Lower Gales Creek Enhancement Planning
Geomorphic Assessment • Technical Study

Table of Contents
1.Introduction
2. Setting
3. Channel Conditions
4. Existing Conditions
5. Salmonid Restoration
6. Reference

4. Existing Conditions

4.1 Bank Stability

An existing conditions assessment of bank stability was conducted for the entire study reach from the Iler Creek confluence downstream to the Stringtown Bridge. Bank stability was evaluated for both the right and left banks separately and each bank was assigned an erosion potential rating from very low to extreme. Erosion potential was determined using an assessment approach adopted from Rosgen (1994). The Rosgen method is based on the assumption that the ability of a stream bank to resist erosion is primarily determined by seven components:

• The ratio of streambank height to bankfull stage,
• The ratio of riparian vegetation rooting depth to streambank height,
• The degree of rooting density,
• The composition of streambank materials,
• Streambank angle,
• Bank material stratigraphy and presence of soil lenses, and
• Bank surface protection afforded by debris, vegetation, or resistant material such as boulders or bedrock.

These seven components are evaluated in the field by measuring reach length, flow distribution, erodibility, bankfull width, bankfull width at two times the bankfull depth (i.e. – channel entrenchment), bank height, bankfull depth, sinuosity, bank angle, percent bank face protected, percent root density, rooting depth from top of bank, bank material particle size, bank material sorting, bank soil stratification, streambed material, and stream gradient. Each field parameter was determined for relatively homogeneous stream and bank segments by averaging each parameter along the segment length. Bank parameters were determined for left and right banks (looking downstream) separately to determine the final index values for each stream segment. The bank erosion potential for each field segment is then determined based on the rating table developed by Rosgen and summarized in Table 2. Adjustments are made to the final score based on bank material and bank stratification to produce a final score for each segment. The final score is then assigned a erosion potential rating of very low, low, moderate, high, very high, and extreme.

Table 2: Bank erosion potential values for measured field parameters. A total score and rating is assigned to each field segment based on the following index values.

Results for each survey segment were then tied to a GIS layer representing the measured stream segments and are displayed on an aerial photograph of the area (Figures 7a, b, and c). Table 3 provides a statistical summary of the bank stability results. Though there appears to be problematic areas of high to moderate erosion potential, no banks were mapped as having a very high or extreme erosion potential. This is primarily due to the fact that the Rosgen system is meant to be a comprehensive assessment approach for all types of landscapes. Consequently, a very high or extreme bank erosion potential rating is often reserved for heavily incised stream channel located in arid watersheds where vegetation is sparse.

Table 3: Summary statistics for bank erosion potential for right and left banks. The results suggest that many of the banks along the Gales Creek study reach are relatively stable, except in constricted channel segments or where headcuts have recently moved through.

Though the Gales Creek study reach has some very steep banks that are actively eroding, they often have some vegetation on them, are only moderately incised, or lack stratification. The primary areas of concern for significant bank erosion and retreat are associated with Reach 5 where relatively recent incision has occurred and the channel appears to be in a channel widening phase (Douglas, 1985). Bank erosion is evident elsewhere throughout the study area but appears to be confined to discrete locations such as on the outside of a meander bend or adjacent to bridges where the flow is constricted. In Reach 5, bank erosion is widespread and is likely to continue.

The Douglas model, which evaluates channel response to a series of land use disturbances, charts incision, bank erosion, and sediment production as a channel proceeds from undisturbed, through agricultural use, and on to urban development. Each of the phases contains a period of incision followed by accelerated bank erosion in response to the disturbance, whereby a new equilibrium is achieved. The period of time, known as the lag, that is required to achieve a new equilibrium depends on the degree of disturbance and watershed conditions that may result in additional hydrologic disturbance. In the Douglas model, the new disturbance is urbanization. Consequently, a new equilibrium is never achieved between the agricultural disturbance and the urban disturbance.

4.2 Hydrology

Though a number of measurement gages have historically measured streamflow on Gales Creek, there appear to be no active gages on the Creek today. The historic gages with the longest period of record on Gales Creek are gage 14204500 – Gales Creek nr Forest Grove, located near the Roderick Road Bridge, and 14204000 – Gales Creek nr Gales (Table 4). Both these gages combined constitute a record of mean daily and annual peak flow measurements that span the years 1935 to 1981. Unfortunately, as is the trend in hydrologic monitoring networks (Rodda, 1998; Lanfear and Hirsch, 1999), these gages were discontinued and do not provide a record of current streamflow conditions through the study reach.

Table 4: Streamflow data using in the geomorphic analysis for Gales Creek. Gage locations are shown in Figure 1.

Flood frequency values for a range of return periods and exceedence probability values were estimated for gage #14204500 and #14204000 and are presented in Table5 and Table 6, respectively. Flood frequency was calculated using historic annual peak discharge estimates for each of the gages according to a standard approach developed by the USGS and outlined in Bulletin 17B (USGS, 1982). Based on this approach, the 100-year discharge at the Roderick Road Bridge was estimated to be approximately 8,660 cfs. The bankfull discharge, which is typically assumed to have a 1.5-year recurrence (Rosgen, 1996; Dunne and Leopold, 1978; Leopold et. al., 1964), was estimated to be approximately 2,707 cfs for gage #14204500 and 1,660 cfs for gage #14204000. The statistical 1.5-year recurrence flow does not correspond with expected flows at bankfull indicators in the field suggesting that channel forming flows occur more frequently than the 1.5-year event. Castro and Jackson (2001) observed bankfull discharge in humid rivers in western Oregon and Washington on a 1.2 year recurrence based on gage records and infield measurements.

Table 5: Flood frequency discharge estimates for gages on Gales Creek. Flood frequencies were calculated using the methods developed by the USGS and published in Bulletin 17B (1982).

MAF = Mean Annual Flood (2.33 year recurrence); BF = Bankfull Flood (1.5 year recurrence

Table 6 summarizes mean daily flow data on Gales Creek, by month, as observed at gage sites #14204500 and #14204000. The data are presented as exceedence probabilities, by month and annually. Exceedence probability can be defined as the percentage of time a particular flow is exceeded. For example, in September at the Roderick Road gage, a flow of 13 cfs is exceeded 50% of the time. These data are very useful in analyzing summer low flow conditions, evaluating a range of options at fish passage improvement sites, and determining diversion methods for in-channel construction or restoration projects. They can also be useful when developing a rough IFIM-type analysis of instream flow and aquatic habitat relationships.

4.3 Channel Morphology / Sediment Transport

4.4.1 Sediment Transport Calculations

There are two distinct components of the sediment load in Gales Creek; these are gravel and sand/fines. Gravel (particles coarser than 2 mm diameter) generally move by sliding, rolling, or saltating (leaping or jumping) along the bed. Sand and fines (particles less than 2 mm diameter) can often be suspended by flow and kept in motion by turbulent eddies in the water column, being transported without significant grain-to-grain contact. These two components of the load are supplied from different sources, transported by different mechanisms, and deposited in different conditions. Because we are primarily interested in channel morphology and how modifications to the channel may influence formation of bed forms, such as bars, the main focus of our analysis was on the bed load portion of the overall sediment load. In addition, we were interested in understanding sediment transport dynamics within the study area with regard to delivery, transport, and deposition of coarse sediment into and out of the study area.

Bed load can be calculated in a variety of ways. The most accurate method would be to conduct detailed field investigations where bed load is measured directly during a range of storm events. Measurements are typically taken by lowering sampling equipment from a bridge and a rating curve, comparing bed load flux to discharge, is generated. This rating curve is then used to estimate daily, monthly, or annual bed load flux by applying the rating curve to a long term flow record. Though this is the preferred, and most accurate, method of estimating bed load flux, year to year variability, an inability to measure bed load during extremely high flows, and measurement variability due to hysteresis, makes this approach very expensive since it requires many years of data collection. Consequently, bed load data collected using these methods are often not available for most rivers. Long-term data sets are often only collected in intensively studied large river systems such as the Colorado, or in smaller water supply watersheds where reservoir sedimentation is a concern.

In the absence of field measured bed load data, the best approach to estimating bed load transport is to use one of a variety of bed load equations (e.g. – Meyer-Peter Muller, Ackers-White, Yang, Laursen, Tofaletti, Parker, etc). For the Gales Creek study reach, the most appropriate method for computing gravel flux is the surface-based relation of Parker (1990). This bed load transport relationship is based on the best available data set on gravel transport from a real gravel river, collected by Milhous (1973) in Oak Creek, Oregon. Parker’s analysis of the Oak Creek data set is based on the understanding that it is the surface material, rather than the subsurface material, that directly exchanges sediment with the bed load. The Parker (1990) relation specifically excludes material less than 2mm in diameter from the analysis because those grain sizes are considered to be transported by a different mechanism. The model has a rather complicated form but accounts for the entire particle size distribution of the bed surface and bed load, and thus accounts for surface armoring, and predicts the composition of bed material and bed load. Details of this model are provided by Parker (1990) and are not elaborated here.

The parameters required to run the Parker model include cross-section geometry, channel slope, discharge, grain size distribution of the bed, and a roughness coefficient. These data were developed through collection of field data and use of a hydraulic model that was run at discrete locations in the channel. One location was chosen at a representative site at the upstream end of the study area (Reach 1), just upstream of the Clear Creek confluence, and another was chosen at a representative site downstream of the Roderick Road Bridge (Reach 4), just upstream of the Roderick Creek confluence (Figure 6). The Reach 1 site was chosen to model bed load transport conditions in the steeper, transport reach located at the upstream end of the study area and was meant to represent a relative estimate of the amount of bed load that was being delivered to the study area from upstream. The Reach 4 site was chosen to model bed load transport conditions within the lower gradient, depositional reach that makes up much of the study area and was meant to represent a relative estimate of the amount of bed load that was being transported downstream. The difference in bed load transport rates between the upstream and downstream site would determine the quantity of coarse sediment being deposited in the study area.

Once the sites were selected, longitudinal profile, cross-section, and bed material data (Wolman, 1954) were collected at each of the sites (Figures 8 and 9). At the Reach 1 site, 1000 feet of channel was surveyed. At the Reach 4 site a total of 750 feet of channel was surveyed. At both sites, cross-section data were collected at distinct changes in hydraulic conditions such as at the top of a riffle and the bottom of a riffle. Where large pools were present an additional cross-section was measured in the middle of the pool. These data were then input into the U.S Army Corps of Engineers HEC-RAS hydraulic modeling system to generate the necessary output parameters under a range of flow conditions.

Output data from the HEC-RAS model was used to as input parameters to the Parker bed load transport model. Bed load flux was computed for a range of discharges to provide an estimate of mean daily bed load flux as it relates to mean daily discharge[1]. We then fitted a regression to the data to compute bed load flux for mean daily flows greater than 150 cfs. Bed load flux was considered to be zero for flows less than 150 cfs. The result of the regression for Reach 1 and Reach 4 was a power-law relationship for flows less than 1,000 cfs and 2,000 cfs, respectively, and a linear relationship for flows greater than 1,000 cfs and 2,000, respectively (Figures 10 and 11).

These relationships were then used to compute daily, annual, and long-term bed load flux using mean daily flow data for the USGS 14204500 - Gales Creek near Forest Grove gage from 1941-1956 and 1971-1981 and the WRD 14204530 from 1995 to 2004. The data was adjusted for drainage area at each of the sites to reflect potential changes in flow due to tributary inputs. The results for years 1940-1956, 1971-1981, and 1995-2004, presented in Table 7, show that movement of bed load is much higher in Reach 1 than in Reach 4. An average annual bed load flux of approximately 15,600 tons per year was estimated for Reach 1 and an average of 960 tons per year for Reach 4, using the Parker bed load transport equation. Typically, all bed load is transported during several discrete runoff events that may last on the order of a few days to a week. Actual bed load flux may be higher given that our calculations used mean daily flow rather than flow hydrographs.

4.4.2 Sediment Transport Dynamics

Our sediment transport analysis focused on two discrete locations in the channel to gain an understanding of the relative differences in bed load transport rates between reaches and how that might affect observed channel patterns, existing and future bank erosion, and past and future management of the study area. The results suggest that a significant amount of sediment delivered to the reach is deposited between the Clear Creek and Roderick Creek confluences, as postulated in Section 4.1 where we defined transport, transitional, and aggradational reaches (Table 1). Given the depositional nature of Reach 2, and the presence of historic and active off-channel gravel mining operations in this reach, much of the sediment being transported to and through Reach 1 is being deposited in this area.

Reach 5, downstream of the Roderick Road Bridge, can also be classified as a depositional reach, with increased sinuosity, presence of large bar forms consisting of gravel and cobble material, and areas of active bank erosion that’s indicative of an area that’s widening in response to aggradation. The difference between sediment deposition in Reach 2 and sediment deposition in Reach 5 is the source of the material. Reach 2 appears to be receiving coarse sediment from the upstream watershed and Clear Creek. This material is transported downstream during high magnitude, low frequency discharge events. The material is primarily derived from large landslides and debris flows.

Conversely, much of the material being deposited in Reach 5 appears to be derived locally from more frequent episodes of bank erosion. Past channel incision, resulting from downstream changes in base level and past confinement of the channel has produced steep banks that are prone to failure during moderate and high flow events. Prior to confinement and channel incision there was a delicate balance between supplied sediment from upstream and transport through the reach. The amount of sediment supplied to the reach most likely exceeded the amount of sediment being transported, producing a sinuous channel with large bar forms and a relatively stable low flow channel. Confinement of the channel and development of floodplain areas, combined with the lowering of base level downstream resulted in an increase in the sediment transport rate with the resulting channel incision. Once the new base level was achieved, which appears to have happened through this reach, excess energy associated with a confined and incised channel began the process of bank erosion that overwhelmed the sediment transport capacity of the channel. The deposited sediment has created new bar forms which have resulted in an increase in the sinuosity of the channel, resulting in additional bank erosion as Gales Creek attempts to build new floodplain and achieve a new equilibrium (Schumm, Harvey, and Watson 1984; Simon and Hupp 1986).

This process of bank erosion, leading to additional bank erosion is referred to as a positive feedback loop. It is difficult to predict the expected width of the channel with any accuracy but additional bank erosion and channel widening is expected, especially in the sections of Reach 5 that are currently straight and have not experienced significant erosion. Though this may seem counterintuitive, material recently eroded from localized bank erosion upstream and deposited in point bars will most likely migrate downstream, causing episodes of bank erosion and widening in unaffected areas.

The negatives associated with this current episode of bank erosion through Reach 5 are obvious. Landowners adjacent to the channel must deal with the unpredictable nature of Gales Creek, the potential loss of usable farmland and structures, sedimentation of downstream aquatic habitat associated with the delivery of fine sediment from eroding banks, and loss of the shading benefits that the streamside vegetation provides when the narrow riparian corridor is removed when a bank fails. There are also potential positives. Namely, channel widening is part of a natural process that the channel is responding to because of past confinement and incision. Presumably, when the channel achieves a new equilibrium width, future bank erosion will be more predictable and not as catastrophic. Additionally, when a bank erodes it not only delivers fine sediment that gets flushed downstream, but also provides clean gravel and large woody material to the channel that helps build physical habitat for fish. Gravel creates salmonids spawning habitat and large woody material can create deep pools and cover habitat for refuge and rearing habitat.

[1] Using mean daily discharge to estimate bed load flux may underestimate total bed load flux since the peak will move considerably more sediment than the daily average. Unfortunately, daily peak data is often not available. Consequently, the bed load estimated calculated for this project should only be used as a relative measure of flux.

5. Salmonid Restoration