WSP05649 - Laserfiche WebLink

Home Browse Search

TOPPING ET AL: COLORADO RIVER SEDIMENT TRANSPORT, 1 concentration arises from how well the spatial distnbution of suspended sediment in a cross section is characterized. An example of this type of error comes from 1957 at the Lees Ferry gage, when only inadequate sampling equipment was available at the gage. During 1957, a D-43 suspended-sediment sampler was the only type of suspended-sediment sampler available for use at the Lees Ferry gage (unpub~shed USGS Lees Ferry gage annual technical file, 1958). Because a D-43 sampler could only sample to a depth of 4.5 m [Guy and No,"",n, 1970], the lower portion of the Colorado River at Lees Ferry could not be sampled properly during higher flows. Thus the measured sediment load at the Lees Ferry gage during the higher portions of the 1957 snowmelt flood was too 'Iow. The USGS corrected this problem during 1958, when the D-43 was replaced by a P-46 sampler, a sampler designed for deeper, higher-velocity flows (unpublished USGS Lees Ferry gage annual technical file, 1958). The third source of error relaIed to determining sediment concentration arises from how well changes in sediment con- centration over time are characterized during periods of rap- idly v8l)'ing flow. During these periods, sediment loads in the intervals between measured concentrations were estimated by fint drawing a curve between the measured concentrations and then multiplying the estimated concentrations along this curve by the discharge during these intervals (see Porterfield [1972] for a detailed description of this procedure). Because typicaJJy only 1 or 2, and sometimes no, samples were collected during large flood events on the Paria and tittle Colorado Rivers, this source of error may easily have been as high as 100% during individual floods on these rivers (Figure 12). This source of error also became significanI on the main stem Colorado River at the Grand Canyon gage during the postdam era when only one sample was collected each day under daily fluctuating flows. Therefore this source of error is probably the largest source of error in the sediment budget for Marble Canyon and upper Grand Canyon. The fourth and fifth sources of error related to determining sediment concentration arise from changes in personnel and laboratory analyses. Because of the complexity of measuring sediment loads in a river, significant errors or changes in the magnitude of error can be introduced into sediment-load data when there is a change in the personnel making the measure- ments [e.g., Allen and Petersen, 1981]. Allen and Petersen found that a difference in sediment load of 30% was possible when the load was calculated using samples collected by per- sonnel with different levels of experience. Also, a signilicant error or change in the magnitude of error can be introduced when there is a change in the personnel making the computa- tions. For example, in computing loads, one person might assume that the maximum sediment concentration occurs at the same time as the flood peak, and another might assume that it occurs at a different time (figure 12). The other place where errors can be introduced into sediment-load data is in the laboratory where the samples are processed, but this is typically the smallest of all of the sources of error. As indicated by the above summary of the sources of mea- surement error, uncertainties in the calculated sediment load on any individual day can be quite large but are difficult to quantify. On the main stem Colorado River they were probably as large as 10-30%, and on the Paria and Little Colorado Rivers, they could easily have been as high as 50-]00%. Be- cause the uncertainty in the mean of a time series is typically much smaller than the uncertainties associated with the indi- 541 vidual measurements (when the various sources of error are uncorrelated), the uncertainties associated with sediment loads over monthly or annual timescales were probably smaller. However, because the error associated with measurements of sediment concentration are aU much larger than the error associated with the computation of Ihe discharge of water, the uncertainties in the sediment loads still had to be much larger than those associated with the computation of the discharge of water. Therefore, perhaps the best way to estimate the minimal probable uncertainty in monthly or annual sediment loads is to multiply the uncertainty in the discharge of water by about a factor of 2. This approach yields minimal uncertainties in the monthly or annual sediment loads in the Colorado, Pari.. and tittle Colorado Rivers of about 5%, 20%, and 10%, respec- tively. However, because of the different locations at which sediment concentrations and the discharge of water were mea- sured, the errors thought by USGS personnel to exist in the tittle Colorado River sediment records ranged from slightly less than 25% to as much as 50%. Thus a reasonable minimal uncertainty in the monthly or annual sediment loads in the tittle Colorado River is also 20%. Therefore, in this paper, uncertainties of 5% are assigned to the measured sediment loads of the Colorado River, and uncertainties of 20% are assigned to the measured sediment loads of the Paria and tittle Colorado Rivers. Admowledple.... This resean:h was funded by the Grand Canyon Monitoring and Research Center. Randy Parker, Paul Kinzel, Ingrid Corson, and Lisa Dierauf helped compile and process the historical data from the Lees Ferry and Graod Canyon gages. Ted Melis, Peter Griffiths, George Tate, Bob Bohannon, Lars Neimi, and Steve Bledsoe helped collect the samples from the predam Oood deposits. Conver- satiODS with Jack Schmidt, Jon Nelson, Steve Wiele, Randy Parker, Peter Wilcock, and Paul Grams helped improve the quality of the science. Jack Schmid~ Jon Nelson, Jim Bennet~ Bill Dietrich, Alan Howard. and Peter Whiting provided thorough reviews of earlier ver- sions of this manuscript. References Allen, P. 8., and D. V. Petersen, A study of the variability of suspended sediment measurements, in Erosion and Sedimml Transport MeIJ- suremenls, Proceedings of lhe Florence Symposium, June 1981, fAHS Publ., 133. 203-211, 1981. Andrews, E. D., The Colorado River: A perspective from Lees Ferry, Arizona, in The Geology of NOrTh America, vol. 0-1, Surface Water Hydrology, edited by M. G. Wolman and H. C. Riggs, pp. 28]-328. Geo!. Soc. of Am., Boulder, Colo., 1990. Andrews, E. D., Sediment transport in the Colorado River basm, in Colorado River Ecology and Dam Management, edited by Comm. on Geosci., Environ., and Resour., pp. 54-74, Washington, D. C, Natl. Acad.. 1991. Andrews, E. D., C. E. Johnson, J. C Schmidt, and M. Gonzales, Topographic evolution of sand bars, in ~ 1996 Controlled Flood in Grand Canyon, Geophys. Monogr. Ser., vol. 110, edited by R. H. Webh et al.. pp. 117-130, AGU. Washington. D. c., 1999. Anima, R. 1., M. S. Marlow, D. M. Rubin, and D. 1. Hogg, Comparison of sand distribution between April 1994 and June 1996 along six reaches of the Colorado River in Grand Canyon, Arizona, u.s. Geol. Surv. Open File Rep., 98-141, 33 pp., 1998. 8eus, S. S., S. W. Carothers, and C. C. Avery, Topographic changes in fluvial terrace deposits used as campsite beaches along the Colorado River in Grand Canyon, J. Ariz.. N~. Acad. Sci., 20, 111-120, 1985. Brooks, N. H., Mechanics of streams with movable beds of fine sand, Trans. Am. Soc. Civ. Eng., 123, 526-594, 1958. Burkham, D_ E., Trends in selected hydraulic variables for the Colo- r;ldn River at Lees Ferry and near Grand Canyon, Arizona-1922- 11):'4, report. 58 pp.. Bur. of Reclam., Glen Canyon Environ. Stud.,