Sedimentation and Geomorphic Changes During and Following the 1980-1983 Eruptions of Mount St

10 新砂防, Vol. 37, No. 2 (133) 昭59. 7.

SEDIMENTATION AND GEOMORPHIC CHANGES DURING AND FOLLOWING THE 1980-1983 ERUPTIONS OF MOUNT ST. HELENS, WASHINGTON* (1)

Richard J. Janda**, David F. Meyer** and Dallas Childers**

INTRODUCTION

After lying in repose for 123 years, Mount St. Helens, formerly known as the Fujiyama of America, started a continuing series of episodic eruptions on March 27, 1980. Based upon the amount of energy released as well as the diversity and volume of volcanic products produced, the May 18, 1980 eruption was the most significant eruption of this series. That eruption killed at least 59 people, caused more than S 900 million of damage and remedial work, and generally disrupted civil works and operations throughout much of the northwestern United States (Sarah Deatherage, written communication, Cowlitz County Dept. Community Development, 1983). Landscape modifications caused by this eruption have caused rapid erosion in the 550km2 devastated area and along downstream lahar-affected channels. The resulting sediment yield has created persistent sedimentation and flooding hazards of unprecedented magnitude for downstream urban areas. Post-eruption flooding along some lahar-affected channels has caused more erosion and deposition than the initial lahars. Numerous prehistoric eruptions at Mount St. Helens (Mullineaux and Crandell, 1981) and historic eruptions at other stratavolcanoes were more energetic and voluminous than the 1980-1983 eruptions of Mount St. Helens. Therefore, recent events at Mount St. Helens should serve as a reminder of the volcanic and hydrologic hazards associated with explosive stratavolcanoes and not be dismissed as a freak catastrophy. On behalf of my coauthors, I am honored to address the 1983 convention of the Erosion- Control Engineering Society of Japan. Members of your society are internationally acclaimed for their contributions to knowledge of basic processes of erosion, sediment transport, and deposition, as well as their application of that knowledge to hazard mitigation. Our paper briefly describes the manner in which recent eruptions at Mount St. Helens caused both immediate and lingering hydrologic hazards and presents examples of the types and magnitudes of post-eruption erosion. Initial attempts at mitigation are described only in passing because they were intended as interim measures, and final solutions are being actively discussed. I shall try to give you an overview of these subjects, but I shall focus most closely on stream channel processes, particularly those along the Toutle River system, because that is where our research is concentrated. The Cowlitz and Columbia Rivers are not included in this discussion because the impacts of natural processes on sediment yield and channel geometry have been largely obscured by intermittent dredging. I hope that this description will help to further the productive dialogue started at the 1982 Japanese-American Symposium on Erosion Control in Volcanic Areas in Se- attle and Vancouver, Washington (Public works Research Institute, 1983).

ACKNOWLEDGEMENTS

The eruptions of Mount St. Helens created unparalleled research opportunities and scientists

*昭和58年度砂防学会研究発表会 ,特別講演,次号に後半を掲載する。 ** U. S. Geological Survey, WRD David A. Johnston Cascades Volcano Observatory 5400 MacArthur Boulevard Vancouver, Washington 98661, U.S.A. SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 11

have been drawn to the volcano in great numbers. This paper reflects the findings of many of these scientists as well as our own research. Publications and field discussions with Randy Dinehart, Harry Glicken, Jon Major, Holly Martinson, Tom Pierson, Kevin Scott, Steve Stockton, Fred Swanson, and Barry Voight were particularly helpful in preparing this paper. Nobuo Anyoji, Hiroshi Ikeya, Takahisa Mizuyama, and Etsuro Shimokawa have been helpful in drawing analogies between erosion at Mount St. Helens and at various Japanese volcanoes. An early draft of this manuscript benefited from reviews by Gary Gallino, C. Newhall, and K. Michael Nolan.

PHYSICAL AND CULTURAL SETTING

Prior to 1980 Mount St. Helens was a symmetrical stratovolcano, rising 2950 meters above sea level and covered with 5.02km2 (0.179km3) of glacial ice (Brugman and Meier, 1981). The Mount St. Helens volcano rests on a drainage divide within an intricately dissected landscape carved into Tertiary volcanic and metavolcanic rocks (Huntting and others, 1961). The major valleys surrounding the volcano were extensively glaciated during Pleistocene glacial periods. Small glacial lakes are common above 1100 m above sea level, and glacial and glacio-fluvial deposits are widespread. Headwater tributaries of the North Fork Toutle River were originally dammed by pyroclastic flows and lahars about 4000 years ago to form Spirit Lake (Mullineaux and Crandell, 1981)-formerly a much photographed reflecting pond for the north flank of Mount St. Helens. Local relief along major valleys is typically between 400 and 700m. Hillslope gradients average 20 to 30 percent in unglaciated areas and are commonly greater than 40 percent in glaciated areas. Broad alluvial reaches of major streams have gradients of 0.005 to 0.03 and are separated by narrow bedrock gorges. The volcano-affected lower Cowlitz River, however, has an average gradient of only 0.0008. The deeply eroded bedrock valleys concentrate most water and sediment discharge from the volcano into the Toutle and Lewis Rivers, rather than allowing discharge to be radially dispersed. The Toutle River is unregulated, but the Lewis River has three hydroelectric and flood control reservoirs. Precipitation is heavy, with average annual amounts ranging from 1140mm near the Columbia River to more than 3200mm on the upper slopes of the volcano. About 75 percent of the annual precipitation falls between October and March (Phillips, 1974). During most winters a persistent snow pack develops above 1000m. Rainfall intensities are moderate with 20mm per hour occurring not more than once every 10 years (Hershfield, 1961). Most major floods are produced by prolonged periods of rain or by warm rain on water-saturated snow pack. Prior to 1980 the now devastated area was cloaked with a dense coniferous forest and used primarily for timber harvest and recreation. The two closest permanent towns are Cougar and Kid Valley (Fig. 1). The greatest concentration of lives and property within the eruption-affected area is along the lower Toutle and Cowlitz Rivers where 50,000 people live and the total assessed property value is about S 2.6 billion (Sarah Deatherage, written communication, Cowlitz County Dept. Community Development, 1983). Additional value is associated with an interstate highway, railroad, and the Columbia River, all of which are vital transportation corridors for the northwestern United States.

IMPACT OF INITIATING ERUPTIONS

The volcanic events responsible for the immediate death and destruction on May 18, 1980 also set the stage for more persistent sedimentation and flooding hazards. After nearly two months of intense outward slope movement and seismicity associated with intrusion of a dacite cryptodome, the north slope of Mount St. Helens collapsed following a magnitude 5.2 earthquake on May 18, 1980, creating one of the largest mass movements of historic time (Voight and others, 1981; 1983)*). This collapse and the subsequent eruption re- moved the top 450m of the formerly symetrical cone and formed a north-facing amphitheater- 12 新砂防, Vol. 37, No. 2 (133) 昭59. 7.

Figure 1. Location map showing localities and measurement locations discussed in text. shaped crater 600m deep. The path of the resulting avalanche of fragmental debris was strongly influenced by local topography. One avalanche lobe rammed into Spirit Lake causing a wave run-up to 260m above the original lake level and raising the lake by about 60m. Another lobe traveled northward for about 7km surmounting a 300-380m ridge at velocities of 50-70ms-1. The main avalanche lobe traveled 23km down the North Fork Toutle River valley in about 10 minntes. An area of 60km2 was buried to an average depth of 45m by about 2.8km3 of hum- mocky, poorly sorted, easily eroded debris; 56-94 percent of the deposit is less than 10mm in diameter with 15-69 percent being sand. Much of the avalanche deposit was deposited in an unsaturated condition. Water from some saturated pockets drained during subsequent lahar formation. The original avalanche surface lacked integrated drainage. Lakes formed in closed depressios on the avalanche deposit and in the lower ends of impounded tributaries.

*) While this paper was in review the following well illustrated description of the North Fork Toutle debris ovalanche deposit was published in Japanese: Ui, T, and Aramaki, S., 1983, Volcanic Dry Avalanche Deposit in 1980 Eruption of Mount St. Helens, U.S.A.: Bulletin Volcanological Society of Japan, vol. 28, p. 289-299. SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 13

The crater, avalanche deposit, and impounded lakes at Mount St. Helens are strikingly similarr to those formed in the July 15, 1888 eruption of Bandi-san (Sekiya and Kikuchi, 1889). My brief inspection of the Bandi-san deposit suggests that it contains far more angular boulders and blocks of coherent andesite than the North Fork Toutle deposit. Those coarse clasts probably contributed to the rapid stabilization and revegetation at Bandi-san. A number of other Japanese analogs of the North Fork Toutle debris avalanche deposit were recently described by Ui (1983) in a way that should help in hazard zonation around other stratavolcanoes. Release of confining pressure following the rockslide-avalanche caused phreatic and magmatic explosions that generated a ground-hugging directed blast (Hoblitt and others, 1981; Moore and Sisson, 1981; Waitt, 1981). The blast rushed outward at velocities of about 200ms-1 felling all the trees and depositing 0.01-1 m of noncohesive sediment over 550km2 of steep mountainous terrane. A 0.3-3.0km-wide zone where trees remained standing but were killed by searing heat forms the outer fringe of the blast-affected area. The directed blast deposit locally remobilized to form flat-topped blast pyroclastic flow deposits more than 10m thick in some valleys. The directed blast deposit consists of 0.19km3 of pebble-to silt-sized fragments of juvenile dacite, older rocks, soil, and shattered wood. Typically the deposit is normally graded with a relatively impermeable silty top. A vertically jetting Plinian eruption column was intiated within minutes of the directed blast and vigorously spewed forth ash of varying textures for about 9 hours (Christiansen and Peterson, 1981), Winds swept the tephra plume east-northeast. A minimum volume of 1.1km3 of uncom- pacted tephra, equivalent to about 0.20-0.25km3 of solid rock, was produced during the May 18, 1980 eruption, causing darkness eastward for more than 200km and depositing visible ash more than 1500km away (Sarna-Wojcicki and others, 1981). Tephra thicknesses, however, were not great; the area receiving 20cm or more tephra was only 16.2km2, and the area receiving 5 cm or more about 1000km2. Outside of the area devastated by the directed blast, the tephra deposits were thin enough to cause only limited mortality to fragile understory vegetation (Swanson and others, 1983 b). The depth, texture, and internal stratigraphy of the tephra varies considerably from place to place. However, the deep tephra accumulations in the Smith Creek and Clearwater Creek areas are predominatly pumiceous gravel. A surficial silty crust was originally present throughout the area affected by the directed blast and deep tephra accumulation, but it was rapidly destroyed by sheet wash and freeze-thaw action (Swanson and others, 1983 b). Starting about noon on May 18, a series of pumiceous pyroclastic flows was erupted inter- mittently for the next 5 hours. The vast majority of these flows spilled northward out of the crater where they covered a fan-shaped area of about 15.5km2 with at least 0.12km3 of deposits (Rowley and others, 1981). Indvidual flow units are 0.25 to 10m thick, and the total thickness is as much as 40m. Smaller pumiceous pyroclastic flows also formed on the outer flanks of the volcano from partial collapse of the eruption column. The surface of pyroclastic flow deposits is commonly armored with cobbles and boulders of pumice and lithic clasts, but the main deposits are predominatly noncohesive gravelly sand (Kuntz and others, 1981). Lahars were generated in three major ways, during the May 18, 1980 eruption (Janda and others, 1981). Selected attributes of eruption-caused lahar at Mount St. Helens are compared in Table 2. Lahars with initial velocities in excess of 30ms-1 developed from channelized parts of the initial pyroclastic surge in tributaries to Smith Creek, Muddy River, Pine Creek, Swift Creek, Kalama River, and South Fork Toutle River (Fig. 1). The transition from gas-mobilized pyroclastic surge to water-mobilized debris flow occurred during 4 to 5km of transport on the lower flanks of the mountain. Peak discharges of these lahars attenuated extremely rapidly away from the mountain(Pierson, 1983; Fairchild and Wigmosta, 1983). Lahars in the sector bounded by Smith Creek and Swift Creek (Fig. 1) remained as viscous slurries capable of transporting high density cobbles and boulders until they flowed into Swift Reservoir and deposited 1.4x107m3 of sediment and water. Enroute these lahars sheared off and abraded riprarian vegetation and 14 新砂防, Vol. 37, No. 2 (133) 昭59. 7. deposited a cohesive coating of clast-studded mud on the trees and valley walls. The flow through the lower 19 km of the Muddy River was still a fully developed debris flow even though the channel gradient was only 0.013. Velocities did, however, slow to 3 to 6ms-1, and the deposits became progressively finer grained and better sorted (Pierson, 1983). In contrast, the explosively generated coarse-grained lahars on the southwestern sector of the volcano between Swift Creek and South Fork Toutle River (Fig. 1) moved only 2.4 to 5.3km from the crater rim before stopping abruptly on gradients of 0.085 to 0.125. Highly sediment laden streamflow which developed during the recession of some of these lahars caused severe erosion and deposition for several kilometers downstream of the lahar termini. The explosively generated lahar in the South Fork Toutle River was particulary interesting rheologically in that the lahar was transformed downstream from a fully developed debris flow into hyperconcentrated streamflow through addition of water and deposition of sediment. The lahar generated by the breakout of a short-lived crater lake on March 19, 1982 underwent a similar transformation (Waitt and others, 1983; Pierson and Scott, in press). Detailed stratigraphic studies at Mount St. Helens and Mount Hood indibate that downstream lahar transformation is a common process. The most voluminous and destructive May 18, 1980 lahar was generated not on the steep vflanks of the volcano, but from slumping and flowing of water-saturated parts of the massiv North Fork Toutle debris avalanche deposit, aided in part by seismically-induced liquifaction (Janda and others, 1981; Fairchild and Dunne, written communication, 1982). The total volume of this lahar at the distal end of the avalanche deposit was about 1.4x108m3 (Fairchild and Wigmosta, 1983), or more than ten times larger than either the lahars that entered Swift Reser- voir or the South Fork Toutle lahar. Gradients in the source area were 0.05-0.08; such gradients are typically associated with debris flow deposition rather than initiation. The North Fork Toutle River lahar was characterized by more sustained flow and less rapid downstream attenuation of peak discharge than the South Fork Toutle River lahar. The peak flow of this lahar traveled 120km to the Columbia River as a fully developed mudflow capable of supporting pebble-and cobble-sized clasts of high density rock. Particle size analyses of peak flow deposits indicate that this flow, like that along the lower Muddy River, became finer grained and better sorted in a downstream direction (Kevin Scott, written communication, 1982), but because of its huge bulk the peak flow was not transformed to hyperconcentrated streamflow. The recessional flow did become progressively more dilute with time, particularly in the lower Cowlitz River (Janda and others, 1981). Pumiceous pyroclastic flows erupted during the afternoon of May 18, 1980 generated additional lahars by melting debris-laden ice and snow in the headwaters of Smith Creek, Muddy River, and South Fork Toutle Rivers. These lahars were smaller and less extensive than those generated by the initial pyroclastic surge in the same basins. Nonetheless, melt-generated lahars would have caused considerable damage had they not been preceded by larger lahars. Except for the fact that the melt-generated lahars were particularly rich in pumice, the deposits of all three types of lahars were remarkably similar. Compared with typical Japanese mud and debris flows (Ikeya, 1981, table 1; Mizuyama, 1981, Fig. 1), the 1980 Mount St. Helens lahars are noteworthy because of their long runout and low gradients of transport. The length-descent ratios for these lahars are large compared with historic lahars from other volcanoes, but fall within the range of prehistoric lahars at Mount St. Helens and at comparable volcanoes (Neall, 1975). Following its cataclysmic May 18, 1980 eruption, Mount St. Helens has had an additional 15 eruptions (Swanson and others, 1983 A). The most recent eruption started on February 2, 1982 and led to a year-long period of persistent dome growth, punctuated by occassional small gas and ash emissions. Only the five eruptions occurring between May 25 and October 16, 1980 were explosive enough to generate pyroclastic flows or significant tephra fall beyond the previously devestated area (Sarna-Wojcicki and others, 1981). These explosive eruptions occurred SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 15 when the crater and flanks of the volcano were free of snow and as a reault no major lahars were generated. The total volume of airfall produced by these eruptions was only 0.063km3 (Sarna-Wojcicki and others, 1981); this was sufficient to cause some transitory increases in suspended-sediment concentration or affected streams but insufficient to cause lingering hydrologic impacts. All ten eruptions since October 16, 1980 involved predominantly passive dome building. However, relatively small explosive events associated with dome extrusion at times of heavy snow accumulations in the crater melted sufficient snow to form lahars on March 19 and April 4-5, 1982 as well as February 2-3, 1983. The March 19, 1982 lahar was sufficiently large to breach and severely erode a major debris retention structure (DRS-Nl in Fig. 1) located 35 km downstream of the crater (Waitt and others, 1983); downstream from this structure, this lahar quickly evolved into a hyperconcentrated flow in a manner similar to that of the May 18, 1980 South Fork Toutle lahar (Waitt and others, 1983; Pierson and Scott, in press). The April 4-5, 1982 and February 2-3, 1983 lahars flowed only 2-4 km beyond the base of the cone.

HYDROLOGIC IMPACTS OF THE INITIAL ERUPTION

The 1980 eruption not only mantled the hillslopes and stream channels surrounding Mount St. Helens with more than 3km3 of readily erodible sediment, it also modified all of the land- based phases of the hydrologic cycle so that stream discharges capable of significant erosion and transport now occur more frequently than prior to the eruption. The post-eruption 6-hour unit hydrograph for the lower Toutle River during the 1980-1981 winter had a peak discharge of 615 m3s-1, 1.75 times larger than prior to the eruption. Additionally, the time to peak was reduced from 15 hours of 9 hours (Orwig and Matheson, 1981). Likewise, early modeling results sug- gested that a mean daily discharge in excess of 570m3s-1 which formerly occurred only once in 5 years, may now occur as often as twice in 5 years (Lettenmeir and Burgess, 1981). These early predictions have been borne out by actual stream gaging in the three years since the May 18, 1980 eruption. By pre-eruption standards the heavily affected Toutle River (TOTW THIW in table 1 and Fig. 1) had, during this period, four flood peak discharges larger than those previously occurring only once in 5 years; three of those peak discharges were larger than those previously occurring only once in 25 years (J. R. Williams, written communication, U. S. Geological Survey, 1983). The Cispus River (CISW in Table 1 and Fig. 1) which received heavy airf all but was otherwise unaffected by the eruption had two flood peaks that exceeded those previously occurring once in 10 years. In contrast, streams in this same areathat were either unaffected by the 1980 eruptions or received only minor airf all (TILW, KECW, SPEW, and HSSW in Table 1 and Fig. 1) experienced, with one exception, only flood peaks less than those occurring once in 5 years. The exception to this generalization is the January24, 1982 peak which exceded the 10-year recurrence flood peak even on some unaffected streams. The eruption-induced increases in storm runoff and peak flood discharge reflect at least three factors-1) decreased infiltration, 2) reduced hydraulic roughness along lahar-affected channels, 3) bulking of flows by increased suspended-sediment concentration. Infiltration immediately following the 1980 eruption was reduced to the point that for modeling purposes Lettenmeir and Burgess (1981) could reasonably assume that the area affected by the directed blast and thick airf all tephra was essentionally impervious. In an area about 20km northwest of Mount St. Helens the original silty crust had infiltration capacities of only 1 to 4 mm hr-1 in August 1980 (Herkelrath and Leavesley, 1981); that rate is about 10 times slower than that of the surficial black ash at Mount Sakurajima (Shimokawa and Taniguchi, 1983). Infiltration capacities at Mount St. Helens increased naturally during 1980 and 1981 because of 1) sheet erosion of surficial silty ash, 2) rill and gully erosion into pre-eruption soils, 3) surface disruption by sprouting vegetation, and 4) mixing of soil layers by freeze-thaw action. Infiltration capacities in August 1981, were 7 to 9mm hr-1 in undisturbed areas 20km northwest of the volcano (George Leavesley, written communication, 1981), and about 13mm hr-1 in undisturbed 16 新砂防, Vol. 37, No. 2 (133) 昭59. 7. SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 17 18 新砂防, Vol. 37, No. 2 (133) 昭59. 7. areas 15km northeast of the volcano (Swanson and others, 1983 B). In both areas widespread salvage logging and scarification caused at least a 2-fold increase in infiltration (Swanson and others, 1983 b). However, these rates remain at least 2 times slower than infilration capacities for undisturbed forest soils prior to the eruption. Pre-eruption capacities typically were in excess of 50mmhr-1, with values in excess of 100mm hr-1 being fairly common (George Leavesley, written communication, 1981; Johnson and Beschta, 1981). Therefore, reduced infiltration continues to contribute to persistent flashy runoff in the Mount St. Helens impact area. A second factor contributing to increased flashiness of runoff at Mount St. Helens is decreased hydraulic roughness of channels affected by major lahars. Prior to the 1980 eruption these streams were characterized by dense riparian forest, pool-and-riffle configuration, cobble and boulder bed material, and occasional large accumulations of coarse woody debris. The lahars reamed out the channels, shoving most of the woody debris and riparian vegetation to the edges of the inundated area. Recessional flows then replaced the pre-eruption course-grained armored layer with more sandy bed material. These changes were most pronounced along the Toutle River system where Orwig and Matheson (1981) thought that they were the dominant reason for the increased flashiness of runoff because of the limited extent of the blast- affected area. How- ever, the roughness of channels within the blast-affected area that were not reamed out by lahars (for example, upper Green River and Clearwater Creek, Fig. 1) may actually have been increased by accumulation of downed timber. Dunne and Leopold (1981) also pointed out that high suspended-sediment concentration (around 100,000mg L-1) also caused the streamflow to be hy- draulically smooth. Research is underway at the Cascades Volcano Observatory to quantify this affect. The third factor contributing to higher flood peak discharges is bulking of flows by high concentrations of suspended sediment. Suspended-sediment concentration did significantly increase, peak discharges associated with 1) the August 27, 1980 breakout of Elk Rock lake, 2) the first significant runoff events along the North Fork Toutle and Muddy Rivers following the 1980 eruption (September 1980 through February 1981) and (3) the February 20, 1982 flood along the North Fork and main stem Toutle River (Dinehart and others, 1981; Table 1). The initially high suspended-sediment concentrations reflect the flushing of sediment from the most unstable temporary storage sites created by the eruption. The flood of February 20, 1982 was associated with the breaching of Jackson Creek lake. Thus, the high suspended-sediment concentration during that flood may reflect particularly vigorous channel erosion on the North Fork Toutle debris avalanche deposit similar to that accompanying the August 27, 1980 breach of Elk Rock lake and the March 19, 1982 lahar-causing crater lake breach. However, sediment bulking usually plays a minor roll in increasing peak flow. For example, the concentration of suspended-sediment at the North Fork Toutle River at Kid Valley (KIDW in Table 1 and Fig. 1), which consistently displays the highest concentration of any Mount St. Helens-affected stream gaging station, exceded 130,000mg L-1 (about 5% solids by volume) only 1% of the time during the 1982 water year; concentrations exceeding 4,800mg L-1 (about 0.2% solids by volume) only 50% of the time during that same interval. Another example is provided by the upper Green River and Shultz Creek which during No- vember 1980 through February 1981 had suspended-sediment concentrations well below 100,000mg even though the blast-affected hilislopes were undergoing vigorous sheet and rill erosion (Lehre and others, 1983). Reduced interception and transpiration associated with the destruction of 550km2 of forest vegetation apparently did not significantly change basin losses (U. S. Army Corps of Engineers, 1983). The impact of the eruption was to change the time distribution of storm runoff, not the total volume of runoff. This apparent constancy of runoff volume reflects in part the fact that precipitation falls primarily as persistent low intensity rainfall during the time of year when SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 19

Table 3. Comparison between sediment yields for Mount St. Helens-affected streams and selected sediment yields from other rapidly eroding areas. Suspended Total Drainage Period Sediment Sediment Area of Yield Yield River (km2) Record (103Mgkm-2yr-1) (103m3km-2yr-1) Reference 20 新砂防, Vol. 37, No. 2 (133) 昭59. 7.

1/ Preliminary data 2/ Estimated assuming 2x106Mgyr-1 contributed by Green River. 3/ Computed assuming 10 percent bedload and 0.67m3/Mg. evapo-transpiration is minimal.

SEDIMENT YIELD

Although present concentrations of suspended sediment are insufficient to add significant bulk to typical flood flows in the Mount St. Helens area, it Toes mark a dramatic departure from pre-eruption conditions. Concentrations are sufficiently high for them, when combined with the high water discharge, to cause suspended-sediment discharges per square kilometer of drainage area that are among the highest values in the world for streams of comparable cite (Table 3). In contrast, measured sediment discharges from comparable drainage basins elsewhere in the northwestern United States suggest that average annual pre-eruption suspended-sediment yields from the major streams draining Mount St. Helens probably 200-600Mgkm-2, or about 2 orders of magnitude lower than post-eruption values. Comparisons of a limited number of pre eruption suspended-sediment concentration samples collected near North Fork Toutle River at Kid Valley (KIDW in Fig. 1) (Collings and Hill, 1973), at Toutle River at Highway 99 (THIW in Fig, 1), and at Cowlitz River at Kelso (KELW) (Dinehart and others, 1981) with post-eruption samples from the same sites (Dinehart and others, 1981) suggest that for any given water discharge between 1981 and 1983 these streams typically transported at least 100 time more SEDIMENTATION AND GEOMORPHIC CHANGES OF MT. ST. HELENS 21

suspended sediment than prior to the eruption. The sediment yield from the 220km2 of the North Fork Toutle River basin (exclussive of the Spirit, Coldwater, and South Castle Lake basins(upstream from Camp Baker (Fig. 1) approaches well documented yields from small rapidly eroding areas that generate debris flows (Table 3). Estimates of the volume of sediment eroded from the debris avalanche deposit vary widely. However, all researchers do agree that the 60 km2 covered by the debris avalanche deposit is the dominant sediment source for the Toutle River. Sediment yields from that area may well exceed those from the debris flow-generating areas. In that regard, it is interesting to note that, except for the May 18, 1980 and March 19, 1982 lahar deposits, mud and debris flow deposits on the North Fork Toutle debris avalanche deposit are uncommon; alluvium is typically stratified and reasonably well sorted. The extremely high sediment yields are the product of frequent and sustained high water discharge, as well as availability of readily eroded sediment. Suspended-sediment concentrations are high but not exceptionally high when compared with concentrations in other high sediment yield streams (Milliman and Meade, 1983). Systematic measurements and calculations of bed load discharge have just started for Mount St. Helens-affected atreams. Bedload measurements and calculations for other high sediment yield streams in the northwestern United States, along with initial calculations using the Meyer- Peter, Meyer-Peter and Muller, Einstein, Colby, and Bagnold procedures suggest that at most Mount St. Helens gaging stations bedload probably accounts for 5 to 10 percent of the total annual sediment discharge, with 25 percent being a plausible upper limit (Janda and Nolan, 1979; Dunne and Leopold, 1981; U. S. Army Corps of Engineers, 1982; Lehre and others, 1983; unpublished U. S. Geological Survey). The amount of bedload therefore appears comparable to the uncertainty inherent in the measured annual suspended-sediment discharge. Nonetheless, the bed load is critical in determining channel geometry and flooding hazards, so research is underway to develop methods to measure bedload transport directly and to try additional calculations, in- cluding formulae that have been used successfully in similar environments in Japan (Mizuyama. 1977). In comparing annual sediment yields (Table 3), the relatively low sediment yields during water year 1981 for KIDW and THIW are stiking. Active sediment control measures in effect during water year 1981 contributed significantly to the relatively low yield. The North Fork Toutle debris retention structure (DRS-NI in Fig. 1) trapped 8.4x106m3 of sediment and an additional 6.4x106m3 was dredged from sediment stabilization basins along the lower North Fork and main stem Toutle River during water year 1981 (Stockton, 1981). Sediment control was pursued much less vigorously in the Toutle basin in subsequent years. Sediment yields for water years 1982 and 1983 partly reflect erosion of more than 2x106m3 of sediment original trapped in DRS-NI or stored in unrevetted spoil piles on flood plains. Although disturbed geomorphic systems commonly display rapid recovery, the high measured yields should not be dismissed as a short-term aberation. Precipitation and stream gages around the volcano indicate that annual rainfall amounts have been normal or slightly above normal, and that major streams have experienced long periods of sustained high flow. However, the devastated area has not been subjected to a major regional storm or flood since the 1980 eruption (Table 1). Since 1980, all daily rainfall amounts at Glenoma 1 W and Cougar 6 E (G1 and Cg in Fig. 1), the two continuing long-term precipitation stations closest to the volcano, have been less than the 24-hour rainfall that is typically equaled or exceded only once in two years (National Oceanographic and Atmospheric Administration, 1980, 1981, 1982, 1983). In the case of the Toutle and Cowlitz Rivers, net erosion of the North Fork Toutle debris avalanche deposit has been less than 5 percent of its total volume. Moreover, over the next several decades, the dominant erosion processes operating on the debris avalanche deposit are unlikely to be significantly slowed by encroaching vegetation. Clasts 150mm in diameter are routinely transported by streams on the avalanche, and larger casts are not sufficiently abundant to develop an armor layer.