USGS Washington Water Science Center
Any use of trade, firm, or product names is for descriptive purposes only and does not imply endorsement by the U.S. Government.
By Christopher A. Curran, Christopher S. Magirl, and Jeffrey J. Duda
Fluvial sediment data were collected from the lower Elwha River over a range of flows and turbidity values about 380 m downstream of the Elwha River at Diversion, near Port Angeles (USGS station No. 12046260, fig. 1). Suspended-sediment samples were collected using the equal-width-increment (EWI) method (Edwards and Glysson, 1999). Suspended-sediment samples were collected using various depth-integrated samplers approved by the Federal Interagency Sedimentation Project (FISP) and routinely used by USGS personnel (Davis, 2005). Most sediment samples were analyzed at the USGS sediment laboratory at the Cascades Volcano Observatory in Vancouver, Washington, to determine the total sediment concentration and percentage of fine-grained particles (less than 0.0625 mm). Two samples were analyzed at the USGS-certified GMA Hydrology fine-sediment laboratory in Placerville, California. All suspended-sediment data collected for the Elwha River are available through the National Water Information System (NWIS, http://waterdata.usgs.gov/wa/nwis/nwisman/?site_no=12046260&agency_cd=USGS).
Turbidity was continuously monitored from September 15, 2011, to September 10, 2013, with the exception of intermittent periods when instrument failure or excessive sedimentation or fouling occurred. Turbidity was measured using a DTS-12 nephelometric turbidity sensor (Forest Technology Systems Ltd., 2012) enclosed within a protective pipe. This mounting arrangement allowed turbidity measurements in an actively flowing part of the river channel and decreased the likelihood of debris accumulation around the sensor face or on the mounting hardware. Turbidity data were recorded at 15-min intervals and transmitted hourly by satellite from the water-quality monitoring station to the USGS Automated Data Processing System and are available at: http://waterdata.usgs.gov/wa/nwis/uv/?site_no=12046260&PARAmeter_cd=00060,00065. The operational range of the DTS-12 sensor reported by the manufacturer is 0–1,500 FNU (Forest Technology Systems Ltd., 2012), however, based on comparisons with turbidity standards and other hand-held turbidity sensors in sediment-laden conditions in the Elwha River (Wagner, 2006), precision of the DTS-12 was diminished for turbidity values greater than 1,200 FNU. For this reason, when river turbidity values exceeded 1,200 FNU, turbidity values were reported as ">1,200 FNU" and corresponding computed values of suspended-sediment concentration were reported as ">5,700 mg/L" or ">5,530 mg/L," for dates before or after February 28, 2013, respectively. USGS protocols for the operation and maintenance of continuous water-quality instruments were followed as outlined by Wagner and others (2006), and the time-series data were processed and reviewed according to established USGS policy for continuous water-quality data. All turbidity data collected for the Elwha River are available through NWIS.
Standard USGS guidelines were followed for using turbidity as a surrogate measurement for suspended-sediment concentration (Rasmussen and others, 2009). Regression equations were developed for computing the concentration of total suspended sediment and the fraction of fine-grained (less than 0.0625 mm) suspended sediment based on measured suspended-sediment concentrations and concurrently measured turbidity. The least-squares regression model was selected based on analysis of diagnostic statistics and model residuals consistent with the procedures outlined in Rasmussen and others (2009). Upper and lower confidence intervals for the regression model, as well as upper and lower prediction intervals that show uncertainty of individual estimates of suspended-sediment concentrations were determined at the 90-percent level. A time-series record of suspended-sediment concentrations was computed by applying the appropriate regression equation using 15-min turbidity time-series data from September 15, 2011, to September 10, 2013. This reporting period brackets the dates when operational turbidity sensors were exchanged with replacement calibrated sensors and checked against standards following USGS protocol, thus ensuring integrity of the turbidity data (Wagner and others, 2006).
The regression model that demonstrated the best normal distribution of residuals and associated homoscedasticity (Helsel and Hirsch, 1992) required a cube-root transformation of the suspended-sediment concentration. However, converting the cube-root regression model back into arithmetic space resulted in a bias (Helsel and Hirsh, 1992). To correct for the bias, a nonparametric “smearing” correction was used when transforming the estimates of suspended-sediment concentration and 90-percent prediction intervals back into arithmetic space (Duan, 1983).