Verification of Turbidity Sensor Linearity Using Dual Standard Dilution Series

A turbidity sensor is often described as linear when its output rises in proportion to the concentration of suspended material. In practice, that assumption needs to be tested across the working range. Optical instruments can respond differently at low and high concentrations because of stray light, multiple scattering, particle settling, detector saturation, fouling, and changes in particle size or colour.

A dual standard dilution series provides a practical way to examine that response. Two independently prepared series are made from suitable turbidity standards, then measured through the same sensor and data-acquisition system. Agreement between the series gives stronger evidence of repeatability than a single dilution run, while differences can reveal preparation error, unstable standards, or a matrix effect.

The approach suits laboratory verification before deploying equipment in Australian rivers, estuaries, dredging zones, aquaculture sites, and coastal monitoring programs. It is especially useful when a sensor will be converted from turbidity units to suspended-solids concentration, since the relationship between optical response and actual sediment mass depends on local material.

Why linearity matters in optical monitoring

Linearity describes how closely the sensor signal follows a straight-line relationship with a known concentration. Depending on the instrument, the measured signal may be reported directly as NTU or FNU, as a frequency, voltage, digital count, or internally calculated turbidity value. The test should therefore record both the displayed measurement and the raw output where that option is available.

A linear response improves confidence in interpolation. If a calibration point is collected at 100 NTU and a sample reads 60 NTU, the result is meaningful only when the sensor behaves predictably between nearby standards. A curved response can cause a substantial concentration error even when the calibration line has a high coefficient of determination. R² is useful, but residuals, replicate spread, slope, intercept, and the location of deviations are more informative.

The relevant range should reflect the intended deployment rather than an arbitrary laboratory span. A clear-water stream may require excellent resolution below 10 NTU, while a dredging plume can move rapidly into hundreds or thousands of units. In the latter case, a sensor may remain operational while its response compresses at the upper end.

Preparing two independent dilution series

Begin with clean vessels, consistent mixing equipment, a calibrated pipette or graduated glassware, and standards with a documented nominal value and expiry date. Commercial polymer or formazin-based standards can be suitable, but their stability, storage requirements, and compatibility with the instrument should be checked. Use the same water quality and temperature conditions for both series, preferably low-turbidity water that does not introduce particles or bubbles.

The two series should be prepared independently rather than by splitting one finished set. For example, Series A and Series B can each contain a blank and five or six target levels covering the operating range. A geometric sequence such as 0, 5, 10, 25, 50, 100, and 200 units is useful for broad screening, while closely spaced points around a regulatory or process threshold can provide a more sensitive check.

Use volumetric calculations consistently. If a stock concentration is diluted into a final volume, the target concentration follows (C_1V_1=C_2V_2). Record the stock identity, lot number, preparation date, water source, final volume, mixing time, and operator. Allow the standard to equilibrate before measurement, but do not leave suspensions standing long enough for settling to alter the intended concentration.

Particle characteristics can affect the result even when the nominal turbidity is identical. A useful discussion of particle response behaviour explains why flocculated and dispersed material may produce different optical signals. That distinction is central when a laboratory standard is being used to represent natural sediment.

Controlling the measurement conditions

Bring the sensor to the required operating temperature and inspect the optical windows before starting. Remove fingerprints, biofilm, salt deposits, and trapped air. A small bubble across the light path can produce a reading that looks like a sudden loss of linearity. For submersible instruments, keep the sensor orientation, immersion depth, and clearance from the vessel wall consistent.

Mix each dilution gently but thoroughly immediately before measurement. Excessive agitation can create bubbles; insufficient mixing allows heavier grains or flocs to settle. A magnetic stirrer, overhead mixer, or standardised manual inversion may be used, provided the method is repeatable. For coarse sediment, note the time between mixing and reading because the settling rate may be part of the measurement uncertainty.

Measure the blank first, then proceed from the lowest to the highest concentration, or use a randomised order when drift is a concern. Take several readings after the displayed value stabilises and calculate a mean with a standard deviation. Repeat the sequence in reverse order if the sensor may foul or exhibit hysteresis. Rinse between levels where carry-over could be significant, especially after high-concentration standards.

Temperature, ambient light, bubbles, vibration, and electrical noise should be controlled or documented. An instrument intended for a Murray–Darling tributary may encounter a very different thermal environment from one installed near the Great Barrier Reef. Recording these conditions makes the verification defensible when results are compared months later.

Analysing the paired response curves

Plot the measured sensor response against the prepared standard value for Series A and Series B. If the instrument reports turbidity directly, use those readings as the response variable. If it produces voltage or counts, retain the raw response and fit the calibration equation in that domain before applying any instrument-specific conversion.

Fit a linear model such as (R=aC+b), where (R) is the measured response and (C) is the nominal standard concentration. Examine the slope, intercept, residuals, confidence intervals, and replicate variability. A blank with a large positive intercept may indicate background scatter, contamination, electronic offset, or incomplete zeroing. A slope change at the upper levels can indicate optical saturation or multiple-scattering effects.

Compare the slopes and intercepts of the two independently prepared series. Their curves should overlap within the combined measurement and preparation uncertainty. Calculate the difference between paired readings at each target level, along with the relative percent difference where the denominator is appropriate. A Bland–Altman-style plot can show whether disagreement grows with concentration, which is often more revealing than a single regression statistic.

The acceptance limit should be established before testing. It may be based on the manufacturer’s specification, a project quality plan, a laboratory precision target, or the uncertainty required for a permit or environmental management program. Avoid declaring success solely because R² exceeds a chosen value; a smooth but curved response can still produce a misleadingly high R².

Using the result for suspended-solids conversion

Turbidity is an optical property, while total suspended solids are determined by mass. The sensor’s response to clay, silt, sand, organic fragments, and flocculated material will vary with particle size, shape, refractive index, colour, and settling state. Consequently, a linear turbidity check does not by itself validate a universal turbidity-to-milligrams-per-litre conversion.

For a site-specific suspended-solids model, collect representative water samples across the expected turbidity range. Keep the sensor reading time-stamped with each sample, then determine suspended solids using a documented gravimetric laboratory method. Regression may be linear, polynomial, segmented, or unsuitable over the whole range. Separate models may be required for dry-weather baseflow, storm runoff, construction disturbance, and dredging conditions.

The distinction matters in Australian applications. Red-brown catchment sediment near the Fitzroy River may scatter light differently from pale marine sand near Gladstone. Wet-season flood pulses in northern Australia can alter particle size and organic content within hours, while a controlled dilution series uses a much more stable material. A verification test should therefore be treated as evidence about the instrument and test medium, not as proof that every field sediment will behave identically.

For plume monitoring around a port or reclamation project, pair optical measurements with occasional laboratory samples and field observations. In a busy Australian market, project specifications may be written by consultants, equipment supplied through an integrator, and compliance data reviewed by a regulator or principal contractor. Clear traceability between sensor output, calibration equation, sample identity, and laboratory result prevents disputes over which conversion was applied.

Documenting a repeatable verification

A useful record allows another technician to reproduce the test without relying on informal knowledge. Include the sensor serial number, firmware, optical configuration, cable or logger, standard certificates, dilution calculations, water quality, temperature, mixing method, measurement order, raw readings, averages, and exceptions. Retain the original data file rather than recording only rounded values in a report.

A compact verification record should include:

• The target concentrations and independently calculated preparation volumes
• Replicate readings, mean values, standard deviations, and blank response
• Linear regression parameters, residual plots, and paired-series differences
• The defined acceptance criteria and final operating range

If the two series disagree, investigate the process before adjusting the calibration. Check pipette accuracy, vessel cleanliness, standard age, settling time, bubbles, sensor fouling, and logger scaling. Repeating one level will not identify a systematic dilution error, so remake both series when preparation is suspect. If disagreement persists, test whether the instrument is stable with a certified check standard and inspect the optical path.

Keep a clear distinction between verification and calibration. Verification asks whether the sensor follows the expected response under defined conditions. Calibration changes the instrument equation or converts its output for a particular application. The manufacturer and support information available through D & A Instruments can help identify the relevant sensing configuration, product history, and technical context before a field procedure is formalised.

A documented dual-series test can then become part of routine quality assurance. Repeat it after sensor servicing, firmware changes, replacement of optical components, or a significant change in deployment conditions. For installations exposed to sediment, marine growth, or frequent handling, trend the slope and blank response over time. That record gives early warning of drift before an apparently plausible turbidity value becomes a questionable environmental result.