Sediment monitoring to support modeling a reservoir sediment flush on a sand-bed river in Northern Nebraska

Abstract

The U.S. Geological Survey (USGS) in cooperation with the U.S. Army Corps of Engineers (USACE), monitored a sediment flush event from Spencer Dam located on the Niobrara River near Spencer, Nebraska, during the fall of 2014. Data collected during the flush was used to validate a one-dimensional sediment transport model developed by the USACE. The USACE surveyed 26 cross sections within the reservoir and as far as 1 kilometer (km) upstream from the reservoir pool to about 10 km downstream from the dam before and after the flushing event to measure erosion and deposition. They also collected surficial sediment samples from sandbars within the reservoir. The USGS assisted USACE in its model validation efforts by collecting sediment data before, during and after the flush using both traditional sampling techniques and a continuous laser-diffraction particle-size analyzer. From the context of longitudinal volumetric change, the model replicated erosion in the upper half of the reservoir within four percent of that observed by survey data and it replicated deposition downstream of the dam within 5 percent. However, the model underpredicted the erosion of the accumulated delta sediments in the reservoir by 43 percent. The timing and magnitude of suspended sediment concentrations produced by the model compared reasonably well to the discrete suspended-sediment sample results. These results indicate cross-sectional survey data and discrete sediment data may be adequate for developing sediment flush models for reservoirs in similar well-sorted sand-bed streams. The USGS installed a continuous particle-size analyzer immediately downstream from the dam. Although the particle-size analyzer was successful in providing a large dataset during the flushing event, based on discrete point samples, it overestimated the amount of fine particles and underrepresented the amount of coarse material. It also required a significant amount of maintenance during the flushing event because of the large sediment load and the rapid bed aggradation. The maintenance issues with the particle-size analyzer along with uncertainty in the correlation to discrete suspended-sediment samples reduced its value for model validation. However, these issues may have been specific to the flushing event at Spencer Dam, which involved a sand-bed dominated stream and a wide channel. It is foreseeable that other sediment flush models developed for different streams with dissimilar sediment gradations may benefit from similar continuous sediment data, but adequate planning and evaluation should be performed.