Field Audits of Optical Sensor Accuracy With Grab Samples
Optical turbidity and suspended-solids sensors provide continuous measurements in conditions where laboratory sampling alone cannot capture rapid changes. A sensor installed near a dredging operation, stormwater outfall, intake, or streambed can record plume movement minute by minute. That continuous record is valuable only when the instrument’s output remains connected to a defensible physical measurement.
A field audit compares sensor readings with carefully collected grab samples analyzed by an established reference method. The purpose is not simply to prove that two numbers match. It is to identify bias, response limitations, fouling, installation effects, and changes in the relationship between optical signal and actual sediment concentration.
The most reliable audits combine site observations, synchronized sampling, laboratory quality control, and transparent data analysis. This approach applies to turbidity monitors, backscatter instruments, transmissometers, and suspended-solids probes used in marine and freshwater environments, including systems integrated into hydrology or OEM platforms.
Define The Measurement Before Visiting The Site
Start by documenting what the optical sensor measures and what the grab-sample laboratory will report. Turbidity may be expressed in NTU, FNU, or another instrument-specific unit, while suspended solids are commonly reported as milligrams per liter after filtration and drying. These quantities are related in some waters, but they are not interchangeable.
An optical sensor responds to how particles scatter or absorb light. Particle size, shape, color, mineralogy, and concentration all affect the signal. A sensor calibrated in a fine, pale sediment may produce a different response in water containing dark organic particles or coarse sand. Reviewing backscatter and transmissometry helps clarify why two optical technologies can respond differently to the same sample.
Write a short audit objective before fieldwork begins. It might be to verify a factory calibration, assess a site-specific conversion from turbidity to total suspended solids, quantify drift over a deployment, or determine whether a sensor is suitable for an operational threshold. The objective determines the sample range, number of observations, and acceptance criteria.
Prepare The Instrument And Sampling Equipment
Inspect the sensor before collecting any comparison data. Record the serial number, firmware or configuration, optical path condition, wiper operation, cable condition, mounting orientation, and date of the last cleaning or calibration. Photograph the installation and note the sensor depth, distance from the bed, flow direction, and nearby structures that could influence water movement.
Download the instrument data before changing settings. Preserve the original time series, configuration file, calibration coefficients, and diagnostic logs as read-only records. Confirm that the sensor clock, sampler clock, and laboratory sample labels use the same time reference. Even a modest time offset can create an apparent accuracy problem during a fast-moving sediment plume.
Prepare clean sample bottles, labels, preservatives where required, a field notebook or electronic form, a depth sampler, a rinse container, gloves, a cooler, and a calibrated thermometer. If the audit includes turbidity measurements made in the field, bring appropriate standards and a clean measurement vessel. Secondary standards can support routine verification between full calibrations; the guidance on secondary standard checks provides useful context for that control step.
Collect Samples At The Sensor’s Measurement Zone
The grab sample must represent the water viewed by the optical path. Sampling from the surface while the sensor measures near the bed is unlikely to produce a meaningful comparison, especially where sediment concentration varies vertically. Position the sampler at the same depth and as close to the sensing volume as practical without disturbing the water around the instrument.
Collect samples across the expected operating range rather than gathering only convenient low-concentration observations. Include background conditions, moderate values, high-concentration events, rising and falling limbs of a plume, and any conditions that operators consider important. A useful audit often requires repeated visits or event-based sampling because a single calm-day dataset cannot reveal performance during challenging conditions.
Avoid disturbing the bed, mooring, pipe, or sediment plume before the sample is taken. Approach from downstream when site conditions allow, and allow turbulence caused by a boat or operator to settle. For a fixed installation, collect a concurrent sample before removing or cleaning the sensor. If the sensor is profiled through the water column, document depth and movement rate for every sample.
Collect duplicate or replicate samples at selected points. Replicates reveal whether variation comes from the sensor or from the water itself. Field blanks and equipment blanks can identify contamination, while a split sample sent to two laboratories can expose laboratory-related variation. Keep every sample traceable to a sensor timestamp, location, depth, and field condition.
Match Sensor Readings To Reference Results
Record the optical output at the exact time of each grab sample, preferably using a short averaging window that represents the collection period. Avoid comparing an instantaneous peak with a laboratory result that represents several minutes of filling, mixing, and transport. For rapidly changing flows, collect a sample while another operator announces the time and sensor value.
The laboratory method should be appropriate for the target parameter and concentration range. For total suspended solids, the laboratory generally filters a known volume, dries the retained material, and calculates mass per unit volume. Record the filter type, sample volume, drying conditions, reporting limit, and any dilution. For turbidity, document the method, instrument, cell handling, and standardization procedure.
Keep the raw results separate from derived results. Do not overwrite sensor data with corrected values, and do not alter laboratory results to make a calibration line look cleaner. A strong audit preserves the original readings and applies any correction through a documented equation or processing step.
| Audit element | Optical sensor record | Grab-sample reference | Why it matters |
|---|---|---|---|
| Time | Timestamp and averaging interval | Collection time and laboratory receipt time | Reveals mismatches during changing conditions |
| Location | Sensor coordinates, depth, and orientation | Sampler coordinates, depth, and position | Tests whether both measurements represent the same water |
| Quantity | Turbidity, backscatter, transmission, or converted solids | Laboratory turbidity or measured suspended solids | Prevents comparison of unlike parameters |
| Quality control | Diagnostics, cleaning status, calibration history | Replicates, blanks, method details, and reporting limits | Separates instrument error from sampling or laboratory error |
| Analysis | Raw output and processing settings | Raw result, dilution, and uncertainty where available | Makes recalculation and review possible |
Evaluate Bias, Precision, And Range
Plot the paired data before calculating a correction. A scatterplot can show curvature, separate particle populations, outliers, hysteresis between rising and falling conditions, or a cluster of values near the detection limit. Plot sensor output against the laboratory result in both directions when a site-specific conversion is being developed, and include units on every axis.
Calculate the difference between each sensor reading and its paired reference value. Useful metrics include mean bias, mean absolute error, root mean square error, relative error where the reference is sufficiently above zero, and the coefficient of determination. R-squared can describe the strength of a relationship, but it does not prove accuracy; a consistently biased sensor may still produce a high R-squared value.
Assess repeatability with duplicate samples and repeated sensor readings. Compare errors across concentration bands rather than reporting only one overall percentage. A five-unit error may be significant near a low regulatory threshold but negligible during a very high sediment event. Establish acceptance limits before reviewing the results whenever the audit supports compliance or procurement decisions.
Look for patterns linked to operating conditions. A sensor may agree during steady flow but diverge when bubbles enter the optical path, when wiper cleaning causes a transient signal, or when coarse particles pass through the sensing volume. Differences between suspended-solids grab samples and optical readings may reflect genuine changes in particle characteristics rather than a simple instrument fault.
Investigate Installation And Water-Quality Effects
Optical accuracy depends heavily on installation. Bubbles, sunlight, biofouling, condensation, scratched windows, loose mounts, and sediment deposits can change the received light. Check whether the instrument is aligned with the flow, exposed to wake turbulence, or positioned too close to a wall, bed, intake, or discharge point. Review diagnostic flags and compare readings before and after cleaning.
A sensor can also be functioning correctly while the grab sample is unrepresentative. Stratification, settling, turbulent suspension, and lateral concentration gradients are common in rivers, channels, reservoirs, and dredging zones. If a vertical profile is part of the audit, document concentration and optical readings at several depths. Related hydrology work on vertical hydraulic gradients illustrates the value of careful depth-resolved measurements when water conditions vary through a profile.
Check environmental variables alongside concentration. Temperature, conductivity, salinity, flow velocity, water level, rainfall, pump status, tide, and weather can help explain changes in the sensor-to-laboratory relationship. In marine settings, salinity and particle composition may shift during tidal cycles. In groundwater or low-turbidity applications, small contamination events or bubbles may dominate the apparent signal.
When an anomaly appears, repeat the comparison before adjusting calibration coefficients. Clean and inspect the sensor, verify the clock, check a known standard if appropriate, and collect another paired sample. A correction applied to data affected by fouling or poor sampling can conceal the underlying problem.
Build A Defensible Audit Record
The audit report should allow another technician to reconstruct what happened. Include the sensor model and serial number, installation diagram or photographs, sampling dates and times, coordinates, depths, weather and flow conditions, cleaning actions, instrument settings, laboratory method, chain-of-custody information, raw data, calculations, graphs, and acceptance criteria.
Separate verification from recalibration. Verification asks whether the instrument meets a defined performance requirement under stated conditions. Recalibration changes the relationship between sensor output and reported value. If a site-specific suspended-solids regression is created, state its valid range, equation, units, sample population, uncertainty, and conditions under which it should not be used.
Use an independent dataset when developing a conversion or correction. Fit the equation with one portion of the paired data and test it with samples collected at different times or under different conditions. This guards against overfitting and shows whether the relationship remains stable as the sediment source or hydraulic regime changes.
Recommended field-audit controls include:
- Synchronize clocks and document the averaging interval before sampling.
- Collect paired samples across low, medium, and high concentration conditions.
- Include duplicates, blanks, and laboratory quality-control records.
- Record depth, position, flow state, weather, fouling, and maintenance at every visit.
- Preserve raw sensor output separately from corrected or converted data.
Turn Results Into Ongoing Monitoring Practice
A field audit is most useful when it becomes part of a planned verification cycle. Set the interval according to deployment conditions, risk, fouling rate, and the consequences of incorrect readings. A clean freshwater installation may need less frequent attention than a sensor operating in a dense dredging plume or a biologically active coastal site.
Use audit results to refine sampling design and maintenance, not merely to assign a pass or fail. If errors increase after several weeks, shorten the cleaning interval. If the relationship changes with sediment type, create separate site or season-specific models. If the sensor is reliable only within a limited range, define that range in the operating procedure and flag values outside it.
Campbell Scientific support can help organizations maintain product records and identify the appropriate management or contact path for legacy D & A Instruments equipment. Keep the technical audit with the project’s environmental and asset records so future operators can distinguish sensor changes from genuine changes in water quality.
A well-designed comparison gives field teams more than a single accuracy percentage. It shows whether the optical measurement represents the sampled water, whether the laboratory result is fit for purpose, and whether the complete monitoring system can support decisions about sediment control, dredging, research, or environmental protection. Begin with synchronized observations, document every condition, and use the evidence to keep continuous optical data connected to the water being measured.