second flood was older than 640 ka. The third flood is tentatively assigned a Bull Lake age, 130 – 150 ka.

Figure 1. Upper Arkansas Valley

The Upper Arkansas Valley drains the Sawatch Range on the west and the Mosquito Range on the east (Fig. 1). Bounding walls of the valley are mostly Precambrian
granite, and the valley floor consists of thick Neogene Dry Union Formation sediments with a veneer of Quaternary alluvium.

Glenn Scott (1975) first recognized huge flood boulders in outwash he considered Pinedale age, and he attributed them to a flood from a glacial dam at Pine Creek (Scott, 1984). McCalpin et al. (in press) mapped lacustrine sediments and shoreline gravels from the glacial lake.

Three Glaciers Damsite
Lake Creek glaciers did not dam the Arkansas River. The map view in Figure 2 shows fairly complete end moraine loops without truncation, and the east side of the Arkansas Valley lacks evidence of glacial impact. Each advance of the glacier may have forced the river to the east, however, because the channel of the modern Arkansas River here is in granite, even though extensive and thick deposits of Neogene Dry Union Formation sand and gravel are only a few hundred feet to the west of the river.

June LS2
Figure 2. Distribution of moraines at the Three Glaciers damsite shows the southern two glaciers dammed the Arkansas River in both Bull Lake and Pinedale times, whereas the Lake Creek glacier apparently did not.

Clear Creek glaciers in both Bull Lake and Pinedale times built fairly equal, large, symmetrical lateral moraines and dammed the Arkansas River (Fig. 2). Both sets of moraines are clearly truncated at their lower ends. A distinctive, unusual landform was created by the glaciers of Clear Creek, herein called a glacial impact scar (or whamout). It is a clearly defined, very steep, concave cliff of unweathered granite cut into the wall on the opposite side of the Arkansas Valley (Fig. 3). The tremendous ice contact pressures may have induced fracturing and glacial plucking, and when the glacier dam failed, torrential discharges of dammed waters would have further scoured the wall, probably by kolk action. The crest of the Bull Lake left lateral moraine projects to an elevation of about 9,350 ft, or 470 ft above the river, and the Pinedale moraine similarly projects to about 9,360 ft.

Pine Creek glaciers were equally effective in damming the Arkansas River; both Pinedale and Bull Lake glaciers dammed the river. The Bull Lake moraines indicate pronounced flaring of the glacier consistent with impact against the far wall, both are truncated, and a well defined glacial impact scar formed (Figs. 2, 3). The Bull Lake glacier in Pine Creek merged with the glacier in Clear Creek, essentially buttressing a common ice dam, with the lake elevation determined by the Clear Creek glacier.

Figure 3. Glacial impact scars, or whamouts, from Clear Creek and Pine Creek glaciers that took “bites” out of the obstructing granite walls. Bottom photo is Pine Creek whamout.

At least two pre-Bull Lake glaciers dammed the Arkansas River, probably here at the Three Glaciers damsite. No evidence of these glacial dams remains, but no other locality in the Upper Arkansas Valley presents evidence of a glacial dam, no other locality is as favorable for a glacial dam as this site, and it is demonstrated that the last two glacial dams formed here.

Three Glaciers Lake
The dammed Arkansas River backed up to the north behind the glacial dam and created Three Glaciers Lake, which was about 600 ft deep at the dam and extended about 12 mi up the Arkansas Valley to the Malta Substation (Fig. 4). Although the lake probably filled and emptied at least four times, the two most recent lakes were at about the same elevation, and information about the two earlier ones is lacking. There are no prominent shorelines, as at Lakes Missoula and Bonneville, probably because there was no spillway from Three Glaciers Lake to maintain a constant lake elevation, and the very limited reach available to prevailing westerlies inhibited formation of energetic waves that would have formed beaches.

Figure 4. Three Glaciers Lake at about 9480 ft elevation. Squared dots show locations of ice-rafted boulder erratics.

Indirect evidence of the lake is provided by two large landslides along the shore of Three Glaciers Lake (Fig. 4). These likely occurred when the dam failed and the lake surface dropped catastrophically, trapping high hydrostatic pressures in saturated sediments that led to slope failure. This is analogous to the 1976 Teton Dam failure where the rapid drawdown in the 17-mi-long reservoir caused more than 200 landslides (Randle et al., 2000).

The Lake Creek glacier flowed into Three Glaciers Lake and calved off icebergs, which contained boulders. One iceberg grounded in shallow water along the eastern shore, at about 9,358 ft, leaving a cluster of ice-rafted boulder erratics (IRBEs) (Fig. 4). Two other IRBE clusters just north of Lake Creek are at about 9,384 ft and 9,402 ft.

McCalpin et al. (in press) mapped lacustrine sediments mantling slopes in the Box Creek embayment (Fig. 4) between about 9,340 ft and 9,480 ft elevation, which is the basis for estimating the Three Glaciers Lake elevation. They describe the sediments as “sand, silt, clay, and minor gravel deposited in shallow water below the shorelines of Three Glaciers Lake”. McCalpin et al. (in press) also mapped shoreline gravels on numerous flat hilltops planed off by wave action along the east shore of the lake between about 9,400 ft and 9,480 ft elevation.

Dam Failure and Catastrophic Floods
The mechanism of dam failure is unknown, but huge boulders extending for miles downstream make it clear that the floods resulted from catastrophic failure and rapid dumping of the lake. The most common failure mechanism for an alpine glacial dam with this configuration is outflow through a marginal breach along the ice – bedrock wall contact (Walder and Costa, 1996). As the level in Three Glaciers lake rose, subglacial leakage would most likely have occurred there, where the ice was thinnest. Enlargement leading to a subaerial breach would result from thermal erosion and ice calving (Walder and Fountain, 1997). Another possible mechanism for dam failure is flotation; when water depth reached about 90 percent of the thickness of the ice, the ice dam became buoyant, and when high-pressure water got under the ice dam, it would have destroyed the dam rapidly.

Peak discharge for the most recent flood has been estimated at 1.62 million cfs (Brugger et al., 2011), using conservative parameters. This is some 600 times greater than the historic average annual peak flood of 2750 cfs at the damsite (USGS, 2012). Three Glaciers Lake would have emptied in less than a day (Brugger et al., 2011).

The broad, flat floor of the Arkansas Valley below Pine Creek likely saw a wall of dirty water deflect off the Pine Creek whamout and tear down the valley. As the valley widens to the south, depth of the floodwater would have diminished, but 2 mi below Pine Creek the flood waters dropped a 4 ft boulder 150 ft up on the side of the valley.

Erosion was severe in the damsite area; floodwaters tore out the lower ends of moraines, and at Pine Creek they undercut the downstream lateral moraines enough to cause the large Pine Creek School landslide. Floodwaters ripped out parts of the granite river bottom, causing an extremely steep gradient at Pine Creek, where the Arkansas River today drops 50 ft in a distance of 1300 ft, or more than 200 ft per mile. This created the notorious Pine Creek Rapid, Class V-VI (Staub, 1988), challenged only by expert kayakers and avoided by rafters. Below the damsite where the valley widens, catastrophic aggradation occurred.

Flood Landforms and Deposits
The floor of the Arkansas River Valley below Pine Creek consists of two flat surfaces, which are here called Terrace 1, the higher, and about 30 ft lower, Terrace 2 (Fig. 5). There is no significant difference between the two terraces other than elevation.

Figure 5. Two terraces composed of flood gravels. Boulders in foreground are at the surface of Terrace 2. Exposed flood gravels of Terrace 1 are 59 ft thick, capped by post-flood fan alluvium; large flood boulder at top is 28 ft by 21 ft by 11+ ft.

The Terrace 1 surface has a median height of 49 ft above the modern Arkansas River. The gravels are more than 59 ft thick 2 mi below Pine Creek and thin to about 15 ft about 7 mi farther downstream. Where the base is seen, flood gravels lie on Dry Union Formation. The terrace is composed of flood gravels — that is, gravels that show no internal structure or stratification and that contain flood boulders. Flood boulders are boulders not transported by ordinary floods, and here they range from about 5 ft to 45 ft in diameter (Fig. 6). Flood boulders are at the very base, where exposed, and they sit at the surface (Figs. 5, 6). In short, the terrace consists of a sheet of flood gravel, probably representing the traction bedload of the floodwaters.

It is important to recognize that this is not alluvium deposited over a span of time; rather it represents a catastrophic flood event that deposited a sheet of flood gravels as a single event, likely in a matter of hours. The terrace landform results from subsequent erosion. Lack of internal stratification, however, allows the possibility that the deposit was produced by more than one catastrophic flood.

The Terrace 2 surface has a median height of 20 ft above the modern Arkansas River. Terrace 2 consists of a sheet of flood gravels, whose apparent thickness varies from about 30 ft just below the damsite to about 20 ft downstream (the base is not exposed). As noted, no significant differences between the two flood gravels are apparent, other than elevation.

Flood boulders of a unique class are herein called flood–transported boulder erratics (FTBEs). These are boulders from the Twin Lakes pluton deposited by floodwaters on the Pine Creek School landslide (see Fig. 2). Such boulders were necessarily deposited by flood waters, because the landslide deposit contains none of this material. The highest FTBE found is at a GPS elevation of 8,740 ft, about 150 ft above the modern Arkansas River. These were probably deposited by the youngest flood.

The age of Terrace 1 flood gravels is not clear. Because more than 60 ft of down-cutting occurred after deposition of Terrace 1 flood gravels, and because Bull Lake glaciers dammed the Arkansas River, they are considered Bull Lake age. Scott (1975) mapped them as Older Pinedale age. The age of the Terrace 2 flood gravels is also unknown. They are considered Pinedale because at least 60 ft of down-cutting had taken place between the two floods, only 20 ft of down-cutting has since occurred, and because Pinedale glaciers dammed the Arkansas River. Scott (1975) mapped them as Younger Pinedale age.

Cosmogenic 10Be exposure ages were determined for flood boulders on both of these terraces that vary only from 17.2 to 20.9 ka (Briner et al., 2010). Four boulders from Terrace 1 give an age of approximately 19 ka and four boulders from Terrace 2 are approximately 18 ka (Briner et al., 2010). It is unclear whether these dates actually represent the true ages of the two flood events, however, as it can be (and is) argued that the younger flood might have deposited flood boulders on top of Terrace 1, or at least might have moved the older flood boulders at the surface, effectively “resetting” the exposure-age clock. The image at the front of this article shows the scoured surface of Terrace 1 (LIDAR imagery from USGS).

Remnants of the two oldest floods are found in discontinuous strath terraces that can be traced down the Arkansas Valley. The oldest flood gravels, about 300 ft above the modern Arkansas River, are about 30 ft thick, on Dry Union Formation, overlain by about 5 ft of fluvial cobble-pebble gravel. They contain clasts of Twin Lakes porphyry, indicating that the glacial dam was at, or above, Clear Creek. These deposits represent the first catastrophic glacial outburst flood. Correlative gravels 11 mi south of Buena Vista, although not flood gravels there, are capped by the Bishop Ash, dated at 759 ka (Sarna-Wojcicki et al., 2000).

June LS6
Figure 6. Flood boulder on lower terrace (from Young et al., 2011).

Flood gravels from the second flood likewise are found in discontinuous strath terrace remnants, generally about 50 feet below the oldest flood deposits. Along the west side of the valley, about 35 ft of flood boulders lie on Dry Union Formation, overlain by about 20 ft of fluvial gravels without flood boulders, whereas in the east the flood deposits lie on Precambrian plutonic rocks. These gravels also contain Twin Lakes porphyry, indicating the damsite was at Clear Creek or above. At one locality flood boulders appear to be overlain by Lava Creek Ash, dated at 639 ka (Lanphere et al., 2002).

Glaciers dammed the Arkansas River twice in pre-Bull Lake time, probably at the Three Glaciers damsite. When these ice dams failed, catastrophic floods deposited flood boulders downstream, where only isolated patches now remain. The first flood was older than 759 ka, and the second was older than 639 ka.

In Bull Lake time, glaciers advanced down the three contiguous tributary valleys, and the Lake Creek glacier probably forced the Arkansas River onto Precambrian granite to the east. The Pine Creek and Clear Creek glaciers merged to form an ice dam about 670 ft high on the upstream face. The dammed Arkansas River created Three Glaciers Lake, about 600 ft deep and 12 mi long. When lake level rose to about 9,480 ft elevation, the ice dam failed, and the lake emptied catastrophically. The outburst flood tore out the end moraines and deposited the detritus as an extensive sheet of flood gravels at least 59 ft thick below the dam. Subsequently, the Arkansas River cut down more than 60 ft, leaving the flood gravels as Terrace 1.

Renewed glaciation in Pinedale time essentially repeated the damming. The glacier in Clear Creek again dammed the Arkansas River, creating a lake about 650 ft deep that would have been of similar elevation and extent as that of Bull Lake time. Dam failure led to another catastrophic flood, which similarly tore out the moraines and deposited flood gravels on a valley bottom now consisting of two levels differing by about 60 ft. Floodwaters deposited boulders as high as 150 ft above the modern Arkansas River and probably also deposited boulders on Terrace 1. Since that flood, the Arkansas River has cut down into the Pinedale flood gravels about 20 ft, leaving the surface as Terrace 2.June LS7

Briner, J.P., Young, N.E., Leonard, E.M., and Lee, Keenan, 2010, A new 10Be chronology of late Pleistocene moraines and glacial outburst flood terraces in the Upper Arkansas River Valley, Colorado [abst.]: Geological Society of America Abstracts with Programs, v. 42, no.5, p. 309.

Brugger, K.A., Leonard, E.M., Lee, Keenan, and Bush, M.A., 2011, Discharge estimates for a glacial outburst paleoflood on the Upper Arkansas River, Colorado, from an ice-dam failure model: Geological Society of America Abstracts with Programs, v. 43, no. 4, p. 10.

Lanphere, M.A., Champion, D.E., Christiansen, R.L., Izett, G.A., and Obradovich, J.D., 2002, Revised ages for tuffs of the Yellowstone Plateau volcanic field: assignment of the Huckleberry Ridge America Bulletin, v. 114, p. 559-568.

McCalpin, J.P., Funk, Jonathan, and Mendel, David, in press, Leadville South quadrangle geologic map, Lake County, Colorado: Colorado Geological Survey, 1:24,000 scale.

Randle, T.J., J.A. Bountry, R. Klinger, and A. Lockhart, 2000, Geomorphology and river hydraulics of the Teton River upstream of Teton Dam, Teton River, Idaho: U.S. Bureau of Reclamation, Technical Service Center, Denver, Colorado, 48 p.

Sarna-Wojcicki, A.M., Pringle, M.S., and Wijbrans, Jan, 2000, New 40Ar/39Ar age of the Bishop Tuff from multiple sites and sedimentation rate calibration of the Matuyama-Brunhes boundary: Journal of Geophysical Research, v. 105, p. 21 431–21 443.

Scott, G. R., 1975, Reconnaissance geologic map of the Buena Vista quadrangle, Chaffee and Park Counties, Colorado: U.S. Geological Survey Map MF-657, 1:62,500 scale.

Scott, G. R., 1984, Part III Pleistocene floods along the Arkansas River, Chaffee County, Colorado, in Nelson, A.R., Shroba, R.R., and Scott, G.R., eds., Quaternary deposits of the Upper Arkansas River Valley, Colorado: Boulder, Colorado, American Quaternary Association, 8th Biennial Meeting, August 16-17, 1984, unpublished guide for Field Trip No. 7, p. 51-57.

Staub, Frank, 1988, The Upper Arkansas River, rapids, history, and nature, mile by mile: Golden, Colorado, Fulcrum, 265 p.

Walder, J.S., and Costa, J.E., 1996, Outburst floods from glacier-dammed lakes: the effect of mode of lake drainage on flood magnitude: Earth Surface Processes and Landforms, v. 21, no. 8, p. 701-723.

Walder, J.S., and Fountain, A.G., 1997, Glacier generated floods, in Destructive Water: Water-caused natural disasters, their abatement and control, IAHS Publication No. 239, p. 107-113. USGS, 2012,

Young, N.E., Briner, J.P., Leonard, E.M., Licciardi, J.M., and Lee, Keenan, 2011, Assessing climatic and nonclimatic forcing of Pinedale glaciation and deglaciation in the Western United States: Geology, v. 39, no. 2, p. 171-174.