Calibrating a suspended-solids sensor with site-specific sediment
A suspended-solids sensor estimates the concentration of particles in water by measuring how those particles interact with light. The instrument may report an optical signal, turbidity, or a calculated suspended-solids concentration, but the relationship between that signal and actual sediment concentration depends heavily on the material in the water. A calibration developed in one river, harbor, lake, or dredging area may produce misleading results somewhere else.
Site-specific calibration connects the sensor’s response to the sediment and water conditions at the monitoring location. The process pairs sensor readings with laboratory measurements of suspended solids from carefully collected samples. When the samples cover the expected operating range and represent the particle sizes, colors, and mineral composition found in the field, the resulting conversion is more defensible and useful for environmental monitoring.
This work is especially important for dredging plume monitoring, sediment transport studies, stormwater investigations, and research in changing freshwater or marine environments. Optical instruments are powerful because they provide rapid, continuous measurements, but they need a local relationship between optical response and mass concentration. The optical sensing technology behind the measurement should therefore be treated as the starting point for calibration, rather than as a universal conversion from turbidity to milligrams per liter.
Why site sediment changes the calibration
A suspended-solids sensor detects the way particles absorb, scatter, or reflect light. Particle size distribution is one of the strongest influences on that response. Fine clay particles can produce a different signal from coarse sand at the same mass concentration because their surface area, shape, and light-scattering behavior differ. Organic particles, algae, shell fragments, and flocculated sediment can also change the measured signal.
Color matters as well. Dark organic material may absorb light, while pale mineral particles can scatter it efficiently. Particle shape and aggregation affect the optical path through the sample, and salinity can influence floc formation in estuarine or coastal water. These variables mean that two samples with identical total suspended solids, or TSS, may generate different sensor readings.
The instrument’s installation also influences the measurement. A sensor mounted near the bed may see larger particles or intermittent resuspension events, while a midwater sensor may measure a more evenly mixed suspension. Flow velocity, bubbles, fouling, ambient light, and wiper performance can create additional differences between the raw optical signal and the actual concentration.
For this reason, a factory calibration or generic turbidity-to-TSS equation is often best used as a preliminary estimate. A local calibration should replace it when quantitative sediment concentration is important, especially when results will support regulatory reporting, dredging limits, process control, or comparisons between monitoring stations.
Prepare a representative sampling design
Begin by defining the concentration range and conditions the sensor must measure. Review historical data, site observations, storm records, dredging schedules, tidal cycles, and planned construction activity. The calibration should include clear water or background conditions as well as the highest realistic sediment loads. If the instrument will operate during extreme events, samples from those conditions are particularly valuable.
Collect water samples as close as possible to the sensor and at the same time as the instrument reading. Record the date, time, location, depth, water level, flow condition, weather, tide, operational activity, and any visible changes in the water. A sample taken several minutes away from the sensor observation may no longer represent the same sediment plume, especially in fast-moving water or near a discharge.
Use a sampling method that matches the monitoring objective. A depth-integrated sample may be suitable for estimating the average concentration through the water column, while a point sample may better represent the conditions at the sensor face. If sediment concentration varies strongly with depth, collect paired samples at multiple elevations rather than assuming that one bottle represents the entire profile.
Use clean containers and avoid unnecessary settling before laboratory processing. Mix samples gently but thoroughly before subsampling so that particles remain distributed. Very vigorous shaking can break fragile flocs, while insufficient mixing can cause coarse particles to remain at the bottom. Note the handling method because sample disturbance can affect the laboratory result.
Build the calibration dataset
The reference value is normally obtained by filtering a known volume of sample, drying the retained material, and weighing the residue. The laboratory result is then expressed as mass per volume, commonly milligrams per liter. Follow a consistent method for filter type, drying temperature, balance resolution, sample volume, and blank correction. If a recognized standard method or laboratory protocol is required, use it consistently for every calibration sample.
At each sampling event, save the raw sensor output rather than recording only a displayed concentration. Depending on the instrument, this may be voltage, counts, optical attenuation, turbidity units, or another diagnostic value. Raw output makes it possible to test alternative equations later and helps identify sensor saturation, unstable readings, or a configuration problem.
Collect enough paired observations to describe the complete response curve. A small number of samples clustered around low concentrations can produce a visually convincing but unreliable equation at high concentrations. Include replicate samples at selected points to estimate laboratory and sampling variability. Replication is particularly helpful when sediment is patchy, flocculated, or rapidly changing.
| Calibration element | What to record | Why it matters |
|---|---|---|
| Sensor response | Raw optical value and displayed reading | Preserves the original measurement for analysis |
| Reference concentration | Laboratory TSS or suspended-sediment result | Provides the mass-based value used for calibration |
| Sample conditions | Depth, flow, tide, weather, and site activity | Explains changes in particle distribution |
| Sediment characteristics | Grain size, color, organic content, or mineral notes | Helps interpret differences in optical response |
| Quality information | Replicates, blanks, holding time, and laboratory method | Quantifies uncertainty and identifies weak data |
| Operating state | Sensor cleaning, orientation, depth, and diagnostics | Separates site effects from instrument problems |
Plot the sensor response against the laboratory concentration before selecting an equation. A straight line may be appropriate over a limited range, but nonlinear behavior is common when particles aggregate, the detector approaches saturation, or the optical response changes at higher concentrations. Do not force a linear model simply because it is easier to implement.
Fit, validate, and document the model
The calibration equation should be based on the relationship that best represents the paired data and the way the instrument will be used. Possible models include linear, polynomial, logarithmic, or segmented relationships. A more complex equation is not automatically better; it should have a physical and statistical justification and remain stable when applied to new measurements.
Inspect residuals, which are the differences between laboratory results and values predicted by the calibration. A pattern in the residuals can show that the model misses curvature or behaves differently at low and high concentrations. Large isolated errors may indicate sample mismatch, bubbles, fouling, laboratory handling issues, or an unusually different sediment type rather than a problem with the entire calibration.
Reserve some paired observations for validation instead of using every sample to fit the equation. The validation data should cover the operating range and, when possible, include different flow or sediment conditions. Compare predicted and laboratory concentrations using absolute error, percentage error where appropriate, bias, and confidence intervals. A high coefficient of determination alone does not prove that the calibration is accurate; a model can show strong correlation while consistently overestimating or underestimating concentration.
Check the lowest measurable range carefully. Near the instrument’s detection limit, background noise and small changes in clean-water optical properties may be a significant proportion of the signal. At the upper end, examine whether the sensor saturates or loses sensitivity. If the relationship changes sharply, use separate validated ranges only when the monitoring system can reliably identify which range applies.
Document the final equation, valid concentration range, units, sensor serial number, firmware or configuration, sampling dates, laboratory method, sediment description, and uncertainty. Include the number of calibration and validation samples, exclusions, model-selection rationale, and any correction for baseline signal. This record allows future users to understand exactly what the reported suspended-solids concentration represents.
Transfer the calibration to field operation
Before deploying the calibrated sensor, inspect the installation and verify that the optical window is clean and undamaged. Confirm the sensor depth, orientation, cable connections, power supply, data logger settings, averaging interval, and time synchronization. A correct equation cannot compensate for an instrument that is exposed to bubbles, sunlight, biofouling, or an unrepresentative flow zone.
Perform a field check with clean water or a suitable low-sediment condition when practical. Compare the live output with the expected baseline, then observe the response during a known disturbance or natural event. The goal is not to recreate the full laboratory calibration but to confirm that the configured instrument behaves consistently with the documented setup.
Schedule verification sampling after deployment. Recheck the sensor when sediment sources change, construction methods vary, seasonal biological material appears, or the instrument is moved. A calibration may remain useful for months or years in a stable setting, but it can become unsuitable after a major change in grain size or sediment origin.
Maintenance records should include cleaning dates, wiper replacement, desiccant changes, diagnostics, unusual readings, and any physical movement of the sensor. When a sudden concentration shift occurs, compare it with raw optical output, nearby flow data, photographs, and independent water samples. This helps distinguish a genuine sediment event from fouling, bubbles, cable problems, or an invalid conversion.
Sediment monitoring can also benefit from complementary measurements. Water level, velocity, conductivity, pressure, and profiles from other instruments can reveal whether the sensor is positioned in a representative flow zone. Guidance on interpreting spatially varying measurements is available in the discussion of groundwater profiler data, and the same principle applies broadly: a concentration value is most useful when its physical sampling location and environmental context are understood.
Keep the calibration defensible over time
A local calibration is a working measurement model, not a permanent property of the sensor. Its reliability depends on the continued similarity between the calibration samples and the sediment reaching the instrument. Establish a review trigger based on time, data quality, site activity, or a change in the distribution of sensor readings.
When collecting verification samples, target conditions that are poorly represented in the original dataset. If the initial calibration was built during calm weather, add samples during high flow. If it was developed during mineral sediment transport, check whether organic-rich material or seasonal algae changes the response. Expanding the dataset deliberately is more useful than collecting many nearly identical samples.
Apply these practices when creating or updating a site-specific sediment calibration:
- Pair laboratory samples with sensor readings as closely in time and location as practical.
- Cover the full expected concentration range, including background and peak conditions.
- Preserve raw sensor data and document the instrument configuration for every sample.
- Use replicate samples and independent validation observations to quantify uncertainty.
- Reassess the equation after changes in sediment source, installation, fouling control, or monitoring purpose.
A calibration should also state its limitations in the data system or project documentation. Specify whether the reported value is total suspended solids, suspended sediment concentration, or an instrument-specific estimate. Explain any exclusions, range limits, smoothing, gap filling, or quality flags so that downstream users do not treat every value as equally certain.
When the work supports dredging compliance or environmental decisions, retain the physical samples or laboratory records according to project requirements. A clear chain of evidence from field observation to laboratory result to sensor configuration makes the monitoring record easier to audit and defend.
Use the resulting model to turn continuous optical observations into meaningful site-specific sediment data, while continuing to verify that the relationship remains valid. For product support, configuration details, and current management information for D & A Instruments equipment, consult Campbell Scientific and maintain the calibration record alongside the instrument documentation.