Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
9 Dams and Mitigation of Their Effects hi; _~ L INTRODUCTION Dam construction in the Pacific Northwest began late in the 1800s when small irrigation reservoirs were constructed on tributaries of the Snake River in Idaho. Early in the twentieth century, the first hydropower dams were con- structed on tributaries of the Columbia, such as the Spokane and Willamette Rivers. During the early 1900s, dam construction moved more rapidly and most of the reservoirs were relatively small (see Figures 3-9 and 3-10~. However, beginning in the late 1930s with the initiation of the construction of Bonneville and Grand Coulee, dam construction proceeded at a more rapid pace, as both the number and storage volume of dams in Washington, Oregon, and Idaho increased. During the 45-year period after the authorization of Bonneville (1933) and Grand Coulee (1935) dams, 14 mainstem Columbia River and 13 mainstem Snake River dams were completed. By the late 1970s, potential sites and public support for major new dams had been virtually exhausted, and the growth phase ended. Dams had been con- structed across the migration routes of most Pacific Northwest salmon runs. They range from irrigation diversions with a hydraulic head of only a few feet to dams at Grand Coulee, Dworshak, and Hells Canyon that are several hundred feet high and completely block upstream and downstream passage of anadromous fish. Adult salmon can pass high dams with the aid of trap-and-haul arrange- ments or even fishways, but the great depths, cross sections, and lengths of the reservoir pools might hinder smolts from finding routes to the sea. Figure 9-1 shows how the reservoir system has affected the average seasonal 226
227 ~- V ~ Ct o ~ o ~ Ct ~ _ 5 U: el~''P"i o . ~ O ~ O Lo O O O O O O O ~ O O O CO COW ~0 C' Ct (000~/SdO) MOSS NV3W (000~/SdO) MO1B NV3W ~ V ~ - c, Cal ~._ r ~ ' , it, ~ t: ~ O O O O O O O O O 0 0 ~ ~ S O O O O O ~ O ~ O .> CM ~ ~ C/2 ~ O ~ ~ ~ V (000~/S39) MO1d NV3W (000~/SBO) MORE NV3W ~ ~ ~ ~ s ~ C~ ~ o ;.L1 V 5: .o _
228 225 _ 1 _ can 175 _ Go 8 150 _ 200 175 Z 125 LL G I CO 100 75 50 _ 25 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST DRAINAGE AREA = 103,200 SQ. Ml. Observed Peak Flowl82,100CFS /1\ sol\ A alar`_, -A / ' / / ~ / \ I N/ \ l 1 ~/'\\1 Unregulated Peak Flow 240, 170 CFS \ \ ,~ FIGURE 9-2 1993 Seals River hydrograph below Lower Granite Dam. Source: Colum- bia River Water Report for 1993. discharge at four points in the Columbia River system: the Columbia River at the international border reflects the effect of Canadian storage, the sum of the Okanogan and Methow rivers reflects regulation by the Lake Okanogan and upstream reservoirs in Canada, the Snake River at its mouth reflects the effects of all Snake River reservoirs, and the Columbia River at The Dalles reflects the effects of all major storage facilities on the Columbia River system except those on the Willamette, Cowlitz, and Lewis rivers. The average seasonal discharge of the Columbia River mainstem has been drastically altered. However, the season- ality of regulated flow of the Snake River has been much less affected (Figure 9- 2~. That is an important distinction for the discussion that follows because of two common misperceptions: that there has been a major seasonal shift in the mean discharge hydrograph of the Snake River, which Figure 9-2 shows is clearly not the case; and that the reservoir storage in the Snake River is much less than on the mainstem of the Columbia total storage in the Snake River system, expressed as a fraction of the mean flow, actually is only slightly less than that of the mainstem Columbia. Because there has not been a major shift in the Snake River hydrograph, it is doubtful a priori that the declines in Snake River salmon stocks are due to or reversible by changes in the seasonality of the flow regime of the Snake River alone. These same salmon must traverse the Columbia River, whose seasonal hydrograph has been substantially altered. Even if flow changes are useful in rehabilitation efforts, they are likely to be insufficient without changes in other human interventions in the salmon's life cycle and habitat.
DAMS AND MITIGATION OF THEIR EFFECTS 229 The major difference between the Snake River and the Columbia mainstem is that much of the Snake River storage is used for irrigation rather than hydro- power generation. The amount of hydrograph shaping (change in the natural hydrography required to meet irrigation requirements is much less than that needed for hydropower because the water-demand peak is in midsummer, typically only 1-2 months after the natural hydrograph peak. However, in contrast with hydro- power, part of the water diverted for irrigation is consumptively used it does not return to the river. In the case of the Snake River, the total consumptive use of water by agriculture 4-5 million acre-feet (MAF) annually constitutes an ap- preciable fraction of the natural flow of the river during the months of highest agricultural demand (about 20% of the flow during the period May-September). That is the basis for the argument that agriculture has greater effects on the managed hydrology of the Snake River than does hydropower. Although the flow regime of regulated rivers usually is less variable over the course of the year than it was before dam construction, water storage in and release from dams can result in large day-to-day or even day-to-night fluctuations in flow and depth. The fluctuations can lead adult salmon to construct their nests in unsuitable places and can strand juveniles. However, intensive studies below Priest Rapids Dam during periods of peaking operation (large diurnal release variations) and load operation (minimal diurnal variations) revealed little or no effect on fall chinook spawning or abundance. In addition to affecting seasonal hydrographs, the reservoir system has had major effects on flow velocities, water chemistry (especially nitrogen supersatu- ration downstream of dams), and stream temperatures. Supersaturation with atmospheric gas, chiefly nitrogen, occurs when water is spilled over high dams. Gas is absorbed into the bloodstream of fish during respiration, especially fish that remain close to the surface. When the gas comes out of solution, bubbles form and can subject the fish to a condition similar to the bends suffered by divers. In some years before development of spill deflectors to prevent deep entrainment of spilled water, gas supersaturation caused extensive mortality (Ebel 19691. River managers now coordinate spill at various projects to reduce risk of serious losses. However, supersaturation can still exceed the high-risk levels (125~o saturation) in years of high river discharge. Reservoirs unintentionally provide thermal storage, as well as water storage, so seasonal variations in stream temperature are reduced in much the same way as seasonal variations in streamflow. In general, storage reservoirs tend to increase winter temperatures and reduce downstream summer temperatures and to cause maximum and minimum temperatures to occur later in the year than in the ab- sence of damming. However, water below Bonneville Dam has shown longer and warmer summer conditions over the last 40-50 years (Quinn and Adams, in prep.~. Figure 9-3 shows the trend in the date when the spring water temperature on the Columbia mainstem exceeded 15.5° since 1938. The upward trend in spring water temperature is consistent with introduction of storage in upstream
230 (a Aug 1~- LO . .. . . . .. ................ May t i I UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST l:5 a, a, x AS ~ Jun 1 ........ , ~.. ~_ ...~.... July - r - ~ . ~. ~. ~J ~ ....~............................. - ·'a_ ~a ~ ~ at_ . ........... '' ''hi-'- ~1 938 ~ 948 195;8 ~ 968 1978 ~ 988 Yew FIGURE 9-3 Spring water temperatures on Columbia mainstem since 1948. reservoirs, particularly Canadian storage, which came on line in the 1970s (see Appendix E'. Although the physical system has not changed since then, the trend apparently continues. That might reflect long-term climatic change or variabil- ity; Lettenmaier et al. (1994) found that air temperatures increased significantly in much of the northwestern United States during 1948-1988, especially in late winter and early spring. The effects of long-term changes in water temperature on salmon depend on a complex interaction of early rearing conditions, emer- gence date, and predator populations. High dams can inundate substantial amounts of spawning and rearing habi- tat. Some salmon, notably chinook, spawn in the mainstems of rivers and hence lose usable area when rivers become reservoirs (e.g., John Day, Priest Rapids, Coulee, and Wells dam pools). In addition, juveniles, particularly chinook, might rear in large rivers or feed there during downstream migration (e.g., the Fraser, Columbia, and Sacramento rivers) (Rich 1920, Levy and Northcote 1982, Chapman et al. 19941. The reservoirs might constitute a reduction in desirable habitat. In the Columbia River, however, reservoir rearing might substitute for lost flowing-river habitat and make up for lost quality and quantity of estuarine habitat (Rondorf et al. 1990, Chapman and Witty 1993~. The cumulative volume of all Pacific Northwest (Idaho, Oregon, Washington, and northern California) reservoir storage was over 50 MAF by 1980 and currently exceeds 65 MAF (Figure 3-101.
DAMS AND MITIGATION OF THEIR EFFECTS 231 The following discussion focuses on intermediate and large dams. Most smaller dams were built and are operated primarily to generate hydroelectric power. Their effects on salmon populations are discussed in Chapter 3. EFFECTS OF DAMS ON SALMON The effect of dams without fish-passage facilities on salmon is clear: the upstream habitat is lost. Such dams block about one-third of the Columbia River watershed to access by anadromous fishes; owing to natural passage barriers, one-third was never accessible. One-third lost to anadromous fish is upstream from Grand Coulee Dam on the Columbia River and the Hell's Canyon complex of dams on the Snake River (Chief Joseph Dam, the reregulating dam for Grand Coulee, also is impassable). Many dams on tributaries are also impassable, such as Mayfield Dam on the Cowlitz River and Round Butte Dam on the Deschutes River (tributaries to the Columbia), Detroit Dam on the North Santiam (tributary to the Willamette), and Dworshak Dam on the North Fork of the Clearwater (tributary to the Snake). The loss of spawning and rearing habitat because of impassable dams is perhaps most acute on the Columbia River system but is by no means restricted to it. Shasta Dam eliminated the upriver runs of salmon on the Sacramento River, and rivers in Puget Sound (e.g., the Skagit River) and the Strait of Juan de Fuca (the Elwha River) have impassable dams as well. On the coast, access to the upper Klamath and Rogue rivers was blocked by Iron Gate and Lost Creek dams, respectively. It is difficult to specify the magnitude of losses attributable to such dams because record-keeping before construction was often poor or nonexistent, and it might not be possible to survey the inundated habitat to estimate potential production. Not all the impassable dams were large hydroelectric dams. Small splash dams, built to back up water and then float logs downriver, were often impassable and sometimes remained in place long enough to obliterate major salmon runs, e.g., in western Oregon and Washington (Sedell and Luchessa 1982) and in the upper Adams River, British Columbia (Williams 19871. Relatively small irriga- tion and hydroelectric dams blocked some salmon migrations early in western development e.g., Black Canyon Dam on the Payette River in Idaho, Grangeville Dam on the south fork of the Clearwater River, and Sunbeam Dam on Idaho's Salmon River. More recently, a common practice at many fish hatch- eries has been to block upstream migration at or near the hatchery to aid in collecting returning adults or to isolate adults, possibly carrying diseases, from the hatchery's water supply. Dam-Related Mortality Even when dams are constructed with fish ladders for upstream passage of salmon, fish can still be delayed. Turbine discharge flows can disorient salmon
232 UPSTREAM: SALMON AND SOCIETY IN THE PA CIFJC NORTHWEST and make it difficult for them to find the small attraction flows that lead to the ladder. Ladder designs have evolved greatly since the early l900s. Early facili- ties often had excessive in-ladder flow rates and turbulence and lacked sufficient resting areas; salmon either avoided them or found them impassable or too de- manding of energy. Flood flows destroyed many wooden fish ladders, e.g., at Condit Dam on the White Salmon River and Grangeville Dam on the south fork of the Clearwater River. Poor concrete quality caused others to fail, such as Sunbeam Dam on the Salmon River. Spillways close to fishway exits tended to pull adults that left the ladders back over the dam and thus caused migration delay (Bjornn and Peery 1992J. Delays might not kill fish, but salmon do not feed on the upstream migration and must use stored energy as efficiently as possible to migrate upstream, mature sexually, and spawn successfully (Gilhousen 19801. Adult salmon can be killed if they drop back through turbine intakes, although the rate of loss of fish that drop back is unknown. Counts at successive dams seem to indicate that deaths occur between dams, although it is rare to observe dead salmon there. Poaching might account for some of the loss. Interdam losses have been estimated at up to 25% for the reach from Bonneville Dam to John Day Dam, but current loss estimates are about 4-5% per project there and elsewhere in the Columbia River system (Chapman et al. 19911. Downstream passage of juveniles through bypass facilities comes at a bio- logical price. These juveniles, hereafter referred to as guided fish, can make contact with deflection-screen surfaces, gatewell walls, the vertical barrier screens in gatewells, the orifice entrance, or portions of the bypass channel or downwell (Figure 9-41. Such encounters can cause impingement, bruising, scale loss (descaling), and stress (Chapman et al. 1991~. Because it is the most quantifiable evidence of damage to fishes, descaling is evaluated by biologists quantitatively and is used to indicate facility problems and fish viability. Stress accompanies passage through bypasses. Fish that hold in currents to resist passing downstream, that contact separators, or that otherwise experience stressful contact (e.g., with equipment) become physiologically stressed. They appear to recover when held for several to 48 hours. However, direct bypass to the river delivers stressed fish to the outfall,1 where they can become prey for birds and fish, especially northern squawfish (Ptychocheilus oregonensis). Bypass systems also concentrate smells in a small area. Smolts from the whole width of the Columbia and Snake rivers (which might flow at 250 and 90 thousand cubic feet per second [kcfs], respectively) for example, are gathered into a bypass channel that flows at only 500 cubic feet per second (cfs). If bypassed directly, many thousands of smelts per hour can be delivered in a small volume of water to the dam tailrace, which provides a concentrated stream of 1The bypass outfall, the water just delivered from the bypass system to the river, delivers fish to the tailrace or the tailrace edge; the tailrace is the entire river flow just downstream from the dam.
DAMS AND MITIGATION OF THEIR EFFECTS ,, C. ~ g.' i` q Gatewell , Bypass channel Vertical barrier screen - ~ ~ :~ tom Am\ {so i portion at top ~ Operating gate- ~ , ~ I '`: ,, -~'=~\ \\ fully raised) _ o I 'gay\ 7 c ~ ,~G~ SHOWS: ~ ~"< ~ °D,>oW ~ , ~: Submersible ~ ~ ~ traveling screen) Fyke nets > ~ I it\ ~ l ~ . FLO\lv \\ ~ ~,`oj~: v.~-` ~ I'm C~ TV c';~¢ C~5 ~ I, a., I: c, . .. ~ - \~ ~ 0 ~ Scat ~1 ~1 ., . t4Og5;` ~°e (~` ~ , ~Ott,; .~ c, ~°~r~ ^~ G ~^- :_. (' ~ _. ~ _ I, 233 ll FIGURE 9-4 Cross section of typical dam and bypass system Fyke nets are used to estimate fish-guidance efficiencies Source: adapted from USCOE 1993:163. prey for predators. At the Bonneville second powerhouse, extensive studies of fall chinook passing through turbines and the bypass revealed that the survivals through the two routes were not very different and that predators were keying on the stream of prey from the bypass outfall (Ledgerwood et al. 1991~. Similar studies have begun at the Bonneville first powerhouse. At other dams equipped with bypasses, total bypass-related mortality has not been thoroughly investi gated. Some investigators consider bypass-caused deaths to include only those which can be observed (carcasses) in the raceways and sampling facilities incor- porated in bypass systems. But those do not provide data on impingement on deflection screens, predation within the gatewell and bypass system, predation caused by the bypass-concentrated stream of prey, stress-related deaths that occur
234 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST after smells leave the outfall area, or predation on stressed fish long after they leave the outfall pipe. Carcass counts typically indicate a mortality of 1% in the bypass system (Koski et al. 19851; some studies indicate that bypass-related mortality averages perhaps 5-7% (Matthews et al. 1987~. The location of the outfall is crucial to bypass-related mortality. Expanded evaluations of bypass-related mortality include the first smolt encounters with deflection screens to a point well downstream from the dam, where fish become free of all physical effects of the bypass. Passage through the bypass system is not immediate in most cases. The National Marine Fisheries Service annually evaluates how quickly smelts pass out of the gatewell and through the orifice on the basis of the fraction of smelts that enter the gatewell and leave it within 24 hours. Orifice-passage efficiency of 70~o is considered satisfactory; this means that some smelts remain well over 24 h, some are delayed in the bypass channel, and others are delayed beneath the separator (if the system incorporates one). Seasonal races and individual populations of salmon migrate from rivers to the ocean at specific times of the year. It is generally believed that photoperiod is the primary cue that triggers migration, although flow, temperature, and social interactions can also influence it (Godin 1982~. The rate at which smells migrate downstream depends on both their swimming speed and their orientation to the velocity of the flow. Smolts migrated downstream in the Columbia River more rapidly before dams were constructed than they do now (Raymond 1979, Rieman et al. 1991~. It also appears that migration is more rapid in years or times of the year when river flow is greater. However, there is dispute over the validity and interpretation of these data, as discussed in the next section. There is some evidence that predators, such as northern squawfish, have increased in abundance in the lower Columbia River. The most important fish predators (squawfish) and birds, such as gulls, are native species. There is some predation by nonnative fishes such as walleye (Stizostedion vitreum) and small- mouth bass (Micropterus dolomieui). Perhaps more important, the reservoirs, tailraces, and bypass outfalls might have improved the river as rearing habitat for these species, and the tailraces and forebays of the dams might lead to an increase in foraging efficiency over that in the undammed river. In addition to possible increases in mortality en route to the ocean that might result from retarded migra- tion, delayed arrival in the estuary or ocean might result in higher mortality or reduced growth. Storage in the upper Columbia River and Snake River has altered the main Columbia River's hydrograph. Sherwood et al. (1990) analyzed monthly mean flows of the Columbia River and found that large-scale manipulation of the flow cycle began around 1969. Since then, monthly mean flow has varied less. Flow damping has resulted in a reduction in average sediment supply to the estuary. Except for times of major floods, residence time of water in the estuary has increased with decreasing salinity. Detritus and nutrient residence has increased; vertical mixing has decreased. Sherwood et al. ( 1990) noted that although hydro
DAMS AND MITIGATION OF THEIR EFFECTS 235 dynamic changes have probably enhanced the pelagic primary productivity in the estuary, the costs have yet to be evaluated. The changes have enhanced estuarine conditions for detritivorous epibenthic and pelagic copepods; the estuary has been converted to a less-energetic microdetritus-based ecosystem with higher organic sedimentation rates. Sherwood et al. (1990) concluded that it is apparent that these changes, and other changes in the fluvial part of the system, have contributed to the dramatic decline in salmon populations. The implications such as have taken place in the Columbia River estuary and water- shed need to be incorporated into contemporary estuarine and shorelands man- agement strategies. In particular, proposals for comprehensive hydroelectric and water withdrawal developments, shoreline modifications, and navigation projects should all be evaluated in terms of potential consequences to the estua- rine ecosystem and resulting effects on other resources, including fisheries, which depend on a highly convolved and biologically diverse estuanne environ- ment. Sherwood et al. (1990) estimated that the estuary of the Columbia River lost 20,000 acres of tidal swamps, 10,000 acres of tidal marshes, and 3,000 acres of tidal flats between 1870 and 1970. They further estimated an 80~o reduction in emergent vegetation production and a l5% decline in benthic algal production. Ebbesmeyer and Tangborn (1993) demonstrated that reservoir storage in the Columbia River had altered the hydrograph by diverting summer flows to winter, which altered coastal sea-surface salinities from California to Alaska. Coastal ocean and estuarine dynamics have changed at various locations along 2,000 km of North Pacific shoreline. The effects of those alterations on trophic dynamics, loss to predators, and migration success are completely unknown. Ebbesmeyer and Tangborn (1993) raised the question of whether homing behavior of salmon might change in response to the altered salinities. If homing were affected, we would expect to see steelhead and salmon from Columbia River hatcheries as strays in coastal rivers of California, Oregon, and Washington. But no evidence of extensive or unusual straying has been found. Time of Travel The effect of time of travel on subyearling chinook (i.e., ones that pass their first winter in the sea, or ocean-type chinook) is less clear than the effects of passing through dams. Ocean-type chinook pass their first winter of life (after emerging from the read) at sea; stream-type chinook spend their. first winter of life in the stream before going to sea. Some studies indicate that subyearlings migrate downstream more quickly with higher flows (Rondorf and Miller 1994), but other studies (e.g., Giorgi et al. 1990, Chapman et al. 1994) do not. Sub- yearling chinook gradually move downstream as they grow-a rearing migration. Rather than the rapid downstream passage of yearling chinook and steelhead, which often reach an average speed of 20 miles/day or more, downstream pas
236 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST sage of subyearling chinook is perhaps 2 miles/day. Some subyearling summer- fall chinook do not enter the sea in the first summer of life, as it is commonly supposed; rather, they do not pass McNary Dam until late fall. Growth rates of subyearlings in mainstem Columbia River reservoirs are very high (Chapman et al. 19943. No information exists on whether subyearlings that reach large size in reservoirs of the Columbia River have higher survival rates than small sub- yearlings that go to sea in July. The effect of migration speed on smolt survival has been inferred from data on system survival and travel-time data acquired during the 1970s. The accuracy, precision, and relevance of those historical estimates are questionable (Giorgi 1993), and for that reason, the data from the 1970s were abandoned in the early 1980s. Smolt survival through the same river reaches would differ today: the smolt cohort in the Snake River has changed from almost 70% wild in the late 1960s to only limo wild today (Park 1993~. Hatchery steelhead and spring chinook predominate in the main Columbia and Willamette rivers, and spring- summer chinook in the Snake River. Wild fish still compose the majority of fall- run and summer-run chinook of the mid-Columbia and in some tributaries that lie in wilderness areas where hatcheries are not used (e.g., the middle fork of the Salmon River and Chamberlain Creek in Idaho) or are considered refugia for wild gene pools (John Day River in Oregon). Not only the makeup of the smolt runs, but the river environment itself has been altered. More turbines have been installed in mainstem dams. John Day and other newer reservoirs have matured limnologically. Fish communities have changed (e.g., exotic species have been introduced), and river management has evolved in response to needs for fish conservation. Reach-specific or system survival studies have not been completed that would allow managers to evaluate modern conditions in the river; they are crucially important and should be pur- sued with vigor and dispatch. Recently, there has been sponsorship of new research, such as reach-specific survival studies that began in 1993 (Snake River Salmon Recovery Team 1993~. If such studies are continued, they would provide steady improvement in the scientific knowledge available for the difficult challenges presented by salmon rehabilitation. For instance, Williams and Matthews (1994) and Steward (1994) critically reviewed the data obtained in the 1970s. They concluded that system and average project mortalities in smelts migrating downstream were overesti- mated and that the data should not be used to estimate current system and project losses. The lack of modern survival data that could be used to evaluate survival through various dam-passage routes and through reservoirs has resulted in dis- sension over the value of high flows for reducing mortality in downstream mi- grants. Many investigators believe that the older data support a need to provide flows of 80-90 kcfs in the Snake River and 200-220 kcfs in the main Columbia. However, many of the same researchers question the gains in survival that the
DAMS AND MITIGATION OF THEIR EFFECTS 237 Fish Passage Center and Columbia Basin Fish and Wildlife Authority indicate will result from providing flows higher than that. Survival studies with passive integrated transponder (PIT) tags in 1993 and 1994 indicate survival near 100% through Lower Granite pool and high survival through Little Goose pool. However, estimated mortality "across the concrete" was higher than expected (Iwamoto et al. 1994, Schiewe 19941. Data on interro- gation rates of PIT-tagged fish that had been trapped and marked in the lower Salmon and Clearwater rivers also support high survival to the first (Lower Granite) dam (data tabulated in Fish Passage Center 19941. Those preliminary results, if confirmed by additional analysis and studies and augmented by reach- specific survival estimates at downriver dams on the Snake and Columbia rivers, would suggest high priority for mitigation efforts directed at increasing survival at the dam rather than speeding fish through pools. If later work shows that the 1993 and 1994 pool-survival estimates were atypically high, drawdown or flow augmentation could be given higher priority. The committee strongly supports acquisition of additional reach-specific survival information at Snake and Co- lumbia river projects. Unscreened Diversion Dams When water is diverted from the river into domestic or agricultural uses, there is always the possibility that downstream-migrating salmon and other fish will be drawn into the diversion channel or pipe or that fish will actively seek it. Many unscreened or inadequately screened diversions have operated on the Co- lumbia River system over the years, and some unscreened or poorly screened diversions remain. An extensive program, funded by BPA, has added new screens and maintained old ones in irrigation-diversion ditches and canals in Columbia River tributaries. No study has balanced the net benefit of screens and compari- son to the lost rearing habitat in irrigation canals and ditches downstream from the screens. Estuarine Dynamics The estuaries of West Coast rivers, both dammed and undammed, might have been changed to the detriment of salmon. Reservoir storage in the upper Columbia and Snake rivers has altered both the seasonal pattern and the charac- teristics of extremes of freshwater entering the estuary, as mentioned above. Since large-scale regulation of the flow cycle began about 1969, average sedi- ment supply to the estuary, which is governed by floods, has been greatly reduced (Simenstad et al. 19921. Although changes in discharge have undoubtedly af- fected estuarine dynamics, many other factors have played roles as well.
238 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST MITIGATION OF DAMS' EFFECTS ON SALMON Before 1980, the Columbia River dams and reservoirs were operated for the primary purposes of irrigation, flood protection, hydropower, navigation, and recreation. As a practical matter, irrigation, flood protection, navigation, and recreation were reasonably compatible with hydropower, and hydropower con- siderations dominated the operating strategy for the system. The Northwest Power Act of 1980 made fishery protection and enhancement one of the key objectives of reservoir-system operations in principle, a major change in oper- ating strategy. In this section, seven measures for mitigating dams' effects on salmon are discussed: fish-passage facilities, predator control, transportation, spill, flow augmentation, reservoir drawdown, and dam removal. Fish-Passage Facilities Fish ladders to pass salmon upstream of dams were inadequate or absent in the early 1900s. As knowledge developed about fish behavior, response to attrac- tion flows, and in-ladder hydraulics, designs improved. Fish facilities at Rock Island Dam (1933), Bonneville Dam (1938), and McNary Dam (1953) greatly improved on the early designs, although instances of fishes' being delayed by hydropower dams and falling back after passage have been noted (Bjornn and Peery 1992, Dauble and Mueller 1993~. Studies continue on adult passage effi- ciency, in-ladder behavior, fallback, and interdam losses. On the Columbia River, the first mainstem dams constructed were equipped with fishways that permitted only adult passage. In particular, Rock Island Dam (near Wenatchee) and Bonneville Dam were equipped with fishways when built. Most mainstem Columbia River dams Bonneville, The Dalles, John Day, McNary, Priest Rapids, Wanapum, Rock Island, Rocky Reach, and Wells- incorporate fishways. Grand Coulee and Chief Joseph dams do not. The four mainstem Snake River dams-Ice Harbor, Lower Monumental, Little Goose, and Lower Granite also have adult fishways. Juvenile passage facilities on most mainstem dams in the Columbia River system use deflection screens that project downward into the intakes of turbines and deflect fish upward from the turbine intake into the gatewell (Figure 9-41. Some deflection screens have traveling screen surfaces and are called submers- ible traveling screens. Recent designs involve bar screens, which are usually constructed of a wedgewire material that is triangular in cross section and pre- sents a flat surface to incident flow. Bar screens are considered most useful where debris and drifting vegetation are less abundant; traveling screens serve best where debris and vegetation would clog a stationary screen. Traveling screens need not operate continuously. Periodic movement of screens can keep them clean and preserve hydraulic configurations in the turbine-intake environ
DAMS AND MITIGATION OF THEIR EFFECTS 239 ment important for fish guidance. Periodic cycling also reduces wear and main- tenance requirements. Deflection-screen designs are based on the premise that downstream-migrat- ing juvenile salmon tend to pass through the turbine intake high in the water column. The screens guide intercepted fish upward into the gatewell. The gatewell's primary purpose is to provide a slot for insertion of a bulkhead to prevent water from passing farther down the turbine intake. Water-surface eleva- tion in the gatewell is the same as that in the reservoir pool. The degree to which deflection screens guide fish up into the gatewell is termed fish-guidance effi- ciency (FGE). FGE is estimated by dividing the number of fish deflected upward into the gatewell by the sum of that number and the number of deflected fish and smelts trapped in fyke nets arrayed below the deflection screen. For example, if the number of fish deflected upward into the gatewell is the same as the number trapped in fyke nets, the FGE is 50~o. FGEs of 70% or greater are considered "acceptable" with agencies, utilities, etc., in discussions about fish passage fa- cilities at dams. Hydroacoustics are often used in connection with FGE studies to determine behavior of fish that encounter or avoid screens in turbine intakes. A vertical flow is developed in the gatewell by permitting water to escape through a vertical barrier screen (VBS). The VBS prevents smolts from leaving the gatewell with the flow that leaves it. Several feet from the water surface within the gatewell, an orifice, usually about 14 in. in diameter, leads to a bypass channel that passes longitudinally through the dam structure normal to river flow. Makeup water (water to provide a reasonable gallery flow at the upstream end of the bypass gallery, where few orifices have had an opportunity to feed the flow) and water from the orifices provide a substantial flow down the bypass channel. Each turbine in the dam has one or two orifices that permit fish to escape to the bypass channel. Flow through each orifice amounts to 9-12 cfs. Thus, by the time the bypass channel contains all orifice flow and initial makeup water, the flow in the channel can reach perhaps 500 cfs or more. The bypass channel lies only a few feet below the reservoir elevation. It carries water by a downwell to near-tailrace elevation. Different types of down- well designs have been used. Some lead to raceways where fish are held for later transportation by barge or truck. Others lead to the river for direct bypass of smells without transportation. Most incorporate fish sampling facilities between the downwell and the raceway or bypass outfall. Where fish are held in raceways for later transport by barge, they must be separated from the large flow produced in the bypass channel and downwell; this is usually accomplished by a separator that consists of screening or wedgewire. The downstream-migrant facilities described above have been installed or are being installed on Bonneville, The Dalles, John Day, McNary, and the four Snake River dams operated by the Corps of Engineers. On the Columbia River upstream from the mouth of the Snake River, where five dams are operated by public utility districts, the only dam with an operating fish bypass system is Wells
240 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST Dam, which, unlike others on the Snake and Columbia rivers, is a hydrocombine design that has the spillways directly over the turbine intakes and therefore can- not have conventional deflection screens in turbine intakes. Wells Dam has a retrofitted spill bypass that encourages juveniles to leave turbine intakes upward and pass downstream in a modest amount of spill from the spill bay. The remaining middle Columbia dams (Rocky Reach, Rock Island, Wanapum, and Priest Rapids) were built without bypass facilities. For the last 12 years, various studies have been under way to facilitate design of juvenile by- passes. Most options involve some form of deflection screen. The FGEs ob- tained at various dams with fyke-net studies have been variable, ranging from less than 10% to more than 90%, depending on the dam and species. FGEs with the Priest Rapids and Wanapum prototype screens have been high enough to encourage further work on design and installation. FGEs at Rocky Reach have been very low. Turbine intakes there lie in a cul-de-sac that has the axis of the turbine-intake mouths parallel to the main flow of the Columbia River, rather than normal to flow. A somewhat similar orientation has prevented prompt application of screening technology to The Dalles Dam on the lower Columbia River. The FGEs obtained at dams with deflection screens differ considerably. They have been low at Bonneville Dam, whose bypass facilities were incorpo- rated in the design rather than being retrofitted, and range from at least 70~o at John Day and McNary dams to 55% at Lower Granite Dam. Predator Control The dynamics of the interaction between squawfish and smelts has received relatively little attention except in the pool of John Day Dam (see Raymond 1979, Rieman et al. 1991), where squawfish predation was estimated to kill millions of downstream migrants each year. Rieman et al. (1991) estimated that 8%, 11%, 7%, 19%, and 61%2 of salmon that entered John Day pool were killed in the months of April through August respectively; steelhead loss averaged 12% per month in April, May, and June. Those estimates provided a platform for launch- ing a major effort to control squawfish in the Columbia River basin. The pro- gram includes bounties and intensive efforts to reduce predator densities. It was designed to reduce the population of large squawfish and is to continue indefi- nitely. Assumed gains from such control measures are incorporated into smolt- survival models. Transportation At Lower Granite, Little Goose, Lower Monumental, and McNary dams, raceways can hold downstream migrants for delivery to barges or trucks for 2Because rounding was used, the sum of the percentages is 1065to.
DAMS AND MITIGATION OF THEIR EFFECTS 241 transport downstream. The transportation concept relies on "leapfrogging" down- river reservoirs and dams to deliver the juveniles to some point downstream from hydropower projects. In the Columbia River system, transportation programs deliver juveniles to the Bonneville Dam tailrace, largely by barge. Barges control water quality, e.g., by preventing the occurrence of gas supersaturation. Trans- port conditions can also be monitored in barge containers. In addition, barges can deliver smalls to varied release points and at varied times to confound predators in the lower Columbia River. Data on this point are limited but intriguing. Solazzi et al. (1991) reported that survival to adult capture of hatchery-produced coho smolts was increased by about 50% when smelts were delivered to Tongue Point, near the mouth of the Columbia River, instead of to the Bonneville Dam tailrace, about 185 km upstream. Other work (e.g., Ledgerwood et al. 1991) showed that based on survival from subyearlings to adults, midriver release at the Bonneville tailrace was superior to shoreline release. Experiments with steelhead have begun. Chinook, both yearling and subyearling, should also be tested for time and place of optimum survival after release from barges. However, effects of transportation on homing must be closely monitored (e.g., Solazzi et al. 1991~. State agencies and American Indian tribes currently oppose barging. Some of the objections center around the unnatural migration conditions (see Chapter 5~. The efficacy of barging has been questioned because Snake River runs declined while barging was being used as a major mitigation tool. However, the committee notes that many Pacific Northwest runs that have not been barged have also declined. Recent transport-benefit ratios (TBRs) of observed survivals of transported compared with observed survivals of inriver migrants have been 1.6: 1.0 in favor of transported fish (1986) and 2.5: 1.0 (1989) (Mundy et al. 1994) for yearling chinook at Lower Granite Dam on the Snake River. Transport of yearlings at McNary Dam has yielded a ratio of 1.55: 1.00 in favor of transported fish (Townsend and Skalski 1994~. For subyearling (ocean-type) chinook, stud- ies at McNary Dam indicate much higher transport-benefit ratios (often over 3.0: 1.0) (Park 19931. Transportation might increase disease in transported smelts, but the evidence to date is equivocal (Pascho and Elliott 1989, Pascho et al. 1993~. However, the transport-benefit ratios indicated by studies in the 1980s suggest that even if disease transfer occurs, the effects are not sufficient to erase the survival advantage received by transported fish, which avoid passing through turbines at hydroelectric projects. The committee relied on the information avail- able on survival of transported fish in relation to in-river migrants and agrees with the Snake River Salmon Recovery Team (1993) that transportation is the mitigation tool of choice until inriver migration shows higher survival rates than does barging fish. Spill For fish-passage purposes, "spill" is water deliberately discharged over dam
242 UPSTREAM: SaLMON AND SOCIETY IN THE PACIFIC NORTHWEST spillways, rather than discharged through turbines or bypass facilities, with the intent of reducing turbine mortality. Spill is an option that has been considered to improve the survival of juveniles passing dams on the lower Snake and lower and middle Columbia rivers. Schoeneman et al. (1961) found that mortality in chinook that were spilled over McNary Dam (Columbia River) and Big Cliff Dam (Santiam River) was less than Do. Other studies have indicated similar low loss rates associated with spill. Like bypass, spill can have a price in terms of gas supersaturation. Gas becomes trapped in bubbles in spilled water that plunges deeply into the stilling basin (the basin of relatively still water where gas bubbles can disperse) down- stream from the dam (Ebel and Raymond 1976~. Air within the plunging water is dissolved at depth because of increased pressure. It later escapes from the water when turbulence causes the gas to be released into the atmosphere. However, turbulence downstream from tailraces is scant because most dams are close to the pool of the next dam downstream, so supersaturation can persist some distance downstream of the spillway. "Flip-lips" (devices that direct spilled water outward to prevent deep plunging of spill flows, which cause supersaturation) have been installed on most dams. However, when spill is great, the lips become largely useless. An independent panel of experts appointed by the National Marine Fisheries Service (NMFS) on dissolved-gas effects recommended that gas super- saturation be limited to 110% (NMFS Independent Panel on Gas Bubble Disease 1994~. Even so, gas supersaturation is often greater than 115% and sometimes as high as 140~o (Fish Passage Center weekly reports) below major dams on the Columbia River and Snake River where spill occurs. Gas supersaturation of up to 143% killed many adult salmon in 1968, when all flow was passed through the completed spillway during construction of John Day Dam (Ebel et al. 19751. An emergency at John Day Dam on May 29, 1990, caused all flow to be spilled again, and water in the tailrace was supersaturated up to 145% (Chapman et al. 1991~. As a result, dissolved-gas saturation at The Dalles Dam (the next dam downstream of John Day) remained greater than 130Yo from May 30 through June 2, 1990 (Fish Passage Center weekly reports 1990), and might have been even higher in some parts of The Dalles pool. Spill is thought to reduce small mortality compared with turbine passage as they migrate across dams. However, because of the way it is measured-from the head of the pool to the dam- and because it is not thought to speed fish through reservoirs, it does not affect reservoir mortality. Flow speed, which often directly corresponds with flow volume, can speed the fish through the reservoir, assuming that they are physiologically ready to move downstream. Thus, spill differs from transportation, which can reduce both dam and reservoir mortality of smelts by carrying them around downstream reservoirs.
DAMS AND MITIGATION OF THEIR EFFECTS Flow Augmentation 243 Of the major effects of the dams on fish, the increase in time of passage through the hydropower projects on the middle Columbia and lower Snake rivers, particularly in the spring, has been identified as a key obstacle to survival of juvenile salmon and steelhead. An attempt to reduce passage time was made through a "water budget," which allocated some upstream storage in the Snake (at Dworshak Reservoir) and in the mainstem Columbia (at Lake Roosevelt, the reservoir of Grand Coulee Dam) to increase spring flows. It was anticipated that most of the additional water released would pass quickly through the downstream run-of-the-river dams (hydropower dams that primarily use riverflow rather than steed water for power generation; see Appendix F) and therefore would not be available for the generation of power during the peak-demand season. However, the excess power produced by the increased flows at the time of the water-budget releases, although of less value, could be sold to meet demands elsewhere (e.g., in California, by use of the direct-current intertie); this compensated for some extent for the loss due to the water-budget releases (Wood 1993) (see Box 9-11. The purpose of flow augmentation is to reduce the travel time through the reservoir system at key times in the salmon life cycle (most proposals are for the period April 15-June 14, although some proposals would augment summer dis- charge as well) to approximate more closely the pre-dam travel times. Flow augmentation was originally termed the "water budget" by the Northwest Power Planning Council (NPPC) in the middle 1980s. The idea was to reserve enough storage at Dworshak Dam's reservoir and Lake Roosevelt (Grand Coulee Dam's reservoir) to meet a flow target of 85 kcfs in the Snake River at Lower Granite in May (about 53% of the mean May flow under pre-dam conditions and about 78~o of mean May flow with regulation; see Chapman et al. 1991) and to meet a target of 134 kcfs in the Columbia at Priest Rapids (about 87~o of the mean May flow for the period 1971-1993, after further alteration of the natural hydrograph by Canadian storages. The projects on the lower Snake and middle and lower Columbia rivers were to operate near "full pool" (the maximum reservoir water level authorized in the projects' design). The storage required to meet the streamflow targets was determined by adding the positive differences between the target monthly flows and the critical- period-rule curve for hydropower targets. This resulted initially in a storage requirement of about 4 MAF. NPPC later determined that sufficient unallocated storage capacity was not available on the Snake; of the total reservoir storage of 11.7 MAF on the Snake, only about 2 MAF at Dworshak Dam is federally controlled and thus potentially available. Therefore, the water-budget storage was increased on the Columbia and decreased on the Snake so that the total was 4.65 MAF. The water budget was to be released during the peak small outmigration period (April 15-June 14~.
244 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST , ,,,, ,, - . ~ ''X"""""'' . ~t 0~= of F~h an~ WE tnve~men~ ....................................... .................................................. ::::::::.::..:::.:::::'..:::..~...~::::! ~:::':~:':'~:':~r''~:'::~n':~:~::'~::~::~:'::~::':~:':::::: . ~n ~ . ~. .................................. ................................................................................................................. .............. . ........... .. ... .... .. :.:::::::::::::::::::.: ~.:::. ~': :':b . ~. ~I .~.,...,..~.,,.,..~n..,... :~ ~ - ~ ~ ~ 00!~':':'~::~t' ' 'A - ~t ~ . ~. ~ ~ ~ ,,....,. , ~ , ....................................................................................................................... . ~ ....................... = . .... ...... ..... ... . .... ... ..... . ... .... . ~ ~ ~ fit| ~ ~e| I| ~ - - be|| en| ~ alley ::::::::::::::::::::::: ::: :::~:~:~:~::~::~:::~:~:::~:::~:::~::::~::::~:~::~m :::::: .......... ~r :: ~:::~:::~::~:: i : ~ an:: \~ ~n ~ - ~ or ~ a . ~. ~. ~. ~. ~d - A:::: . : ::::. ~ :: In.. I. ~ :::::::::: ::::: . ~. ~. ~. ~.~. ::::::::::::::::::::::::: :::::::~.::::::::::::::::..::::::~:: :: ::: ::: in :: ................ ~ En ~ 0~ ~ WOK ~ ~ - ~ num m ( 1 N ~ - §~ t~ l ~ ~OO ............................................................. ~ ' '.' ' ' '"' - "'~'e~''S0~'NP~"~'' " - '~"'~''T~'0"'"'' . ~, ~ ~t =~.~.n - m ~ ~ E~ p - m . . ~- '-a ~ - ~S ~ .,.,.,.,.,.,.,.,., , . ~. ~, ~.,.~., ~., ~, ~.,.,~,.,. ~. ............. ................................... ... ............................................... ............................................................................................... · a~ ~-1~ ~ .. ...... . . ~. . : . .:: . : x i - - .~ ~ s a ~s ~..~....~ B^~n S8~"~'~ $ ~ ...................................................... . . ..~ ..~. _ ~ ~ o.~ - '' SX'''' '"'pi--'"" - ' ~ - than Mod ~ F~ - ~ ~O ~ i ~ v ~ ~E a-=~e~s,. Ohm me ....................... , .,, . = ............................................................................................................................................................. ............................. ~ ~ Is =~ .. ~ ~ ~ - ~d ..................................... ....................... ................................................................ .... . ~. ~t l~ se O.T a~ ~ mod.= ~S ~ h~ ......................... .............................................................................................................. ....... ::::::::::: =. ~ ~ ..~...~. - ::.~::::~::~:'::::~::::~:':~::~:::~::::~:::~":'1::~ ':::::::::::: I::::::::::::: . ~., ~. Id ........... ... . Wood (1993) argued that the water budget has failed for four reasons. First, the base power flows, which the water budget was designed to augment, are not necessarily maintained during low-flow years on the Snake. Second, the man- agement agencies, notably the U.S. Army Corps of Engineers, view the water budget as "a cooperative arrangement," not a mandate, so water-budget volumes are not guaranteed, especially in low-flow years, when they are most needed.
DAMS AND MITIGATION OF THEIR EFFECTS 245 Third, other operating considerations have higher priority, e.g., for secondary power release and in some cases refill of downstream reservoirs. Fourth, BPA mitigates the cost of the water budget by selling the resulting excess spring power, then buying it back in the summer from other sources, thus causing
246 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST abnormally low summer flows in the months after the period of spring enhance ment. More recently, the interagency Systems Operation Review (1992) consid- ered a base-case flow-augmentation alternative, in which Dworshak Dam would provide at least an additional 0.30 MAF release in the spring and 0.47 MAF release in addition to 0.43 MAF release from the Upper Snake River Dam (note that these are releases, and not storage allocations). In addition, up to 3 MAF of spring flow augmentation would occur on the Columbia; this would be achieved in part through winter flood-control shifts from the Snake River dams to Grand Coulee Dam in some low-flow years. The effectiveness of flow-augmentation alternatives has not been demon- strated. On the basis of modeling studies that evaluated the effectiveness of several flow-augmentation alternatives for improvement of juvenile survival, the Systems Operation Review (1992) found that "flow augmentation alternatives produce similar results in all three drainages (Snake, Lower Columbia, Upper Columbia), providing negligible survival benefits for inriver migrants." Flow augmentation might be useful if it provides sufficient water to reduce dam-related mortality (e.g., by spill). However, it is unable to reduce the water-particle travel times through the pools in average-flow years by more than a few days prob- ably biologically insignificant beyond the levels already achieved by NPPC's 85-kcfs Lower Granite Dam target. It might well be important that the system operating policy treats the NPPC "targets" of 85 kcfs at Lower Granite Dam and 134 kcfs at Priest Rapids Dam as operating constraints, rather than operating targets; i.e., these targets are given precedence over power production. Flow augmentation should be implemented in such a way that targets are met in all years when storage is available, not just in average and above-average years. Merely focusing on average years is insufficient, as was shown in the years after 1986. In some of those years, spring flows were very low; for example, average May discharge at Lower Granite Dam in 1990 was only 68.2 kcfs. Before Snake River sockeye and chinook salmon were listed under the Endangered Species Act (ESA), the Fish Passage Center allocated the water budget on the basis of num- bers of fish passing through the system. It concentrated mitigative efforts on hatchery fish, which tend to move in a relatively narrow period. Wild fish from some tributaries move through the Snake River over a much longer period- some from as early as mid-April, others as late as the end of June (Chapman et al. 1991). Reservoir Drawdown A different approach to increasing survival has been suggested in response to the ESA listing of the Snake River sockeye and spring-summer and fall chinook and the general perception among fishery managers that the water budget has failed to stop the decline of Snake River salmon runs. Although the water-budget
DAMS AND MITIGATION OF THEIR EFFECTS 247 approach can reduce the time of travel through the reservoir system (in particular, the lower Snake and middle Columbia River reservoirs, all of which are run-of- the-river Esee Appendix Fly, the effect is at best modest a few days at mean springtime flows. Average travel time through a reservoir is roughly equal to discharge (volume per unit time) divided by the average cross-sectional area. Therefore, drawdown would reduce travel times by lowering reservoir levels (subject to control of dissolved gases; see the discussion of spill) to decrease the cross-sectional area, hence increasing the water-particle velocities through the run-of-the-river reservoirs. As an extreme case, drawdown could be made to the pre-dam or original river channel. For example, a drawdown of zero corresponds to full operating pool (essen- tially the present operating condition), and, for a particular dam, 110 ft of draw- down corresponds to the undammed river channel. Harza and Associates (1994), in reviewing a range of drawdown options considered by the interagency Systems Operation Review (BPA et al. 1992), recommended that only three be considered in detail: natural river, deep drawdown to some level below minimum operating pool, and maximum drawdown (to spill crest). Harza found that for the maximum drawdown option [to the] natural river ~channel], the maxi mum travel time advantage (that is, the decrease in travel time through the Lower Snake River reservoir system for the drawn down pools relative to travel time with present pool levels) is 7 days at 100 kcfs, 14 days at 50 kcfs, and 29 days at 25 kcfs. The latter condition (25 kcfs flow) applies less than 2% of the time in the spring migration season (Apnl 15 - June 15), although it is not clear whether this figure includes the effect of the water budget. The System Operation Review found that the natural river options "decrease travel time (from the mouth of the Salmon River to Bonneville Dam) between 7- 10 days, depending on the stock." However, effects are limited to Snake River populations since the action is restricted to the Snake River Basin. Most of the studies, including Harza's (1994), that have considered a draw- down option have concluded that drawdown to any elevation greater than natural river level is not likely to have biologically significant benefits, in comparison with other alternatives for improving juvenile survival (such as flow augmenta- tion and barging). However, reduction of the reservoir only to minimum operat- ing pool would allow hydropower turbines to continue to operate and would not require reconstruction of fish-passage facilities, such as fish ladders. In addition to the economic costs associated with elimination of hydropower production, navigation, and recreation during the natural-river drawdown peri- ods, a number of complications accompany drawdown. These include at least the following: · Loss of spawning habitat for nonsalmon in the reservoir pools and tribu- taries that directly enter the pools.
248 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST Concentration of predators in the relatively small channel volume. · Necessity to reconstruct passage facilities for use during the drawdown and refill period, which would otherwise be unusable at reservoir levels below . . . minimum operating pool. · Increased mortality due to operation of turbines at less than maximum efficiency during the drawdown and refill periods. · Loss of rearing habitat for subyearling fall chinook. Direct economic costs are also associated with reconstruction of fish-passage facilities, irrigation withdrawal intakes, and channel stabilization. The Snake River Salmon Recovery Team (1993) found, on the basis of model predictions, that the natural-river drawdown option would produce the highest inriver survival of yearling migrants in the Snake River basin. This option also would have the potential to increase survival over that predicted for transportation. However, the team added that "the tother] drawdown alternatives are highly uncertain, and even the most optimistic juvenile passage assumptions associated with a four pool drawdown fail to improve survival values of Snake River stock beyond what is achievable with juvenile transportation." Models that incorporate all life stages of salmon and their responses to various sources of mortality have the potential to be helpful in comparing alterna- tive management and environmental scenarios and as guides to research. The committee has not evaluated any of the models that have been developed on these lines (examples include models produced by the Columbia River Salmon Passage (CRISP) project (Center for Quantitative Science, University of Washington, undated), but it encourages their development and especially the collection of reliable data for them. Dam Removal Although dams are seemingly permanent (albeit recentJ features of the North- west riverine environment, like all artificial structures, they have a finite engi- neering and economic life expectancy. Structural criteria for dam safety have changed greatly in the 85 or so years since construction of the first high dams in the region, notably to include passage of extreme floods and resistance to earth- quakes. Some smaller dams have already undergone significant modifications for these reasons. In addition, dams trap sediment, which significantly reduces their active storage capacity and economic value. Although sedimentation in most Columbia River reservoirs is minor compared to dams on rivers elsewhere, which carry higher sediment loads under natural conditions, the economic life of all reservoirs is ultimately affected by sedimentation. A number of older, low dams elsewhere in the U.S. (notably in the East and Midwest) have been removed because of sedimentation. Where dams are a significant contributor to the decline of salmon runs, dam
DAMS AND MITIGATION OF THEIR EFFECTS 249 removal is an obvious rehabilitation alternative. Like the construction process that created the dams, dam removal would be a major engineering undertaking with major environmental consequences. It would be naive to expect that re- moval of a dam would allow a stream quickly to revert to its natural state. After removing the structure, in most cases, there would be major and long-lasting downstream effects due to movement of sediment stored behind the structure, and engineered rechannelization of the former reservoir bed would almost cer- tainly be necessary. The Elwha River Proposal The Elwha River dams provide a useful case study that gives some idea of the magnitude of the dam removal problem. The Elwha River drains 831 km2 of the Olympic Mountains, WA, and discharges to the Strait of Juan de Fuca. The Elwha Dam, about 8 km from the mouth of the river, was constructed from 1910 to 1913, and the Glines Canyon Dam, about 21 km from the mouth of the river, was constructed from 1925 to 1927. The entire firm energy productions of about 19 MW is presently used at a pulp mill in nearby Port Angeles. Glines Canyon Dam lies in the Olympic National Park, which was created in 1937, as does 83~o of the Elwha River drainage. Neither dam has fish-passage facilities, and the lower dam (Elwha) has never had a federal license to operate. In 1992, Congress authorized the secretary of the interior to acquire the dams and remove them if he determined that their removal was necessary to "the full restoration of the Elwha River ecosystem and native anadromous fisheries" (PL 102-495, Elwha River Ecosystem and Fisheries Restoration Act). A report pur- suant to the act (USDI 1994) found that "The removal of the Elwha and Glines Canyon Dams is the only alternative that would result . . . fin meeting the goals of the act]". The report also conducted a preliminary engineering analysis of dam removal alternatives, reviewed briefly here. The major issues in the Elwha River project are removal of the structures, management of sediment and erosion during and following dam removal, and re- establishing and protecting a channel within the presently inundated area. Al- though various combinations of dam removal (e.g., only Glines Canyon) and provision of fish passage facilities (at Elwha Dam) were considered, only the option of removal of both dams is reviewed here, because the report concluded that it is the only option that would meet the goals of the Elwha Act. Removal of the dams is complicated by the necessity to provide a stream channel during the dam removal process, which is estimated to require about 18 months. The channel and temporary discharge structure must be sufficient to 3Firm energy production is the amount of energy that could be produced during the worst-case historical conditions (known as the critical period), with all other demands on the system (e.g., irrigation) fixed.
250 UPSTREAM: SALMON AND SOCIETY IN THE PACIFIC NORTHWEST pass floods safely during this period. The options considered for diversion of the river were a) construction of a diversion tunnel, b) construction of a surface diversion channel, c) construction of a low-level diversion through the dam struc- ture; and d) progressive construction of notches through the dam along with top- down removal of the structure. The report did not conclude which option was preferable (although comparative costs were estimated), since the removal op- tions affect sediment management and channel reconstruction as well. Further, site-specific constraints preclude application of some of the options at both sites. For instance, the Elwha Dam was back-filled with large rock and gravel as a result of a structural failure of the dam foundation during the initial filling. Therefore, options c and d are not feasible at this dam. Construction of a surface diversion channel at Glines Canyon Dam is considered infeasible due to con- struction staging problems and high cost associated with the need to excavate much of the channel in bedrock. The lowest-cost alternative is a combination of low-level diversion through the dam, or progressive notching of the dam and top-down removal at Glines Canyon, and construction of a surface diversion channel at Elwha Dam. If top- down removal were used at Glines Canyon, it would be accomplished by con- struction of multiple notches 15.2 m (50 It) deep by 5.2 m (17 It) wide and temporary gates, in nine increments of 3.8 m (12.5 ft) each. Dam material above the notch would be removed by barge. At the last step, the dam base at the bottom of the gorge would be removed by cableway or boom crane during sum- mer low-flow conditions. The low-level diversion alternative would make use of a small outlet valve that was installed at the base of the dam during construction. This outlet would be enlarged, and the reservoir level lowered to its level to allow "dry" removal of most of the dam. Removal of the remaining portion of the dam structure would be the same as with the top-down approach. At Elwha Dam, the surface channel alternative would isolate the north spillway structure, which would be removed "dry," then the stream channel would be diverted in a second stage to a channel excavated through the existing north spillway site, and the main concrete dam section as well as power plants and south spillway would be removed. Perhaps the most difficult aspect of the project would be sediment removal and management. An estimated 8.6 million m3 of sediment has been deposited behind Glines Canyon Dam in the 70 years since the reservoir was filled, and 2- 3 million m3 has been deposited behind the Elwha Dam (mostly before comple- tion of the upstream dam). A delta has been formed at the head of Lake Mills (formed by Glines Canyon Dam), which is about 21 m deep. The delta consists mostly of coarse materials (small sand and larger); fine sediment is more gener- ally distributed throughout the reservoirs. Three alternatives, and combinations thereof, have been evaluated for sediment management: removal, erosion, and retention. Removal would be accomplished either after draining of the reservoirs by trucking to an upland site or to disposal in salt water, or (before draining of the
DAMS AND MITIGATION OF THEIR EFFECTS 251 reservoirs) by pipe transport of dredging spoils as a slurry to salt water or to an upland site. Both of the removal approaches are comparatively costly, require a long time for completion (as much as 9 yearsJ, and would have significant ancillary envi- ronmental effects (e.g., to the disposal site). They have the advantage that the final topography of the presently inundated areas could be made to resemble pre- dam conditions closely. The erosion alternative would allow the river to transport the accumulated material downstream. Resuspension of the materials would be augmented by dredging or other mechanical means. Fine and smaller coarse material would be transported by the river to salt water, and larger materials would likely be rede- posited downstream. Although this alternative has the advantage of low cost, it would result in high levels of suspended sediment downstream, and almost cer- tainly would extensively damage downstream fish habitat. Further, it was esti- mated that approximately 20 years would be required to remove all of the mate- rial from Lake Mills. An alternative of dredging the Lake Mills delta material and disposing of the dredged material over Glines Canyon Dam before removal was also investigated. This alternative would effectively store the coarse materi- als in Lake Aldwell (formed by Elwha Dam), and allow the river to transport the fine sediments downstream. This alternative would reduce the period necessary for stabilization of the channel system with respect to sediment movement and would eliminate the need for an upland disposal site. Because there is insuffi- cient space in Lake Aldwell for disposal of all of the Lake Mills delta material, the Lake Mills bed would not be completely restored to natural topography. The retention approach would relocate materials deposited in the old (pre- dam) river channel elsewhere within the lake beds, but would otherwise leave the accumulated sediment in its present location. Two retention alternatives were considered. The first would use hydraulic dredging before dam removal to re- store the original channel through the Lake Mills delta. The channel would be dredged to the original river-bed elevation. The second alternative would remove less material by dredging, and would allow the river to erode the delta to form a new channel after removal of the dams. However, unlike the erosion alternatives, no attempt would be made to remove sediment away from the original channel. The hydraulic dredging alternative is expected to have the least downstream effect, as much of the fine sediment disturbed during channel excavation would settle out in the reservoirs, rather than being transported downstream. Resettling would be enhanced through use of impermeable silt curtains, and by using the coarser dredged material to form containment cells, that would aid in dewatering of fine sediments. Subsequent to dewatering, the sediment would be graded, compacted, and revegetated. The estimated total costs of dam removal, sediment management, and reveg- etation (but excluding the cost of the hydropower loss) was estimated to range from about $70 million to $240 million. The lowest cost alternatives were those
252 UPSTREAM: SALMON AND SOCIETY IN THE PA CIFIC NORTHWEST that would remove sediment only from the former river channel, use either the notch or low level diversion at Glines Canyon Dam, and a surface diversion channel at Elwha Dam. Most of the cost of the restoration project (about 75% for the lowest-cost alternative, and well over 90% for the higher-cost alternatives) would be associ- ated with sediment management. Although the least-cost alternatives do not restore the topography of the inundated areas to their original contours, and might result in less spawning habitat in these areas, they have the advantage (in addition to cost) of greatly reducing the movement of fine sediment downstream. Further- more, only a relatively small amount of the spawning habitat that would be re- opened as a result of dam removal is in the inundated area. The dam removal project would itself be a significant engineering project. In addition to the removal of the dam structure, it would involve (depending on the alternative selected) construction of temporary roads, channels, and diversion dams, as well as extensive dredging and channel work. The greatest uncertainty in performance of the different alternatives (and their cost) is undoubtedly asso- ciated with sediment management, and this uncertainty is in turn greatest for alternatives that rely on channel transport of sediment (as opposed to removal or retention). Although there has been some experience with rechannelization via dredging in a manner similar to one of the sediment retention alternatives by British Columbia Hydro in the case of a reconstructed dam, there is essentially no comparable experience with any of the other approaches. In addition, there remains, after project completion, the potential problem of sediment movement due to slope failures induced by channel migration, which might well require permanent channel structures, such as levees, at least for the sediment-retention alternatives. Applicability of Experience How applicable is the Elwha River experience to potential removal of dams in the Columbia River system? At this point, the strongest candidates for dam removal in the Columbia system appear to be the run-of-the-river dams in the middle Columbia and lower Snake, which are the targets of current flow-aug- mentation studies. The most obvious differences between the Elwha River and middle Columbia and Snake dams are the size of the structures and the climate. Annual precipitation in the Elwha River Basin is about 1,700 mm/yr; in the middle Columbia and lower Snake it is in the range 200-400 mm. The humid climate in the Elwha River Basin would aid revegetation efforts; it is quite likely that a native forest could ultimately be restored in the presently inundated areas (albeit with soil enhancement), whether or not the original contours were re- tained. In the arid middle Columbia and lower Snake region, revegetation (and hence erosion control) in the presently inundated areas would pose a much greater challenge due to the probable necessity to irrigate, at least in the short term. In
DAMS AND MITIGATION OF THEIR EFFECTS 253 addition, the channel slopes and river gradients in the case of the much larger middle Columbia and lower Snake River dams are lower than the Elwha, which is a steep mountain stream, especially at Glines Canyon Dam. Therefore, the inundated area and length of inundated channel is much larger for the middle Columbia and lower Snake River dams relative to the Elwha dams, with expected restoration costs for the channel and inundated areas proportionately larger, as well. On the other hand, the amount of sediment deposited behind the middle Columbia and lower Snake dams is likely to be less, because these dams were constructed after upstream dams had already trapped much of the sediment that would otherwise have been deposited. Nonetheless, the costs of removing a middle Columbia or lower Snake River dam would be much larger than those estimated for the Elwha River dams. Selection of Mitigation Alternatives Many entities including the Fish Passage Center, CBFWA, the Idaho De- partment of Fish and Game, the Oregon Department of Fish and Wildlife, and the U.S. Fish and Wildlife Service have recommended spill of water at Snake River dams, elimination of transportation, flow increases, and drawdown of the Snake River reservoirs to increase water velocities. Opposing that position have been the Corps of Engineers, NMFS, utilities, and others. The Snake River Salmon Recovery Team (1993) extensively modeled miti- gation alternatives and their effect on the proportion of Snake River smelts that would arrive at a point downstream from Bonneville Dam. The team concluded that no combination of spill, flow augmentation, and drawdown within the limits imposed by present dam structures would produce survivals close to those ob- tained by transporting smolts from collector dams to the Bonneville tailrace. The team modeled a transport-benefit ratio (TBR) of 2:1. That ratio was based on an average of TBRs obtained in transport tests in 1986 (1.6:1) and 1989 (2.5:1 J. The study years represented modern river management and community structure. The team found that for all flow regimes, including high discharges (151 kcfs in the Snake and 401 kcfs in the Columbia), a mitigation program using transportation as the main tool was most effective. Only drawdown of the Snake River reservoirs to river grade, which would take years to design and build, potentially offers higher survival than transportation. No investigator to date has provided the Columbia River region with experi- mental results that demonstrate higher survival of inriver migrants than trans- ported migrants at any discharge level. Until such experimental data become available, transportation should continue to be used. However, it is essential that managers use an adaptive (experimental) management approach and avoid tak- ing any action that jeopardizes all of the fish in a stream. For example, if some fish in a stream are transported downstream, the action should be designed so its effectiveness can be assessed and compared with other alternatives, such as spill.
