9568 - Laserfiche WebLink

Home Browse Search

I I I I I I I I I I I I I I I I I I I Both Ferguson (1986) and Miller (1984) provide relatively simple bias correction factors derived from the calibration data to improve rating curves based on log transformed data. Note that Koch and Smillie (1986) commented on Ferguson's paper, pointing out that such bias correction factors are based on the premise that the scatter about the rating curve (the residual values) are normally distributed, which may not always true. Ferguson responded to this comment by identifying four specific assumptions that must be met (including the assumption of normally distributed residuals), any of which can influence the accuracy of the resulting rating curve. Citing Walling and Webb (1981) and Church et al. (1985), Ferguson concludes that the bias correction factor generally provides improvement in accuracy, even though substantial imprecision may remain due to potential violation of other assumptions. In an attempt to improve the rating curves in this report, the bias correction proposed by Ferguson was used when the cahbration data were readily available (Deerlodge Park and Mathers Hole). For the remaining stations the rating curve coefficients reported in the literature were utilized, none of which were bias corrected. For all results, the rating relationships were evaluated for consistency. For example, the rating curve for total load should plot as the maximum enveloping curve of all rating curves. If this was not the case, as happened with the development of bias corrected rating curves for Deerlodge Park and Mathers Hole, the original data was more closely considered and adjustments made in the data set used for analysis. 5.2.2 Ratin~ Curve Results. Mavbell Gage Rating curve information for the Maybell gage was not readily available in the literature. For purposes of this report it was not considered necessary to acquire USGS records and develop rating curves. This task could be completed in subsequent phases of the investigation, as required. Lilv Gage Andrews (1980) derived a rating curve for suspended load at the Lily gage based on measured data, and then derived a total load relation by calculating the bed load with the Meyer-Peter, Muller bedload equation. Given Andrews' suspended and bedload rating curves, and particle size analysis of suspended load data (pers. comm, 1989), rating curves for wash load and suspended sand load were derived. Specifically, particle size analysis of the suspended sediment particle measurements were used to compute a discharge weighted mean of the percent suspended sediment coarser than 0.062 mm (37 percent). Use of the discharge weighted mean accounted for the variation of sand sized material with discharge and was based on 185 samples. A wash load rating curve was then calculated as 63 percent of the measured suspended load rating curve. A suspended sand load rating curve was derived by adding 37 percent of the suspended load rating curve to the results from the bedload rating curve. A total load rating curve was derived by summing results from the suspended load rating curve and bedload rating curves. The resulting rating curve coefficients are given in Table 5.1. Note that bias correction was not applied to these results. 5-5