November-December 2011

Bioretention Filters

Part 1: What do we know?

Tweet

Bioretention has been in use for two decades. A large body of laboratory and field data provides indications of the performance. It is therefore appropriate to synthesize this information so as to provide insight into when and when not bioretention should be used. This first of two articles on bioretention focuses on configuration, processes, and design criteria. The second part focuses on hydrologic and treatment performance.

Overview
Bioretention is not new. It was used by the Incas of Peru to protect the steep slopes of Machu Pichu. The water was stored in gravel and rock sublayers to be “pumped” out by the vegetation via evapotranspiration.

The basic configuration is shown in Figure 1. Bioretention is a sorptive filter. Alternatively, it could be considered either an infiltration system or a biological system. However, bioretention is best viewed as a filter—specifically, a modification and therefore an extension of the sand filter, but one in which the amendments are mixed rather than placed as a layer, as is the case with the older organic filter in which a layer of peat was placed on or beneath a sand layer. Thus, bioretention is a sand filter to which amendments are added to remove dissolved pollutants, a capability presumed to not be present in un-amended sand. The unit processes are listed in Table 1.

Figure 1. Basic configuration

We should not presume that if stormwater enters the soil that all of each metal is removed before reaching groundwater and subsequently the lake or stream.

It is unfortunate that the word bioretention is used to label a unit operation. The more appropriate use of this word is for a group of unit processes involving biological organisms, carrying out processes such as photosynthesis, nitrification-denitrification, and sulfate reduction. Were this its use, it would be recognized that bioretention occurs in most unit operations. Examples are wet ponds, constructed wetlands, and even sand filters.

In keeping with the objective of using names that explicitly convey the basic nature of a unit operation, the more appropriate label is vegetated amended sand filter. However, it is unlikely that the descriptor bioretention as currently used will change. Some clarification is afforded by the use of the label bioretention filter, which conveys the basic nature of the treatment system and is therefore used in this article.

Bioretention filters are generally recommended for small drainage areas such as cul-de-sacs, circle drives at building entrances, and small parking lots. However, they have been installed to treat drainage areas as large as about 10 acres.

Rain gardens are a “sister” to bioretention filters, used on the individual home scale. The difference is that with rain gardens, no underdrains are used, and the area of the filter is sized to ensure full infiltration. Consequently, their size can range to 30% of the impervious area draining to them, reflecting generally poorly infiltrating soils. Commonly, only loam soil is used as the filter media, excluding the inclusion of sand. In contrast, about 5% to 10% of the development is consumed by bioretention filters.

Configurations
Bioretention filters involve constructing a depression beneath the current ground level sufficient in size to contain the water quality design storm to meet the treatment performance goal. The soil bed immediately below this elevation is removed and replaced with a specified media, commonly a mixture of sand, compost, and loam soil. Early practice specified adding phosphorus, nitrogen, and potassium for plant growth. It is now recognized that stormwater provides these nutrients. The system may include pretreatment such as a catch basin with a sump or a grass slope (strip) leading to the filter in the bottom of the treatment area.

The lower chamber shown in Figure 1 beneath the underdrains provides two benefits. It enhances infiltration by storing stormwater between storms, and it promotes nitrogen removal through the transformation of nitrate to nitrogen gas if a carbon source is added.

Bioretention filters may or may not have underdrains. Underdrains are generally used in poorly to moderately well-drained soils. They are not generally used in well-drained soils. However, some have recommended their use even in well-drained soils to better handle the occasional large event.

Bioretention filters come in two general configurations: cell and swale. A cell is generally rectangular with low length-to-width ratio (say, five or less), oval, or undulating conforming to a landscaped area. A bioretention swale, commonly called a dry swale, has a large length-to-width ratio, generally greater than 10, and is rectangular and flat, without a longitudinal slope, in comparison to a sloped grass swale.

A bioretention swale when placed in well-drained soils has also been referred to as bioinfiltration, blurring the distinction between bioretention cells and infiltration basins. A preferable label for what is now called the dry swale is simply bioretention filter, making no distinction between cells and swales, given that the performance will be the same, or alternatively bioretention filter swale.

There is also a rooftop drain by Americast in which the basic Filterra product is used. Its parent filter has been tested in the field and shows it can meet the performance goal of 80% total suspended solids (TSS) removal.

A dry swale is essentially a narrow bioretention unit, treating stormwater from the street. The cell is placed between the street and sidewalk, treating street runoff. It appears to have a problem of litter and sand clogging the inlet. Clogging likely occurs because the water must turn 90 degrees to enter the filter. The outcome is not enough water entering the cell to meet the particular volume performance goal.

Design Elements and Specifications
Types of specifications include:

• filter media
• media depth
• operating water depth
• filtration rate
• drawdown time
• mulch layer on the surface
• vegetated surface

Media. The original media specification (50 to 60% sand, 20 to 30% topsoil, and 20 to 30% leaf compost) has evolved into over a dozen different specifications as indicated in Table 2. The specifications were obtained from manuals, reports, and articles. The most commonly used mix appears to be sand with native soil (loam, sandy loam, or loamy sand) and compost or peat as the organic amendments. Compost is favored over peat given the higher cost of peat.

The mixes presented in Table 2 were developed for wet climate regions. For cold climate regions, the clay content must be kept low, less than 2%, to avoid solid freezing. For semi-arid climates, native plants are habituated to a soil organic content that is lower than in wet climates. It may therefore be advisable to reduce the organic content by reducing the compost percentage with more sand and/or more native soil.

It is not known if the differing compositions in Table 2 result in different levels of performance. One study provides insight. The researchers studied eight combinations of four media ingredients: sand, clay, soil, and compost. With respect to nitrogen removal, 100% washed sand gave the best performance, with removals of 50 to 75% depending on the form of nitrogen. Mixes with compost and/or soil had negative removals, with the sand/compost/soil mix being the worst. All mixes achieved a high removal of copper.

However, for zinc two mixes were best: 100% sand, and sand/compost/topsoil at 70/10/20%. The chemistry of the sand surface was not investigated. However, the removal rates of metals suggest an oxide coating. The unanswered question from this study is sorption capacity.

Another study found similar results. Phosphorus and nitrogen removal was positive only with pure sand as the media. All combinations that included loam soil had negative performance for the two nutrients. It thus appears that a standard sand filter offers better performance. A second study evaluated various mix combinations of soil and sand covered by mulch. Mixtures that were dominated by either soil or compost were the least effective at removing sediment, mostly likely because of loss of fine solids from these two materials. All other pollutants were substantially removed.

One study found the best performing mix was 10% organics, 10% clays, 40% fine sand, and 40% gravel with a filtration rate of 3.5 inches per hour. Pure sand was not investigated. The engineer must take care in the selection of compost, particularly if the media specification is only sand and compost (i.e., no loam soil), as it may be contaminated with metals, pesticides, or other persistent anthropogenic organics.

A field study found an increase in the concentrations of metals with the addition of compost to roadway shoulders. Leachate studies found an increase in dissolved metals and nutrients. The studied composts were made from community green yard waste (metals release) and fallen deciduous leaves (minor nutrient release).

The compost should be relatively mature, following the production specifications of the US Composting Council. It also should be free of excessive phosphorus and nitrogen.

A benefit of incorporating organic matter into rather than on the soil surface is enhancement of conditions for nitrifying and denitrifying bacteria, although experience with infiltration and bioretention filter systems suggests the latter activity is not significant. Incorporation improves conditions for predators of enteric bacteria and viruses. It is also possible that the vegetation itself provides sufficient organic matter. However, given the common applications of bioretention, organic litter such as leaves and fallen branches are removed for
aesthetic reasons.

Another factor is particle size. Using properly aged compost minimizes its rate of degradation in the facility. These concerns aside, the mulch protects the plants assisting in the retention of water. Compost particles with a diameter of 150 µm have about twice the sorption capacity of those with a diameter of 4,750 µm, directly related to the respective surface areas of the two sizes. The compost in the media presumably degrades gradually and its role in removing pollutants, particularly dissolved metals and pesticides, is replaced by organic material produced by the plants.

To ensure a homogenous mix of the three ingredients, it has been recommended that the media be prepared by commercial providers of topsoil and that these providers be required to periodically test their mixes for hydraulic conductivity. Mixing at the site with a front-end loader is not considered to be effective and should be discouraged.

There may be a tendency to minimize the amount of clay out of concern for its effect on the filtration rate. However, clay plays an important function regarding pollutant removal. It sorbs ammonia, phosphorus, and bacteria and viruses. Once sorbed, the ammonia is modified to nitrate by specialized bacteria, reopening the sorption sites. This likely occurs between storms when there is plenty of oxygen in the media, which the nitrifying bacteria require.

If iron or aluminum oxide or calcite/apatite is present on the clay or sand, dissolved phosphorus and metals are removed. It is likely that this phosphorus is also available to plants, possibly reopening the sorption sites as with ammonia.

Bacteria and viruses, once sorbed to clay, are susceptible to predation by other bacteria and fungi, although the clay provides some protection from predation. Differences in sand and clay surface chemistry might account for some of the differences in phosphorus and nitrogen removal observed among studies. All elements of the media, sand and clay, compost, and the native soil should be examined separately in every field study, with respect to individual performance, chemistry, and sorptive capacity.

The common specification for sand is ASTM C33 fine aggregate available at concrete mix companies. ASTM C33 contains from zero to 10% clay/silt, and loam soil contains from 10 to 30% clays. Hence, a mixture of 50% ASTM C33, 25% compost, and 25% loam soil would contain from 5% to 12% clay. Given the important role of clay, the percentage of clay should be specified, commensurate with the design filtration rate. Studies on the quantitative effects of clay with respect to the pollutant removal are needed to arrive at a suitable specification.

Care must be exercised with the selected loam soils. Early in the implementation of the Prince George’s County design manual, there was a 50% failure rate due to variability of the sandy loam amendment. Another study found a 25% failure rate for the same reason.

However, what specifically was wrong about the loam soil was not given. Presumably it was the filtration rate characteristics. It likely had too much clay relative to the desired filtration rate. A sandy loam or loamy sand is likely more satisfactory. The second issue is phosphorus release, which will be presented fully when discussing performance.

Media Depth. The original depth specification for the media was 4 feet. As indicated in Table 3, it is recommended that the depth of the media vary with the targeted pollutant. The needs of the plants with respect to the depth of the media must also be considered: 12 inches (0.3 meter) for turf grass and forbs, 18 inches (0.5 meter) for shrubs, and 30 inches (0.8 meter) for trees. It has been found that depth within the range of 2 to 4 feet is not a factor in performance for the removal of nitrogen, phosphorus, or zinc.

However, an earlier study found that performance-depth relationships were consistent with Table 3. This study also found a slight improvement in metals removal at a depth of 36 inches (1 meter). Hence, where a water body requires high-quality effluent, as is the case with a total maximum daily load (TMDL) plan, a depth of 3 to 4 feet might be used.

Water Depth. The original operating water depth of 6 inches has evolved to 12 inches. Eighteen inches has been recommended. There are essentially no data on the impact of these greater water depths and drawdown times on the health of plants. The original design criterion of only 6 inches was conservative given the lack of information in the early 1990s. However, anecdotal evidence suggests that 12 and possibly 18 inches is feasible. The operating depth clearly has a direct bearing on the size of the bioretention filter: the greater the depth, the smaller the filter area. However, this presents a conflict in design criteria: the smaller the filter area relative to the drainage area, the less the reduction in stormwater volume by infiltration and evapotranspiration (I/ET).

Filtration Rate. The commonly recommended filtration rate is 1 to 3 inches per hour. However, inasmuch as the lower limit for infiltration basins and trenches is 0.5 inch per hour, there is no reason why this should not be the lower limit for bioretention filters as well. Below 0.5 inch per hour, underdrains would be installed. It is likely that the removal of dissolved metals, sediment, and pesticides is independent of filtration rate given the ease with which these pollutants are removed. As noted below, however, this is apparently not the case with nitrogen.

Reduction of the more difficult pollutants such as phosphorus, nitrogen, bacteria, and temperature may be affected by the filtration rate. Concern has not been raised regarding the filtration rate being excessive. Table 3 lists recently recommended maximum filtration rates. Notice that lower maximum rates are recommended for the pollutants known to be more difficult to remove. However, higher filtration rates have been observed with good performance: 15 inches per hour and 50 to 100 inches per hour.

Drawdown Time. Drawdown times range from 24 to 72 hours depending on the design manual. An early laboratory study found no difference in performance between three and 12 hours. Effluent concentrations for nitrogen were cut in half by increasing the drawdown time from 24 to 72 hours. Another study found an improvement in nitrogen removal with an increase from two to four days, but a decrease as well as more inconsistent performance when increasing the drawdown time more than four days.

Mulch. A surface mulch layer of several inches is commonly recommended. It improves metals removal, captures much if not all of the incoming sediment, and adds to the well-being of the plants. It also assists in the removal of bacteria and viruses and xenobiotics. Given the important role of mulch in retaining the sediment, it must be replaced—that is, racked from the site—not added. If it is not replaced it will degrade, putting the sediment on the surface of the filter, resulting in clogging.

Mulch is essentially compost. It should be made from uncontaminated raw materials, low in metals and nutrients. This suggests the use of fallen, deciduous leaves rather than green biowaste. It should be mature and well composted, inasmuch as further breakdown at the site results in loss of nutrients and metals sorbed from the stormwater as well as original to the mulch. Regardless, it will gradually break down.

There have been many laboratory studies of the benefits of mulch to performance. It has been found effective at removing dissolved metals, total petroleum hydrocarbons (TPHs) and polycyclic aromatic hydrocarbons (PAHs), sediment, pesticides, and other persistent anthropogenic organics. However, the mulch degrades over time, evident in the recommendation that fresh mulch be changed every six to 12 months.

It is therefore not clear that the mulch should be counted on with respect to pollutant removal, given that it degrades. Regardless, the remaining mulch on the surface should be removed to remove the above pollutants, particularly with respect to accumulated sediment, prior to the placement of fresh mulch. The nature of the release of previously sorbed pollutants from a degrading mulch has not been researched; whether they are released in aqueous form or as colloids or likely both.

Colloids can be carried through the media as has been observed in soils. Dissolved copper binds with dissolved organics, a complex that is highly mobile. Another factor to consider is the particle size of the compost. Mulches act as physical barriers that dissipate the erosive force of rain drops, protecting the soil at the surface, which in turn improves permeability. The dark mulches used in bioretention units adsorb more solar energy, thereby increasing the soil temperature.

Vegetation. Native forbs and shrubs appear to be the most common plant cover. Trees may be included. Turf grass is also used, but should be discouraged as the tendency to water and fertilize is greater than with forbs and shrubs, resulting in more water leaving the filter than entering during the growth season. Frequent watering in the dry periods can eliminate the hydrologic benefits.

One original design criterion was to specify a minimum filter area of 600 square feet (56 square meters), with a minimum width of 15 feet (4.5 meters) and a minimum length of 40 feet (12.2 meters). The objective was creation of an adequate microclimate. Newer manuals now allow a minimum area of 200 square feet (18 square meters). While native plants are preferred, it should be recognized that they may not be used to frequent inundation. This may be particularly true in semi-arid environments.

Plants provide many benefits. The first benefit to consider is pollutant removal by photosynthesis: the use of nitrogen, phosphorus, and metals for growth. But to keep the unit process of photosynthesis effective, the foliage must be periodically harvested. The experience from forest research suggests that much of the nitrogen and phosphorus taken up by plants in bioretention filters is lost through leaf fall.

The fallen leaves degrade, with the nutrients re-entering the soil. As a consequence, net removal, particularly of nitrogen and phosphorus, over the long term may be low if fallen vegetative materials are not removed. It is therefore important to remove dead material from the surface of the filter.

Many studies have been conducted using small skill units called mesocosms to ascertain the contribution of the plants in removing pollutants. Bare filters are compared to vegetated filters. The difference in performance implies the benefit of the plants. Another approach is to compare seasonal differences with field units. The percentage removal difference has been most dramatic with nitrogen. The vegetated units’ removals of nitrogen in one study were 63, 77, and 77%, respectively, in comparison to minus 10, 10, and 25%, respectively, for the barren media. The ratio for orthophosphate was on the order of 0.1 for the vegetated and 0.25 for the barren gravel and loam, but about 0.1 for the sand, suggesting an oxide or calcium coating. However, the chemistry of the sand was not defined. The lack of clay in the gravel and little clay in the loam soil (3%) may have contributed to poor showings for phosphorus. Also, most of the phosphorus was particulate. Consequently, the improvements in removal by the vegetation may have simply represented a better ability to strain the particulates.

In one study, the tap water had 0.59 mg/L of nitrogen versus 5.44 mg/L in the stormwater, and dissolved phosphorus of 0.01 versus 0.02 mg/L, respectively. Flushing experiments with tap water showed little release of phosphorus, but again 95% of the phosphorus was particulate. The return to tap water resulted in a flushing of the nitrogen. Once exposed to higher levels of the two nutrients in the stormwater, the plants might respond by taking in quantities in excess of their metabolic needs.

The authors of the study concluded that “the most important effect of vegetation . . . is retention of nutrients, in particular nitrogen that would otherwise leach . . . after an interevent dry period. It appears that organic nutrients are mineralized . . . during these . . . periods, and the products of mineralization (nitrate and phosphate) are highly soluble and prone to leaching” absent vegetation. The authors further noted “uptake of nutrients in vegetated mesocosms may be due to the higher microbial activity and population [evidenced] by the lower redox potentials of the vegetated media.”

A study using mesocosms found that the vegetated systems performed better than their barren counterparts for the most part. Three different media were studied: gravel, sand, and loam soil individually rather than as a mixture. Nitrogen removal gradually decreased in both the vegetated and barren mesocosms for all three media, from an effluent/influent ratio of 0.2 to 0.7 for the barren and slightly less than 0.1 to 0.3 for the vegetated units, but appeared to be leveling half way through the test runs.

There is a question as to whether short-term mesocosm studies represent full-size units in the field. Possible biases with each mesocosm study are shown in Table 4. In the studies noted above, both tap water and synthetic stormwater were run through the filters in the order of tap-stormwater for several weeks. The possible bias with this approach, however, is the exposure of the media to relatively low concentrations of nutrients in the tap water immediately before the passage of stormwater.

A study was made of two field units, comparing seasonal uptake of nitrogen and phosphorus as a way to ascertain the benefits of vegetation. Table 5 presents seasonal data of two field filters from one study. The conclusion that must be drawn is that the data are inconsistent regarding the effect of plants. The benefits of vegetation are therefore not clear from this particular study. Note that the effluent concentration is higher in the winter than in the summer. The increase can be attributed to the effect of evapotranspiration, which will concentrate the pollutant, and leaching in the case of nitrogen from the loam soil. Another study of bioretention filters found they are less prone to seasonal performance variations than conventional stormwater management systems. There may be no contradiction in this statement inasmuch as the other systems evaluated at the test site and subject to considerable differences in seasonal performance. The study was conducted in a cold climate area.

In general, in comparison to the results of mesocosms, actual field units do not appear work as well. Regardless, uptake by plants will be short lived unless the vegetation is periodically harvested. However, experience from actual wetlands indicates that harvesting the foliage removes little of either nutrient or metals. In wetlands most of the removed metals, upward of 95%, are found in the soil and sorbed to the roots of the plants. This appears to be the case for bioretention filters.

Hence, the dominant role of the plants may be indirect. They provide an ecosystem conducive to microbial growth. However, if this is the case, the mechanism of removal by bacteria will end at some point with die-off equaling growth in the filter. Despite these positive results, particularly with respect to nitrogen removal, the performance of field units has not been as good as found in the laboratory generally.

Another indirect unit process involving plants relates to metals. Dissolved metals, such as zinc and copper, sorb to iron oxide on the roots of the plants. This is called plaque; the natural process is a defensive mechanism protecting the plant from excessive uptake of various metals, which can be toxic. This mechanism is likely why on the order of 95% of the removed metals are located in the soil, associated with plant roots.

In dying, plants also provide organic matter to which dissolved metals—and dissolved pesticides and other anthropogenic organic—sorb. Perhaps the most important benefit is I/ET, which will be discussed further in part two of this article. Two studies suggest that plant species selection affects performance, as has been found to be the case with wetland plants, contrasting tall sedge (Carex appressa) and lavender tea tree (Melaleuca ericifolia). Examples of tolerant species include red maple, black ash, green ash, eastern larch, black willow, and bald cypress. However, lists are not consistent. The designer should check with the local extension service.

Author's Bio: Gary R. Minton, Ph.D., P.E., is an independent consultant on stormwater treatment with Resource Planning Associates. He is the author of the book Stormwater Treatment: Biological, Chemical, and Engineering Principles.