Stormwater from urban runoff causes contamination of our waterways. Understanding the nature of stormwater contamination is important to designing, installing, and maintaining effective stormwater treatment systems (STS). Inherent variability in STS performance, however, can result in unmet water quality improvements.

Variability in STS performance occurs through contaminant generation, operation of STSs, and sample collection and processing. Build-up processes have a large effect on contaminant concentrations in runoff, as they relate to land use variability and atmospheric deposition, and are not fully understood on their own. Rainfall intensity and total depth affect contaminant wash-off processes, which are further affected by surface characteristics as well as the specific form of the contaminants (i.e. particulate or dissolved). Variability due to treatment systems is due to differences in design, operation, and maintenance regimes. When monitoring techniques are finally deployed to quantify untreated and treated contaminant concentrations, variability is expected from field collection of samples and laboratory analyses.

This study aimed to identify and quantify the factors impacting variability in the event mean concentration (EMC) of untreated and treated runoff by monitoring the performance of proprietary STSs. Through an improved understanding of the sources of variation, water quality improvement work can be more focused and effective.

Ten unique treatment systems were monitored, consisting of four unique types of proprietary systems. An experimental media, crushed mussel shells, was used within two different types of the proprietary STSs. To reduce the effect of land use changes on load variability in a field-based experiment, the monitored catchments consisted of roof surfaces and one small mixed catchment. The roofs contributed predominantly dissolved zinc, while Zn and total suspended solids (TSS) were targeted at the mixed catchment.

Sampling techniques included flow-proportional auto-sampling and whole-of-event capture. To observe water quality fluctuations throughout a storm at the mixed catchment, aliquots from two auto-samplers were processed individually. A novel, whole-of-event capture system was used at the roof catchments that monitored flow and split the runoff into untreated and treated flows.

Identification and quantification of factors affecting variability was carried out through a mixture of qualitative and quantitative assessments. STS condition at the time of treatment was recorded to qualify its impact on performance variability. To determine the effect of influential variables on STS performance, multiple linear regressions were used. Results were verified using a classification and regression tree (CART) analysis, and through coefficient values from the multiple linear regression.

Variability in treatment performance was influenced primarily by the concentration of untreated contaminant. Untreated EMC variability was observed in all catchments, and is an inherent characteristic of stormwater. The variability of untreated galvanised roof runoff was primarily due to the age of the surfaces themselves, with more variation observed at the older roof surface. Outside of surface type and condition, rainfall depth and precipitation intensity, both peak and average, most significantly affected the dissolved Zn EMC and total load.

The mussel shell media within the two types of treatment systems performed well and averaged near 80 % removal efficiency of dissolved zinc, with mean untreated EMCs around 200 µg/L. A 10 % decline in removal efficiency was correlated to the cumulative 500 mm of rainfall depth treated by the smaller, downpipe treatment systems.

Faults in design, installation, and a lack of timely maintenance were the largest sources of treatment performance variability. Consistent monitoring demonstrated that alternative maintenance schedules, such as surveillance of volumetric load treated or special inspections following shock events, will likely improve the performance of STSs.

EMC measurement uncertainty originates primarily from the use of different monitoring techniques. The contribution to uncertainty from laboratory processing techniques is relatively low, compared to the use of different field-based monitoring techniques. The field-based impact can be reduced through consistent sampling methods that take flow-weighted aliquots across the entire duration of a runoff event. Uncertainty decreases with a focus on flow-proportional sampling throughout the entire storm duration, rather than a high frequency of samples taken over one segment.

Variability is inherent in stormwater treatment system performance and is linked to the variability of the untreated runoff quality itself. Oversizing STSs may decrease performance variability, and increase resilience in the face of larger, more intense storms due to climate change. The importance of maintenance of existing STSs cannot be over emphasised. Traditional, temporal-based maintenance regimes can be improved to include more consistent visual checks or remote monitoring. This has the potential to increase performance of STSs and use allocated funds more effectively. The largest gains to STS performance are likely to come from contaminant load identification and targeted design, followed by proper installation and regular maintenance.