7910 - Laserfiche WebLink

Home Browse Search

C. The range of extrapolation was limited to a minimum flow ?- of 0.77 times the minimum measured flow point, to a maximum flow of 1.30 times the maximum measured flow. 3. If three or more points were used to define the rating curve, the range of extrapolations was limited to a minimum flow which was 0.4 times the minimum flow used to define the curve, to a maximum flow of 2.5 times the maximum flow used in the rating curve. Oata which have been subjected to the preceding-selection and limitation process will henceforth be termed ref' data. The error associated with the number of points used to establish a stage-discharge relationship is shown graphically in Figures 12 and 13 for Oak Creek in Oregon, and the Yakima River near Umtanum, Washington. Figure 12 shows the mean error of the stage-discharge predictions, using all combinations and permutations of data points. Figure 13 shows the errors associated with the number of points used to establish the stage- discharge relationship, using only refined data as defined previously. Tables 1 and 2, and Figures 12 and 13 show some interesting trends. When no limitations are placed on the data used for calibration, nor on the extent of extrapolations, the use of a two-point rating curve may in fact be no better than using Manning's equation. However, the use of a three-point rating curve almost always produces more reliable results than either the two-point system or Manning's equation. Figures 12 and 13 show that at some number of points on the rating curve, there is little improvement in the accuracy of the predictions by adding more points. If unlimited or unbounded data, such as that used in Table 1, is used, this limit is approached with three or four stage-discharge- points. If the suggested refinements are made when selecting data points and extrapolating the data, the limit of reliability is approached with only two data points. However, the useful range of extrapolated discharges is smaller when two points are used, so in most cases it may be desirable to use three, even if the reliability may not improve significantly. Velocity Predictions The methods presented for calculating the velocity distribution in a stream are based on equations which have been accepted by hydraulic engineers for calculating the mean velocity of the channel. These con- cepts have not been generally applied when the channel is subdivided into a series of channel segments. These methods have been used before by others, but the discussion of errors resulting from the use of a par- ticular approach has been limited. Elser (1976) conducted a.brief field test of the Manning equation for predicting channel segment velocities in a large prairie river. Velocity measurements were made for five channel segments, at five 3S