Scientific Investigations Report 2007–5164
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2007–5164
Daily mean discharge and turbidity during water years 2002–04 are indicated for all sites in figure 6. With the exception of the site downstream of Cougar Reservoir (CGRO), there typically were only a few periods at any site that resulted in elevated turbidities (>10 FNU, for example) during the study period, from January 2003 through September 2004. Maximum daily mean discharge in the South Fork above Cougar Reservoir (SFCO) was about 4,100 cubic feet per second (ft3/s) each year, whereas maximum discharge downstream of Cougar Reservoir (at CGRO) was constrained by the geometry of the dam’s temporary bypass structure to about 3,000 ft3/s, and generally was less than 2,000 ft3/s. In the mainstem above the South Fork (at MRBO), maximum daily mean discharges were about 8,000 ft3/s, and at VIDA, the most downstream site studied, daily mean discharges were about 10,000–14,000 ft3/s.
Turbidity levels in the basin are characterized by generally low values (< 1-2 FNU) during most periods, punctuated by short spikes of elevated values during storms or other (commonly storm-related) events. Turbidities were highest downstream of Cougar Reservoir during several events; however, background turbidity in Blue River downstream of Blue River Lake during the winter drawdown period typically was elevated over that in the summer or at other sites. Turbidity at VIDA was intermediate to values measured at the regulated upstream stations (CGRO and BLUE) and to those at unregulated upstream stations (MRBO and SFCO), indicating a direct effect on the mainstem from Cougar and Blue River Reservoirs. The turbidity signal at VIDA tended to respond rapidly to inputs from the regulated tributaries, and VIDA frequently had substantially higher turbidities than those measured upstream at MRBO.
Suspended-sediment concentration data from storm samplings are presented in table 2, along with the turbidity values obtained from the continuous monitors corresponding to the time of sampling. Although the largest storms during the study period generally were sampled, there were not as many large storms with elevated turbidity as had been anticipated. Therefore, the number of SSC samples obtained from each station was variable and generally less than originally intended. Monitored turbidity values that were less than 1 and their corresponding SSC data were not included in the regression equations because of the high variability of turbidity in the range of 0–1 FNU. Only the station downstream of Cougar Reservoir on the South Fork McKenzie (CGRO) had more than 20 samples that could be used for a regression.
Suspended-sediment concentrations measured at the different stations were highly variable and depended on the flow and weather conditions at the time of sampling. The highest sampled SSC from CGRO (266 mg/L), during the storm of January 30, 2003, occurred following the collapse of a diversion tunnel and supporting fill material within the Cougar Dam construction zone, and may not have represented typical sediments released from the reservoir during high-turbidity events. The fraction of suspended sediment with diameter less than 0.063 mm was 99 percent, which is higher than for most other samples from that station, perhaps reflecting the character of the fill material used to construct the coffer dam (J. Britton, U.S. Army Corps of Engineers, written commun., 2006). The SSC values at SFCO were highest (351 mg/L) during the event on December 13, 2003, when movement of the streambed was audible and discharge exceeded 5,300 ft3/s. The fine fraction (< 0.063 mm) during this sampling was only 46–48 percent, indicating that the energy of the stream was high and that a mix of fine- and coarse-grained material was being transported.
At least one other sample from each station had fine fractions that were relatively low (less than 80 percent), even with flows or SSCs that also were somewhat low. These data indicate that concentrations of larger particles were higher than fine particles, which typically is not the case during low flows. It is unknown what may have contributed to these anomalous size fractions; possibilities include sampling or laboratory errors, which would be magnified at low concentrations. These data were not removed from the regressions because information to confirm any errors was not available and because their effect on the regressions generally was small.
In general, the percentage of suspended sediment <0.063 mm, shown in figure 7 as the medians of all samples from each station, indicates differences in the character of the sediment being transported at the individual stations. The unregulated stations (SFCO and MRBO) tended to have a lower percentage of sediment as fine material, around 50–60 percent, and higher variability (table 2), whereas the regulated stations (BLUE and CGRO) typically had more than 90 percent of the suspended sediment as fine material, with less variation. The percentage of fine material was intermediate at VIDA, where a large component of the discharge originates from unregulated upstream reaches (represented by MRBO), but where suspended-sediment concentrations are apparently disproportionately influenced by releases from Blue River and Cougar Reservoirs. The differences in standard deviation of fine fraction among the sites indicate a dependence of the sediment character at a particular site on the stream’s energy, a primary difference between regulated and unregulated sites (Collier and others, 1996). Standard deviations in fine fractions were 11 percent at both regulated sites, 17 percent at both unregulated sites, and intermediate (13 percent) at VIDA. However, this finding should be treated as hypothesis rather than fact owing to different numbers of samples among sites and relatively few samples at some sites.
The regression equations and associated statistics for all stations are shown in table 3. With the exception of the regression at Blue River, SSC–turbidity correlations from all sites had r2 values greater than 90 percent, meaning that turbidity explained more than 90 percent of the variability in SSC at those sites.
Although sampling was targeted at events with high SSC and high turbidities in order to develop regressions that encompass the range of measured turbidity values at each site, maximum sampled turbidities were less than the maximums recorded by the continuous monitors at all stations (table 3). The estimation of SSC from regressions on turbidity, at turbidities beyond the maximum sampled value, is assumed by linear extension of the model, but remains unverified. Thus, during brief periods, estimated records of suspended-sediment concentration and load are tenuous and should be used qualitatively. This is especially true because a high percentage of the suspended-sediment load tends to be transported during the largest events, which are not entirely represented by the turbidity-SSC regressions generated. However, most of the instances of recorded turbidities greater than those sampled were anomalous events and are summarized as follows:
The only site with substantial time periods (days, rather than hours) where recorded turbidities exceeded the maximum sampled was Blue River below Blue River Reservoir. At that station, the maximum value sampled was 45 FNU, but there were three periods during autumn 2003 when recorded turbidities exceeded this sampled value. On November 17, one recorded value was 57 FNU. During the period from November 27 at 3:30 a.m. to November 29 at 10:30 p.m., a period of about 2.75 days, recorded turbidity values ranged from 45 to 103 FNU, with a median of 56 NFU. Finally, following the storm of December 13, 2003, recorded turbidities increased sharply due to construction work to stabilize a landslide that had covered State Highway 15 along the east shore of the reservoir—mud and debris removed from the road entered the upstream arm of the reservoir around 10 a.m., and by noon the turbidity values began to increase as the plume made its way through the reservoir and out the dam. This activity resulted in turbidities greater than 45 FNU (maximum 70 and median 61 FNU) for slighly longer than 1 day.
Suspended-sediment and turbidity data from each site, along with corresponding linear and power-function regression lines (from table 3), are shown in figure 8. Although SSC computations for estimation of loads are performed using power functions from table 3, linear plots are included in figure 8A to demonstrate the differences in slopes among stations. These differences potentially represent differences in the character of the sediment as it affects the optical quality of the turbidity measurement from station to station. Regression slopes for the unregulated stations were higher, indicating a higher concentration of suspended sediment is being transported for a given turbidity than at the regulated stations. This finding is consistent with data in figure 7 showing that the percentage of fine material in the suspended phase was higher at the regulated sites than at the unregulated sites. It also is consistent with theoretical aspects of turbidity monitoring, wherein smaller particles tend to cause a proportionally higher scattering of light, translating to higher nephelometric turbidity, than do larger particles (Brumberger and others, 1968; Sadar, 1998; Anderson, 2004). Color also can affect turbidity readings; however, the instruments used, with a near-infrared light source, are moderately robust with regard to the influence of color on turbidity (Anderson, 2004). In this case, the slope for VIDA also is similar to those from CGRO and BLUE, indicating that the sediment characteristics at VIDA may be determined more by sediment exported from the South Fork and Blue River than from sediment transported from reaches farther upstream.
Although the potential reasons for differences in slopes among the sites are consistent with other data in this study, they must be treated as hypotheses requiring further testing because the number of samples available for the analysis was relatively small. Additional data from the unregulated stations (MRBO and SFCO) might have resulted in modified turbidity–SSC regressions that were less distinct than those from the regulated sites; alternatively, the relations observed at CGRO may have been strongly influenced by events associated with the construction project itself (for example, the collapse of the Rush Creek Diversion or erosion of upstream deltaic sediments) and may not be valid when the reservoir is operating normally following completion of the construction project and refilling of the reservoir.
By using instantaneous turbidity data from the McKenzie River Basin continuous monitors and the derived turbidity-SSC regressions (table 3), instantaneous concentrations of suspended sediment were estimated on a thirty-minute basis for the periods of record at each site during water years 2002–04. From these estimated concentrations and instantaneous discharge data, SSLs were estimated on an instantaneous (every thirty minutes), daily mean, and annual basis as previously described. Annual discharge and suspended-sediment loads are presented in table 4. For MRBO, BLUE, and VIDA, where turbidity monitors were not installed until January, 2003, water year 2004 is the only water year with complete data and therefore the only year for which annual loads could be calculated. Partial-year loads were calculated for water year 2003 beginning in January 2003, when turbidimeters were installed at BLUE, MRBO, and VIDA. For comparison purposes, a partial-year record covering the same period was calculated for SFCO and CGRO.
Estimated annual loads were variable from year to year at all stations (table 4), depending on events occurring in each year. For instance, the load at SFCO during 2002 was 6,800 tons (or about 7,500 metric tons), most of which (>65 percent) was transported during one storm on April 13–16, 2002 (see fig. 6, appendix A), during the initial drawdown period. The same storm represented the first time the newly exposed deltaic sediments in the upper reaches of the Cougar Reservoir pool had been subjected to scouring flows since the reservoir was filled in 1953, and caused a large amount of erosion and channel changes through the valley bottom within the drawn down reaches of the reservoir and tributaries. The combination of a large influx of sediment from both the upstream reaches of the South Fork, upstream of the reservoir, and from erosion within the reservoir’s drawdown area, resulted in a high turbidity peak and the downstream transport of large amounts of suspended sediment. Much of the eroded sediment appears to have passed through the reservoir and into the South Fork below Cougar Dam. Of the estimated 17,000 tons that was released from Cougar Dam (measured at CGRO) during 2002 (table 4), about 14 percent (2,400 tons) was transported during the storm of April 13-16, 2002. Subsequently, high turbidity released from the dam during spring 2002 (fig. 3) resulted in a daily median load of about 240 tons of suspended sediment until mid-June. Some of this load likely was the result of persistently suspended fine particles within the reservoir pool from the storm of April 13‑16, 2002, and some smaller subsequent events, and some was from the nearly continuous evolution of the stream channel resulting from everyday flows in the exposed arms of the reservoir as the channel reached a new equilibrium position.
During water year 2003, additional sediment transport from the reservoir occurred during storms near the end of December 2002 and again at the end of January 2003 (fig. 6); however, the overall annual suspended-sediment load at CGRO was reduced compared to that in water year 2002 (table 4). Although the high daily median turbidity at CGRO on January 31 was significantly affected by the collapse of the Rush Creek Diversion, the storm also caused a major reconfiguration and downcutting of the upstream river channel through the exposed deltaic sediments (fig. 9), which was therefore another important source for high turbidity on that date. During this storm, the river pulled away from the left bank of the South Fork McKenzie River within the drawn-down portion of the reservoir near Terwilliger Hot Springs, effectively straightening through several large channel features and previous meanders. Similar changes happened in the side channels, particularly the East Fork of the McKenzie River, a tributary to the South Fork in the middle of Cougar Reservoir, a few miles downstream of the Terwilliger overlook, and selected other locations where sloughing or other erosion occurred during the same storm. These changes resulted in a prolonged release of high-turbidity water from the reservoir and transport downstream into the McKenzie River past the gaging station at Vida. On the basis of instantaneous sediment concentrations computed from turbidity–SSC relations, the period January 30–February 4 accounted for the downstream transport of 36 percent of the annual load at CGRO during water year 2003. Similarly, 28 percent of the annual load moved from December 27, 2002, to January 7, 2003. During both of these events, poorly mixed plumes of elevated turbidity were plainly visible in the mainstem downstream of the mouth of the South Fork McKenzie River.
The suspended-sediment load was markedly decreased during water year 2004 at CGRO compared to the preceding 2 years (fig. 10), despite the fact that annual discharge during water year 2004 increased by almost 50 percent from water year 2003 (table 4). This decrease in load was likely a result of the stabilization of the channel and banks by plant growth in the exposed arms of the reservoir, and the establishment of an equilibrium channel configuration. Scouring flows were experienced during water year 2004, including a bed moving event at the unregulated sites in December 2003, but with the major channel changes that took place in the reservoir in water year 2003, there was little opportunity for large-scale erosion in 2004 as had occurred previously. Annual discharge was slightly decreased upstream of the reservoir in water year 2004, but slightly increased at VIDA.
Similarly, annual discharge at SFCO, upstream of Cougar Reservoir, was slightly lower (5 percent) in water year 2004 than in water year 2003 (table 4), but suspended-sediment load increased slightly (7 percent) (fig. 10). The large storm in April 2002 (see fig. 6) contributed to a much larger annual suspended-sediment load (6,800 tons) and average SSC (10.7 mg/L) upstream of Cougar Reservoir than in the following years, despite annual discharge being slightly lower than in 2003 and almost identical to that in 2004. This event illustrates the large potential annual variability at a given site even with annual discharges remaining equivalent. The event also occurred during the initial drawdown period at Cougar Reservoir, as discussed previously, and likely contributed to the increased release of suspended sediment from the reservoir to the McKenzie River in spring 2002, as a result of both increased incoming load and erosion of the newly exposed deltaic sediments in the upper reaches of the reservoir.
Based on the differences in annual SSLs and discharge-weighted SSCs upstream and downstream of Cougar Reservoir (fig. 10), the reservoir evidently acted as a net sediment source during construction, particularly in 2002 and 2003. The difference between inflowing and outflowing loads was more than 10,000 tons in 2002, decreasing to 8,000 and 1,000 tons in 2003 and 2004, respectively. Much of this increased export likely originated from the erosion of the streambed within the exposed delta areas. However, the exact amount of erosion cannot be determined from this analysis because it does not account for re-settling of sediments within the remaining reservoir pool near the dam, nor would it differentiate between other potential sources, such as the Rush Creek diversion failure or slumping and resuspension from the reservoir’s steep side slopes.
Annual discharge-weighted SSC estimates, which also are corrected for partial records in table 4, provide another means for comparison. During most years, discharge-weighted SSC was lowest at the unregulated stations, MRBO and SFCO (around 2–3 and 4–6 mg/L, respectively, disregarding 2002 at SFCO), the regulated stations, CGRO and BLUE, had the highest concentrations, and those at VIDA were intermediate or similar to unregulated stations. The high annual SSC at CGRO during water year 2003 is partly a result of the large spike in SSC from the Rush Creek Diversion failure in January 2003, together with the substantially decreased discharge from the reservoir during that year. During 2004, when the newly formed channels through the exposed arms of the reservoir had apparently stabilized, the discharge-weighted annual SSCs entering and exiting Cougar Reservoir from stations SFCO and CGRO were equivalent (4.9 and 5.0 mg/L, respectively (fig. 10)), indicating the overall suspended-sediment contribution from Cougar Reservoir to the mainstem had been decreased to essentially background concentrations. However, on the basis of the previous discussion and data in table 2, the suspended sediment exiting Cougar Reservoir most likely had a higher percentage of fine material than did the suspended sediment entering the reservoir.
From a mass-balance standpoint, the monitoring network provided a reasonably complete accounting of the water sources in the study area and provides an opportunity to examine sources and fate of transported sediment. However, the late start to turbidity and SSC monitoring prevents a thorough estimate of deposition and transport during water year 2002, the initial year of the construction project and likely the most affected. Using water year 2004 as a test case, because it is the only water year with complete record at all sites, the sum of upstream annual discharges from MRBO, CGRO, and BLUE was 2,673,700 acre-ft (or 3.3×109 cubic meters). This is within 3 percent of the measured annual discharge of 2,743,000 acre-ft at VIDA (table 4). The most significant unmeasured inflow between the South Fork McKenzie River and the VIDA gaging station is Quartz Creek, a Western Cascades tributary on the south side of the river that generally has low flows but can swell during rainstorms to relatively high flows. Quartz Creek is likely the source of most of the 3-percent difference in summed discharges compared to that measured at VIDA.
In contrast to the 3-percent difference in measured, compared to summed, discharge at VIDA, the difference in suspended-sediment load from measured (16,900 tons) and summed (13,100 tons) estimates at VIDA is about 23 percent during water year 2004 (table 4). The difference is positive, indicating either unmeasured sources of suspended-sediment load between MRBO and VIDA, or that the errors in turbidity-SSC relations underestimate the upstream sources, overestimate the load at VIDA, or a combination of both. Deposition of the transported load in the reach between the South Fork and VIDA was apparently not significant during water year 2004, which is reasonable considering that stream velocities are relatively high in that reach and that suspended sediment measured at CGRO, BLUE, and VIDA tended to have relatively high fractions of fine material (medians 93, 91, and 85 percent, respectively). Quartz Creek may be a disproportionately large source of suspended sediment during winter rains, as there are large areas within the drainage basin that have been clear-cut and are potentially subject to erosion. The turbidity-SSC relations do not have as many data points as was initially desired, particularly at MRBO and BLUE, in part because of a lack of appropriate storms during the study period, so turbidity-based estimates of SSC could be biased low at the upstream sites or high at VIDA.
Similar conclusions may be drawn by using the partial record (January 29–September 30) from 2003 shown in table 4. The sum of upstream discharges from MRBO, CGRO, and BLUE (1,688,800 acre-ft) is within 7 percent of the partial-year discharge at VIDA, whereas the sum of upstream suspended-sediment loads is 32 percent less than the load at VIDA. These discrepancies indicate a possible sediment source, as opposed to large-scale deposition, between MRBO and VIDA. The partial-year suspended-sediment load from CGRO (about 7,400 tons) accounts for about 37 percent of the load measured at VIDA (about 20,000 tons) during the same period.
Sampling for ΣDDx and other organochlorine compounds in transport was conducted during three storms: February 3, March 7–8, and December 13–15, 2003 (table 5). Organochlorines were not detected in any water samples (whole-water or suspended sediment only). However, Practical Quantitation Limits (PQL—see U.S. Environmental Protection Agency, 2006) were variable, ranging from relatively low (less than 0.0005 µg/L) to relatively high (0.0052 µg/L). These concentrations are equal to 0.5–5.2 parts per trillion. Considering that the USEPA aquatic-life criterion is 0.001 µg/L, or 1 part per trillion, analysis of some of these samples had decreased likelihood of detecting ΣDDx, especially when turbidities were relatively low; however, when turbidities were high, the increased mass of sediment in the water column would be expected to carry more sorbed ΣDDx, and the likelihood of detection increased. No other organochlorine compounds were detected in any storm samples (appendix B).
At least one infiltration bag was retrieved intact from each location, and more than one in two cases. Issues preventing all bags from being retrieved intact included lost bags and retrieval cables that broke during the process of winching the bags out of the bed. One bag, from South Fork McKenzie River above Cougar Reservoir (SFCO), was found downstream during the deployment period following the large storm of December 13, 2003. During this storm, rocks and boulders were audibly being moved and large logs were being transported downstream at SFCO. The streambed likely was mobilized at this site, thus ripping the infiltration bag from its deployed position. Another bag from the same location was never found, despite the use of a metal detector, although it is unknown whether it too was ripped out of the bed or if aggradation and bed movement had merely buried its retrieval cables. During the initial retrievals, some cables broke, having rusted during the deployment period, and the bags could only be retrieved by digging them up which disturbed the bed and rendered the results qualitative. In order to retrieve the remaining infiltration bags without breaking the cables, a tension equalizer was devised that equally distributed the force from the tripod-mounted winch into each cable. Subsequent extractions were all successful with this device employed. Future deployments of infiltration bags will have a greater chance of successful retrieval if stronger cables (three-sixteenths or quarter-inch, stainless steel) are used.
The mass and percentage of all sediment, fine sediment, and ΣDDx in each is shown in table 6. The total mass of sediment retrieved ranged from about 23 to 43 kilograms (or about 51–95 pounds) among all bags, with the lowest amount at SFCO and the highest at MRBO. Amounts of sand and smaller (<2 mm in diameter) material ranged from 3 to 14 percent by weight among all bags, and fine material (silt and clay fractions, < 0.063 mm in diameter) ranged from about 65 to 415 grams (0.14–0.91 pounds), or 0.2–1.2 percent by weight, among all bags. Where two or more bags were retrieved, replication of sediment concentrations was good; at VIDA, standard deviations of the total fines and percent fines were about 15.2 grams (g) and 0.1 percent, or 7.1 and 16 percent of the averages, respectively; at MRBO the relative differences were 0.8 g and 0.02 percent, or about 0.9 and 9 percent of average, respectively. These results indicate that the method is sound and that variability of fine sediment deposition within a given area of similar hydraulic and geomorphological properties is low compared to variability between sites.
Although stream velocity may be an important factor contributing to deposition into the streambed, data were unavailable to fully assess the velocity differences among sites, particularly at high flows. At low flow, during deployment and retrieval periods for the infiltration bags, stream velocities were likely similar among sites because all sites were selected on the basis of a set of similar characteristics. These included maximum depths of 10–13 cm, and gravel beds at the head of a small riffle where downwelling would be likely. During high-flow periods, velocity differences among sites were undoubtedly important, on the basis of differences in peak flows and the gradients of the streams in different locations, as well as the moderating effect on flow and velocities exerted by Blue and Cougar Reservoirs. Data on ΣDDx were insufficient to assess method variability because of the lack of detections.
Despite the low variability of the deposited fine material within an individual site, variability was substantial among site types. Fine sediment (<0.063 mm) generally constituted less than 0.3 percent (about 0.23–0.28 percent) of the sediment in the infiltration bag at unregulated sites (MRBO and SFCO), but was greater than 1 percent at sites downstream of reservoirs, with the greatest deposition (1.2 percent) occurring at Blue River downstream of Blue River Reservoir (BLUE) (table 6, fig. 11). Deposition at VIDA (0.6 percent) was intermediate to that at the regulated and unregulated sites, which is consistent with the fact that it is located just a few miles downstream of the mouths of both Blue River and the South Fork McKenzie River. Apparently the sediment regime at VIDA was strongly influenced by sediment loads from the South Fork and from Blue River during the study period. In contrast to fine materials, more sand- and smaller-sized sediment (< 2 mm) was observed in infiltration bags from unregulated sites (9 to 12 percent) than at the regulated sites (about 3 to 6 percent). Deposition of <2-mm sediment in infiltration bags at VIDA was similar to unregulated sites (about 10–14 percent).
DDD and DDE were detected in deposited sediment from infiltration bags downstream of Cougar Reservoir, in concentrations of about 0.18–2.4 nanograms per liter (ng/L, or parts per trillion) and on the mainstem at VIDA (1.93 ng/L). This included sediment from one infiltration bag (CGRO) that had to be dug out manually due to broken retrieval cables, and which is therefore treated as qualitative from the standpoint of determining a total mass of ΣDDx in the infiltration bag. None of the parent compound, DDT, was detected in either of these samples, in contrast to the findings of the USACE’s initial sampling of streambank and lakebed sediments in which an average of about 15 percent of the detected ΣDDx was as DDT (U.S. Army Corps of Engineers, 2003). No other organochlorine compounds were detected in any samples (appendix C).
U.S. Department of the Interior | U.S. Geological Survey
URL: https://pubs.usgs.gov/sir/2007/5164
Page Contact Information: Publications Team
Page Last Modified: Thursday, 01-Dec-2016 19:49:38 EST
