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

GENERAL STATEMENT

Fundamental to the analysis of subaerial sediment gravity flows at Mount Rainier is the recognition of two distinct types of debris flows that differ significantly in both texture and origin. Thus it is possible to deduce the origin of an ancient flow deposit, even one far from the volcano, from its texture, in particular the texture of the matrix phase of the characteristically bimodal flows. An earlier investigation at Mount St. Helens (Scott, 1988b) reported two distinctive types of debris flows: (1) relatively clay-rich flows that traveled long distances as debris flows and (2) more granular flows that began mainly as streamflow, then bulked (increased volume by incorporating sediment) to form hyperconcentrated streamflow, and continued to incorporate eroded sediment until a debris flow was formed. Distally, the sequence of flow types was reversed. These two types of debris flows are the cohesive and noncohesive debris flows, respectively, of this report.

A subdivision of only mudflows (commonly defined as silt- and clay-rich debris flows) into cohesive and noncohesive types was made by Russian investigators (Kurdin, 1973, p. 311). However, noncohesive flow was defined as containing 80 percent sediment by weight (approximately 60 percent by volume), and the particles were described as being deposited and sorted as flow velocity slowed (Kurdin, 1973, p. 315). Although the sediment content was that of debris flow, the behavior described is mainly characteristic, not of debris flow, but of hyperconcentrated streamflow as noted by Costa (1984, p. 289). The subdivision of sediment-gravity flows and deposits into the cohesive and cohesionless classes of soil mechanics (for example, Postma, 1986) is probably only theoretically useful. Several workers have implicitly or explicitly questioned the reliability of clay or matrix content (for example, Lowe, 1979, 1982; Nemec and Steel, 1984) as the main criterion for distinguishing between ancient deposits of cohesive and cohesionless debris flows. This criterion is probably not useful for deposits of subaqueous debris flows, for which only the texture may be known and nothing at all may be known about the involvement of processes such as fluidization, escaping pore fluids, and the modified grain flow of Lowe (1976).

Differences in behavior between flows having different matrix properties are more obvious where, as at a volcano, the deposit of a modern or postglacial flow can be seen from beginning to end, and each textural, stratigraphic, and morphologic nuance can be known. In some modern (1982) volcanic flows, even the postdepositional changes in matrix character can be defined (Scott, 1988b). Noncohesive debris flow, as defined here, is flow that retains sufficient strength (albeit with lower matrix cohesiveness than cohesive flow) to produce the diagnostic characteristics of debris flow deposits: transversely and longitudinally convex flow fronts, lateral levees, buoyed dense megaclasts, and a texture of commonly dispersed casts, pebble size or coarser, in a finer grained granular matrix. Despite these similarities, however, the behavior of noncohesive debris flows differs radically from that of their cohesive counterparts.

Neither type of debris flow has truly irreversible sediment entrainment, one of Hooke's (1967) criteria for distinguishing a debris flow from a water flow; both cohesive and noncohesive types may leave coarse, clast-supported whaleback bars at sites of rapid energy loss (Scott, 1988b). Only noncohesive flows lose coarse sediment at a rate sufficient to cause transformation downstream to more dilute flow types, initially hyperconcentrated flow. Some cohesive flows show a slight textural change in the direction of that transformation, probably by particle settling within the rigid central plug of the flow, and by periodic loss of coarse clasts at sites of energy loss where dispersive and other particle-impact stresses are minimal. However, cohesive flows at both Mount Rainier and Mount St. Helens traveled well over 100 km and did not transform. The net results of deposition of sediment, and its addition (bulking) in reaches of high shear stress at flow boundaries, are discussed in the subsequent section on the Osceola Mudflow.

Debris flow behavior correlates strongly with particle-size distribution (size classes shown in fig. 2), especially clay content (Hampton, 1975; Middleton and Hampton, 1976; Qian and others, 1980, fig. 3; Costa, 1984; Pierson and Costa, 1987). The description of debris flows as cohesive or noncohesive is intended to reflect an important empirical difference in behavior related to clay content, and thus to matrix cohesiveness. Silt content also contributes to the cohesiveness of a flow but is normally proportional to the clay content. Both cohesive and noncohesive debris flows have a matrix phase and a coarse-sediment phase (dispersed phase of Fisher and Schmincke, 1984). The coarse sediment is dispersed throughout the matrix phase in most cases, but not all. Figure 2 shows particle-size distributions for representative cohesive and noncohesive flow deposits and for the other flow types into which the noncohesive debris flows normally transform downstream.

The existence of a spectrum of debris flow behavior is implicit (see Lowe, 1979) in the original Coulomb viscoplastic model of a debris flow (Johnson, 1965; Yano and Daido, 1965) and, regardless of rheological model, a range of behavior in response to varying sediment properties and content has been observed as noted above. Some flows are clearly dominated by viscoplastic behavior resulting primarily from momentum exchange within a "sticky" fine-grained matrix (cohesive flows of this report); others are more granular flows dominated by momentum exchange between coarser particles (noncohesive flows of this report) that are, however, still part of the matrix. Sand-size sediment dominates the matrix of noncohesive flows at both Mount Rainier and Mount St. Helens.

Although the two flow behaviors are distinct, basic mechanisms of particle support likely overlap. A greater abundance of particle collisions probably explains the more pronounced shear-related boundary features and cataclasis in the noncohesive flows as well as the transformations of those flows to and from other flow types. It may also explain some previously divergent assumptions of debris flow rheology, and it clearly accounts for some of the difficulties in modeling their behavior. The fundamental distinction is pragmatic, however, and a view of debris flows as having only two distinct types of momentum exchange and particle support does not fully consider the evidence of other dynamic interactions between their solid and fluid constituents (Iverson and Denlinger, 1987). Nevertheless, the distinction based on clay content is highly useful and it rationalizes many of the features and behavioral variations of lahars. Based on observations at volcanoes around the Pacific Rim by the senior writer, the distinction is generally applicable.

Figure 2. Cumulative curves of particle sizes in a typical cohesive lahar (Electron Mudflow) and noncohesive lahar (National Lahar). Other flow types into which the noncohesive debris flows normally transform downstream are also shown. (click on image for an enlargement in a new window)

The observation that a species of debris flow transforms to and from other flow types also can add greatly to our knowledge of a volcano's past. For example, the presence of large noncohesive flows and their downstream transformations, or of synchronous flows in more than one watershed, can indicate unrecognized eruptions (magmatic or phreatic) or shallow landslides mobilized to debris flow. Also, the identification of debris-flow-related deposits is aided by knowing that noncohesive flows attenuate more rapidly than cohesive flows and that a noncohesive debris flow upstream may be marked on flood plains downstream by deposits of hyperconcentrated or normal streamflow, which are less conspicuous than debris flow deposits (Scott, 1989).

The formation of debris flows from flood surges is the dominant formative process at some Cascade Range stratovolcanoes, such as Mount St. Helens (Scott, 1988b), but apparently is less common in other environments. The process probably does not involve pure autosuspension (Bagnold, 1962; Southard and Mackintosh; 1981) and is greatly facilitated by large sediment contributions from bed and bank mobilization. The efficacy of the process is dramatically illustrated at Mount St. Helens by the huge lahar (PC 1) that consists almost entirely of stream-rounded alluvium (Scott, 1988a). In constructing a conceptual model of debris flow formation, Johnson (1984, p. 331) cited one example (Jahns, 1949) in which a debris flow resulted from the bulking of sediment from channel erosion by a clay-water mixture. Costa (1984) also cited several cases in which this mechanism probably occurred, and Rodolfo (1989) documented the process for rain-induced lahars. A surge from a moraine-dammed-lake breakout quickly bulked to debris flow in the Bol'shaya Almatinka River in Russia and continued to enlarge downstream (Yesenov and Degovets, 1979). The same mechanism formed debris flows from breakouts of moraine-dammed lakes at the Three Sisters volcanoes, Oregon (Laenen and others, 1987, 1992). The requisite factor both for bulking to debris flow and for continued enlargement is an abundance of loose, poorly sorted volcaniclastic and morainal sediment on steep slopes.

Usage here of the popular but variously defined term "lahar" for volcanic debris flows corresponds, with one exception, to its application by Crandell (for example, 1971) throughout the Cascade Range. Glacial-outburst floods that bulk to debris flows but lack evidence of triggering by volcanism are not here called lahars; the term is reserved for, and most usefully applied to, flows that are directly or indirectly related to volcanism rather than merely the alpine environment. Other characteristics, such as composition or angularity of debris, are not definitive. Even origin on a volcano is not a reliable criterion, for in some cases bulking may not produce a debris flow until the surge is beyond the volcanic edifice. Some details of terminology are discussed by Scott (1988b). To be consistent with most formal and informal usage in the Cascade Range, the term is applied here, as by Crandell (1971), to both the flow and the deposit. Future workers may wish to conform to the definition recommended by the 1988 Penrose Conference on Volcanic Influences on Terrestrial Sedimentation (Smith and Fritz, 1989), which is essentially the definition used here except that the flow deposit is excluded.

COHESIVE DEBRIS FLOWS (MORE THAN 3 TO 5 PERCENT CLAY)

The largest lahars at Mount Rainier were recognized as relatively clay-rich by Crandell (1971), who logically hypothesized that a clay content of about 5 percent or more reflected an origin directly from large landslides. The clay is an alteration product of the hydrothermal system of the volcano. These flows remained debris flows to their termini. At Mount St. Helens, the critical clay content that characterized the nontransforming cohesive flows was a minimum of 3 percent. The apparent difference in critical clay content between the two volcanoes, as it affects flow behavior, mainly reflects sampling procedure and is discussed below. Hydrothermal alteration is more intense at Mount Rainier, probably as a result of its greater age. This difference is the probable cause of the dominance of large cohesive flows at Mount Rainier, whereas large noncohesive flows dominate the record at Mount St. Helens (Scott, 1988a, 1989; Major and Scott, 1988). The original and type example of a cohesive flow is the 1980 North Fork Lahar at Mount St. Helens (Scott, 1988b).

The differences in behavior of cohesive and noncohesive debris flows correlate strongly with the texture of the matrix phase: the matrix of cohesive debris flows is a mix of sand, silt, and at least 3 percent clay; that of noncohesive debris flows is silty sand with commonly about 1 percent clay. In cohesive debris flows, (1) grain interaction is cushioned by the adhering clay aggregates, thereby reducing near-boundary shear and other particle interactions recorded by the boundary features characteristic of noncohesive debris flows; and (2) the clayey matrix retards each of the following: (a) the settling of coarse particles, (b) the differential movement of all coarse-phase particles (which produces the well-developed normal and inverse grading in noncohesive flows), and (c) the miscibility of the flow with associated streamflow. The latter effects prevent or greatly retard the transformation of a cohesive debris flow to hyperconcentrated streamflow. These conclusions are empirical; the actual physics and chemistry of clay in the matrix remain to be investigated. For example, clay content may affect the viscosity of the pore fluid and, therefore, the hydraulic diffusivity of that fluid through the granular phase (Iverson, 1989). Such movement may be slight in cohesive flows, where interparticle attractive forces can dominate, but in the noncohesive regime, the character of the medium around colliding and abrading particles of the matrix must be important.

Clay minerals compose most clay-size sediment, but their proportion is variable in lahars. They compose 75 percent of the clay-size sediment in the largest lahar from Mount Rainier (Crandell, 1971). Clay minerals are layer-lattice silicates with powerful surface forces that can provide cohesion and strength to the entire flow. Clay aggregates in turn adhere to sand in the matrix as well as the coarse-phase clasts. Varieties of clay minerals reported from the edifice of Mount Rainier and the lahars derived from it include kaolinite, montmorillonite, smectite, halloysite, illite, and pyrophylite (Crandell, 1971; Frank, 1985).

The failed sectors of the volcano contained enough water and clay to provide uninterrupted mobility as they rapidly disaggregated, first to a debris avalanche and then to a lahar. A debris avalanche is a rapid flow of rock debris (Vanes, 1978), wet or dry, commonly containing many large megaclasts. Studies of the May 18, 1980, eruption at Mount St. Helens suggest that some cohesive flows may have been derived from the surface of an immobilized debris avalanche. The surficial portion of the huge 1980 debris avalanche at Mount St. Helens was saturated by dewatering after emplacement, thereby forming a critical mass of ponded muddy debris which began flowing as a broadly peaked lahar wave several hours later. All large-scale debris avalanches recorded by known deposits at Mount Rainier mobilized directly to lahars. The only debris avalanches known to have yielded lahars secondarily were small examples of shallow origin.

NONCOHESIVE DEBRIS FLOWS (LESS THAN 3 TO 5 PERCENT CLAY)

These more granular debris flows commonly represent the middle segments of meltwater flood surges (either volcanically or climatically induced) that both begin and end as streamflow surges. They are generally better sorted and maybe finer grained on average than the larger, boulder-rich cohesive debris flows (fig. 3). The initial water surges incorporate most sediment from stream-channel deposits that are partly depleted of fine sediment as a result of selective or hydraulic sorting by fluvial processes acting both subglacially and proglacially. Consequently, through bulking, the surges form debris flows that have an average clay content of only about 1 percent. The original and type example of a noncohesive flow is the lahar of March 19—20, 1982, at Mount St. Helens (Scott, 1988b).

Figure 3. Sorting versus mean grain size for selected cohesive and noncohesive debris flows. Mean grain size and sorting values correspond to Mz and sG of Folk (1980). (click on image for an enlargement in a new window)

On the volcano, glacial-outburst flows are characterized by a relatively high lahar-bulking factor (LBF; the percent age of sediment added to the flow beyond the point of origin as revealed from cast roundness or composition; Scott, 1988b). Noncohesive debris flows originating as slides of relatively unaltered volcaniclastic debris have a distinctively lower LBF. This origin resembles, at smaller scale, the process by which the large cohesive lahars are formed from deep-seated failures.

In comparison with the deposits of cohesive lahars, those of noncohesive lahars document more intense particle interaction, especially near flow boundaries, where the group of boundary features described by Scott (1988b) records the effects of shearing on particles and their size distributions. These features include a distinctive sole layer, inverse graded bedding, a lahar-abraded pavement, truncated size distributions, and grain cataclasis. Such features are clearly best developed in the noncohesive flows, but are not exclusive to them.

A common but not consistent distinction between the debris flow types is a generally higher rate of attenuation of the noncohesive flows. The granularity of noncohesive flows increases their miscibility with overrun streamflow, a factor leading to their downstream transformations to hyperconcentrated flow. The effect is illustrated by the increase in transformation rate at sites of significant tributary inflow (Scott, 1988b, fig. 37). In effect, the flow becomes diluted and loses strength, and the fluid phase progressively outruns the sediment phase. The sediment component of the noncohesive debris flows is more readily deposited than that of the cohesive flows. Once transformation occurs, the peak sediment concentration characteristically lags behind peak discharge (Scott, 1988b), as in some storm-flood peaks (Guy, 1970). In cohesive debris flows, the entire mixture remains coherent and relatively constant in texture, although systematic down stream change can occur (Scott, 1988b).

HYPERCONCENTRATED FLOWS

Hyperconcentrated flow is an important flow type at most Cascade Range volcanoes. The history of its recognition and the criteria by which its deposits are recognized are described elsewhere (Scott, 1988b). The most obvious feature that differentiates these deposits from debris flow deposits is their undispersed, entirely clast-supported texture. They are distinguished from flood-surge and normal streamflow deposits by poor development of stratification, sorting in a range intermediate between those of debris-flow and flood-surge deposits, locally well-developed inverse or normal grading, and the local presence of dewatering structures such as the dish structure of Wentworth (1967) and Lowe and LoPiccolo (1974). Dish structure is previously reported only from deposits of sediment-gravity flows in the deep marine environment.

Bulking of a flood surge to a debris flow commonly occurs on the steep flanks of the volcano in confined channels. In this setting, the granular deposits of the hyperconcentrated flow interval are thin and rarely preserved. Channel steepness and an abundance of unstable detritus commonly result in rapid bulking in a short increment of channel. The debulking, in contrast, may occur over a long distance. In the streams draining Mount Rainier, the longest documented intervals of hyperconcentrated transport in single flows occurred in the Nisqually River from Longmire to below Alder Reservoir, more than 40 km. and in the White River from near the base of the volcano to beyond the boundary of the Puget Sound lowland, over 70 km.

A distinctive facies distinguishes the interval where debris flow transforms to hyperconcentrated flow (Scott, 1988b, fig. 10). This "transition facies" begins to form as the front of the flood wave transforms, continues as the change works its way progressively back through the debris flow, and ends at the point where the entire wave becomes hyperconcentrated flow. The preserved record of this transition interval, therefore, consists of downstream-thickening hyperconcentrated flow deposits overlain by downstream-thinning debris flow deposits. This transition facies thus documents the origin of hyperconcentrated flow from an upstream debris flow.

At Mount St. Helens, the hyperconcentrated flows were described as lahar-runout flows and interpreted as evidence of upstream lahars, based on the presence of the transition facies (Scott, 1988b). The record of ancient and modern flows at Mount Rainier confirms that most, if not all, of the significant hyperconcentrated flows there had such an origin. At most Cascade Range stratovolcanoes, the steep slopes and abundance of volcaniclastic sediment assure that any significant flood surge will bulk to debris flow.