Interpreting turbidity time series during storm events

Storms turn a relatively stable water-quality record into a fast-moving account of erosion, runoff, transport and deposition. A turbidity sensor may register a sharp rise within minutes of rainfall, yet the highest suspended-sediment concentration can occur hours later, after tributaries connect, banks fail or a flood wave reaches the monitoring point. Reading the record properly therefore requires more than identifying the largest value.

For sediment load estimation, turbidity must be treated as a time-varying proxy for suspended solids and combined with discharge, site conditions and a defensible calibration. The result is an estimate of the mass transported through a cross-section during a defined event. Optical instruments from D & A Instruments are designed for water-quality and sediment-monitoring applications in marine and freshwater environments, but the quality of the estimate depends equally on deployment, validation and interpretation.

Start with a clean event record

A storm-event time series should contain turbidity, water level or flow, timestamps, instrument status and, where possible, rainfall intensity. Aligning all channels to the same time zone and clock is essential. A sensor clock that is ten minutes out of synchronisation can distort the apparent relationship between rainfall, discharge and suspended solids, particularly in small urban catchments where the response is rapid.

Begin by marking the pre-storm baseline. This establishes the background turbidity before runoff enters the stream and helps distinguish a genuine event response from gradual drift. The baseline may be elevated after an earlier storm, during a construction period or in an estuary affected by tidal resuspension. A storm should generally be split from the preceding event only after turbidity and flow have returned to an agreed threshold for a suitable separation period.

Raw optical readings need quality control before they are converted into concentration. Remove values recorded during sensor startup, maintenance, communication failure or diagnostic alarms. Inspect sudden vertical jumps, long flat lines and impossible negative values. A sharp peak may represent a real sediment pulse, but it may also be caused by a leaf passing directly across the optical path, air bubbles, wiper movement or a temporarily exposed sensor.

The sensor location matters as much as the data file. A probe mounted too close to the bed can measure local scour rather than the cross-sectional average, while a shallow installation may be exposed during falling water levels. In Australian waterways, floating vegetation, sticks and fine organic material can be especially common after intense summer rainfall. Photographs, field notes and maintenance records provide useful evidence when an unusual feature appears in the trace.

Read the shape of the storm response

The rising limb often shows the first flush of readily available material. Fine particles from roads, drains, cleared ground and channel margins can produce a rapid turbidity increase before discharge reaches its maximum. If the turbidity peak precedes the flow peak, the catchment may have a short sediment-delivery pathway or a limited supply that is exhausted early in the event.

A delayed turbidity peak has a different meaning. It may indicate that sediment is being supplied from distant sub-catchments, that tributary inflows arrive later, or that increasing velocity is progressively eroding the bed and banks. In a large basin such as the Murray–Darling, travel times can extend well beyond the rainfall burst observed at one gauge. In steep coastal catchments near Brisbane or Cairns, the delay may be much shorter and the hydrograph far more abrupt.

Plot turbidity against discharge as well as against time. A clockwise loop, in which concentration is higher on the rising limb than at the same discharge on the falling limb, commonly indicates a nearby or rapidly available sediment source. An anticlockwise loop can point to delayed delivery from upstream areas, remobilisation from floodplains or changing hydraulic conditions. Multiple loops may reveal several sediment pulses within one storm.

The largest turbidity value does not necessarily represent the largest sediment load. Load depends on concentration and water volume, so a broad period of moderate turbidity can transport more mass than a brief spike. This is particularly important when comparing a short, intense thunderstorm with a long frontal rainfall event affecting a large Australian catchment.

Convert optical response into suspended solids

Turbidity is measured through the scattering or attenuation of light, whereas suspended-sediment concentration is a mass per unit volume. The relationship between the two depends on particle size, mineral composition, colour, shape and organic content. A calibration developed from clear quartz-rich sediment may perform poorly when the storm carries dark clay, iron-stained material or organic debris.

Collect water samples across the full range of event conditions, including the rising limb, peak and recession. Laboratory results for total suspended solids or suspended sediment concentration can then be paired with the sensor value recorded at the sampling time. A linear relation may be adequate over a narrow range, but many sites require a log-linear or segmented regression. The selected model should be based on residuals and independent validation rather than convenience.

For a sample with concentration (C) and a sensor reading (T), a basic calibration might be written as (C=aT+b). If the relation is fitted in logarithmic space, retransformation bias must be considered before reporting concentrations. Separate calibrations may be justified for different sediment regimes, especially when stormwater, estuarine resuspension and catchment runoff reach the same monitoring point.

Instrument configuration also affects comparability. A turbidity monitor with a fixed optical path can respond differently from a laboratory benchtop measurement, and readings reported in NTU or FNU should not be treated as interchangeable without checking the instrument method. The weather station integration information is useful when rainfall, wind and atmospheric observations need to be interpreted alongside the water-quality record.

Calculate event sediment load

Once the turbidity series has been converted to concentration, combine it with discharge at matching time steps. The basic event load is the time integral of concentration multiplied by flow:

[ M=\int_{t_0}^{t_1} C(t)Q(t),dt ]

Here, (M) is mass, (C(t)) is suspended-sediment concentration, and (Q(t)) is discharge. If concentration is in milligrams per litre and flow is in cubic metres per second, the result must be converted carefully: one cubic metre contains 1,000 litres, and the time increment in seconds determines the final mass. A spreadsheet or script should retain the units in every intermediate calculation.

For regularly sampled data, the trapezoidal method is often suitable:

[ M \approx \sum \frac{C_iQ_i+C_{i+1}Q_{i+1}}{2}\Delta t ]

The interval should be short enough to capture the fastest changes in the event. Five-minute data may be appropriate for a small urban stream, while fifteen- or thirty-minute data may be acceptable for a slower river. If a narrow turbidity peak is sampled only once, interpolation can substantially underestimate its contribution. Conversely, noisy high-frequency readings can inflate the load unless quality-control rules are applied.

Discharge is another major source of uncertainty. A water-level-to-flow rating curve may be unreliable during overbank flow, scour, backwater or rapidly changing flood conditions. Check whether the rating curve covers the observed stage range and whether hysteresis affects the relationship between rising and falling water levels. Where direct flow measurements are unavailable, report the sediment estimate as a range based on plausible discharge scenarios rather than presenting excessive precision.

Define the event boundaries consistently. Starting too early includes background transport; ending too soon omits the long recession tail, when lower concentrations may still be carried by substantial flow. Record the separation threshold, start and end times, calibration equation, missing-data treatment and unit conversions so that another analyst can reproduce the calculation.

Manage uncertainty and long-term sensor performance

Storm-event sediment estimates combine uncertainty from sampling, calibration, sensor response, discharge measurement and data gaps. A useful report separates these sources instead of assigning a single vague confidence statement. Bootstrap resampling of calibration pairs, alternative rating curves and sensitivity tests on event boundaries can show which assumptions matter most.

High turbidity can exceed the useful range of an optical sensor, producing a clipped or flattened peak. In that case, the missing concentration cannot be recovered reliably by simply extending a linear calibration. Dilution samples, a second measurement range or a site-specific high-concentration calibration may be needed. The same caution applies to very low readings near the instrument’s detection limit, where small electronic or optical changes can have a large relative effect.

Biofouling is a persistent issue in warm, nutrient-rich waters. Algae, bacterial films and fine sediment on the optical window gradually alter the signal and may create a slow upward drift that resembles increasing background turbidity. Practical controls include regular cleaning, antifouling hardware, protective placement and comparisons with grab samples. Guidance on fouling mitigation strategies is particularly relevant for Australian lakes, irrigation channels and tropical waterways.

Interpret the results in their catchment setting. A storm monitored near Sydney may combine impervious urban runoff with creek-bank erosion, while a northern Queensland event may follow monsoonal rain or a tropical cyclone and carry substantial material into coastal waters. For projects affecting the Great Barrier Reef, event loads can be compared with rainfall totals, land-use records and sampling from receiving waters. In every setting, the strongest interpretation comes from linking the turbidity trace to rainfall, hydrograph shape, field observations and sediment characteristics rather than treating a single peak as the whole story.