In Reply Refer To:
Mail Stop 415

October 28, 2009

OFFICE OF SURFACE WATER TECHNICAL MEMORANDUM 2010.01 and OFFICE OF WATER QUALITY MEMORANDUM 2010.04

SUBJECT: Publication of Techniques and Methods Report Book 3, Chapter C4 “Guidelines and Procedures for Computing Time-Series Suspended-Sediment Concentrations and Loads from In-Stream Turbidity-Sensor and Streamflow Data,” by Patrick P. Rasmussen, John R. Gray, G. Douglas Glysson, and Andrew C. Ziegler

The purposes of this memorandum are to:

1. announce the on-line availability of Techniques and Methods Report Book 3, Chapter C4 (T&M 3C4; Rasmussen and others (2009)), “Guidelines and procedures for computing time-series suspended-sediment concentrations (SSCs) and suspended-sediment loads (SSL) from in-stream turbidity-sensor and streamflow data," at http://pubs.usgs.gov/tm/tm3c4/,
2. provide policy, with background information, for use of turbidity as a suspended-sediment surrogate to compute unit and daily records of SSCs and SSLs for subsequent storage in the
   National Water Information System (NWIS), and
3. provide a synopsis of the guidelines and steps required to compute a suspended-sediment record described in detail in the subject report.

The guidelines presented in T&M 3C4 represent an alternative to the technique for computing daily SSC and SSL records described in Techniques of Water-Resources Investigations Report 3, Chapter C3 (Porterfield, 1972). Both the guidelines presented in T&M 3C4, and in Porterfield (1972), are acceptable for use in computing, storing, and making available for public release daily-value SSC and SSL data in the Automatic Data Processing System (ADAPS) of the NWIS.

Turbidity – an expression of the apparent optical properties of a water sample that cause light rays to be scattered and absorbed rather than transmitted in straight lines – is the most common in-situ surrogate measurement used to compute SSCs in U.S. rivers. T&M 3C4 provides site-specific guidelines for use of continuously recorded turbidity and streamflow data, in concert with manually collected SSC calibration data, to compute continuous unit- and daily-value records of SSCs and SSLs.

The analytical and processing steps described in these guidelines are fundamentally similar to those for computing daily records of streamflow (Rantz and others, 1982) and water-quality (Wagner and others, 2000) at a stream gage. Computation of SSCs and SSLs is predicated on the availability of approved streamflow and turbidity time-series data and SSC calibration data.

Both the Porterfield (1972) and T&M 3C4 methods are used for computing SSLs from the SSC time series; however, the methods differ as follows:

• Porterfield’s (1972) method is predicated on the manual (or computer-assisted) development of a temporal SSC time series from sample data. Development of the SSC time series normally requires estimation of SSC values associated with most of the unit-value streamflow time series. Often, through subjective interpolation and extrapolation, most of the SSC time series is estimated by the hydrographer, even when using automated techniques such as the Graphical Constituent Loading Analysis System (Koltun and others, 2004). Results from these techniques are appropriate to compute daily values of SSL. However, the uncertainties associated with the SSC- and SSL-computed values are undefined.
• The T&M 3C4 method uses the turbidity time series calibrated with SSC samples based on linear-regression techniques to compute a SSC time series associated with a streamflow time series, thereby eliminating the routine need for interpolation or extrapolation of SSC time-series values by the hydrographer. The uncertainty of the resulting computed time series of SSC is defined statistically based on regression-relation statistics.

The continuity in the surrogate SSC time series and the statistically based uncertainty characteristics of the T&M 3C4 method represent major improvements over Porterfield’s (1972) method.
For the T&M 3C4 method, quality-assured, continuous in-stream turbidity data, or both continuous turbidity and streamflow data when deemed appropriate, are used to compute SSC values using linear regression analysis between in-stream data and SSC values from collected SSC samples. This computation is predicated on the availability of a sufficiently reliable empirical relation between turbidity and SSC, or turbidity and both SSC and streamflow.

Two types of empirical relations using turbidity as the explanatory variable may be derived:

1. a simple linear regression (SLR) model using turbidity and SSC data derived from in-stream samples obtain by methods described by Edwards and Glysson (1999) and Nolan et al. (2005); or, if the model standard percentage error ≥ 20 percent,
2. a multiple linear regression (MLR) model using instantaneous turbidity and streamflow measurements and the sampled SSC values.

Computed time-series SSC values are multiplied subsequently by their paired streamflow time-series values and a units-conversion factor for computation of unit and daily values of SSLs.

The record-approval process must include review by a hydrographer experienced in developing and applying suspended-sediment-surrogate regression models. As with streamflow and continuous water-quality records, the characteristics of the model and important results from the record-computation process must be summarized in a station analysis.

Unit and daily-value SSC and SSL data computed by these guidelines and all SSC values used in the development of the regression model are stored in the NWIS. SLR models with turbidity as the explanatory variable can be entered into ADAPS for automatic computation of SSC values. Currently (2009), ADAPS does not allow for any MLR or unit-value SSL computations using streamflow. MLR and SSL unit-value data must be computed outside of ADAPS and then loaded to ADAPS through DECODES. Alternatively, for automated computation of SSC (using a SLR or MLR model) and SSL, the National Real-Time Water-Quality web page (http://nrtwq.usgs.gov/) can be used. Computed SSC and SSL and the associated uncertainties for all stored unit values of the explanatory variables are then available through this resource. Longer-term storage of SSC- and SSL-computed values should be stored in the NWIS after being computed using the NRTWQ web page. These data also can be loaded into ADAPS using DECODES.

The effectiveness of these guidelines for a given site is dependent on a number of factors, including the quality of the turbidity record; the exclusive use of a single turbidimeter or optical scatterance (OBS) meter manufacturer, make, and model for the study period; range in site SSCs and particle sizes; and the amount, range, and adequacy of the calibration data. SSC values computed using these guidelines are unreliable when in-stream turbidity measurements exceed the upper measurement range of the in-situ turbidimeter; however, there are available technologies and approaches that will improve the reliability of such data, including the measurement of turbidity using an OBS meter.

SUSPENDED-SEDIMENT LOAD COMPUTATON STEPS: The generalized steps for computing unit value SSL from continuous turbidity and streamflow data from a streamgage follow:

1. Ensure that the turbidity, streamflow, and suspended-sediment data sets for the site are free of spurious data and are otherwise reliable.
2. Develop a SLR model between concurrent instantaneous measurements relating the explanatory variable turbidity to the response variable, manually collected suspended-sediment concentrations.
3. Evaluate the SLR model standard percentage error.
   
   1. SLR analysis is preferred for sites where turbidity is the measure most strongly correlated with SSC and where model standard percentage error is <20 percent.
   2. If the model standarad percentage error is ≥ 20 percent, then a MLR model using of turbidity and streamflow as the second explanatory variable should be developed and evaluated.
4. Evaluate the MLR model: The streamflow variable is significant in the MLR model if the p-value of the partial F statistic is less than a predetermined alpha value (α) of 0.025.
   (Note: If both models are unacceptable, do not compute unit-value and daily sediment-load records using this procedure.)
5. Compute SSC time-series values using the selected model and the appropriate continuous data.
6. Compute unit and daily time-series suspended-sediment loads as the product of each paired computed unit concentration and streamflow values, and a units-conversion constant.
7. Compute and display on-line uncertainty statistics for the computed suspended-sediment concentration values based on the uncertainty associated with the regression model. This is available for computed SSC and SSL through the Internet at http://nrtwq.usgs.gov
8. Retrieve unit values of SSC and SSL from NRTWQ website and load them to ADAPS via DECODES.

Once an acceptable regression model is developed, it can be used to compute SSC beyond the period of record used in model development if there is proper continuation of sample collection and analysis. Maintenance of a long-term SSC record requires ongoing collection of turbidity and streamflow time-series data and sample collection for re-analysis and verification of the current SSC regression model. Regression models to compute SSC at a site should never be considered static, but rather considered to represent a finite period in a continually dynamic system in which additional data will help verify any change in SSL, type, and source.

RELEVANT PUBLICATIONS:

Edwards and Glysson, 1999, Field methods for measurement of fluvial sediment: U.S. Geological Survey Techniques of Water-Resources Investigations Book 3, Chapter C2, 89 p.(http://pubs.usgs.gov/twri/twri3-c2/).

Koltun, G.F., Eberle, Michael, Gray, J.R., and Glysson, G.D., 2006, User’s Manual for the Graphical Constituent Loading Analysis System (GCLAS): U.S. Geological Survey Techniques and Methods Book 4, Chap. C1, 50 p. (http://pubs.er.usgs.gov/usgspubs/tm/tm4C1).

Nolan, K.M., Gray, J.R., and Glysson, G.D., 2005, Introduction to suspended-sediment sampling: U.S. Geological Survey Scientific Investigations Report 2005-5077 (available on CD-ROM and at: http://pubs.er.usgs.gov/pubs/sir/sir20055077).

Porterfield, George, 1972, Computation of fluvial-sediment discharge: U.S. Geological Survey Techniques of Water-Resources Investigations Book 3, Chapter C3, 66 p. (http://pubs.usgs.gov/twri/twri3-c3/).

Rantz, SE. and others, 1982, Measurement and computation of streamflow: volume 2. Computation of Discharge: U.S. Geological Survey Water-Supply Paper 2175, pp. 285-631 (http://pubs.usgs.gov/wsp/wsp2175/).

Rasmussen, P.P., Gray, J.R., Glysson, G.D., and Ziegler, A.C., 2009, Guidelines and procedures for computing time-series suspended-sediment concentration and loads from in-stream turbidity sensors and streamflow data: U.S. Geological Survey Techniques and Methods Report 3 C3, 54 p.(http://pubs.usgs.gov/tm/tm3c4/).

Wagner, R.J., Boulger, R.W., Jr., Oblinger, C.J., and Smith, B.A., 2006, Guidelines and standard procedures for continu­ous water-quality monitors—Station operation, record computation, and data reporting: U.S. Geological Survey Tech­niques and Methods 1–D3, 51 p. plus 8 attachments (http://pubs.water.usgs.gov/tm1d3/).

************************************************************************