Robin M. Harris and Caty F. Clifton¹

ABSTRACT: The Umatilla River in Northeastern Oregon provides multiple beneficial uses, including native salmon habitat, water supply, and recreation. Water quality issues include developing TMDLs for sediment and temperature. The Forest Service operates three monitoring sites in the upper watershed with 34 years of streamflow records and 11 years of suspended sediment records. These stations were analyzed, with a focus on quantifying the annual and seasonal sediment loads and on the relationships between suspended sediment and discharge. We found: 1) high spatial and temporal variability in annual sediment loads between stations, and at the same station year to year, and 2) a lack of correlation between streamflow and sediment indicating complex streamflow and sediment supply relationships. Recommendations include future analysis to evaluate the reliability of automatic fixed point sampling by collecting concurrent depth integrated samples; future analysis to determine the frequency of sampling needed to characterize the sediment parameters; and, adding bedload sampling to determine bedload contribution to total load.

KEY TERMS: Sediment yield; sediment rating equations

INTRODUCTION

The Umatilla Barometer Watershed, located in the Blue Mountains in Northeastern Oregon, was established in 1965 as part of the national USDA Forest Service Barometer Watershed Program. Seven monitoring stations were established on the Umatilla National Forest; three mainstem stations located on the North Fork Umatilla River, the South Fork Umatilla River, and the Umatilla River at Corporation, and four stations in the High Ridge Evaluation Area. The High Ridge Evaluation Area is located in the headwaters of Buck Creek, which drains into the lower South Fork Umatilla River below the gaging station. Extensive physical descriptions of the study area can be found in the Umatilla Barometer Watershed Analysis (Ross and Moore, 1965), the Umatilla Barometer Watershed, Survey-Analysis-Plan (Spandenberg, 1971), and the Watershed Hydrology for the Umatilla-Meacham Ecosystem Analysis (Clifton, 1996).

Current Issues

The development of Total Maximum Daily Loads (TMDLs) for the Umatilla River is currently in progress to address the water quality issues of sediment and temperature. Water quality monitoring is an important component in the development and validation of TMDLs. The Umatilla River is listed as water quality limited for sediment on the basis of Oregon Department of Fish and Wildlife (ODFW) stream survey information. Ten years of suspended sediment data are available for three USDA Forest Service operated gaging stations on the upper Umatilla River, and several years of data are available at three stations on the Umatilla Indian Reservation, operated by the Confederated Tribes of the Umatilla Indian Reservation (CTUIR) and the U.S. Geological Survey. These long term records can be used to identify annual and seasonal variability in sediment loads, and approximate baseline conditions for the upper watershed.

The Umatilla TMDL Technical Committee, composed of local natural resource and water quality professionals, implemented an extensive winter sediment monitoring project throughout the Umatilla Basin in 1997-98. The purpose of the project was to characterize mainstem and tributary contributions to sediment loads (Lambert and Harris, 1998). Winter weather conditions during the extensive monitoring effort were not generally representative of long term average winter weather. The ten years of sediment data (1987-97) described in this report illustrate longer term variability in the annual, seasonal, and storm sediment loads transported by the upper river.

Note on Terminology

Total suspended solids (TSS), is a concentration in mg/l, used to approximate the suspended sediment transported by the river, and used interchangeably with suspended sediment throughout this report. Suspended sediment load refers to that part of the sediment load carried in suspension by turbulent motion. Sediment discharge and transport rate refer to the mass of sediment passing a stream cross section in a unit of time (Williams et al, 1988). Unit suspended sediment load is an average area distribution, arrived at by dividing the annual sediment load by watershed area.

METHODS

Stream stage was recorded at the three gaging stations by float-sensors in stilling wells using Fisher-Porter analog-to-digital recorders. Stage was recorded on punch tapes in 15 minute intervals, and converted by a translator. Stage data were converted to discharge using the program HYDRA. Discharge measurements were used to develop rating equations. Operations and maintenance of the gaging stations were performed mainly by one technician over the period of record, and data analyzed following U. S. Geological Survey standard methods (Rantz and others, 1982).

Water samples are collected using battery-operated pumping samplers. The intakes are positioned in the deepest part of the channel, off the streambed. Samplers are programmed to collect a daily composite sample consisting of four samples drawn at six hour intervals. The samples are analyzed for specific conductance, TSS, and turbidity at the Umatilla National Forest Water Laboratory in Pendleton, Oregon. An estimate of total dissolved solids is calculated using specific conductance multiplied by 0.5.

Average daily suspended sediment loads are determined by summing the daily loads from the year and dividing by the number of days sampled. A total annual load is determined by multiplying the average daily load by the number of days in the year. This method of accounting for data gaps was preferred when the sampled points represented conditions for the year. This approach was not used if the data gaps spanned an entire season. Annual unit loads are calculated by dividing the annual load by the watershed area.

RESULTS AND DISCUSSION

Discharge

The North Fork and South Fork Umatilla exhibit different flow regimes, as illustrated by Figure 1. Watershed characteristics contribute to differences in peak flows, water yield, and low flows. The North Fork watershed is characterized by higher elevations, deeper soils, greater snow accumulations, and bedrock units oriented away from streamflow, compared to the South Fork. The North Fork tends to have lower peak flows but greater discharge over most of the year. Low flows average 30 cfs. The east-west orientation of the valley provides topographic shading, producing cooler water temperatures. The North Fork flows largely from a designated wilderness, with no streamside roads in the main valley, and intact riparian vegetation cover providing shade to the stream, further lowering water temperatures. In contrast, the South Fork watershed is characterized by relatively shallow soils, lower average elevations, and a south to north channel orientation, factors that result in lower, warmer baseflows of about 7 to 10 cfs. The road system along the South Fork Umatilla and Thomas Creek open the canopy, allowing for more solar radiation, which contributes to warmer

temperatures.

Overall, annual maximum discharge is lower on the North Fork compared to the South Fork, for the same storm event. Annual water yields were higher on the North Fork in nine out of nineteen years measured, or about half of the time.

Discharge at Umatilla at Corporation consistently reflects the sum of the South Fork and North Fork, with a small component that is attributed to Buck Creek. Low flow at Umatilla River at Corporation, at 40-50 cfs, reflects the importance of the North Fork contribution to streamflow during the late summer.

Figure 1. North Fork Umatilla River and South Fork Umatilla River

Hydrographs, Water Year 1994.

Suspended Sediment

Unit sediment loads (in mass per unit area) represent sediment production distributed over a watershed, and are useful in comparing temporal and spatial variability. Annual unit suspended sediment loads were calculated for each station (Table 1). Annual loads for the Umatilla River at Corporation ranged from 14 tons/mi²/yr in WY97, to 197 tons/mi²/yr in WY93, an order of magnitude difference in annual load. For WY88 and WY89, limited data which was not representative of the entire year was available to determine annual loads; these years were excluded from the analysis. In general, year to year variability in suspended sediment loads is the result of variability in weather conditions, storm events, sediment sources, and storage on hillslopes, floodplains, and channels.

Table 1. Annual suspended unit loads (tons/mi²) - Umatilla Barometer Watershed.

N/A Not applicable, measurements not representative of entire year
* Limited data available

In eight out of ten years, the North Fork Umatilla River produced more suspended sediment per square mile than the South Fork Umatilla River. Watershed characteristics contributing to higher unit suspended sediment loads in the North Fork watershed include higher precipitation, deeper soils, and higher sustained stream velocities. Total sediment loads are unknown for both streams without sampling the bedload component of the total sediment load. The South Fork, with its "flashier" discharge regime, may move a higher proportion of bedload than the North Fork.

The total annual suspended load at the Umatilla River was expected to be slightly higher than the sum of the North Fork and South Fork because of the ungaged contribution of Buck Creek. In seven out of eight years, total suspended load at the Umatilla River was higher than North Fork and South Fork combined, ranging from 2.2 times higher in WY95, to 1.02 times higher in WY92. The greater load could be accounted for in part by bank erosion below the North Fork and South Fork gages, by contribution from Buck Creek, and by the variability inherent in fixed-point sampling. In WY97 the total suspended load at the Umatilla River was lower than North Fork and South Fork combined (North Fork + South Fork was 1.5 times higher than Umatilla River at Corporation). Deposition below the North Fork and South Fork gages would account for the lower total suspended load at Corporation. Floodplain deposition and development of a mid channel bar above the Umatilla River at Corporation were features observed after winter floods in WY96 and WY97.

Sediment Rating Curves

Our purpose was to develop rating equations that link discharge to sediment. If acceptable predictive equations can be developed, then continuous streamflow records could be used to estimate sediment transport. In a sediment rating equation, discharge (independent variable) is used to predict sediment (dependent variable). To develop a rating equation, a regression of sediment to discharge is used. The regression yields the equation of the best-fit line and the coefficient of determination (R²). The R² describes the extent to which the regression accounts for the variance in observed values of the dependent variable.

To develop suspended sediment rating equations, TSS values are plotted against daily discharge on a log-log scale (Ketcheson, 1986). At each station, TSS values were analyzed on an annual, seasonal, and selected runoff event basis. In addition, all data points collected over the sample period were plotted together. None of the R² values were above 0.08 at the 95 percent confidence level and the correlations were not statistically significant. All methods of grouping the values showed high variability. A representative plot of TSS to discharge shows the general positive trend of increasing sediment with discharge but wide scatter of data points with many extreme low values of sediment across the range of discharge (Figure 2). Statistically significant rating equations could not be developed for TSS to discharge because of low explained variance.

Figure 2. South Fork Umatilla River, Winter, Water Years 1987-1997.

Complications to the development of sediment rating equations include hysteresis, where the sediment to discharge relationship changes from the rising limb to the falling limb of a storm hydrograph. Hysteresis results from changing availability of sediment (Dunne and Leopold, 1978). Sediment availability is subject to seasonal influences and watershed conditions (Ketcheson, 1986 and Williams et al, 1988).

Automated Pumping Samplers

Fixed-point automated pumping samplers offer high sampling frequency with low maintenance, yet there are problems associated with using these devices. The intake of the sampler is small relative to the cross section. The intake is located at a single point and does not vary with stage height. One assumption with automatic samplers is the sampled point represents the entire cross section. Since concentration gradients occur throughout the cross section, fixed point sampling only estimates the actual sediment transported through the cross section. Generally, the most representative point is above the thalweg, at least 10 cm above the streambed (Thomas, 1985).

When pumping a sample, the fixed point automatic sampler intake alters the water velocity, potentially altering the sediment concentration of the sample (Thomas, 1985). Automatic samplers should be calibrated with depth-integrated samples at each site over a wide range of streamflows to determine the relationship between cross sectional suspended sediment concentration and fixed point suspended sediment concentration. In low flow conditions, the initial backflushing of the intake line to purge any collected water or sediment may stir up sediment or algae on the streambed prior to pumping the water sample. During low flow conditions, the intakes at the three Umatilla stations are located in flowing water, minimizing this potential error source.

Fixed-interval sampling may miss peak flow events in quick response streams. Since storm events span days, setting the sampling interval at six hours is a reasonable approximation of the daily suspended sediment contribution.

SUMMARY AND RECOMMENDATIONS

High year-to-year and between station variability in sediment yields was evident from the analysis. For example, the Umatilla River at Corporation ranged from 14 to 196 tons/ mi², an order of magnitude difference in annual unit sediment loads. Comparing the tributary streams, the North Fork produces more sediment than the South Fork, even though the South Fork is a larger watershed and exhibits greater peak flows. The majority of the annual suspended load is contributed during spring snowmelt. Statistically significant rating equations for TSS to discharge could not be developed because of high variability. This data is useful to the TMDL analysis because it shows the variability of the sediment supply from the upper watershed. More data is needed for the lower Umatilla basin to determine if sediment and discharge relationships could be used.

Recommendations to improve the reliability of sediment data include: evaluating the accuracy of fixed-point sampling with concurrent depth-integrated measurements; determining the frequency of sampling needed to characterize the sediment parameters; and, measuring bedload to determine transport rates and contributions to total sediment loads.

ACKNOWLEDGMENTS

The Umatilla Barometer Watershed project was maintained through the dedication of Ed Calame and Greg Holden (retired USFS). This work was supported in part through an interagency agreement with the Environmental Protection Agency, Seattle WA.

REFERENCES

Clifton, C., 1996. Watershed Hydrology: Ecosystem Analysis of the Umatilla and Meacham Watersheds. Report on file Umatilla National Forest, Pendleton, OR, 52 p.

Dunne, T. and L. B. Leopold, 1978. Water in Environmental Planning. W.H. Freeman and Co., San Francisco, CA, 818 p.

Ketcheson, G. L., 1986. Sediment Rating Equations: An Evaluation for Streams in the Idaho Batholith. USFS General Technical Report INT-213, 12 p.

Lambert, B. and R. Harris, 1998. Total Maximum Daily Load (TMDL)Winter 1998 Sediment Monitoring Final Report, Total Suspended Solids. Report on file Umatilla National Forest, Pendleton, OR, 21 p.

Rantz, S. E. and others, 1982. Measurement and Computation of Streamflow: Volume 1. Measurement of Stage and Discharge, and Volume 2. Computation of Discharge. U. S. Geological Survey Water Supply Paper 2175, 631 p.

Ross, R. N. and M. R. Moore, 1965. Watershed Analysis: Umatilla River Barometer Watershed. Report on file Umatilla National Forest, Pendleton, OR, 145 p.

Spangenberg, N. E., 1971. Umatilla Barometer Watershed Survey-Analysis-Plan. Report on file Umatilla National Forest, Pendleton, OR, 98 p.

Thomas, R. B., 1985. Measuring Suspended Sediment in Small Mountain Streams. USFS General Technical Report PSW-83.