Fanno Creek Study Reach

Fanno Creek, one of several major tributaries to the Tualatin River, originates within the City of Portland and flows for about 24 kilometers (km) through the Washington County suburbs of Beaverton, Tigard, and Durham before its confluence with the Tualatin River (fig. 1). The Fanno Creek Basin is approximately 86-percent developed, 13-percent forest and shrubs, and 1-percent wetlands (U.S. Geological Survey, 2011). The soils generally are highly consolidated silts and clays, and infiltration rates are relatively low, with moderately high soil phosphorus availability (Kurahashi and Associates, Inc., 1997).

The streamflow in Fanno Creek is typical of an urban stream in the Pacific Northwest, with flashy, high flows occurring during winter rain events and low summer flows dominated by groundwater inputs (McCarthy, 2000; Sobieszczyk and others, 2018). Mean monthly streamflow measured at the Fanno Creek at Durham, Oregon streamgage (14206950; U.S. Geological Survey, 2020; fig. 1) in December and January (the typical high-flow months) for water years 1994–2019 was 2.9 and 2.7 cubic meters per second (m³/s), respectively, with a maximum peak streamflow of 47.6 m³/s. Low streamflows typically occur in July and August, with mean monthly flows of 0.2 m³/s in both months and a minimum daily mean flow of 0.03 m³/s.

The Fanno Creek study reach is at the upstream section of Greenway Park (fig. 2). Greenway Park is an urban park frequented by bikers, commuters, and nearby residents. Keith and others (2014) reported that the Greenway Park segment of Fanno Creek was one of the largest potential sinks for fine sediment along Fanno Creek, owing to its low and wide floodplain and vegetation that increases channel roughness and decreases stream velocity. The study reach begins at the bridge on SW Hall Boulevard (Upstream at Fanno Creek [UF]) and has a drainage area of 26.7 km² at this upstream boundary (U.S. Geological Survey, 2021; table 1). The reach extends 560 meters (m) downstream to the ponded site at SW Tuckerwood Court (Ponded at Fanno Creek [PF]), then 390 m to the farthest downstream site at SW Pearson Court (Downstream at Fanno Creek [DF]; fig. 2) for a total length of about 940 m. The channel between SW Hall Boulevard and the south pond is narrow (2–3 m), with mean depths of about 1.5 m, and is shaded by some riparian vegetation.

From 2008 to 2012, Clean Water Services re-meandered the channel (once straightened for agricultural purposes), creating a new channel and leaving the remnant old channel (referred to as the “old channel”; fig. 2) as a floodplain feature. The old channel is shaded, narrow, and generally can be waded, except for some areas upstream from the two beaver dams. The entire old channel remains wet year-round, owing to stormwater run-off, contributions from groundwater, and backwater from the two beaver dams. Water velocity throughout the study reach and old channel is slow during low-flow periods (less than 0.01 meter per second [m/s]; White and others, 2025b). As part of restoration efforts, large woody debris also was added to the main channel, and native vegetation was planted (Smith and Ory, 2005). Starting in 2012, beavers began colonizing the reach and built several dams and a beaver lodge.

One dam (referred to as the “long dam”) was 30-m long and extended onto the floodplain, creating a large pond (referred to as the south pond; fig. 2). During high-flow events, the south pond can flood the main channel, nearby paths, riparian areas, and the old channel. The north pond and adjacent Koll Center Wetland are other floodplain features in the reach that were present prior to the construction of beaver dams. The long dam creates the south pond, which is approximately 7,900 square meters (m²) in wetted surface area (Smith and others, 2025). Here, the main channel is still narrow and deep, and the ponded area is generally shallow (0.5 m on average) with some riparian shade. Downstream from the south pond to the end of the study reach, the channel is again narrow, deep, and shaded by riparian vegetation with one beaver dam that spans the channel upstream from DF (fig. 2).

During high flows, Fanno Creek occasionally overtops its banks, and water flows into the Koll Center Wetland to the southeast (fig. 2), as observed during suspended-sediment sample collection and streamflow measurements. This high-flow, overland connection between Fanno Creek and the Koll Center Wetland complicated the ability to obtain representative SS samples and streamflow measurements and may lead to some underestimation of the SSLs at DF during those events. The overland flow occurred during an estimated 6 days during the study (Poor, 2018).

Bronson Creek Study Reach

Bronson Creek originates in the Tualatin Mountains on the eastern part of the Tualatin River Basin (fig. 1). It flows for about 10.5 km before it merges with Beaverton Creek, which flows into Rock Creek, and then continues to the confluence with the Tualatin River near Hillsboro. The Bronson Creek drainage encompasses an urbanizing region of Washington County and is about 67-percent developed land, 23-percent forest and shrubs, 4-percent herbaceous, 1-percent wetland, and 5-percent other land uses (U.S. Geological Survey, 2011). Unlike the Fanno Creek Basin, the Bronson Creek Basin does not have a long-term USGS streamflow station, and hence no available record of streamflow data. However, like the Fanno Creek Basin, increasing flows in Bronson Creek occur during rainfall runoff events primarily in the fall and winter months. The Bronson Creek Basin is about 12.7 km² and includes parts of the cities of Portland, Beaverton, and Hillsboro (U.S. Geological Survey, 2021). Downstream from the confluence with Beaverton Creek, the channel gradient decreases, and the floodplain widens, causing sediment deposition into the basin and adjacent wetland areas.

Clean Water Services planted riparian vegetation and added large woody debris to the channel within the study reach from 2004 to 2019 to improve water quality and enhance the stream and floodplain habitats (Smith and Ory, 2005). Beaver dam building is believed to have begun sometime around 2010 following restoration activities. Between NW Laidlaw and NW Kaiser Roads, at least six beaver dams were present along the main channel at the start of the study, and beavers built a new dam during the study near NW Laidlaw Road (fig. 3). The total number of beaver dams between the monitoring site at Upstream at Bronson Creek (UB) to NW Kaiser Road is unknown, owing to the dense stand of reed canary grass (Phalaris arundinacea) that hindered site access.

The Bronson Creek study reach has an upstream contributing area of about 5.2 km² at its upstream boundary (U.S. Geological Survey, 2021; table 2) and continues downstream through a 2,300-m reach surrounded by a vegetated riparian zone and residential properties (fig. 3). The study reach begins at the monitoring site at Farther Upstream at Bronson Creek (FUB) and flows for about 650 m to the first continuous monitoring location at Upstream at Bronson Creek (UB), then 630 m to the ponded site at Ponded at Bronson Creek (PB), and then 980 m to the final monitoring location at NW Kaiser Road (Downstream at Bronson Creek [DB]; fig. 3). The upstream part of the Bronson Creek drainage basin near the forested foothills is confined to a narrow (2-m wide) channel with various depths and ponded by multiple beaver dams (fig. 3). In the downstream part of the reach, Bronson Creek splits into braided channels across the wide floodplain and wetland area near Northwest Kaiser Road (fig. 3). While water velocity is slow during low flows (0.01 m/s; White and others, 2025b), the stream banks often remained full at PB and DB.

Comparisons of Turbidity Within the Study Reaches

Turbidity data were evaluated for each study reach to gain insights into the effects of beaver dams on longitudinal changes in sediment transport. Differences from upstream to downstream sites could indicate whether the dams were acting as a sink (positive difference) or source (negative difference) for turbidity and SS.

The peak turbidity values measured at the continuous monitoring sites during storm events were compared for a given reach. A discrete storm event was characterized by a rising limb and falling limb in the hydrograph and a peak streamflow that exceeded 5.7 m³/s at Fanno Creek at Durham, Oregon (USGS streamgage 14206950; U.S. Geological Survey, 2020). A storm event was not included in the comparison if one continuous site along a reach was missing the potential maximum turbidity measurement because of sensor fouling or equipment malfunction that resulted in missing data. Because Bronson Creek does not have a USGS streamgage, a visual inspection of the time-series water-level data measured at DB during an event verified that a corresponding storm event hydrograph occurred in the Bronson study area as well.

Estimations of Suspended-Sediment Loads

Many studies have used continuous turbidity measurements, streamflow measurements, and SSC samples to calculate suspended-sediment loads (SSLs) in rivers (Wass and others, 1997; Anderson, 2007; Bragg and others, 2007; Chanson and others, 2008; Schenk and Bragg, 2014; Sobieszczyk and others, 2015). This process involves continuously measuring turbidity and streamflow while simultaneously collecting water samples, which are later analyzed for SSC.

Ordinary least squares (OLS) regression analysis was used to describe the relation between continuous turbidity and SSC. Site-specific linear regression models were developed, and models within a reach were tested for statistical significance. The resulting reach- or site-specific regression subsequently was used to calculate continuous SSC (Rasmussen and others, 2008). The SSC and streamflow values were used to compute monthly and annual SSLs by multiplying the calculated SSC by the respective streamflow (Rasmussen and others, 2009).

Ideally, at least 25–30 SSC samples would be used for a given sediment-turbidity regression to provide sufficient statistical power (Rasmussen and others, 2009). For this investigation, we collected nine samples at each site along both study reaches. Although the total number of samples is relevant, the distribution of data over the range of observed turbidity, SSC, and streamflow values for the site is most important (Rasmussen and others, 2009). SSC samples collected from the four sites are representative of conditions during low, medium, and high streamflows (fig. 5A–D) and cover the full range of turbidity and SSC during the study period.

For this study, the analyses necessary to compute SSLs upstream and downstream from the study reaches included:

Estimation of Turbidity Data Gaps

Data values were deleted or are missing from the continuous turbidity datasets from both reaches owing to sensor fouling, sensor failure, or other equipment malfunctions, which created gaps in the continuous record. Most instances of missing data occurred on falling limbs of the hydrographs following a storm event that transported sediment and resulted in sensor fouling or burial. Deleted and missing data were less than 20 percent of the data at each monitoring site: UF, 18 percent; DF, 13 percent; UB, 13 percent; and DB, 19 percent.

Multiple approaches were considered to fill the turbidity data gaps for the Fanno Creek sites, including relating the site turbidity (for example, at UF) with streamflow data from Fanno Creek at Durham, Oregon (fig. 1), local water-level data, and site turbidity from PF and DF. Site turbidity and streamflow at the Durham station were fairly well correlated (UF, coefficient of determination [R²]=0.60; DF, R²=0.59), as was site turbidity and local water-level data (UF, R²=0.67; DF, R²=0.62). However, the strongest correlations were between turbidity measurements at the upstream, ponded, and downstream locations (fig. 7). Thus, turbidity data gaps at UF were first filled using the ordinary least squares (OLS) regression between turbidity at PF and UF (R²=0.92; mean absolute error [MAE]=5.03; fig. 7A). After that step, 4 percent of turbidity data were still missing from the UF dataset, which occurred when turbidity data were missing from UF and PF. The regression between DF and UF turbidity (R²=0.78; MAE=8.80; fig. 7B) was used to estimate the remaining missing values at UF. Similarly, missing turbidity data at DF were filled using the regression between turbidity at PF and DF (R²=0.79; MAE=7.24; fig. 7C), when possible, and then the UF-DF regression (R²=0.78; MAE=8.01; fig. 7D), when needed. After applying these regression equations to fill turbidity data gaps, UF and DF were missing less than 1 percent of the turbidity dataset, which occurred when data were missing from all three sites.

Like the Fanno Creek sites, turbidity was not strongly correlated with water level for the UB and DB sites (OLS regression R²=0.03 and 0.14, respectively). The lack of correlation between water level and turbidity at UB may have been a result of the beaver dam located just downstream from the continuous monitor. DB was in a wide floodplain, and the creek within the channel was near bank full capacity, even during low-flow periods. Therefore, when water level increased at the DB site, water likely spread over the wide floodplain and sediment dropped out of suspension, changing the expected water-level-turbidity relation for a free-flowing stream.

Methods used to fill turbidity data gaps for the Fanno Creek reach were applied to the Bronson Creek reach, and regression equations of turbidity measurements among sites were developed (fig. 8). Data gaps at UB were first filled using the regression between turbidity at UB and PB (R²=0.69, fig. 8A). If PB data also were missing, the regression equation between turbidity at UB and DB (R²=0.30) was used to estimate missing turbidity values at UB (fig. 8B). Turbidity gaps at DB were filled using the regression between turbidity at DB and PB (R²=0.45; fig. 8C), when possible, and then using the regression between turbidity at DB and UB (R²=0.30) if PB turbidity data were unavailable (fig. 8D). Final UB and DB turbidity datasets were missing less than 1 percent of a complete dataset. Estimated turbidity values for the Fanno Creek and Bronson Creek sites introduced uncertainty in the turbidity dataset and, therefore, in the SSL calculations (app. 3).

Relation Between Suspended-Sediment Concentrations and Turbidity

OLS regressions were used to compare SSCs, and average turbidity values measured during the water sample collection period from each site (Fanno and Bronson Creeks, upstream and downstream locations), creating four regression models. All regressions were forced to fit through zero to remove negative results. An Analysis of Covariance (ANCOVA) test was used to compare relations among upstream and downstream sites (significance level α=0.05). The null hypothesis stated that there was no difference between the simple linear regression models; in other words, the data from upstream and downstream sites within the study reach could be described using one regression line. For an ANCOVA test, the data are assumed to be independent and normally distributed with the same variance. Some statistical-test assumptions were violated because measurements collected from upstream to downstream were not independent. Despite the violation of assumptions, ANCOVA tests were used to accentuate the relative differences among groups.

Similar regression slopes at UF and DF (1.04 and 1.01, respectively; fig. 9A) indicate that the relation between SSC and turbidity does not change along the beaver-affected reach and that the same type of suspended material (such as clay) is found at both sites. ANCOVA results showed that the two regressions were not significantly different (probability value p=0.58), and that the data could be explained with one model. ANCOVA results were used as another piece of evidence that, when considered with the very similar slopes, indicate that the regressions are not significantly different. Therefore, the UF and DF datasets were combined to increase sample size (fig.9B; R²=0.83; MAE=39.91; sample size n=18).

The site regression that was formed with SSC data collected at FUB and turbidity data collected at UB is referred to as the UB regression. Two pairs of SSC/turbidity values were not included in the UB regression because the site turbidity sensor was fouled when the SS samples were collected. However, an additional SS sample was collected at the upstream site on February 16, 2017, approximately 1 hour after the primary sample was collected. This sample was included in the UB regression because conditions were changing quickly, and it represented a different sample.

At Bronson Creek, the upstream and downstream regressions differed significantly from each other (p=0.02); this difference can be seen visually in figure 10 and was confirmed with the ANCOVA test. These different regressions indicate that the beaver reach affected the SSC/turbidity relation, potentially by altering the particle-size distributions. The expansive floodplain, reed canary grass, and multiple beaver dams in the reach affected the suspended material as the water flowed from upstream to downstream. Less sediment was flowing past DB than UB per unit of turbidity, and the suspended material at DB may have been mostly fine material in contrast to larger particles that likely would drop out of suspension as stream velocity decreased over the period of the storm hydrograph. These datasets were not combined, resulting in separate regressions for UB (R²=0.91; MAE=127.30; n=8) and DB (R²=0.94; MAE=7.31; n=9; fig. 10). The applicable regression model was used with each site’s continuous turbidity data to compute continuous SSCs.

The high SSC value (1,380 milligrams per liter [mg/L]) at UB may seem erroneous, but it was included in the analyses for two reasons. First, the sample was collected at FUB while the corresponding turbidity value was measured onsite at UB; the sample and turbidity measurement indicate that it was a high SS event. Second, the sample was collected on March 15, 2017, and the SSCs at DB and UF on that date were also the highest measured during the study at those sites. The sample was collected on the falling limb of the hydrograph at UB, and the concentration and turbidity values could have potentially been higher if the sample had been collected a few hours earlier during the rising limb of the storm hydrograph.

Water-Level-Streamflow Rating Curves

Most USGS rating curves are developed over time, and the rating is continually modified and updated as additional streamflow measurements are collected. However, ratings curves for this project were developed using only the streamflow measurements and water-level readings collected during the study period. Water levels collected with pressure transducers were paired with corresponding streamflow measurements. Water-level-streamflow rating curves were developed using the USGS rating-development tool.

A water-level-streamflow rating curve could not be developed using the continuous water-level data collected at UB and the streamflow measurements collected at FUB because of the beaver dam constructed downstream from the pressure transducer at UB. The dam resulted in an inconsistent relation between water level and streamflow, making the use of a rating curve unsuitable for this study. Thus, streamflow at UB was estimated by using the drainage-area ratio method (DARM) based on the rating developed at DB (eq. 1). The DARM assumes that the streamflow for a site of interest can be estimated by applying the ratio of the drainage area for the site of interest to the drainage area of a nearby site (Emerson and others, 2005). Because of the upstream influence of the south pond and long dam in the Fanno Creek study reach (fig. 2), the DARM also was used to estimate streamflow at the downstream Fanno Creek site (DF) using the rating developed at UF (eq. 2). Thus, the DARM used to estimate streamflow at UB is given by:

The DARM used to estimate streamflow at DF is given by:

The drainage areas of DB and UB are 8.05 and 5.30 km², respectively, and a ratio of 0.66 was applied to the continuous streamflow at DB for estimating streamflow at UB. The drainage areas of UF and DF are 3.98 and 4.05 km², respectively, and a ratio of 1.02 was applied to the continuous streamflow at UF for estimating streamflow at DF. Streamflow measurements from both study reaches are listed in appendixes 2 and 3.

Suspended-Sediment Load Calculations

SSLs were calculated from the modeled SSCs and estimated streamflow using the following equation:

Loads were summed by month and water year (2017) for comparison purposes. Potential sources of error in the computed SSLs were assessed and documented in appendix 3. Sources of error in this analysis included:

Probable overall error associated with the daily computations of SSLs was 59.8, 56.4, 69.4, and 52.3 percent at UF, DF, FUB/UB, and DB, respectively (table 3.1). For all sites, most of the error in the daily SSL calculations was a result of the regression models relating SSC to turbidity (table 3.1). The second greatest source of error was the relation used to develop continuous estimates of streamflow. Error associated with the annual load computations ranged from 1.8 to 4.3 percent (table 3.1).

Calculation of Stored Sediment in the South Pond of the Fanno Creek Reach

The volume of sediment stored in the south pond along the Fanno Creek reach (fig. 2) was determined using bed elevation and EAR surveys. Initial inspection of the two datasets indicated that changes in bed elevation occurred in the main channel and western and northern sides of the pond during the 4 months between the two surveys. These changes likely were a result of ongoing beaver activity in the study reach. Thus, subsequent analyses were limited to the eastern part of the south pond (fig. 11) where changes in bed elevation were not evident in the data. The area of analysis captured about 60 percent of the pond wetted surface area. The bed elevation points and EAR survey points within the area of analysis were converted to triangulated irregular networks and then converted to a raster using standard tools in ArcGIS. The EAR raster was then subtracted from the bed elevation raster (refer to fig. 6) to create a sediment-depth raster.

Sediment volumes in the eastern part of the south pond of Fanno Creek were estimated using two methods. First, the volume of stored sediment was determined by multiplying the sum of the sediment depths by the 0.5-meter resolution raster, using linear interpolation. Second, the volume of stored sediment was estimated by multiplying the pond area by mean sediment depth so that results could be compared to those reported previously by Butler and Malanson (2005). These two estimates of sediment volume were converted to mass (in metric tons) by applying a bulk density of 0.23 grams per cubic centimeter (g/cm³). A bulk density of 0.23 g/cm³ was selected based on site knowledge and reported literature values for similar sites (Puttock and others, 2018).

Fanno Creek Study Reach

For the period of record, turbidity values measured at UF and PF were similar and overlapped (fig. 12). However, turbidity values measured at PF and DF varied substantially. Turbidity was affected by the differences in stream characteristics along the reach. The stream remained within the channel between UF and PF, which likely explains the similar turbidity measurements in these sections of Fanno Creek. In contrast, the beaver dams downstream from PF caused the creek to overflow its banks and form the large south pond (fig. 2) upstream from DF.

During storm events (n=25), the peak turbidity values measured at DF were substantially less than at PF (fig. 13). The median peak turbidity values measured during storm events were 197, 212, and 138 FNU at UF, PF, and DF, respectively (fig. 13). Additional evidence of this pattern can be seen during a storm event on March 15, 2017, where the peak turbidity values were 257, 281, and 204 FNU from upstream to downstream (fig. 14). These peak storm values suggest that sediment was dropping out of suspension in the large south pond between PF and DF.

A pronounced diel pattern of higher turbidity during the night and early morning was observed in DF data during periods of low flow compared to UF and PF turbidity data (fig. 15). The pattern suggests that the area between PF and DF was affected by nocturnal animal activity and bioturbation, such as dam building and maintenance by beavers and nutria. Although the turbidity values were higher and suggested that more sediment was in suspension at DF, this activity occurred during baseflow conditions when streamflow and velocity were very low.

Bronson Creek Study Reach

Turbidity values generally decreased from UB to DB along the Bronson Creek reach during storm events (figs. 16–18) suggesting that SS was deposited on the floodplain between the UB and PB sites and between the PB and DB sites. Turbidity values at UB were often much greater than DB values (for example, 1,000 FNU higher), suggesting substantial sediment deposition along the reach between UB and DB. The decreasing downstream turbidity values suggest that beaver dams in the Bronson Creek study reach, especially dams at the two downstream sites, served as a sink for sediment.

Peak turbidity values measured during storm events (n=23) at UB were higher than at PB and DB (fig. 17). Median peak turbidity values measured during storm events were 327, 159, and 62 FNU at UB, PB, and DB, respectively. Further evidence of decreasing turbidity from upstream to downstream can be observed during a storm event on March 15, 2017, where turbidity peaked at 864 FNU at UB, 550 FNU at PB, and 185 FNU at DB (fig. 18). The timing of the turbidity peak varied among the sites indicating the travel time of water through the 1,610-m study reach. During the March 15 event, stormwater traveled from UB to PB (630 m) in about 30 minutes and then from PB to DB (980 m) in about 2.5 hours (Smith and others, 2025).

Turbidity measured at DB during low-flow conditions (July–September 2017) was higher than turbidity measured at PB and UB (fig. 16). The diel pattern in turbidity observed in the Fanno Creek study area (fig. 15) was not observed in the Bronson Creek study area. The specific cause of the elevated turbidity readings during summer conditions (and negligible streamflow) at DB is unknown but could be attributed to beaver activity or increased phytoplankton in the warmer, slower-moving water (Carpenter and others, 2022).

Fanno Creek Study Reach

SSCs for UF and DF are shown in figure 19A and listed in appendix 2. The first eight samples collected at UF and DF were collected at estimated streamflows of 0.06–8.60 m³/s and 0.09–7.10 m³/s, respectively, and during the rising and falling limbs of storm events (fig. 5). Suspended-sediment concentrations for the storm events (samples 1–8) ranged from 27 to 301 mg/L at UF and 21 to 191 mg/L at DF (fig. 19A). Concentrations for the UF and DF sites were similar during four storm events (samples 1, 2, 5, 7). However, concentrations at the UF site were 72–126 mg/L higher than those at DF during three storm events (samples 3, 4, 6) but 42 mg/L lower than at DF during one storm event (sample 8). The greater SSC measured at DF in sample 8 was likely an artifact of sampling at UF and then DF as the streamflow increased. Overall, there was an average decrease of about 20 percent between the upstream and downstream sites. The baseflow sample (sample 9) at UF and DF had comparable SSCs (26 and 22 mg/L, respectively; fig. 19A) and streamflows (0.06 at UF and 0.09 m³/s, respectively; fig. 5). Because there was only one sample to compare SSCs during baseflow, it is unclear if this similarity in SSC was consistent during all baseflow conditions. However, the volume of sediment transported during baseflow conditions is a small fraction compared to sediment transport during storm events and higher streamflow conditions.

Bronson Creek Study Reach

The SSC results for FUB and DB are shown in figure 19B and listed in appendix 2. The first eight samples collected at FUB and DB were collected during storm events with estimated streamflows of 0.01–1.66 m³/s and 0.05–2.35 m³/s, respectively, and during the rising and falling limbs of storm events (fig. 5). The SSC values for the storm events (samples 1–8) ranged from 22 to 1,380 mg/L at FUB and 8 to 121 mg/L at DB. Concentrations at the upstream and downstream sites were similar during three storm events (samples 1, 7, 8). Concentrations at the upstream site ranged from 142 to 206 mg/L during four storm events (samples 2, 3, 4, 5) and measured 1,380 mg/L during one storm event (sample 6; app. 2). Overall, there was an average decrease of about 67 percent between upstream and downstream sites. The baseflow samples (sample 9) at the upstream and downstream sites were both 9 mg/L and were collected during estimated flows of 0.01 at FUB and 0.05 m³/s at DB.

Quality Assurance of Suspended-Sediment Samples

Replicate samples were submitted to the USGS Cascades Volcano Observatory Sediment Laboratory to assess accuracy and reproducibility of analytical SSC results. The replicate samples were collected concurrently using EWI methods to eliminate the potential error of splitting or subsampling the primary sample. A total of eight replicate samples were submitted for analyses. The relative percent difference between the samples ranged from 1 to 12.2 percent with a mean of 6.4 percent (app. 1). The highest two relative percent difference comparisons were from replicate samples collected in October 2017 from FUB (12.2 percent) and at DF (12.0 percent), and the lowest was from UF in March 2017 (1.0 percent). The remaining comparisons were all 7.1 percent or less. Overall, analytical results of the replicate samples showed good reproducibility by the Cascades Volcano Observatory Sediment Laboratory.

Fanno Creek Study Reach

SSLs were calculated for the Fanno Creek study reach when both water level and turbidity data were collected (June 22, 2016–November 29, 2017; fig. 20A). High SSLs occurred during autumn 2016 through spring 2017 and in autumn 2017, showing that the loads were associated with the effects of increased stream velocity and shear stress. In water year 2017, estimated monthly SSLs ranged from 1 to 578 t at UF and 1 to 519 t at DF (fig. 20B). Monthly SSLs often were greater at UF than DF during October–April, and SSLs were small to negligible at both sites during summer months. During water year 2017, the estimated total SSLs at UF and DF were 1,370±27.4 t and 1,230±22.1 t, respectively (table 3; app. 3). The difference in SSLs shows that about 11 percent (about 140 t) of the SS entering the Fanno Creek study area deposited within the reach.

Bronson Creek Study Reach

SSLs were calculated for the Bronson Creek reach from July 1, 2016, through November 29, 2017 (fig. 21A). Like the Fanno Creek reach, SSLs in Bronson Creek were higher in the wetter months (October–April) than in summer months. Estimated monthly SSLs in water year 2017 ranged from 1 to 534 t at UB and 1 to 62 t at DB (fig. 21B). Monthly SSLs were substantially greater at UB than at DB in all months except for July and August 2017. The highest monthly load calculated at UB was in February, and that load was more than eight times greater than at DB. The SSLs at UB and DB in July and August were minimal (less than 2 t at both sites) when streamflow was lowest and resulted in minimal sediment transport.

In water year 2017, approximately 1,600±69 t entered the Bronson Creek study reach and approximately 180±5.8 t exited the study reach, indicating that 89 percent of the sediment entering the Bronson Creek study area was deposited within the reach (table 3; app. 3). The high percentage of deposition is likely a result of floodplain characteristics: (1) an abundance of reed canary grass which increases channel roughness, slowing water velocity and leading to increased deposition (Martinez and McDowell, 2016); (2) change in slope through the reach; (3) depositional and wetland areas; and (4) the presence of several beaver dam complexes within the reach (fig. 3).

There were notable instances when instantaneous SSLs were substantially greater entering the Bronson Creek study reach compared to the Fanno Creek study reach (fig. 22A). However, moderate to high streamflow events occurred more frequently in the Fanno Creek reach (fig. 22B), owing to the reach’s position in the valley bottom and higher percentage of impervious area (tables 1–2). Although instantaneous loads were sometimes much higher at UB than UF, the magnitude of SSLs entering the two reaches during water year 2017 were comparable owing to the frequency and duration of storm runoff events in Fanno Creek (table 3).