Considerable sources of uncertainty are involved with hydrograph production from stage measurements and stage-discharge rating curves. In this study, efforts were made to minimize uncertainty due to systematic errors by maintenance and calibration of the current meter, consistent procedures and performance of all discharge measurements by the same person. The magnitude of uncertainty due systematic errors is unknown.
In contrast, several sources of random errors were better known, and reduced by the relatively large number of actual discharge measurements. Examples of random error include those due to channel bed and water surface roughness (e.g. cobbles and waves), variation in the number and position of discharge measurement verticals, fluctuations in current velocity and the use of rating curves. Only one set of two replicate discharge measurements were made - at low discharge - which varied by 2.6 percent. The standard errors, or standard deviations of the means, for three years of rating curves determined by regression ranged from 0.004 to 0.04 m^3 s^-1. Although the sum of systematic and random errors cannot be determined, the overall uncertainty of hourly discharge values is considered acceptable for the purposes of this study. The relative changes in discharge are of greater interest than the accuracy of the measurements.
Four possible sources of uncertainty were identified in the measurement of stream water temperature (see section 3a for equipment details). These were: (1) measurement error; (2) solar loading; (3) error due to reduced circulation at the point of measurement; and (4) further heating of the water between the point of measurement and the lake (350 m in 1991 and 1992).
SSC measurement uncertainty was contributed by each step in the sampling and processing sequence. Uncertainty was due to both random and systematic sources, some of which can only be estimated. Nonetheless, it is essential to determine the combined uncertainties, particularly because methodological changes occurred from year to year.
Total SSC uncertainties are largely the result of error propagation. Uncertainties were determined for net sediment weight and sample volume (quantities measured directly), and then for the range of calculated SSC values.. Net sediment weight uncertainty takes into account electronic balance nonlinearity and lack of precision, along with filter preweight variability (1991) and loss of extractables. The random and systematic components of uncertainty were not rigorously distinguished in the uncertainty calculations; most sources were taken to be independent and random, and hence combined in quadrature (following Taylor, 1982).
In 1990, inadequate filter preweight precision overwhelmed all other contributions to uncertainty, resulting in absolute SSC uncertainty up to +/- 25 mg L^-1. The SSC values calculated, however, indicate that the actual uncertainty may be significantly less; for example, during low discharge intervals, calculated SSC values are between 0 and 10 mg L^-1, as during subsequent years.
Uncertainty in the 1991 calculations was primarily due to calibration problems in the balance used. Absolute SSC uncertainty was +/- 3.2 to 3.4 mg L^-1. Estimating a mean weight for filters, rather than accurately preweighing them also contributed uncertainty, although both filter types were relatively consistent in weight. 1991 uncertainty was reduced by virtue of greater accuracy in the sample volume measurement.
Uncertainty of SSC determinations was reduced in 1992 by preweighing filters, and absolute uncertainty was calculated to be +/- 1.9 to 4.0 mg L^-1. Use of the miscalibrated balance for these weights remained the greatest source of SSC uncertainty up to approximately 270 mg L^-1, when volume measurement error assumed a greater proportion.