San Francisco Bay Delta

Measuring Suspended-sediments and Water Quality


Checking field equipment Continuous water-quality measurements are collected at stations throughout the San Francisco Bay (Bay) and Sacramento-San Joaquin Rivers Delta (Delta) using multiparameter water quality sondes. The sondes are usually deployed in the water by suspension from a stainless-steel cable anchored to the bottom (Figure 1). The sondes are equipped with sensors to measure water level (pressure sensor), temperature, specific conductance, turbidity (optical), and dissolved oxygen (optical). Data are recorded every 15 minutes and are retrieved either real time via cellular telemetry or by manual download during routine station visits.

Biological activity and growth (biofouling) interfere with sensor readings, requiring affected data to be corrected or deleted. Biofouling increases with time and generally is greatest during spring and summer; the degree of biofouling is site dependent, causing data return to vary among stations. Biofouling is mitigated by routine sensor cleaning and anti-fouling sensor equipment such as wipers for optical sensors. Every 2-5 weeks, each station is visited to clean, calibrate, and download the instruments. Sensor performance is ensured by comparing sensor output with known values; such comparison is used to identify sensor drift, calibration errors, or malfunction. For temperature, sensor output is compared to that from a NIST traceable thermistor. For specific conductance and turbidity, sensor output is checked against and, if needed, calibrated to, solutions of known value (standards). For dissolved oxygen, sensor output is checked in water-saturated air. All field procedures and data evaluation are performed according to USGS guidelines1.

Suspended-sediment concentration (SSC) measurements

field equipmentTo develop time series of SSC, we measure turbidity continuously (15-min sampling interval) and relate turbidity sensor readings to in situ SSC determined from periodic water samples. Turbidity is an optical property of the water and is a measure of the relative clarity of the water. We use optical turbidity sensors that measure the intensity of light scattered at 90 degrees between a light-emitting diode and a high-sensitivity photodiode detector; output is in formazin nephelometric units (FNU). A self-cleaning wiper helps prevent biological growth from interfering with sensor readings.

Water samples are used to relate observed turbidity to in situ SSC via a regression model. Water samples are collected by one of two methods. For sites where point SSC is desired, a Van Dorn-style sampler is lowered to the depth of the sensor by a reel and crane assembly and then triggered to collect a water sample while the sensor is collecting data. For sites at which suspended-sediment flux (sediment mass per unit time) is desired, cross-sectionally averaged SSC measurements are made using the equal discharge increment method2. In this method, the channel cross section is divided into five areas of equal discharge, which are sampled using depth-integrated sampler (D-74 or D-96). Water samples are analyzed to determine SSC at the USGS Sediment Laboratory in Santa Cruz, California. Suspended sediment includes all particles in the sample that do not pass through a 0.45-micrometer membrane filter and concentrations are recorded in milligrams per liter (mg/L). The filtrate is rinsed with de-ionized water to remove salts, and the insoluble material and filter are dried at 103°C then weighed.

The final step in computing continuous SSC is to develop a regression model relating turbidity to SSC. For each water sample collected, turbidity sensor output at the time of sample collection is related to measured SSC from the laboratory analysis. In a typical year, 10-20 samples are collected for each sensor; samples from prior years are used in the calibration when feasible. The regression model is developed for each sensor using one of three models: simple linear regression; simple linear regression using log-transformed data; or the nonparametric repeated median method3,4. For each sensor, all three models are computed and best model is selected after analyzing errors and model residuals.


1 Wagner, R. J.; R. W. Boulger Jr, C. J. Oblinger, B. A. Smith, 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.

2 Edwards, T. K., and G. D. Glysson, 1999. Field methods for measurement of fluvial sediment: U.S. Geological Survey Techniques of Water-Resources Investigations, book 3, chap. C2, 80 p.

3 Rasmussen, P. P., J. R. Gray, G. D. Glysson, and A. C. Ziegler, 2009. Guidelines and procedures for computing time-series suspended-sediment concentrations and loads from in-stream turbidity-sensor and streamflow data: U.S. Geological Survey Techniques and Methods, book 3, chap. C4, 52 p.

4 Siegel, A. R., 1982. Robust regression using repeated medians. Biometrika 69, 242244.

SF Bay Projects

California Water Conditions

Real-Time California Streamflow Conditions