655 18 Implementation of FWS Wetlands The purpose of this chapter is to provide information that will increase chances for successful wetland establishment and to help with understanding and overcoming difculties that may arise during construction and start-up of wetland treatment systems. Wetland treatment system establishment can be subdi- vided into three topics: (1) physical design, (2) construction, and (3) vegetation establishment. Physical design encom- passes all the choices of how the wetland basin is built, and how the water is conveyed to and through the system. With the exception of urban stormwater wetlands (Schueler, 1992), there is no single comprehensive literature source that explains options for layout and conguration of the wetland cells that typically comprise a treatment wetland. Site civil construction is a broad discipline of engineering and has been rened through centuries of experience. This chapter does not attempt to discuss all aspects of this engi- neering practice but rather describes the highlights of site civil construction that are most relevant to wetland treatment system construction. A brief general discussion of wetland site civil construction is provided by Tomljanovich and Perez (1989). All aspects of wetland construction are discussed in detail in the Wetlands Engineering Handbook (U.S. Army Corps of Engineers, 2000). This handbook discusses engi- neering procedures for establishing necessary hydrologic conditions, geotechnical design and soils handling for site modication, selecting appropriate vegetation and planting schemes, and establishing substrate conditions conducive to the desired functions. It also discusses baseline assessments of existing site conditions, monitoring strategies to determine long-term success, and contracting considerations. However, the objective of this handbook is the construction of mitigation/ habitat wetlands, not treatment wetlands. Establishment of wetland vegetation seems deceptively simple, but is not, and is frequently outside of the experi- ence of many site civil contractors. Therefore, this important aspect of treatment wetland establishment should generally be subcontracted to specialists, nursery owner/operators, or wetland builders. 18.1 PHYSICAL DESIGN The size of the wetland has been determined, at least ini- tially, by the procedures of the preceding Chapter 17. The general characteristics of the site have been investigated in Chapter 16. There remain a number of design decisions on arrangement, vertical and horizontal placement, and meth- ods of moving and controlling water. SITING General conditions of the potential site will have been con- sidered during the establishment of the basis of design. In general, there will remain a number of possibilities for the location of the system within the overall site connes. Condi- tions that should be evaluated during physical design of a wet- land treatment system include geography, soils and geology, runoff water volume and chemistry, ecological and socioeco- nomic factors, and ancillary regulations. The importance of each of these conditions may vary, but all should be inves- tigated to some extent. Detailed studies may be needed to determine the importance of those site conditions that affect technical feasibility. Floodplains and Floodways A hypothetical siting situation is shown in Figure 18.1, which indicates that the regional waterways, a river and its tributary, are geographic features to be reckoned with. Constructed wetlands located in oodplains with extreme ood condi- tions, such as along major rivers, must have dikes that are designed to allow the passage of oods and/or that are sized to exclude ood waters. Constructed treatment wetlands have typically been allowed in oodplains under constrained con- ditions. Designs should avoid damage due to high-frequency recurrence events, and therefore would usually be built to withstand the 25-year ood event without impairment of func- tion. Similarly, wetland facilities would usually be designed to withstand the 100-year ood event without severe damage. However, oodplains may be jurisdictional wetlands, thus cre- ating regulatory questions. Nevertheless, many FWS wetland systems have been constructed in oodplains such as those at Columbia, Missouri; Brawley and Imperial, California; and Tres Rios, Arizona. If a wetland is built in a oodplain, it is necessary to demonstrate that its presence will not back up oodwaters upstream of the project. In other words, the project should not block the oodway of the river. For most major rivers of the United States, there exist published maps showing the boundaries of the 100-year oodplain and the oodway. OTHER REGULATORY CONCERNS Wastewater treatment and disposal are regulated by an ever increasing number of federal, state, and local laws, rules, ordinances, and standards. In some cases, the most chal- lenging part of implementing a wetland treatment project is complying with regulations through the permitting process. In fact, regulations may hinder innovative project design © 2009 by Taylor & Francis Group, LLC 656 Treatment Wetlands for wetland treatment systems, even when the new design improves upon existing technology. Permitting an innovative project may delay implementation and increase cost with no incremental benet for environmental protection. As a result of regulatory constraints, designers may opt for conventional technology with well-known deciencies instead of technol- ogy that is perceived to have a greater risk—but with poten- tially high environmental benets. A detailed knowledge of pertinent regulations is essential to evaluate the feasibility of a wetland treatment project. This section provides an overview of the regulations that affect the use of wetlands for wastewater treatment in the United States as well as examples of some specic state situations. An updated and more detailed survey of federal, state, and local ordinances should be conducted to determine those that might be relevant to specic projects. Endangered Species Under the Endangered Species Act, the U.S. Fish and Wild- li fe Service is mandated to identify and list plant and ani- mal species that are threatened or endangered by extinction or that are considered likely to be threatened in the future. Threatened and endangered species cannot be harmed, killed, or otherwise negatively affected by human activities. The U.S. Fish and Wildlife Service reviews potential impacts to federally listed species and can veto Clean Water Act per- mits (such as a Section 404 wetland ll permit or an NPDES permit) by issuing a “jeopardy” opinion. The potential for occurrence of threatened or endangered species inhabiting a project site should be considered during project planning. Typically, a list of threatened and endan- gered species that could occur at a project location is com- piled from federal and state natural resource agency records. Road Road Road Floodplain Site 2 Site 3 Floodplain Floodplain Floodway Floodway Floodway Road City Rural Development Site 1 FIGURE 18.1 Siting and ood concerns for a hypothetical treatment wetlands project. Communities are adjacent to a river and its tributary. The oodway occupies the central valley corridors and is bounded by the oodplain. Three potential constructed wetland sites are indicated, one in the oodplain. Potential project impacts on any of these species are assessed. If negative impacts on a species are anticipated during this review, it may be prudent to conduct a eld survey to ascer- tain whether that species actually occurs at the proposed site. This conrmation step is important in eliminating the chance of nding a “fatal aw” after considerable planning, design, or construction monies have been spent. Experience indicates that the presence of federally pro- tected species does not always necessitate substantial changes to wetland treatment project siting or design. In a number of cases, the discharge of treated wastewater efuents to constructed wetlands has enhanced the population of listed wetland-dependent species by creating additional habitat and food resources. For example, populations of bald eagles, wood storks, and snail kites, all federally protected bird spe- cies, have been increased in and around the 485-ha Orlando Easterly constructed wetland treatment system in Florida. Thus, the potential impacts of a project on protected species must be evaluated on a species-by-species basis. As a further illustration, the operation of the 2,400-ha STA1E treatment wetland in south Florida was prevented for a period of weeks because of the nesting activities of burrowing owls (Athene cunicularin oridana) and stilts (Himantopus mexicanus) (Figure 18.2). CULTURAL RESOURCES The U.S. National Historic Preservation Act requires that sites of signicance for the early peoples of America be preserved. That normally involves avoidance of construction activities that might disrupt ancient village sites and burial grounds. For example, a historic archaeological site was excluded from the Coyote Hills, California, urban stormwater wetland (Figure 18.3). The Musselwhite, Ontario, mine-water treatment © 2009 by Taylor & Francis Group, LLC Implementation of FWS Wetlands 657 wetland was sited with due recognition of the line-of-sight from the regional First Nations burial ground. LAYOUT AND CONFIGURATION The area determined to achieve the project treatment goals may typically be arranged in a number of ways. The princi- pal decisions are 1. How many independent ow paths? 2. How many cells in each? 3. What aspect ratio for the cells or ow path? There is a need to set the compartmentalization of the sys- tem. The number of ow paths in the design of surface ow wetlands is based on considerations of redundancy, maintenance, and topography. All constructed wetland treatment systems should have at least two cells that can operate in parallel to allow for operational exibility (cell FIGURE 18.2 Nesting black-necked stilts (Himantopus mexicanus) caused brief stoppage of the STA1E project in Florida. FIGURE 18.3 This archaeological site was excluded from the Coyote Hills urban stormwater wetland. resting, rotation of ows, or maintenance). The ability to take cells out of service is usually viewed as very desir- able. Maintenance activities do not normally require fre- quent shutdowns, but it is usually a good idea to allow at least two ow paths so that complete system bypass is not needed. Having at least two parallel cells is especially important because of unexpected events such as vegetation die-off, pretreatment failures and subsequent wetland con- tamination, and berm and other structural failures. Mul- tiple ow paths allow the loading rate to be manipulated to meet varying inow water quality. Also, parallel ow paths allow cells to be drained for replanting, rodent con- trol, harvesting, burning, leak patching, or other possible operational controls. In the extreme long term, replacement of structures and piping become necessary. Some of the older FWS treatment wetlands are now reaching this point in their service life. Large systems may protably incor- porate more than two ow paths for purposes of internal ow control and to accommodate the site boundaries, site © 2009 by Taylor & Francis Group, LLC 658 Treatment Wetlands topography, and desired aspect ratios. However, multiplic- ity of inlet and outlet control structures can add signicant cost to the overall project. Compartmentalization Compartmentalization interacts with the design-area calcu- lation because more cells and higher aspect ratios lead to higher tanks-in-series (NTIS) numbers, and hence to higher PTIS numbers. As seen in the previous chapters, higher PTIS does not help much for low desired-removal percent- ages but is a necessity if very high percentages are required. Aspect ratio is also a determinant of NTIS, with fewer for low length-to-width ratios. A single FWS wetland cell of modest aspect ratio (1 < L/W < 5) is likely to behave as three or four NTIS in tracer testing. When compounded with the weathering factor that inuences most wastewater parameter reductions, it is likely that a P-value for the P-k-C* model will be two or three. That may be perfectly adequate if the treatment does not attempt to achieve outlet concentrations close to C*, but as the efuent target approaches C*, the value of P becomes more and more important. If the compu- tation is sensitive to P, then it becomes desirable to use cells in series. Use of three cells in series will lead to P-values between six and nine, beyond which is a region of diminish- ing returns. Site constraints of steeply sloped ground may man- date terraced, multicell design. Because it is not possible to maintain uniform level and shallow depths on steep slopes, the change in elevation is accomplished via drop structures between cells in series. Inman et al. (2001; 2003) reported that site slopes of up to 25% called for 22 cells arranged on three separate ow paths at the Clayton County, Georgia, wastewater-polishing wetland facility. That project required Deep Zone Emergent Vegetation Control Structure In Out Berm A Cell Aspect: 2:1 NTIS: 4 Vegetated Area: 0.90 Wetted Area: 1.00 Control Structures: 2 In Out B Cell Aspect: 1:2 NTIS: 6 Vegetated Area: 0.70 Wetted Area: 0.90 Control Structures: 4 C Cell Aspect: 1:1 NTIS: 9 Vegetated Area: 0.65 Wetted Area: 0.85 Control Structures: 8 In Out In Out FIGURE 18.4 Options for cells in series and parallel. NTIS are rough estimates. Areas are examples only; actual numbers are size-dependent. the movement of 420,000 m 3 of earth over the 22 ha of wet- lands or close to 2 m of cut and ll over the site area. Costs increase as the number of cells increases. The number of cells required in each ow path must be deter- mined by balancing the cost of more cells with the need for high-hydraulic efciency. The ratio of berm area to wetland surface area increases with more cells. Figure 18.4 shows the design variables associated with increasing cells in series for a hypothetical xed rectangular project footprint. Com- partmentalization increases the NTIS for the footprint. The number of transfer structures increases with the number of cells, but the vegetated area becomes smaller due to more berm area and more water distribution zone areas. If the total area were not xed, the total footprint would increase with compartmentalization. For an extreme example of dual-ow paths and multiple cells, see Figure B.24 in Appendix B for the Orlando Easterly project, which contains 34 structures for 17 cells on two ow paths. Internal Cell Arrangement The arrangements of internal features of a single wetland cell have very large implications for the efciency of the cell. Both the volumetric efciency, e V , and the detention time dis- tribution efciency, e DTD , are very sensitive to the vegetation patterns within the cell. It is probably these factors that give rise to the majority of the intersystem scatter in performance data. Each individual wetland cell may be contoured in a num- ber of different ways. The goals of internal physical design are to provide efciency of treatment, adequate conveyance of water, and perhaps habitat values. Wetlands have a ten- dency to channelize from points of inlet to points of outlet. If permitted, this operational feature reduces the gross areal efciency of the wetland. Control of the bottom elevation and © 2009 by Taylor & Francis Group, LLC Implementation of FWS Wetlands 659 vegetation density can, in principle, prevent poor ow distri- bution. But in practice, the bottom of the wetland can neither be constructed nor maintained at tolerances that promote full areal contacting. Nevertheless, care must be taken to degrade any preexisting ditches, roads, or berms on the site, because these will exert possibly undesirable ow control in the FWS wetland. There are no hard and fast rules for design for high-areal efciency for an individual wetland cell. Some general rec- ommendations are 1. Avoid very small length-to-width ratios. 2. Avoid blind spots in corners. 3. Avoid unvegetated short-circuit paths. 4. Reestablish ow distribution at intermediate points in a ow path. 5. Maintain good bottom uniformity during con- struction and start-up: minimize the formation of topographic channels parallel to ow. Figure 18.5 illustrates some of these ideas. Aspect Ratio The length-to-width (aspect) ratio is important in basin design because of its effect on ow distribution and hydraulic short circuiting. Theoretically, a constructed wetland with a high- aspect ratio is not better for treatment than one with a lower aspect ratio, as long as ow is distributed effectively. It has been speculated that long, narrow ow paths are closer to plug ow than short, wide ow paths. But the interior micro- channels of a uniform depth, uniformly vegetated FWS wet- land have small dimensions; recirculation eddies are limited by depth (on the order of 0.3 m for a FWS wetland) or by the lateral spacing of plant stems or clumps of stems (also a fraction of a meter). Thus, the effective length-to-width ratio is predetermined and widening the wetland only adds more parallel channels. (a) Bad. Open water channel from inlet to outlet. (d) Better yet. Inlet spreader ditch and outlet collector ditch. (e) Still Better. Spreader, collector and redistribution ditches. (b) Poor. Large corner zones not in flow path. (c) Better. Multiple inlets and outlets. FIGURE 18.5 Concepts of cell internals: inlet distribution, outlet collection, and internal redistribution. (From Kadlec and Knight (1996) Treatment Wetlands. First Edition, CRC Press, Boca Raton, Florida.) There is a large literature dealing with the effects of aspect ratio (length-to-width, L:W) on the behavior of inert tracers and on the performance of ponds and wetlands for pollutant reduction as a function of aspect ratio. Thackson et al. (1987) presented data that illustrate the importance of L: W on the effective volume (e V ) of unvegetated ponds, showing an increase with L:W. Since that time, computational capa- bilities have permitted theoretical exploration of the values of e V and e DTD for basins of different shapes and inlet congura- tions (see Chapter 2 for a discussion of these efciencies). For instance, Persson et al. (1999) investigated the effect of L:W using the two-dimensional, depth-integrated code of MIKE- 21 (Warren and Bach, 1992). Jenkins and Greenway (2005) also investigated empty ponds using the two-dimensional, depth-integrated code of TDFLOW (Jenkins and Keller, 1990). Both numerical studies utilized point inlets and point outlets, and found a theoretical increase in e DTD (increase in NTIS) with increasing L:W, up to 10–30 TIS for L:W = 15. However, both studies neglect wind mixing, which may be the dominant driving force in open water systems. Just as importantly, increasing aspect ratio will decrease the inefciency of utilizing a point inlet or outlet, but there are other steps that may be taken to minimize poor inlet or outlet distribution of ows. A large aspect ratio tends to improve volumetric efciency, quantied by the ratio of actual to theoretical detention time. In other words, it is not possible to get closer to the plug-ow model, but it is possible to utilize the entire wetland area. For empty ponds, a L:W of over 10 may be needed to minimize the effect of jet ow short-circuiting and corner dead zones (Figure 18.6). FWS treatment wetland data show that “longer and nar- rower” does not necessarily mean less effect of mixing, and hence a closer approximation to plug ow. For example, the tracer studies at Richmond, Australia, showed NTIS = 4.6 for an open water unit of L:W = 25 and NTIS = 5.0 for a Myriophyllum FWS wetland of the same aspect (Bavor et al., 1988). A number of tracer tests of the wetland cells © 2009 by Taylor & Francis Group, LLC 660 Treatment Wetlands 7B and 9B, both with L:W = 12.5, at the Sacramento, Califor- nia, project found NTIS = 4.7 ± 0.8 and 5.0 ± 1.4, respectively (Nolte and Associates, 1998b). On the other hand, the data from the Champion, Florida, project show NTIS = 11 for a FWS wetland with L:W = 10 (Knight et al., 1994). Likewise, the South Florida Water Management District (SFWMD) periphyton treatment wetland project found NTIS = 9 for L:W = 5 and NTIS = 25 for L:W = 45 (CH2M Hill, 2003a). The effect of aspect ratio in these studies is confused by issues of wind, vegetation patterns, and deep zones. Thus, it is dangerous to assume that aspect ratio alone can provide the desired improvements in efciency. The literature contains a few “recommendations” for L:W ratios to be used in design. For instance, Crites et al. (2006) recommend 2:1 < L:W < 4:1. U.S. EPA (2000a) comments that, in general, FWS wetlands are built with L:W > 1:1. These appear to be reasonable suggestions that go with the recognition that lower system aspect ratios may be used provided cells in parallel are used. However, there is not much quantiable treatment performance motivation to prefer one aspect ratio over another, as long as the design stays in a reasonable range, such as 2 < L:W < 10. However, there are sometimes limitations on aspect ratio due to the hydraulic prole (see Chapter 2 and subsequent sections of this chapter). If one looks beyond the volumetric improvements, data from FWS wetland studies indicate somewhat better perfor- mance for higher length-to-width ratios (Herskowitz, 1986; Knight et al., 1994; CH2M Hill, 2003a), but the margin is not large when the pollutant reductions are low-to-moderate (0–50%). Higher aspect ratios increase the area of external berms that must be constructed to enclose a given wetland area. Therefore, economics may argue for low L:W ratios (Knight, 1987). Other methods for maintaining effective ow distribu- tion, such as deep zones, should be considered as alternatives to reduce the need for higher length-to-width ratios. 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.1 1 10 100 Aspect Ratio, L:W Volumetric Efficiency, e V FIGURE 18.6 The effect of aspect ratio on the volumetric efciency of unvegetated ponds with point inlets and point outlets. (Simulation data from Jenkins and Greenway (2005) Ecological Engineering 25(1): 61–72.) Sinuousity One of those alternatives is to use smaller internal divider berms to create a sinuous ow path of high-aspect ratio (Figure 18.7). An extreme example of this design is the Tucush, Peru, wetland with 14 cross-passes in each of two ow paths (see Figure 16.1). The interior berms create longer ow paths at the expense of cost and area. Not only do the berms subtract from the wetted area but the frequent ow reversals invite dead zones in the corners of the switchbacks. These corner zones may be minimized by orienting the ow in the long dimension of the cell. FIGURE 18.7 Options for interior berms creating sinuosity. NTIS are rough estimates. Areas are examples only; actual numbers are size-dependent. A Cell Aspect: 2:1 NTIS: 4 Flow Area: 0.90 Wetted Area: 1.00 Control Structures: 2 B Cell Aspect: 20:1 NTIS: 12 Flow Area: 0.80 Wetted Area: 0.90 Control Structures: 2 C Cell Aspect: 12:1 NTIS: 15 Flow Area: 0.60 Wetted Area: 0.90 Control Structures: 2 © 2009 by Taylor & Francis Group, LLC Implementation of FWS Wetlands 661 Computational uid mechanics, in conjunction with rst-order reaction rate kinetics, was used by Abbas et al. (2006) to simulate the effect of bafes creating sinuosity in ponds of different aspect ratios. These calculations were vertically averaged, and thus might be considered repre- sentative of unvegetated wetlands as well. Higher length- to-width was effective in promoting biochemical oxygen demand (BOD) reduction, but more internal bafes were yet more effective. Ratios 1:1 < L:W < 100:1 were investi- gated for point inlets and outlets. The swing in BOD reduc- tions was from 10 to 95%. Field testing of pathogen reduction was done in a side- by-side test at Lidsey, United Kingdom, utilizing bafed and unbafed ponds, 1 m deep, with point inlets and outlets (Bracho et al., 2006). The unbafed pond had L:W = 9:1; the adjacent pond of the same shape and size had three longi- tudinal bafes, creating L:W = 79:1. Tracer testing indicated NTIS = 2.1 and e V = 0.33 for the unbafed pond, and NTIS = 7.3 and e V = 0.38 for the bafed pond. The low efciencies were attributed to “jet ow short-circuiting.” The reduction of fecal coliforms was 82.1% in the unbafed pond and 99.4% in the bafed pond. Interestingly, the use of divider berms to create back- and-forth sinuous ow paths has been found to provide low- volumetric efciency while showing moderate-to-high NTIS values in some systems. Tracer tests at the sinuous Hillsdale, Michigan, site showed NTIS = 4.3 and 7.4, and e V = 0.33 and 0.51, and for the east cell of L:W ≈ 16 (four cross-passes in a 1:1 footprint); and NTIS = 37, and e V = 0.34, and for the west cell of L:W ≈ 18 (ve cross-passes in a 1.6:1 footprint). The reason was found to be the fact that the ow path became channelized (deeper) in the center portion of the sinuous path, leaving excluded sides and corners. A side-by-side comparison of sinuous and nonsinuous cells for phosphorus removal was conducted by the SFWMD (Figure 18.8) (CH2M Hill, 2003a). The hydraulics of the sin- uous cell were superior with NTIS = 25 versus NTIS = 9 for the nonsinuous cell. However, the volumetric efciency was only 58% versus 88% for the nonsinuous cell. The phospho- rus performance of the sinuous cell was superior to that of the nonsinuous cell with higher k-values. Divider berms are very prone to cross-berm leakage, which may create bad short-circuits. This was the case at the SFWMD project. Also, the divider berms occupy some mea- surable fraction of the total footprint, which detracts from overall areal effectiveness. Results to date indicate that inter- nal sinuosity may have little net effect. Flow Distribution Both eld results and computational studies have shown that point inlets and point outlets are not a good idea. Pond studies are not a good model in this regard, because inlets to ponds are very frequently point inlets. The corner zones of the pond are often not easily accessed by the incoming water. However, the presence of vegetation in FWS wetlands can lead to even more difculties for water to reach corner zones. Computational uid mechanics suggests that if the water is distributed uniformly from side-to-side of the wetland inlet, rather than at a point, there will be a signicant improve- ment of the hydraulics (Persson et al., 1999). The changed distribution is forecast to increase NTIS from 1.5 to 6.7, and e V from 79 to 89% for a basin of L:W 2:1. Wörman and Kronnäs (2005) computed that as the width of the water inlet increases relative to the wetland width, both e V and NTIS should increase. Stairs (1993) experimentally determined the effect of inlet cross-spreading by performing tracer tests both before and after the removal of a perforated pipe spreader. With the spreader, NTIS = 4.47 and e V = 83.6%. Without the spreader, for a point inlet, NTIS = 3.27 and e V = 66.5%. Thus, the theo- retical results of simulations have been veried in the eld. Cell 2 Cell 1 FIGURE 18.8 Cells 1 and 2 of the South Florida Water Management District periphyton treatment wetland project. Cell 1 had an aspect ratio of 5:1, whereas Cell 2 had L:W = 45:1, because of a three-pass sinuous design. Both were two hectares in size and are shown empty in this photo. Cell 1 tracer tested at NTIS = 9, whereas Cell 2 tested at NTIS = 25. (Photo courtesy CH2M Hill.) © 2009 by Taylor & Francis Group, LLC 662 Treatment Wetlands The collection of the water at the wetland outlet poses the same problems of corner dead zones. Therefore, collection across the wetland width is advantageous. Bathymetry: Deep Zones and Speed Bumps Wetlands have a tendency to channelize from points of inlet to points of outlet. If permitted, this operational fea- ture reduces the gross areal efciency of the wetland (see Fi gure 4.21, for instance). Control of the bottom elevation and vegetation density can, in principle, prevent poor ow distri- bution. But in practice, the bottom of the wetland can neither be constructed nor maintained at tolerances that promote full areal contacting. Bottoms do not start completely level, and a number of processes contribute to the formation of local highs and lows such as soil heaving and sediment deposition. Nevertheless, care must be taken to degrade any preexist- ing ditches, roads, or berms on the site because these will exert possibly undesirable ow control in the FWS wetland. Dierberg et al. (2005) demonstrated the poorer treatment that occurs along such channelized short circuits. Compartmentalization involves emergent berms with transfer structures, but such cross-berms may also be con- structed so as to be always or occasionally under water. For example, the treatment wetland at Bulwer Island, Australia, was constructed with a repeated pattern of 0.4 l 0.2 l 1.0 m depths along the ow path (Simi and Mitchell, 1999). The intent of this design was to break up longitudinal channeling and to promote different microenvironments. Underwater cross berms have also been placed in some of the stormwater treatment areas in South Florida. It is not known if these have positive effects on treatment. The use of cross-benches was forecasted to promote good hydraulics, NTIS = 23 and e V = 0.80, by Persson et al. (1999). Deep zones in surface ow constructed wetlands are sup- posed to serve several purposes (Knight and Iverson, 1990). These deeper areas extend below the bottom of the vegetated basin areas by at least 1 m to exclude the development of ro oted macrophytes (Figures 18.9 and 18.10). Such unveg- etated cross ditches provide a low-resistance path for water to move laterally and provide a nearly constant head across the wetland. They also provide for extra detention time, but in a deep water zone. Such ditches often become covered with duckweed (Lemna spp.) and can be used by wetland birds and sh as reliable habitat. These redistribution ditches change the potential for short-circuiting within the wetland, because high-speed rivulets are intercepted and mixed with quiescent water in the deep zone. However, the redistribution ditch adds a potential for wind mixing that compensates the reduced short circuiting. But water is more effectively distrib- uted over the wetland, improving the gross areal efciency to some extent. Lightbody et al. (2007) provided evidence from laboratory models and computations that indicated improve- ment due to deep zones at high degrees of removal. However, that improvement was contingent on the assumption that seri- ous short-circuiting occurred in the vegetated sections of the system. Knight et al. (1994) operated a set of six treatment wet- lands receiving treated paper-pulp mill efuent for two years. Two pairs of these received the same inuent water, one of each pair containing two internal deep zones comprising 25% and 45% of the area (Table 18.1). There was variability in ow (different HLRs) and aspect ratios (2.5 and 10) between the pairs. Tracer data from that project were here reanalyzed and characterized by a tanks-in-series (TIS) detention time distribution. The effect of deep zones was to decrease the number of TIS (Table18.1). And, for the larger cell pair (A and B), the volumetric efciency was improved by deep- zone addition. The load reductions for all contaminants were increased by the inclusion of deep zones but by only a very small margin for the higher aspect ratio pair (E and F). One Deep Zone Emergent Vegetation Control Structure In Out Berm A Cell Aspect: 2:1 NTIS: 4 Vegetated Area: 0.90 Wetted Area: 1.00 Control Structures: 2 In Out B Cell Aspect: 2:1 NTIS: 4 Vegetated Area: 0.70 Wetted Area: 1.00 Control Structures: 2 C Cell Aspect: 4:1 NTIS: 6 Vegetated Area: 0.65 Wetted Area: 0.90 Control Structures: 4 In Out In Out FIGURE 18.9 Options for deep zones in series and parallel. NTIS are rough estimates. Areas are examples only; actual numbers are size-dependent. © 2009 by Taylor & Francis Group, LLC Implementation of FWS Wetlands 663 FIGURE 18.10 The Saginaw, Michigan, treatment wetland system contains two ow paths, each bifurcated by one cross-deep zone. interpretation of these results is that the net effect of better efciency and worse NTIS results from interception of short circuits in the wide cell but that these are less important for the high-aspect cells. Thus, wide cells prot from deep zones but narrow cells do not. Another important feature of these results was the fact that the added volume provided by deep zones resulted in smaller volumetric rate constants, an obser- vation in agreement with similar results detailed in Part I. The artifact of L:W was not present in a study of 12 research wetlands operated under varying conditions at a site west of the city of Phoenix, Arizona (Kadlec, 2006a). These were constructed as a triplicated design with zero, one, two, and three internal deep zones, all containing an inlet distribu- tion and an outlet collection deep zone, together comprising 12.5–35% of the wetland areas. There were no differences in the internal hydraulics with deep-zone numbers. Deep-zone numbers in the wetlands did not affect nitrogen treatment performance. No differences with deep-zone numbers were found for temperature, dissolved oxygen, pH, or nitrogen removals or rate constants. In this study, there was no large treatment benet or detriment of incorporating internal deep zones in free water surface wetlands. Moore and Niswander (1997) operated a set of six treatment wetlands receiving diluted dairy wastewater for two years. Two wetlands had a central internal deep zone comprising 45% of the area, whereas the other four had no internal deep zone. The authors concluded that deep-center sections did not show any signicant impact on treatment efciency. Data for the second year, past the startup period, are s h own in Table18.2, as reported in NADB v.2 (NADB data- base, 1998). None of the outlet concentrations are different at the 5% level. The wetland areas were all identical and, hence, the hydraulic loading to all six wetlands was uniform at 3.95 cm/d. For a presumed 4 TIS hydraulic pattern, the areal k-values do not differ at the 5% level. For instance, for total Kjeldahl nitrogen (TKN), k = 10.1 ± 1.1 m/yr compared to TABLE 18.1 Performance of the Champion Wetlands, July 1991–June 1993 Wetland A B E F Area (m 2 ) 4,048 4,048 1,012 1,012 Aspect 2.47 2.47 9.88 9.88 DZ% 5 25 5 45 Mean depth (m) 0.27 0.38 0.27 0.58 HLR (cm/d) 3.16 3.38 7.82 8.13 Nominal HRT (d) 8.54 11.23 3.45 7.13 NTIS 4.6 3.6 10.7 2.0 Volume efciency 0.63 0.81 0.91 0.74 Inlet TN (mg/L) 9.1 9.1 9.1 9.1 Outlet TN (mg/L) 3.5 2.5 4.7 4.7 TN LR (g/m 2 ·yr) 65 81 126 131 Inlet NH 4 -N (mg/L) 3.4 3.4 3.4 3.4 Outlet NH 4 -N (mg/L) 0.7 0.3 2.0 2.0 NH 4 -N LR (g/m 2 ·yr) 31 38 40 42 Inlet NO x -N (mg/L) 0.90 0.90 0.90 0.9 Outlet NO x -N (mg/L) 0.13 0.11 0.13 0.12 NO x -N LR (g/m 2 ·yr) 9 10 22 23 Inlet BOD (mg/L) 21 21 21 21 Outlet BOD (mg/L) 10 7 13 13 BOD LR (g/m 2 ·yr) 127 173 228 237 Inlet TSS (mg/L) 58.6 58.6 58.6 58.6 Outlet TSS (mg/L) 10.1 6.9 12.9 13.6 TSS LR (g/m 2 ·yr) 559 638 1304 1335 Inlet TP (mg/L) 0.97 0.97 0.97 0.97 Outlet TP (mg/L) 0.37 0.28 0.58 0.59 TP LR (g/m 2 ·yr) 6.9 8.5 11.1 11.3 Source: Hydraulic and concentration data are from Knight et al. (1994) TAPPI Journal 77(5): 240–245; the DTD analyses are from this work. © 2009 by Taylor & Francis Group, LLC 664 Treatment Wetlands k = 11.0 ± 1.1 m/yr for the two cells with deep zones. How- ever, the inclusion of the deep zone added to the water vol- ume, and increased the average depth from 30 to 75 cm, and consequently increasing the nominal detention time from 7.6 to 19 days. As a consequence, the k V values for TKN were different at the 5% level, 0.093 ± 0.01 d −1 for no deep zones versus 0.040 ± 0.004 d −1 for deep zones. As for the Champion study, these Oregon State University results suggest that the extra depth added by deep zones carries a penalty in volu- metric performance. Importantly, all three studies demonstrate a penalty for deep zones in terms of lower volumetric rate constants for systems with deep zones. The presumption of constant k V , which is reiterated in much of the treatment wetland literature (Crites and Tchobanoglous, 1998; U.S. EPA, 1999; U.S. EPA, 2000a; Water Environment Federation, 2001), is, therefore, inappropriate for design of FWS wetlands with deep zones. This further substantiates the growing body of knowledge that wetland areal models are to be preferred over volumetric models for FWS wetlands, in general. For deep zones, the added water volume and its attendant-added detention time do not contribute proportionately to pollutant removal. Taken together, these studies do not demonstrate a clear advantage or a clear disadvantage for deep zones. Other wet- land attributes, such as aspect ratio, vegetation community type, and compartmentalization, may be equally or more important for water quality improvement. Nevertheless, deep zones may be important for habitat value, may provide for increased reaeration, and may provide opportunity for patho- gen reduction via UV radiation. The amount of open water that may be included in a cell remains a controversial issue. Based upon anecdotal evidence, U.S. EPA (2000a) contends that the central part of a ow path must be open water/SAV to provide oxygenation for ammonia reduction. However, Kadlec (2005e) found no particular support for that view in an examination of data from many systems, including ponds as the open water extreme. Open water sections in a cell pro- vide better habitat value than does a uniform dense stand of emergent plants (Weller, 1978). However, habitat values must be balanced by the potential for wildlife to interfere with treatment. Excessive waterfowl use can add pathogens to the water. Muskrats, nutria, and beavers want the deep water for refuge and can seriously impair the vegetation, hydraulics, and possibly treatment (Kadlec et al., 2007). An important function of an inlet deep zone is the set- tling of particulate matter that might otherwise create dif- culties in the inlets of vegetated sections. This has been discussed in Chapter 7 but deserves consideration here as part of the process of developing a layout for the treatment wetland. It is useful to remove large sediment loads before they reach the vegetation, so that the eventual cleanout is from a pond rather than from vegetation. The Imperial, California, system employs a 3.9-ha, two-cell sedimenta- t i on basin before wetlands (Figure 18.11). Over a four-year operating period, the basins trapped 3,820 tons of sediments, amounting to approximately 10 cm of accretion. Those solids would have been detrimental to the inlet zone of the 4.7-ha follow-on wetlands. Table 7.8 indicates that it is easily pos- sible to settle river-borne solids in just a few days detention in a presettling basin with removals of over 90%. The siz- ing of this inlet element of the treatment train is dependent upon the settling characteristics of the specic solids in ques- tion, as well as the accumulation rate of the solids. An inlet deep zone built for ow distribution may ll up quickly and require emptying at frequent intervals (see Figure 7.13). A simple mass balance sufces for the sizing of sedimentation basin storage for the desired loading and emptying frequency, whereas a settling calculation determines the area needed for a given basin-operating depth. Fringing, Banded, and Clumped Vegetation Many natural wetlands are riverine, meaning that the wet- land contains a central stream channel, which may meander through a zone of emergent vegetation. Indeed, this wetland form is so common that it is mistakenly presumed to be a good prototype for a treatment wetland. Nothing could be further from the truth. A central channel will convey most of the water, and the fringing vegetation will effectively receive ow only on rising ood conditions and discharge it on the ebbing stage. Therefore, congurations similar to Fi gure 18.12A are usually to be avoided. Simply stated, if the water needs to go through the plants, do not plant them off to the side of the ow path. Jenkins and Greenway (2005) TABLE 18.2 Oregon State University Pilot Wetland Performance, Means, and Standard Deviations among Wetlands Parameter Inlet Outlet (N = 4) No IDZ Outlet (N = 2) One IDZ pH 7.50 7.11 7.25 0.05 0.03 DO 2.35 0.15 0.19 0.12 0.06 NH 4 -N 148 84 78 46 TKN 204 107 102 76 TSS 653 195 202 22 6 BOD 888 307 267 39 7 TP 36.5 19.6 18.4 1.1 1.5 Note: IDZ inlet deep zone; data are averages of biweekly data, January 1994–February 1995. Source: Data from NADB database (1998) North American Treatment Wetland Database (NADB), Version 2.0. Compiled by CH2M Hill. Gainesville, Florida. © 2009 by Taylor & Francis Group, LLC [...]... 55 are trees, and 18 are vines The NADB v.2 lists 593 macrophytic plant species reported from constructed treatment wetlands and 427 species from natural treatment wetlands Emergent herbaceous macrophytes account for 501 species in constructed treatment wetlands and 290 species in natural treatment wetlands A significant variety of tree and shrub species occur in a few constructed wetlands (primarily... of potted plants is the high initial cost They are not only more expensive to grow but also expensive to transport to the newly constructed wetland Potted plants should be in at least 10-cm pots The bound soil/root mass should be the same size as the container (otherwise, the product could be a 5-cm plant in a 10-cm pot) Plants in peat pots should be well rooted through the sides and bottom of the pot, ... the wetlands This is the case in South Florida, where water conservation wetlands and stormwater treatment areas are so configured Recycle inevitably requires pumping Treatment advantages that may be gained by this procedure are, therefore, expensive and sometimes are more economically accomplished by increasing the size of the wetland 18. 5 STORMWATER WETLANDS The physical design of event-driven wetlands. .. Gated pipe (Figure 18. 24) Implementation of FWS Wetlands FIGURE 18. 21 Inlet structures include point discharges, perhaps with pre-aeration at the Olentangy Wetlands in Ohio • • • • • Multiple tees Perforated pipe Large rock zone Spreader swale Castellated weir Figure 18. 25 shows several approaches that have been used to distribute inflows and create sheet flow conditions in constructed wetlands These often... species A more diverse mix of plant species will be better able to accommodate changes in water quality and flow 684 Treatment Wetlands TABLE 18. 3 Percent Removals for the Oregon State Side-by-Side Animal Waste Treatment Cells with Different Vegetation Types Vegetation Type BOD COD TSS TKN NH4-N TP Grass/Cattail/Bulrush (60:30:10) mean s.e 57 7 46 6 52 12 48 5 39 7 48 6 Grass/Cattail/Bulrush (50:30:20)... exterior slurry Implementation of FWS Wetlands 669 FIGURE 18. 17 Design considerations for constructed wetland berms (From Kadlec and Knight (1996) Treatment Wetlands First Edition, CRC Press, Boca Raton, Florida.) wall tied into deeper, low-permeability sediments can be used to limit offsite infiltration Berm slope is dictated by geotechnical considerations and a slope-stability analysis Minimum berm slopes... rap 30 cm Rooting soil 60 cm 30 cm 30 cm 10 cm Dike soil Liner protection (Sand) FIGURE 18. 18 Typical constructed wetland exterior berm with a liner © 2009 by Taylor & Francis Group, LLC Impervious liner 670 Treatment Wetlands FIGURE 18. 19 Wetland berm protection using stone riprap at the Lake Nebagamon, Wisconsin, treatment wetland This divider berm is not drivable Photo shows wetland just after planting... groundwater Outlet structures should be designed to allow backflooding (not through-flooding) of wetland cells 672 Treatment Wetlands FIGURE 18. 20 Wetland dike erosion occurring due to wave action at Stormwater Treatment Area 1W in Florida Winds at the site have a long fetch across this large wetland cell Hurricane-force winds caused this damage Damage to the wetland vegetation is not likely for inundation... suppliers who can propagate the types and quantities of plants required for constructing large wetland treatment systems Plant propagules that are frequently used to establish constructed wetlands for wastewater treatment include seeds, bare-root seedlings, rhizomes, greenhousegrown potted seedlings, and field-harvested plants If the wetland is to be planted, the cost and availability of plant materials must... Columbia, Missouri, treatment wetlands (Brunner et al., 1992; Brunner and Kadlec, 1993) The conceptual configuration was selected to be three banks of cells in parallel The available lands were bounded by streams, roads, towns, railroads, and the floodway of the Missouri River As a consequence, the actual 666 Treatment Wetlands FIGURE 18. 13 Aerial view of the Brawley, California, treatment wetland The . (Figure 18. 3). The Musselwhite, Ontario, mine-water treatment © 2009 by Taylor & Francis Group, LLC Implementation of FWS Wetlands 657 wetland was sited with due recognition of the line-of-sight. high-areal efciency for an individual wetland cell. Some general rec- ommendations are 1. Avoid very small length-to-width ratios. 2. Avoid blind spots in corners. 3. Avoid unvegetated short-circuit. NH 4 -N (mg/L) 3.4 3.4 3.4 3.4 Outlet NH 4 -N (mg/L) 0.7 0.3 2.0 2.0 NH 4 -N LR (g/m 2 ·yr) 31 38 40 42 Inlet NO x -N (mg/L) 0.90 0.90 0.90 0.9 Outlet NO x -N (mg/L) 0.13 0.11 0.13 0.12 NO x -N