This project was designed to create and evaluate a novel bioswale design, utilizing a treatment train of sand, plants, and sorbents for the removal of polycyclic aromatic hydrocarbons (PAHs) and their toxic degradation products, polychlorinated biphenyls (PCBs), per- and polyfluoroalkyl substances (PFAS), and heavy metals (Cu2+ and Zn2+) from impacted stormwater at relevant Department of Defense sites. The overall goals of this project were to: (1) develop innovative stormwater control and treatment technologies that improve stormwater management, prevent sediment recontamination, and add to the existing water supply, and to (2) develop watershed modeling of new stormwater control processes that focus on sediment-related contaminants to provide information on the efficiency needed and the number of systems deployed to prevent sediment recontamination and increase stormwater harvesting.

Technical Approach

The research was structured around four tasks. Task 1 utilized laboratory-scale batch isotherm and column experiments to characterize the removal of PAHs, PCBs, PFASs, Cu2+, and Zn2+ by commercially available sorbents. Additionally, the column studies characterized the potential desorption of chemicals during periods of interrupted hydraulic loading. The sorbents tested included bioswale soil (25% mint compost, 25% municipal compost, and 50% native clay soil), granulated activated carbon, Biochar Basic™ (sourced from Douglas fir), EarthLite™ stormwater filter media (composite of hazelnut shell biochar, expanded shale, compost, and oyster shells), and RemBind® media (mixture of activated carbon, aluminum hydroxide, and kaolin clay).

In Task 2, treatment of real stormwater by 10 native and regional plants was investigated in a series of greenhouse experiments. The primary goal of Task 2.1 was to screen the ability of ten different bioswale plants, five grasses, and five broad-leafed plants for their ability to remove heavy metals (Cu2+ and Zn2+), PAHs, PCBs, and five model PFAS (PFBA, PFOA, PFHxS, PFOS, and FtSaB) from stormwater collected from the bioswale research facility. Plant-free mesocosms (soil only) were tested as negative controls. The plants represented a variety of root structures, leaf structures, growth rates, and nutrient requirements and are currently in use at our pilot-scale research bioswale facility. All experiments were conducted in laboratory-scale mesocosms under greenhouse conditions (70 °F set point, 60-85 °F range, natural lighting). All plants and plant-free controls were tested in sextuplet 2.4 L mesocosms. Each mesocosm contained 2.2 kg of modified bioswale soil (50% sand (ASTM C-33 spec), 25% native clay soil, 12.5% mint compost, and 12.5% municipal compost). All mesocosms were irrigated with collected stormwater until fully saturated. The mesocosms remained saturated for 5 h before being drained. Irrigation occurred once per week for 10 weeks. Influent and effluent contaminant concentrations were quantified for each irrigation event.

In Task 3, a set of field investigations were performed. A multi-stage treatment train intended to be used in a Low Impact Development (LID) model for on-side stormwater intake and remediation system was installed. This system intended to characterize transport of sediment-associated PFAS in LID using sediment tracer studies, temporally characterize plant stress in bioretention cells during wet and dry periods in the field, assess hydrologic performance of treatment train over multiple storms, and assess performance of chemical of concern treatment within the treatment train over multiple storms. Installation further included tracer studies to characterize spatial transport of PFAS-associated sediments, and design-build-evaluation of novel treatment train for passive treatment of sediments, nutrients, heavy metals, and PAHs.

Task 4 focused on field data collection for evaluation of treatment train and calibration and validation of the Storm Water Management Model (SWMM) for simulation of treatment train hydrology and fate and transport processes. Data from multiple storm events and prior tasks were assessed and assembled to simulate and address fate and transport based on pump controlled inflows for 24 hour durations. Improvement in further simulation of interactions between plants, water, chemicals of concern, microbes and microecologies, and soil was investigated. Mechanistic equations in existing phytoremediation models (e.g., Phyto-DSS, BALANS, Dynamic factor method, and Hung and Mackay model) were evaluated for different chemicals of concern, sorbents, and plants, and for integration into proposed compartment model.


Of the five sorbents tested (bioswale soil, granulated activated carbon, Biochar Basic™, EarthLite™ and RemBind®), Biochar Basic™ demonstrated superior removal of heavy metals and RemBind® demonstrated superior removal of PFAS under clean synthetic stormwater conditions. Under real stormwater conditions, all supports demonstrated significant decreases in removal capabilities and suggested that a pretreatment stage may be necessary. Additional experiments were conducted to determine which water quality constituents were key to remove during pretreatment.

Ten plant species located in the research bioswale (both grasses and broad leafed) were evaluated for their ability to remove heavy metals. In general, broad-leafed plants showed superior Zn2+ removal (especially checker mallow, yarrow, and salal) while grasses showed superior Cu2+ removal (especially slough sedge and dense sedge). A pilot-scale implementation utilizing sorbent column studies generated new protocols to characterize the capture and movement of sediments through the bioswale using silt sized (~20 mm diameter) fluorescent-magnetic tracers and fluorescent probes.

This work is some of the first to consider how biofilm formation and desorption affects the removal of heavy metals, nutrients, and DOC from real stormwater. Results from this work demonstrate that the environmental biofilms can form on sand, GAC, and biochar with the greatest biofilm density forming on GAC and the lowest biofilm density forming on sand. The formation of these environmental biofilms increased the sorptive capacity of sand, GAC, and biochar to remove copper and zinc from stormwater.

Using a tracer study, sediment-borne contaminant load on a per-size-fraction basis was quantified. The results showed contaminant load increased with decreasing particle size, as would be expected due to the larger surface area associated with smaller particles, with one exception of Fe which had greatest contaminant load for sizes 25-75 μm.

A new model for fine sediment transport in LID-like conditions was developed. The model also allows for representation of the “clogging” effect that occurs when high concentrations of fine sediment are present in the influent stormwater. Overall, the treatment train system reduced turbidity from an average of 94 NTU at the influent to an average of 14 NTU at the effluent in the six monitored storms. A tendency of a slight increase in turbidity was, however, observed after Rembind treatment train for all six storms, which could be attributed to leaching fines. Across 6 storms, residence time of LID was about 90 minutes when the number of antecedent dry days were ≥ 1. Biochar and Rembind Cells’ residence time varied from 10 - 37 minutes. The overall treatment system yielded significant reductions in TSS event mean concentrations, ranging from 84% to 97% across the six monitored storms.

The treatment train reduced total copper and total zinc by 17% - 84% and 54% - 90%, respectively. For three monitored storms, the treatment train was also consistent in removing dissolved copper by modest amount (5% - 30%) and dissolved zinc by 47% - 75%.


Together, the results of this research indicate that Green Infrastructure can be developed and used to remove a wide array of potential pollutants at military sites, in order to minimize the recontamination of remediated sediment. (Project Completion - 2023)


Prettyman, K., M. Babbar-Sebens, C.E. Parrish, and J.M. Babbar-Sebens. 2021. A Feasibility Study of Uninhabited Aircraft Systems for Rapid and Cost-Effective Plant Stress Monitoring at Green Stormwater Infrastructure Facilities. Journal of Hydroinformatics, 23(3):417-437doi.org/10.2166/hydro.2020.195.

Wells, J. and G. Wilson. 2022. Sediment Deposition Characteristics in Shallow Saturation
Excess Overland Flow for Urban Stormwater Applications. Hydrological Processes, 36(8):e14647. doi.org/10.1002/hyp.14647.