Geological Survey Professional Paper 1547 Sedimentology, Behavior, and Hazards of Debris Flows at Mount Rainier Washington

FLOWS OF LOW MAGNITUDE AND HIGH FREQUENCY (LESS THAN 100 YEARS)

The smallest, most frequent debris flows and their derivative runout flows are common in a few river systems at Mount Rainier, but rare in others. These flows have several general characteristics: (1) They tend to occur in clusters within periods of several years (such as those that occurred in the periods of 1967-70 and 1986-92), and decades may separate the clusters; (2) the debris flows that originate as glacial-outburst surges have been historically most common in late summer and fall; (3) the flows are uniformly noncohesive, forming from flood surges and in most cases transforming downstream through hyperconcentrated flow to normal streamflow; (4) this transformation is rapid, occurring at the base of the volcano, and so the flows attenuate rapidly (fig. 13) and are typically contained within stream channels beyond that point; and (5) the flows have a variety of glacier-related origins and interactions; the largest flows occur during or just after periods of precipitation, which may trigger collapse of the stagnant terminal ice resulting from Neoglacial recession. Walder and Driedger (1994) have prepared a detailed analysis of the effects of outburst floods and the debris flows formed by them.

Lakes dammed by terminal Neoglacial moraines are not a large hazard at Mount Rainier. Unlike the numerous moraine-dammed lakes on some Oregon volcanoes (Laenen and others, 1987, 1992), the lakes on Mount Rainier either are cirque lakes with bedrock sills or are dammed by old moraines and have highly stable outlets, having broken out long ago. However, a landslide into a lake, as has occurred at Lake George in the Tahoma Creek watershed, could catastrophically displace enough water to create a significant surge that may bulk to debris flow.

A second and more hazardous type of frequent flow is a debris avalanche, which is not likely to extend far from the volcano, unlike the large debris avalanches that most commonly transform to lahars. This flow type and its possible mobilization to a lahar are discussed in a later section.

RECURRENCE INTERVAL OF SMALL DEBRIS FLOWS

Many small streamflow surges originate on the volcano. They have occurred at a rate of at least one per year between 1986 and 1992 (Walder and Driedger, 1993). Most are glacier-related, either as subglacial outbursts or supraglacial outbursts of ponds dammed by saturated modern moraines. Others are the result of temporary impoundment of streams by landslides, commonly in Neoglacial lateral moraines. Some of the surges are not large enough to erode the coarse bed material and thus do not bulk to debris flows.

The larger surges of this type, especially those triggered by precipitation, are competent to erode all bed material, including boulders several meters in diameter. These surges bulk to debris flows and may be divided into precipitation-induced events and clear-weather events (fig. 14). The best-known examples are along the Nisqually River and its tributaries, Tahoma and Kautz Creeks. At least 20 such events between 1925 and 1990 were of sufficient size to inundate areas of inactive flood plain in those watersheds and pose a local hazard to hikers. (Flows for which only the year is known are not shown in figure 14.) Walder and Driedger (1993) have prepared a guide to their hazards and occurrences for park visitors.

Between 1986 and 1988 there were eight major flows from Mount Rainier—five on Tahoma Creek and three on Kautz Creek—and at least two smaller flows (on Tahoma Creek in late August 1987). A similar cluster of flows between 1967 and 1970 on Tahoma Creek was ascribed by Crandell (1971, p. 60) to possible geothermal activity. Although no increase in geothermal activity was known to accompany the latest cluster, Frank (1985) reported the presence of heated ground and sub-boiling-point fumaroles on the South Tahoma and Kautz Glacier headwalls. Study of the Tahoma Creek deposits before the initial 1986 flow verified that the lapse in reported flows between 1970 and 1986 represents a true lack of significant flows in that drainage, not just a lack of observations.

Some flows of this type have occurred unseen and unrecorded in other drainages, even since construction of the Wonderland Trail, which circumnavigates the volcano near its base. Deposits are covered or eroded by those of later flows of similar noncohesive texture. The record of 20 significant, flood-plain-inundating debris flows since 1925 is clearly a minimum in the Nisqually River headwaters. An appropriate composite recurrence interval for planning purposes in those watersheds is approximately two years. At present (1994), however, even though the cluster of flows on Tahoma Creek that began in 1986 may be tapering off, it is reasonable to assume that at least one flow can be anticipated each year. Each flow's area of inundation is likely to extend only locally outside that of the previous members of the cluster. However, a detailed study focusing on origins and valley responses of these flows concludes that long-term predictions of flow frequency in the watershed are not possible (Walder and Driedger, 1994).

SEASONAL DISTRIBUTION

A significant factor mitigating the hazards of these flows is that they tend to occur in late summer and fall (fig. 14) after most back-country tourist use. The mean date of the known flows in figure 14 is September 7. In addition, the largest flows recorded historically in each of the three major valleys in the Nisqually headwaters have occurred in October—October 2 for Kautz Creek, the 25th for the upper Nisqually River, and either the 15th (highest volume) or 26th (highest discharge) for Tahoma Creek. Each of those flows began as a precipitation-induced surge, and two were amplified by the collapse of areas of stagnant ice. Although these surges probably were amplified by subglacial water, the glaciers served mainly as conduits and temporary reservoirs of storm runoff.

Many clear-weather flows are logically ascribed to subglacial storage of meltwater. These flows tend to occur earlier in the year than the precipitation-induced flows (fig. 14). The previous cluster of flows, from 1967 to 1970, occurred in the relatively narrow time interval of August 20 to September 23 (Crandell, 1971). This time of occurrence suggests an origin as glacial-outburst floods induced by warm-weather melting. Other evidence cited by Crandell (1971) suggests a geothermal origin, and we assume that possibility exists.

Figure 14. Seasonal distribution of debris flows, hyperconcentrated flows, and floods in glacier-fed tributaries of the Nisqually River from 1925 to 1990. Only flows with known dates are shown; many others, smaller or of unknown date, are not shown. Some data from Crandell (1971); Erdmann and Johnson (1953); Richardson (1968); J. J. Major (U.S. Geological Survey, written commun., 1985); and M. Carney, S. V. Scott, and D. J. Sharlow (National Park Service, oral commun., 1986—1987).

FLOW TEXTURE AND FORMATIVE TRANSFORMATIONS

Whether their origin is from precipitation or meltwater, the flows bulk rapidly through hyperconcentrated flow to debris flow. These transformations have occurred on the moraine-covered surface of the glacier for surges that exited above the terminus (fig. 15), or in the unvegetated, proglacial valleys for subglacial surges that emerged at the terminus. The proglacial valleys contain vast amounts of reworked sediment of morainal and volcaniclastic origin. This sediment readily bulks, mainly by mobilization of unstable bed and bank material, into the surges from the glaciers to yield debris flows that are uniformly noncohesive in texture and contain 1 percent or less of clay in 10 examples, including the 1947 flows on Kautz Creek.

FLOW DYNAMICS AND TRANSFORMATIONS

Two flows in the Tahoma Creek valley were studied as models to analyze the dynamics and transformations of a precipitation-induced flow and a clear-weather flow: the former occurred October 26, 1986, and the latter, June 29, 1987. By chance, the watershed and stream channel had been studied immediately before each event, and they were restudied afterwards. Both flows were noncohesive, although the deposits of the clear-weather flow contain slightly more clay. The mean grain size and sorting of the flow matrixes are shown in figure 3.

A significant part of the October 1986 flow originated from a sinkhole-like collapse near the active glacier terminus, which is presently just below a crevassed ice fall at an altitude of about 1,830 m (6,000 ft) to 2,260 m (7,400 ft). The dimensions of the collapse were estimated from an aircraft as 9 by 15 m (R. Dunnagan, National Park Service, oral commun., 1986). The precipitation-induced surge bulked as it flowed across the top of the stagnant, moraine-covered lower portion of the glacier. Part of the flow may have entered a small sinkhole (fig. 16A), and the remainder was apparently dammed temporarily on the surface before cutting a channel along the west side of the glacier. The flow probably was already a debris flow at that point, as indicated by boulder levees and deposit texture on the glacier surface (fig. 16A). As the two distributaries (subglacial and supraglacial) rejoined below the stagnant-ice terminus at 1,510 m (4,960 ft) altitude, bulking continued to enlarge the flow in the channel incised in the Neoglacial moraine. Bulking was amplified by the collapse of debris-rich ice at the front of the stagnant part of the glacier, and the resulting material may have dammed the main channel.

The 1986 flow had the highest discharge of any flow in the 1986—88 period (fig. 13). The flow volume, however, was exceeded by the flow or series of flows that occurred on October 15, 1988. The deposits of that flow (or flows) inundated the entire valley floor, 0.2 km in width at the site of the former picnic area, and occurred in greatest volume at a point farther downstream than any other flow in the 1986—88 group. This observation suggests a correlation between flow size and the distance of the locus of deposition from the mountain, which is also suggested by the data shown in figure 13. Although peak discharge was higher on October 26, 1986, the volume of sediment transported on October 15, 1988, was greater, accomplished by either a broader flow wave or by multiple flows.

Figure 15. Active front of the South Tahoma Glacier 5 days before (A) and 1 day after (B) the clear-weather glacial-outburst flood and debris flow of June 19, 1987. Arrow in B points to dark areas of collapse, source of at least part of the flow.

Figure 16. Area of stagnant, moraine-covered lower part of South Tahoma Glacier 5 days before (A) and 1 day after (B) the flow of June 29, 1987. Location 0.9 km downstream from active front of glacier shown in figure 15. In A, note flow of part of supraglacial stream into sinkhole (left arrow) and fresh scarps and factures in alluvium (extending from lower left toward sinkhole), corresponding to crevasses in underlying stagnant ice. Boulder berms delineate flow of October 26, 1986 (right arrow). B shows incision into he debris-rich stagnant ice by the June 29 flow.

Figure 17. Ice clast, more than 1 m in maximum dimension, included with andesite clasts of similar size in lobate boulder front of flow of October 26, 1986, Tahoma Creek drainage.

The June 1987 flow originated from the base of the icefall (fig. 15B), at the end of a week of completely clear weather that marked the beginning of a severe drought period. (This drought also resulted in small debris flows along Tahoma Creek on August 28 and 31 and one of moderate size on September 23.) Lateral deposits of the June flow were silt-rich as the surge issued from the ice fall, and bulking to debris flow occurred on the surface of the stagnant ice above the site of the previously existing sinkhole. Lateral erosion of stagnant ice triggered an ice-block avalanche into the channel, and blocks of mixed ice and rock several meters in diameter were transported. Figure 16 shows the channel several days before and the day after the flow. Below the lateral ice avalanche, the flow triggered a spectacular collapse of the stagnant glacier surface from above the sinkhole to the terminus. Rapid incision into the debris-rich ice then led to further bulking and enlargement of the flow wave.

The depositional patterns of the 1986 and 1987 flows were nearly identical. The deposits were thickest within 0.5 km of the inundated picnic area (a campground before inundation in 1967) along Tahoma Creek. Boulder fronts as much as 3.5 m high (eroded or buried in 1988) represented the "frozen" termini of convex lobes of the coarse front of the flow. As movement of each lobe ceased, its deposits diverted flow from the following segment of the wave to one side. Each new surge successively stopped, diverting the following portion of the wave, and so on in a chain reaction. Distal surges in the flow were thereby created from a single flood wave, as shown by the existence of only a single berm of deposits upstream. The coarsest boulder fronts of each flow contained as much as 10 percent clasts of ice and frozen ground (fig. 17).

Each flow front was lower and finer grained than the preceding lobe. At a point in this progressive longitudinal "sampling" of both the 1986 and 1987 flows, the transformation to hyperconcentrated flow was reached, and the successive deposition of debris flow lobes ended. The point in each case was about 0.5 km downstream from the former picnic area. The pattern documents the progressive fining, improvement in sorting, and decline in strength (shown by loss of dispersed large clasts) longitudinally within the flow wave (fig. 18).

Figure 18. Cumulative curves of particle sizes within successive boulder fronts and hyperconcentrated-flow deposits formed during transformation of the Tahoma Creek debris flow of October 26, 1986. After deposition of the lobate fronts, only the hyperconcentrated tail of the flow continued downstream. Down-channel distance from the peak of Mount Rainier is shown for each deposit. (click on image for an enlargement in a new window)

The tail of the flow wave clearly consisted of hyperconcentrated flow. Deposits having the texture characteristic of that flow type accreted to the sides of the debris flow channels and distributaries at levels lower than those achieved by the debris flow levees. Both the continuity in the successively finer and lower debris flow fronts and the textural transformation to hyperconcentrated flow indicate fractionation of a single flow wave. Some of these events have been interpreted as a series of separate flows because the differences in flow within a single flood wave, as well as the creation of distal surges, were not recognized. With the exception of the 1947 Kautz Creek flows, which were clearly separated, most of the flows in this category of magnitude and frequency began as single flood waves. The distal surges described above are variants of the surges resulting from temporary damming of a confined channel by the coarse boulder front of a flow. (See Pierson, 1980; and Costa, 1984.)

For both the 1986 and 1987 flows, the hyperconcentrated flow deposits of the receding flood wave overlie the sole layers of distributary debris flow channels (fig. 19). These highly compacted layers of pebbles dispersed in a silty sand matrix are identical to the Type II sole layers at the bases of lahars formed in 1980 at Mount St. Helens (Scott, 1988b).

Figure 19. Debris-flow levees and underlying sole layers from recent flows along Tahoma Creek. A, Distributary channel of the October 1986 debris flow as it appeared in May 1987. Darker, compacted sole layer, in middle, overlies lighter channel bed material. Sole layer is 30-50 cm thick. Flow lines formed by recessional hyperconcentrated flow are visible just below the coarse debris at the tops of the levees. B, Main channel of the June 1987 debris flow at same site in July 1987. Sole layer, accreted to channel sides, is being eroded. Bank topped by levee is 3.5 m high.

After the hyperconcentrated tail of the main flow wave had passed, a small secondary debris flow was formed through dewatering of the coarse debris-flow deposits. Pore fluid draining from the coarse flow fronts contained sufficient silt (15 percent of deposits) and clay (2 percent of deposits) to yield a 1-cm-thick deposit in downstream channel thalwegs. The elevated deposit margins indicate strength in the range of debris flow. Most of the deposit is sand (75 percent, fig. 20) and, like a sole layer in a subsequently active channel, is unlikely to be preserved. This deposit is a smaller version of the large lahar formed from the 1980 debris avalanche in the North Fork Toutle River at Mount St. Helens; it is likewise similar to the lahar formed from the main 1963 debris avalanche from Little Tahoma Peak into the White River drainage (fig. 20).

Figure 20. Cumulative curves of particle sizes of debris flows derived by dewatering from the main 1963 debris avalanche in the White River valley and the June 1987 debris flow in Tahoma Creek, compared with cumulative curves of the primary deposits. (click on image for an enlargement in a new window)

The deposit textures of the 1963 debris avalanche, the 1987 debris flow, and the flows derived from each by dewatering are illustrated in figure 20. The slope of the cumulative curve of each derivative flow is very similar to that of the finer part of the source flow. The dewatering process thus removes part of the matrix of the primary deposit but, unlike the more common direct transformation of the debris flows to hyperconcentrated flows, produces another, relatively small debris flow. The ability of the dewatering process to produce large flows is documented by the 1980 lahar in the North Fork Toutle River at Mount St. Helens. No such origin, however, can be established at Mount Rainier for any debris flow larger than the relatively small 1963 example. This conclusion confirms our belief (and that of Crandell, 1971) that the large sector collapses at Mount Rainier continued directly as debris flows for long distances, rather than yielding thick, hummocky masses immobilized nearer the volcano.