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

Managing Turbulence in Optical Backscatter Measurements

Optical backscatter sensors are valuable tools for measuring suspended sediment, turbidity and particle transport in rivers, estuaries, coastal waters and dredging zones. They provide rapid observations that can reveal changes a bottle sample may miss, especially when a sediment plume moves past a fixed monitoring point.

The same sensitivity that makes an optical sensor useful can create a variable signal. Turbulent eddies lift sediment, mix particles through the measurement volume and carry air bubbles or organic matter across the optical path. A sensor may therefore record genuine short-term changes alongside fluctuations caused by the measurement environment.

Accurate results depend on treating variability as information rather than simply smoothing it away. The appropriate averaging period, quality-control method and deployment arrangement should reflect the waterbody, the particle population and the decision the data must support. This is particularly important in Australian projects, where conditions can shift rapidly between a dry-season river, a tropical wet-season creek and a busy working harbour.

Why turbulence changes optical backscatter

An optical backscatter sensor sends light into the water and measures the light scattered back by particles. The measured response depends on particle concentration, size, shape, colour, mineral composition and the geometry of the sensor. Turbulence changes how those particles occupy the sensing volume from moment to moment, so a steady mean concentration can still produce an uneven signal.

An eddy may carry a dense packet of sediment past the optical window, followed by relatively clear water. Bed shear can lift coarse grains for a few seconds, while finer particles remain suspended for much longer. Flocs may break apart under strong shear and reform as energy declines. These processes create spikes, rapid falls and changing variance that a simple arithmetic average cannot always represent well.

Bubbles are a separate source of interference. Breaking waves, propeller wash, hydraulic jumps and aerated discharges can introduce highly reflective gas interfaces into the optical path. A bubble-related spike may be much larger than a sediment-related fluctuation, yet it does not indicate a corresponding increase in suspended-solids concentration. The D & A Instruments range provides useful context for optical sensing applications across marine and freshwater environments.

Where variability appears in Australian waters

A dredging plume near the Great Barrier Reef may contain fine mineral sediment, organic particles and flocs that respond differently to tides and vessel movements. In a Port Hedland or Pilbara port, ship loading, tug activity and strong currents can create short-lived bursts of resuspension. A sensor positioned near the seabed may therefore show much greater variance during vessel manoeuvres than during an equivalent tidal stage without traffic.

River systems have their own patterns. In the Murray–Darling Basin, low flows can leave sensors exposed to localised bed disturbance, algal material or stratification, while a storm pulse can quickly change both concentration and particle size. In the Top End, wet-season runoff may produce a large plume with rapidly evolving sediment characteristics. A reading averaged across several hours can conceal the sharp arrival of that first flush.

Urban waterways add another layer. A turbidity monitor in Sydney Harbour, the Swan River or Moreton Bay may be affected by ferry wakes, recreational boats, wind chop and tidal exchanges. Local field teams may describe a brief disturbance as a “boat hit” or a “rough arvo”, but the data system needs a repeatable rule for identifying whether that disturbance is a valid environmental event, an interference episode or both.

Australian monitoring programmes also operate through a mixed market of consultants, port authorities, regulators, research groups, dredging contractors and OEM integrators. Procurement may specify a compliance average, while an engineer needs peak concentrations for plume modelling. Averaging must therefore be designed around the reporting purpose rather than selected as a default instrument setting.

Selecting an averaging window

Start by defining the time scale of the process being measured. If the objective is a compliance statistic over 15 minutes, the raw sensor record should usually be collected at a shorter interval and then aggregated into that reporting period. If the objective is to detect a dredger-induced plume front, a long average may erase the event entirely. The measurement interval and reporting interval are related, but they are not the same thing.

A useful approach is burst sampling: record at a relatively high frequency for a defined period, calculate quality-controlled summaries, then repeat the burst at regular intervals. This preserves short-term structure while controlling data volume and power use. In a fixed station, a moving average can show the underlying trend, whereas non-overlapping block averages are easier to audit and compare with laboratory samples.

The averaging window should also relate to water movement. If the sensor views a volume that is replaced every few seconds, averaging over only one or two seconds may describe individual particle packets rather than the local water mass. If the window is much longer than the passage of a plume, it can dilute the maximum. Flow velocity, sensor position, path length and expected eddy turnover all help determine a defensible interval.

Practical averaging rules

Averaging should never be used to disguise a poorly placed sensor. If a probe sits directly in propeller wash or intermittently emerges into air, statistical treatment cannot restore reliable concentration data. Physical changes to mounting, orientation or depth are usually more effective than increasingly complex filters.

Handling spikes and non-uniform samples

The arithmetic mean remains useful because mass loads and many regulatory metrics are based on average concentration. It becomes misleading when a record contains bubbles, fouling, electronic artefacts or a highly skewed particle distribution. A good workflow calculates several descriptors: mean, median, standard deviation, percentile values, coefficient of variation and the proportion of data rejected by quality-control rules.

Robust methods are particularly helpful when turbulence produces occasional high readings. The median describes the typical sample in a skewed burst, while a trimmed mean reduces the influence of extreme observations. A rolling median can identify a baseline without allowing one large spike to shift it for the entire window. These methods should be reported clearly, since a median concentration and a mean concentration answer different questions.

Spike rejection needs evidence. A sudden increase that coincides with a pump cycle, bubble detector flag, loss of signal or implausible rate of change may be removed. A sudden increase that coincides with high flow, a dredge move or a verified grab sample may be a real sediment event. Automatic thresholds should therefore be supported by field observations, auxiliary sensors and deployment notes.

Raw optical backscatter data often need conversion into a sediment concentration time series through calibration and processing steps. The data-processing guide explains why filtering, calibration and validation should be treated as separate stages rather than folded into one opaque calculation.

Matching calibration to turbulent conditions

Optical response is rarely universal across all sites. A sensor calibrated with fine estuarine mud may respond differently in quartz-rich sand, iron-stained Pilbara sediment or organic-rich freshwater material. Particle size distribution, mineralogy, shape and flocculation affect the relationship between backscatter and suspended-solids concentration.

Site calibration should cover the range of conditions likely to be encountered. Collect water samples across low, medium and high backscatter periods, including rising and falling limbs of a storm or dredging event where possible. Laboratory concentration results can then be compared with synchronised sensor readings. If the relationship changes with depth, flow direction or tidal stage, separate calibration groups may be justified.

Turbulence makes timing important. A bottle sample taken several minutes after a sensor spike may represent a different particle population. Record the exact sampling time, sensor depth, flow conditions and nearby activity. Replicate samples are valuable when the water is visibly patchy, because a single bottle may not represent the volume observed by the optical instrument.

Sensor maintenance is part of calibration control. Biofouling, sediment coating, scratched windows and a loose mounting frame can alter the optical geometry. In warm Australian waters, fouling may develop quickly, while floodwaters can carry debris that damages exposed equipment. A deployment log should record cleaning, wiper operation, depth changes, battery events and any period when the instrument was moved.

Field checks that protect data quality

For product-specific operating details, installation questions and support arrangements, the frequently asked questions provide a practical reference. Campbell Scientific now supports the product line with contact and product-management information, which is relevant for Australian users planning long deployments or integrating sensors into a wider monitoring system.

Designing a defensible monitoring record

A defensible record separates observation, processing and interpretation. Store the original timestamped signal, instrument status data and any diagnostic flags before producing averaged values. Keep the chosen window length, rejection criteria, calibration equation and units with the processed dataset. This makes it possible to reproduce a result months later when a project enters review or an approval condition is audited.

Synchronisation is equally important. Optical data should share a reliable clock with flow meters, pressure sensors, acoustic instruments, pump logs or dredging activity records. A few minutes of clock drift can make a real relationship appear weak, particularly when a plume front is moving quickly. Use time zones and daylight-saving settings consistently, especially when data from remote Australian stations are combined with laboratory or contractor records.

Depth profiling can reveal whether turbulence is local or spread through the water column. A sensor near the bed may capture resuspension, while one near the surface may respond to wind mixing, rainfall or vessel wakes. In stratified estuaries, a single depth average can be difficult to interpret. Profiles or multiple fixed sensors help distinguish vertical structure from random noise.

The final product should state what each value represents. A one-minute mean, a 15-minute compliance average, a burst median and a maximum raw reading are different observations. Presenting them together, with flags and supporting hydrological information, allows a regulator, researcher or project engineer to see both the prevailing condition and the short-lived events that shaped it.

Averaging is most effective when it follows the physics of the site. Turbulence-driven variability can reveal mixing, resuspension and plume movement, while unsuitable smoothing can hide precisely those processes. With appropriate sampling frequency, robust quality control, site-matched calibration and transparent reporting, optical backscatter measurements can support reliable decisions across Australia’s rivers, ports, estuaries and coastal works.