Estimating sediment yield with optical backscatter sensors
Optical backscatter sensors provide a practical way to track fine sediment in moving water without collecting a sample every minute. Their output responds to particles that scatter light, so it can act as a continuous proxy for suspended sediment concentration when calibrated for the stream, sediment and sensor. Small catchments make that work both valuable and demanding: short, intense runoff events can move most of the annual load before a routine visit.
This article explains a field-to-budget workflow for Australian teams using optical backscatter sensors to estimate suspended sediment yield from small catchment streams. It links turbidity to laboratory-measured solids, combines concentration with discharge, addresses the timing of samples and runoff, and keeps uncertainty visible. The result is not a laboratory instrument used as a black box, but a defensible estimate of how much sediment leaves a catchment during a storm, a month or a water year.
From optical signal to sediment concentration
An optical backscatter sensor directs light into the water and measures light returned from suspended particles. The signal depends on particle concentration, size, shape, colour and refractive index, as well as the sensor’s optical geometry. Clay and fine silt can produce a strong response; coarse sand may settle rapidly or contribute little even when its mass is significant. Bubbles, algae, dissolved colour and organic detritus can also change the reading without adding mineral sediment.
Accordingly, turbidity in nephelometric turbidity units is an optical property, not a sediment mass. A site-specific relationship is needed to translate it into total suspended solids concentration. The relationship may be represented as SSC = a × turbidity^b, although the best form should be selected from paired observations rather than assumed. A high correlation does not guarantee a low bias, so residuals, limits of validity and performance at very low and very high concentrations all need attention.
Sensor selection also matters. Instruments based on 880 nm infrared backscatter are widely used for turbidity because many mineral sediments respond strongly at that wavelength, but wavelength alone does not remove site effects. A sensor with good wet calibration traceability, a suitable optical range and replaceable wiper or cleaning capability is usually more valuable than one with a sophisticated digital output but poor field maintenance.
Building a reliable turbidity–sediment relationship
A useful calibration set contains paired turbidity and laboratory suspended-solids results collected across the range the logger will see. Include clear-weather baseflow, ordinary runoff, storm peaks and recession, and sample rising as well as falling limbs. Samples collected only during one flood are unlikely to define annual behaviour. Well-distributed pairs are generally more valuable than many duplicates from one narrow flow window.
Each pair should share a timestamp and depth, and the laboratory result should come from a documented filtration and drying method. The guide to paired-data turbidity regression explains suitable transformations, influential observations and checks on the fitted relationship. Log turbidity and laboratory concentration before fitting, inspect residuals against discharge and time, and retain rejected points with reasons rather than silently removing them.
Validate the relationship with samples not used to fit it. Quantify confidence intervals for concentration and sediment load, and record the concentration and discharge range over which interpolation is acceptable. Extrapolation beyond storm peaks should be flagged. If turbidity frequently plateaus because the instrument is saturated or water contains large amounts of non-sediment scatterers, the rating remains unsuitable even when the regression statistics look strong.
Designing representative sediment sampling
At each visit, take samples with a depth-integrated sampler where practicable, because near-bed concentration may differ from the surface. A hand-held grab at one point can bias the rating towards either coarser near-bed material or the clearer upper layer. Record the exact time, stage, discharge estimate, sensor reading and sampling location; ambiguous timing creates false scatter when the event curve is reconstructed.
Run field blanks, duplicates and occasional split samples through the laboratory’s quality-control system. Use pre-weighed compatible filters, a consistent drying temperature and a balance traceable to Australian laboratory requirements. Distinguish total suspended solids from fixed or mineral suspended solids where organic material is important. That distinction can reveal a rainfall or fire-related change in optical response and prevent algal or detrital material being credited as eroded soil.
Schedule sampling around forecast rainfall where safe, because the most informative pairs occur while concentration is changing quickly. If access is restricted, automatic samplers can collect time-composited samples, but intake flow and settling still require checks. Keep bottles chilled where required, avoid long storage delays and document any screen, filter or preservative change, since each analytical decision alters the calibration data.
Capturing runoff in small Australian streams
Continuous logging is essential because small streams can rise and peak within minutes. A stage or pressure sensor supplies the timing and shape of the hydrograph; a velocity sensor or validated stage–discharge rating supplies flow. Sample at several stages, measure velocity across the section and check for bank erosion, debris, gravel movement and changes in channel roughness. Discharge is not a secondary detail: concentration multiplied by an uncertain flow creates a correspondingly uncertain load.
Greater Melbourne’s peri-urban creeks illustrate the problem. Rain falling on roofs, roads and compacted development areas can produce a sharp turbidity pulse that arrives before a rural catchment’s more gradual sediment wave. Around Adelaide and the Adelaide Hills, dry summers and intense winter storms can separate long periods of little transport from short sediment-laden events. Instrument sites must therefore remain through the dry interval and be ready before forecast rain, not activated only after the channel is already muddy.
Murray–Darling tributaries present another Australian pattern: irrigation return flow, bank erosion and episodic rainfall can produce loads that vary by reach and season. In remote areas, solar power, telemetry and safe maintenance access can determine whether a complete record is possible. Data gaps should be retained in the audit trail and filled with measured events or clearly labelled estimates, never treated as zero sediment simply because the sensor stopped.
Converting concentration into sediment yield
First convert the sensor-derived concentration to a consistent mass unit, then multiply by discharge. With concentration C in grams per cubic metre and discharge Q in cubic metres per second, C × Q gives the instantaneous suspended load in grams per second. Divide by 1,000 for kilograms per second. A simple event total in tonnes can be calculated as 0.0864 × Σ(CᵢQᵢΔtᵢ), using intervals Δtᵢ in seconds. The unit conversion is a common source of order-of-magnitude error.
Integrate over the full rising, peak and falling sequence. Trapezoidal integration is usually adequate when intervals are short relative to the hydrograph, but concentrated and sloping stages need closer spacing. Do not assume a fixed relationship between peak concentration and peak discharge; sediment and water peaks can be offset. If a continuous rating is unavailable, estimate concentrations at measured points, interpolate conservatively and carry the resulting uncertainty through the load calculation.
Sediment yield is the accumulated suspended load divided by drainage area to express a mass per unit area, or it may be reported as a total mass for the event or period. Clearly distinguish suspended-sediment yield from total erosion, which also includes bed load and bank material not measured by the optical sensor. A small catchment budget should cover storm events and routine periods, report sampling coverage and distinguish measured, interpolated and extrapolated contributions.
Checking accuracy and maintaining the instrument
Optical sensors drift or become biased through biofilm, mineral coatings, scratched windows, trapped bubbles and cable or wiper faults. Check the window before and after cleaning, compare the reading with a portable turbidity meter or reference sample, inspect the wiper stroke and confirm that timestamps remain aligned. A structured optical sensor audit turns these observations into a repeatable field record and helps separate gradual calibration drift from abrupt fouling.
Use a control chart of turbidity against laboratory concentration or against stable reference readings. A change in slope may indicate that the sediment mix has changed; an offset may indicate fouling or sensor drift. Clean and recalibrate only after the event that revealed the problem has been sampled, or retain enough reference data to revise the rating later. Record every intervention, including time, water condition, method and person responsible.
For Australian buyers, local support and spare-part planning are practical procurement issues. D & A Instruments’ product information and contact arrangements are now supported by Campbell Scientific, which can be considered when defining maintenance and calibration access. Imported instruments also pass through Australian Customs and may attract GST, freight and clearance costs; factor those expenses, as well as lead time for replacement optics, into the total monitoring budget.
Recommendations for dependable sediment budgets
A dependable sediment budget depends on a traceable chain from raw optical signal to reported yield. The following measures provide a practical minimum for small Australian catchments:
- Calibrate against laboratory total suspended solids across baseflow, storm and recession conditions, then reserve independent samples for validation.
- Pair every concentration observation with defensible discharge data and retain raw signals, calibration records and event metadata.
- Log continuously across forecast storms, record data gaps and sample both rising and falling limbs.
- Inspect optics regularly, test for fouling and establish a maintenance and calibration-audit schedule.
- Report units, sampling coverage, uncertainty and the contribution from extrapolated periods beside the final load.
A credible annual result is therefore not simply a multiplied turbidity series. It is an event-resolved sediment budget with traceable calibration, defensible flow, documented maintenance and a clear account of what was measured versus inferred. That record supports comparison between catchments, evaluation of remediation works and better estimates of sediment pressure on Australian rivers and reservoirs.