Scientific Investigations Report 2006–5060
U.S. GEOLOGICAL SURVEY
Scientific Investigations Report 2006–5060
Managed according to the fill curve, Hagg Lake water surface levels generally rise in the winter as rainfall accumulates and demand for water is low (fig. 4). The water surface elevation typically reaches a maximum in late May and declines through the summer and autumn as downstream users call for water releases. The lake reached a minimum water surface elevation in these model scenarios between mid-November and mid-December.
The year 2002 base case, scenario 0, was simulated using the lake’s original fill curve. Scenarios 1 through 4 used the new fill curve for the 6.1 m (20 ft) dam raise. For these scenarios, inflows were sufficient to meet that target maximum elevation. Scenarios 5 through 8 used the new fill curve for the 12.2 m (40 ft) dam raise, but natural inflows were insufficient in 2002 to reach the target maximum elevation; the maximum elevation was 2.8 m lower than the target in late May; this unfilled storage corresponds to about 11.5 percent of the total potential storage. With extra inflow from the Sain Creek tunnel, scenarios 9 through 13 started the summer season with a full lake using the 12.2 m (40 ft) dam raise. Scenarios 14 [12.2 m (40 ft) raise] and 15 [7.6 m (25 ft) raise] in 2002 also started the summer with a full lake, due to augmented inflows via pump-back. The year 2001 base case, scenario 0, simulated with the lake’s original fill curve, did not fill the lake due to drought conditions (fig. 5). Even with pump-back in scenarios 14 and 15 for 2001, the lake did not reach the new fill curve for the 12.2 m (40 ft) or 7.6 m (25 ft) dam raises because available water for pumping was limited, again due to the drought.
Water surface elevations resulting from the two Sain Creek tunnel inflow scenarios, 9 (low inflow rate) and 10 (high inflow rate), were similar despite the difference in inflows. Both levels of inflow augmentation were sufficient to fill the lake, and excess water was simply spilled from the lake. These simulations were based on 2002, a normal hydrologic year; it is possible that the higher inflows would be necessary to fill the lake in a year with lower than normal inflows.
Water surface elevations for scenarios with initial water demands just after construction (“a”) and maximum water demands (“b”) generally were identical from January through the end of May (figs. 4, 5), as releases during this time were mostly for instream flow requirements and spills to keep the water surface elevation on the fill curve. Once summer water deliveries began, the water surface elevations for the “a” and “b” scenarios diverged. Note that the future maximum water demands caused the lake stage to be very low in November, thus making it more difficult to fill the lake for the next year without augmented inflows via a Sain Creek tunnel or from pump-back.
Changes in the water temperature regime can fundamentally alter a lake’s water quality. For instance, chemical reactions and algal growth rates are temperature dependent. The distribution of fish in a lake can change, as water temperature is an important habitat factor (Ferguson, 1958; Cherry and others, 1977). The duration and depth of temperature stratification also can affect the time and volume of water that is isolated during the summer, affecting anoxia and ammonia production and accumulation in the lake’s hypolimnion.
In the 2002 base case, the annual average lake temperature was 10.9°C (table 3). All 2002 dam-raise scenarios produced cooler annual average temperatures, ranging from 8.7 to 10.7°C. Scenarios with selective withdrawal, which allowed some of the warmer water near the top of the lake to be discharged downstream, were cooler [9.4°C, standard deviation (SD) 0.3°C], than those with only the original fixed-elevation outlet (10.3°C, SD 0.2°C). The maximum lake surface water temperature (averaged over the top 2 m) in the base case was 26.6°C. In the 2002 dam-raise scenarios, the peak surface water temperature ranged between 25.3 and 26.7°C.
Hagg Lake undergoes an annual cycle of thermal stratification. In this study, a thermocline was defined to be present when a temperature change of at least 1°C over 1 m depth was continuously present. The depth of the thermocline was defined as the depth with the greatest temperature gradient when a thermocline was present. Scenarios for 2002 using selective withdrawal had a shallower thermocline (7.1 m, SD 0.6 m) than those with only one fixed-elevation outlet (9.6 m, SD 0.9 m) (figs. 6, 7 and table 3). Selective withdrawal exported warm water from near the surface of the lake, limiting epilimnion development; scenarios with one fixed outlet at a mid-to-lower elevation in the lake stored more summer heat at the lake surface and allowed the epilimnion to develop undisturbed.
The thermocline depth deepened through the summer into autumn for those 2002 scenarios with only one fixed-elevation outlet (scenario 10b in fig. 6). In all scenarios, the elevation of the lowest lake outlet appeared to provide a lower limit for the thermocline depth. This result also was found in modeling by Casamitjana and others (2003) for reservoirs where summer inflow rates were small compared to withdrawals.
Scenarios for 2002 with the 12.2 m (40 ft) raise and extra inflow from the upper Tualatin River via the Sain Creek tunnel (10–13) and from pump-back (14) resulted in lower annual average water temperatures compared to their counterpart scenarios without extra inflow (scenarios 5–8 for the former and 6 for the latter). The difference averaged about -0.5°C for the Sain Creek tunnel and -0.4°C with pump-back. All scenarios with extra inflows also had shorter durations of thermal stratification and shallower thermoclines.
The 2001 scenarios (14 and 15) that combined the effects of selective withdrawal and pump-back resulted in cooler annual average temperatures compared to the base case; the difference averaged –0.8°C. These scenarios also produced longer periods of thermal stratification and a shallower thermocline relative to the 2001 base case.
All “a” scenarios, which resulted in a deeper lake in October–December than the “b” scenarios, had a later turnover date than their “b” counterparts, by an average of 15 days in 2002 and 9 days in 2001. Given the model’s high degree of accuracy in simulating lake circulation and heat-exchange processes, the model’s predictions of the turnover date also should be accurate. The turnover date, of course, is influenced greatly by meteorological conditions, so variations in the weather can easily push the turnover date to earlier or later dates.
Water temperatures in Scoggins Creek downstream of the dam are heavily influenced by the presence of Hagg Lake. The temperature maximum in Scoggins Creek upstream of the reservoir, for example, typically occurs in late July (fig. 3). Downstream of the dam, the peak temperature in Scoggins Creek usually occurs in late September. This shift in the peak temperature occurs because Hagg Lake releases relatively cold water from near the bottom of the lake in spring and summer, then releases stored summer heat in autumn as the lake is drawn down. Keeping a more natural seasonal temperature pattern downstream of a dam is often desirable for aquatic biota (fish). Some dams have been designed or retrofitted with selective withdrawal towers, which allow discharge waters to be blended from different lake elevations with different temperatures to match a more natural seasonal temperature pattern. As described previously, selective withdrawal capability was modeled for Hagg Lake in some scenarios.
Scenarios using only the original, fixed-elevation outlet (scenarios 0, 1, 5, 9, 10), did not have selective withdrawal capability. For these runs, the simulated outflow temperature was cooler than the target temperature through August (or September), and warmer than the target for most of the rest of the year (fig. 8). During a short part of the year, outflow temperatures from some of these scenarios exceeded the maximum water temperature criteria for Scoggins Creek (fig. 8). The scenarios with larger withdrawals, the “b” scenarios using maximum water demands, had greater exceedances of the temperature criteria than those that kept more water in the lake through the year. These exceedances can be quantified in terms of “degree days” (Bartholow and others, 2001) (table 4), where 1 degree day represents 1°C above the standard for a period of one day. Note that these comparisons do not use the 7dADM water temperature; this analysis was not meant to be a critical analysis of compliance or violation of the temperature standard, but a simple indication of a scenario’s ability to keep the downstream temperatures less than the relevant temperature criterion.
Scenarios with selective withdrawal capability were run in an attempt to meet the downstream temperature target. Scenarios 2, 6, 11, 14, and 15, which used both the original fixed outlet and one sliding-gate (variable-elevation) outlet, were best able to match the target outflow temperatures (figs. 8, 9). Late in the year, however, the water left in the reservoir was not cold enough to reach the target; the lake “ran out” of cold water. None of the 2002 outflows from these scenarios violated the downstream temperature criteria (figs. 8, 9; table 4). Scenarios 3, 7 and 12, simulating two fixed-elevation outlets, allowed blending of water from two different depths until the lake’s stage declined below the top outlet, thus ending selective withdrawal capability. Outflow water temperatures exceeded the criteria to some extent in all these scenarios (fig. 8; table 4). Scenarios 4, 8 and 13 simulated only one outlet, but it was a sliding-gate (variable-elevation) outlet. These scenarios matched the general seasonal pattern of the target temperature (as in scenarios 2, 6, 11, 14, 15), but because only one elevation adjustment per day was allowed and because the outlet was usually positioned in the middle of the thermocline, the temperature of the outflow varied greatly both above and below the target temperature. This variability led to occasional exceedances of the downstream water temperature criteria in the summer (fig. 8; table 4).
On an annual average basis, outflows from 2002 scenarios with selective withdrawal were warmer (11.6°C, SD 0.8°C) than those using just the original fixed-elevation outlet (8.9°C, SD 1.2°C). The effort to match a more natural seasonal temperature pattern in the outflow causes a large amount of heat to be exported from the lake’s epilimnion during the summer. The scenarios that were best able to match the downstream temperature targets were those that could access both warm water near the lake surface and colder water in the deeper part of the lake.
Scenarios for 2002 with extra inflow from the Sain Creek tunnel or from pump-back resulted in lower annual average outflow water temperatures compared to their counterpart scenarios without the extra inflow. The difference averaged –1.2°C.
The 2001 scenarios that combined the effects of selective withdrawal and pump-back resulted in warmer outflow temperatures compared to the base case, as expected; that difference averaged +1.2°C. These scenarios also reduced the degree-days over the water temperature criteria, so that “a” scenarios were less than the criteria at all times, and “b” scenarios exceeded the criteria by less than 1 degree‑day; using the simulated 7dADM water temperature, no exceedances of the standard were simulated.
Adequate concentrations of dissolved oxygen in a lake are critical to the health of its aquatic biota. When Hagg Lake is thermally stratified, the hypolimnion loses contact with the atmosphere; dissolved oxygen in these deep isolated waters can become depleted due to aerobic decomposition of organic material (in the water column and surficial sediments) and due to ammonia nitrification. As the hypolimnion becomes anoxic, it becomes an inhospitable environment for many biota. In addition, loss of oxygen stimulates a range of reactions such as the hydrolysis and deamination of proteins to produce ammonia, the reduction and dissolution of iron and manganese oxides, and the desorption of any orthophosphate that had been adsorbed to those oxides.
All 2002 scenarios resulted in fewer days with low (<1 mg/L) dissolved oxygen in the lake (0 to 53.2 days), compared to the base case (64.2 days, table 3). Managing the lake releases in the 2002 scenarios to meet a downstream temperature target using selective withdrawal also affected the extent of in-lake anoxia. The extent of anoxia was greatest in 2002 scenarios that used only the original fixed-elevation outlet (figs. 10, 11; table 3). With only one fixed-elevation outlet, the model simulated an average of 46.1 days (SD 8.6 days) where dissolved oxygen concentrations were less than 1 mg/L somewhere in the lake. Using selective withdrawal, fewer days with those conditions occurred (16.3 days, SD 12.8 days). Clearly, the spatial and temporal extent of hypolimnetic anoxia in Hagg Lake could to some extent be controlled and minimized through careful utilization of selective withdrawal. The annual average volume of the lake with dissolved oxygen less than 1 mg/L was 0.36 percent (SD 0.18 percent) using only the original fixed outlet, and 0.08 percent (SD 0.11 percent) considering all scenarios with selective withdrawal. All these results were substantially lower than the base case, at 1.26 percent.
Different water levels in the 2002 model scenarios affected the timing and extent of anoxia in the lake’s hypolimnion. Lake turnover marks the end of anoxia; the turnover date was calculated on the basis of a breakdown in the dissolved oxygen gradient: when at least a 1 mg/L difference in dissolved oxygen concentration over 1 m depth was no longer present. Turnover occurred on average on day 329.5 (November 25; SD 2.6 days) for the 2002 “b” scenarios, the scenarios with maximum future water demands, which left less water in the lake. The turnover date for the 2002 base case was similar, on day 329.2. Turnover occurred more than 2 weeks later for the 2002 “a” scenarios [average day 344.5 (December 10); SD 5.3 days], the scenarios which left more water in the lake. Supporting the idea that taking more water from the lake leads to an earlier turnover, it also was found that earlier turnover, on day 317.1 (November 13) occurred during the drought year, 2001, which had the lowest level ever recorded in Hagg Lake.
Scenarios for 2002 with extra inflow via the Sain Creek tunnel resulted in higher annual average dissolved oxygen concentrations compared to their counterpart scenarios with the dam raise but without extra inflow; the difference averaged about 0.3 mg/L. There was little difference in dissolved oxygen concentrations after adding extra inflows via pump-back compared to their counterpart scenarios without pumping for 2002.
The 2001 scenarios that combined the effects of selective withdrawal and pumping resulted in higher annual average dissolved oxygen concentrations relative to the base case; that difference averaged +0.3 mg/L. The scenarios also resulted in a lower lake volume with low dissolved oxygen concentrations compared to the base case; that difference averaged –1.8 percent. These scenarios produced, on average, 33 fewer days with low dissolved oxygen, and the turnover occurred later in the year, trends that are similar to those from 2002. These results are consistent with the idea that having more water in the lake leads to later turnover.
Previous modeling work showed that waters withdrawn from Hagg Lake have a seasonal cycle in dissolved oxygen concentrations, with high concentrations in winter and a minimum in late summer/autumn that typically is less than 5 mg/L. However, the released waters are well-oxygenated in a turbulent aeration basin, causing the water in the creek downstream of Hagg Lake to be well-oxygenated and close to saturation (Sullivan and Rounds, 2005). With the structural changes under consideration, some of the outflow may be routed directly into a pipe rather than into the aeration basin. Dissolved oxygen concentrations in the outflow, therefore, are important to consider, especially because other chemical constituents such as ammonia, dissolved iron and manganese, and phosphorus may be present in heightened levels in water with low dissolved oxygen. Furthermore, if CWS flow-augmentation water is released directly to a pipe, some subsequent method of aeration may be necessary before that water is returned to the river.
Similar to the temporal patterns of dissolved oxygen in the lake, the outflow concentrations in all 2002 scenarios were highest in winter and spring, and reached their lowest concentrations in autumn (figs. 12, 13). Scenarios with the “a” water demands reached lower minimum dissolved oxygen concentrations in the outflow (2.71 mg/L; SD 1.44 mg/L), with minima later in the year (day 327.4; SD 17.8 days), compared to scenarios with “b” maximum water demands, which had higher minimum concentrations (4.17 mg/L; SD 1.12 mg/L) and earlier dates for those minima (day 290.6; SD 17.6 days). This pattern is a direct consequence of the later turnover of the lake in the “a” scenarios, where the near-bottom waters remain isolated and are exposed to oxygen-consuming reactions for a longer period of time. No attempt was made in these scenarios to avoid low concentrations of dissolved oxygen when positioning the sliding-gate outlets (see Appendix A).
Scenarios for 2002 with extra inflow from the Sain Creek tunnel and from pump-back resulted in higher annual average outflow dissolved oxygen concentrations compared to their counterpart scenarios with the dam raise but without extra inflow. The difference averaged about +1.18 mg/L for the inflows from the Sain Creek tunnel and +0.75 mg/L for the pump-back scenario. The outflow minimum dissolved oxygen concentrations also increased in both of these scenario groups; they increased by an average of +1.25 mg/L with inflows from the Sain Creek tunnel and by +0.55 mg/L using pump-back. When compared to the 2002 base case, scenarios with pump‑back water had minimum dissolved oxygen concentrations that were lower by an average of 1.8 mg/L.
The 2001 scenarios that combined the effects of selective withdrawal and pumping resulted in lower minimum dissolved oxygen concentrations compared to the base case; the minimums changed by an average of –4.4 mg/L. The lowest dissolved oxygen concentration in the outflow for all scenarios was 0.18 mg/L for scenario 14a in 2001.
High concentrations of ammonia can be toxic to biota, can lead to low dissolved oxygen concentrations via nitrification, and generally are not desirable in water designated for Scoggins Creek or downstream users. In drinking water treatment plants, ammonia can affect disinfection efficiency, result in nitrite formation, cause the failure of manganese removal filters, and produce taste and odor problems (World Health Organization, 2003). In Hagg Lake, concentrations are low through much of the lake, but substantial concentrations of ammonia can accumulate in an anoxic hypolimnion (Sullivan and Rounds, 2005).
The 2002 base case had an annual average ammonia concentration of 12.9 µg/L as N and a maximum hypolimnetic ammonia concentration of 251 µg/L as N. All other 2002 scenarios resulted in lower annual average ammonia concentrations (10.1 to 12.2 µg/L as N; table 5) and lower maximum hypolimnetic ammonia concentrations (10.6 to 229 µg/L as N; table 5). The decrease in ammonia concentrations correlated well with shortened periods of anoxia in the hypolimnion, except for the 2002 pump-back scenarios 14 and 15 (fig. 14).
Selective withdrawal helped to minimize the ammonia concentration in the hypolimnion (figs. 15, 16; table 5). The annual average ammonia concentration for scenarios using only the original fixed-elevation outlet was 11.6 µg/L as N (SD 0.3 µg/L as N); for scenarios with selective withdrawal, the annual average concentration was 10.8 µg/L as N (SD 0.7 µg/L as N).
Scenarios for 2002 with extra inflow from pump-back resulted in higher annual average ammonia concentrations compared to their counterpart scenarios without extra inflow; the difference averaged about +1.4 µg/L as N. There was little difference in the annual average ammonia concentration after adding extra inflows via the Sain Creek tunnel for 2002.
The 2001 scenarios, which combined the effects of selective withdrawal and pump-back, resulted in lower annual average ammonia concentrations compared to the base case; the decrease averaged –1.8 µg/L as N.
Concentrations of ammonia in the outflow for 2002 scenarios were low, with annual averages generally ranging between 9.3 and 13.3 µg/L as N (table 4). With only one fixed outlet, the highest concentration of ammonia in the outflow occurred during turnover (fig. 17, scenarios 0 and 10), when the ammonia concentrated in the hypolimnion became mixed through the lake. However, annual average concentrations of ammonia in the outflow were not distinctly different among the various scenarios (figs. 17, 18).
Scenarios for 2002 with extra inflow from the Sain Creek tunnel resulted in a small (–0.5 µg/L as N) decrease in the annual average outflow ammonia concentration compared to their counterpart scenarios without extra inflow. On the other hand, extra inflow from pump-back resulted in an increase in the annual average outflow ammonia concentration (by +1.3 µg/L as N).
Excluding some pump-back scenarios, 2002 scenarios had lower annual average chlorophyll a concentrations in the photic zone (1.05 to 1.54 µg/L) compared to the 2002 base case (1.65 µg/L). A sensitivity analysis of the Hagg Lake model (Sullivan and Rounds, 2005) showed that blue-green algae were temperature sensitive, preferring warmer water, so it is possible that the overall cooler lake temperatures produced in the model scenarios caused a decrease in algal growth. Export of algae downstream in scenarios with selective withdrawal, which withdrew water from the epilimnion where algae grew, also could have contributed to lower chlorophyll a concentrations. Indeed, 2002 scenarios with selective withdrawal capability had shorter periods of blue-green algal concentrations greater than 0.5 g/m3 (40.5 days; SD 19.9 days), compared to those using just the original fixed-elevation outlet (67.9 days; SD 2.9 days).
All 2002 scenarios had higher average whole-lake orthophosphate concentrations (5.22 to 9.27 µg/L as P) compared to the 2002 base case (5.05 µg/L as P). Algae use orthophosphate for growth, and, except for higher orthophosphate concentrations with the pump-back option (described below), some of the higher orthophosphate concentrations probably are directly related to the lower chlorophyll a and algal concentrations in the scenarios.
Scenarios for 2002 with extra inflow from pump-back resulted in higher photic-zone average chlorophyll a and whole-lake average orthophosphate concentrations compared to their counterpart scenarios with the dam raise, but without extra inflow. The difference was +0.4 µg/L for chlorophyll a and +3.8 µg/L as P for orthophosphate. This is a 69 percent increase in whole-lake average orthophosphate; average total phosphorus concentrations also increased, by about 39 percent. The increase in phosphorus loading produced an increase in the duration of the blue-green algal bloom and the aforementioned increase in chlorophyll a. In this model, the greatest uncertainty in the model’s predictions involves algal dynamics, and this should be considered when interpreting the results. However, the potential of increasing the size and duration of algal blooms in the reservoir is a concern to downstream water users (taste and odor issues, for instance) and should be considered in implementing or managing this option. There was little difference in the photic-zone average chlorophyll a or whole-lake average orthophosphate concentration for the scenarios with added inflows from a Sain Creek tunnel compared to their counterpart scenarios with the dam raise, but without extra inflow.
The 2001 scenarios that combined the effects of selective withdrawal and pump-back resulted in slightly higher whole‑lake average orthophosphate concentrations and slightly lower photic-zone average chlorophyll a concentrations compared to the base case; the differences were +0.5 µg/L as P and –0.1 µg/L, respectively. Blue-green blooms did not exceed 0.5 g/m3 in these scenarios, possibly due to cooler epilimnetic temperatures. The increase in orthophosphate was not as great in 2001, in part because the available pump-back water had lower orthophosphate concentrations compared to 2002.
Use of selective withdrawal typically increased the concentration of chlorophyll a in the 2002 scenario outflows compared to the base case and other scenarios using the original outlet. The annual average outflow chlorophyll a concentration was 0.83 µg/L (SD 0.17 µg/L) for scenarios with selective withdrawal and 0.31 µg/L (SD 0.14 µg/L) for scenarios using only the original fixed-elevation outlet. This result is directly attributable to the fact that selective withdrawal drew water from near the lake’s surface, where the algae generally reside (because they require light for growth), while the water at the depth of the original outlet typically was well below the photic zone and therefore had a lower chlorophyll a concentration.
Conversely, the use of selective withdrawal decreased the concentration of orthophosphate in the scenario outflows (average 5.36 µg/L as P; SD 0.42 µg/L as P; excluding pump-back runs) compared to the base case. The scenarios using only the original fixed outlet had higher concentrations of orthophosphate (average 6.93 µg/L as P; SD 0.36 µg/L as P) compared to the base case.
The lower chlorophyll a and higher orthophosphate concentrations from scenarios with only the original fixed outlet, as well as the general inverse relation between orthophosphate and chlorophyll a in the outflow for the scenarios, is illustrated in figure 19. In that figure, the points for scenarios with only one fixed outlet (0, 1, 5, 9, and 10) all lie on or near the same line. The relative position of these points is due to the average depth of the withdrawals and the fact that higher algae concentrations and therefore lower orthophosphate concentrations are found at shallower depths. Scenario 0, the base case, is the shallowest of these scenarios, followed by scenario 1 [only a 6.1 m (20 ft) dam raise] and scenarios 5, 9, and 10. The “a” scenarios result in a deeper lake than the “b” scenarios, which helps explain the rest of this trend. Most of the selective withdrawal scenarios lie in the opposite direction, due to the increased export of shallow water where the algae grow.
Scenarios for 2002 with extra inflow from the pump-back option produced increases in both chlorophyll a and orthophosphate in the outflow (fig. 19), indicating again that the importation of extra phosphorus led to slightly higher levels of algal growth in that year. The 12.2 m (40 ft) dam raise with pump-back (scenario 14) produced higher outflow orthophosphate concentrations than the 7.6 m (25 ft) dam raise with pump-back (scenario 15), due to the fact that the 12.2 m (40 ft) dam raise required more pump-back of the higher phosphorus Tualatin River water to fill the larger reservoir. Extra inflow from the Sain Creek tunnel did not appreciably change either the annual average chlorophyll a or orthophosphate concentration in the outflow in scenarios that used that option.
The 2001 scenarios that combined the effects of selective withdrawal and pump-back resulted in slightly higher average outflow orthophosphate concentrations and slightly lower chlorophyll a concentrations compared to the base case; the changes averaged +0.7 µg/L as P and –0.3 µg/L, respectively.
For more information about USGS activities in Oregon, visit the USGS Oregon Water Science Center home page.
U.S.
Department of the Interior | U.S. Geological
Survey
Persistent URL: https://pubs.water.usgs.gov/sir20065060
Page Contact Information: Publications
Team
Page Last Modified:
Thursday, 01-Dec-2016 19:05:03 EST
