Summary and Conclusions

Regression techniques, mass and energy balances, measured water temperatures upstream of reservoirs, and modeled and historic rates of summertime warming with downstream distance were used to estimate the streamflows and water temperatures that likely would occur at 12 major dam sites in the Willamette River basin in the absence of those dams and any upstream dams. The dam sites include River Mill Dam on the Clackamas River as well as all major dams of the Corps of Engineers’ Willamette Project. Because of short residence times in two re-regulating reservoirs, the effects of Big Cliff and Detroit Dams on the North Santiam River, and Dexter and Lookout Point Dams on the Middle Fork Willamette River, were combined and treated as single dam sites for this analysis. These without-dam streamflows and water temperatures were estimated for calendar years 2001 and 2002 to match the general time frame used in the Willamette River water-temperature Total Maximum Daily Load (TMDL) analysis. Estimated errors for the without-dam temperature estimates ranged from 0.5 to roughly 1.5 °C for most of the dam sites in mid-summer, depending on the methods used, the availability of measured temperatures upstream of the reservoirs, and the amount of summertime warming applied through the reservoir reach.

The downstream effects of the dams were evaluated with a suite of CE-QUAL-W2 flow and temperature models by imposing both with- and without-dam streamflows and water temperatures at the dam sites as upstream boundary conditions. The models included the entire Willamette River and selected major tributaries up to their first major dams, were constructed and calibrated originally for the Willamette River water-temperature TMDL, and required several small modifications to allow them to run under the greatly changed without-dam streamflow conditions. The modeling time period for the TMDL and this analysis focused on the summertime low-flow period from April (2002) or June (2001) through October. Results were analyzed using the 7-day average of the daily maximum (7dADM) temperature, the statistic specified in the State of Oregon maximum water-temperature standard.

Many of the larger, taller dams in the Willamette River basin that are operated for flood control and hydropower generation have characteristic effects on streamflow and water temperature at the dam sites. The dams tend to reduce peak flows during precipitation or snowmelt time periods (for flood control) and augment streamflow during the summer low‑flow period (for downstream navigation, irrigation, and other needs). Because the taller dams typically release water from a mid-depth or deeper outlet in the upstream impoundment, that water tends to be colder in mid-summer than it would be without the dam. In September or October, a large amount of water is released from many of these dams to make room for flood storage, which can bring warmer surface waters down to the elevation of the outlet, thus releasing the warmest water of the year during a time period when the river without the dam would be cooler because of shorter days and colder air temperatures. Many of the taller dams, therefore, tend to shift the annual maximum water temperature from July/August to September/October, and can release water that generally is 6 to 10 °C cooler or warmer than what would occur without the dams.

The temperature effects of each dam are specific to the characteristics and operation of each structure. Starting in 2005, for example, Cougar Dam on the SF McKenzie River no longer produces a seasonal thermal shift because it was retrofitted with a selective withdrawal tower that allows operators to produce a more natural seasonal temperature pattern downstream of the dam. Detroit and Big Cliff Dams on the North Santiam River produced the typical seasonal shift in maximum temperature during the time period of this study, but were operated differently in 2008 and 2009, releasing water from multiple depths in Detroit Lake to produce a more natural pattern of downstream temperatures. A similar effect of releasing water from different outlets was apparent in the temperatures at Fall Creek Dam during the study period, where abrupt changes in the release temperature were indicative of a change in dam operations. Shorter dams, such as River Mill Dam on the Clackamas River and Fern Ridge Dam on the Long Tom River, do not exhibit a seasonal maximum temperature shift because the upstream reservoirs are not deep enough to provide a significant pool of cold hypolimnetic water at the dam outlet.

The degree to which the reservoirs are filled at the beginning of the summer season also is important in determining the temperature effects of the dams. Precipitation during the winter of 2000–01 was abnormally low and insufficient to fill most of the larger reservoirs in the Willamette River basin. The lower water level resulted in a shallower release, closer to the warmer epilimnion during summer, when using the normal release point from these dams. As a result, water released in mid-summer from several of the Willamette Project dams (Big Cliff/Detroit, Blue River, Dexter/Lookout Point, and Cottage Grove Dams in particular) was significantly warmer in 2001 than in 2002.

Estimation of without-dam water temperatures at the dam sites showed that the sites can be grouped according to their upstream drainage characteristics and elevation, a finding that is consistent with recent research by Tague and others (2007) for the McKenzie River basin. Sites that derive most of their flow from areas in the High Cascades, for example, tend to have cooler mid-summer temperatures because that water originates from snowmelt and cold spring complexes high in the mountains. This is the case for dam sites on the SF McKenzie River (Cougar Dam), the North Santiam River (Big Cliff and Detroit Dams), and to a slightly lesser degree, the upper reaches of the MF Willamette River (Hills Creek Dam) and the Clackamas River (River Mill Dam), although the latter site is at a lower elevation. Each of these sites produces without-dam mid-summer 7dADM temperature maxima less than 20 °C. The SF McKenzie River site is the coldest, with annual maximum 7dADM temperatures less than 15.5 °C, because it derives the greatest amount of flow from the High Cascades and is located at a relatively high elevation.

Sites that derive their flow from the Western Cascades, Coast Range, and Willamette Lowlands, areas that are affected more by rainfall than by snowmelt and lack the cold and consistent springs present in the High Cascades, tend to have warmer mid-summer temperatures, with annual maximum 7dADM temperatures greater than 21 °C and as high as 25 °C. Sites at higher elevations, such as the Blue River dam site, tend to produce mid-summer temperatures at the low end of this range, whereas sites at lower elevations, such as Fern Ridge on the Long Tom River, Cottage Grove on the Coast Fork Willamette River, Foster on the South Santiam River, and Dorena on the Row River, tend to have warmer without-dam temperature estimates.

In order to address temperature issues related to dams and climate change, it is critical to maintain long-term year-round temperature monitors at key unregulated sites in several characteristic physiographic areas of the Willamette River basin. Currently, Blowout Creek near Detroit (USGS station 14180300) is one of the only long-term (10+ years), year-round temperature monitoring sites that is representative of conditions in the Western Cascades physiographic region. Several long-term, year-round temperature monitors exist at sites that are representative of conditions in High Cascades streams, such as the SF McKenzie River above Cougar Reservoir (USGS station 14159200), Breitenbush River above French Creek (USGS station 14179000), and the North Santiam River below Boulder Creek (USGS station 14178000). The latter two sites are upstream of Detroit Lake in the North Santiam River basin. New sites for temperature monitoring are being established at many sites throughout the Willamette River basin, including sites upstream of key reservoirs. Data from those sites will prove helpful for assessing basin-specific thermal characteristics, trends, and the effects of dams, but these new sites still have a relatively short period of record. Until datasets with longer period of records have been collected at these new sites, it is critical to maintain the few long-term temperature-monitoring sites that do exist.

Downstream of the dam sites, model results show that the effects of the dams persist to the mouth of the Willamette River, although the thermal effects diminish with downstream distance. The downstream thermal effects are caused by changes in both streamflow and water temperature at the dam sites. Without the dams, greatly decreased streamflow during the late summer months results in longer travel times through the river network, allowing additional time for the river to gain (or lose) heat from (to) its surroundings. Late-summer streamflow in the Willamette River is augmented by dam releases, increasing the flow at Albany and at Salem by approximately a factor of 2 or more during August, September, and October in 2001 and 2002. No attempt was made to separate the effects of altered streamflows relative to altered upstream temperatures on downstream water temperatures, but the effect of changes in streamflow are likely to be significant.

The thermal effects of the dams are greatest at the dam sites, where the 7dADM temperatures are as much as 6 to 10 °C cooler or warmer compared to what would occur without the dams. Downstream, the effects decrease, but are still in the 0.5 to 1.0 °C range near the mouth of the Willamette River, much larger than the cumulative point-source heating effect (< 0.3 °C) regulated by the Willamette River temperature TMDL. Upstream of the Clackamas River but downstream of the Santiam River, the 7dADM temperature changes are on the order of 1 °C, sometimes as high as 2.5 °C. Upstream of the Santiam River confluence, the effects of upstream dams on 7dADM temperatures in the Willamette River are on the order of 1.5 °C but sometimes as high as 3 to 4 °C. Clearly, the dams have an important effect on both streamflow and water temperature in the Willamette River and many of its major tributaries. Changes in dam operations have been used in recent years to greatly decrease those effects in the SF McKenzie River (Cougar Dam) and the North Santiam River (Detroit Dam). A quantification of the thermal effects of the dams both at the dam sites and downstream will be helpful as dam operators, resource managers, and regulators work to improve water quality and restore critical fisheries in the Willamette River basin.

First posted August 17, 2010