Hydrographs of hourly discharge were produced for each field season. Considerable effort was spent, both in the field and in data processing, to create accurate hydrographs. The discharge values were determined by two methods:
Discharge values for the 1990 period from 1000 h June 9 to 1200 h June 12 (DY 163) were determined from a plot of measured discharges (n = 11). Intermediate values were interpolated, and estimated based on field notes regarding relative changes in stage and channel geometry.
The composite stage record for the balance of the 1990 field season was assembled from manual observations that made up the standardized gage heights, interpolations between the readings, and estimations for intervals without observations. The standardized heights were first plotted for 6 day periods. Linear interpolation was then used to fill in hourly values where the stage trend was apparent, while other periods required estimation based upon field notes and/or continuation of trends. The estimated values were in all cases conservative, in the sense that values were not arbitrarily extrapolated beyond constraining observations. (GRAPH: example of this procedure in 1991)
To produce rating curves for 1990, the measured discharge values (n = 36) were each numbered on a plot of standardized height against discharge. On this plot, three different curves were suggested, demonstrating a shift to the left during the field season. Rating curves were then fit to these points by least squares regression (r^2 = 0.98 to 1.00).
During the time periods to which rating curves were fit to standardized gage heights, only the last curve required extrapolation beyond the measured values of stage and discharge. With this curve, however, the highest standardized gage height was 2.25 and 5.0 cm higher than the stages of the two highest measured discharges. To extend the curve for this period higher, an additional, artificial stage-discharge point was plotted, which produced a regression line parallel to the other curves. The last time period was defined by two equations, one for high stage and one for low stage, again to obtain a better fit by the rating curve. When plotted, these two curves do not appear distinct.
The regression equations defining each rating curve were then applied to the standardized gage heights, producing predicted discharge values. The resulting hydrograph, when plotted with actual measured discharges at the average time each was made, demonstrated that the computed hydrograph closely reflected the measured data.
In 1991, hourly discharge for the period 1600 h June 4 to 1500 h June 9 was determined from a plot of 4 to 6 measured discharges each day, along with manual height observations and field notes on changes in channel geometry. Because the snow constrained channel was changing rapidly during this period, height observations were comparable for only short intervals.
Discharge during the period 1600 h June 9 to 1100 h June 18 was determined from direct measurements, and short period rating curves using both manual and recorded gage heights. Discharge was measured 4 to 6 times each day (except 3 times on June 12). On all days except June 10 and 12, a discharge measurement was made close to the daily peak - according to the water-stage recorder - which provides confidence in the predicted values through this period.
Residual snow and ice constrained the left edge of the stream until June 18, when the channel became relatively stable in cross-sectional area. Considerable aggradation of the channel bed also occurred following the peak on June 16. The composite stage record therefore begins in 1991 at 1200 h June 18. Heights were taken from the water-stage recorder, manual observations, and interpolations. (GRAPH: stage height compilation, 1991)
Discharge measurements made after 1130 h June 18 determined the single rating curve used for the 1991 field season (r^2 = 1.00). This curve was applied without extrapolation to the hourly composite stage record from 1200 h June 18 to 0800 h July 12.
The only period of the 1992 field season where the hydrograph was not derived from rating curves was between 1330 h on June 22 and 2300 h June 25. There were 10 discharge measurements made during this period, all at less than 0.075 m^3 s^-1. Hourly values of discharge during this period were interpolated and estimated from these measurements, as well as subjective observations of stage.
The composite hourly heights for the 1992 field season were compiled from the barrel gage (for stage above 24.7 cm), and the tube gage (see section 3). The record spans the entire field period after 2300 h June 25, and agrees well with the manual record, excepting one short period discussed below.
The height record produced by the tube gage, before the other gage was installed, appears to have been adversely influenced by hydraulic effects at water levels above about 27 cm. As a result, for periods between June 26 and July 2 when stage height was above 27-28 cm, the recorded heights could not be used directly. Instead, discharge measurements made at high gage heights during the next period were used to predict the correct gage height at the time of 10 discharge measurements made between June 26 and July 2, by third order regression. These new heights were plotted on the recorder chart, and adjustments were made to the height record above 27-28 cm. Several considerations justified this procedure: (1) the recorded stage heights greater than about 28 cm were highly suspect in relation to discharge values, as also noted in the field; (2) the discharge measurements for which heights were predicted were not used subsequently to develop rating curves; and (3) discharge was measured near each daily maximum during the period, so extrapolations beyond lower (recorded) and upper (predicted) height values were not done.
Rating curves for 1992 were again determined from a plot of standardized gage heights against discharge, from 63 discharge measurements. (GRAPH: 1992 rating curve) These measurements exhibited a small amount of scatter relative to those of the two previous years. Nonetheless, five stage-discharge curves were discriminated, made up of: (A) values recorded by the barrel gage, when stage ranged from 24-38 cm; (B) values following the peak flow of July 25, which resulted in a curve shift to the right; (C) all values when stage was less than 22 cm; (D) early values from the tube gage, with heights between 19 and 27 cm; and (E) a subset of values used for curve A, for application to early tube gage heights greater than 27 cm. Proxy values were used for the latter curve due to the recording problem identified above.
The five rating curves were applied to stage heights close to the range of heights from which they were developed. The greatest extrapolation was by the earliest curve, only 1 cm higher than the highest stage of measured discharges. The resultant hydrograph agrees well with measured discharge values when plotted together.
The concentration of suspended sediment (SSC) in mg L-1 was determined for each sample as:
where the sediment dry mass was obtained as detailed below, and the sample volume was obtained by filtrate measurement in the field. Post-field processing of the filtered sediments and filters occurred in the University of Massachusetts Quaternary Lab. This section considers the procedures and assumptions associated with determining the dry mass of sediment for each sample. The uncertainty in each term of the equation above is also discussed.
The upper size limit of suspended sediment collected was taken as 2 mm. Particles in transit coarser than 2 mm are usually bedload (Church and Gilbert, 1975), in contrast to those smaller than 63 µ m (0.063 mm) which are suspended, and often termed wash load. Although some studies use only particles finer than 63 mm in the determination of SSC, 2 mm has commonly been used by studies working with high energy glaciofluvial systems (e.g. Fenn and Gomez, 1989).
The choice of 2 mm as the upper size limit for suspended sediments was supported by a size distribution analysis of material collected from the stream channel. Among three samples from the channel edge (i.e. freshly deposited sediments exposed at low water level), the mean size ranged from 0.29 to 0.5 mm, and 94 to 97 percent of the distributions were coarser than 2 mm. Evidently, at discharge levels prior to the collection of these samples, considerable material up to 2 mm was in transit.
Processing of filtered sediments in the lab began by separating particles larger than 2 mm from each sample. This was done by manual inspection, and testing large grain passage through a 2 mm mesh sieve. A small proportion of suspended sediment samples collected in 1990 and 1992 contained particles larger than 2 mm (3.8 and 6.0 percent), several of which were 5 mm, measured on the long (a) axis. During the 1991 field period, only one sample contained a particle larger than 2 mm.
When the fraction of each sample larger than 2 mm was separated, filtered samples were oven dried in preparation for weighing. The greater than 2 mm fraction was dried and weighed separately each year. In 1990 and 1991, samples were oven dried at 50° C for 15 minutes, and left to equilibrate for 4 hours at ambient room temperature and humidity (20° C, 19 to 24 percent). The 1992 filtered sediments and filters dried at 50° C for 30 minutes, and equilibrated at ambient temperature and humidity for 60 minutes (23° C, 24 percent). Equilibration after drying was done under ambient, low humidity conditions, rather than in a desiccator. This maintained more uniform conditions, and reduced the possibility of spillage, given the large number of samples processed.
The variables relating to suspended sediment sample acquisition, filtering, and processing were entered into a spreadsheet, for calculation of SSC. (TABLE: sample SSC calculation) This method allowed the SSC data to be stratified in multiple ways: by variables of time, discharge, or position in cross section. A sample time-SSC plot over a three day period of demonstrates that suspended sediment concentrations between the measured values can be estimated by linear interpolation. (GRAPH: three-day plot of measured SSC)