Urbanisation leads to increases in stormwater runoff, resulting in elevated contaminant (e.g. metal, sediment, and nutrient pollutant) loads, decreased local infiltration and greater peak flow intensities. Heavy metal contaminants of concern, primarily copper (Cu), lead (Pb), and zinc (Zn), originate from a variety of sources including wear-and-tear of vehicle parts, corrosion of alloy roofs, legacy petroleum contamination, and multifarious construction practices. Different technologies have been used to mitigate stormwater runoff, ranging from traditional drainage networks fitted with concrete proprietary devices (e.g. vortex sediment separators and filters) to more environmentally integrated sustainable solutions. Rain gardens, a type of Sustainable Urban Drainage System (SUDS) or Water Sensitive Urban Design (WSUD), are employed to control stormwater peak flows and runoff volumes and simultaneously reduce contaminant loads to neighbouring waterways through biologically-active landscaped design. Despite increases in use of rain gardens as a best management practice (BMP) to treat urban stormwater runoff, there is a dearth of knowledge about their treatment and infiltration performance worldwide. It is believed that incorporating topsoil into rain garden design is likely to improve contaminant removal efficiencies (Davis et al. 2001; ARC 2003; Fletcher et al. 2004; Carpenter and Hallam 2010), but design recommendations are not informed by performance data which is limiting. Performance data is necessary for understanding the long-term responses of bioinfiltrative treatment systems and for modelling efforts aiming to predict their mitigation behaviour (Fletcher et al. 2004). In order to evaluate the influence of substrate composition on stormwater treatment and hydraulic effectiveness in rain gardens, mesocosm-scale (180 L, 0.17 m2) laboratory systems were established. Saturated (constant-head) hydraulic conductivity was determined before and after contaminant (Cu, Zn, Pb and nutrients) removal experiments on three rain garden systems comprising various proportions of organic topsoil. Raw stormwater runoff from a neighbouring Christchurch city catchment was collected, characterised, and applied in the removal efficiency experiments. The system with only topsoil had the lowest saturated hydraulic conductivity (160 mm/hr initial to 164 mm/hr final) and poorest metal (Cu, Zn) removal efficiency (Cu 0.3%, Zn 60.5% and Pb 89.5%) at a ‘standard’ contaminant loading rate (Cu = 5.99 ± 0.73 µg/min, Zn = 57.89 ± 6.06 µg/min, Pb = 13.65 ± 2.80 µg/min). The sand-only system demonstrated good metal removal (Cu 56.4%, Zn 73.5%, and Pb 81.6%) with hydraulic conductivity (up to 805 mm/hr) adequate for practical implementation (i.e. greater than the 13 mm/hr minimum requirement (ARC 2003; MDE 2009; SFPUC 2009)). Overall, total metal amounts in the effluent were <50% of influent loads for all experiments, with the exception of Cu in the topsoil-only system, whose removal was negligible (0.3%). Greater metal removal was observed when effluent pH was elevated (up to pH 7.38). The pH increase (from an initial pH of 6.23 in raw stormwater) was provided by the calcareous sand, whereas the topsoil-only system lacked an alkaline source. Consequently, organic topsoil had poorer contaminant removal due to higher dissolved metal fractions, which are more difficult to immobilise at the lower pH. The relationship between pH and dissolved fraction was highly significant (Pearson’s Correlation, p < 0.0001, df = 74) for Cu, Zn, and Pb. Mesocosm-scale systems were then re-established with a calcareous substrate supplement to quantify the effects of pH augmentation on contaminant removal and hydraulic efficiencies. Mussel shells, a waste product from the shell-fish industry, were employed in two different volumetric proportions. Metal removal efficiency was increased in systems with mussel shells (Cu up to 46.6%, Zn up to 80.2%, Pb up to 88.7%) compared to the topsoil-only system (Cu 27.5%, Zn 55.5%, Pb 81.0%). Larger increases in removal efficiency were seen for Cu and Zn because increases in pH from mussel shell enhanced particulate fractions, which are easier to remove in filtration systems, while Pb is mainly in the particulate form at influent pH (Morrison et al. 1990). Effluent from systems with mussel shells also had higher hardness (hardness up to 101.7 mg/L as CaCO3) compared with 22.4 mg/L as CaCO3 in topsoil-only effluent. Hardness reduces metal ecotoxicity (Hyne et al. 2005). Results of these experiments show that mussel shells are a promising rain garden substrate capable of increasing metal removal efficiency and also decreasing metal ecotoxicity in effluent of bioinfiltration systems. Concurrently, an operational field-scale “rain garden” (42 m3; 60 m2) in Christchurch was monitored for hydraulic throughput and contaminant removal. The field system performed extremely well at mitigating peak flows, detaining water throughout storm events and removing total suspended solids (TSS) (90.6% average removal). However, the system failed to reduce effluent median total metal concentrations (Cu = 15.9 µg/L, Zn = 139.6 µg/L, Pb = 11.7 µg/L) below relevant ANZECC guidelines (Cu = 1.8 µg/L, Zn = 15.0 µg/L, Pb = 5.6 µg/L) highlighting the opportunity to optimise these field designs to improve metal removal.

This thesis project monitors the quantity and quality of stormwater entering and leaving a bioretention system in Coralville, Iowa. Bioretention is among many engineered solutions designed to provide treatment for runoff that might otherwise be drained directly to a body of water. Increased quantities of stormwater can impact stream morphology, degrade aesthetics, increase flood frequency, peak flow, and peak duration; as well as increased sedimentation and sediment transport. Decreases in water quality can impair fish or other aquatic populations, and increase the treatment requirements for downstream intakes. The number of communities, presently 47, affected by stormwater control ordinances increases as the Environmental Protection Agency continues to require smaller Municipal Storm Sewer Systems to adhere to National Pollutant Discharge Elimination System permits. The City of Coralville is setting an example by using infiltration practices to treat runoff from a 4-lane divided thoroughfare. Preliminary monitoring shows that the system in Coralville provides an average reduction in effluent temperature g of 3.7°C, an average reduction in peak flow of 2 cfs, and an average peak delay of 45 minutes. The project provides infiltrative treatment for the water quality volume and the empirical curve number for the project is 77.4. The urban runoff to the project is within literature values and the pollutant concentrations in the project effluent are below legal limits.

Stormwater runoff from highways is a source of pollution to surface water bodies and groundwater. This project develops a bio-detention treatment and harvesting system that is incorporated into roadside swales. The bio-detention system uses Bold & Gold™, a type of biosorption activated media (BAM), to remove nutrients from simulated highway runoff and then store the water in underground vaults for infiltration, controlled discharge, and/or irrigation and other non-potable applications. In order to design a bio-detention system, media characteristics and media/water quality relationships are required. Media characteristics determined through testing include: specific gravity, permeability, infiltration, maximum dry density, moisture content of maximum dry density, and particle-size distribution. One of the goals of this experiment is to compare the nitrogen and phosphorous species concentrations in the effluent of BAM to sandy soil for simulated highway runoff. Field scale experiments are done on an elevated test bed that simulates a typical roadway with a swale. The swale portion of the test bed is split into halves using BAM and sandy soil. The simulated stormwater flows over a concrete section, which simulates a roadway, and then over either sod covered sandy soil or BAM. One, one and a half, and three inch storms are each simulated three times with a duration of 30 minutes each. During the simulated storm event, initial samples of the runoff (influent) are taken. The test bed is allowed to drain for two hours after the rainfall event and then samples of each of the net effluents are taken. In addition to the field scale water quality testing, column tests are also preformed on the sandy soil and Bold & Gold™ without sod present. Sod farms typically use fertilizer to increase production, thus it is reasonable to assume that the sod will leach nutrients into the soils on the iv test bed, especially during the initial test runs. The purpose of the column tests is to obtain a general idea of what percentage removals of total phosphorus and total nitrogen are obtained by the sandy soil and Bold & Gold™. It is shown that the Bold & Gold™ media effluent has significantly lower concentrations of total nitrogen and total phosphorus compared to the effluent of the sandy soil based on an 80% confidence level. The Bold & Gold™ has a 41% lower average effluent concentration of total nitrogen than the sandy soil. The Bold & Gold™ media has a 78% lower average effluent concentration of total phosphorus than the sandy soil. Using both the column test data in combination with the field scale data, it is determined that the Bold & Gold™ BAM system has a total phosphorus removal efficiency of 71%. The removal efficiency is increased when stormwater harvesting is considered. A total phosphorus reduction of 94% is achieved in the bio-detention & harvesting swale system sample design problem