Bioretention systems, which mimic natural hydrology and reduce volume of stormwater runoff, are a preferred solution for meeting water balance objectives, but lack of knowledge about the long-term performance of these systems hinders their wider adoption. This study was a field survey of mature (>3 years and up to 10 years post-construction) bioretention cells across Ontario, Canada. The survey involved visual inspections, determination of soil physical parameters and soil-water interaction parameters, infiltration capacity testing and synthetic drawdown testing. Results indicate that infiltration capacity remains above the recommended minimum of 25 mm/hr, likely due to high content soils and development of soil structure due to biological factors over time. The drawdown times for three sites ranged from 5 minutes to 6 hours, much less than the maximum allowed drawdown time of 24–48 hours. Ksat (saturated hydraulic conductivity) was only moderately negatively correlated with age, and where data existed on KSat at the beginning of operation, KSat improved for six out of nine sites. Soil-water interaction properties more closely resembled loam soils than sandy soils, which may be due to the development of a soil structure over time. We recommend conducting visual inspections regularly over infiltration capacity testing for quick determination of maintenance needs.

  • Hydrology of bioretention continues to perform well over long-term (up to 10 years).

  • Media infiltration rates remain high and increased for 6/9 sites.

  • Drawdown times much faster than design.

  • Soil structure developed over time likely contributes to more water holding capacity available for plant uptake.

Streams in urban areas have become degraded due to erosion and increased pollutant loads, a phenomenon known as ‘urban stream syndrome’ (Walsh et al. 2005). Over the past twenty years, watershed managers and practitioners have been moving to more nature-based and distributed systems for managing regularly recurring rainfall events, by capturing them and returning them to infiltration and evapotranspiration (Bradford & Gharabaghi 2004). In North America, this trend comes under the term ‘Low Impact Development’ (LID) (Fletcher et al. 2015), of which bioretention is one technology.

Bioretention is a small depressed area consisting of engineered soil media and vegetation which allows stormwater runoff to percolate slowly through (Liu et al. 2014). Bioretention has been shown to reduce significant amounts of surface water runoff (Hatt et al. 2009; Davis et al. 2012; Khan et al. 2012; Winston et al. 2016; Johnson & Hunt 2019), which improves water quality largely by reducing the load of contaminants into surface water bodies. Bioretention has many co-benefits beyond stormwater management such as carbon sequestration in plants and trees (Kavehei et al. 2018), evaporative and shading cooling to reduce the urban heat island effect (Jamei & Tapper 2019), and improved quality of life and land values (Foster et al. 2011).

In Ontario, bioretention can be used to meet water balance criteria (Aquafor Beech Ltd 2017). Such guidelines are expected to become part of provincial requirements (Environmental Registry of Ontario 2017) and are increasingly required by local regulators (e.g. Lake Simcoe Region Conservation Authority (2016), City of Toronto (2019)). Despite over two decades of practice, widespread adoption of bioretention is not occurring, with installations mostly consisting of one-off demonstration projects. A barrier to wider adoption is the uncertainty around LID cost, long-term maintenance and performance (Roy et al. 2008). For example, other stormwater infrastructure, such as sewer pipes, stormwater management ponds and catchbasins, have been used in our urban environments for more than four decades, so maintenance and replacement costs are well known. Ponds, pipes and catchbasins are maintained by the municipal owners, with annual budget allocated to inspection and maintenance of the infrastructure. Though guidelines exist for maintenance of LID practices (Gulliver et al. 2010; Credit Valley Conservation 2015), there are no specific timelines listed for replacement of media, and media replacement is recommended only when media is ponded/saturated for more than 48 hours following a rain event.

To address the gap in knowledge on long term maintenance of bioretention, research has been conducted on simulated systems and in the field (where possible). Some laboratory studies show that the media becomes clogged as the bioretention systems ages (Khan et al. 2012; Le Coustumer et al. 2012). But laboratory studies, by nature of their design, often omit vital field parameters that can prevent clogging, such as the influence of vegetation and roots, soil microbiota, freeze/thaw cycles, or the influence of wetting and drying cycles on soil formation. It is thus not entirely surprising that field surveys of bioretention completed in the United States (Minnesota and North Carolina) (Asleson et al. 2009; Wardynski & Hunt 2012) and Germany (Kluge et al. 2018) have instead reported that the infiltration capacity of bioretention was maintained or improved compared to post-construction. Kluge et al. (2018) found that there was no correlation between age and infiltration rate, instead the main correlation was between soil type and infiltration rate. The field surveys revealed other data about long-term performance, such as in Wardynski & Hunt (2012), where the authors found that the bioretention cells were not constructed as designed, and so the retention volume was much lower than expected. Asleson et al. (2009) found that the majority of surveyed bioretention systems were functioning well by measuring the infiltration capacity and the drawdown time.

The previous surveys of mature bioretention systems represent bioretention systems in temperate climates (Wardynski & Hunt 2012; Kluge et al. 2018) and one in continental climate (Asleson et al. 2009). Bioretention systems originated and have been largely studied in temperate climate zones (Spraakman et al. 2020a), and additional research is needed on their performance in the continental climate zone, which is marked by colder winters and hot-humid summers. This study also reflects guidelines and practices in place in Ontario, Canada, which has different media specifications and planting regimens than in Minnesota, North Carolina or Germany. In addition, design and construction expertise remains quite localized, and the choices of a particular designer may impact overall functioning.

The purpose of this study is to evaluate the mature hydrologic performance of bioretention cells in the continental climate of Southern Ontario, Canada. In assessing several soil physical properties (water retention at different tensions, plant available water, porosity, particle size distribution, saturated hydraulic conductivity), we sought to determine if age influenced the hydrologic capacity and soil structure of the bioretention media. Increasing the amount of data practitioners have on mature bioretention system will hopefully lead to increased confidence in our estimates for lifetime expectancy and better planning for determining maintenance and rehabilitation practices.

This study was a survey of bioretention systems in Southern Ontario, Canada. Monitoring activities included visual assessments, soil sampling of bioretention media, infiltration capacity testing and synthetic drawdown testing. This methodology was adapted from Asleson et al. (2009) and Gulliver et al. (2010). All statistical data analysis was performed in R (R Core Team 2017). Summary statistics, normality test (Shapiro-Wilks) and Spearman correlation tests were taken using the base R package. Mapping figure was prepared using QGIS (version 3.12) and openly available map data (DMTI Spatial Inc. 2014; OpenStreetMap contributors 2021). Figures were prepared using the ggplot2 package (Wickham 2016).

Site selection

A mature site was defined as being three years or more post-construction, as previous findings have indicated that bioretention media needs more than two years to settle (Spraakman et al. 2020b), and that the majority of bioretention research is based on systems that are less than 3 years post-construction (Spraakman et al. 2020a). Through research project partners (municipalities, conservation authorities and one consulting firm), 29 sites were identified that met the criteria of three years or more post-construction and with contributing drainage areas that were undisturbed (i.e. no active construction) (see Figure 1). Initial site visits were performed in April-May 2019 to determine if the sites were suitable for inclusion in this study. After initial visits, seven sites were excluded from this study for the following reasons: the sites did not receive stormwater runoff due to the inlet being above the incoming grade (3 sites), the sites were altered and were no longer functioning as bioretention systems (2 sites), permission to study the site was not granted (1 site), or the site received roof runoff only, which was not comparable to other stormwater-receiving sites (1 site). As many of the sites in this study were early installations and demonstration projects, there was inexperience amongst designers and contractors which led to flaws such as the inlet being above incoming grade.

Figure 1

Map of 30 mature bioretention sites identified for this study.

Figure 1

Map of 30 mature bioretention sites identified for this study.

The characteristics of the selected sites are shown in Table 1. Infiltration capacity testing was performed at 20 sites. Infiltration testing was not completed at two sites because the soil media was saturated (sites RC and OP). Visual inspections and infiltration capacity tests were conducted between May and August 2019. Synthetic drawdown tests were performed at three sites during October 2018 and October and November 2019. Water truck access, necessary for the synthetic drawdown tests, was not possible for most sites, either due to accessibility of inlet or permission from site owners.

Table 1

Site characteristics, provided by partners

Site IDAge (years)Number of cellsLand use descriptionDrainage area (ha)Area of cell(s) (m2)Drainage area ratioaMedia typebMedia depth (m)Vegetation
WW Residential 2.3 4,400 5.2 Filter media 0.53 Inside swale: sod
Along slopes: trees and shrubs 
ED Road runoff, permeable pavement 0.65 145 45 Filter media 0.45 1–2 deciduous trees or shrubs per cell, plus a mix of perennial flowers 
RC Parking lot 0.5 100 50 Native soil (not described) 1.05 Wetland plants (cattails), shrubs and trees 
IM Parking lot, combination of asphalt and permeable pavement 0.61 174 35 Filter media 0.5 Ornamental flowers, ornamental shrubs, native trees 
KV Parking lot, rubber pavers 0.027 33 8.0 62% sand
38% fines 
0.4 Flowering native perennials, shrubsc 
ER Parking lot, permeable interlocking concrete pavement 0.23 251 9.1 Filter media 0.6 Flowering native perennials, shrubsc 
I1 Parking lot, asphalt 0.043 26 16 Filter media+Imbrium Sorbtive Media (5% by volume) (commercial product) 0.5 Flowering native perennials, shrubsc 
I2 Parking lot, asphalt 0.043 26 16 Filter media 0.5 Flowering native perennials, shrubsc 
I3 Parking lot, asphalt 0.043 26 16 Red sand 0.5 Flowering native perennials, shrubsc 
KKd 18 Municipal road, downtown core 0.1 123 79 69% sand
31% fines 
0.7 Ornamental flowers contained within pots within the sandy media 
A1 Municipal road 0.22 100 22 Filter media+Iron filings (5% by weight) 0.85 Perennial flowers and grassesc 
A2 Parking lot, asphalt 0.31 100 31 Filter media 0.85 Perennial flowers and grassesc 
A3 Municipal road 0.08 50 16 Filter media+Imbrium Sorbtive Media (5% by volume) (commercial product) 0.5 Perennial flowers and grassesc 
UMR Regional road 4.02 1,600 25 Filter media 0.8 Trees, shrubs, grasses 
S1 Municipal road 0.086 73 12 Filter media 0.5 Perennial flowers and grassesc 
S2 Municipal road 0.074 85 8.7 Filter media 0.5 Perennial flowers and grassesc 
S3 Municipal road 0.304 86 35 Filter media 0.5 Perennial flowers and grassesc 
BFC1 Regional airport 0.13 59 22 Filter media 0.5 Sod 
BFC2 Regional airport 0.29 98 30 Filter media 0.5 Sod 
OP Parking lot, permeable pavement 0.19 111 17 Filter media 0.63 Perennial flowers and grasses 
UC Parking lot, asphalt 0.28 228 12 Filter media 0.88 Grasses, shrubs, trees 
PC 10 Parking lot, asphalt 0.77 235 33 Filter media 0.35 Grasses, shrubs, trees 
Site IDAge (years)Number of cellsLand use descriptionDrainage area (ha)Area of cell(s) (m2)Drainage area ratioaMedia typebMedia depth (m)Vegetation
WW Residential 2.3 4,400 5.2 Filter media 0.53 Inside swale: sod
Along slopes: trees and shrubs 
ED Road runoff, permeable pavement 0.65 145 45 Filter media 0.45 1–2 deciduous trees or shrubs per cell, plus a mix of perennial flowers 
RC Parking lot 0.5 100 50 Native soil (not described) 1.05 Wetland plants (cattails), shrubs and trees 
IM Parking lot, combination of asphalt and permeable pavement 0.61 174 35 Filter media 0.5 Ornamental flowers, ornamental shrubs, native trees 
KV Parking lot, rubber pavers 0.027 33 8.0 62% sand
38% fines 
0.4 Flowering native perennials, shrubsc 
ER Parking lot, permeable interlocking concrete pavement 0.23 251 9.1 Filter media 0.6 Flowering native perennials, shrubsc 
I1 Parking lot, asphalt 0.043 26 16 Filter media+Imbrium Sorbtive Media (5% by volume) (commercial product) 0.5 Flowering native perennials, shrubsc 
I2 Parking lot, asphalt 0.043 26 16 Filter media 0.5 Flowering native perennials, shrubsc 
I3 Parking lot, asphalt 0.043 26 16 Red sand 0.5 Flowering native perennials, shrubsc 
KKd 18 Municipal road, downtown core 0.1 123 79 69% sand
31% fines 
0.7 Ornamental flowers contained within pots within the sandy media 
A1 Municipal road 0.22 100 22 Filter media+Iron filings (5% by weight) 0.85 Perennial flowers and grassesc 
A2 Parking lot, asphalt 0.31 100 31 Filter media 0.85 Perennial flowers and grassesc 
A3 Municipal road 0.08 50 16 Filter media+Imbrium Sorbtive Media (5% by volume) (commercial product) 0.5 Perennial flowers and grassesc 
UMR Regional road 4.02 1,600 25 Filter media 0.8 Trees, shrubs, grasses 
S1 Municipal road 0.086 73 12 Filter media 0.5 Perennial flowers and grassesc 
S2 Municipal road 0.074 85 8.7 Filter media 0.5 Perennial flowers and grassesc 
S3 Municipal road 0.304 86 35 Filter media 0.5 Perennial flowers and grassesc 
BFC1 Regional airport 0.13 59 22 Filter media 0.5 Sod 
BFC2 Regional airport 0.29 98 30 Filter media 0.5 Sod 
OP Parking lot, permeable pavement 0.19 111 17 Filter media 0.63 Perennial flowers and grasses 
UC Parking lot, asphalt 0.28 228 12 Filter media 0.88 Grasses, shrubs, trees 
PC 10 Parking lot, asphalt 0.77 235 33 Filter media 0.35 Grasses, shrubs, trees 

Notes:

aDrainage area ratio is the ratio of the drainage area to the bioretention cell area.

bFilter media represents the standard mix specified in Credit Valley Conservation and Toronto and Region Conservation Authority (2011): 85–88% sand, 8–12% fines, organic matter 3–5%.

cFull planting list is available in Supplementary Materials.

dDrainage area and cell area for KK cells is representative of only one cell, not all 18 cells.

Partners on this project supplied design drawings, as-constructed drawings, planting plans, and monitoring assessments. Publicly available reports and monitoring documents were available for sites implemented by Credit Valley Conservation Authority (Credit Valley Conservation 2020) and the Toronto and Region Conservation Authority (Toronto and Region Conservation Authority 2020b). Several reports and drawings prepared by the consulting firm partner (Aquafor Beech Ltd) were not publicly available.

Site visits and visual inspections

Visual inspections followed the methodology in Gulliver et al. (2010). A copy of the field form used for the visual inspection is available in Supplementary Materials. The site visits included checking the inlet and outlet for signs of damage or debris/sediment buildup and collecting soil samples for later analysis. Soil samples were taken in the top 10–30 cm of media, after removing mulch and other deposited organic matter. Visual inspections were performed at 22 sites. Soil testing was performed at 21 sites. Soil testing was not completed at site RC due to a large amount of organic materials (plant detritus and plant roots) prevented access to soil media.

Soil processing and physical properties

Soil samples were collected at the site and analyzed at University of Toronto laboratories. Particle size analysis using dry sieving and hydrometer method was conducted following ASTM D422-63 (ASTM 2007). Upon returning to the lab, the soil was air dried and crushed before sieving. Sieve sizes 19, 4.75, 2, 1.18, 0.6, 0.3, 0.15 and 0.075 mm were used. The mass retained on each sieve was measured, then the percent passing was calculated using the total mass, resulting in a particle size distribution curve. The portion of sand, silt and clay were calculated as the percent of particles above diameter 0.05 mm, between 0.002- and 0.05-mm diameter, and less than 0.002-mm diameter, respectively (ASTM 2017). The particle size diameter for 60, 30 and 10% passing (D60, D30 and D10, respectively) are used in calculating the coefficient of uniformity (Cu) and coefficient of conformity (Cc) and are obtained from the particle size distribution (ASTM 2017). Particle size density (ρs) was determined using a stereopycnometer device (Quantachrome) using compressed hydrogen. A single 90.59 cm3 sample at 10–30 cm depth was taken in a steel ring using a bulk density sampler (AMS Inc. 2012), which was used for determination of bulk density (ρb) and organic matter content (%OM) (ASTM 2014).

The following formulas were used for determining soil physical properties (AMS Inc. 2012; ASTM 2014, 2017):
formula
(1)
formula
(2)
formula
(3)
formula
(4)
formula
(5)
where md is the air-dried mass in grams, and mf is the mass after burning in a furnace at 450 °C for 8 hours following ASTM D2974-14.

Soil-water interaction properties

A point soil sample was taken at 10–30 cm depth for the HYPROP device (Meter Group Inc.) using a 250 cm3 sample ring. The HYPROP uses the soil evaporation method to determine volumetric water content (ϴ) at various soil tensions (Ψm) (UMS), and the measurements are then fitted to a curve using the HYPROP-FIT Software (Pertassek et al. 2015). Hydraulic conductivity at various water contents and soil tensions was measured by the change in mass as the soil sample dries due to evaporation (UMS). The fitted equations used by the various soil samples included the van Genuchten unimodal model (van Genuchten 1980), the bimodal variant of the van Genuchten model (Durner 1994), or the Peters-Durner-Iden (PDI) variant of the van Genuchten model (Peters 2013). The model with the lowest root mean square error on the volumetric water content variable was selected for each sample. Both the van Genuchten model and the PDI variant of the van Genuchten model describe unimodal pore distribution and generally homogeneous soils, while the bimodal variant describes a heterogeneous pore structure with two dominant pore sizes. In bimodal pore structures, larger pore sizes are the interparticle voids, and are responsible for water transmission and aeration, and the smaller pore sizes are intraparticle voids, and are responsible for water retention for plants (Durner 1994).

Fitting parameters used in the above-mentioned van Genuchten equations (α, n and w) are given the subscript ‘1’ for interparticle voids and subscript ‘2’ for intraparticle voids. The fitting parameter α (units cm−1) is a scaling factor representing pore size maximum, and n and w are dimensionless curve-shape parameters where n>1 and 0<w<1. The portion of pores in the interparticle (w1) and intraparticle (w2) range provide information on the dominant water flow path. The fitted water retention curve also provides information on volumetric water content (ϴ) at hydraulically-important matric potentials (Ψm), such as field capacity (Ψm=−316 Pa or pF 2.5) and wilting point (Ψm=−15,850 Pa or pF 4.2) (Lambers et al. 2008). The plant available water (PAW) is the difference between water contents at field capacity (water content after the soil is saturated and before it freely drains) and permanent wilting point (the minimum water content a plant is able to access water from soil) (Lambers et al. 2008).

The pore size radius (r) was calculated using the matric potential (Ψm, or water head), which was then plotted on the x-axis using a logarithmic scale, then the density distribution of pore size radius was calculated using the derivative of the water retention curve function, then plotted on the y-axis. This method was adopted from Basile et al. (2006) and Liu & Fassman-Beck (2018). The equations used for pore size distribution were:
formula
(6)
where r is the pore size radius (m), σ is the surface tension of water in N/m and Ψm is the matric potential in Pa, and
formula
(7)
where θ is the volumetric water content.

Infiltration capacity testing

The saturated hydraulic conductivity (KSat) was measured using the auger hole method and the Guelph permeameter device (Soil Moisture Equipment Corporation, Model 2800K1). At each site, 5–17 measurements were taken, depending on the size of the bioretention cell. The mean number of measurements across all sites was 12. The smallest sites were KK, I1, I2 and I3, so 6–9 measurements were taken at each. The largest site was UMR, with 15 measurements taken at this site. In Asleson et al. (2009), 4–19 measurements would be required to obtain a KSat value within 10% of the true mean. Using the Guelph Permeameter involves measuring the water level decrease in a consistent time period (30 seconds or 1 minute) at constant heads of 5 and 10 cm. The formulas for converting between the rate of fall (cm/min) and KSat are described in detail in Elrick & Reynolds (1992). Guelph Permeameter measurements were taken in the centerline of flow, and not along the edges or sides of the bioretention systems, as Gnanaraj (2018) showed that this was where the lowest KSat values were measured. All KSat measurements were taken within a single day at a given site, and all sites were visited between May and July 2019, except for KV which had KSat testing performed in July 2017.

Synthetic drawdown testing

Three sites were selected for synthetic drawdown testing due to their ease of access for a water truck and permissions granted by partner organizations: KV, A3 and S2. The amount of water added to the bioretention cell was approximately equivalent to the 25-mm storm depth applied to the drainage area, following the methodology outlined in Gulliver et al. (2010). For KV, water was added from a cistern containing primarily roof runoff from a nearby building. For A3 and S2, water was sourced from the local municipality and added to the cell via a tanker truck. The water trucks available did not use gauge or flow meters, so the amount of water entering the bioretention cell was an estimate based on the capacity of the water truck. For example, 18.5 m3 was needed for the S2 bioretention cell, and the local water truck company had a truck of 4,000 gallons (15.14 m3), and so one truck load was unloaded onto the cell, the truck departed the site to refill, then returned and unloaded approximately one-half of the water truck onto the cell. Thus, S2 was loaded with approximately 20–23 m3 loaded over 1.5 hours.

Visual assessments and soil physical parameters

Two sites out of the 22 sites included in the visual inspection contained saturated soils, which was likely due to a design flaw in these early installations. This was indicated by the presence of ponded water during dry conditions (more than 24 hours after the last rainfall event) and by the presence of wetland vegetation (cattails, Typha angustifolia or Typha latifolia, see Figure 2). Half of the visited sites required minor maintenance, primarily clearing debris and/or sediment that had accumulated at or near the inlet. The main findings and recommendations of the visual assessments at each site are shown in Table 2.

Table 2

Visual inspection results

Site IDInlet conditionOutlet conditionStanding water (Y/N)Sediment accumulation (Y/N)Requires maintenance? (Y/N)Type of maintenance neededSoil classification
WW Clogged partially
-sediment accumulation in concrete inlet, not affecting flow through 
Clear
-large catchbasin receives flow from underdrain 
Minor: clear sediment from inlet Well-graded sand with silt and gravel 
ED Clogged partially
some inlets (2/6) were covered in debris and leaves but most were clear 
Clear, outlet is via an overflow, one per cell, which appeared clean at every cell Y, near inlet of each basin Minor: clear sediment from inlet Poorly graded sand with silt 
RC Clear Clear Major: investigate reasons for saturated soils. May need to be re-installed. Unable to collect soil sample 
IM Clear Clear Y, at inlet above rocks Minor: clear sediment from inlet Poorly graded sand with silt 
KV Clear Clear Continue routine maintenance. Poorly graded sand with silt 
ER Clear, sheet flow from parking lot N/A cannot see outlet, undergroumd Continue routine maintenance. Well-graded sand with silt 
I1 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand with silt 
I2 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand 
I3 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand 
KK Clogged partially
-leaves, cigarettes
some inlet grates are broken 
Clear Y, cigarettes, garbage Minor: clear sediment from inlet, repair inlet grate Well-graded sand with silt and gravel 
A1 Clogged partially, filled w/sediment but not impeding flow Clear
sludge at bottom of catchbasin 
Minor: clear sediment from inlet Poorly graded sand 
A2 Clogged partially
Lots of leaves blocking flow, rocks asphalt covered in leaves 
Clear higher than surroundings Y leaves Minor: clear sediment from inlet Poorly graded sand 
A3 Clogged partially, leaves sediment build up Clear
just freshly mulched 
Y leaves Minor: clear sediment from inlet Poorly graded sand 
UMR Clear, some grasses No underdrain Continue routine maintenance. Well-graded sand with silt 
S1 Clogged partially, w/leaves and debris Clear Y leaves
build up 
Minor: clear sediment from inlet Poorly graded sand 
S2 Clogged partially, leaves, sediment build up Clear Y, leaves Minor: clear sediment from inlet Well-graded sand with silt 
S3 Clogged partially, leaves and debris Clogged partially by mulch Minor: clear sediment from inlet Poorly graded sand 
BFC1 Clear, some partial covering of inlet with fresh-cut grass Clear Continue routine maintenance. Poorly graded sand with silt 
BFC2 Clear, some partial covering of inlet with fresh-cut grass Clear Continue routine maintenance. Poorly graded sand 
OP Clogged partially, good and alive grasses present Clear, grate at inlet, underdrain is actively flowing from standing water Y, around inlet Major: determine reason for saturated soils, clear sediment from inlet Poorly graded sand with silt 
UC Clear Clear Continue routine maintenance. Poorly graded sand 
PC Clear, curbs cuts from parking lot Unable to see, catchbasin at rain garden has restriction inside Continue routine maintenance. Poorly graded sand 
Site IDInlet conditionOutlet conditionStanding water (Y/N)Sediment accumulation (Y/N)Requires maintenance? (Y/N)Type of maintenance neededSoil classification
WW Clogged partially
-sediment accumulation in concrete inlet, not affecting flow through 
Clear
-large catchbasin receives flow from underdrain 
Minor: clear sediment from inlet Well-graded sand with silt and gravel 
ED Clogged partially
some inlets (2/6) were covered in debris and leaves but most were clear 
Clear, outlet is via an overflow, one per cell, which appeared clean at every cell Y, near inlet of each basin Minor: clear sediment from inlet Poorly graded sand with silt 
RC Clear Clear Major: investigate reasons for saturated soils. May need to be re-installed. Unable to collect soil sample 
IM Clear Clear Y, at inlet above rocks Minor: clear sediment from inlet Poorly graded sand with silt 
KV Clear Clear Continue routine maintenance. Poorly graded sand with silt 
ER Clear, sheet flow from parking lot N/A cannot see outlet, undergroumd Continue routine maintenance. Well-graded sand with silt 
I1 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand with silt 
I2 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand 
I3 Clear, sheet flow Clear Continue routine maintenance. Poorly graded sand 
KK Clogged partially
-leaves, cigarettes
some inlet grates are broken 
Clear Y, cigarettes, garbage Minor: clear sediment from inlet, repair inlet grate Well-graded sand with silt and gravel 
A1 Clogged partially, filled w/sediment but not impeding flow Clear
sludge at bottom of catchbasin 
Minor: clear sediment from inlet Poorly graded sand 
A2 Clogged partially
Lots of leaves blocking flow, rocks asphalt covered in leaves 
Clear higher than surroundings Y leaves Minor: clear sediment from inlet Poorly graded sand 
A3 Clogged partially, leaves sediment build up Clear
just freshly mulched 
Y leaves Minor: clear sediment from inlet Poorly graded sand 
UMR Clear, some grasses No underdrain Continue routine maintenance. Well-graded sand with silt 
S1 Clogged partially, w/leaves and debris Clear Y leaves
build up 
Minor: clear sediment from inlet Poorly graded sand 
S2 Clogged partially, leaves, sediment build up Clear Y, leaves Minor: clear sediment from inlet Well-graded sand with silt 
S3 Clogged partially, leaves and debris Clogged partially by mulch Minor: clear sediment from inlet Poorly graded sand 
BFC1 Clear, some partial covering of inlet with fresh-cut grass Clear Continue routine maintenance. Poorly graded sand with silt 
BFC2 Clear, some partial covering of inlet with fresh-cut grass Clear Continue routine maintenance. Poorly graded sand 
OP Clogged partially, good and alive grasses present Clear, grate at inlet, underdrain is actively flowing from standing water Y, around inlet Major: determine reason for saturated soils, clear sediment from inlet Poorly graded sand with silt 
UC Clear Clear Continue routine maintenance. Poorly graded sand 
PC Clear, curbs cuts from parking lot Unable to see, catchbasin at rain garden has restriction inside Continue routine maintenance. Poorly graded sand 
Figure 2

Sites with saturated soil conditions, as indicated by presence of wetland vegetation. (a) Site RC (9 years maturity). (b) Site OP (7 years maturity).

Figure 2

Sites with saturated soil conditions, as indicated by presence of wetland vegetation. (a) Site RC (9 years maturity). (b) Site OP (7 years maturity).

Soil classification showed overall sandy soils, as 20 sites had a soil texture of sand (sand content greater than 80%) and one site was loamy sand (sand content 77%). 24% of sites had well-graded sand with a coefficient of uniformity (Cu) ≥6.0 and 1.0≤coefficient of conformity (Cc) ≤3.0 (ASTM 2017), whereas 76% (16/21) of sites had poorly graded sands (Table 2). Poorly-graded sands would be expected in bioretention systems, as typically these soils are not compacted to allow for good drainage. The mean soil organic content of 7% was within an acceptable range of 3–8% (Credit Valley Conservation & Toronto and Region Conservation Authority 2011). The porosity, ranged from 0.37 to 0.88 with a mean of 0.52±0.12, was overall much higher than design guidance of 0.4, a typical porosity used for calculating actual storage volume. The summary statistics for soil physical parameters are shown in Table 3, and all soil parameters and particle size distributions for all sites are included in the Supplementary Materials.

Table 3

Soil sampling statistics, n=20

ParameterMinMedianMeanStandard deviationMax
% Sand 73 88 86 96 
% Silt 
% Clay 
D60 (mm) 0.17 0.52 0.55 0.17 0.9 
D30 (mm) 0.1 0.28 0.26 0.09 0.5 
D10 (mm) 0.01 0.095 0.11 0.05 0.2 
Cu 2.13 5.5 7.66 10.0 50 
Cc 0.43 1.15 1.73 2.50 12.5 
φ 0.37 0.51 0.52 0.12 0.88 
% OM 28 
ParameterMinMedianMeanStandard deviationMax
% Sand 73 88 86 96 
% Silt 
% Clay 
D60 (mm) 0.17 0.52 0.55 0.17 0.9 
D30 (mm) 0.1 0.28 0.26 0.09 0.5 
D10 (mm) 0.01 0.095 0.11 0.05 0.2 
Cu 2.13 5.5 7.66 10.0 50 
Cc 0.43 1.15 1.73 2.50 12.5 
φ 0.37 0.51 0.52 0.12 0.88 
% OM 28 

Soil-water interaction properties

The soil moisture release curve was measured for 19 soil samples. An example water retention curve, pore radius distribution and hydraulic conductivity curve are shown in Figure 3, with curves for all sites shown in Supplementary Materials. The saturated hydraulic conductivity estimated with the HYPROP is also shown in Supplementary Materials. Model-fitted soil-water parameters are shown in Table 4.

Table 4

Soil-water interaction properties, by site and with summary statistics for all sites

Site NameWRC Model Fit% (v/v)
Interparticle voids
Intraparticle voids
% (v/v)
ϴrϴsα1 (cm−1)n1w1r (μm)α2 (cm−1)n2w2r (μm)ϴ (pF=2.5)ϴ (pF=4.2)PAW
WW Bimodal van Genuchten 17 44 0.05 2.15 0.77 59 0.00 8.07 0.23 24.4 17.2 7.2 
ED Bimodal van Genuchten 12 45 0.05 2.14 0.84 60 0.50 1.01 0.17  17.9 10.4 7.5 
IM Bimodal van Genuchten 24 70 0.01 1.92 0.48 47 0.04 4.68 0.52 30.7 7.0 23.7 
KV Bimodal van Genuchten 22 56 0.36 1.62 0.31 453 0.00 1.44 0.69 40.3 26.0 14.3 
ER Bimodal van Genuchten 29 49 0.02 1.88 0.67 377 0.32 2.19 0.34 27 31.5 29.0 2.5 
I1 Bimodal van Genuchten 16 57 0.05 15.0 0.83 68 0.00 2.30 0.18 22.9 16.5 6.4 
I2 Unimodal van Genuchten 21 61 0.19 1.59 1.00 164     24.8 21.7 3.1 
I3 PDI-variant of bimodal van Genuchten 28 56 0.04 6.49 0.31 54 0.04 2.58 0.69  22.8 13.6 9.2 
KK Bimodal van Genuchten 22 70 0.01 2.29 0.24 216 0.30 1.57 0.76 29.9 22.2 7.6 
A1 Bimodal van Genuchten 17 64 0.11 2.29 0.40 143 0.00 2.23 0.61 37.7 17.0 20.7 
A2 Unimodal van Genuchten 28 59 0.04 2.31 1.00 59     28.7 17.3 11.5 
A3 Bimodal van Genuchten 28 61 0.02 1.24 0.69 94 0.07 3.94 0.31  25.7 10.3 15.4 
UMR Bimodal van Genuchten 13 64 0.04 4.01 0.18 59 0.50 1.18 0.82  29.9 21.2 8.6 
S1 Bimodal van Genuchten 19 66 0.04 3.65 0.60 58 0.01 2.05 0.40 28.0 19.2 8.8 
S2 Unimodal van Genuchten 15 60 0.04 2.29 1.00 41     18.0 10.3 7.7 
S3 Bimodal van Genuchten 57 0.06 2.43 0.37 62 0.03 1.15 0.63  28.6 17.9 10.7 
BFC1 PDI-variant of bimodal van Genuchten 17 49 0.02 2.82 0.55 43 0.03 11.9 0.45  13.9 8.1 5.8 
BFC2 PDI-variant of bimodal van Genuchten 20 51 0.03 1.93 0.45 50 0.04 4.63 0.55  18.0 9.9 8.1 
PC PDI-variant of bimodal van Genuchten 66 0.08 1.13 0.82 108 0.08 2.97 0.18  30.1 13.5 26.2 
Minimum 44    41    14 
Median 19 59    60    28 17 
Mean 19 58    117    27 16 11 
Standard Deviation    116    8.32 
Maximum 29 70    453    27 40 29 26 
Site NameWRC Model Fit% (v/v)
Interparticle voids
Intraparticle voids
% (v/v)
ϴrϴsα1 (cm−1)n1w1r (μm)α2 (cm−1)n2w2r (μm)ϴ (pF=2.5)ϴ (pF=4.2)PAW
WW Bimodal van Genuchten 17 44 0.05 2.15 0.77 59 0.00 8.07 0.23 24.4 17.2 7.2 
ED Bimodal van Genuchten 12 45 0.05 2.14 0.84 60 0.50 1.01 0.17  17.9 10.4 7.5 
IM Bimodal van Genuchten 24 70 0.01 1.92 0.48 47 0.04 4.68 0.52 30.7 7.0 23.7 
KV Bimodal van Genuchten 22 56 0.36 1.62 0.31 453 0.00 1.44 0.69 40.3 26.0 14.3 
ER Bimodal van Genuchten 29 49 0.02 1.88 0.67 377 0.32 2.19 0.34 27 31.5 29.0 2.5 
I1 Bimodal van Genuchten 16 57 0.05 15.0 0.83 68 0.00 2.30 0.18 22.9 16.5 6.4 
I2 Unimodal van Genuchten 21 61 0.19 1.59 1.00 164     24.8 21.7 3.1 
I3 PDI-variant of bimodal van Genuchten 28 56 0.04 6.49 0.31 54 0.04 2.58 0.69  22.8 13.6 9.2 
KK Bimodal van Genuchten 22 70 0.01 2.29 0.24 216 0.30 1.57 0.76 29.9 22.2 7.6 
A1 Bimodal van Genuchten 17 64 0.11 2.29 0.40 143 0.00 2.23 0.61 37.7 17.0 20.7 
A2 Unimodal van Genuchten 28 59 0.04 2.31 1.00 59     28.7 17.3 11.5 
A3 Bimodal van Genuchten 28 61 0.02 1.24 0.69 94 0.07 3.94 0.31  25.7 10.3 15.4 
UMR Bimodal van Genuchten 13 64 0.04 4.01 0.18 59 0.50 1.18 0.82  29.9 21.2 8.6 
S1 Bimodal van Genuchten 19 66 0.04 3.65 0.60 58 0.01 2.05 0.40 28.0 19.2 8.8 
S2 Unimodal van Genuchten 15 60 0.04 2.29 1.00 41     18.0 10.3 7.7 
S3 Bimodal van Genuchten 57 0.06 2.43 0.37 62 0.03 1.15 0.63  28.6 17.9 10.7 
BFC1 PDI-variant of bimodal van Genuchten 17 49 0.02 2.82 0.55 43 0.03 11.9 0.45  13.9 8.1 5.8 
BFC2 PDI-variant of bimodal van Genuchten 20 51 0.03 1.93 0.45 50 0.04 4.63 0.55  18.0 9.9 8.1 
PC PDI-variant of bimodal van Genuchten 66 0.08 1.13 0.82 108 0.08 2.97 0.18  30.1 13.5 26.2 
Minimum 44    41    14 
Median 19 59    60    28 17 
Mean 19 58    117    27 16 11 
Standard Deviation    116    8.32 
Maximum 29 70    453    27 40 29 26 
Figure 3

Site WW, Water Retention Curve, volumetric water content vs. matric potential (pF) (a), Pore Size Distribution Curve, density distribution of pore size (dimensionless) vs. pore size (μm) (b) and hydraulic conductivity (K, mm/hr) vs. water content (%) (c).

Figure 3

Site WW, Water Retention Curve, volumetric water content vs. matric potential (pF) (a), Pore Size Distribution Curve, density distribution of pore size (dimensionless) vs. pore size (μm) (b) and hydraulic conductivity (K, mm/hr) vs. water content (%) (c).

The majority (16/19) of water retention curves had the bimodal model as the curve of best fit, meaning that most samples had a heterogeneous pore size distribution. All of the soils that were classified as poorly graded (see Table 2) had a bimodal model for the water retention curve of best fit. Only a few soils were dominated by either interparticle or intraparticle voids (w1>0.8 at sites ED, I1 and PC, and w2>0.8 at site UMR). The dominant pore size radius for interparticle voids and intraparticle voids are shown in Table 4 and graphically in the pore size distribution figures in Supplementary Materials. Pore radii between 0.2 and 30 μm hold the most water that is available for plant uptake, whereas pore sizes greater than 30 μm are too large to hold water and instead drain the water quite quickly (Lambers et al. 2008). All interparticle pore sizes were above 30 μm, and only 11/19 sites had soils with intraparticle pore sizes between 0.2 and 30 μm. Permanent wilting point was an average of 16% and PAW had a mean of 11%, which is in the range of a loam soil (Lambers et al. 2008). Most of the soils sampled have intraparticle pores, which retain water for plants, and a PAW and wilting point similar to a loam soil, but unlike a loam also have large pores that encourage fast drainage. This is similar to other studies on engineered media (green roof media (Hill et al. 2019) and comparisons of green roof and bioretention media (Liu & Fassman-Beck 2018)), have reported predominantly bimodal water retention curves and large interparticle pore sizes that facilitate fast drainage.

The unsaturated hydraulic conductivity measured by the HYPROP device was typically between 0 and 1 mm/hr (see Figure 3(c) and Supplementary Materials), whereas the hydraulic conductivity curves estimated a saturated hydraulic conductivity orders of magnitude lower than measured in the field, and behaved very similarly to a power function. For example, the saturated hydraulic conductivity estimated by the HYPROP for WW (Figure 3) was 1.55 mm/hr, and the field saturated hydraulic conductivity for WW (see Figure 4) was 46 mm/hr. The median HYPROP KSat estimate was 9.0 mm/hr, and the median field KSat measurement was 177 mm/hr. Hydraulic conductivity was estimated using the difference in tension at the two different levels of tensiometers within the soil sample between when the soil was fully saturated and when it was nearly dry (UMS). As all samples were of sandy soil, the soil samples all dried out within a few days, and there was often only a minimal difference in tensions between the two tensiometers. Therefore, most samples only had a few data points for hydraulic conductivity. Therefore, the hydraulic conductivity curves should only be used cautiously for engineered media, and are not likely to provide reasonable estimates of saturated hydraulic conductivity.

Figure 4

Infiltration Rate Boxplots for all sites.

Figure 4

Infiltration Rate Boxplots for all sites.

Infiltration capacity testing

The boxplot of results for each site is shown in Figure 4. The minimum KSat recommended for bioretention is 25 mm/hr (Credit Valley Conservation & Toronto and Region Conservation Authority 2011), and all sites were at or above 25 mm/hr at the median per site. The median KSat was used as data were not normally distributed overall. The median KSat across all measurements was 176 mm/hr. The lowest KSat values were observed randomly throughout the cell, and there was no obvious correlation between location within the cell and KSat. Where possible, the saturated hydraulic conductivity measured in 2019 was compared to post-construction conditions, shown in Table 5. The KSat increased for six out of nine sites and decreased at one site.

Table 5

Comparison of KSat measured post-construction and in 2019

SiteKSat post-construction (mm/hr)Bioretention age (years)Median KSat 2019 (mm/hr)Change
WW 44 46 – 
ED 133 152 – 
KK 76 175 ↑ 
A1 78 182 ↑ 
A2 67 109 ↑ 
A3 109 289 ↑ 
UMR 137 52 ↓ 
UC 69 96 ↑ 
PC 122 10 175 ↑ 
SiteKSat post-construction (mm/hr)Bioretention age (years)Median KSat 2019 (mm/hr)Change
WW 44 46 – 
ED 133 152 – 
KK 76 175 ↑ 
A1 78 182 ↑ 
A2 67 109 ↑ 
A3 109 289 ↑ 
UMR 137 52 ↓ 
UC 69 96 ↑ 
PC 122 10 175 ↑ 

Correlation between soil, age, and infiltration

Correlation tests were conducted for all numerical parameters, and are shown in Table 6, with highlighting indicating whether the Spearman correlation test found correlation at significant levels (p-value<0.10). There was a moderate negative correlation between age and median KSat and between age and porosity (see Figure 5), meaning that KSat and porosity appeared to decrease slightly with age. Interestingly, KSat was not correlated with D10, which is often used in pedotransfer equations for determining KSat (Chapuis 2012), but KSat was moderately positively correlated with porosity and organic matter, and moderately negatively correlated with water content at field capacity (ϴ(pF=2.5)) and water content at wilting point (ϴ(pF=4.2)). A strong positive correlation was found between porosity and organic content. The drainage area ratio was not correlated with any parameter. Several moderate correlations were found that are not the focus of the research in this paper but are reported in case these parameters are of use to future researchers. PAW was moderately correlated with soil structure parameters (% clay, D10, Cu), but not with porosity, organic content or KSat. Other moderate correlations included: organic content and saturated water content, porosity and weighting factor (w1), and porosity and water content at wilting point (ϴ(pF=4.2).

Table 6

Pearson correlation coefficients between various site and soil parameters

 
 
Figure 5

Scatterplots for KSat vs Age (a) and Porosity vs Age (b), including Spearman rho coefficient.

Figure 5

Scatterplots for KSat vs Age (a) and Porosity vs Age (b), including Spearman rho coefficient.

Synthetic drawdown testing

Synthetic drawdown testing was conducted at three sites (see Figure 6). Though all three sites contained underdrains, one site (KV) functioned as a fully-infiltrating system, as the valve on the underdrain remained closed for the drawdown test. KV had the longest drawdown time at 6 hours and a natural recession limb, likely due to the closed underdrain. S2 had the next longest drawdown time of 4 hours, and during the test, no flow was observed in the underdrain. The A3 system had flow in the underdrain immediately, and a drawdown time of just 5 minutes, which suggests there was short-circuiting and system failure due to its inability to retain runoff. A3 and S2 had recession curves that are nearly vertical, showing the rapid release from the system. Two out of the three systems were able to completely retain the 25-mm-equivalent storm volume. The drawdown times for all systems were much less than the required drawdown time of 24–48 hours (Credit Valley Conservation & Toronto and Region Conservation Authority 2011), showing that these systems are still exfiltrating quite quickly.

Figure 6

Drawdown curves for synthetic drawdown tests at sites S2, KV and A3.KV had a closed valve at the underdrain at the time of the drawdown test.

Figure 6

Drawdown curves for synthetic drawdown tests at sites S2, KV and A3.KV had a closed valve at the underdrain at the time of the drawdown test.

Long term performance

From a hydrologic perspective, bioretention systems are functioning well over the long term based on results from 22 mature bioretention sites across Ontario that were up to 10 years old. KSat for 20/22 sites was at or above 25 mm/hr at the median, the recommended minimum according to local guidelines (Credit Valley Conservation & Toronto and Region Conservation Authority 2011). Similarly maintained or improved KSat was observed in other bioretention surveys (Asleson et al. 2009; Wardynski & Hunt 2012; Kluge et al. 2018). In a study of eight bioretention systems in Minnesota, Asleson et al. (2009) measured KSat at above 30 mm/hr for all sites at the median. In a similar study of 43 bioretention systems in North Carolina, 98% of systems had media with adequate or high permeability, as measured using a constant-head test on soil cores (Wardynski & Hunt 2012). The study with the oldest bioretention cells (10–22 years across 48 cells) found that 44/48 sites had KSat values that exceeded local minimum guidelines (Kluge et al. 2018).

Through this study and others (Jenkins et al. 2010; Khan et al. 2012; Le Coustumer et al. 2012; Paus et al. 2013), it remains uncertain as to whether KSat improves or deteriorates with age. KSat was moderately negatively correlated with age (Spearman correlation coefficient of −0.49), but KSat increased for 6/9 sites where post-construction KSat data was available. In other field studies where KSat was monitored over time, researchers found no change over time (Jenkins et al. 2010) or an increase in KSat over time (Paus et al. 2013). The field results are surprising compared to laboratory testing of simulated aging, where simulated runoff on columns or mesocosms showed significant decreases in KSat values over the course of the experiments in Le Coustumer et al. (2012) and in Khan et al. (2012). Laboratory experiments occur on short time scales, such as weeks to months, and do not have the same type of dry periods between wet periods as field systems. The simulated experiments do not allow sufficient time for slow biological, soil formation processes or preferential flow pathways to become evident and are likely severely underestimating the systems' ability to maintain infiltration rates.

In bioretention systems, the living part of these engineered systems is likely responsible for maintaining high hydrologic capacity. Soil structure is influenced by roots, soil fauna, microorganisms, environmental variables (e.g. wetting and drying cycles) and inorganic binding agents (Six et al. 2004). Plant roots have been shown to increase KSat by creating preferential flow paths and increasing porosity (Rasse et al. 2000; Hatt et al. 2009; Muerdter et al. 2018). Freezing and thawing in cold climates has also been shown to increase the size of macropores responsible for fast drainage (Ding et al. 2019) and infiltration rates (Denich et al. 2013).

The aging of bioretention media is also the development of a soil profile, a process known as pedogenesis. Pedogenesis starts when vegetation establishes in the soil, adding organic matter to the surface via litter deposition, root exudates, growth and dieback, which creates an organic horizon at the soil surface over the course of several years (Buol 2006). Ayers & Kangas (2018) observed the process of pedogenesis in bioretention at 10 field sites and argue that pedogensis maintains soil porosity, which in turn maintains infiltration, combatting the transport of fines and sediment from runoff.

The fast-draining nature of sand (all sites had sandy soils) appeared to be counter-acted by pedogenesis for sites in this study. A majority of soils had intraparticle pore sizes that encouraged retention for plant uptake, and the wilting point and PAW were within the range of a loam soil instead of a sand soil. There was a strong positive correlation between organic matter and porosity in this study, which suggests that adding organic matter and pedogensis maintains a more suitable soil environment for plants over time. However, this study is lacking soil-water interaction properties (e.g. field capacity, wilting point, plant available water) from the time when the media was first installed, and so further research is needed to determine if these soil properties have changed over time.

In several other mature bioretention studies, performance appeared to improve over time, which may also be due to soil pedogenesis and structural processes maintained by vegetation and soil fauna. In Johnson & Hunt (2019), volume reduction and total phosphorous reduction both increased after 17 years of operation. In Spraakman et al. (2020b), water quality at the effluent was much more stable after 4–5 years of operation than at the post-construction phase. Lastly, Willard et al. (2017) conducted a similar study on a bioretention system with 7 years of operation, and found similar performance for volume reduction and contaminant reduction in both the post-construction and 7-years operation conditions. Though the literature appears mixed on KSat improving or deteriorating with age, other performance factors of bioretention shows agreement on improvement over time (Willard et al. 2017; Johnson & Hunt 2019; Spraakman et al. 2020b).

Monitoring and maintenance requirements

The most needed maintenance task was to clear the inlet of sediment and debris. The start of a storm event carries the most sediment, referred to as the ‘first flush’ (Sansalone & Buchberger 1997). Research on spatial variation of media within bioretention systems shows that fines and contaminants such as metals accumulate near the inlet and in the top 10 cm of media (Paus et al. 2013; Lucke & Nichols 2015; Johnson & Hunt 2016; Kluge et al. 2018), indicating that there is very little migration of ‘first flush’ contaminants beyond the inlet and top of media zone. The increase of fines and contaminants near the inlet and in the top 10 cm of media may not be impacting KSat, as KSat in this study was not correlated with location inside the cell. Similarly, Asleson et al. (2009) observed low KSat values near the inlet, but also distributed randomly throughout the cell. Also, in Gnanaraj (2018) study of the KV site also used in this study, KSat was higher in the centerline of flow and lower along the side slopes, but there was no strong correlation between KSat and distance away from the inlet. In a tracer test also at the KV site, only 30% of the soil volume was used (Gu et al. In Prep.), which lends evidence to the notion that sediment and contaminants build up only at the inlet of the bioretention system.

As bioretention systems are distributed throughout a catchment and have a small drainage area, the amount of sediment that builds up in each cell is minimal if the catchment remains undisturbed (i.e. little construction activity or loose soils), particularly when compared to the larger catchment areas of stormwater management ponds. Many of the sites in this study were constructed as demonstration sites, thus maintenance activities were not always assigned in a systematic way, and for several sites, there had been no maintenance activities since construction. Clearing sediment from the inlet and removing the top 10 cm only near the inlet are minimal tasks in terms of cost, time and equipment needed.

Based on our experience with this monitoring methodology, we recommend conducting frequent visual inspections (every 6 months or 1 year), and less frequent synthetic drawdown tests (every 5–7 years), as well as continuous water level logging wherever possible. Synthetic drawdown tests should also be conducted at the assumption stage, to verify that the systems are functioning as designed. This is particularly relevant for site A3, which exhibited short-circuiting that we believe is due to the shallow media and the underdrain being located near to the inlet for the system, both of which are design flaws. We do not recommend investing significant time and energy into infiltration capacity testing, unless synthetic drawdown tests are not possible at the site due to access restrictions. From visual inspections, we were able to determine if the site had ponded water or not, and what maintenance activities needed to be performed. This is sufficient information for an operator to be able to determine if the system is functioning. Operators can then determine the exact monitoring and maintenance frequency (e.g. how quickly inlets become clogged with sediment, and therefore how often they should be cleaned), and more accurately understand the costs of these activities. Although synthetic drawdown testing may be difficult to coordinate, it provides superior information on exact retention volumes and whether short-circuiting is occurring, which infiltration capacity testing does not provide. Infiltration capacity testing is at least 4 hours for one person to complete (e.g. measurements should be at 10–15 locations and each location will take approximately 15–20 minutes) and must be done under similar antecedent moisture conditions and time of year to be truly comparable year-to-year. Also, the infiltration capacity measured is only for the bioretention media, and does not tell the operator about the length of time it takes for a rainfall event to soak into the native ground, or whether that volume is completely retained or instead goes to the underdrain. Synthetic drawdown testing will give you the information on retention and drawdown time. Synthetic drawdown testing requires a level logger in place, and so we recommend installing a monitoring well during construction of a new bioretention cell.

Design

In the three sites with synthetic drawdown testing, the drawdown time (6 hours or less) was much less than the required drawdown time of 24–48 hours. With infiltration capacity testing, the median across all measurements was 176 mm/hr, and so drawdown at all sites would also be similarly fast. All soils were classified as sandy, and the water release curves indicated that most water was freely drained and only an average of 11% was available for plant uptake. These soil characteristics are the result of bioretention media guidelines in the local area requiring sand content above 80% (Credit Valley Conservation & Toronto and Region Conservation Authority 2011). Other jurisdictions require less sand and/or have methods for amending native soils (CIRIA 2015; Minnesota Pollution Control Agency 2020) which allows for more flexibility for designers and contractors to select materials that are available closer to the subject site. Soil with a high sand content is not always available close to the construction site, such as at sites KV, I1, I2, I3, where the local soils were all silty clay. Guidelines requiring a high sand content mean that material must be imported from elsewhere in the province or country, and that displaced material be transported to a landfill, which is not likely to be an environmentally sustainable solution. Utilizing more of the readily-available soils, which are likely to contain more silt and clay, could increase plant available water, increase the drawdown time at the same time, and possibly provide a healthier environment for a wider variety of plants. Also, retaining water for a longer period of time inside the bioretention cell may increase water treatment, by allowing more time for chemical and biological treatment processes to take place. The total water retention via porosity could remain relatively unchanged if design standards decreased sand content slightly, though sand should still be greater than silt and clay contents.

The fast drawdown times measured in this study indicate that an underdrain may not have been necessary, as clearly the native soils were able to exfiltrate the runoff water quite quickly. At site KV, where an underdrain was in place but a valve was closed during the synthetic runoff test, the native soils were silty clay. In a study in Minnesota where synthetic runoff tests were also conducted, all three sites had drawdown times less than 4 hours, including one that contained a native soil layer of silt loam (Asleson et al. 2009). According to local guidelines, a safety factor 2–3 should be applied to native soil KSat (Toronto and Region Conservation Authority 2020a), and any KSat less than 15 mm/hr should have an underdrain. With adding underdrains where these are not necessary, small but intense rainfall events could trigger flow in the underdrain, leading to more surface runoff and less retention than designed. To truly maximize retention, we recommend removing or at least reducing the safety factor applied to KSat and installing underdrains with closed valves that can be opened if the cell is not able to retain small rainfall events regularly.

The evidence is building that bioretention systems will continue to perform well, from a hydrologic perspective, for up to 10 years, given the age of the oldest site in this study. The bioretention systems surveyed in this study received little to no maintenance overall, and were located in a cold climate, and are still functioning well to retain stormwater runoff years after installation. This is likely due to the development of the media's structure over time from deposition of organic matter, activity of soil fauna, and environmental factors such as wetting/drying and freeze/thaw. Field studies such as this one indicate that infiltration capacity of bioretention is maintained or increased over time, which is contrary to laboratory studies which show clogging and decreased KSat over time. The difference may be in the development of the soil structure over time, which does not occur under the short time scales and artificial conditions within the laboratory.

We recommend that owners maintain a regular schedule of visual inspections. Also, we think designers should enhance retention even further by using media with less sand content or capping/altering underdrains. This study adds to the literature on field performance of mature bioretention systems, showing that hydrologic performance is maintained and that routine and minor monitoring and maintenance are all that is needed for continued long-term functioning.

Further research is needed to determine at which point these systems fail and methods for rehabilitation, to assist operators with creating better long-term cost forecasts. We recommend that operators track monitoring and maintenance using similar methods to those outlined in this study and share this with researchers. Further collaboration between practicing engineers, operators and researchers is needed to advance the long-term performance and operation needs of bioretention systems.

All relevant data are included in the paper or its Supplementary Information.

AMS Inc.
2012
Bulk Density Sampler
.
Aquafor Beech Ltd.
2017
Draft Low Impact Development (LID) Stormwater Management Guidance Manual
. Aquafor Beech, Missisauga, Ontario.
Asleson
B. C.
,
Nestingen
R. S.
,
Gulliver
J. S.
,
Hozalski
R. M.
&
Nieber
J. L.
2009
Performance assessment of rain gardens
.
Journal of the American Water Resources Association
45
(
4
),
1019
1031
.
ASTM
2007
D422-63 Standard Test Method for Particle-Size Analysis of Soils
.
ASTM International
,
West Conshohocken, PA
.
ASTM
2014
D2974-14 Standard Test Methods for Moisture, Ash, and Organic Matter of Peat and Other Organic Soils
.
ASTM International
,
West Conshohocken, PA
.
ASTM
2017
D2487-17 Standard Practice for Classification of Soils for Engineering Purposes (Unified Soil Classification System)
.
ASTM International
,
West Conshohocken, PA
.
Ayers
E. M.
&
Kangas
P.
2018
Soil layer development and biota in bioretention
.
Water
10
(
11
), 1587.
Basile
A.
,
Coppola
A.
,
De Mascellis
R.
,
Mele
G.
&
Terribile
F.
2006
A comparative analysis of the pore system in volcanic soils by means of water-retention measurements and image analysis
. In:
Soil of Volcanic Regions in Europe
(
Arnalds
O.
, ed.).
Springer
,
Berlin
.
Bradford
A.
&
Gharabaghi
B.
2004
Evolution of Ontario's stormwater management planning and design guidance
.
Water Quality Research Journal of Canada
39
(
4
),
343
355
.
Buol
S.
2006
Pedogenic processes and pathways of horizon differentiation. In R. Scalenghe (author) & G. Certini (ed.), Soils: Basic Concepts and Future Challenges. (R. Scalenghe (author) & G. Certini, eds.), Cambridge University Press, Cambridge, pp. 11–22. doi:10.1017/CBO9780511535802.003.
Chapuis
R. P.
2012
Predicting the saturated hydraulic conductivity of soils: a review
.
Bulletin of Engineering Geology and the Environment
71
(
3
),
401
434
.
CIRIA
2015
The SUDS Manual, CIRIA Report No. C753
.
Dundee
,
Scotland
.
Credit Valley Conservation
2015
Lessons Learned: CVC Stormwater Management and Low Impact Development Monitoring and Performance Assessment Guide
. Credit Valley Conservation, Missisauga, Ontario.
Credit Valley Conservation
2020
LID Monitoring Technical Reports
.
Credit Valley Conservation and Toronto and Region Conservation Authority
2011
Low Impact Development Stormwater Management Planning and Design Guide
. Credit Valley Conservation, Missisauga, Ontario.
Davis
A. P.
,
Traver
R. G.
,
Hunt
W. F.
,
Lee
R.
,
Brown
R. A.
&
Olszewski
J. M.
2012
Hydrologic performance of bioretention storm – water control measures
.
Journal of Hydrologic Engineering
17
(
5
),
604
614
.
Ding
B.
,
Rezanezhad
F.
,
Gharedaghloo
B.
,
Van Cappellen
P.
&
Passeport
E.
2019
Bioretention cells under cold climate conditions: effects of freezing and thawing on water infiltration, soil structure, and nutrient removal
.
Science of the Total Environment
649
,
749
759
.
DMTI Spatial Inc.
2014
CanMap RouteLogistics
.
DMTI Spatial, Markham, ON
.
Elrick
D. E.
&
Reynolds
W. D.
1992
Methods for analyzing constant-head well permeameter data
.
Soil Science Society of America Journal
56
(
1
),
320
323
.
Environmental Registry of Ontario
2017
Low Impact Development Stormwater Management Guidance Manual
.
Fletcher
T. D.
,
Shuster
W.
,
Hunt
W. F.
,
Ashley
R.
,
Butler
D.
,
Arthur
S.
,
Trowsdale
S.
,
Barraud
S.
,
Semadeni-Davies
A.
,
Bertrand-Krajewski
J.-L.
,
Steen Mikkelsen
P.
,
Rivard
G.
,
Uhl
M.
,
Dagenais
D.
&
Viklander
M.
2015
SUDS, LID, BMPs, WSUD and more – the evolution and application of terminology surrounding urban drainage
.
Urban Water Journal
12
(
7
),
525
542
.
Foster
J.
,
Lowe
A.
&
Winkelman
S.
2011
The value of green infrastructure for urban climate adaptation
.
The Center for Clean Air Policy
, Washington, DC.
Gnanaraj
A. R. J.
2018
Spatial and Temporal Analysis of Hydraulic Conductivity, Snow Depth and Soil Properties of a Bioretention System
.
Master of Applied Science
,
Department of Civil and Mineral Engineering, University of Toronto
,
Toronto
,
Canada
.
Gu
X.
,
Rodgers
T. F. M.
,
Spraakman
S.
,
Van Seters
T.
,
Flick
R.
,
Diamond
M. M. L.
,
Drake
J.
&
Passeport
E.
(In Prep)
.
Trace Organic Contaminant Transfer and Transformation in Bioretention Cells: A Field Tracer Test with Benzotriazole
.
Gulliver
J. S.
,
Erickson
A. J.
&
Weiss
P. T.
2010
Stormwater Treatment: Assessment and Maintenance
.
University of Minnesota, St. Anthony Falls Laboratory
,
Minneapolis, MN
.
Hatt
B. E.
,
Fletcher
T. D.
&
Deletic
A.
2009
Hydrologic and pollutant removal performance of stormwater biofiltration systems at the field scale
.
Journal of Hydrology
365
(
3–4
),
310
321
.
Jamei
E.
&
Tapper
N.
2019
WSUD and urban heat island effect mitigation
. In:
Approaches to Water Sensitive Urban Design
(A. K. Sharma, T. Gardner & D. Begbie, eds), Elsevier, Oxford, pp.
381
407
.
Jenkins
J. K. G.
,
Wadzuk
B. M.
&
Welker
A. L.
2010
Fines accumulation and distribution in a storm-water rain garden nine years postconstruction
.
Journal of Irrigation and Drainage Engineering
136
(
12
),
862
869
.
Kavehei
E.
,
Jenkins
G. A.
,
Adame
M. F.
&
Lemckert
C.
2018
Carbon sequestration potential for mitigating the carbon footprint of green stormwater infrastructure
.
Renewable and Sustainable Energy Reviews
94
,
1179
1191
.
Khan
U. T.
,
Valeo
C.
,
Chu
A.
&
van Duin
B.
2012
Bioretention cell efficacy in cold climates: part 1 — hydrologic performance
.
Canadian Journal of Civil Engineering
39
(
11
),
1210
1221
.
Kluge
B.
,
Markert
A.
,
Facklam
M.
,
Sommer
H.
,
Kaiser
M.
,
Pallasch
M.
&
Wessolek
G.
2018
Metal accumulation and hydraulic performance of bioretention systems after long-term operation
.
Journal of Soils and Sediments
18
(
2
),
431
444
.
Lake Simcoe Region Conservation Authority
2016
LSRCA Technical Guidelines for Stormwater Management Submissions
. Lake Simcoe Region Conservation Authority, Newmarket, Ontario.
Lambers
H.
,
Chapin
F. S.
&
Pons
T. L.
2008
Plant water relations
. In:
Plant Physiological Ecology
(Lambers, H., Chapin, F. S. & Pons, T. L., eds), Springer Science + Business Media, New York, NY, pp.
101
150
.
Le Coustumer
S.
,
Fletcher
T. D.
,
Deletic
A.
,
Barraud
S.
&
Poelsma
P.
2012
The influence of design parameters on clogging of stormwater biofilters: a large-scale column study
.
Water Research
46
(
20
),
6743
6752
.
Lucke
T.
&
Nichols
P. W. B.
2015
The pollution removal and stormwater reduction performance of street-side bioretention basins after ten years in operation
.
Science of the Total Environment
536
(
August
),
784
792
.
Minnesota Pollution Control Agency
2020
Design Criteria for Bioretention
.
Muerdter
C. P.
,
Wong
C. K.
&
LeFevre
G. H.
2018
Emerging investigator series: the role of vegetation in bioretention for stormwater treatment in the built environment: pollutant removal, hydrologic function, and ancillary benefits
.
Environmental Science: Water Research & Technology
4
(
5
),
592
612
.
OpenStreetMap contributors
2021
OpenStreetMap
.
Paus
K. H.
,
Morgan
J.
,
Gulliver
J. S.
,
Leiknes
T.
&
Hozalski
R. M.
2013
Assessment of the hydraulic and toxic metal removal capacities of bioretention cells after 2 to 8 years of service
.
Water, Air and Soil Pollution
225
,
1803
.
Pertassek
T.
,
Peters
A.
&
Durner
W.
2015
HYPROP-FIT Software User's Manual, V.3.0
.
METER Group AG
,
Munchen
,
Germany
.
Rasse
D. P.
,
Smucker
A. J. M.
&
Santos
D.
2000
Alfalfa root and shoot mulching effects on soil hydraulic properties and aggregation
.
Soil Science Society of America Journal
64
(
2
),
725
731
.
R Core Team
2017
R: A Language and Environment for Statistical Computing
.
R Foundation for Statistical Computing
.
Roy
A. H.
,
Wenger
S. J.
,
Fletcher
T. D.
,
Walsh
C. J.
,
Ladson
A. R.
,
Shuster
W. D.
,
Thurston
H. W.
&
Brown
R. R.
2008
Impediments and solutions to sustainable, watershed-scale urban stormwater management: lessons from Australia and the United States
.
Environmental Management
42
(
2
),
344
359
.
Sansalone
J. J.
&
Buchberger
S. G.
1997
Partitioning and first flush of metals in urban roadway storm water
.
Journal of Environmental Engineering
123
(
2
),
134
143
.
Spraakman
S.
,
Rodgers
T. F. M.
,
Monri-Fung
H.
,
Nowicki
A.
,
Diamond
M. L.
,
Passeport
E.
,
Thuna
M.
&
Drake
J.
2020a
A need for standardized reporting: a scoping review of bioretention research 2000–2019
.
Water
12
(
11
), 3122.
Toronto and Region Conservation Authority
2020a
Design Infiltration Rate
.
Toronto and Region Conservation Authority
2020b
Low Impact Development
.
van Genuchten
M. T.
1980
A closed-form equation for predicting the hydraulic conductivity of unsaturated soils
.
Soil Science Society of America Journal
44
(
5
),
892
898
.
Walsh
C. J.
,
Roy
A. H.
,
Feminella
J. W.
,
Cottingham
P. D.
,
Groffman
P. M.
&
Morgan
R. P.
2005
The urban stream syndrome: current knowledge and the search for a cure
.
Journal of the North American Benthological Society
24
(
3
),
706
723
.
Wardynski
B. J.
&
Hunt
W. F.
2012
Are bioretention cells being installed per design standards in North Carolina? A field study
.
Journal of Environmental Engineering
138
(
12
),
1210
1217
.
Wickham
H.
2016
ggplot2: Elegant Graphics for Data Analysis
.
Springer-Verlag
,
New York, NY
.
Willard
L. L.
,
Wynn-Thompson
T.
,
Krometis
L. H.
,
Neher
T. P.
&
Badgley
B. D.
2017
Does it pay to be mature? Evaluation of bioretention cell performance seven years postconstruction
.
Journal of Environmental Engineering
143
(
9
),
04017041
.
1–10
.
This is an Open Access article distributed under the terms of the Creative Commons Attribution Licence (CC BY-NC-ND 4.0), which permits copying and redistribution for non-commercial purposes with no derivatives, provided the original work is properly cited (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Supplementary data