Turbidity & suspended solids instrumentation, historically based in Port Townsend, WA Product line now supported through Campbell Scientific, Inc.
A bold yellow angular logo mark set against a deep dark background, sharp and modern
D & A Instruments
Turbidity monitors & water-quality instrumentation

How to calibrate optical backscatter sensors with field sediment samples

Optical backscatter (OBS) sensors estimate suspended sediment concentration by measuring how particles scatter or reflect light. The instrument response is immediate and useful for tracking dredging plumes, stormwater pulses, river transport, mine-site discharge and coastal disturbance. It does not directly measure milligrams per litre, however. A site-specific relationship between the optical signal and suspended solids must be established.

A multi-point calibration uses field sediment samples collected across the expected concentration range. Each sample is paired with the sensor output recorded at the same time, then analysed for total suspended solids (TSS) or suspended sediment concentration by a recognised gravimetric method. The resulting dataset is used to calculate a calibration equation and assess its uncertainty.

The quality of the result depends as much on sampling design as on the sensor. Sediment mineralogy, particle size, organic content, colour, air bubbles, fouling and flow conditions can all change the optical response. In Australia, a calibration developed in a clear Victorian river may perform poorly in a turbid Queensland flood channel or near a Western Australian mine, so local samples and local operating conditions matter.

Define the measurement range and sampling conditions

Begin by specifying what the sensor is expected to measure. Review historic turbidity or TSS data, construction schedules, dredging activity, rainfall records and discharge limits. The calibration should cover ordinary background water as well as the elevated concentrations likely during a plume, wet-season runoff event or operational upset. A narrow cluster of low-concentration samples can produce an attractive regression that fails when conditions change.

Aim for at least five concentration levels, with several replicate samples at each level where practical. The exact number depends on the project, but a useful spread usually includes the lowest reliable sensor response, normal operating conditions, moderate elevation and the upper end of the monitoring range. Avoid creating all high-concentration points by adding settled sediment to a bucket unless the test is specifically intended for a controlled laboratory check. Artificially mixed samples may have a different particle-size distribution from the natural suspension.

Record the conditions surrounding every sample. Useful fields include date, time, GPS position, depth, water level, flow velocity, weather, recent rainfall, tide, sensor orientation and visible plume characteristics. In tidal environments around Sydney Harbour, Darwin or the Port of Brisbane, collect the phase of tide and salinity as well. In inland systems affected by irrigation or episodic storms, the flow regime may be more important than the calendar season.

The sample plan should also reflect the intended monitoring location. A sensor mounted near the bed can see coarser particles than one positioned in the middle of the water column. A fixed instrument beside a dredge may experience rapidly changing concentrations, while a sensor at a compliance boundary may require cross-sectional sampling to relate its reading to the broader discharge.

Prepare the optical sensor before deployment

Inspect the sensing window, wiper, cable, connector and mounting hardware before calibration. Remove biofilm, sediment film and grease using the manufacturer’s approved cleaning method. Do not scratch the optical surface or use a solvent that can damage seals or plastics. Check that the sensor has stable power and that the logger records the raw optical output, rather than only a converted turbidity or TSS value.

Confirm the instrument’s measurement units and configuration. OBS devices may report counts, voltage, frequency or a proprietary signal. The calibration cannot be reproduced reliably if the raw response, gain setting, integration time or averaging interval is unknown. Save the firmware version and configuration file, and document whether the sensor is using a factory curve, a site-specific equation or an unscaled output.

Perform a clean-water or zero check in accordance with the equipment documentation. A zero check does not prove that the sensor is accurate across the field range; it simply identifies offset, electronic drift or contamination at the low end. If the instrument has multiple optical ranges, choose the setting that avoids saturation at the highest expected concentration while retaining adequate sensitivity at background levels.

Bubbles are a frequent source of false readings. Position the sensor so air cannot collect over the optical window, and allow turbulence to settle if samples are taken from a container. In the field, mount the instrument firmly and note whether sunlight, wave action or passing vessels could affect the measurement. A stable mechanical arrangement is particularly important for long deployments in exposed Australian estuaries and coastal worksites.

Collect paired samples across the range

The sensor reading and the physical sample must represent the same water parcel as closely as possible. Record the optical signal immediately before and after taking each sample, then use the mean or another documented summary for that sampling interval. If the concentration is changing rapidly, shorten the interval and collect several paired samples rather than treating a long period as one point.

Use clean, labelled bottles appropriate for the laboratory method. Rinse containers with sample water when the laboratory procedure permits, but do not rinse pre-preserved bottles or containers supplied for a specific analytical method. Fill bottles without trapping air where possible, cap them promptly and keep them cool and dark. Record the bottle identification, volume, sampler, location and sensor output in a field sheet or electronic form.

Collect enough water for duplicate analysis, laboratory quality control and possible re-testing. Very low concentrations may require a larger filtered volume, whereas high concentrations can overload a filter or produce an uneven deposit. Ask the laboratory to confirm the required bottle type, holding time, filtration method, drying temperature and reporting units before fieldwork begins.

Sampling technique must match the sensor position. A grab sample at the surface is not a valid reference for a sensor fixed 500 millimetres above the bed if the water column is stratified. For rivers and channels, take depth-integrated or profile samples when the goal is to estimate mean suspended solids. For dredging, sample within and outside the plume, while documenting the cutter, pump or spoil-disposal activity that produced each condition.

Determine suspended solids in the laboratory

The reference value is commonly obtained by filtering a known sample volume, drying the retained material to a specified constant mass and calculating the concentration. The laboratory should report the result as mg/L, or another agreed unit, and state the method used. TSS and total suspended sediment can differ because methods, filter types, drying conditions and treatment of coarse material vary, so the calibration dataset must use one consistent analytical basis.

Calculate concentration from the net mass gain of the filter divided by the sample volume. Include method blanks, duplicate samples and laboratory control checks. At low concentrations, the mass difference may be close to the balance detection limit; at high concentrations, insufficient mixing or overloaded filters may create bias. A duplicate that differs substantially from its companion should be investigated rather than averaged automatically.

Mix each sample thoroughly immediately before subsampling, while avoiding vigorous shaking that creates foam or changes particle aggregation. Sediment settles quickly, especially in calm bottles. If the sensor output was averaged over a moving water column but the bottle was allowed to settle before filtration, the laboratory value will no longer represent the recorded optical measurement.

Keep particle character in the dataset. Fine clay, pale quartz, dark organic matter, iron-rich laterite and shell fragments can produce different backscatter for the same mass concentration. This matters in Australian settings ranging from red-brown Pilbara mine runoff to fine tropical mud near Cairns and clay-rich catchments feeding the Murray–Darling system. A single calibration may need separate equations for distinct sediment sources or seasons.

Build and test the calibration equation

Create a paired dataset with one row for each valid observation: time, raw OBS signal, laboratory concentration and relevant field notes. Plot concentration against the optical signal before fitting a model. The graph can reveal a zero offset, saturation, curvature, outliers or separate clusters caused by different sediment types. Do not delete unusual points solely because they weaken the correlation; first check the sample identity, laboratory record and field conditions.

A linear model is often a practical starting point:

TSS = a × optical signal + b

Here, a is the slope and b is the intercept. Some sensors and sediment types require a polynomial, logarithmic or segmented relationship, particularly when the response becomes non-linear at high concentration. Choose the simplest model that describes the data without producing implausible behaviour outside the measured range. A high coefficient of determination alone is not proof of a sound calibration.

Reserve some paired observations for validation, or use cross-validation when the dataset is small. Report bias, root mean square error, mean absolute error and prediction intervals across the operating range. Check residuals against concentration and time. A model that is accurate at 10 mg/L but poor above 500 mg/L is unsuitable if high-concentration plume events are the main management concern.

Treat extrapolation as a risk. If the field instrument records a value above the highest calibration sample, flag the derived concentration or report that it is outside the validated range. Recalibrate when the sediment source changes substantially, the sensor is relocated, a dredging method changes, or routine checks show drift. In projects governed by state or territory environmental approvals, keep the equation, raw data, laboratory certificates and validation results as part of the monitoring record.

Verify performance in the field and maintain traceability

After fitting the equation, return the sensor to representative water and collect independent check samples. These should not simply repeat the points used to calculate the curve. Compare the converted sensor result with the laboratory concentration and investigate systematic differences. A consistent high bias may indicate fouling or an unsuitable intercept; scattered differences may reflect heterogeneous sediment or poor pairing between the sample and sensor reading.

Include field duplicates, equipment blanks where appropriate and periodic reference checks. Inspect the optical window during every service visit and record cleaning, desiccant replacement, cable repairs and configuration changes. Trend the raw signal in clean water and in a stable reference suspension if the project has suitable control material. Store timestamps in a consistent timezone and check logger clocks before deployment, especially when multiple instruments are used around a large worksite.

Australian monitoring programs often need to connect technical data with state environmental licence conditions, project approval requirements and the ANZG 2018 water-quality framework. The relevant limit may concern turbidity, suspended solids, visual plume extent or ecological protection rather than the sensor’s raw output. Confirm the reporting requirement with the project’s environmental manager and laboratory, and use a NATA-accredited laboratory where the contract or regulator requires accredited results.

Mining and infrastructure teams should also relate calibration work to the operational source of sediment. Guidance on mining monitoring guidance can help frame issues such as changing particle characteristics, runoff controls and continuous sensor deployment. In remote Western Australian or Northern Territory locations, plan for spare wipers, desiccant, batteries, clean bottles and duplicate instruments because access delays can be longer than the calibration window.

A clear calibration record should identify the sensor serial number, mounting depth, site, sampling dates, sediment source, laboratory method, model equation, valid range, uncertainty and next review date. Product-management or technical support information is available through contact Campbell Scientific, which supports the D & A Instruments product line. With this documentation, an optical backscatter sensor becomes a defensible suspended-solids monitoring instrument rather than a device producing unexplained site-specific numbers.