Hydrology

Key Elements

• Since 1906, daily mean streamflow has been measured by the USGS at 29 active and now inactive sites.
• The Carmen-Smith–Trail Bridge hydropower dam complex decreased the frequency of large floods (greater than 10-year recurrence interval) in Reach 1 by 71 percent. No significant change has occurred in other environmental flow components.
• Large floods on the South Fork McKenzie (Reach 3) below Cougar Dam have been eliminated. Small floods (5–10 year recurrence interval) decreased in frequency and magnitude. The magnitude of high flow pulses decreased and of low flows increased. The timing of the lowest annual streamflow shifted from September to March. Monthly streamflows from February to May decreased, and monthly streamflows from July to November increased.
• Small and large floods below Blue River Dam (Reach 5) have been eliminated. Low flows have increased and the timing of the lowest annual streamflow has shifted from September to March. Monthly streamflows have decreased from February to May and increased from July to November.
• The frequency and magnitude of small and large floods has decreased in Reaches 7–12 (downstream of the dams). Low flows have increased in magnitude. No significant change in the timing of the annual lowest streamflow has occurred. Monthly streamflows during March–May have decreased and have increased during July–November.

Methods

Streamflow Data

The USGS has collected daily mean streamflow data at 29 streamflow-gaging stations in the McKenzie River basin since 1906 (table 5). Thirteen of these stations were active through water year 2008. The stations with the longest streamflow time series are the McKenzie River at McKenzie Bridge (14159000: 1910–1994) and the McKenzie River near Vida (14162500: 1925–2008). These daily streamflow time series were used to estimate a daily streamflow time series for each of the 12 study reaches.

Measured and Estimated Streamflow

Pre- and post-dam streamflow time series were assembled for each reach. Nine of the 12 study reaches included active or inactive USGS streamflow-gaging stations located at midpoints of the reaches and were considered representative of streamflow for the entire reach. It was necessary to extend existing records for most gaged reaches and to estimate streamflows for Reaches 4, 6, and 9, which did not have a streamflow-gaging station (table 6). Streamflow records for Reaches 8 and 10 were not extended into the pre-dam period because of limited data and because streamflow in those reaches have been diverted through the Leaburg and Waterville Canals since the 1930s. Additional details of the streamflow estimation procedure are included in appendix A.

Computed Unregulated Streamflow

In addition to the USGS measured streamflow data, the USACE computed unregulated daily streamflow time series for water years 1936–2004 at locations on the South Fork McKenzie River near Rainbow, Oregon [below Cougar Dam] (14159500), Blue River at Blue River, Oregon [below Blue River Dam] (14162200), and the McKenzie River near Vida, Oregon (14162500) (Julie Ammann, U.S. Army Corps of Engineers, written commun., 2008). These time series are an estimate of streamflow at these three locations if the two USACE dams had not been constructed. For this study, these time series were used to evaluate the hydrologic effect of the dams by comparing pre- and post-dam streamflow conditions.

The time series for the South Fork McKenzie River and Blue River sites were computed using correlations with nearby unregulated USGS streamflow records in McKenzie and Middle Fork River basins. Details on how the unregulated time series were computed are provided in appendix B. For this study, these two time series were extended by an additional 4 years (water years 2005–08) based on the USACE methods. The daily streamflow time series for the unregulated McKenzie River near Vida (14162500) was computed by first routing measured South Fork McKenzie (14159500) and Blue River (14152500) daily mean streamflows to McKenzie River near Vida (14162500). These two routed daily time series were then subtracted from measured Vida (14162500) daily mean streamflows to compute a time series for the local drainage basin upstream of Vida and downstream of the dams. The local time series and the computed unregulated Cougar Dam (14159500) and Blue River Dam (14152500) time series were then summed into a single daily time series.

Simulated Regulated Streamflow

USACE also simulated regulated daily streamflow time series for water years 1937–2004 at the Cougar and Blue River Dams and the Vida streamflow-gaging station using the USACE HEC–ResSim model (Julie Ammann, U.S. Army Corps of Engineers, written commun., 2009). In the simulations, the two dams were assumed to have existed for the entire 68-year period and operated consistent with releases from the dams based on the reservoir operation plans in accordance with the Biological Opinion released by NMFS in 2008 (National Marine Fisheries Service, 2008b). For this study, these time series were used to compare measured pre-dam streamflow conditions with hypothetical post-dam streamflow conditions (based on the Biological Opinion operation plans).

Bankfull Streamflow Estimation

In geomorphology, bankfull streamflow is often used as determinant of the size and shape of a river channel and can be defined as the discharge just contained within the banks. Using USGS measured stage and discharge data, measurement notes, photographs, and rating curves, bankfull stage and streamflows were estimated at the seven study reaches that have streamflow-gaging stations (table 7). The estimates have substantial uncertainty and are not necessarily representative of the entire reach. Bankfull streamflow estimates also were made for Reaches 4, 6, and 9, which do not have streamflow- gaging stations, based on an interpolation of estimates from the upstream and downstream reaches. Estimates were not made for Leaburg and Walterville power canal reaches (Reaches 8 and 10) because of the flow diversion from the river.

The National Weather Service also uses the same estimate of bankfull stage (8 ft) and streamflow (20,000 ft³/s) for the USGS streamflow-gaging station near Vida (14167500) (Reach 7) (National Weather Service, 2009). Since the construction of Cougar Dam in 1963 and Blue River Dam in 1969, USACE regulates streamflows at the Vida gaging station within the bankfull level to avoid flood damage. Daily mean streamflows at the gaging station have exceeded the bankfull streamflow estimate of 20,000 ft³/s in only 5 years during water years 1969–2008.

Indicators of Hydrologic Alteration

A statistical analysis of daily mean streamflow time series was used to quantify hydrologic changes resulting from construction and operation of dams and diversion canals in McKenzie River basin using the Indicators of Hydrologic Alteration (IHA) statistical software package developed by The Nature Conservancy (2007). The software package allows the user to analyze a daily mean streamflow time series as a single time period or divided into pre- and post-dam periods. For each user-defined time period, the software package computes 33 IHA metrics, which include mean monthly streamflows, annual streamflow extremes (1-day maximum and 7-day minimum), Julian day timing of streamflow extremes, low and high streamflow pulses, rise and fall rates, and the number hydrologic reversals (table 8). These metrics have specific linkages to ecosystem processes and functions.

The software package also separates daily mean streamflows into five categories, which are referred to as Environmental Flow Components (EFC), by using a simple algorithm. EFCs include extreme low flows, low flows, high-flow pulses, small floods (2-year events), and large floods (10-year events). The algorithm first separates all daily streamflows into either the high-flow pulse or low-flow groups. The highest 25 percent of daily streamflows (that is, greater than the 75th streamflow percentile) are assigned to the high-flow pulse group, and streamflows less than the median of the streamflows are assigned to the low-flow group. Daily mean streamflows between the 50th and 75th percentiles are assigned to the high-flow pulse group if that streamflow increased by 25 percent or more above the streamflow of the previous day, which also signifies the start of a high-flow pulse event. The high-flow pulse event ends when streamflows decrease by less than 10 percent per day. The high-flow pulse events are then categorized as a “small” or “large” flood if it has a recurrence interval (annual probability) of at least 2 years (50 percent) and less than 10 years (10 percent) or greater than 10 years (10 percent), respectively. IHA uses Weibull plotting positions to determine annual flood frequencies. This method is different from the Bulletin 17B Log Pearson III peak streamflow method (Interagency Advisory Committee on Water Data, 1981). Of the remaining streamflows that were categorized as low flows, an extreme low flow is a streamflow in the lowest 10 percent of the low flows. In this study, the time series were divided into pre- and post-dam periods, and the thresholds for the five categories (or components) are based on the pre-dam period data, which reflect unaltered streamflow conditions.

For each user-defined time period, the software package computes 34 EFC metrics (in addition to the 33 IHA metrics), which include the duration (days), magnitude (maximum or minimum streamflow of the event), timing (Julian day of the peak or minimum), and rise and fall rates for each of the five EFC categories (table 9). Complete IHA and EFC analysis results for all 12 reaches are included in appendix C.

The Indicators of Hydrologic Alteration software includes a “significance count” as a means of testing if the difference between pre- and post-impact period metrics is significant. The significance count value ranges from 0 to 1. A significance count value close to 0 indicates that the difference between the pre- and post-impact periods is highly significant. The value can be interpreted similarly to a p-value used in statistical tests. For this study, significance count values less than or equal to 0.05 were considered significant. Additional information about the test is included in appendix C.

Additional metrics used in the analyses of hydrologic alteration caused by dams that were not included in the IHA software package were pre- and post-dam period 2-, 10-, 50-, and 100-year flood statistics computed from annual peak streamflow data using the Bulletin 17B Log Pearson III method (Interagency Advisory Committee on Water Data, 1981) and streamflow durations (10th, 50th, and 90th percent exceedances) computed from daily mean streamflow data.

Hydrology Results and Discussion

The effects of dams and canals on streamflow regime are described below for selected reaches in the McKenzie River basin. Statistical metrics computed using the IHA software compare the pre- and post-dam periods in the context of different environmental flow components, such as low streamflows, high-flow pulses, small floods, and large floods. Graphical comparisons of pre- and post-streamflow regulation include mean daily streamflow plots based on measured and estimated daily mean streamflow data. Mean daily streamflow plots are created by computing the mean of all daily mean streamflows for each calendar day from a record of a site. (This is different from daily mean streamflow, which is the computed mean streamflow on a specific day). Because a mean daily streamflow plot dampens the magnitude of floods, comparisons of measured daily mean streamflows and USACE computed unregulated daily streamflows for a single water year (1995) also are included. Water year 1995 data were used in the daily mean streamflow comparison plots because it approximates an average year in the historic record based on a comparison of mean annual streamflows. Mean annual streamflows for water year 1995 and the period of record (water years 1925–2008) for the McKenzie River near Vida streamflow-gaging station (14162500) are 4,026 and 4,039 ft³/s, respectively. The driest and wettest mean annual streamflows for this period were 2,447 ft³/s (water year 1977) and 6,211 ft³/s (water year 1956), respectively.

A reach by reach summary of the hydrologic changes to the environmental flow components is provided in table 10. The Carmen-Smith–Trail Bridge Dam complex has caused minor hydrologic alterations in Reach 1 in the form of fewer floods. More profound effects have occurred in Reaches 3 and 5 downstream of the Cougar and Blue River Dams, respectively; large floods have been eliminated in these reaches. Streamflows have decreased from February to May, and increased from July to November. The hydrologic alterations created by the upper basin dams have propagated downstream to the Willamette River confluence; however, they are less pronounced in each consecutive downstream reach.

The effect of the dams on the magnitude and frequency of major floods in Reach 7 can be seen in the streamflow record of the McKenzie River near Vida streamflow-gaging station (14162500), which has been in operation since 1924 (fig. 7). Since the early 1960s, the dams have effectively kept most floods below the bankfull streamflow at the Vida gaging station of 20,000 ft³/s. Without the dams, the USACE estimated that the maximum daily mean streamflow of the floods of December 1964 and February 1996 would have been 62,338 and 47,622 ft³/s, respectively.

Monthly precipitation data at the Eugene airport from the pre- and post-dam periods, water years 1936–1962 and 1963–2008, were evaluated to determine whether climate was a contributing factor to changes in streamflow and not just the effects of dams. Based on a Wilcox rank-sum test, there was no significant difference in monthly precipitation between the two periods. The p-values for all months, with the exception of November, were greater than 0.05 (table 11).

McKenzie River—Reach 1

Mean daily streamflow data for pre- and post-dam periods (water years 1936–62 and 1963–2008) from the streamflow-gaging station at McKenzie River below Trail Bridge Dam, near Belknap Springs, Oregon (14158850), show that December streamflows slightly decreased and that January and March streamflows slightly increased in the post-dam period (fig. 8). Overall the effect of the Carmen-Smith–Trail Bridge dam complex was a modulation of the annual hydrograph. This also is evident in a comparison of water year 1995 measured regulated and computed unregulated daily mean streamflows (fig. 9). Although the timing of streamflow events remained constant, the magnitude of the high-flow events decreased and their streamflow recessions were elevated under regulated conditions. Summer streamflow for water year 1995 was greater under regulated conditions. These hydrologic effects are consistent with the operation of the Carmen-Smith–Trail Bridge Dam complex, which was constructed uniquely for hydropower production and has minimal total reservoir storage.

Downstream of the McKenzie River below Trail Bridge, near Belknap Springs, Oregon, streamflow-gaging station (14158850), the McKenzie River at McKenzie Bridge streamflow-gaging station (14159000), near the lower end of Reach 1, was in operation during water years 1911–94. Post-dam period (water years 1963–94) 2-, 10-, 50-, and 100-year peak streamflows, computed using the Bulletin 17B Log Pearson III method, were slightly (less than 10 percent) less than corresponding peak streamflows for the pre-dam period (water years 1911–62) (table 12). (It was not possible to compute peak streamflows for the for the gaging station at McKenzie River below Trail Bridge, near Belknap Springs, gaging station (14158850) because data collection began in water year 1960, which was only 3 years prior to the construction of the Carmen-Smith–Trail Bridge Dam complex (table 6). It was only possible to extend daily streamflows for this record and not annual peak streamflows.)

At the streamflow-gaging station below Trail Bridge Dam (14158850), the medians of the annual 1-day maximum and 7-day minimum streamflows for the post-dam period are slightly lower than during the pre-dam period (table 13). However, the IHA “significance count” p-values for both metrics were greater than 0.05, and the difference was not statistically significant. The median monthly streamflows from both periods show minor differences without a consistent pattern of increase or decrease (table 14). The differences were all statistically insignificant with the exception of January. Changes in the 10th, 50th, and 90th percent streamflow exceedances between the pre- and post-dam periods also were relatively minor (table 15). However, the frequency of large (10-year or greater) flood events (based on daily mean streamflow and not annual peak streamflow data) decreased during the post-dam period by 71 percent (table 15).

South Fork McKenzie River—Reach 3

Constructed in the early 1960s, Cougar Dam has reduced the magnitude of flood events, increased summer and early fall low flows in the South Fork McKenzie River (Reach 3), and changed the annual distribution of streamflow (figs. 10 and 11). Although December and January mean daily streamflows appear relatively unaffected, streamflows from February to May decreased as water is stored in the reservoir for summer releases. The date of the annual minimum streamflow shifted from August and September to March, April, and July. Daily streamflow releases from Cougar Dam for water year 1995 also show abrupt rises and falls that do not follow unregulated streamflow conditions (fig. 11).

For the pre-dam period (water years 1946–63), the 100-year flood magnitude was 34,430 ft³/s; however, for the post-dam period (water years 1964–2008), it decreased by 75 percent to 8,451 ft³/s (table 12). The median of the annual 1-day maximum daily streamflows decreased from 6,321 to 4,410 ft³/s, a difference that is statistically significant (table 13). On a monthly basis, there was a near-significant (0.08 or less) decrease in streamflows from February to May and a significant increase from August to November (table 14). High flows (10th percent streamflow exceedance) decreased by 16 percent and low flows (90th percent streamflow exceedance) increased by 18 percent (table 15). The frequency of small floods (5–10 year recurrence interval) decreased by 90 percent and large floods (greater than 10-year recurrence interval) were eliminated. The median of the annual 7-day minimum daily mean streamflows increased from 214 to 265 ft³/s (table 13), a difference that is statistically significant.

The medians of almost all monthly streamflows and 1-day annual maximum streamflow (but not the annual 7-day minimum streamflow) computed from the USACE computed unregulated streamflow data for the post-dam period (water years 1964–2008) were not significantly different from the pre-dam period (water years 1936–1963) medians (tables 13 and 14). This, along with the lack of significant difference in the pre- and post-dam period monthly precipitation data from the Eugene airport (table 11), indicate that climate is less of a factor than the dams in explaining the significant difference between the pre- and post-dam period observed streamflow data medians.

Blue River—Reach 5

Reach 5 extends from the Blue River Dam, completed in 1968, to the confluence with the South Fork McKenzie River. Like Cougar Dam, Blue River Dam is used for flood control in addition to other purposes (although not hydropower). Mean streamflow for Blue River is approximately one-half the mean streamflow for South Fork McKenzie River. Like Cougar Dam, Blue River Dam has a similar effect on the annual distribution of daily streamflows (fig. 12). Streamflows decreased in the spring and increased from the summer through the fall. Streamflow releases from Blue River Dam in water year 1995 also have abrupt rises and falls that are in contrast to the computed unregulated daily streamflow hydrograph (fig. 13). The 100-year flood magnitude for the pre-dam period (water years 1936–65) was 19,520 ft³/s. However, the 100-year flood magnitude for the post-dam period (water years 1969–2008) decreased by 76 percent to 4,677 ft³/s (table 12). Small and large floods, defined by the pre-dam period data, have been eliminated since the dam was constructed (table 15). Similar to the pre- and post-dam changes in median monthly streamflows in the South Fork McKenzie River (Reach 3), there was a significant decrease and significant increase in streamflows from February to April and July to November, respectively (table 14). The median of the annual 1-day maximum daily streamflows decreased from 4,982 to 3,010 ft³/s, and the median of the annual 7-day minimum daily streamflows increased from 23 to 47 ft³/s (table 13). The pre- and post-dam period difference for both these metrics was significant. Changes in median to high streamflows (10th and 50th percent streamflow exceedances) between the pre- and post-dam periods were relatively minor. However, the 90th percent streamflow exceedance significantly increased by 59 percent from 32 to 51 ft³/s (table 15).

When the medians of all monthly streamflows, 1-day annual maximum streamflow, and 7-day annual minimum streamflow computed from the USACE computed unregulated streamflow data for the post-dam period (water years 1969–2008) were compared with corresponding medians based on the pre-dam period (water years 1936–1968) measured streamflow data, their differences were not significant (tables 13 and 14). Pre- and post-dam period monthly precipitation data at the Eugene airport also were not significantly different (table 11). This would indicate that climate is less of a factor than the dams in explaining the significant difference between the pre- and post-dam period medians of measured streamflow data.

McKenzie River near Vida—Reach 7

Changes in the annual distribution of daily streamflow for the Vida streamflow record are similar to but less pronounced than those for the South Fork McKenzie River below Cougar Dam and Blue River mean daily streamflow hydrographs (fig. 14). Streamflows decreased in the spring and increased during the summer and fall. The water year 1995 daily mean streamflow hydrograph shows a major reduction in peak flows (fig. 15); however, the effects of regulation are dampened in Reach 7 in the water year 1995 non-flood period compared to similar periods of the water year 1995 South Fork McKenzie River and Blue River hydrographs (figs. 11 and 13).

The 100-year flood magnitude for the pre-dam period (water years 1925–1962) was 69,970 ft³/s at the Vida streamflow-gaging station. The 100-year flood magnitude for the post-dam period (water years 1963–2008) decreased by 41 percent to 41,390 ft³/s (table 12). The median of the annual 1-day maximum daily streamflows significantly decreased from 23,800 to 14,000 ft³/s (table 13). The frequency of small and large floods (based daily mean streamflow data) during the two periods decreased by 95 and 72 percent, respectively (table 15).

Although the dams decreased flooding, the median of the annual 7-day minimum daily streamflows significantly increased from 1,553 to 1,951 ft³/s (table 13). Low flows (90th percentile streamflow exceedance) increased by 35 percent, from 1,600 to 2,167 ft³/s; however, the median date of the annual minimum streamflow remained almost unchanged (table 15). The standard deviation of the daily streamflows decreased by 19 percent, indicating that streamflows were slightly less variable during the post-dam period (table 13). Median monthly streamflow changes in Reach 7 were similar to changes in Reaches 3 and 5, as there was a near significant decrease and significant increase in streamflows from March to May and July to November, respectively (table 14).

Similar to Reaches 3 and 5, the medians of all almost the monthly streamflows, 1-day annual maximum streamflow, and 7-day annual minimum streamflow computed from the USACE computed unregulated streamflow data for the post-dam period (water years 1963–2004) for Reach 7 were not significantly different from the pre-dam period (water years 1925–1962) streamflow medians (tables 13 and 14). This would indicate that climate is less of a factor than the dams in explaining the significant difference between the pre- and post-dam period streamflow medians.

Leaburg and Walterville Canals—Reaches 8 
and 10

Reaches 8 and 10 extend the length of the Leaburg and Walterville canals, respectively. The canals have reduced streamflows in Reaches 8 and 10 throughout the year by 1,000 to 2,000 ft³/s (figs. 16 and 17). If both canals did not exist with the current operation of the upper basin dams (Carmen-Smith–Trail Bridge, Cougar, and Blue River), minimum streamflows from July to October in Reaches 8 and 10 would be between 2,500 and 3,000 ft³/s. Winter and spring streamflows would be higher if both canals and the upper basin dams did not exist. Flood events would be more frequent and have greater magnitudes. Minimum summer streamflow would be approximately 2,000 ft³/s (figs. 16 and 17).

Computing pre-dam and post-dam period streamflow metrics for Reaches 8 and 10 was not possible because of limited daily streamflow data. The USGS streamflow-gaging stations below Leaburg Dam, near Leaburg (14163150) and near Walterville (14163900) measures streamflow only in the river channel (canal bypass). Both gaging stations have been in operation only during the post-dam period since water year 1990. Although EWEB has measured streamflow on the Leaburg and Walterville canals since 1998, and the USGS measured streamflow data in the Waterville canal during 1927–33, simultaneous streamflow records of the canal and the canal bypass during the pre-dam period prior to 1963 do not exist for either reach.

McKenzie River—Reach 12

The cumulative effect of the upper basin dams and canals on McKenzie River streamflow is evident in Reach 12 near the Willamette River confluence (fig. 18). Spring streamflows from February to May are lower and summer streamflows from July to September are higher in the post-dam period than in the pre-dam period. The magnitudes of high flow events also are lower in the post-dam period than in the pre-dam period (fig. 19).

The median 7-day minimum annual streamflow significantly increased in the post-dam period from 1,744 to 2,131 ft³/s (table 13). Another low-flow metric (90th percent streamflow exceedance) increased by 31 percent (table 15). However, the pre- and post-dam period medians of annual 1-day maximum streamflow significantly decreased, from 34,700 to 22,200 ft³/s (table 13). The frequency of small (2-year to less than 10-year) floods (based on daily mean streamflow data) during the two periods decreased by 76 percent. However, the median date of the annual minimum streamflow remained relatively unchanged (table 15). Median monthly streamflows from March through May significantly decreased as a consequence of streamflow regulation during the post-dam period. Median monthly streamflows from July through November significantly increased (table 14).

First posted February 8, 2010