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



The dynamics of debris flows are described here in hydrologic and hydraulic terms, because that nomenclature is appropriate, and because it is familiar to land-use planners and civil engineers dealing with structures subject to inundation. What we believe is the best example of each of the flow types that pose risk (table 6) is described in this section, its dynamics are presented (table 7), and its flow cross sections are portrayed (pl. 1). There is no substitute for the description of real-world flow behavior, through the case-history approach when dealing with engineering problems involving complex phenomena. No rheologic model deals with the spectrum of debris flow behavior at Mount Rainier.


Velocities (table 7) are based on measurements of runup on obstacles to flow, or on superelevation of flow around bends. Johnson (1970), Costa (1984), Fairchild (1985), Pierson (1985), and Scott (1988b) have analyzed or commented on field applications of this method and the accuracy of the results. The behavior of the large cohesive lahars near the boundary of the Puget Sound lowland is critical, so special attention was given to the measurements near that point. Several sets of measurements defined flow above 15 m/s and approaching 20 m/s. No runup measurement was ideal, as in the case of a steep bare surface (a frictionless surface is assumed) normal to flow. We believe that 20 m/s is a conservative minimum velocity for the cohesive lahars at the lowland boundary. A velocity in the range of 25 to 30 m/s was obtained from runup of the branch of the Osceola Mudflow in the West Fork White River where it entered the White River valley at a high angle, and the estimate of velocity of at least 40 m/s for the Round Pass Mudflow near the base of the volcano was noted in the section on that lahar. In general, runup measurements were more readily obtained than measurements of superelevation in bends for both cohesive and noncohesive lahars.

Table 7. Characteristics of design- or planning-case lahars.

[Characteristics determined a described in text; N.A. = not applicable]

Characteristic Maximum lahar Case I Case II Case III

Debris flow typeCohesive Cohesive Noncohesive Usually cohesive.1
Recurrence interval (yrs) ~10,000 500-1,000 100-500 <100
Volume at lowland boundary (x 106 m3). >3,000 230 60 (Puyallup R.)
65 (Carbon R.)
Mean flow velocity (m/s):
   Base of volcano 2 >40 2 >30 10 2 >30
   Lowland boundary >20 ~20 ~7 N.A.
   1 km on lowland ~10 ~8 ~3-4 N.A.
Cross-sectional area of flow at lowland boundary (m2). ~90,000 ~16,000 1,000 (Puyallup R.)
1,200 (Carbon R.)
Peak discharge at lowland boundary (m3/s). >1,800,000 ~320,000 7,700 (Puyallup R.)
8,400 (Carbon R.)
Flow depth (m):
   Base of volcano ~200 ~50 15 55
   Lowland boundary ~100 22 8 N.A.
   1 km on lowland ~30 ~10 1-3 N.A.
Sediment concentration at lowland boundary (percent by volume).3 >60 >60 ~40 (Puyallup R.)
~45 (Carbon R.)
Extent (or inundation area) To Puget Sound or Columbia R. (in Cowlitz R. drainage). Inundation of 36km2 (Electron) to ~50 km2 (modern recurence of same flow). All active flood plains (except Cowlitz R.) above reservoirs, if present; otherwise upstream of Puyallup. Runout phases of noncohesive lahar could extend an additional 10 km.

1Some maybe partly or entirely noncohensive depending on source area.
2Estimated by comparison with similar flows at other volcanoes.
3Estimated using the linear relation observed between sorting and concentration at Mount St. Helens (Scott, 1988b).

Cross sections were defined by the distribution of deposits. Unlike floods, debris flows leave deposits accreting on valley sides to the level of peak flow. Delineation of the highest peak flow deposit, equivalent to the high water mark of a flood, was commonly confirmed at multiple points, and the cross sections of flows were measured only where the valley-bottom deposit of the same flow was known. In a few instances, the thickness of valley fill of a flow was extrapolated longitudinally. As noted above (under "Flow magnitude and frequency"), markedly concave flow surfaces in sharp bends may have the effect of exaggerating both cross-sectional areas and the discharges calculated from them (Webb and others 1989, p. 22, table 10). However, of the sites used for calculating the discharges shown in table 7, none has a radius of curvature sufficient to cause this effect.

Flow wave volumes were estimated from the volumes of their deposits. The volume was in most cases not increased to account for loss of water because, in the case of the cohesive debris flows, a large proportion of the deposits probably remained saturated, and in the case of the noncohesive debris flows, the water content of the flow wave was largely interstitial between grains in nearly continuous contact Volumetric comparisons worked well for the Electron Mudflow where the distal end is defined and the deposits were extensively angered (Crandell, 1971). It is less exact where flows continued into Puget Sound or, in the case of the runout phases of the noncohesive lahars, where the flow wave was diluted eventually to streamflow. Corrections for loss of deposits by erosion are an additional source of error, but reconstructions of the original depositional surfaces were possible for some flows.


The term "maximum lahar" is substituted for the "worst-case flow" of hydrologic analysis, because there can always be a flow worse than that defined as the worst case. We also prefer the term to "most-extreme lahar," used for a moraine-dammed-lake breakout in which the most-extreme case is displacement of an entire lake by a snow or debris avalanche (Laenen and others, 1992). The true "worst-case" or "most-extreme" analog at Rainier is the improbable removal of the entire edifice. "Maximum lahar" is analogous to the "maximum mudflow" or "maximum credible mudflows" used in forecasts of lahars at Mount St Helens (for example, U.S. Army Corps of Engineers, 1985). The term is intended to imply that, although larger flows are possible, they are so unlikely they need not be considered.

The Osceola Mudflow (Crandell, 1971), is the maximum lahar at Mount Rainier. Cross sections of this flow are shown on plate 1, and its dynamics are described in table 7. The inundation area of the actual flow is easily discernible on the Puget Sound lowland (Crandell, 1963b, 1971), although the flow was difficult to define upstream (noted in sections 0—1 and 0—2, pl. 1). The inundation area of a modern cohesive lahar of the same size could extend to Puget Sound, through Tacoma along the Puyallup River and through Seattle by way of the Green River system and the Duwamish Waterway. The lower resistance to flow of the modern unforested river valleys would allow a recurrence of this flow to go farther and faster than did the original flow approximately 5,000 radiocarbon years ago. Relative sea level in the Duwamish Embayment of Puget Sound was higher at that time, and the flow entered the sound farther upstream. In a well 6 km northwest of Auburn, deposits of the flow occur 85 m beneath present sea level and are 7 m thick (Luzier, 1969, p. 14); submarine deposition is probable.

The Osceola Mudflow had a volume many times that of the next largest cohesive lahar. We accept Crandell's (1971) estimate of 2—3 km3 for the present volume, but the original volume may have been as much as twice that amount if subsequent erosion and a possibly larger original submarine extent are taken into account. A lahar this size has occurred only once in postglacial time, within the last 10,000 years. When compared with all other large cohesive lahars, it is a statistical outlier. It is tentatively assigned a recurrence interval of 10,000 years. Thus, for illustrative purposes, its probability approximates that of a return of glaciation to the Puget Sound lowland. One or more events of at least this size have a 1 percent chance of occurring within a century (Reich, 1973). An event of this frequency is not normally considered in hazards planning, but Latter and others (1981) propose that it should be. In modern risk analysis, such an event is described as one of "low probability and high consequences," with the implication that the risk may be unacceptable at even very small probabilities.

The record at Mount Rainier indicates that the most probable recurrence of a maximum lahar will be a debris avalanche that transforms directly to a lahar on or near the volcano. Primary transformation is not a certainty, however, and Crandell (1988, fig. 18) calculates the probable runout distances in each river system at Mount Rainier of untransformed debris avalanches with a volume of at least 1 km3. The hazards of untransformed debris avalanches are discussed by many (including Crandell, 1988; Scheidegger, 1973; Siebert and others, 1987; and Francis and Self, 1987); the risks from debris avalanches are generally greater than those from lahars, mainly because of higher flow velocities. A debris avalanche from Mount Rainier would probably be at least partially saturated; such a flow would have the potential to yield a large secondary lahar as did the 1980 example at Mount St Helens. If a primary debris avalanche occurs, downstream warnings of a subsequent lahar would be necessary. The lag time at Mount St. Helens from avalanche emplacement to lahar initiation was about five hours. This sequence of events can be regarded as a much less probable variant of both the maximum lahar and Case I, below.


Case I is a large cohesive debris flow having a recurrence interval of 500 to 1,000 years and is the appropriate case for long-term planning in the watersheds draining Mount Rainier. Even one event (or more) equal to or greater than a flow with a 1,000-yr recurrence interval has a 9.5 percent probability of occurring at least once in the next century (Reich, 1973).

If the Osceola Mudflow is excluded on the grounds that it is a statistical outlier of this flow type, several smaller cohesive lahars form a discrete population. The most recent and best defined of these flows is the Electron Mudflow. The importance of this flow to hazard analysis has long been recognized (Crandell, 1971; Cullen, 1977; and Cullen Tanaka, 1983). The lahar is here assigned a magnitude and frequency, and its dynamics are specified at the margin of the Puget Sound lowland, where risk increases greatly (table 7, pl. 1).

The volume of the Electron Mudflow deposits on the Puget Sound lowland was satisfactorily determined by Crandell (1971, p. 57) as slightly more than 183 million m3. The flow deposit is overlain by reworked deposits of the flow. Its original volume (table 7) is estimated by assuming deposition near the levels of the highest medial flow deposits. This assumption is based on the downstream behavior of the cohesive lahar originating in the North Fork Toutle River at Mount St. Helens in 1980.

The risk of this type of flow surpasses that of all smaller but more frequent flows. Moreover, the risk is increased by the lack of a clear association with major episodes of volcanic activity which could provide a warning (Crandell, 1971; Scott and Janda, 1987). Such flows may be triggered by nonmagmatic seismicity, by steam eruptions, or just by gravity in places where a failure plane has been lubricated by clay and geothermal pore fluids. No assumption of precursor volcanic activity can be made in planning for these flows. This is a conservative approach that is consistent with the available evidence.

Sector collapses of the size that produce cohesive lahars can occur on any side of the volcano (Frank, 1985, p. 181). Given the lack of evidence that one flow of this type will stabilize the affected sector of the volcano thereafter, this is the best assumption. Potential effects on downstream areas differ only slightly among watersheds. The main complicating factor is the presence of reservoirs in three of the five major watersheds.

A modern recurrence of a large cohesive lahar will inundate a larger area of the Puget Sound lowland than did the prehistoric flows because of the greatly reduced friction on deforested flood plains. The distribution of a modern flow can be predicted by estimating the deposit thickness on unforested flood plains and distributing the design volume at the mountain front over the corresponding area. On the basis of the behavior of the 1980 cohesive lahar at Mount St Helens, which traversed clearcut and forested flood plains, the modern thickness would be close to 70 percent of the prehistoric thickness. Some additional bulking of the flow on the cleared flood plains will reduce its natural rate of attenuation, but erosion will probably be concentrated in active channels as it was under forested conditions. Thus the inundation area of a modern flow of the same type and same original volume as the Electron Mudflow could increase to approximately 50 km2 (compared to the 36-km2 area of the Electron). A similar but somewhat larger volume will just be spread over a larger area.

The flow record indicates that the most probable recurrence of Case I will be a debris avalanche that transforms to a lahar on or near the volcano. As noted for the "maximum lahar," this origin is not a certainty. Untransformed debris avalanches (Crandell, 1971, fig. 18) can be regarded as a much less probable variant of Case I. Because such flows have not occurred at Mount Rainier, it is impossible to specify probable magnitudes or frequencies except by means of examples at other volcanoes, as Crandell has done.


Case II is a noncohesive flow represented by the National Lahar, which with its runout phases is a suitable example of this category that can be extrapolated to all watersheds. The recurrence interval of noncohesive flows in the size range of the National is near the lower end of the 100- to 500-yr range and thus is analogous to the 100-yr flood, one widely considered for structure design and flood-plain management. Flow cross sections (pl. 1) can be applied upstream from reservoirs.

Comparison shows that this design debris flow will be larger than design water floods in upstream reaches, but will be smaller downstream. This difference is explained by the continuous attenuation of a lahar or lahar-runout flow, as compared with the typical amplification of a meteorologic flood as tributary inflows increase downstream. Measured flood-carrying capacities of the Puyallup, White, and Carbon Rivers on the Puget Sound lowland illustrate this trend (Prych, 1987). Nonetheless, the noncohesive debris flows increase the risk of flood-plain inundation throughout a river system without reservoirs. Upstream, the lahar subpopulation presents more risk than meteorologic floods. Inundation levels can be estimated by adjusting the cross-sectional areas (pl. 1) for the attenuation, as shown, due to distance from the volcano.

The flow wave in this design and planning case consists of hyperconcentrated flow during much of the flow interval beyond the base of the volcano. Hyperconcentrated flow probably will persist to the boundary of the Puget Sound lowland, but will transform to normal streamflow rapidly beyond that point because of rapid loss of sediment from the flow wave on flood-plain surfaces. The runout phases of the National Lahar (figs. 2, 11B, and 11C) are representative of changes expected in future noncohesive flows.

If Case I presents the greatest total risk, should Case II be considered as well as Case I in any part of a drainage? The answer here is affirmative, because of the distinction between planning for the longest term that is cost-effective, as in a land-use evaluation contingent on Case I, and designing for a flow with a high degree of probability during the life of an individual structure. For example, a flow equal to or greater than the event with a recurrence interval of 100 years has a 64 percent probability of occurring at least once in the next century (Reich, 1973).

An additional rationale for the application of Case II is its probably greater association with precursory volcanic activity than Case I. In the event of impending eruptive activity forecast by a monitoring network, Case II is the minimal flow event that logically can be expected. Each river system contains enough glacial ice to provide meltwater capable of producing a noncohesive lahar of this size.


Case III is a relatively small debris avalanche, originating as a landslide, that probably will transform to a debris flow. Two moderate-sized and several small debris and rock avalanches have occurred since 1900. The largest of these came within a kilometer of the White River Campground in 1963, albeit at a time (December) when the campground was closed (Crandell and Fahnestock, 1965). Neither moderate-sized flow transformed directly to a lahar, but both produced small debris flows by dewatering and slumping of their surfaces.

The origin of the Case III flow is the same as that of both Case I and the maximum lahar, but the smaller Case III examples occur much more frequently on the volcano. They probably will recur without warning and certainly will move at high velocity. As the case history best exemplifying this flow, the debris avalanche yielding the Tahoma Lahar (pl. 1) is the most appropriate example. At least part of that debris avalanche transformed to a hummocky lahar with a flow depth of as much as 55 m in confined canyons on the volcano and 20 m shortly beyond the base of the volcano. Subsequent attenuation was rapid.

Greatly adding to the risk from these flows is their high velocities, almost certainly well in excess of 30 m/s (67 mph). The velocity of the largest avalanche from Little Tahoma Peak was at least 35 to 40 m/s (Crandell and Fahnestock, 1965). A velocity of about 50 m/s, which was increasing at the point of measurement, was reported for the 1980 debris avalanche at Mount St Helens (Voight and others, 1981).

The best outcome of planning for such a rapid flow can only be to minimize exposure. Risk associated with a runout flow like that possibly developed from the Tahoma Lahar will be much less than that of Case II. Within the Park, however, consideration can be given to siting new campgrounds and facilities above the flow depths of the Tahoma flow wave (pl. 1), and in other drainages, above its extrapolated cross-sectional area adjusted for distance from the summit.


Hazard-zone analysis is another approach to assessing the composite risk of all flows, including those smaller and more frequent than the above cases. The approach involves the determination of past inundation levels from tephra layers, vegetation, and fan and flood-plain morphology. The combination of criteria yields areas of inundation over several time intervals useful for land-use planning. Hazard zones delineated by deposits and dendrochronology have effectively defined the risks of small, high-frequency flows at Mount Shasta (Osterkamp and others, 1985). A similar analysis can be useful in siting individual facilities in Mount Rainier National Park. Hazard zones, however, are not calculated or presented in this report The zones can be determined, as needed, for specific locations according to the following guidelines.


Tephra set W, deposited over most of the National Park, can be treated as a paleohydrologic crest-stage gage by measuring the height to which the layer has been truncated by flow against valley side slopes. Set W consists of two layers, deposited in A.D. 1480 and 1482 (Yamaguchi, 1983 and 1985). The first of the two layers predominates at Mount Rainier. Its eroded lower margin records the highest level of flow since deposition in at least the upper part of each drainage surrounding the mountain. Where a drainage did not convey a significant flow like the Tahoma Lahar in the last 500 years, the level to which set W is eroded provides an estimate of the inundation potential without a major eruption or sector collapse. Zone I is thus defined. Because zone I defines only the most recent time interval, preference in planning should be given to the planning and design case histories selected with the perspective of longer time periods.


The area inundated since the beginning of the 20th century can be established by the historical record and by dendrochronology; such data can provide a good approximation of the level to which flow has extended in the last century. In most cases, glacial-outwash flow patterns subsequent to the start of Neoglacial recession about 60 to 200 years ago (Sigafoos and Hendricks, 1961 and 1972; Burbank, 1981) are discernible from vegetation patterns on aerial photographs.


This area encompasses the modern alluvial fans at the base of the volcano, active flood plains, and marginal areas subject to lateral erosion. The boundaries would seem obvious except on the alluvial fans, whose surfaces are broad and convex and locally support a mature forest. The fans record debris flow deposition triggered by the decreasing slope and expansion in reaches at the base of the volcano. Debulking is rapid in such areas, which explains why the smaller flows attenuate rapidly (fig. 13).


Lateral (or bank) erosion is an additional hazard down stream along the streams with normal flood plains. The high terrace bordering the Nisqually River upstream from Longmire (figs. 9 and 11A; set W on surface) is being cut by normal fluvial erosion, requiring some resiting of trails. Progressive lateral erosion tends to be localized and is controlled by factors such as channel pattern (Brice and Blodgett, 1978, chap. 4), bank material (Schumm 1960, 1961), and vegetation (Scott, 1981).

Channel pattern normally will be the determining factor in localizing erosion at banks cut against a terrace. A meandering pattern will result in erosion at the outsides of bends, for example. Braided streams in the proglacial environments may regularly impinge against bedrock valley walls, and where streams are confined by steep-sided Neoglacial moraines, erosion can be large, episodic, and unpredictable in location. Trails and climbing routes along crests of Neoglacial moraines are subject to mass failures triggered by lateral erosion at the base of the moraine. Most of the major streams fall in the category of "streams wider at bends," which, according to Brice and Blodgett (1978), have greater lateral instability than either equiwidth streams or those with random variation.

The White River shows the effects of a cohesive valley fill, the deposits of the Osceola Mudflow, in reducing lateral erosion. Where the active channel is incised into the valley fill, an uncommonly low width/depth ratio results (in the range of 4-7 at locations downstream from the national park). This low ratio is a function (inverse) of the high silt and clay content of the central-valley facies of the Osceola and is consistent with the findings of Schumm (1960) elsewhere.

<<< Previous <<< Contents >>> Next >>>

Last Updated: 01-Mar-2005