Pre-eruption observations of bed material in the Toutle River and the upper Lewis River tributaries are rare and mostly anecdotal. A partial list of streambed observations was excerpted from comments on discharge measurements made at the Toutle River near Silver Lake from September 1969 to September 1981 ( table 15 ). Streambeds were described as composed of cobbles until after May 18, 1980, when channel descriptions for the next year usually mentioned sand and silt. The anecdotal comments concur among various individuals. Descriptions of turbid flow and floating trees and woody debris also appear in the pre-eruption storm measurements.
Surficial bed material in the Toutle River consisted of sand and fine gravel through the major storms of 1980. Equipment designed for sampling sandy beds worked well in these streams at low and medium flows. Sampling points in the cross section were located at centroids of equal discharge, which were determined for suspended-sediment sampling. Some sample sets were composited before analysis of particle size. Median particle size was determined for bed-material samples (both individual and composited sets) by interpolation to D50 and was plotted by time (figs. 71-74). In the figures, individual samples of bed material from the same set are connected with a vertical line.
Median particle sizes of bed-material samples from the North Fork Toutle and the mainstem Toutle River range from 0.1 to 100 mm ( fig. 71 ). Median particle sizes of samples from the North Fork Toutle River at Kid Valley are distributed through the sand and gravel ranges. Variability within sample sets from the North Fork Toutle River increased noticeably between 1982 and 1986 (fig. 71). Variability also increased in bed-material samples from the Toutle River at Tower Road, especially when contrasting samples from 1981 with later years. More than half the samples from Toutle River at Tower Road have median particle sizes in the sand range.
Bed material at the Toutle River at Highway 99 was primarily sand and fine gravel through 1983, when sampling operations at that station ceased. Median particle sizes at the Cowlitz River at Castle Rock were mostly in the sand range ( fig. 72 ). An annual cycle of coarsening, coinciding with increasing uniformity in the cross section, is discernible in water years 1982 and 1983.
At the Muddy River, the average D50 of bed material sampled from 1981 to 1989 remained in the range of coarse sands ( fig. 73 ). Although the first few sets of samples were primarily fine and medium sands, other samples ranged from sand to gravel throughout the study period.
Streambeds having a stable, gravelly surface during moderate discharges may be covered with sand following storm flow. At other times, high stream velocities over sandy material in a particular reach can leave a coarsened bed. During periods of extreme sediment discharge, the processes of bedform growth, sand transport, and pavement formation alter bed-material size distributions quickly. Bed-material samples collected during two storm flows (February 1982 and February 1986) are contrasted in figure 74 . Median particle sizes of bed-material samples were all in the sand range following the 1982 storm flow at the Toutle River at Tower Road, when bed-material samples had previously included fine- and medium gravel. No such increase in sandy samples was measured following the storm flow of February 1986 at the North Fork Toutle River at Kid Valley. Instead, two samples of coarse gravel were collected 2 days after bed-material samples were fine gravel or finer.
Only sand and gravel can be sampled representatively by a BM-54 bed-material sampler. Streambeds with sediment larger than gravel are not well represented by the BM-54 bed-material data. The field notes for bed-material samples include many instances where bed material was too coarse for reliable sampling. Although sand and gravel may have been present in patches on the streambed, the small clamshell bucket (10.7 in3) on the BM-54 bed-material sampler would not close in the presence of protuding cobbles. Size distributions of bed material in the gravel-paved streams were poorly documented where streambeds were not penetrable by the BM-54 bed-material sampler. These limitations have been noted in the published tabulations of bed-material data (Dinehart, 1992b).
Pavement formed gradually on the Toutle River streambeds, but was not fully quantified by bed- material samples. Sediment beneath the pavement (subpavement) is usually released and transported during pavement-disrupting flows. During summer low flow in 1987, 1989, and 1990, subpavement material in the stream channel at the North Fork Toutle River at Kid Valley was sampled by excavation in exposed gravel bars. Size distributions of the subpavement samples are plotted on a geometric scale for clarity ( fig. 75 ). The bimodal size distribution of gravel deposits is evident where a minimum amount of coarse sand and fine gravel is shown. In 1990, streambeds at gaging stations on the Muddy River, the Green River, Clearwater Creek, and the North Fork, the South Fork, and the mainstem Toutle Rivers consisted of gravel and cobbles, with occasional boulders, interspersed with poorly sorted sand. Sand was found mainly in the subpavement material and as slackwater deposits during low-flow periods. Cobbles were predominant on exposed channel bars that had sand and fine gravel on their surfaces in the early 1980s.
