Abstract Classifications of flowing sediment-water mixtures have, in the past, been based primarily on relative, qualitative differences in the style and rate of movement as well as on morphology and sedimentology of deposits. A more quantitative and physically relevant classification is presented here, based on thresholds in rheologic behavior. The classification is constructed on a two-dimensional matrix in which flows are located according to deformation rate (mean velocity) and sediment concentration, with composition of the mixture constant. Three major rheologic boundaries are crossed as sediment concentration increases from 0 (clear water) to 100 percent (dry sediment): (1) the acquisition of a yield strength—the transition from liquid “normal streamflow” to plastic “hyperconcentrated streamflow”; (2) an abrupt increase in yield strength coinciding with the onset of liquefaction behavior—the transition to “slurry flow”; and (3) the loss of the ability to liquefy—the transition of “granular flow.” These three rheologic boundaries shift according to particle-size distribution and composition of the mixture. Processes controlling flow behavior depend on deformation rate (velocity). Rate-independent frictional and viscous forces dominate at lower velocities and in finer grained mixtures; rate-dependent inertial forces dominate at higher velocities and in coarser grained mixtures. As velocity increases, grain-support mechanisms change from low-energy varieties (buoyancy, cohesion, structural support) to progressively higher energy mechanisms (turbulence, dispersive stress, fluidization). Existing nomenclatures of geologic flow phenomena can fit within this rheologic classification. The morphology and sedimentology of flow deposits commonly can be used to deduce rheologic behavior, but caution needs to be exercised in inferring processes from deposits.

Abstract Japanese concepts of modeling debris flow are thoroughly reviewed in this chapter. Many Japanese models, ranging from highly theoretical non-Newtonian fluid models to the very simple empirical relations of Bingham and Bagnold, are evaluated in terms of accuracy, generality, and practical usefulness, and are compared with a generalized viscoplastic fluid model described herein. Most debris flow formulas and criteria presently used in Japan are closely related to those developed by Takahashi on the basis of Bagnold’s “dispersive” pressure concept. Although the generality of Bagnold’s model is still at issue, Japanese scientists apparently have accepted Takahashi’s debris flow formulas and criteria. For example, Takahashi’s velocity profile for (steady) uniform debris flow in wide channels is only valid for grain-intertia regime. Applying Takahaski’s solution to modeling other flow regimes than the grain-inertia may thus have to adjust the value of Bagnold’s numerical constant in order to better fit the computed velocities to the measured ones. This and many other aspects of debris-flow modeling concepts in Japan are critically examined. An appraisal of the present status of Japanese research in debris flow modeling helps determine the direction of future efforts in debris flow research.

Abstract The thousands of debris flows that mobilized from shallow slides in the San Francisco Bay region during the rainstorm of January 3-5,1982, left evidence of the range in soil textures susceptible to mobilization and of differences in completeness and speed of mobilization. These differences in mobilization are related to a broad range in the ratio of saturated water content to liquid limit, which we have used as an approximate index of mobilization potential. To understand such differences in mobilization, we have explored the transformation from slide to flow, using relations among inplace void ratio, void ratio needed for flow from the slide scar, and the steady-state line. These relations define two principal means of direct transformation from slide to flow: contractive soil behavior, which commonly results in liquefaction, and dilative soil behavior, which in many cases probably results in partial mobilization of the slide mass. These means of mobilization determine the completeness of mobilization of slides and the time required for mobilization; they also influence the thickness and lumpiness of deposits, as well as the travel distance of debris flows. These relations permit means of mobilization to be predicted in both an approximate and a precise manner through soil testing.

Abstract Very large rock avalanches, involving more than about 10 6 m 3 of rock debris, exhibit anomalously low coefficients of friction. Consequently they travel much farther than conventional slope-stability criteria predict. Such long-runout landslides ( sturz-strom ) include the catastrophic Elm (1881), Frank (1903), and Sherman Glacier (1964) events. Attempts to explain this behavior have considered water or air lubrication, local steam generation, or even the formation of melt layers within the rock debris. Discovery of deposits of such landslides on Mars and the moon, however, appears to rule out the fundamental involvement of volatiles or atmospheric gases in the flow mechanism. It appears that large, high-frequency pressure fluctuations due to irregularities in the flow of the debris may locally relieve overburden stresses in the rock mass and allow rapid pseudoviscous flow of even dry rock debris. If the avalanche volume is large enough, the rate of production of this vibrational (acoustic) energy exceeds its loss rate, and sustained motion is possible. Small-scale laboratory experiments have verified theoretical predictions of the rheology of such acoustically fluidized debris. This rheology is consistent with the rate and pattern of observed large rock avalanches. Although much work remains to be done, acoustic fluidization is the most plausible explanation of the fluidity of large, dry debris avalanches.

Abstract There were two lahars that reached the Toutle River during the eruption of Mount St. Helens, Washington, on May 18,1980. The North Fork lahar was much larger than the South Fork lahar, had a much more rectangular hydrograph shape, and was much more destructive. Hydrographs (graphs of discharge versus time) constructed for both lahars demonstrate that differences between the lahars existed as close to the lahar sources as measurements were made, indicating that differences in processes that initiated the lahars must have been responsible for observed disparities between the lahars. The South Fork lahar was apparently generated when a laterally directed pyroclas-tic cloud triggered slab snow avalanches, and then rapidly incorporated and melted the snow. The North Fork lahar was generated from a small portion of avalanche debris in which ice was comminuted to an abnormally small size. This ice melted rapidly and saturated the host avalanche debris, which then liquefied during a long harmonic tremor event. The North Fork lahar differed greatly from the South Fork lahar because of significant dissimilarities between the pyroclastic cloud and harmonic tremor sequence that were directly responsible tor the characteristics of each lahar. Because differences in lahar characteristics were ultimately responsible for the contrast in destructiveness, I have concluded that the process of initiation is an extremely important factor controlling downchannel destruction. The importance of initiation must be accounted for in the quantitative analysis of lahar hazard.

Abstract Development within the mountains of coastal British Columbia has recently increased the exposure of people and facilities to debris flows. Attempts to specify weather conditions under which debris flows are apt to occur—such as threshold precipitation—appear not to work because of the highly contingent nature of the flows. Debris must exist in unstable position in or near the channel, and conditions prior to the flow may strongly condition the necessary trigger to mobilize it. Events have been observed in the following circumstances: locally concentrated rainfall with high antecedent moisture and no snowmelt (the “classical case”); uniformly distributed, moderate rainfall with snowmelt; low intensity rainfall and heavy snow-melt; and heavy rainfall onto deeply frozen, but thawing, ground. A weather-based warning threshold for the British Columbia coast would be fairly complex. At present, such a system would include the substantial probability of issuing nuisance predictions of nonoccurring events. There is an indication that the incidence of debris flows has increased since 1980. Reasons why this might be so are investigated. Aside from the occurrance of four very wet years since then, no clear meteorologic correlation can be made.

Abstract In early November 1977, a storm system that formed in the Gulf of Mexico moved northeastward into the Appalachian Mountains. It produced intense (as much as 102 mm/hr) and heavy (200-300 mm) rainfall that set off debris avalanching in steep terrain of the Pisgah National Forest, North Carolina. Antecedent rainfall during September and October was 177 percent of normal and the wettest on record for these 2 months. The storm began on 2 November, and rainfall was relatively continuous and even (20-50 mm/day) for the next 3 days. The long-duration rainfall was capped by intense convective downpours the night of 5-6 November when debris avalanching occurred. Peak intensities measured at 15 gauges near Asheville, North Carolina, ranged from 21 to 102 mm/hr, with nearly half exceeding 75 mm/hr. Return intervals for peak intensity rainfall in the range of 75 to 102 mm/hr are 50 to 200+ yr. Total storm rainfall for these gauges ranged from 35 to 250 mm, with peak 24-hr rainfalls of 30 to 180 mm. Rainfall intensities for 1-, 3-, 6-, 12-, and 24-hr periods at a gauge near one avalanching site were 69,137,159,164, and 180 mm, respectively. Development of the storm was monitored by GOES infrared satellite imagery in real time, and flash flood warnings were issued. Debris avalanching and high stormflow produced peak stream flows with return periods ranging from 20 to 100+ yr. The largest debris avalanches occurred on steep slopes (70% +), started at high elevations (900-1,100 m) in shallow residual soils (less than 1 m deep), had tracks commonly greater than 700 m, and carried a volume of material averaging 2,500 m 3 per avalanche.

Abstract Rainfall intensity and duration of storms has been shown to influence the triggering of debris flows. After examining storm records of the San Francisco Bay region, documenting when debris flows occurred, and measuring piezometric levels in shallow hillside soils, continuous high-intensity rainfall was found to play a key role in building pore-water pressures that trigger debris flows. Debris flows in 10 storms between 1975 and 1984 in a 10-km 2 area near La Honda, California, were examined, and their rainfall records compared to the records of other storms to determine the antecedent conditions and the levels of continuous, high-intensity rainfall necessary for triggering debris flows. No flows were triggered before 28 cm of rainfall had accumulated each season, which suggests that prestorm soil-moisture conditions are important. After this sufficient antecedent rainfall, a threshold of rainfall duration and intensity—which accounted for triggering at least one debris flow per storm within the study area—was identified. The number of debris flows increased in storms with intensity and duration characteristics significantly above this threshold. By studying where debris flows initiated in storms of different intensity and duration, debris flow susceptibility was found to depend on soil thickness and hillside concavity and steepness. Moderate intensity storms of long duration triggered complex soil slump/debris flows in thick soils on concave slopes below large drainage areas, whereas high-intensity storms of short duration caused complex soil slide/debris flows in thinner soils without respect to size of drainage area. From these observations, an empirical model based on geology, hydrology, and topography is proposed to account for the triggering of debris flows at selective sites by storms with different combinations of intensity and duration once the antecedent and intensity-duration thresholds are exceeded.

Abstract Debris flows following Are are a common, but poorly understood, problem in southern California. Research to date suggests that they result from greatly accelerated rates of surface erosion by both wet and dry processes during the days and weeks following a fire. Significant amounts of hillslope debris are delivered to stream channels during the fire by a process called dry ravel. An important feature of postfire erosion is the rapid development of extensive rill networks on hillslopes. These rill networks are linked to a layer of water-repellent soil that forms a few millimeters below the ground surface during the fire. These rill networks result from numerous, tiny debris flows that occur on the hillslopes during the early storms. The rill networks form rapidly, often in a matter of minutes, and provide an efficient means for transporting surface runoff to stream channels. This helps explain why postfire debris flows often occur during very small storms and after short periods of rainfall.

Abstract This chapter presents a method by which morphometric criteria can be used to obtain a rapid first-approximation of potential debris flow hazard on alluvial fans in the Canadian Rocky Mountain Front Ranges. Geomorphic and sedimentologic evidence indicates that many fans are affected by debris flow processes. Such fans generally are steeper than 4° and have small, steep first- or second-order drainage basins with Melton’s ruggedness number ( R ) more than 0.25 to 0.3. Fans not prone to debris flows are dominated by fluvial processes and have gentler slopes in less rugged third-order or higher drainage basins. This morphometric approach should have wide applicability for continuously graded basins in unglacierized regions.

Abstract The central and southern Appalachian region experiences intense rainfall events that trigger episodes of debris slides and debris flows. High rainfalls may be preceded by wet periods, normal conditions, or droughts, and still result in rapid mass movements. Most slides and flows occur in existing hillslope depressions and move downslope. The bedrock-soil contact is the most common movement interface, although slippage and flowage are also common in deep soils. Lithologic, structural, soil, vegetative, and land-use influences on mass movements are identifiable in some areas, yet not apparent in others. Better data on precipitation thresholds, movement mechanisms, and slide and flow precursors are urgently needed. Accelerating tourism growth rates and development of mountainous areas are accompanied by greater losses of human property and life caused by slope failures. The dangers of rapid debris slides and flows threaten increasing numbers of people in developing areas.

Abstract Debris fans in low-order Appalachian Mountain drainage basins can be used to estimate the return periods between catastrophic debris flow events such as the Hurricane Camille storm of 1969 in Virginia. Debris fans in Davis Creek, Virginia, have been the sites of repeated debris flow deposition at least three times during the last 11,000 years. Debris flow frequency estimates are possible if individual events can be recognized in the fan stratigraphy. Discrimination of events is based on the recognition of paleosols, and on abrupt changes in sediment texture and in matrix composition at suspected event boundaries. Major controls on slope stability appear to include the orientation of the slope, bedrock structure, and presence of colluvial hollows at the sites prior to slope failures. Hollows are sites of between-event accumulation of colluvium, and are areas of subsurface water concentration during heavy rains. Tropical air masses seem to have been a factor in most historical Appalachian debris flows. The early Holocene initiation of debris flow activity on the central Virginia fans appears to coincide with paleoclimatic data, indicating the commencement of conditions that permitted the invasion of tropical moisture into the region at the close of Pleistocene time.

Abstract Debris flow activity in the Whitney Creek basin of Mount Shasta is caused by incisement of soft pyroclastic beds in upper fan areas, and is the dominant late Holocene geomorphic process. A variety of geologic and botanical techniques permit the dating of many debris flows. These methods aid in the interpretation of recent denudation rates and late Quaternary geomorphic changes at Whitney Creek gorge. Geologic techniques used for dating and interpreting debris flows included carbon-isotope analyses of wood and charcoal samples, stratigraphic relations, analysis of aerial photography, and particle-size analyses of sediment deposits. Relatively recent debris flows were dated dendrochronologically using tree ages, eccentric growth-ring patterns following tree tilting by a debris flow, suppression and release sequences, and corrasion scars caused by debris flow impacts on tree trunks. Results indicate intense debris flow activity along upper Whitney Creek during recent centuries; a minimum of 10 debris flows are identified for the last 420 yr. Sediment yields and denudation rates estimated from debris flow frequency and volume data suggest that activity has been most intense in the last five centuries. Sediment thicknesses on lower parts of the Whitney Creek fan appear sufficient to account only for deposition rates during late Holocene time. If present rates of deposition had prevailed throughout Holocene time, the average thicknesses of the lower fan deposits would be at least eight times greater than they are.

Abstract Hollows are the concave-out portions of hillslopes not occupied by channels. The topographic convergence in hollows forces colluvial debris to accumulate and causes shallow subsurface runoff to be concentrated during storms. Consequently, hollows are more susceptible to landsliding than side slopes and constitute important mappable source areas of debris flows. Hollows can be extremely subtle topographic features that require recognition in the field; these subtle hollows are commonly tributary to larger hollows, and greatly increase the density of mappable debris flow sources. In a study area in Marin County, California, hollows are spaced 20 to 60 m apart along the slope, resulting in a density of 25 to 35 km of hollow axis per km 2 . Even the subtle hollows can produce debris flows capable of destroying houses, particularly when large trees are carried by a flow. Mitigation measures that focus on draining the main hollow axis may be inadequate because of the destructive ability of debris flows shed from small tributary swales and from side slopes. Road runoff discharged onto hollows can trigger landsliding and gullying, but this problem can be prevented by extending culverts downslope to stream channels. Along the drainage network, from subtle tributary hollows to major hollows, and to first-order channels where many additional hollows enter, the recurrence interval of debris flow events probably systematically decreases as the number of upslope sources increases, perhaps reaching the lowest recurrence interval on second-order channels. Farther downstream, debris flows may occur less frequently. A greater emphasis on hollows as debris flow source areas and as paths for flows from upslope should make a significant contribution toward identifying the hazard to existing structures and toward improved siting of new development.

Abstract A major debris flow occurred on January 4,1982, in the Oddstad Boulevard area of Pacifica, California. The flow emanated from a previously unrecognized colluvium-filled swale (one of many making up first-order drainages in the region), moved down a 21°, 172-m-long slope, and extended into an urban area. The failure involved the upper 4.5 m of a 6.1-m-thick colluvial section in the upper of two bedrock basins underlying the swale. Soil-stratigraphic measurements show that upper-basin colluvium accreted slowly to form a cumulic soil profile, characterized by thick surface (mollic epipedon) and subsoil (argillic) horizons. An approximately 500-yr-old mean residence time (MRT) radiocarbon date from the prefailure mollic epipedon indicates that the average sedimentation rate was about 0.6 m/1,000 yr and, accordingly, that colluviation began at least 8,000 to 10,000 yr ago. In contrast, the lower basin is characterized by at least four pre-1982 slide deposits. These deposits emanated almost wholly from within the lower basin, and are distinguished by clast lithology and angularity, and by the local presence of capping buried paleosols. Radiocarbon MRT dates of approximately 2 to 3 ka for the upper, older debris flows, and the presence of a moderately developed argillic horizon on an underlying flow, suggest that lower basin failure recurrence is on the order of 1,000 to 4,000 yr. A simple, three-stage evolutionary model for the Oddstad swale is postulated for engineering-geologic comparisons with swales elsewhere: (1) initial basin incision by fluvial processes in late Pleistocene time; (2) change of climatic regime and resultant colluvial filling of the upper basin in Holocene time; and (3) exhumation and renewed fluvial incision of the upper basin following the 1982 debris flow.