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

Risk analysis is a generic term for methods that support decision-making by quantifying consequences (magnitude and extent of lahars, for example) and the probabilities of their occurrence (frequency of lahars) (National Research Council, 1988). Which of the types of flows in table 5 pose sufficient risk to influence downstream hazards planning? An initial premise is that volcanic debris flows and their transformations can be treated like other hydrologic hazards. That is, flow events of equivalent frequency require the same planning awareness, whether the flow wave consists of sediment moving interstitial water (debris flow, and the upper range of hyperconcentrated streamflow) or water moving sediment (floods, and the lower range of hyperconcentrated flow). Floods and volcanically induced flows can be treated as separate components of a mixed population with a minor overlap in their scales of magnitude: that is, the high end of the flood scale overlaps the low end of the volcanic flow scale. Pragmatically, the risks are additive. The chief practical differences between inundation by floods and inundation by lahars are the destructive impact forces of a lahar and the long-term effects of its deposits as contrasted with the ephemeral inundation by a flood.

"Hazard" refers both to the agent and to the potential for harm posed by that agent. Also, risk can be said to exist when something of value is at jeopardy. Thus, in the general case of volcanic hazards (Dibble and others, 1985):

(1) RISK = HAZARD x VALUE x VULNERABILITY,

where HAZARD is an event of known probability, VALUE is the economic assessment of loss, and VULNERABILITY reflects susceptibility for harm, which may vary for different things affected by the same hazard. The inclusion of the latter term is extremely valuable in assessing volcanic flow hazards.

A similar approach to the dangers of volcanic flows is:

(2) RISK = FLOW MAGNITUDE x FLOW FREQUENCY x VALUE x VULNERABILITY,

where each flow subpopulation can be treated separately and ranked by the risk it poses. Although the results (table 6) are qualitative, they clearly separate the differing risk of each flow type and provide a logical basis for the quantitative analysis of individual case histories of flows that represent the flow types that pose the greatest risk (pl. 1). In this initial ranking, MAGNITUDE is replaced by a convenient surrogate, area of inundation, which is based on the extent of the flows as established by their deposits (tables 2, 3). FREQUENCY is the probability of each flow type, or the inverse of the recurrence interval. VALUE is also proportional to inundation area, but its inclusion is necessary to assess the relative risks of different size flows. At Mount Rainier, population and property values increase downstream in each watershed, approximately exponentially, but with a large increase as flow reaches the Puget Sound lowland (data from Pierce and Thurston Counties, Washington). Consequently, including a VALUE term correctly emphasizes the catastrophic potential of the larger flows.

The VULNERABILITY factor in equation 2 significantly affects the danger of certain flow types. That is, vulnerability to a flow type is reduced if there is the probability of a warning in the form of volcanic activity precursory to the flows. People and movable objects in the path of rapid debris avalanches at Mount Rainier are far more vulnerable than those near the attenuating debris flows of glacial-outwash or rainfall origin. Vulnerability also depends on probable reservoir levels and whether they can be drawn down in the event of a warning. For example, vulnerability is reduced by the fact that Mud Mountain Reservoir on the White River is solely a flood-control structure and is thus normally empty.

No single flow type and origin will pose the greatest hazard throughout an entire river system. On the highly populated Puget Sound lowland, the huge sector-collapse debris avalanches mobilized as lahars (flow 1, table 6) pose the greatest danger. In valleys on and immediately adjacent to the volcano, noncohesive lahars (flows 2, 3, or 5, table 6) and debris avalanches (flow 4, table 6) pose the greatest danger. And, for hikers along proglacial streams on the volcano, a debris flow formed from a glacial-outburst flood (flow 6, table 6) is the greatest statistical risk.

FLOW FREQUENCY AND RISK AT MOUNT RAINIER

This discussion focuses on flows of the frequencies most commonly used in long-term hydrologic planning—100 and 500 years (Brice, 1981). These recurrence intervals correspond to probabilities of 1 percent and 0.2 percent per year. By contrast, Latter and others (1981) believe it is "desirable" to incorporate events with recurrence intervals of 1,000 and perhaps 10,000 years when assessing volcanic risk. Although practice is variable, design frequency for bridges on primary roads is commonly 50 years, with some states using a 50-yr flood for the bridge superstructure and a 100-yr flood for the substructure (Brice, 1981). Flow frequencies for structures such as reservoirs and power plants are commonly lower (that is, return periods are higher) than these values and are commonly controlled by economic factors (Linsley and others, 1958).

In occurrence, lahars at Mount Rainier differ from those at Mount St. Helens in an important way. The latter have a significant tendency to cluster in groups, and their time distribution can be analyzed both in an eruptive period, as at present, or over any other time interval. At Mount Rainier, in contrast, both volcanism and lahars are scattered throughout postglacial time (tables 2—4). Therefore, the occurrence of one large lahar does not increase the odds of a second, as it does during the modern eruptive period at Mount St. Helens. The assumption of basically random occurrence, as in flood analysis, is probably valid at Rainier.

All recurrence intervals discussed here are based on mountain-wide occurrences over undivided intervals of postglacial time. This dispersion of risk, rather than its definition within each river system, reflects the uncertainty in knowing what river system or systems will experience the next major lahar. For example, the Carbon River system records the lowest frequency of lahars. However, considering the modern topography and structure of the volcano, that river system may have substantial risk of conveying part or most of a huge, sector-collapse lahar. The river system also contains a large volume of glacial ice that, although covered with insulating rockslide debris, is subject to melting and thus to the formation of noncohesive lahars. The example illustrates the need to reassess risk once the location of any precursor intrusive activity is evident. For example, volcanic activity affecting the Carbon River sector will pose an extreme risk of large debris flows.

Conversely, the White River system illustrates the possible temporary reduction in risk of a second large lahar following a significant sector collapse and before edifice reconstruction. The crater remaining after the Osceola Mudflow is now largely infilled, however, and the original failure plane could facilitate renewed failure. Correlations between changes in risk and the occurrence of flows are complicated, perhaps hopelessly so, by the lack of knowledge of hydrothermal alteration and structure within the edifice. Supporting evidence of a temporary risk reduction is not definitive and, at Rainier, cohesive flows have recurred in the same drainage.

Other factors also support a volcano-wide risk assessment: (1) large cohesive flows have recurred from a single drainage; (2) a single flow has affected more than one drainage (Osceola and Round Pass Mudflows); (3) three of the river systems, the White, Puyallup, and Carbon Rivers, join downstream within range of Rainier lahars; and (4) a major explosive eruption like that at Mount St. Helens in 1980 would produce lahars in all of the main drainages. The situation is largely analogous to arid-zone flood-hazard mapping where, although only one sector of an alluvial fan will probably be affected by any given flood, all parts must be considered potentially prone to inundation (Scott and others, 1987; Scott, 1992).

The length of time needed to evaluate frequency depends on flow size—the smaller the flow type, the shorter the time span needed to establish recurrence interval statistically. The time intervals selected, such as the post-Y time interval used for the definition of noncohesive lahars, are in part a function of geological convenience, but each is sufficient to define flow probability. Even if older and smaller flows are eroded or obscured (a possibility given the number of postglacial episodes of aggradation and degradation in entire river systems), the analysis is not affected substantially.

Lahars are far more numerous than episodes of known volcanic activity (producing juvenile eruptive products) at Mount Rainier. Neither cohesive nor noncohesive lahars correlate well with volcanism, and many of the latter probably resulted from geothermal heat flux and steam eruptions. The noncohesive lahars that formed by bulking of meltwater surges are not obviously linked to the most clearly recorded eruptions, those producing tephra. Prevailing west winds have distributed Rainier tephras on the east side of the volcano (Mullineaux, 1974), yet the Cowlitz River (east side) has a sparser record of lahars than all other Rainier drainages except the Carbon River (northwest side), and the large number of flows in the Nisqually River (southwest side) is similar to that in the White River (northeast side). The lack of a clearly recorded association of the large cohesive lahars with known volcanism is discussed by Crandell (1971), and his conclusion is reinforced by the ages of the additional cohesive lahars reported here.

DEBRIS FLOWS AND SUMMIT-CONE VOLCANISM

Along with the small glacial-outburst flows, a previously unrecognized grouping of lahars is an exception to the general lack of time-clustering of flows at Mount Rainier. Noncohesive lahars and derivative lahar-runout flows (tables 3 and 4) occurred throughout the post-Y time interval, as described in the section on flows of intermediate size and frequency. The deposits of these flows form much of the fill in the White and Nisqually River valleys recognized by Crandell (1971) and then believed to have had a normal fluvial origin. The last such flow occurred in 1947. There is, however, a clear concentration of flows late within the post-C, pre-W time interval. The radiocarbon dates in table 4 define flow activity that peaked between about 2,200 and 800 radiocarbon years ago, and particularly in the last 500-600 radiocarbon years of that interval. The interval is bounded by calendar ages of about 2,250 to 710 years (Stuiver and Becker, 1986).

This interval of flow activity does not coincide with an eruptive period as defined at Mount St Helens (Mullineaux, 1986). Rather, it overlaps the assumed end of lava and pyroclastic flow activity during building of the summit cone above the east rim of the volcano (fig. 4; Fiske and others, 1963, p. 80). Summit-cone lava flows are believed to have occurred between about 2,100 and 1,200 calendar years ago (Crandell, 1971, p. 14). Either lahar-producing activity associated with the construction of the summit cone continued later than believed, or later pulses of geothermal heat or steam eruptions created major meltwater surges. Geothermal activity at the modern summit (Frank and Friedman, 1974) produces only local melting.