Fall 1998

North Santiam River Turbidity Study, 1996-1997

Deigh Bates

USDA Forest Service, Willamette National Forest Eugene, OR

Katherine Willis

City of Salem Public Works, Salem, OR

Frederick Swanson, Ph.D.

USDA Forest Service, Pacific Northwest Research Station, Corvallis, OR

J. Reed Glasmann, Ph.D.

Willamette Geological Services, Corvallis, OR

David Halemeier

USDA Forest Service, Willamette National Forest, Detroit, OR

Hank Wujcik

City of Salem Public Works, Salem, OR

Editor's Note: Several figures would not reproduce electronically. See hardcopy for these.

Abstract: Following the flooding of February 1996, the North Santiam River below Detroit Reservoir carried an extraordinarily high level of turbidity that lasted for months. The Willamette National Forest, Pacific NW Research Station, Oregon State University, and the City of Salem undertook a joint effort to determine the composition and source of the persistent turbidity.

The primary causes of the persistent turbidity were smectite clays. These constituents were found to have multiple sources in broadly identifiable locations throughout the North Santiam River subbasin. The study reaffirms earlier findings that earthflows and other erosion processes affecting deeply weathered, smectitic deposits of the ancestral Western Cascades are the main sources of these materials in rivers and streams. These deposits are present, but less prevalent, in the younger High Cascades.

Streamflow in subsequent storm events has been of lesser magnitude and has not resulted in significant modification of channel patterns established during the February event. Absence of severe persistent turbidity subsequent to the flood of February 1996 may indicate that the river channels were effectively winnowed of the major smectitic deposits. Strategies for minimizing the production and effects of persistent turbidity include avoidance of earthflow deposits, restoration of disturbed sites, maintenance and enhancement of riparian ecosystems, construction of stream structures to accommodate floods and changes in reservoir operations.

 

Introduction and Conceptual Framework:

Following a large mid-winter storm and subsequent flooding in early February 1996 the waters of Detroit Reservoir and the North Santiam River downstream of the reservoir experienced high and persistent levels of turbidity. Concern was expressed in the spring of 1996 by City of Salem officials to personnel of the Detroit Ranger District, Willamette National Forest, about the high levels of turbidity that persisted at the Salem water intake for months after the February 1996 flood event. Following a meeting on March 5, 1996 between representatives of the City and the Detroit District, Bill Funk, Detroit District Ranger, asked that a study be conducted to investigate the sources of the turbidity, potential management relationships and remedial or corrective measures that could be employed.

This study, reported here, initiated a cooperative effort among technical specialists for the Willamette N.F., City of Salem, Pacific Northwest Research Station, and Oregon State University.

Broadly speaking, potential sediment and turbidity sources varied by geographic location and involved a range of processes:

1. Forest lands managed by the Forest Service upstream of Detroit Reservoir such as:

  • Surface erosion from roads and bare hill slopes

  • Landslides and debris flows

  • Channel erosion of earthflow toes.

  • Erosion of channel, floodplain and terrace deposits

2. Detroit Reservoir

  • Erosion of native soil along the reservoir shoreline

  • Sediment deposited in the reservoir since its establishment being remobilized as a secondary sediment/turbidity source.

  • Turbidity delivered to the reservoir during the flood of February 1996 that remained in suspension and delivered to downstream locations over a protracted period because of the release schedule of the reservoir.

3. Forest lands downstream of Detroit Reservoir and upstream of Salem water intake.

4. Other sources, such as Willamette Valley alluvium and agricultural lands along the North Santiam River valley.

We considered all potentially significant source areas and processes of sediment delivery upstream of the Salem water supply intake. This study examined geology, soils, land use, and other factors across ownerships.

 

History and Land Use:

The Santiam River, comprised of the North and South forks, is a major tributary of the Willamette River, entering the Willamette at River Mile 109.0. The North Santiam River, whose major tributaries are the Little North Santiam River, Breitenbush River, and Blowout Creek, drains ~1,984 km2 (766 mi2). The North Santiam River watershed serves as the municipal water supply for numerous communities, including Salem, Turner, Stayton, Lyons, Mehama, Gates, Detroit and Idanha. The largest of these systems, the city of Salem, draws water for its drinking water treatment plant from the North Santiam River at approximately River Mile 31, located 29.9 miles below Detroit Dam and Reservoir. Detroit Dam, completed in 1953, is located at River Mile 60.9 on the North Santiam River (Corps of Engineers, 1947). The reservoir is operated by the US Army Corps of Engineers (COE) for power production, flood control and recreation. The watershed area upstream of the reservoir is 1,134 km2 (438 mi2).

Ownership of the watershed upstream of the reservoir is:

National Forest 102,755 hectares (253,905 acres) 90%

Private 9,910 hectares (24,487 acres) 9%

State of Oregon 520 hectares (1,286 acres) 1%

Ownership distribution within the entire North Santiam subbasin is:

National Forest 119,494 hectares (295,266 acres) 60%

Private & Other 57,296 hectares (141,576 acres) 29%

State of Oregon 12,628 hectares (31,204 acres) 6%

Bureau of Land Mgmt. 8,474 hectares (20,940 acres) 4%

US Fish & Wildlife Service 552 hectares (1,363 acres) <1%

Jones and Grant (1996) report that the Breitenbush and North Santiam River watersheds, on National Forest land above Detroit Reservoir, were logged at a rate of 0.25- 0.5% of land area per year from 1955 until 1990. Over 90% of the harvest has been concentrated in what is termed the transient snow zone, below 1200 meters (3,937 ft), because of wilderness and other land allocation designations at higher elevations. By 1990 cumulative clearcuts in the North Santiam River watershed, excluding the Breitenbush River watershed, totaled 12% on National Forest land. Both watersheds are primarily in National Forest ownership; the Breitenbush River contains 0.3% private land, and the Upper North Santiam River contains 7%.

 

The Storm and Flood (Taylor, 1996)

The fall and early winter months of 1995-1996 set the stage for the hydrologic conditions that helped to generate the flood event of early February 1996. During that time large amounts of precipitation occurred in the Coast Range and on the west slopes of the Cascade Mountains. The Portland, Oregon airport received 67.95 cm (27.46 inches) of rain during the period October 1, 1995 through January 1996, 141% of the normal or average precipitation for this period of 49.53 cm (19.5 inches). This pattern was true for other stations around the Willamette River system. Eugene experienced 105.54 cm (41.55 inches) of rain which is 148% of the normal amount of 71.45 cm (28.13 inches). Detroit Dam received 171.37 cm (67.47 inches) of precipitation in the same period which is 145% of the normal amount of 117.86 cm (46.40 inches). The soils in the watershed were saturated as a result.

Snowfall in the 1995-96 winter was almost non-existent until the middle of January when large amounts of snow began building up in the Cascades. Most snowpack measuring sites (SNOTEL) in the Willamette River basin were reporting 112% of normal snow water content by the end of January. At this point streams had normal winter flows.

There was an intense cold spell during the week of January 29 with many Willamette Valley weather stations reporting lows in the teens for 4-5 consecutive days, especially in the northern part of the valley. A moderate storm during the evening of February 3 into February 4 dropped freezing rain throughout the Willamette Valley. On February 6, a dramatic weather shift occurred. A strong, warm, subtropical jet stream reached Oregon with a very humid air mass that brought record rainfall. In the Willamette Valley, daily minimum air temperatures were higher than normal maximum values for early February. Nighttime lows in the mid-50's were common. Freezing levels went to 7,000 - 8,000 feet and rain fell in the Cascade mountain passes.

Large amounts of warm rain and snowmelt combined to produce stream flow levels that approached those experienced in the 1964 flood event, one of the largest on record. Data collected by the US Geological Survey indicate the magnitude and recurrence interval, or return frequency, of peak flows during this event, in the Santiam River system (Table 1).

Unregulated basins, (those above the dams and storage reservoirs) in the upper Santiam River system are estimated to have experienced a 30- to 50-year storm event.

In Figure 3, the upper graph displays stream flow in kcfs (thousands of cubic feet per second) for both inflow and outflow on the left axis, and precipitation in inches on the right axis displayed as bars coming from the top of the chart. The horizontal axis displays time in months beginning on October 1, 1995. The lower graph shows the water surface elevation in feet, of Detroit Reservoir during the same period and is plotted with the COE operational rule curve for dam operations. The reservoir filled to full pool water level in three days during the flood event effectively capturing most of the incoming sediment from upstream or in-reservoir sources. Suspended sediment was then released after the flood as the COE drew down the reservoir to meet their management requirements to stay on the operational rule curve in anticipation of possible additional flooding. About March 1 they began filling the reservoir for summer recreational activities.

 

Table 1 Peak streamflow at selected Santiam River gaging stations. (USGS, WDR OR-96-1)

The February 1996 storm produced extremely high turbidity in the North Santiam River below Detroit Dam during and following the storm event. Turbidity readings as high as 140 ntu1 were measured at the City of Salem's Geren Island intake. This level of turbidity was essentially untreatable by the City's slow sand filtration system, which is designed to treat water less than 10 ntu. Raw water turbidity exceeded 10 ntu from February 6 through March 19 and finished water turbidity, that is, post-treatment by the City of Salem, ranged from 0.4 to 4.4 ntu. Two months following the February event, river turbidity continued to exceed 5 ntu (Figure 2) and finished water turbidity exceeded 1 ntu, the standard for drinking water turbidity set by the Environmental Protection Agency. The City obtained a waiver from the Oregon Health Division to allow delivery of the higher turbidity water. The raw water contained a large quantity of colloidal particles that remained in suspension (Willis, 1996).

Other municipal watersheds on the west slopes of the Cascades experienced similar high flow and high turbidity during the February Storm. Layng Creek is the major source of water for the City of Cottage Grove, Oregon. Grab samples of turbidity at the Cottage Grove treatment plant resulted in the following turbidity levels (K.Smith, Umpqua N.F., Personal Communication):

February 8th: 145 ntu February 9th: 166 ntu

February 10th: 61 ntu February 11th: 18 ntu

Similar samples gathered in Portland Oregon's Bull Run watershed showed that the maximum turbidity reading was 35 ntu in the North Fork Bull Run river on February 7 (A.Smart, Mt. Hood N.F.,Personal Communication).

The Portland District of the U.S. Army Corps of Engineers (Ruffing, 1996) performed turbidity analysis of streams flowing into Detroit Reservoir and conducted vertical profiles of turbidity concentrations in the reservoir on two separate periods, February 14-15 and February 21, 1996. The results indicated levels of turbidity in the tributary streams of 12 - 21 ntu on February 14 and in several reservoir profiles turbidity levels ranged from 55 ntu at the surface to 389 ntu at a depth of 78.33 meters (257 feet) measured on February 15 and 21. Additionally, they conducted gradation tests, i.e., separating the suspended sediment into soil size categories; 75% of the material was silt/clay and the settling rate was very slow.

 

Introduction to Turbidity

Turbidity in natural waters is typically caused by fine inorganic and organic particles suspended within the water column. Sediment is delivered to streams in a variety of ways (Figure 4). A discussion of turbidity can be separated into two parts: The mineralogical character of the materials comprising the turbidity and the delivery processes of that material to water.

Clay Mineralogy

The character of the materials that contribute to persistent turbidity in the Cascade Range has been the subject of study for over 25 years. Most of the material that exhibits long-term suspension in natural waters does so because of its extremely small particle size. Clay particles are defined as sediment that are <2 µm in diameter. Within this size class, several groups of different clay minerals occur, each with different physical and chemical properties that influence their potential for persistent suspension in water. Among these different clay mineral groups, the smectite clays are common in unstable soils and hydrothermally (hot groundwater) altered volcanic rocks of the Cascade Range. Smectite clays typically show the finest particle size distribution of naturally-occurring clay minerals (particles <0.05 µm), and have electrically-charged surfaces that permit the adsorption and absorption of water and other substances. These tiny, charged surfaces interact with other molecules and are kept in suspension through Brownian motion (molecular vibration); however, smectite particles are sensitive to coagulation by altering the chemistry of the suspending medium. If coagulation can be induced, the larger clumps or flocs may settle out of suspension or may be removed via filtration.

A second group of clay particles that contribute to persistent turbidity is amorphous or poorly crystalline materials. These materials commonly form by rapid weathering of volcanic glass in soils and are a common component of many ash-rich soils of the High and Western Cascades. Amorphous clay particles include water- and silica-rich gels and poorly crystalline aluminum, silica, and iron-hydrates. Depending on their chemistry and progress towards crystal organization, these particles may display a wide range of suspension characteristics.

Other clay mineral groups occur that show less potential for persistent suspension. The kaolin group of clays includes particles with relatively large size and neutral electrical charge. Kaolin forms under conditions of strong leaching and can develop in well-drained surface soils or in acidic hydrothermal environments. Kaolin particles generally fall in the <20 µm size range, although highly disordered kaolin (or a related mineral halloysite) may occur as smaller particles. Because of their relatively large particle size and absence of strong surface charge, kaolin group clays are less prone to remaining in suspension due to Brownian motion and settle more readily than clays of the smectite group.

Clay minerals of the mica and chlorite groups are common components of high temperature hydrothermal environments in the Western Cascades. Both vermiculite and chlorite may develop from a smectite precursor, given proper chemical conditions, and generally form larger clay particles that do not contribute greatly to persistent turbidity. Weathering processes in the soil may transform mica to vermiculite, kaolinite, or smectite. With each transformation, there is a decrease in particle size and an accompanying increase in the suspendability of the newly formed clay particles. Chlorite is one of the more stable clay minerals in soils and occurs in relatively large particles with low surface charge. If formed by weathering processes, chlorite particles are small and display some of the chemical properties of smectite group clays, but when formed through regional hydrothermal alteration, the particles are larger and less prone to prolonged dispersion in water.

The clay size fraction of suspended sediments also includes minerals that are not classified with clay minerals. Tiny particles of quartz, feldspar, pyroxene, or other primary minerals that occur in unaltered volcanic rocks may temporarily exist in suspension, but because of size/charge properties, they settle rapidly in quiet water. Hydrothermally altered volcanic rocks may also contain zeolites, iron oxides, iron sulfides, metal ores, or aluminum hydroxides (gibbsite). Many of these minerals show limited geographic distribution related to a specific geologic environment (e.g., mineralized ore vein). Some minerals, such as gibbsite, are natural flocculants and actually help remove suspended clays from natural waters.

These different primary and secondary minerals may all occur in the suspended load of a stream, but once entering a reservoir, a natural segregation occurs which concentrates the smallest, higher-charge clays in suspension. Thus a mixed mineralogical system in the water column generally becomes less heterogeneous with time as kaolin, mica, chlorite and various primary minerals settle from suspension, leaving smectite and poorly crystalline clay particles as components of persistent turbidity.

Delivery Processes

Landslides are shallow, rapid movements of soil and rock materials in thin soils generally less than ~2 meters deep overlying steep bedrock or compacted glacial material. Landslides can transform into debris flows, defined as highly mobile slurries of soil, rock, vegetation and water that can travel up to thousands of meters from their initiation point and usually occur in steep (>6 degrees) and confined mountain channels (Benda, 1997). Debris flows are a minor source of smectite clays but do deliver amorphous, hydrated halloysite and chloritic intergrade clays to stream channels.

Earthflows are large (to many hectares), deepseated (failure planes may be 515 m below the surface), slowmoving (movement rates of millimeters to several meters per year) mass movements. Earthflows move in a slow and episodic manner over a long period of time, although rapid movement of large blocks of material within the flow is common (Benda, 1997). Earthflows are a major source of smectite clays and additionally deliver zeolite and cristobalite clay types.

A third sediment delivery mechanism is bank erosion along mainstem and tributary channels during high flow events. This may include erosion of earthflow toes and previously deposited alluvium in terraces and floodplains. The magnitude of the erosion is related to the magnitude of the flow, i.e., the greater the flow the more bank erosion is likely. Smectite clays can be delivered to stream channels in substantial quantities from bank erosion along the toes of earthflows and in alluvial deposits.

Surface erosion from roads and bare soil areas is a fourth mechanism that contributes sediment to stream channels in a watershed. Only a small portion of the sediment from this mechanism contains smectite clay.

Once sediment enters flowing water it is rapidly separated into components of the stream's suspended and bed load. The suspended load generally consists of silt and clay particles that are too small to settle quickly under the influence of gravity, whereas the bed load consists of sand and larger-sized particles that bounce and roll downstream. Within the suspended portion of a sediment load, the various particles can be segregated into those that readily settle from suspension in quiet water and those that persist in suspension for weeks or months. It is this later group of particles that contribute to persistent turbidity in streams and reservoirs.

 

Relevant Past Work

This study builds on the work of Youngberg et al. (1971, 1975) on turbidity in Hills Creek reservoir. The studies by Youngberg et al. were initiated in response to concern about persistent turbidity similar to the Salem water supply issue. They observed higher turbidity levels in streams draining weak, pyroclastic bedrock and unstable soil. Streams with the highest turbidity levels drained areas with large earthflows and road failures. "Soils developed from andesite, basalt, pumice, and/or morainal deposits are generally coarsetextured and smectites are low or absent" (Youngberg et al. 1975, p. 282).

Youngberg et al. (1971, 1975) used a "fingerprint" study approach by analyzing water samples to determine clay minerals causing persistent turbidity, and then identifying source areas and delivery processes of these particular minerals.

Youngberg et al. (1971, 1975) argued that smectite (montmorillonitegroup clays) and amorphous solid materials are causes of persistent turbidity in streams draining hydrothermally altered rocks of the Cascade Range. They also noted that geologic units and associated soils containing smectite and amorphous solid materials are prone to soil mass movement by earthflows. They defined earthflows as large (to many hectares), deepseated (failure planes may be 515 m below the surface), slowmoving (movement rates of millimeters to several meters per year) landslides. The presence of zeolites in turbid water samples indicated sediment sources from below the soil profile, such as deeper levels of large landslides. Youngberg et al. (1971, 1975) also argued that shallow, rapid sliding, including the high frequency of sliding from roads, contributed to delivery of soil to streams tributary to Hills Creek Reservoir.

Taskey (1978) and Taskey et al. (1978) conducted studies of clay mineralogy and landscape stability in the western Cascades of Oregonmainly in the Willamette National Forest. They concluded that slope failures are "caused by (soil) discontinuities between smectite rich materials below the contact, and perched water tables and occurrences of halloysite and amorphous materials above" (Taskey et al. 1978, p. 191). They also observed amorphous material and halloysite in stable areas and in steep, debrisslide prone terrain as well as earthflow terrain. Differences between stable and unstable sites may result from variation in the proportions of these clay minerals and hydrologic properties of the sites. In summary, these authors suggest that there is some potential to distinguish sediment source areas and processes by their clay mineralogy, but "fingerprinting" turbidity sources is not simple and sharply defined.

 

Methods

In an effort to locate sources and types of turbidity, X-ray diffraction and transmission electron microscopy were used to characterize the mineral assemblage of samples taken throughout the North Santiam River subbasin.

Since clay minerals are typically too small to be observed and characterized by optical methods, X-ray diffraction analysis is the most commonly used method for identifying the mineralogy of fine soil (<2 µm) samples. The distance between atoms in the crystal structure of most naturally-occurring minerals is about the same as the wavelength of X-rays; therefore, a focused beam of X-rays directed at a clay crystal will yield a characteristic diffraction pattern that gives information about the internal structure of the material. Different minerals within a soil sample will produce different diffraction patterns. By carefully analyzing X-ray subsequently enters a stream channel or a stream erodes an adjacent terrace (Glasmann, 1997). Suspended Sediment samples were collected in the reservoirs using a Van Doren sampling bottle or were hand grab sampled, in the case of the two Breitenbush River samples . Delta Sediment samples were collected using a 2 inch PVC pipe pushed by hand as far as possible into the reservoir bottom sediment. These pipes were then capped and delivered for analysis. Soil and Suspended Sediment samples were collected in 18.9 liter (5 gal) plastic buckets and sealed for delivery. Ten milliliters of household bleach was added to the Suspended Sediment samples to retard biological activity.

The initial phase of sampling included Suspended Sediment and Soil samples taken over a limited geographic area. Initially no samples were taken below the reservoir with the exception of SWR 3/96 taken in March 1996 at the Salem City Water Treatment Plant intake. The second phase of sampling was conducted to widen the geographic area of sampling, i.e., below the reservoir, and to take advantage of the lowered reservoir pool level that exposed deltaic sediments at the mouths of major tributaries. Lastly, a third phase of sampling was done to examine the clay mineral contributions from eroding stream banks along the main North Santiam River below Big Cliff dam.

 

Results

A complete discussion of the results from each sampled location is available in Clay Mineralogy of Soils and Suspended Sediments within the North Santiam River Drainage: Preliminary Report, Glasmann (1997); Clay Mineralogy of Soils and Suspended Sediments Within the North Santiam River Drainage: Progress Report, Glasmann (1997) and Clay Mineralogy of Soils Exposed in Riverbank Cuts below Big Cliff Reservoir, North Santiam River Drainage, Glasmann (1997). Copies of the above documents are available at the Willamette N.F. Supervisor's Office or the City of Salem Public Works Department.

Contrasting X-ray diffraction patterns reveals the differences in mineralogical makeup of the watersheds upstream and downstream of the reservoir. The role of the reservoir in both homogenizing incoming sediment and metering it out through time to the lower river is also revealed diffraction patterns from different soils, their mineralogical assemblage may be identified in a semi-quantitative manner.

Similarly, the small size of clay particles precludes normal optical characterization with visible light microscopes. To view clay particles, a microscope which uses a much smaller wavelength of light must be used. The electron microscope uses a beam of focused electrons to illuminate tiny specimens in much the same way as a normal light microscope is used to study larger materials. In transmission electron microscopy, a beam of electrons is passed through a thin sample to produce an image from the material. The electrons interact with the material in a number of ways, yielding information about the chemistry, crystallinity, and shape of the sample. These features may be used to help identify sample mineralogy and characterize differences between various samples.

Three different types of samples were taken: 1) Suspended Sediment samples taken from rivers or the reservoir represent a mineralogical mix of sediment in the stream and reservoir derived from a variety of upstream erosional processes and/or landforms, 2) Soil samples from a specific landform, such as a debris flow, and 3) Delta Sediment samples taken from delta areas at the mouths of tributaries to the reservoir that closely represent the sediment derived from a particular watershed (Figure 5). Suspended Sediment samples were collected from Detroit Reservoir, Big Cliff Reservoir, the Breitenbush River and the North Fork of the Breitenbush River. Soil samples were collected throughout the subbasin. Selected sites included active debris flow channels, stream bank failures, rotational slumps, and road failures. Recent Sediment samples were all taken within the exposed area of Detroit Reservoir during the winter of 1997 when the Corps of Engineers lowered the reservoir water level 7.62 meters (25 feet) below minimum flood control pool level. This was done at the request of the City of Salem in an attempt to flush remaining turbid water from the reservoir.

A comparison of X-ray diffraction patterns from Suspended Sediment samples (Sample SWR 3/96 and DR 7/22/96) from Detroit Reservoir and the City of Salem Water Intake near Stayton yielded a very close match (Figure 6, next page). The results indicate similar clay mineralogy for the two samples. Sample SWR 3/96 was collected from the Salem intake in March 1996 and sample DR 7/22/96 from a depth of 67.1 meters (below the thermocline) in Detroit Reservoir on July 22, 1996. The raw water sample from the Salem intake is a very complex mineralogical mixture dominated by smectite and amorphous material. The Detroit Reservoir sample is very similar differing primarily in the lower abundance of kaolinite and chlorite (Glasmann, 1997). These results show that the mineralogical makeup of water released from Detroit Dam in the month following the February flood was essentially the same as water continuing to be released from the reservoir in July.

No particular Soil sample examined in the first phase of sampling, which was limited to areas above the dam, provided an exact match or even close match to the clay assemblage of persistent turbidity in Detroit Reservoir. This suggests that the reservoir acts as a homogenizing catch basin wherein fine-particle, higher charged clays (smectite, amorphous materials, halloysite) remain in suspension while the larger particle, lower charged clays (kaolinite, chlorite, illite) settle from suspension. Furthermore the results of the Salem water intake sample indicate contributions of kaolinite and chlorite from sources below the dam. This was confirmed by the last two samples taken (Phase 3) from eroding streambanks below Detroit Dam (NSRFB & NSRBLMPL). While these two samples show appreciable smectite content they are also the probable below-dam source of the chlorite clays that were present at the Salem Water intake (Glasmann, 1997).

In order to more specifically determine the watersheds
contributing to the persistent turbidity, a second phase of sampling was conducted (Table 3). These samples included Delta Sediment at the mouths of each of the major tributaries to Detroit Reservoir as well as an expanded array of Soil samples, bothabove and below the reservoir. Of the samples in Table 3 listed as Delta Sediment, two in particular are instructive: Mouth of Blowout Creek (FS-5) and Hoover Campground Mud Flat (FS-6)(Figure 6a). Glasmann (April,1997)(Figure 6b) states "The smectitic character of sediments deposited in the mouth of Blowout Creek contrast markedly with the chloritic clay assemblage in Cumley and Kinney Creek bays..., but is very similar to the mineralogy of reservoir suspended sediments." Additionally, he states "Sediments exposed during low pool conditions are dominantly deltaic materials derived from the North Santiam. The clay mineral assemblage of these sediments is dominated by smectite, with minor amounts of halloysite, zeolite, chlorite, and traces of illite and iron oxide.... The Hoover mud flat sediment is mineralogically identical to materials deposited in Blowout Creek bay and shows strong similarity to suspended sediment from both Detroit and Big Cliff reservoirs" (Glasmann, April 1997).

Soils from areas characterized by large scale, deep-seated earthflows yield a smectite-dominated clay mineral assemblage, e.g., Blowout Creek and Straight Creek sites, FS-5 and SCTDT-6, (Figure 5).

Upon erosion and dispersion these soils contribute smectite-rich suspended sediment. Clays from both sites (FS-5 and SCTDT-6) are derived from areas of hydrothermal alteration within identified earthflows and, as discussed earlier, these are believed to be major sources of the smectitic sediments that produced the high and persistent turbidity in Detroit Reservoir. These areas of large earthflow terrains are common within the western Cascades and exist in all of the
WMC Networker Fall 1998 watersheds examined in this subbasin and in others throughout the Willamette National Forest. A map of the North Santiam River subbasin (attached) shows that, earthflow terrains occur in distinct zones both upstream and downstream of Detroit Reservoir.

Areas downstream of Detroit Reservoir were also found to produce smectite and amorphous sediment. The clay mineral assemblage from the sample taken from the Little North Fork Debris Slide (FS-9) is dominated by smectite, with minor amounts of hydrated halloysite, cristobalite, and plagioclase. This clay assemblage is similar to that developed in earthflows in the Detroit Reservoir watershed, indicating that smectitic clays were added, during the flood, from areas below Detroit Reservoir. While clays from these downstream areas did not contribute, in a major way, to the persistent levels of turbidity experienced at Salem, i.e., for many weeks after the February flood, they did add to the initial sediment load impacting the treatment plant, and will continue to serve as a future source for such sediment.

A secondary source of turbidity-producing sediment was identified as the reworking of sediments stored in stream valleys. Stream discharge from storms of various magnitudes access different portions of these sediments, i.e., major storm flows access areas that lesser storm flows do not. Braudrick and Grant (1997) mapped channel bar dimensions and elevations before and after the February 1996 flood event on a section of the Breitenbush River. They studied a 600 meter section of the North Fork of the Breitenbush River above the Jefferson Bridge, road 4685, composed of cobble bed material derived from glacio-fluvial terraces 4 to 20 meters high. They noted large changes in channel configuration. Previously abandoned side channels became activated while previously active ones were abandoned as the stream moved across this large floodplain, often in response to the influence of large individual pieces of wood or the formation or presence of debris jams. The actively eroding channel margin or terrace changed from past locations, indicating that new sediment sources were activated. Sample NFBR-1, North Fork Breitenbush River above Road 4685 Bridge, contained a trace amount of smectite clay. Hydrated halloysite, another small clay particle, was dominant in these deposits.

Source patterns of persistent turbidity vary at several scales over the North Santiam River subbasin. The High Cascades area has low levels of clays that produce significant persistent turbidity due to young rocks and lack of hydrothermal alteration. Soils derived from the older Western Cascades, on the other hand, contain high levels of persistent-turbidity clays. High rates of delivery of this material occur at earthflow toes and along streambanks with deeply weathered sediment deposits. Consequently, it is possible to delineate sediment sources at several levels of importance and scale.

 

Discussion

While it is not possible to define with complete certainty the scenario of timing and delivery of sediment to the Salem intake, the following represents a plausible sequence of events based upon the investigations of the sediment sources done in this study.

Sediment was delivered to or mobilized in streams from the toes of earthflows, the reworking of in-channel and valley floor sediments, debris slides and debris flows from steep slopes and other sources such as road surfaces (Steps 1,5 &6
Figure 7). In the upper watershed a relatively complex mineralogical assemblage of sediment, including a variety of clay types, entered Detroit Reservoir (Step 2- Figure 7). The sediment that persisted in suspension, was dominately smectite. Most of the other types of clay, i.e., the chlorite, illite, and kaolinite, settled out of suspension in the reservoir in a short period of time, forming muddy deltas along the reservoir margin.

During the flood event much of the sediment entering Detroit Reservoir was trapped as a consequence of the Corps of Engineers flood control efforts (Figure 3). During the main flood event below the dam clay-enriched sediment, including smectites, was moved out of various tributaries, especially the Little North Santiam River (Step 5 - Figure 7) from debris flow and earthflow sources similar to those sampled in watersheds above Detroit Reservoir. Streamflow from the Little North Santiam River, although only 17% of the total watershed area, represented 64% of the total flow at the North Santiam River at Mehama gage on February 7, 1996, the day the flood peaked. During the flood, channel erosion below the dam (Step 6 - Figure 7) added additional sediment and turbidity to the mainstem of the North Santiam River above Salem's water intake. Thus it is highly likely that the initial surge of sediment received at the Salem intake came primarily from below-dam sources.

The turbidity that persisted for several months, however, came from the smectite rich waters of Detroit Reservoir being parsed out through time, as the Corps of Engineers lowered the reservoir to attain their normal operating level for mid-February (Figure 3). The smectite clays stayed in suspension and were transported through Detroit Dam to Big Cliff Reservoir and into the lower river (Steps 3&4 - Figure 7). The initial X-ray diffraction analysis of suspended sediment samples taken in March 1996 at the Salem intake and in July 1996 at depth in Detroit Reservoir showed almost identical properties, supporting the idea that sediment-rich waters from the reservoir were being released over time, thus creating the persistent turbidity experienced in the lower river. Following the flood, flows below the dam stayed artificially high as the Corps dropped the reservoir levels. Streambank erosion below the dam from these flows may have added additional sediment to the downstream mix and contributed to the persistent turbidity problems experienced at Salem.

 

Summary

This study highlights the complex nature of the turbidity issue in the North Santiam River subbasin. The problem of persistent turbidity involves a complex interaction among sources of turbidity causing sediment, delivery and transport processes, temporary storage and fractionation of turbid water in the reservoir, and release of turbid water through time.

  • Smectite clay is the major component of persistent turbidity in reservoirs on the North Santiam River and in the Salem City intake water, and as shown by this study, has multiple sources in broadly identifiable locations within the North Santiam watershed.

  • Levels of turbidity in streams flowing into the reservoir dropped to 12-21 ntu in the days immediately following the February flood. However, reservoir levels of turbidity stayed high well into the summer months. (Ruffing, 1996) The temporary storage and release of turbid waters from the reservoir contributed to the persistent turbidity experienced at the Salem intake.
    · The mineralogy of lake sediment reflects the regional importance of deep-seated slope failures (earthflows) and river incision into deeply weathered, smectitic deposits of the ancestral Western Cascades (Glasmann, 1997). The removal of the toes of these deep-seated failures during high flow events may be the primary source of smectite clays contributing to persistent turbidity.

  • "The smectite-dominated character of the clay assemblage of lake (Detroit Reservoir) sediments is not representative of the mineralogical diversity of surface soils and shallow soil failures in the Detroit watershed, indicating that lake sedimentation is generally unrelated to shallow soil-erosional processes." (Glasmann, 1997)

  • "February 1996 flooding was associated with major reworking of sediments stored in stream valleys and dumped a tremendous amount of material into Detroit Reservoir. Smectitic clay remained in suspension in the reservoir through the summer and continued to affect downstream water quality for many months after the major flood event." (Glasmann, 1997)

  • Streamflow in subsequent storm events has been of lesser magnitude and has not resulted in significant modification of channel patterns established during the February event. The absence of severe persistent turbidity in post-February storm events apparently indicates that stream channels were efficiently winnowed of the turbidity causing smectite which 'supercharged' the February flood (Glasmann, 1997). The City of Salem did note high turbidity at its intake from November 1996 through January 1997 following high flow events during those months. However, this turbidity was not as high in colloidal particles (smectite) nor was it sustained as long as that following the February 1996 event (Willis, unpublished data, 1997).

  • Detroit/Big Cliff reservoirs act like settling basins, receiving sediment of diverse character, homogenizing the suspended component, and allowing coarser sediment phases to settle from suspension.

  • Turbidity on the lower North Santiam River is a function of several sources that differ in frequency and distribution. The Little North Santiam contributes considerable suspended sediment and adversely affects Salem intake water turbidity, albeit on a limited scale related to episodic peak discharge events. While the turbidity derived from the Little North Santiam watershed initially impacts the Salem water intake during high flow events, it is quickly flushed through the system and does not contribute to the persistent turbidity.

 

Management Recommendations

The magnitude of the February 1996 flood presented extreme challenges for management of the watersheds throughout the North Santiam River subbasin. Planning for such events, and events of lesser magnitude, will be an ongoing challenge for forest, reservoir and water supply managers.

Strategies for reducing erosion from forest management activities in order to minimize the effects of persistent turbidity events in the future can take three possible tracks: Minimizing impacts to sediment-sources, restoration of sediment-source sites, and construction and re-construction of structures such as culverts with specific consideration for potential turbidity problems. Additional field work and data analysis could provide information useful for determining
where specific measures might be most effective in improving water quality.

Earthflow terrains, major sources of persistent turbidity, are currently mapped and recorded in the Willamette N.F. GIS geology data layer and in the Soil Resource Inventory data layer. The Soil Resource Inventory data layer has had extensive field verification and is generally more accurate than the coarser mapping done for the geology data layer. Further development of the GIS data within the Willamette National Forest and development of such information for areas within the subbasin, but outside the National Forest, would be useful for forest management efforts. These data should be an integral element of any of the following strategies.

Field investigations and mapping of earthflow terrains in the subbasin can afford planning teams the necessary information to avoid activities in unstable areas likely to produce excessive persistent turbidity. This strategy is basically one of avoidance or at least minimizing disturbance of the areas. This passive approach recognizes the size and complexity of such features and the inherent costs and difficulties in engineering a solution aimed at limiting the input of sediment from these earthflows into adjacent streams. Thus, forest management activities such as timber harvesting, road construction and recreational facility development should be conducted with an awareness of the geologic characteristics of the land.

Restoration activities associated with earthflows are somewhat limited simply due to their large size and the mechanisms at work in the earthflow. Confining a channel with an armored and compacted road, opposite an earthflow toe, may push the channel against such toes and aggravate sediment problems even during moderate flow events. Where roads have constricted channels at the toes of earthflows, and the roads are no longer needed, it would be prudent to remove such road segments thus widening the channel and allowing it to move more freely across its valley floor.

A strategy that can be applied to either re-construction or new construction is to anticipate flood effects when building stream-crossing structures and culverts. Recognizing that it may never be feasible to engineer stream-crossing structures that can pass all combinations and quantities of water, sediment and debris during large flood events, such as February 1996, it is prudent to design these crossings so an individual failure will not trigger a series of additional structure failures. The failure of a single culvert, while locally catastrophic, is not as significant, in most cases, as the suite of effects that can be caused by a cascade of linked failures originating from a single source. This "snowball" effect can result in extensive damage to both structures and watersheds and resultant water quality. By anticipating flood effects, the consequences from a single crossing failure can be limited.

Maintenance and enhancement of riparian ecosystems through silvicultural activities is an important restoration activity. Healthy riparian vegetation anchors streambanks and retains sediment on floodplains during high flows, however it should not be expected, nor is it desirable, that all sediment can be controlled by riparian vegetation.


Further Investigations

The effectiveness of each of the above approaches will be limited by the availability of information on the processes at work. Monitoring sediment influx and character in relation to storm runoff for the major streams feeding the Detroit reservoir would help define the significant contributors to reservoir suspended sediment. Proposals to conduct such monitoring are currently being discussed as a joint effort between the U.S. Geological Survey, City of Salem, U.S. Army Corps of Engineers and the Willamette National Forest.

Detailed stream and hillslope field surveys may help to quantify sources of potential sedimentation and determine which management changes or actions can alleviate contributions of sediment from such sources.

Continued study of the potential for increases in peak flows, as a result of roading and timber harvest, should also address the potential consequences of increased bank erosion into earthflow toes and terrace deposits, a primary source of persistent turbidity.

The development of a sediment budget to quantify the volumes of sediment production by watershed and perhaps sub-watershed within the North Santiam subbasin would provide complementary information to this study. Sediment loading is likely to be addressed to some degree by a currently funded study by Hulse, Whitelaw and Grant (1997). This study will involve mapping the geomorphological and land-use characteristics of the North Santiam River sub-basin and determine the incremental contribution of sediment from natural versus managed landscapes.

The role of the reservoir in the delivery of turbid water should be studied further to determine how best to manage the level at which water is removed. Since turbid water has a higher density than clear water, at equivalent temperatures, it tends to occupy deeper zones in the water column. Dams, such as Detroit, which are used for power production have their water inlet as low as feasible to take advantage of greater hydraulic pressure and thus greater power production. Retrofitting the dam with a stand tower to draw water from various elevations would provide greater flexibility to influence downstream turbidity levels through dam operations and should provide for full power production.-

 

Acknowledgments

The authors wish to thank Bill Funk, Detroit District Ranger and Frank Mauldin, Salem Public Works Director for their continued interest in and support of this work. Additionally, we wish to thank Tina Schweickert, Steve Sundseth, Rollie Baxter, and Jim West from the City of Salem, and Herb Wick, Rich Stem and Darrell Kenops from the Willamette National Forest, and Dr. Gordon Grant, PNW Research Station for their guidance and document review. Finally, thanks go to Tom Queer and Tim Sherman, City of Salem and Dave Klug, Detroit R.D., who participated extensively in the collection of the samples and to George Lienkaemper, PNW Research Station and Henry Jennings, Willamette N.F. for their GIS assistance.

 

1. ntu - nephelometric turbidity unit - see Standard Methods 17th Edition - pg 2-13 for definition

Literature Cited

Benda, L., Veldhuisen, C., Miller, D., Miller, L.R., 1997, Slope Instability for Forest Land Managers, Earth Systems Insti
tute, Seattle, WA.

Braudrick, C.A., Grant, G.E., 1997, Map of Channel morphology and wood location for the North Fork of the Breitenbush River, above Jefferson Bridge, Oregon State University, unpublished.

Jones, J.A., Grant, G.E., 1996, Peak Flow Responses to Clearcutting and Roads in Small and Large Basins, Western Cascades, Oregon, Water Resources Research, Vol. 32, No.4, Pages 959-974.

Glasmann, J.R., Clay Mineralogy of Soils and Suspended Sediments within the North Santiam River Drainage: Preliminary Report, Willamette Geological Service, WGS019706, 1997.

Glasmann, J.R., Clay Mineralogy of Soils and Suspended Sediments within the North Santiam River Drainage: Progress Report, Willamette Geological Service, WGS049704, 1997.

Glasmann, J.R., Clay Mineralogy of Soils Exposed in Riverbank Cuts below Big Cliff Reservoir, North Santiam River Drainage, Willamette Geological Service, WGS049704, 1997.

Hulse, David W., Whitelaw, E., Grant, G., 1997, Establishing correlations between upland forest management activities and the economic consequences of stream turbidity in municipal watersheds, Proposal to U.S Environmental Protection Agency, Washington, D.C.

Peterson, C.D., Hamilton, D.M., Burns, S.F., 1995, Sediment Deposition in Reservoir No. 1, Bull Run Watershed, Oregon, Portland State University, Portland, Oregon.

Portland District Corps of Engineers, October 15, 1947, Review of Survey Report Willamette River and Tributaries, Oregon, Appendix D, Dams and Reservoirs, pp D-210 - D-215

Ruffing, F.E., Britton, J.L., Flint, N.A., 1996, City of Salem Water Supply - February 96 Flood, 2nd Annual Pacific Northwest Water Issues Conference, 8 pages.

Swanson, F.J., G.E. Grant. 1982. Rates of soil erosion by surface and mass erosion processes in the Willamette National Forest. Report to the Willamette National Forest, Eugene, OR. 50 p.

Swanson, F.J., R.J. Janda, T. Dunne, D.N. Swanson, eds. 1982. Sediment budgets and routing in forested drainage basins. Gen. Tech. Rep. PNW141. Portland OR, USDA, Forest Service. 165 p.

Taskey, R.D. 1978. Relationships of clay mineralogy to landscape stability in western Oregon. Ph.D. Thesis. Oregon State Univ., Corvallis, OR.

Taskey, R.D., M.E. Harward, C.T. Youngberg. 1978. Relationship of clay mineralogy to landscape stability. In: Youngberg, C.T. Ed. Forest soils and land use: Proceedings of the 5th North American Forest Soils Conf. 1978 August; Fort Collins, CO. Colorado State U.; Dept. of Forest and Wood Science. 140164.

Taylor, George H., The Great Flood of 1996, Internet Address: http://www.ocs.orst.edu/Flood2.html, April 1996.

USDA Forest Service, Region Six and USDI, Bureau of Land Management, 1996, Storms and Floods of the Winter of 1995-1996 - An Assessment of Effects on USDA-Forest Service and USDI-Bureau of Land Management Lands. 15 p.

U.S. Geological Survey, Water Data Report: OR-96-1.

Willis, Katherine, April 1996. Flooding, Landslides affect Salem Water Treatment - City of Salem, Water Quality Report.

Willis, Katherine, 1997, Unpublished data from City of Salem Water Treatment Reports.

Youngberg, C.T., M.E. Harward, G.H. Simonson, D. Rai. 1975. Nature and causes of stream turbidity in a mountain watershed. In: Forest soils and Forest land management. In: Bernier, B and C.H. Winget, eds. Proceedings of the 4th North American Forest Soils Conf.. Laval Univ.,

Quebec. August 1973. Les Presses de l'Universite Laval, Quebec Canada.267282.

Youngberg, C.T., M.E. Harward, G.H. Simonson, D. Rai, P.C. Klingeman, D .W. Larson, H.K. Phinney, J.R. Bell. 1971. Hills Creek Reservoir turbidity study. Water Resources Research Institute. Oregon State Univ., Corvallis, OR. Report WRRI