Revisiting Design Criteria for Stormwater Treatment Systems, Part 4
Removal of dissolved pollutants by wet basins
It has been 25 years since the first community in the United States established the requirement for the postdevelopment treatment of stormwater from new developments. Since then, many manuals and handbooks have been published by local, regional, and state governments identifying acceptable treatment systems and design criteria. Initially, few of these design criteria were supported by laboratory or field research. With numerous studies completed over the past two decades, it is timely to reexamine some of these criteria.
The first three articles (Part One, Part Two, Part Three) in this series focused on the removal of settleable sediment with attached pollutants by three systems: basins, fine-media filters, and flow-through swales. The final two articles consider how design criteria should differ when the removal of dissolved pollutants is also an objective. This article considers wet basins, and the final article considers fine-media filters and flow-through swales.
Background
Pollutants with commonly 50% or more as dissolved include zinc, copper, cadmium, and nitrogen (Minton 2002, Pitt et al. 2004). Phosphorus is generally less, 25% to 50%. While most pesticides and hydrocarbons are substantially attached to settleable solids (sediment), the dissolved concentrations may still be toxic. Some pesticides and herbicides such as atrazine are very soluble. Table 1 presents median values for national data (Pitt et al. 2004). The ranges reflect differences between land uses. Generally, zinc represents 75% of the metals.
Separation into dissolved and particulate fractions is a physical, not a chemical, definition. Dissolved fractions are the constituents that pass through the laboratory filter, typically 0.45 microns. Small particles, or colloids, pass through the filter with pollutants attached. In wet basins, soluble pollutants are removed by sorption/precipitation in the soil and by plant uptake. Colloids do not settle well, if at all, and are essentially unfilterable. However, they attach to plant surfaces and may coagulate and settle naturally given time. For this article we make this distinction between dissolved and soluble constituents: The former pass through a laboratory filter, and the latter are the free ions or ionic complexes of the pollutant.
The many forms of each pollutant complicate their removal. Dissolved nitrogen is present as nitrate, ammonia, and organic-bound ammonia, but primarily the last two. Dissolved phosphorus is present as orthophosphate, free or bound in organic matter. Metals are present in ionic complexes and sorbed to soluble humic organics.
Our focus is how design criteria differ for surface wet basins: ponds and constructed wetlands. Minimal removal of dissolved pollutants is expected in extended detention basins and wet vaults, the former because the stormwater leaves the facility too quickly, the latter because of the lack of soils and biota. The presumed benefit of wetponds and constructed wetlands is the removal of dissolved pollutants. There are potentially adverse effects of wet basins: greater space, mosquitoes, thermal enrichment, and invasive species. This suggests that wet basins be selected and sized with caution and that design criteria recognize potential adverse impacts.
Configurations
Manuals contain many configurations, divided into wetponds and wetlands. In general, the surface area of a wetpond is 75% or more deep open water. A wetland is the reverse with 15% to 25% deep open water. There are three general schemes of plant coverage: along the lateral edges (fringe) with open water between inlet and outlet (a pond); complete coverage (a shallow wetland), perhaps with open water forebays and afterbays; and alternating transverse bands of deep open water and shallow vegetated benches (pond/wetland). There are also “treatment train” concepts such as a wetpond followed by a shallow wetland.
Shallow wet basins placed on soil are surface flow wetlands; the stormwater passes through the facility above the soil. Diffusion of the pollutants within saturated soil is very slow. To enhance the water’s contact with plant roots, gravel is used in lieu of soil. At the facility inlet, the stormwater is directed downward and through the gravel bed. The configuration is called subsurface flow (SSF). This system has additional advantages. The invert is placed a few inches below the top of the gravel to remove mosquito habitat. SSF wetlands may be attractive in regions with cold winters, avoiding the effects of freezing, and in semi-arid areas, reducing evaporation. Gravel provides a firm work area to harvest foliage. The bed is shallow, 2 to 3 feet (0.6 to 1 meter). Improving the water’s direct contact with the roots and gravel might reduce the residence time, and therefore basin volume may be less. However, the media offsets the reduction in liquid volume. An additional concern is clogging of the media over time by incoming sediment.
Each of the configurations described above may have either a free-flowing or a restricted outlet to create an extended detention (ED) layer above the permanent pool. Whether dividing the treatment volume into a wet pool and an ED layer benefits performance has not been definitively established. The ED layer is of value in configurations with fringe vegetation, given its adverse effect on hydraulic efficiency (Wong et al. 1999). The ED layer may improve hydraulic efficiency by forcing stormwater into thickly vegetated areas and into the corners of the basin. The hydraulic benefit may be most relevant in climatic regions with high-intensity rainfall.
Conversely, the disadvantage of the ED layer is that much less stormwater is retained between storms. Given the slowest of the processes of dissolved pollutant removal, retaining stormwater between storms is important. The more stormwater retained between storms, the greater the load reduction of dissolved pollutants. Some recommend the basin volume be primarily extended detention with less than 20% as wet pool (Wong et al. 1999). The objective is periodic drying with enhanced aerobic degradation of organic matter, believed necessary for long-term performance, particularly for phosphorus. However, this necessitates a substantial drawdown time of about 200 hours (Wong et al. 1999), presumably to give a residence time similar to wet basins.
Whether a particular configuration has a clear advantage is unknown. A constructed wetland is perceived to be better than a wetpond because of the greater vegetation coverage, but the validity of this perception has not been definitively established. Wetlands—fully vegetated or with alternating deep/shallow transverse bands—are likely more effective than ponds at removing suspended solids during storms and colloidal material between storms (Wong et al. 1999, Breen and Wong 2000). This may be the most appropriate configuration if removing dissolved nitrogen is the objective (Stearman et al. 2003). In contrast, a fully vegetated system may be necessary for maximum phosphorus removal. However, whether wetlands are more effective than wetponds at removing dissolved pollutants has not been established. The marginal benefit, if it exists, may not be justified, particularly considering the potentially greater adverse effects previously listed.
Removal Processes
It is commonly believed that plant growth is the primary removal process. This may be true initially when plant infilling occurs. However, research with stormwater and wastewater wet basins shows that most of the pollutants accumulate in the soil or are transformed and removed by bacterial processes. What plants remove is located primarily in their roots. Plants provide indirect benefits as described below.
The biological and chemical processes of a wet basin are illustrated in Figure 1 (Minton 2002), which indicates there are many processes other than uptake by plants that may be important removers of pollutants. Note in Figure 1 the thin surface layer of aerobic soil. In natural wetlands the thickness may be as much as an inch (25 millimeters), though its thickness in stormwater basins is unknown. It may decrease as the facility matures with the accumulation of organic matter. The thin aerobic layer is very important to the removal of soluble phosphorus and the transformation of organic nitrogen and ammonia to nitrate. Beneath this layer the soil is anaerobic because the consumption of dissolved oxygen by bacteria is more rapid than diffusion from above.
Metals sorb to humic organics and aluminum/ferric iron/manganese oxides in the upper aerobic layer. Metals also diffuse lower into the soil where they complex with sulfide produced by sulfate reducing bacteria. Pesticides sorb to clays and humic organics. Bacteria in the soil are likely dominant removers of nitrogen. Bacteria change organic nitrogen to ammonia (ammonification) and ammonia to nitrate (nitrification) under aerobic conditions. Some ammonia is lost directly to the atmosphere by volatilization, but this is significant only if the pH is high, which can occur during summer afternoons with algal activity. Other bacteria change the nitrate to nitrogen gas (denitrification) but only in the presence of low dissolved oxygen. Anaerobic conditions enhance the removal of some pesticides and toxic organics (Kao, Wang, and Wu 2001; Stearman et al. 2003).
In treatment wetlands, the thin aerobic surface layer may be more dynamic than in natural wetlands, frequently becoming anaerobic. During the summer high water temperatures increase the rate of degradation of organic matter by bacteria. The water becomes anaerobic (Huneycutt 2002) particularly in deeper wetponds with thermal stratification. Soluble phosphorus and metals previously removed are released to the overlying water through the dissolution of ferric iron phosphate (McKee and McKevlin 1993) and ferric-manganese-metal complexes (Reddy and Patrick 1983), respectively. Contributing to this dynamic is algal growth stimulated by nitrogen and phosphorus from the stormwater. During the day algal growth produces dissolved oxygen. The pH is driven upwards from the consumption of carbon dioxide, beneficial to calcium phosphate precipitation, and ammonia volatilization. However, at night the algae respire, consuming the dissolved oxygen. The result is diurnal swing at the soil surface: aerobic-anaerobic-aerobic. As a consequence, the performance of stormwater basins may not be maximized (Diab, Kochba, and Avnimelech 1993; Herskowitz, Black, and Lewandowsk 1987).
Plants use what we consider to be pollutants as nutrients for growth. Their concentration in plant biomass responds directly to the concentration in the water: The higher the concentration of the nutrient in the water, the greater the concentration in the biomass, to a point that is likely above the concentrations of nutrients observed in stormwater (Reddy and DeBusk 1987). The relationship varies significantly depending on plant species, plant health, and the season. Regardless, research consistently shows that the bulk of removed metals are found in the soil and that most nitrogen removal is by bacteria (Kadlec and Knight 1996, Minton 2002). However, the proportion removed by direct plant uptake and soil processes depends on the chemistry of the soil and the maturity of the basin.
The primary benefit of rooted plants is indirect. When they die, plants provide organic matter to which pollutants—in particular metals, pesticides, and hydrocarbons—sorb. Bacteria that nitrify and denitrify require carbon from decaying organic matter for growth. Plants pump dissolved oxygen from their foliage downward to create an aerobic zone around their roots, increasing the volume of aerobic soil. The effect varies with the species, plant health, and season (Urbanc-Bercic and Gaberscik 1999). This process benefits the formation of nitrate, the sorption/precipitation of phosphorus, and the sorption of metals to aluminum/ferric iron/manganese oxide complexes. Plants increase the removal of pesticides by supporting increased densities of bacteria (Thullen, Sartoris, and Walton 2002).
Dissolved pollutant removal occurs in the overlying water. Free-floating algae consume pollutants as nutrients, die, and sink to the bottom. However, their role may be counterproductive, because the bulk of the algae may exit the basin during storms, taking pollutants that would otherwise have diffused into the soil, the preferable route. Algae’s most important role may be at the surface of plant foliage. A biofilm of bacteria and algae grows on the plant foliage, consuming nutrients; sub-micron colloids attach to this film (Lawrence and Breen 1998). Plant surfaces have a much higher density of nitrifying bacteria than the water. This may be particularly important if the diurnal swing of the thin aerobic soil layer previously described occurs (Davido and Conway 1989).
It is commonly stated that the wet pool must not be allowed to disappear between storms through infiltration and/or evaporation. Loss is prevented with a liner and a drainage area of sufficient size to provide base flow. As a consequence, wet basins are discouraged for small drainage areas and in semi-arid climates. Temporary exposure of the normally anaerobic soil to dissolved oxygen significantly increases the degradation of accumulated organic matter, releasing pollutants. A portion of the sorbed metals and phosphorus may be transformed to forms more susceptible to desorption (Olila, Reddy, and Stites 1997; Saeki, Okazaki, and Matsumoto 1993). It is perhaps feared that the released pollutants are lost in subsequent storms. However, they likely migrate deeper in the soil where permanent sequestering occurs (Urbanc-Bercic and Gaberscik 2001). Regardless, it has not been established that periodic drying reduces performance. The volume of a wet basin is substantially greater than the volume of most storms. The first storm following a dry period partially fills the basin, providing time for released pollutants to be re-sequestered. As noted previously, some recommend that the basin be designed to dry frequently. A constant base flow can be detrimental. What is removed during storms may be lost in the base flow from desorption and/or plant litter (Oberts 1997). There is no evidence to suggest that wet basins should be discouraged for small drainage areas and in semi-arid areas.
A wetland/wetpond system is complex. Only a small portion of the pollutants in the effluent of the system during each storm is from the influent of that storm. Most has moved through the soil and plants in a continual process of uptake and release, escaping the facility with each storm and in the base flow if present. The dynamic turnover also suggests that plants establish background concentrations below which further reduction cannot be achieved. Experience from wastewater treatment suggests 1 to 2 mg/L for nitrogen (Kadlec and Knight 1996). Total phosphorus (TP) background may be 0.010 to 0.050 mg/L (Richardson and Qian 1999). A survey of 13 wet basins found a mean concentration of TP of 0.080 mg/L (Carr and Kehoe 1997). The background concentrations are a reflection of plant activity and the resistance of the dissolved organically bound pollutants to degradation (Newman and Lynch 2001).
Alternative Design Criteria
Stormwater manuals do not distinguish the removal of total suspended solids (TSS) from the removal of dissolved pollutants. Furthermore, manuals specify the volume of wetponds and wetlands the same as the volume for extended detention basins. In the first article of this series it was established that a wet basin can be considerably smaller yet meet the typical TSS performance goal of 80%.
As the bulk of the sediment is removed during each storm, performance is best related to the ratio of the volume of the basin to the runoff volume of the mean storm (Vb/Vr) (USEPA 1986). This approach is appropriate for wet basins where the sole objective is the removal of TSS and attached pollutants. In contrast, it is likely the removal of dissolved pollutants occurs primarily between storms. Therefore, the volume of the basin is best related to the annual runoff depth. Sizing will likely differ for each pollutant type: phosphorus, nitrogen, and metals. If the management goal is the general removal of dissolved pollutants, the sizing criterion that gives the largest basin takes precedence.
Several alternatives for a basic sizing criterion are discussed below: storm runoff depth, hydraulic residence time, hydraulic loading rate, pollutant loading rate, kinetic removal rate, and process model.
Storm Runoff Depth: This design criterion is currently specified in most manuals to size both extended detention basins and wet basins. A common design depth is 1 inch to treat 90% of stormwater generated over time. The volume, V, is determined by V = RaRcA where Ra is the runoff design depth, Rc a runoff coefficient to account for pervious surfaces, and A the drainage area. Different depths can be specified for different objectives: TSS or dissolved. However, as noted in the first article, the concept is not relevant to wet basins. A design criterion that more explicitly relates basin size to dissolved pollutant removal is perhaps more appropriate.
Hydraulic Residence Time (HRT): Soil and plant processes are slow, as is diffusion in saturated soils, suggesting the importance of time. Larger basins with an HRT of 14 days have been recommended for improved phosphorus removal (Hartigan 1988). The volume of the basin is determined by V = RaRcAHRT/365 where Ra is the annual runoff depth, Rc the runoff coefficient, A the drainage area, and HRT the hydraulic residence time in days. HRT is not the residence time during each storm . It represents the average time the water resides in the basin, including the time between storms. HRT has been referred to as the average annual residence time.
Annual rainfall runoff in areas of interest ranges from 10 to 60 inches (25 to 150 centimeters) (USEPA 1989). With an HRT of 14 days, the unit volume (V/A) ranges from 0.4 to 2.4 inches (1 to 6 centimeters). This provides volume ratios (Vb/Vr) of about 0.7 to 3. In the first article, a Vb/Vr of 1 was suggested as adequate for the removal of TSS. In semi-arid areas with low annual rainfall, like the Southwest and the Rocky Mountain states, the HRT for a Vb/Vr of 1 is about equal to or substantially greater than 14 days. In contrast, the HRT is less than seven days in wetter regions. In wetter regions the basin must be larger, by a factor of two or three, for an HRT of 14 days. The HRT as represented here is the nominal HRT—that is, the basin volume divided by the average inflow rate. The actual HRT is less, reflecting hydraulic inefficiency. How much less depends on the configuration.
Hydraulic Loading Rate (HLR): If soil and rooted plants are central to pollutant removal, the surface area of the basin should be explicitly recognized. This is accomplished by considering the areal loading rate. There are two options: HLR and pollutant loading rate (PLR). The first is discussed here. The facility surface area, Af, is determined by Af = RaRcA / 365HLR where Rais the annual runoff depth, Rcthe runoff coefficient, Athe drainage area, and HLRthe hydraulic loading rate in centimeters per day. HLR does not refer to the loading rate during a storm but rather the average rate over a year, or the average annual hydraulic loading rate. The numerator represents the total annual volume of stormwater entering the facility. The design method implies the selection of a depth that gives a residence time, HRT, for satisfactory performance.
Limited studies (Abtew et al. 2004, Bulc and Slak 2003, Newman and Lynch 2001) of stormwater wetlands suggest satisfactory nutrient removal at HLR values of 1 to 10 cm/day, which is similar to the more substantial experience with wastewater treatment (Kadlec and Knight 1996). A wetland with an HLR of 2 cm/day with an average depth of 1 foot (0.3 meters) has an HRT of about 14 days. At these design values, the surface area of the wet basin varies widely across the United States, from about 100 to 800 square feet per impervious acre (approximately 22 to 175 m2/ha), the surface area increasing with annual rainfall. For a wetpond with an average depth of 5 feet (1.5 meters), the HLR is 11 cm/day for an HRT of about 14 days. The surface area is proportionally less.
Pollutant Loading Rate: Using the same value for any of the design criteria introduced previously implies a constant median pollutant concentration. However, the concentration differs between climatic regions and between land uses within each region (Pitt et al. 2004). The alternative design criteria introduced to this point obscures what is perhaps the more relevant consideration, the pollutant loading per unit surface area or PLR. The facility surface area, Af , is determined by Af = RaRcAPm/PLR where Ra is the annual runoff depth, Rc the runoff coefficient, A the drainage area, Pm the median influent pollutant concentration, and PLR . The numerator represents the total annual pollutant load to the facility. The design criterion implies the selection of a depth that gives a residence time, or HRT, for satisfactory performance. Like HRT, the PLR and HLR presented here are nominal criteria. The actual values differ from the nominal to the extent of hydraulic inefficiency.
The PLR differs with the pollutant: nitrogen, phosphorus, and metals (in the aggregate) given their different median concentrations. Performance studies will likely show the unit area needed for each of the pollutants is the same. Not clear at this point is whether the PLR should be based on the total or the dissolved concentration. A portion of the particulate form solubilizes after entering the basin. Little is known about the potential for solubilization, but there are methods available for analysis (Minton 2002).
PLR allows consideration of a question not yet discussed: short- versus long-term performance. It is likely that wet basins remove nitrogen and metals over the long term, improving as the facility matures. This does not appear to be true for phosphorus. The chemical sorption capacity is reached; how soon depends on the soil chemistry. There is no evidence that as iron and aluminum in the influent accumulate in the basin they factor in phosphorus removal. Absent harvesting of vegetation, the facility must be sized based on the production of plant debris resistant to long-term degradation. Extensive experience (Richardson and Qian 1999) with wastewater wetlands suggests a sequestering rate of 0.25 to 1 g/m2/yr, with a frequently suggested value of 0.5.
Limited studies of stormwater wetlands seem consistent with wastewater experience (Abtew et al. 2004, Moustafa et al. 1996). One study found 90% removal at a PLR of about 0.50 g/m2/yr, decreasing to 63% removal at about 4.2 (Abtew et al. 2004). Mean annual outflow concentrations ranged from 25 to 53 _g/L (micrograms per liter). The system continued to remove phosphorus after eight years. In contrast, another pond-wetland facility experienced loadings of about 5 g/m2/yr. It removed 85% of the TP during its first year but none when tested 10 years later (Oberts 1997). While removal during storms continued, phosphorus was lost in the base flow.
Consider these assumptions: PLR of 0.5 g/m2/yr; medium concentrations of TP by climatic region (Pitt et al. 2004); and 50% of TP bioavailable (dissolved plus desorption). The surface area of a wet basin ranges from about 500 to 3,000 square feet per impervious acre (about 110 to 650 m2/ha), about 1% to 6% of the drainage area depending on the climatic region. The HRT for an average basin depth of 2 feet (0.6 meters) is about four to 16 days. Perhaps sorptive media can be added to soil or gravel to allow reduction in surface area.
It may be feasible to decrease the surface area if harvesting occurs regularly. But this decreases the HRT if depth is held constant. Perhaps the HRT can be less than 14 days. It may be that the greater the surface area in proportion to the volume, the lower the HRT needed to achieve the desired performance. Phosphorus in foliage is 0.1 to 0.4 g/m2 in natural wetlands (Kadlec and Knight 1996) but likely higher in stormwater wetlands (Reddy and DeBusk 1987). The effectiveness and appropriate harvest methods require further study as to their effect on short-term performance (Gu et al. 2001, Urbanc-Bercic and Gaberscik 1999).
Loss of long-term performance is not likely with nitrogen and metals. For these pollutants, a wetpond with transverse shallow vegetated benches may perform as well, requiring less space than a shallow wetland at the same HRT. It is possible that a wet basin sized for TSS removal need not be larger to obtain the maximum practical removal of nitrogen or metals, depending on the climatic region. With a lower limit of 1 to 2 mg/L, modest removal for nitrogen is expected (England 2001). Consequently, a shallow marsh wetland (low PLR) may gain little and may decrease performance. Based on the experience of wastewater wetlands (Kadlec and Knight 1996) and recognizing the lower concentrations in stormwater, PLR values for nitrogen of 3 to 4 g/m2/yr may be appropriate. However, because of the effect of plants on the background concentration, higher loadings are likely possible without degradation of effluent quality.
Kinetic Removal Rate: Pollutants are removed at a kinetic removal, or reaction, rate in a wet basin. The concentration decreases exponentially as the water passes through the basin. Reduction is greatest in the entry area of the basin, gradually decreasing toward the outlet. Kinetic equations are commonly used in the sizing of wastewater wetlands. The simplest form of a kinetic equation is Ce = Cie-kHRT where Ceis the effluent concentration, Cithe influent concentration, k the kinetic removal rate coefficient, and HRTthe hydraulic residence time. The equation assumes a first-order reaction rate; that is, for the same value of kHRT, the removal efficiency is the same irrespective of the influent concentration. Research suggests that kis dependent on influent concentration (Kadlec 2000). Therefore, values from wastewater treatment are not applicable. A study of experimental stormwater wetlands found kfor phosphorus to vary by a factor of six (Bays et al. 2001). The factors affecting the value included presence/absence of vegetation, water depth, type of soil, HLR, and PLR. No relationships between the factors were developed.
More complex equations than the one above include influent concentration, HRT, background concentration, HLR, and/or kinetic rates other than first order. The approach is complicated further by the multiple forms of nitrogen and phosphorus that have different kinetic removal rates. The concept may not be useful for either nitrogen or metals. As noted previously, influent and background concentrations are similar. The limitation with metals is their low influent concentrations, except for zinc.
Process Model: All previous design criteria are empirical “black box” constructs within which the multiple processes occur. These can be called implicit models. Alternatively, the major processes are explicitly described mathematically (Fletcher 2004; Lawrence and Breen 1998; Thullen, Sartoris, and Walton 2002). Using appropriate input coefficients for each process, the system is sized as a function of desired performance.
One model includes the removal of all pollutant forms: particulate, colloidal, and soluble, by sedimentation, attachment, sorption/precipitation, and plant and biofilm uptake (Lawrence and Breen 1998). Presented in Figures 2 and 3 are example outputs for settleable solids and TP, respectively. Three sizes of sediments are shown: Fine represents the colloidal. While TP is shown, removal of soluble and colloidal fractions is simulated explicitly. The figures indicate an HRT less than 15 days may be sufficient. The process model has the advantage of showing key relationships explicitly, important to understanding the effect of size on performance.
A concern with process models is the many decisions to be made on input coefficients for the particular climatic region. Studies by region are sparse or nonexistent. Studies to date of both stormwater and wastewater suggest substantial variation in the coefficients, reflective of a complex chemical-biological ecosystem. But this observation is valid for all alternative design criteria introduced.
Final Observations
Regardless of the design criterion selected, arriving at the appropriate design value is a daunting task. The processes of wet basins are complex, particularly those of wetlands. Drawing general conclusions from specific studies is fraught with uncertainty. Studies must be comprehensive, structured to consider all processes and the condition of the soil before saturation. Evaluations must occur over an extended period of time as the test facility matures. The ability to draw firm conclusions from field or laboratory studies is complicated by the differing forms of the pollutants of interest and the low influent concentrations near the detection limit. Few existing published studies provide the necessary information to calculate values for the alternative design criteria discussed in this article. It is hoped this discussion will spur researches to include the necessary information in their reports.
Bigger is not necessarily better. As suggested by Figures 2 and 3, a point of diminishing returns exists at which the incremental increase in the size of a basin does not provide a demonstrable increase in performance. The gradual slopes in Figures 2 and 3 indicate that the point is not obvious. Regardless, the design coefficients should be based on incremental cost versus incremental performance, not incremental size.
Bigger can be worse. A basin surface area in excess of what is needed for effective removal of dissolved pollutants might result in the increase of relevant pollutants in the effluent. There is the possibility of too great a mass of emergent plants with loss of the pollutant bound to soluble and particulate organic matter emanating from the plants. Annual cleaning of the forebay may be appropriate to decrease solubilization of pollutants from imported organic matter, thereby reducing loading on the wet basin.
Given the importance of the soil and the effect of its variation in chemistry on performance (Hsieh and Coultas 1989, Newman and Pietro 2001), soils should be tested prior to flooding. Agricultural lands heavily fertilized may be saturated with phosphorus or nitrogen (Pant et al. 2002). Phosphorus and metals original to the soil may be lost as previously described upon basin filling. As a consequence, the wet basin may be a net source rather than a sink. Laboratory methods exist that define susceptibility to release (Minton 2002). Testing may indicate corrective actions to be taken in the design.
Summary
When considering setting goals for the removal of phosphorus, nitrogen, or metals, the importance of the dissolved fraction should be recognized. The sizing criterion should be distinct from that used to size wet basins for the removal of TSS with associated particulate pollutants.
A shallow marsh wetland is likely the most appropriate wet basin configuration for dissolved phosphorus. Whether this configuration is appropriate for dissolved nitrogen and metals has not been demonstrated; a wetpond primarily of deep water with alternating transverse bands of shallow vegetated benches might be more appropriate. It requires less space, and a lower HRT than for phosphorus may be feasible.
It is has not been definitively demonstrated that splitting the basin volume into permanent wet pool with extended detention is appropriate. To improve hydraulic efficiency, it is likely appropriate for a shallow marsh but not for the suggested wetpond configuration.
The relative sizes of the three basin types—extended detention, wetpond, and wetland—will differ with the selected design criteria and climate. Enhanced treatment wet basins as proposed here may be similar in volume to extended detention basins as currently sized but likely with larger surface areas. Except in semi-arid regions, wet basins for the removal of dissolved pollutants will require larger volumes, larger surface areas, or both than if the sole focus is 80% removal of TSS.
Advertisement
Process models should be used to structure pilot and full-scale field studies. With criteria specific to different climatic regions, regulators can use process models to conduct rational cost-benefit analyses. Ideally, the understanding of such models as well as the alternative design criteria discussed in this article will result in the inclusion of all necessary information in performance studies. This is essential if we are to understand the relationship between each alternative criterion and performance.
It is proposed that pollutant loading rate combined with an appropriate minimum hydraulic residence time be used as the basis for sizing wet basins. The other design criteria may do as well, but the suggested combination provides a more explicit view of the key processes except for a process model. With time, laboratory and field studies will provide definitive design values for each pollutant type, although the value of 0.5 g/m2/yr is likely appropriate for phosphorus. We will also likely find that the cost-effective unit volume and unit area requirements differ with each pollutant. Given the complexity of the processes in wet basins, particularly wetlands, professional judgment will always be central to the selection of design values.
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.
May-June 2005
Revisiting Design Criteria for Stormwater Treatment Systems, Part 4
Removal of dissolved pollutants by wet basins
It has been 25 years since the first community in the United States established the requirement for the postdevelopment treatment of stormwater from new developments. Since then, many manuals and handbooks have been published by local, regional, and state governments identifying acceptable treatment systems and design criteria. Initially, few of these design criteria were supported by laboratory or field research. With numerous studies completed over the past two decades, it is timely to reexamine some of these criteria.
The first three articles (Part One, Part Two, Part Three) in this series focused on the removal of settleable sediment with attached pollutants by three systems: basins, fine-media filters, and flow-through swales. The final two articles consider how design criteria should differ when the removal of dissolved pollutants is also an objective. This article considers wet basins, and the final article considers fine-media filters and flow-through swales.
Background
Pollutants with commonly 50% or more as dissolved include zinc, copper, cadmium, and nitrogen (Minton 2002, Pitt et al. 2004). Phosphorus is generally less, 25% to 50%. While most pesticides and hydrocarbons are substantially attached to settleable solids (sediment), the dissolved concentrations may still be toxic. Some pesticides and herbicides such as atrazine are very soluble. Table 1 presents median values for national data (Pitt et al. 2004). The ranges reflect differences between land uses. Generally, zinc represents 75% of the metals.
Separation into dissolved and particulate fractions is a physical, not a chemical, definition. Dissolved fractions are the constituents that pass through the laboratory filter, typically 0.45 microns. Small particles, or colloids, pass through the filter with pollutants attached. In wet basins, soluble pollutants are removed by sorption/precipitation in the soil and by plant uptake. Colloids do not settle well, if at all, and are essentially unfilterable. However, they attach to plant surfaces and may coagulate and settle naturally given time. For this article we make this distinction between dissolved and soluble constituents: The former pass through a laboratory filter, and the latter are the free ions or ionic complexes of the pollutant.
The many forms of each pollutant complicate their removal. Dissolved nitrogen is present as nitrate, ammonia, and organic-bound ammonia, but primarily the last two. Dissolved phosphorus is present as orthophosphate, free or bound in organic matter. Metals are present in ionic complexes and sorbed to soluble humic organics.
Our focus is how design criteria differ for surface wet basins: ponds and constructed wetlands. Minimal removal of dissolved pollutants is expected in extended detention basins and wet vaults, the former because the stormwater leaves the facility too quickly, the latter because of the lack of soils and biota. The presumed benefit of wetponds and constructed wetlands is the removal of dissolved pollutants. There are potentially adverse effects of wet basins: greater space, mosquitoes, thermal enrichment, and invasive species. This suggests that wet basins be selected and sized with caution and that design criteria recognize potential adverse impacts.
Configurations
Manuals contain many configurations, divided into wetponds and wetlands. In general, the surface area of a wetpond is 75% or more deep open water. A wetland is the reverse with 15% to 25% deep open water. There are three general schemes of plant coverage: along the lateral edges (fringe) with open water between inlet and outlet (a pond); complete coverage (a shallow wetland), perhaps with open water forebays and afterbays; and alternating transverse bands of deep open water and shallow vegetated benches (pond/wetland). There are also “treatment train” concepts such as a wetpond followed by a shallow wetland.
Shallow wet basins placed on soil are surface flow wetlands; the stormwater passes through the facility above the soil. Diffusion of the pollutants within saturated soil is very slow. To enhance the water’s contact with plant roots, gravel is used in lieu of soil. At the facility inlet, the stormwater is directed downward and through the gravel bed. The configuration is called subsurface flow (SSF). This system has additional advantages. The invert is placed a few inches below the top of the gravel to remove mosquito habitat. SSF wetlands may be attractive in regions with cold winters, avoiding the effects of freezing, and in semi-arid areas, reducing evaporation. Gravel provides a firm work area to harvest foliage. The bed is shallow, 2 to 3 feet (0.6 to 1 meter). Improving the water’s direct contact with the roots and gravel might reduce the residence time, and therefore basin volume may be less. However, the media offsets the reduction in liquid volume. An additional concern is clogging of the media over time by incoming sediment.
Each of the configurations described above may have either a free-flowing or a restricted outlet to create an extended detention (ED) layer above the permanent pool. Whether dividing the treatment volume into a wet pool and an ED layer benefits performance has not been definitively established. The ED layer is of value in configurations with fringe vegetation, given its adverse effect on hydraulic efficiency (Wong et al. 1999). The ED layer may improve hydraulic efficiency by forcing stormwater into thickly vegetated areas and into the corners of the basin. The hydraulic benefit may be most relevant in climatic regions with high-intensity rainfall.
Conversely, the disadvantage of the ED layer is that much less stormwater is retained between storms. Given the slowest of the processes of dissolved pollutant removal, retaining stormwater between storms is important. The more stormwater retained between storms, the greater the load reduction of dissolved pollutants. Some recommend the basin volume be primarily extended detention with less than 20% as wet pool (Wong et al. 1999). The objective is periodic drying with enhanced aerobic degradation of organic matter, believed necessary for long-term performance, particularly for phosphorus. However, this necessitates a substantial drawdown time of about 200 hours (Wong et al. 1999), presumably to give a residence time similar to wet basins.
Whether a particular configuration has a clear advantage is unknown. A constructed wetland is perceived to be better than a wetpond because of the greater vegetation coverage, but the validity of this perception has not been definitively established. Wetlands—fully vegetated or with alternating deep/shallow transverse bands—are likely more effective than ponds at removing suspended solids during storms and colloidal material between storms (Wong et al. 1999, Breen and Wong 2000). This may be the most appropriate configuration if removing dissolved nitrogen is the objective (Stearman et al. 2003). In contrast, a fully vegetated system may be necessary for maximum phosphorus removal. However, whether wetlands are more effective than wetponds at removing dissolved pollutants has not been established. The marginal benefit, if it exists, may not be justified, particularly considering the potentially greater adverse effects previously listed.
Removal Processes
It is commonly believed that plant growth is the primary removal process. This may be true initially when plant infilling occurs. However, research with stormwater and wastewater wet basins shows that most of the pollutants accumulate in the soil or are transformed and removed by bacterial processes. What plants remove is located primarily in their roots. Plants provide indirect benefits as described below.
The biological and chemical processes of a wet basin are illustrated in Figure 1 (Minton 2002), which indicates there are many processes other than uptake by plants that may be important removers of pollutants. Note in Figure 1 the thin surface layer of aerobic soil. In natural wetlands the thickness may be as much as an inch (25 millimeters), though its thickness in stormwater basins is unknown. It may decrease as the facility matures with the accumulation of organic matter. The thin aerobic layer is very important to the removal of soluble phosphorus and the transformation of organic nitrogen and ammonia to nitrate. Beneath this layer the soil is anaerobic because the consumption of dissolved oxygen by bacteria is more rapid than diffusion from above.
Metals sorb to humic organics and aluminum/ferric iron/manganese oxides in the upper aerobic layer. Metals also diffuse lower into the soil where they complex with sulfide produced by sulfate reducing bacteria. Pesticides sorb to clays and humic organics. Bacteria in the soil are likely dominant removers of nitrogen. Bacteria change organic nitrogen to ammonia (ammonification) and ammonia to nitrate (nitrification) under aerobic conditions. Some ammonia is lost directly to the atmosphere by volatilization, but this is significant only if the pH is high, which can occur during summer afternoons with algal activity. Other bacteria change the nitrate to nitrogen gas (denitrification) but only in the presence of low dissolved oxygen. Anaerobic conditions enhance the removal of some pesticides and toxic organics (Kao, Wang, and Wu 2001; Stearman et al. 2003).
In treatment wetlands, the thin aerobic surface layer may be more dynamic than in natural wetlands, frequently becoming anaerobic. During the summer high water temperatures increase the rate of degradation of organic matter by bacteria. The water becomes anaerobic (Huneycutt 2002) particularly in deeper wetponds with thermal stratification. Soluble phosphorus and metals previously removed are released to the overlying water through the dissolution of ferric iron phosphate (McKee and McKevlin 1993) and ferric-manganese-metal complexes (Reddy and Patrick 1983), respectively. Contributing to this dynamic is algal growth stimulated by nitrogen and phosphorus from the stormwater. During the day algal growth produces dissolved oxygen. The pH is driven upwards from the consumption of carbon dioxide, beneficial to calcium phosphate precipitation, and ammonia volatilization. However, at night the algae respire, consuming the dissolved oxygen. The result is diurnal swing at the soil surface: aerobic-anaerobic-aerobic. As a consequence, the performance of stormwater basins may not be maximized (Diab, Kochba, and Avnimelech 1993; Herskowitz, Black, and Lewandowsk 1987).
Plants use what we consider to be pollutants as nutrients for growth. Their concentration in plant biomass responds directly to the concentration in the water: The higher the concentration of the nutrient in the water, the greater the concentration in the biomass, to a point that is likely above the concentrations of nutrients observed in stormwater (Reddy and DeBusk 1987). The relationship varies significantly depending on plant species, plant health, and the season. Regardless, research consistently shows that the bulk of removed metals are found in the soil and that most nitrogen removal is by bacteria (Kadlec and Knight 1996, Minton 2002). However, the proportion removed by direct plant uptake and soil processes depends on the chemistry of the soil and the maturity of the basin.
The primary benefit of rooted plants is indirect. When they die, plants provide organic matter to which pollutants—in particular metals, pesticides, and hydrocarbons—sorb. Bacteria that nitrify and denitrify require carbon from decaying organic matter for growth. Plants pump dissolved oxygen from their foliage downward to create an aerobic zone around their roots, increasing the volume of aerobic soil. The effect varies with the species, plant health, and season (Urbanc-Bercic and Gaberscik 1999). This process benefits the formation of nitrate, the sorption/precipitation of phosphorus, and the sorption of metals to aluminum/ferric iron/manganese oxide complexes. Plants increase the removal of pesticides by supporting increased densities of bacteria (Thullen, Sartoris, and Walton 2002).
Dissolved pollutant removal occurs in the overlying water. Free-floating algae consume pollutants as nutrients, die, and sink to the bottom. However, their role may be counterproductive, because the bulk of the algae may exit the basin during storms, taking pollutants that would otherwise have diffused into the soil, the preferable route. Algae’s most important role may be at the surface of plant foliage. A biofilm of bacteria and algae grows on the plant foliage, consuming nutrients; sub-micron colloids attach to this film (Lawrence and Breen 1998). Plant surfaces have a much higher density of nitrifying bacteria than the water. This may be particularly important if the diurnal swing of the thin aerobic soil layer previously described occurs (Davido and Conway 1989).
It is commonly stated that the wet pool must not be allowed to disappear between storms through infiltration and/or evaporation. Loss is prevented with a liner and a drainage area of sufficient size to provide base flow. As a consequence, wet basins are discouraged for small drainage areas and in semi-arid climates. Temporary exposure of the normally anaerobic soil to dissolved oxygen significantly increases the degradation of accumulated organic matter, releasing pollutants. A portion of the sorbed metals and phosphorus may be transformed to forms more susceptible to desorption (Olila, Reddy, and Stites 1997; Saeki, Okazaki, and Matsumoto 1993). It is perhaps feared that the released pollutants are lost in subsequent storms. However, they likely migrate deeper in the soil where permanent sequestering occurs (Urbanc-Bercic and Gaberscik 2001). Regardless, it has not been established that periodic drying reduces performance. The volume of a wet basin is substantially greater than the volume of most storms. The first storm following a dry period partially fills the basin, providing time for released pollutants to be re-sequestered. As noted previously, some recommend that the basin be designed to dry frequently. A constant base flow can be detrimental. What is removed during storms may be lost in the base flow from desorption and/or plant litter (Oberts 1997). There is no evidence to suggest that wet basins should be discouraged for small drainage areas and in semi-arid areas.
A wetland/wetpond system is complex. Only a small portion of the pollutants in the effluent of the system during each storm is from the influent of that storm. Most has moved through the soil and plants in a continual process of uptake and release, escaping the facility with each storm and in the base flow if present. The dynamic turnover also suggests that plants establish background concentrations below which further reduction cannot be achieved. Experience from wastewater treatment suggests 1 to 2 mg/L for nitrogen (Kadlec and Knight 1996). Total phosphorus (TP) background may be 0.010 to 0.050 mg/L (Richardson and Qian 1999). A survey of 13 wet basins found a mean concentration of TP of 0.080 mg/L (Carr and Kehoe 1997). The background concentrations are a reflection of plant activity and the resistance of the dissolved organically bound pollutants to degradation (Newman and Lynch 2001).
Alternative Design Criteria
Stormwater manuals do not distinguish the removal of total suspended solids (TSS) from the removal of dissolved pollutants. Furthermore, manuals specify the volume of wetponds and wetlands the same as the volume for extended detention basins. In the first article of this series it was established that a wet basin can be considerably smaller yet meet the typical TSS performance goal of 80%.
As the bulk of the sediment is removed during each storm, performance is best related to the ratio of the volume of the basin to the runoff volume of the mean storm (Vb/Vr) (USEPA 1986). This approach is appropriate for wet basins where the sole objective is the removal of TSS and attached pollutants. In contrast, it is likely the removal of dissolved pollutants occurs primarily between storms. Therefore, the volume of the basin is best related to the annual runoff depth. Sizing will likely differ for each pollutant type: phosphorus, nitrogen, and metals. If the management goal is the general removal of dissolved pollutants, the sizing criterion that gives the largest basin takes precedence.
Several alternatives for a basic sizing criterion are discussed below: storm runoff depth, hydraulic residence time, hydraulic loading rate, pollutant loading rate, kinetic removal rate, and process model.
Storm Runoff Depth: This design criterion is currently specified in most manuals to size both extended detention basins and wet basins. A common design depth is 1 inch to treat 90% of stormwater generated over time. The volume, V, is determined by V = RaRcA where Ra is the runoff design depth, Rc a runoff coefficient to account for pervious surfaces, and A the drainage area. Different depths can be specified for different objectives: TSS or dissolved. However, as noted in the first article, the concept is not relevant to wet basins. A design criterion that more explicitly relates basin size to dissolved pollutant removal is perhaps more appropriate.
Hydraulic Residence Time (HRT): Soil and plant processes are slow, as is diffusion in saturated soils, suggesting the importance of time. Larger basins with an HRT of 14 days have been recommended for improved phosphorus removal (Hartigan 1988). The volume of the basin is determined by V = RaRcAHRT/365 where Ra is the annual runoff depth, Rc the runoff coefficient, A the drainage area, and HRT the hydraulic residence time in days. HRT is not the residence time during each storm . It represents the average time the water resides in the basin, including the time between storms. HRT has been referred to as the average annual residence time.
Annual rainfall runoff in areas of interest ranges from 10 to 60 inches (25 to 150 centimeters) (USEPA 1989). With an HRT of 14 days, the unit volume (V/A) ranges from 0.4 to 2.4 inches (1 to 6 centimeters). This provides volume ratios (Vb/Vr) of about 0.7 to 3. In the first article, a Vb/Vr of 1 was suggested as adequate for the removal of TSS. In semi-arid areas with low annual rainfall, like the Southwest and the Rocky Mountain states, the HRT for a Vb/Vr of 1 is about equal to or substantially greater than 14 days. In contrast, the HRT is less than seven days in wetter regions. In wetter regions the basin must be larger, by a factor of two or three, for an HRT of 14 days. The HRT as represented here is the nominal HRT—that is, the basin volume divided by the average inflow rate. The actual HRT is less, reflecting hydraulic inefficiency. How much less depends on the configuration.
Hydraulic Loading Rate (HLR): If soil and rooted plants are central to pollutant removal, the surface area of the basin should be explicitly recognized. This is accomplished by considering the areal loading rate. There are two options: HLR and pollutant loading rate (PLR). The first is discussed here. The facility surface area, Af, is determined by Af = RaRcA / 365HLR where Rais the annual runoff depth, Rcthe runoff coefficient, Athe drainage area, and HLRthe hydraulic loading rate in centimeters per day. HLR does not refer to the loading rate during a storm but rather the average rate over a year, or the average annual hydraulic loading rate. The numerator represents the total annual volume of stormwater entering the facility. The design method implies the selection of a depth that gives a residence time, HRT, for satisfactory performance.
Limited studies (Abtew et al. 2004, Bulc and Slak 2003, Newman and Lynch 2001) of stormwater wetlands suggest satisfactory nutrient removal at HLR values of 1 to 10 cm/day, which is similar to the more substantial experience with wastewater treatment (Kadlec and Knight 1996). A wetland with an HLR of 2 cm/day with an average depth of 1 foot (0.3 meters) has an HRT of about 14 days. At these design values, the surface area of the wet basin varies widely across the United States, from about 100 to 800 square feet per impervious acre (approximately 22 to 175 m2/ha), the surface area increasing with annual rainfall. For a wetpond with an average depth of 5 feet (1.5 meters), the HLR is 11 cm/day for an HRT of about 14 days. The surface area is proportionally less.
Pollutant Loading Rate: Using the same value for any of the design criteria introduced previously implies a constant median pollutant concentration. However, the concentration differs between climatic regions and between land uses within each region (Pitt et al. 2004). The alternative design criteria introduced to this point obscures what is perhaps the more relevant consideration, the pollutant loading per unit surface area or PLR. The facility surface area, Af , is determined by Af = RaRcAPm/PLR where Ra is the annual runoff depth, Rc the runoff coefficient, A the drainage area, Pm the median influent pollutant concentration, and PLR . The numerator represents the total annual pollutant load to the facility. The design criterion implies the selection of a depth that gives a residence time, or HRT, for satisfactory performance. Like HRT, the PLR and HLR presented here are nominal criteria. The actual values differ from the nominal to the extent of hydraulic inefficiency.
The PLR differs with the pollutant: nitrogen, phosphorus, and metals (in the aggregate) given their different median concentrations. Performance studies will likely show the unit area needed for each of the pollutants is the same. Not clear at this point is whether the PLR should be based on the total or the dissolved concentration. A portion of the particulate form solubilizes after entering the basin. Little is known about the potential for solubilization, but there are methods available for analysis (Minton 2002).
PLR allows consideration of a question not yet discussed: short- versus long-term performance. It is likely that wet basins remove nitrogen and metals over the long term, improving as the facility matures. This does not appear to be true for phosphorus. The chemical sorption capacity is reached; how soon depends on the soil chemistry. There is no evidence that as iron and aluminum in the influent accumulate in the basin they factor in phosphorus removal. Absent harvesting of vegetation, the facility must be sized based on the production of plant debris resistant to long-term degradation. Extensive experience (Richardson and Qian 1999) with wastewater wetlands suggests a sequestering rate of 0.25 to 1 g/m2/yr, with a frequently suggested value of 0.5.
Limited studies of stormwater wetlands seem consistent with wastewater experience (Abtew et al. 2004, Moustafa et al. 1996). One study found 90% removal at a PLR of about 0.50 g/m2/yr, decreasing to 63% removal at about 4.2 (Abtew et al. 2004). Mean annual outflow concentrations ranged from 25 to 53 _g/L (micrograms per liter). The system continued to remove phosphorus after eight years. In contrast, another pond-wetland facility experienced loadings of about 5 g/m2/yr. It removed 85% of the TP during its first year but none when tested 10 years later (Oberts 1997). While removal during storms continued, phosphorus was lost in the base flow.
Consider these assumptions: PLR of 0.5 g/m2/yr; medium concentrations of TP by climatic region (Pitt et al. 2004); and 50% of TP bioavailable (dissolved plus desorption). The surface area of a wet basin ranges from about 500 to 3,000 square feet per impervious acre (about 110 to 650 m2/ha), about 1% to 6% of the drainage area depending on the climatic region. The HRT for an average basin depth of 2 feet (0.6 meters) is about four to 16 days. Perhaps sorptive media can be added to soil or gravel to allow reduction in surface area.
It may be feasible to decrease the surface area if harvesting occurs regularly. But this decreases the HRT if depth is held constant. Perhaps the HRT can be less than 14 days. It may be that the greater the surface area in proportion to the volume, the lower the HRT needed to achieve the desired performance. Phosphorus in foliage is 0.1 to 0.4 g/m2 in natural wetlands (Kadlec and Knight 1996) but likely higher in stormwater wetlands (Reddy and DeBusk 1987). The effectiveness and appropriate harvest methods require further study as to their effect on short-term performance (Gu et al. 2001, Urbanc-Bercic and Gaberscik 1999).
Loss of long-term performance is not likely with nitrogen and metals. For these pollutants, a wetpond with transverse shallow vegetated benches may perform as well, requiring less space than a shallow wetland at the same HRT. It is possible that a wet basin sized for TSS removal need not be larger to obtain the maximum practical removal of nitrogen or metals, depending on the climatic region. With a lower limit of 1 to 2 mg/L, modest removal for nitrogen is expected (England 2001). Consequently, a shallow marsh wetland (low PLR) may gain little and may decrease performance. Based on the experience of wastewater wetlands (Kadlec and Knight 1996) and recognizing the lower concentrations in stormwater, PLR values for nitrogen of 3 to 4 g/m2/yr may be appropriate. However, because of the effect of plants on the background concentration, higher loadings are likely possible without degradation of effluent quality.
Kinetic Removal Rate: Pollutants are removed at a kinetic removal, or reaction, rate in a wet basin. The concentration decreases exponentially as the water passes through the basin. Reduction is greatest in the entry area of the basin, gradually decreasing toward the outlet. Kinetic equations are commonly used in the sizing of wastewater wetlands. The simplest form of a kinetic equation is Ce = Cie-kHRT where Ceis the effluent concentration, Cithe influent concentration, k the kinetic removal rate coefficient, and HRTthe hydraulic residence time. The equation assumes a first-order reaction rate; that is, for the same value of kHRT, the removal efficiency is the same irrespective of the influent concentration. Research suggests that kis dependent on influent concentration (Kadlec 2000). Therefore, values from wastewater treatment are not applicable. A study of experimental stormwater wetlands found kfor phosphorus to vary by a factor of six (Bays et al. 2001). The factors affecting the value included presence/absence of vegetation, water depth, type of soil, HLR, and PLR. No relationships between the factors were developed.
More complex equations than the one above include influent concentration, HRT, background concentration, HLR, and/or kinetic rates other than first order. The approach is complicated further by the multiple forms of nitrogen and phosphorus that have different kinetic removal rates. The concept may not be useful for either nitrogen or metals. As noted previously, influent and background concentrations are similar. The limitation with metals is their low influent concentrations, except for zinc.
Process Model: All previous design criteria are empirical “black box” constructs within which the multiple processes occur. These can be called implicit models. Alternatively, the major processes are explicitly described mathematically (Fletcher 2004; Lawrence and Breen 1998; Thullen, Sartoris, and Walton 2002). Using appropriate input coefficients for each process, the system is sized as a function of desired performance.
One model includes the removal of all pollutant forms: particulate, colloidal, and soluble, by sedimentation, attachment, sorption/precipitation, and plant and biofilm uptake (Lawrence and Breen 1998). Presented in Figures 2 and 3 are example outputs for settleable solids and TP, respectively. Three sizes of sediments are shown: Fine represents the colloidal. While TP is shown, removal of soluble and colloidal fractions is simulated explicitly. The figures indicate an HRT less than 15 days may be sufficient. The process model has the advantage of showing key relationships explicitly, important to understanding the effect of size on performance.
A concern with process models is the many decisions to be made on input coefficients for the particular climatic region. Studies by region are sparse or nonexistent. Studies to date of both stormwater and wastewater suggest substantial variation in the coefficients, reflective of a complex chemical-biological ecosystem. But this observation is valid for all alternative design criteria introduced.
Final Observations
Regardless of the design criterion selected, arriving at the appropriate design value is a daunting task. The processes of wet basins are complex, particularly those of wetlands. Drawing general conclusions from specific studies is fraught with uncertainty. Studies must be comprehensive, structured to consider all processes and the condition of the soil before saturation. Evaluations must occur over an extended period of time as the test facility matures. The ability to draw firm conclusions from field or laboratory studies is complicated by the differing forms of the pollutants of interest and the low influent concentrations near the detection limit. Few existing published studies provide the necessary information to calculate values for the alternative design criteria discussed in this article. It is hoped this discussion will spur researches to include the necessary information in their reports.
Bigger is not necessarily better. As suggested by Figures 2 and 3, a point of diminishing returns exists at which the incremental increase in the size of a basin does not provide a demonstrable increase in performance. The gradual slopes in Figures 2 and 3 indicate that the point is not obvious. Regardless, the design coefficients should be based on incremental cost versus incremental performance, not incremental size.
Bigger can be worse. A basin surface area in excess of what is needed for effective removal of dissolved pollutants might result in the increase of relevant pollutants in the effluent. There is the possibility of too great a mass of emergent plants with loss of the pollutant bound to soluble and particulate organic matter emanating from the plants. Annual cleaning of the forebay may be appropriate to decrease solubilization of pollutants from imported organic matter, thereby reducing loading on the wet basin.
Given the importance of the soil and the effect of its variation in chemistry on performance (Hsieh and Coultas 1989, Newman and Pietro 2001), soils should be tested prior to flooding. Agricultural lands heavily fertilized may be saturated with phosphorus or nitrogen (Pant et al. 2002). Phosphorus and metals original to the soil may be lost as previously described upon basin filling. As a consequence, the wet basin may be a net source rather than a sink. Laboratory methods exist that define susceptibility to release (Minton 2002). Testing may indicate corrective actions to be taken in the design.
Summary
When considering setting goals for the removal of phosphorus, nitrogen, or metals, the importance of the dissolved fraction should be recognized. The sizing criterion should be distinct from that used to size wet basins for the removal of TSS with associated particulate pollutants.
A shallow marsh wetland is likely the most appropriate wet basin configuration for dissolved phosphorus. Whether this configuration is appropriate for dissolved nitrogen and metals has not been demonstrated; a wetpond primarily of deep water with alternating transverse bands of shallow vegetated benches might be more appropriate. It requires less space, and a lower HRT than for phosphorus may be feasible.
It is has not been definitively demonstrated that splitting the basin volume into permanent wet pool with extended detention is appropriate. To improve hydraulic efficiency, it is likely appropriate for a shallow marsh but not for the suggested wetpond configuration.
The relative sizes of the three basin types—extended detention, wetpond, and wetland—will differ with the selected design criteria and climate. Enhanced treatment wet basins as proposed here may be similar in volume to extended detention basins as currently sized but likely with larger surface areas. Except in semi-arid regions, wet basins for the removal of dissolved pollutants will require larger volumes, larger surface areas, or both than if the sole focus is 80% removal of TSS.
Process models should be used to structure pilot and full-scale field studies. With criteria specific to different climatic regions, regulators can use process models to conduct rational cost-benefit analyses. Ideally, the understanding of such models as well as the alternative design criteria discussed in this article will result in the inclusion of all necessary information in performance studies. This is essential if we are to understand the relationship between each alternative criterion and performance.
It is proposed that pollutant loading rate combined with an appropriate minimum hydraulic residence time be used as the basis for sizing wet basins. The other design criteria may do as well, but the suggested combination provides a more explicit view of the key processes except for a process model. With time, laboratory and field studies will provide definitive design values for each pollutant type, although the value of 0.5 g/m2/yr is likely appropriate for phosphorus. We will also likely find that the cost-effective unit volume and unit area requirements differ with each pollutant. Given the complexity of the processes in wet basins, particularly wetlands, professional judgment will always be central to the selection of design values.