EPA Victoria Pollution Profiling Program
Passive samplers and networked sensors in urban stormwater
Industrial pollution is a major problem across Australia. The nature of the business activity increases the risk of toxicants making their way down to local rivers and creeks. By identifying areas that are most at risk of generating stormwater pollution, more targeted education and enforcement programs can be implemented.
Point source pollution continues to be one of the main drivers of ecological stress on urban waterways. Heavy metals, hydrocarbons and pesticides from urban catchments make their way through stormwater networks into waterways. These pollutants can flush into receiving waterways and can accumulate in sediments where they can persist and cause toxicity to benthic fauna. Stress and toxicity from pollutants can have major consequences for the long-term health of aquatic ecosystems.
As part of EPA Victoria's (EPA) Research and Development Program in 2020-2021, new real-time wireless water quality sensors and stormwater passive samplers were trialled to detect pollutants flowing from the catchment into Merlynston Creek. This catchment was chosen because of its history of poor water quality associated with several industrial fires, including SKM in 2017 and Campbellfield Industrial Services in 2019.
Currently, most water quality monitoring programs involves the collection of discrete grab samples, which are reflective of pollution concentrations at that point of time. Further sampling based on laboratory analysis requires field staff to collect water from a single point in time. While EPA also receives pollution reports from the community, these pollution events have often already passed before EPA can respond and attend a site.
Passive sampling and in-situ monitoring allows for sampling to occur over a longer time-scale providing more data points over time across the catchment. These new technologies use real-time sensors to track concentrations of pollutants (including ammonia, copper and hydrocarbons) to identify pulse pollution events, trends and potential sources. Automatic alarms triggered via the visualisation dashboard will enable responsive actions based on 24/7 real-time pollution data.
Isolating high risk areas within a catchment using StormscoutTM technology will allow EPA to develop targeted programs focusing on high-risk sub-catchments and allow local management authorities to deliver programs more efficiently or to initiate more intensive monitoring. This type of surveillance may be critical to EPA's regulatory response, allowing for problem sites to be identified and for pollution to be tracked to the source.
The aim of this project was to develop pollution profiles for the Merlynston Creek upper catchment area and identify areas most suitable for real-time monitoring of high-risk pollutants.
Bio2Lab deployed StormScoutTM samplers# into stormwater drains at 10 locations for five weeks (Figure 1). Samplers were always deployed for one week, constituting a sampling event. After each sampling event, media from the sampler was collected, processed and analysed for common persistent contaminants*, and a new sampler deployed.
All chemical analysis was performed by consulting laboratories accredited by the National Association of Testing Laboratories Australia (NATA). We converted contaminant concentrations into a simple colour-coded pollution risk rating for ease of interpretation, with colour ratings for each pollutant based on concentrations detected. The site risk score was determined by average pollutant concentrations across the five weeks of sampling. Pollution risk ratings used in this report are based on summary statistics derived from stormwater pollutant concentrations detected in StormScoutTM samplers over a 12-year period (see links to the interactive dashboard below). For each site, measured concentrations were compared to the summary statistics and grouped accordingly.
Real-time stormwater pollution sensors
Real-time pollution sensors were deployed at seven locations in the Merlynston Creek catchment. Data was continually streamed via the 4G network every 15 minutes to a data visualisation dashboard (see links to the interactive dashboard below). Ammonia, pH and copper were monitored from 12th October 2020 - 12th March 2021 at three locations along Merylnston Creek. Water was collected periodically for spot-checking purposes. Volatile organic compounds (VOCs)and water level were monitored in real-time at four locations in the stormwater network across the catchment (Figure 3).
*Note: mercury was not included in the analysis. This was due to a change in the type of StormScoutTM sampler used from “StormScout - Collect” to “StormScout- Adsorb”. As this sampler is not effective at accumulating mercury, it was decided to omit it from the list of parameters to be analysed.
# StormScouts use a artifical media comprised of organic carbon and other media specifically designed to adsorb pollutrants in stormwater over time (7-14 days).
Ammonia Passive Samplers
Ammonia passive samplers were deployed for ten weeks at three key sites in Merlynston Creek (Link Drive, Maffra St and Fawkner). Fabrication and analysis of ammonia passive samplers was completed at the University of Melbourne, School of Chemistry. Samplers were deployed over a 7-day period. Analysis was done by flow system with a limit of detection (LOD) for NH4+ of 0.08 mg/L to 0.13 mg/L. Quality control included multiple sample (trip and field) blanks and the mean concentration blank was subtracted from the passive samplers that were deployed into the creek. Measurement uncertainty was estimated between 0.8 – 7.1% against quality control standards (0.5 and 2 NH4+ mg/L). The total ammonium (NH4+) in each sampler was converted to a time weighted average concentration (CTWA) (see Almeida et al. 2016; based on equation CTWA = CRS/0.618). This value is reflective of the amount of ammonia accumulated in the device during the deployment time and gives us a better understanding of pollutant concentrations over large temporal scales when compared to grab sampling techniques. The final concentrations (CTWA) were reported as ammonia-N (NH3-N mg/L).
Almeida, M.I.G., Silva, A.M., Coleman, R.A., Pettigrove, V.J., Cattrall, R.W. and Kolev, S.D., 2016. Development of a passive sampler based on a polymer inclusion membrane for total ammonia monitoring in freshwaters. Analytical and bioanalytical chemistry, 408(12), pp.3213-3222.
Šraj, L.O.C., Almeida, M.I.G., McKelvie, I.D. and Kolev, S.D., 2017. Determination of trace levels of ammonia in marine waters using a simple environmentally-friendly ammonia (SEA) analyser. Marine Chemistry, 194, pp.133-145.
Automatic alarms triggered via the visualisation dashboard will enable responsive actions based on 24/7 real-time pollution data.
Real-time pollution sensors can now be used to monitor contaminants (including ammonia, copper and hydrocarbons), detect pulse pollution events, highlight trends and identify potential sources.
Results and discussion
Passive samplers were deployed for five weeks at 10 pre-selected locations (Figure 1). Laboratory results from analysis of the media collected from each passive sampler were converted to pollution risk ratings, providing a catchment level pollution profile. Clicking on each site in the interactive dashboard provides site level pollution profiles and the associated risk from each pollutant as assessed against thresholds.
The background reference site, 116 Maffra St, had the lowest levels of pollution in the catchment. In contrast, both 82 Maffra St and Maffra St Main drain outfall had the highest levels of contamination. The sub-catchment associated with 82 Maffra St is likely to be a significant source of heavy metals, especially zinc, copper and lead. It is also likely that the very high levels of pollution found at the outfall is originating from this catchment and accumulating from further up in the catchment. It is also evident that the catchments associated with Lexton St and Kyabram St is not generating high levels of pollution. Total Petroleum Hydrocarbon is a major concern in this catchment with many pulse pollution events detected while on-site. The results from both the passive sampling project and the real-time VOC monitoring indicate that the source of these pulse events is likely originating from Lisa Place at the top of the catchment.
Further investigations focused on the main arm of the Maffra St drain are warranted. The mobility of the real-time sensor hubs trialled in parallel to this project would allow EPA to locate the source of the pollution impacting Merlynston Creek.
Figure 1 - Site locations for the StormScout sampler program
Ammonia passive sampling
There were clear differences in the concentration of ammonium between sites, with Maffra consistently higher than the other two (Figure 2). The elevated concentrations of ammonium at Maffra occurred in conjunction with substantial sources of other contaminants such as the trace metal copper, shown in StormScouts, spot water samples and real-time sensors.
Ammonium at Link Drive showed substantial variation, with elevated concentrations similar to Maffra occurring around 50% of the time, while lower concentrations similar to Fawkner occurring around 30% of the time. This pattern was also reflected in spot samples.
Ammonium concentrations at Fawkner were consistently an order of magnitude lower than both Maffra and Link Dr, indicating substantial dilution occurring over the 8 km stretch of creek, which includes Jack Roper Lake and several retarding basins.
Incorporating ammonium passive samplers as part of future studies is recommended. Additional samplers deployed at multiple sites within catchments allows for a quick and approach for isolating and sourcing contamination.
Figure 2 - Bar plot showing ammonia concentrations from the ammonia passive samplers
Real-time water quality monitoring
Real-time pollution sensors were deployed at seven locations (Figure 3). Data was streamed every 15 minutes to a data visualisation dashboard. High pollution levels triggered alarms that were sent to EPA personal SMS or email. This enabled EPA to prioritise and plan response to pollution events.
Wireless sensors were deployed for one year (March 2020 to March 2021) at different locations throughout the Merlynston Creek catchment. Results showed repeated pollution events with daily fluctuations and weekly patterns. Wireless alarms alerted EPA to high concentrations of various pollutants in particular VOCs. Spikes in pollution often occurred on Friday afternoon, suggesting washdowns at work sites. The patterns would allow EPA to plan investigations based on recurring pollution trends detected by the sensors. Lab spot checks performed throughout the project also showed that the data the sensors provided was accurate and reliable. Click on the image below to view the real-time data visualisation dashboard.
Figure 3 - Site locations for the real-time pollution monitoring sites in Merlynston Creek and stormwater drains situated in the Coolaroo industrial estate
Click on the image below to view the interactive pollution profiling dashboard
Click on the image below to view the real-time monitoring visualisation dashboard
This research and development project showed how coupling real-time pollution sensors to a visualisation dashboard with alerts enabled EPA to take quicker action to control flows, investigate pollution events and find pollution sources. The sensors were able to detect increases in contaminants and inform EPA Victoria through automatic alerts generated from the dashboard. This informed EPA's inspection strategy and improved response times to pollution events. The sensors were also used to monitor water quality following the 2020 MRI fire, which allowed EPA to better assess site clean-up actions.
Operationalising certain components and continuation of trials to better understand how real-time pollution sensors can improve areas of EPA's core business is recommended. For instance, installation of real-time VOC and level sensors into high-risk industrial sites would provide 24/7 monitoring capability. Configuring alerts to better predict the type of pollution event is also achievable. Improvement in data analytics will allow triggers to better predict the magnitude of pollution events. For instance, the signature of an industrial fire would be different to a small dumping event. Triggers could be configured to alert EPA to different types of pollution events.
This EPA Research and Development project showed clearly how sensors could enable EPA to quickly respond to pollution events and inform inspection strategy. It also showed how deploying passive sampling technology into high-risk catchments can prioritise risks and trace pollutants back to their source, thereby reducing contaminants and improving the health of waterways across Victoria.
Acknowledgement of Country
We acknowledge Aboriginal people as the first peoples and Traditional custodians of the land and water on which we live, work and depend. We pay respect to Aboriginal Elders past and present.
We pay respect to how Country has been protected and cared for by Aboriginal people over many tens of thousands of years.
We recognise the unique spiritual and cultural significance of land, water and all that is in the environment and the continuing connection and aspirations for Country of Aboriginal people and Traditional custodians.
Bio2Lab – Steve Marshall and Dr David Sharley
EPA – Simon Sharp and Dr Lenka O’Connor Sraj
We are grateful for the support, assistance and advice provided from EPA by Paul Leahy, Minna Saaristo, Leon Metzeling, Alice Phung, Caroline Martino, Simone Warner, Andrea Hinwood, Sam LeRay, Jess Keyt, Scott Daniels, Tim Turnbull, Jeremy Settle and John Rees.
You can contact Bio2Lab at firstname.lastname@example.org or by calling 0407011074 to discuss your next water quality monitoring project.
How to cite: Marshall, S., Sharley, D. J., Sharp, S., Sraj, L.O., Kolev, S., & Almeida, M.I.G.S (2021) Passive samplers and networked sensors in urban stormwater. A report to EPA Victoria, Report No. 083-01, Bio2Lab Pty Ltd.