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

Building A Reliable Long-Term Turbidity Dataset

Long-term turbidity monitoring supports decisions about dredging, catchment health, mine-water controls, river restoration and marine construction. A dataset may contain years of readings, yet still be difficult to defend if calibration records are incomplete, sensor fouling is undocumented, or changes in equipment are mixed with genuine changes in water quality. A quality assurance plan turns routine measurements into evidence that can withstand technical review.

For Australian projects, the plan needs to reflect strong seasonal contrasts, remote sites and different water types. A sensor deployed near the Great Barrier Reef may experience tidal movement, fine resuspended sediment and intense wet-season runoff, while a Murray–Darling Basin station may face low flows, algal material and long periods between visits. Good quality assurance connects instrument performance, field practice, laboratory work and data management from the beginning.

Define The Measurement And Its Purpose

Start by stating what the turbidity dataset is intended to represent. A compliance programme may need reliable exceedance counts above a project limit, while an environmental research programme may be interested in seasonal patterns, storm responses or relationships between turbidity and suspended sediment concentration. These purposes affect sampling frequency, acceptable uncertainty, sensor placement and the level of documentation required.

Turbidity is an optical response, not a direct mass measurement. It is commonly reported in NTU or FNU, depending on the measurement method and reporting convention. Two instruments can produce different values in the same water because particle size, colour, mineral composition, organic material and sensor geometry influence scattered light. The quality plan should specify the reporting unit, instrument configuration, measurement interval, averaging method and any conversion to suspended solids.

The monitoring objective should also identify the population represented by each reading. A near-bed sensor in a tidal channel does not describe the whole water column, and a fixed instrument in a sheltered creek may miss short-lived sediment pulses. Record the site coordinates, mounting height, depth, flow conditions and reasons for selecting the location. Technical information from D & A Instruments can help frame decisions about optical sensors and their use in marine and freshwater environments.

Design Sampling Around Natural Variability

A robust sampling design captures the processes that are expected to change turbidity. Continuous logging is useful for identifying storm peaks, dredging events, tidal cycles and rapid drawdown responses. However, a high-frequency logger does not automatically provide a representative dataset. The installation needs to be positioned where water movement is adequate, the sensor remains submerged, and the measured signal relates to the management question.

In northern Australia, a wet-season discharge event can alter water quality within hours, followed by weeks of declining flows. A schedule based only on monthly site visits may miss the most important part of the record. In coastal Queensland, the timing of tides and dredging activity may be as important as rainfall. In the Murray–Darling system, irrigation releases, low-flow conditions and algal blooms can create different optical conditions from those observed during high-flow events.

Define routine and event-based sampling separately. Routine measurements establish baseline conditions, while event sampling can be triggered by rainfall, rising water level, equipment alarms or a construction activity. Where a project has several stations, document whether they are intended as upstream references, downstream impact sites, source-area stations or recovery locations. This prevents later interpretation from treating all stations as interchangeable.

A parallel grab-sampling programme is valuable when suspended sediment concentration is required. Samples should cover low, medium and high turbidity conditions, since a single laboratory comparison may produce a misleading conversion. Collect samples close to the sensor, record the time difference precisely and include storm or operational events that challenge the normal calibration range.

Control Calibration And Field Performance

Calibration should be treated as a continuing process rather than a commissioning task. Before deployment, inspect the optical window, cable, connectors, wiper or anti-fouling components and data logger settings. Confirm the serial number, firmware version, calibration date and expected measurement range. A bench check with suitable standards can identify a damaged or unstable instrument before it reaches a remote station.

Field verification should combine an instrument check with an independent observation. Depending on the project, this may involve a turbidity standard, a cleaned sensor comparison, a duplicate instrument or a water sample analysed by an approved laboratory. The method, standard value, temperature, time, operator and acceptance limits should be recorded. A field check is evidence about performance at that moment; it does not prove that every preceding reading was correct.

Cleaning intervals should reflect the site rather than a fixed calendar. Warm, nutrient-rich water can encourage biofouling, while fine clay may coat an optical window after a flood. In Darwin or Far North Queensland, a sensor may need more frequent inspection during productive periods than during the dry season. Record the condition of the sensor before cleaning, since removing fouling without documenting it can conceal the likely start of measurement bias.

Establish rules for drift, failed checks and out-of-range values. For example, a result outside the acceptance limit might require a repeat check, review of nearby readings, comparison with a reference instrument or flagging of the affected interval. Corrections should never overwrite the original data. Preserve the raw value, the reason for the adjustment, the person authorising it and the calculation used to produce the qualified value.

Manage Sediment Relationships And Uncertainty

A turbidity-to-suspended-solids relationship is usually site-specific and can change over time. A catchment may deliver coarse mineral particles during one flow event and fine organic-rich material during another. Dredging can produce a particle distribution different from natural runoff. For that reason, regression equations should be based on samples that represent the range of conditions in the dataset rather than borrowed from another river or estuary.

Laboratory samples need their own quality controls, including chain of custody, preservation requirements, analytical method, reporting limit and laboratory quality status. In Australia, the laboratory’s accreditation and scope should be appropriate to the analysis, particularly where results support regulatory reporting or contractual obligations. Compare laboratory results with sensor readings collected at the same location and as close in time as practical.

Quantify uncertainty in a way that matches the intended use. Relevant components may include sensor repeatability, calibration-standard uncertainty, fouling, temperature effects, sampling variation, laboratory error and the uncertainty of the regression model. An exceedance close to a trigger value should be interpreted differently from a reading far above it. The plan should define how borderline results are reported and whether a confidence interval or decision rule is required.

Use terminology consistently across field sheets, databases and reports. Terms such as calibration, verification, validation, detection limit, drift and suspended solids can be used differently by different teams. A water-quality glossary of terms provides a useful reference when establishing shared definitions for field technicians, data analysts, consultants and regulators.

Build A Traceable Data Workflow

Every record should retain enough metadata to explain how a value was produced. At minimum, store the station identifier, timestamp and time zone, sensor serial number, depth, logger configuration, raw reading, processed reading, quality flag and maintenance history. Store rainfall, water level, flow, salinity or conductivity where these variables help interpret the optical signal.

Synchronise clocks before deployment and check them during servicing. Time errors can make a valid sensor appear inconsistent with rainfall, tide or a dredging shift. Use a consistent daylight-saving rule in Australia and make the convention explicit. UTC may be suitable for centralised systems, while local time can be useful for field operations; either approach is defensible when it is documented and applied consistently.

Data should pass through controlled stages: raw download, technical review, approved dataset and published or reported extract. Each stage should have access controls and a clear version identifier. Automated checks can identify impossible values, duplicated timestamps, flat-lined signals, abrupt jumps, excessive gaps and readings recorded while a sensor was out of water. Automated flags should prompt investigation, not silently delete observations.

A practical data review should examine both individual values and patterns. Plot turbidity against water level, rainfall, flow, tide and operational events. A slowly rising baseline may indicate fouling, while a sudden step change may indicate a relocation, cable issue or change in particle type. A perfectly stable signal during a highly variable flow period can be as suspicious as an extreme spike.

Records Worth Keeping

Govern Reviews And Long-Term Changes

Assign responsibility for each part of the quality system. Field staff should be authorised to identify fouling, damage and installation problems, while a data custodian should control imports, flags and corrections. A technical reviewer should assess calibration performance, unusual patterns and the suitability of any suspended-solids conversion. Separating data entry from approval reduces the chance that an unnoticed mistake becomes part of the official record.

Set review intervals around the project risk. A construction compliance dataset may need daily automated checks and rapid review of trigger exceedances. A research network may use weekly screening with monthly technical review. Annual assessments should examine sensor performance, missing-data rates, calibration stability, changes in site hydraulics and whether the monitoring design still represents the stated objective.

Long-term records often outlive individual instruments, contractors and software systems. When a sensor model changes, run the old and new instruments together for an overlap period where feasible. Document changes to optical geometry, firmware, calibration standards, mounting height, logging interval and data processing. A change in the median value after replacement should not be interpreted as an environmental trend until equipment comparability has been assessed.

Plan for remote access and difficult retrievals. Stations in outback Queensland, northern wetlands or offshore areas may remain unattended longer than intended because of flood conditions, access restrictions or vessel availability. Spare parts, desiccant, batteries, protective housings and duplicate sensors can reduce data loss. The plan should state how gaps are handled, when interpolation is prohibited and how uncertainty increases when an instrument is not checked on schedule.

Apply Clear Flags And Reporting Rules

Quality flags should distinguish different conditions instead of reducing every concern to “bad data”. Useful categories may include accepted, estimated, suspect, rejected, below instrument range, above instrument range, sensor exposed, maintenance period and affected by fouling. The flag definitions must be stable across the full project so that a future analyst can interpret older records without relying on personal knowledge.

Do not remove unusual observations simply because they are inconvenient. A very high value may represent a genuine flood plume, propeller wash or dredging pulse. Investigate it against field notes, photographs, rainfall, water levels, instrument diagnostics and nearby stations. If the value is retained, explain the supporting evidence. If it is rejected, preserve the original record and state the technical basis for rejection.

Reports should separate measured data from derived products. A graph of raw turbidity, a quality-controlled time series, a daily percentile summary and a suspended-solids estimate answer different questions. State the proportion of missing or qualified data, the calibration period, the laboratory relationship used and any thresholds applied. For Australian approval and monitoring programmes, transparent treatment of trigger exceedances is particularly important because stakeholders may include regulators, Traditional Owners, port authorities, contractors and local communities.

At the end of each reporting period, archive a data package that another competent person can reproduce. Include the final dataset, raw files, metadata, quality-control scripts or formulas, calibration documentation, laboratory records, field logs and a change history. A long-term turbidity record becomes far more valuable when its provenance remains clear through staff changes, instrument replacements and future reanalysis.