The legacy from past sampling programs often leaves us with 1) data not available at the scale of relevance (e.g. events), 2) no downstream data to provide information on source and pathways of movement, 3) insufficient information for transport and routing models, and 4) lack of information on errors and natural variability to provide estimates of uncertainty in concentration and load calculations. To answer these questions, better experimental designs for projects are needed and data collection sites from a variety of investigations will have to be integrated in order to provide sufficient gages for analysis of transport data from relatively large basins with complex land-use patterns. At data collection sites ancillary information on the physical characteristics of the sampling site and the reach need to be collected. Sediment transport is strongly dependent on the physical properties of the reach and inadequate information about stream hydraulics can often negate expensive sediment sampling. Repetitive sampling at certain sites is needed to establish spatial (cross-section) and temporal variability. Detailed sampling and the use of surrogate measurements to provide continuous record of sediment characteristics can provide information on errors in samplers, sampling methods, and analysis. The detection of trends in sediment data depends on understanding the variability in the data and the ramifications of changes in samplers or sampling methods.
Past sediment studies have provided a number of daily suspended-sediment stations where data were collected for a variety of objectives. These stations were never part of a network and were never coordinated with respect to data collection or large-scale analysis. These gaging stations do, however, provide a substantial database with long-term, detailed, sediment data at specific locations and it would be worthwhile to provide a wider distribution of this data to scientists. Therefore, these daily stations are being put on a separate database with access from the internet. This database represents 1,593 stations that have an average period of record of 5.3 years. The period of record for individual stations range from one day to 44 years, and 13 percent of the stations have 10 years or more of record. The drainage area for these stations ranges from 0.002 to 1,893,800 sq km and the median drainage area for all stations is 767 sq km.
The database was screened for errors that were easily identified by a computer algorithm. This check on daily entries included: 1) Missing streamflow value, 2) concentration greater than zero when streamflow and sediment discharge are both zero, 3) concentration zero when streamflow and sediment discharge are greater than zero, 4) only sediment discharge reported, 5) streamflow zero when concentration and sediment discharge are greater than zero, and 6) computed daily loads are in excess of ± 30 percent of the reported sediment discharge in the database. Although some of the entries reported under this algorithm may not be in error, this screening provides some indication of problems and provides a flag on sites in the database that may need further review. This screening procedure identified problems with 3.5 percent of all days of record. Missing streamflow represents about one third of the days with errors which suggests that much of this error can be eliminated.
Environmental concerns often requires establishing a link between the upstream watershed and the sediment station. A result of this link is the search for time trends in the data. The identification of time trends in sediment discharge may signal changes in upstream land use patterns or identify the success or failure of some remedial action. The linkage is: 1) from changes in sediment discharge derived from land use change, 2) to transfer of the sediment discharge downstream (Walling, 1983; Ongley, 1987), 3) to the detection of incremental change in sediment discharge at a downstream station. These linkages from upstream source to a downstream measurement site are not direct and the controls on these linkages are variable in space and time. Parker and Osterkamp (1995) used a stochastic equation to generate a time series of annual suspended-sediment discharges that increased linearly. Using statistics from 24 daily suspended-sediment gaging stations with drainage areas from 1,606 to 1,805,230 sq km, this study determined that trends may not be detected with changes in annual sediment discharge of 20 percent or less. Thus, it may be more constructive to couple gaging station measurements with routine assessments of upstream channels to identify trends.
Changes in samplers, sampling methods and/or laboratory techniques can alter data sufficiently to produce an apparent trend in sediment discharge. One of the most startling examples of a trend produced by changes in data collection was reported by Topping and others (1996) in the Colorado River basin where sediment records indicated a basin-wide decrease in measured sediment loads of about 40-50 percent in the early 1940Ős. A number of papers have attempted to explain this apparent decrease and the trend interpreted as a result of changes in climate, land use and intrinsic tributary geomorphic processes. Unraveling this change in sediment discharge is difficult but Topping and others (1996) suggested that the change was due to changes in samplers, sampling technique, and laboratory methods. This example serves as a warning for situations today where a number of samplers and methods are in use and methods are constantly being modified to allow for some sampling problem.
An example of sampler change is the use at a number of sites of the depth-integrating suspended-sediment sampler model D-77 in place of the model D-74. At these gaging stations, suspended-sediment sampling was routinely done with a D-74. As interest in water chemistry became more prevalent, there was a need to have larger volumes of water for analysis and sampling was done using the D-77. Because of the larger size of the D-77, the distance between the nozzle and the bottom of the sampler is 177 mm whereas the same measure on the D-74 is 103 mm (Federal Interagency Sedimentation Project, 1986). Such a change in the extent of the unsampled zone can have a substantial effect on sampling sand. Using the Rouse equation and assuming a water temperature of 20 degrees C, a flow depth of 3.66 m, and a Mannings n of 0.03, differences between the two samplers is shown to vary by over 10 percent for suspended-sediment sizes of 0.063 mm and by 100 percent for 0.250 mm size suspended material.
Field comparisons between the two samplers have been done at three gaging stations on the Colorado River: near Dotsero, Co, near Cameo, Co, and near the Colorado - Utah state line as part of the studies in the Upper Colorado National Water Quality Assessment. Eight to ten sample comparisons were made at each gaging station and the concentrations from the D-74 averaged 16 percent higher at the Dotsero and Cameo gages and 24 percent higher at the state line gage. These differences are complicated by the fact that the D-77 composite samples were split using a cone splitter to obtain the sediment sample. This process could also introduce error.
Ongley, E.D., 1987, Scale effects in fluvial sediment-associated chemical data. Hydrol. Processes, 1, 171-179.
Osterkamp, W.R., Day, T.J., and Parker, R.S., 1992, A sediment monitoring program for North America: IAHS publ. 210, Erosion and sediment transport monitoring programmes in rivers basins, p. 391-396.
Parker, R. S., and Osterkamp, W.R., 1995, Identifying trends in sediment discharge from alterations in upstream land use: International Association of Hydrological Sciences Publication 226, Effects of scale on interpretation and management of sediment and water quality, p. 207-213.
Topping, D.J., Parker, R.S., Nelson, J.M., and Bennett, J.P., 1996, The apparent mid-20th century sediment load decrease in the Colorado River basin -- an investigation of the mechanics of the Colorado River Sampler: Geological Society of America Abstracts and Program, v. 28, no. 7, p. A-261.
Walling, D.E., 1983, The sediment delivery problem: Jour. of Hydrology, v. 65, p. 209-237.
Contributions from Other Federal Agencies
Contribution from the USGS