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

Correcting turbidity data when bubbles disrupt high-energy flows

Turbidity sensors are often deployed where water is moving rapidly: below dredge heads, near propellers, in spillways, around hydraulic structures, and in energetic coastal or riverine environments. These sites generate suspended sediment signals that are valuable for plume tracking and environmental compliance, yet the same turbulence can introduce air bubbles into the optical path. The result may be a sudden, exaggerated rise in reported turbidity that has little relationship to the actual sediment concentration.

Bubble interference is a measurement problem rather than a simple instrument failure. An optical turbidity monitor detects how light is scattered or attenuated within the sampled water volume. Air-water interfaces can scatter light far more strongly than many sediment particles, causing spikes, dropouts, unstable readings, or apparent sediment events that move with the flow.

A reliable correction therefore combines sensor placement, diagnostic flags, signal processing, and field validation. The goal is not to erase every rapid change. Genuine sediment pulses can be equally fast. Instead, the correction should identify observations that are physically consistent with entrained air, preserve valid high-frequency sediment information, and document every decision applied to the data.

Why bubbles create false turbidity signals

Most optical turbidity instruments emit light into the water and measure the light scattered back toward a detector or transmitted through a defined path. Suspended mineral particles, organic matter, and plankton affect that light field according to their size, shape, concentration, and optical properties. Calibration converts the measured response into a turbidity value, commonly expressed in NTU or another instrument-specific unit.

A bubble has a very different optical behavior. Its gas-water boundary produces a strong refractive-index contrast, and its curved surface can redirect light toward the detector. A bubble passing directly through the measurement volume may create a short, intense peak. A cloud of bubbles can produce sustained elevation, rapid oscillations, or periods in which the detector receives an unpredictable combination of scattered and blocked light.

High-energy flow increases the likelihood of this interference through several mechanisms. Turbulence entrains air at the surface, hydraulic jumps and plunging jets draw air below the waterline, and dredging equipment can inject air through leaks, agitation, or propeller action. Bubbles may also collect around a poorly oriented sensor, mounting frame, cable, wiper, or protective cage.

The optical response depends on the sensor design and flow conditions. A backscatter turbidity sensor may show sharp positive excursions when bubbles cross its field of view, while a transmissometer can experience attenuation or signal loss. A suspended-solids probe with multiple optical wavelengths may reveal inconsistent channel behavior. These differences make a universal numerical correction unreliable.

Recognizing bubble-contaminated observations

The first step is to distinguish bubble artifacts from real changes in suspended sediment concentration. Bubble interference commonly appears as narrow spikes with steep rise and fall times, repeated bursts synchronized with wave or pump cycles, or clusters of unstable values followed by a return to a previous baseline. A genuine sediment plume can also change quickly, so timing alone should not determine whether a record is rejected.

Use every available diagnostic channel. Many instruments provide optical intensity, detector counts, quality flags, signal strength, gain state, or separate forward- and backscatter measurements. When turbidity rises while the raw optical signal approaches saturation, drops outside its normal range, or becomes erratic, air entrainment becomes more plausible. A bubble event may also affect optical channels differently, producing a ratio that is inconsistent with the calibration relationship for sediment.

Compare the turbidity record with flow velocity, depth, pressure, acoustic backscatter, pump status, vessel position, and other operational data. If spikes occur whenever a cutter head starts, a thruster changes speed, or a sensor crosses a turbulent interface, the event has a useful physical explanation. If the same pattern appears at multiple sensors separated across the flow, it may represent a real plume rather than a local bubble passing through one optical path.

Visual inspection remains useful during deployment. Video, diver observations, transparent flow cells, or a temporary above-water inspection can reveal whether bubbles collect on the sensing face. Instrument logs should retain the original time series, because an apparently suspicious value may become interpretable when viewed against site activity and water level.

For instrument-specific behavior, review the manufacturer’s documentation and technical support resources, including the sensor FAQs. D & A Instruments systems are associated with optical monitoring in marine and freshwater applications, while Campbell Scientific provides current product-management and support information for the product line.

Improve the measurement before correcting the record

Signal processing cannot fully compensate for a sensor that is consistently sitting in an aerated pocket. Physical installation is the most effective first control. Place the sensing volume below the surface but away from plunging inflows, propeller wash, discharge jets, and structures that shed vortices. Align the sensor with the dominant flow where practical, and avoid creating a mounting geometry that traps air against the optics.

A downward-looking orientation can reduce the chance that rising bubbles remain on the optical window, though the correct orientation depends on the instrument, depth, and sediment distribution. A small angle may help bubbles slide past the sensing face rather than accumulate on it. Protective cages should have sufficient open area to avoid both bubble retention and local turbulence.

Keep the optical window clean and inspect wipers, brackets, cable fairings, and nearby surfaces. Biofouling, sediment deposits, and scratches can create symptoms that resemble air interference. A clean sensor installed in a stable frame makes later data screening more defensible. In dredging work, record the relationship between the instrument and the cutter, discharge, or plume boundary so that the measurement zone is understood.

Sampling rate also matters. A slow logging interval can average a short bubble event into an apparently moderate turbidity increase. A high-frequency record makes the event easier to identify, provided storage and telemetry capacity are sufficient. Store raw or minimally processed measurements alongside the reported series whenever possible.

Calibration should be treated carefully. A turbidity calibration made with sediment standards does not automatically describe an aerated flow. Do not recalibrate the instrument against bubble-contaminated samples, and do not use a blanket scale factor to reduce suspicious data. Bubble effects are usually episodic, nonlinear, and dependent on the position and size of the bubbles.

Select a correction method that matches the evidence

A practical workflow begins by defining a valid-data rule before reviewing the most interesting events. Set reasonable limits for turbidity, raw optical signal, rate of change, and diagnostic status. These limits should come from laboratory behavior, clean-water tests, site history, and the instrument’s documented operating range rather than from an arbitrary percentile of the contaminated record.

When a quality flag or optical diagnostic identifies aeration, the preferred action is to mark the observation as suspect rather than silently replace it. For short gaps, interpolation may be acceptable for visualization or a low-frequency summary, provided the gap duration and method are recorded. For compliance calculations or mass-load estimates, retaining a missing value may be more honest than inventing a concentration.

Median filters and Hampel filters can remove isolated spikes while preserving broad plume changes. A moving median is often more resistant to bubble outliers than a moving mean. However, filter width must reflect the time scale of the process. An overly wide window can flatten a real sediment pulse, shift its apparent timing, and understate peak exposure.

A rule-based correction can combine several indicators: a high turbidity value, an extreme first derivative, a concurrent optical signal anomaly, and a short duration. A confidence score is often more informative than a simple accepted/rejected label. For example, observations can be classified as valid, probable bubble interference, indeterminate, or invalid. The classification should remain attached to each timestamp in the exported dataset.

Correction approach Best use Main risk Recommended output
Physical relocation or shielding Persistent aeration at the installation point Moving the sensor away from the representative plume Revalidated raw measurements
Diagnostic thresholding Instruments with optical or quality channels False rejection when sediment also causes signal changes Quality flag plus original value
Median or Hampel filtering Isolated, short-duration spikes Removal of genuine narrow sediment events Filtered series and filter parameters
Gap interpolation Short, clearly identified bubble gaps Creating unsupported values during major plume events Estimated value marked as imputed
Model-based correction Repeated, well-characterized aeration patterns Overfitting to one flow condition Corrected estimate with uncertainty
Data exclusion Severe saturation or optical blockage Loss of important event information Missing value and exclusion reason

Validate corrections with independent evidence

A corrected turbidity record should be tested against observations that were not used to create the correction rule. Collect discrete water samples during low, moderate, and high-flow conditions, including periods with suspected aeration. Laboratory suspended-solids results can help determine whether a high optical reading represents a real concentration increase, although bottle sampling may miss short-lived events.

Use a second sensing method when possible. Acoustic backscatter, a transmissometer, a duplicate optical probe, or a sensor mounted at a different depth can provide valuable comparison. Agreement between independent methods supports the sediment interpretation. A response isolated to one optical instrument, especially when paired with raw-signal instability, strengthens the bubble-interference diagnosis.

Review the time alignment of all data streams. A bubble burst at the instrument may occur seconds after a pump command, vessel movement, or velocity change. Clock drift, telemetry buffering, and different averaging intervals can create false disagreement between sensors. Synchronize timestamps and preserve the original sampling intervals before applying resampling or smoothing.

Performance metrics should describe both the cleaning process and its effect. Report the percentage of observations flagged, the duration of the longest gap, the number of high-turbidity events retained, and any change in mean, maximum, percentile, or cumulative load. A correction that reduces the maximum value by 80 percent deserves a clear physical explanation and supporting evidence.

For long deployments, maintain an event log with sensor depth, orientation, cleaning dates, weather, discharge conditions, equipment activity, and maintenance. This record turns data processing from an opaque editing step into a traceable measurement procedure. It also helps determine whether a new mounting arrangement has genuinely reduced aeration.

Apply a disciplined field workflow

The following practices help maintain the distinction between a corrected measurement and an unsupported guess:

A field protocol should also include a clean-water or quiescent-water check where practical. This establishes the sensor’s baseline response and confirms that unexpected instability is associated with the flow rather than electronics, fouling, or a damaged optical window. Repeat the check after retrieval because sediment abrasion and deposits can change the optical response over time.

When turbidity monitoring is integrated with mobile platforms, the flow around the sensor deserves special attention. A drone, boat, or remotely operated vehicle can disturb the water and generate its own bubbles. Aerial sampling can improve spatial coverage, but the sensor’s position relative to the vehicle and the water surface must be documented. Guidance on a drone sampling workflow is useful when planning an integrated survey rather than treating the optical measurement as independent of the platform.

Preserve uncertainty in reported turbidity

Bubble correction should produce more than a visually smooth chart. Every processed value needs a clear status: measured and accepted, measured but suspect, filtered, interpolated, or excluded. If a model estimates the likely sediment concentration during a bubble event, its uncertainty should be wider than that of a directly validated measurement.

For regulatory or project reporting, separate the raw maximum from the corrected maximum and explain the screening criteria. This prevents a correction from appearing to conceal a real exceedance. If the evidence cannot distinguish a bubble artifact from a short sediment pulse, retain the event as indeterminate and discuss it as a limitation.

Mass loading calculations require particular care because short peaks can dominate the result. Compute loads from accepted and flagged data separately where possible, then present the range created by alternative treatments. For example, a conservative estimate may retain all high values, while a screened estimate excludes observations with confirmed optical saturation. The difference between those results is useful information about measurement confidence.

The best correction is usually a combination of better hydraulics, better diagnostics, and transparent processing. Optical turbidity data can remain highly valuable in dredging plume monitoring, environmental research, hydrology, defense work, and OEM systems when the record shows how air entrainment was evaluated rather than simply hidden.

Use the resulting workflow as a controlled part of the monitoring system: inspect the installation, capture diagnostic channels, validate suspicious events, and preserve a complete audit trail. Apply those steps to the next deployment so bubble interference becomes a measurable source of uncertainty instead of an unexplained spike in the final dataset.