THE HYDROLOGY OF STORMWATER MANAGEMENT TECHNOLOGIES: DETENTION vs. INFILTRATION Thomas H. Cahill, P. E., Michele C. Adams, P. E., and Andrew Potts Introduction and Background Over the past twenty-five years, the primary method of stormwater management for new development projects has focused on peak rate control of runoff using the detention basin (DB). These earthen basins are usually constructed at a lower elevation on the parcel, downgradient from the development and most earth-moving activities. Detention basins also frequently function as a temporary erosion and sediment control structure during earth disturbance, with later conversion to a detention structure. The size of the structure is based on the storage volume required to control (i.e., hold constant) the peak rate of runoff from the site (usually estimated in cubic feet per second, or cfs). That is, the post-development peak rate of runoff from the site, which impacts receiving swales and stream channels, is controlled so that the post-development peak rate is no greater than pre-development peak rate. This approach to stormwater management was thought to be adequate to solve potential flooding problems, generally regarded as the primary stormwater concern. The rates of runoff prior to and following development are usually estimated by a method of runoff analysis described as the Cover-Complex procedure, also referred to as TR-55 and TR- 20, developed by the Natural Resources Conservation Service (then the Soil Conservation Service; USDA, SCS, 1968) some thirty-five years ago to analyze stormwater runoff from natural landscapes. The TR-55 Cover-Complex method has been developed as a computer model, the current version of which is TR-20, and is generally available to the professional community. The basic theory behind this model is that the soil mantle, with different types of vegetative cover or man-made surfaces, receives incident rainfall over a period of time. As the rain infiltrates into the soil surface during initial rainfall, little or no runoff is produced. During continued rainfall, the rate of infiltration diminishes, depending on the physical form of the soil and the land surface. After some time the vegetated surfaces of the ground reach temporary saturation, at which point additional rainfall immediately runs off of the saturated surface, moving down-gradient at a rate based on the slope of the land and the texture of the land surface. When a site is analyzed for land cover conditions before and after development, the new impervious surfaces convert virtually all of the direct rainfall to immediate runoff, producing a runoff flow that is both more rapid and much greater in volume. Rainfall on a rooftop or pavement is immediately discharged across the surface, with sufficient kinetic energy to scour and suspend particulate matter and solubilize other accumulated material from the surface. The mechanisms of transport as part of the surface runoff and the various chemical forms that comprise what is called non-point source pollution (NPS) has been the subject of extensive research over the past thirty years (Cahill, et al; 1975, 1976, 1991), but is not yet clearly defined. Various technologies have been developed to remove these NPS pollutants from the runoff, with mixed results (USEPA, 1993). 1
Both the volume of runoff and concentration of various NPS pollutants contained in that runoff are entirely dependent on the land surface. On a naturally vegetated soil mantle, covered with woodland or other vegetation, the process is generally understood, but the exact amount of runoff produced under varying conditions of climate, rainfall intensity and soil mantle is derived from observation and empirical formulas (Maidment, 1992). In fact, the Cover-Complex method was developed by the SCS specifically to understand the effects of land cultivation practices and erosion control measures with various soils, and later evolved as a tool for impervious cover runoff analysis. It is intuitively obvious that runoff is a function of rainfall, but the development of a complete equation that explains that process is still a work in progress, and has been for over a century. The Hydrologic Cycle In order to understand the process of runoff, it is first necessary to consider the hydrologic cycle and the proportion of that cycle that produces surface runoff. The explanation of this natural cycle of water movement in the atmosphere, land surface and sub-surface and ultimate return to the atmosphere is well illustrated in any basic science text. Under natural conditions, only a small fraction of the total rainfall in almost any physiographic region results in direct surface runoff following a storm. With a developed soil mantle and vegetated land surface, about half of the rainfall in any significant event is returned to the atmosphere as evaporation and transpiration, with these mechanisms variable by season and climate. The actual percentage of rainfall that becomes immediate runoff from the land surface is quite small for most storms under natural conditions, on the order of 20% or less. When we cover the soil with structure or pavement, creating a land surface that is impervious to the rainfall, we dramatically alter the hydrologic cycle. The ability of the soil mantle to infiltrate virtually all of the rainfall is prevented, and water that would have been returned to the atmosphere or percolated slowly into the soil mantle to recharge the groundwater table, or subsurface zone of saturation, is converted into immediate and direct runoff. Figures 1 and 2 illustrate, in simplistic fashion, the net impact of impervious surfaces. Here the amount of runoff from a parcel of land is increased fivefold in volume by the direct discharge of rainfall from all such surfaces. In the Piedmont Physiographic Region of the eastern United States, the annual increase in runoff can exceed 36 inches (3 feet) in an average year for every unit area of impervious surface. That is, a square foot of woodland, cleared and paved, will add three cubic feet of additional runoff during an average year. In addition, if the soil mantle surrounding the new rooftops and pavement is disturbed, re-compacted and re-vegetated in a grass, the new surface does not retain any significant portion of the natural infiltration of the parent soil, and can be considered as semi-impervious. The process of pollutant transport is also greatly influenced by an impervious or reduced pervious surface. Those materials applied, spilled, dripped or decomposing on an impervious surface will be washed from that surface with runoff, whereas the same materials on a soil mantle during rainfall will largely pass through the soil and be subject to a complex microbial process of biochemical, physical and chemical transformations that will largely render these materials harmless or reduce their mobility, binding them to molecules in the soil. This is not the case for some soluble pollutants, such as Nitrate, which move through the soil unchanged. 2
Figure 1. The Hydrologic Cycle in the Eastern Piedmont Region on a Unit Area Basis Figure 2. The Hydrologic Cycle As Altered by Typical Land Development 3
The Cover-Complex Method of Runoff Analysis In the Cover-Complex Model, this net change in hydrologic response is approximated by estimating the change in surface cover conditions produced by a mix of pervious, semiimpervious and impervious surfaces. The procedure develops a weighted factor described as the Cover factor (C). For impervious surfaces, this factor is almost 100% or 0.98, indicating that nearly every drop of rainfall on that surface will be immediately converted to runoff (with some loss by evaporation, even during rainfall). As a site or catchment is analyzed, the combination of land surfaces are measured and weighed, and the resultant C-value estimated. Thus as the value of C increases, the proportion of rainfall that becomes immediate runoff increases. The original development of runoff factors by the SCS involved field-testing of various types of soil and the development of a classification into what is called Hydrologic Soil Groups or HSG s. Virtually all of the basic types of soil found in the U.S. have been classified in this way, and four alpha codes are used to rank a soil series by its natural tendency to infiltrate rainfall. Soils that are HSG A are very permeable, and allow for almost total infiltration of rainfall amounts. The great majority of soils found in the eastern U.S. are group B or C, with low-lying wetland soils or flood plain soils classified as group D. Thus, the development of a C-value for a given landform involves the consideration of both the HSG type and the vegetative land cover. This process evolved into a set of graphic figures and mathematical functions included in the TR-20 Manual. Since the hydrologic cycle begins with rainfall, the pattern and intensity of rainfall over the duration of a storm is the initial assumption of any runoff model. Analysis of the change in hydrologic response produced by development on a parcel begins with an estimated pattern of rainfall for a given region. A 24-hour rainfall pattern described as an S curve is assumed, beginning with an extended period of fairly light precipitation during the initial 10 to 11 hours of the storm. More intense precipitation develops at hour 11:30 and continues for about 1 hour, with a gradual diminution for the balance of the 24-hour storm period. A graph of time versus rainfall rate illustrates this relationship (Figure 3). During initial rainfall, the soil mantle infiltrates most of the precipitation and effectively reduces the runoff, a process described as the Initial Abstraction (I a ), estimated in inches. Empirical field studies by SCS relate this term to the much greater potential maximum retention (or infiltration) S, also in inches. (I a = 0.2S). This term (S) is derived from the soil-cover descriptive value, cover number (CN), where S = (1000/CN) 10. The total runoff (Q) produced during a given rainfall (P) can then be estimated as the primary equation: Q = (P-I a ) 2 / (P-I a ) + S Substitutions allow the development of a series of nomographs (Figure 4) for manual analysis of parameters or algorithms for the model. The method is highly empirical, based on extensive field studies of actual test plots with rainfall simulation, combined with many years of largescale measurements under different cover conditions. Since the mechanism of soil infiltration does vary (diminish) as rainfall progresses and surface saturation is achieved, the method is a reasonable approximation of the relationship between precipitation and runoff as controlled by land surface and vegetative cover. 4
8 7 7.3 Inches Rainfall (inches) 6 5 4 3 3.3 Inches 2-Year Storm 100-Year Storm 2 1 0 0 4 8 12 16 20 24 Time (Hrs) Figure 3. SCS Type II Rainfall Distribution for the 2-year and 100-year Storms in the Northern Piedmont Region (USDA, SCS, TR-55) Figure 4. Graphic Expression of the Rainfall-Runoff Relationship as Developed in the Cover-Complex Model (USDA, SCS, TR-55, 1986) 5
When an impervious surface such as a pavement is built over the soil mantle, other hydraulic models are used to describe the movement of the stormwater on a flat surface, where gravity provides the energy to displace the liquid. The actual pathways of runoff movement over a much rougher vegetated land surface can be extremely complex, and many of our analogies for surface friction and length of the flow pathway are only approximations of the real world, so that this overland flow routing component of the model is the least accurate. However, it remains the best analogy presently available for estimation of runoff rate and volume increases produced by land development. Runoff Rate vs. Volume The stormwater management methods compared in this paper, detention and infiltration, are fundamentally different in form and theory. With detention design, increased runoff is collected and held for several hours and subsequently released to the receiving stream or watercourse. The critical design parameter is the estimated hydrograph (Figure 5) or pattern of runoff anticipated at the downstream end of the site, and shows the rate of runoff over time, with the area under the curve representing the actual volume of runoff. The same figure can be estimated for conditions following land development, and then the net benefit of attenuating the peak rate by using the detention basin storage capacity can be estimated by the same method (Figure 6). Temporary storage of runoff in the detention basin can control the rate of runoff so that it does not exceed the rate that existed prior to disturbance for certain storms. The total volume increase, however, is not mitigated. In theory, this rate reduction should prevent any increase in erosive energy in the receiving stream channels. However, the period during which bank full flow conditions will occur is significantly greater for any given event, and thus the erosive impact is still greater than prior to development. As an initial design criteria identified in the early 1970 s, this detention basin approach made sense as an initial management method. However, subsequent analysis has demonstrated that this limited stormwater management method is inadequate to alleviate the related water resource impacts of land development. Base Flow Reduction Additional impacts center on two issues; the loss of rainfall volume that normally would have infiltrated the soil mantle and recharged the groundwater system (see Figure 2) and the land surface pollution that is scoured by (or dissolved in) the runoff from the land surface, described as non-point source (NPS) pollution. Under natural conditions with a vegetated land surface, a major fraction of the rainfall that infiltrated the soil mantle would be taken up by the plants (especially woody vegetation) and returned to the atmosphere as transpiration. The remaining portion of infiltrating rainfall would percolate slowly through the soil mantle until it reaches the zone of saturation or water table in the underlying soil or bedrock aquifer. This continuous (but variable) replenishment of the groundwater is vital to the natural hydrologic system. While the amount of water added to the groundwater varies with physiographic conditions, season and climate, the local surface stream system is dependent on this source for maintaining flow during periods without precipitation. This reduction of groundwater recharge expresses itself as a reduction in stream base flow weeks or months after rainfall, and is seldom recognized as the result of impervious surfaces built within the watershed. 6
Figure 5. Hydrograph Showing the Impact of Development: Both Peak Rate and Volume Increase Dramatically Figure 6. Effect of Detention on Hydrograph: Peak Rate may be Controlled but Volume Increase is not Mitigated 7
When we observe a natural stream or river, we seldom consider that most of the flow is coming from the groundwater storage throughout the watershed. Countless seeps and springs along the many single streams (considered first order streams) drain from the headwaters and are discharges of groundwater to the land surface. When it has not rained in more than two days, virtually the entire flow of a surface system is provided in this fashion. Figure 7 illustrates the role of groundwater recharge in sustaining surface streams. Without continuous replenishment by surface infiltration, this recharge can be greatly diminished or in some extreme situations, eliminated. Our high-density urban environments represent this type of extreme land development. The constructed impervious surfaces are virtually continuous, and nearly all of the incident rainfall is collected in gutters, inlets and structures, conveyed outside the developed area, and discharged by storm sewers to surface channels. Many of these channels are remnant natural streams, now deprived of their base flow supply source and degraded to dry ditches. In many of our older urban communities, we have built pipes in the streambeds to serve as conveyance systems for the storm flow, and frequently buried these pipes, filling in the associated flood plain for use as developed land. As our communities age, the constant and continuing movement of water in the sub-surface erodes these piping systems and produces collapse of surface structures built within or over the original stream channel. While the reduction or elimination of base flow as a result of the reduction of groundwater recharge is seldom considered a stormwater management issue, it is now recognized to be as important as the traditional issue of flooding. One can define this lost infiltration using the same relationship defining runoff (Figure 8), based on the covercomplex equations. Nonpoint Source Pollution The third major impact of stormwater runoff, the transport of pollution from the land surface to natural waters, has been a subject of concern for over thirty years, although legislative action to control and reduce this pollution input has been quite limited. The emphasis during the 1960 s focused on reducing, eliminating or treating the discharges of wastewaters from our communities and industries, a subject collectively described as point source pollution. Subsequent water quality studies revealed that point sources comprised only a portion of the total pollutant load received by our streams, rivers and estuaries. The balance of pollutant loads come from the land surface; chemicals we apply, spill, drip or otherwise allow to accumulate on the land in a variety of forms. During rainfall, much of this pollution is scoured by rainfall from both impervious and pervious surfaces, to form a turbid mix of natural detritus (vegetative and organic materials) and anthropogenic waste. While the composition of this runoff varies by surface area, land use and climate, the pollutants are largely transported with the stormwater runoff in either a soluble or a particulate form. Thus during stormwater runoff periods, usually amounting to some thirty days a year, water quality is dominated by NPS pollutants (Cahill, et al, 1975). The selection of stormwater management technologies must now consider the water quality benefits of those materials and methods. Without question, the infiltration of stormwater is by far a better method of reducing NPS pollution in stormwater, as compared to any method based on surface detention systems, or the modification of such basins by permanent storage or vegetation, except for Nitrate, where removal of accumulated biomass is practiced. 8
Figure 7. Base Flow is Sustained in the Surface Streams by Infiltrated Rainfall Figure 8. Infiltration that is Lost as Curve Number Increases 9
Hypothetical Design of a Stormwater Management System The issues of stormwater management in the 21 st century have evolved well beyond the criterion of peak rate control that guided the design methods of the past twenty-five years. The following example is intended to demonstrate both the differences in design parameters and resultant water resource benefits achieved by the two basic methods under consideration: detention and infiltration. In order to provide a fair comparison between methods, a hypothetical design is introduced to allow comparison with previous designs considered or performed. For this example, the land development proposed will consist of a 3-acre (1.21 hectare) parcel of land, presently in wooded cover. The underlying soil mantle will be assumed to be either entirely HSG B or C, with both examples applied. The location of the site is assumed to be the Piedmont plateau of the eastern U.S., with a well-distributed annual rainfall of 45 inches per year. The patterns of rainfall and frequency will be as summarized in Table 1. Figure 9 illustrates that a 2-year frequency rainfall represents (and includes) over 95% of the total rain in any given period, typical of this region. The proposed development will consist of a one-acre building surrounded by an additional 1.2 acres of impervious surfaces, most of which are in the form of a one acre parking lot adjacent to the building. Under normal zoning criteria in many municipalities, this would therefore be considered to result in a land development of 73 % cover, in excess of many (but not all) local densities. However, this is not of any concern in this example, nor does it influence the resultant comparison of methods. The site is assumed to be on a gently sloping (3%) landscape, with the building on the upper portion and the parking positioned down-slope, with the balance of the parcel below the parking (Figures 10a and 10b). Detention Design The land development plan will include the construction of a surface detention basin, situated in the rear of the parcel, requiring the removal of remaining woodland. Both roof drains and surface runoff from most of the new impervious pavement will be collected and conveyed to this location by storm inlets and sewers. The plan of the basin will be constrained by the available remnant land (0.8 acres). Any dimensional setbacks required by municipal code will not be considered. For peak attenuation design, the volume of the basin will be sized to hold sufficient runoff from the site so that there is no increase in the peak rate of runoff from the parcel following development. The application of the TR-20 Model to this site produces a set of hydrographs similar to Figure 5, based on Tables 2 and 3. The volume of the detention basin required can also be roughly estimated at 62% of the net increase in runoff volume during the 100-year frequency rainfall of 7.3 inches in 24 hours (6.59 runoff). Including only the structure (1 acre) and the pavement (1.2 acres) and estimating the pond itself at 0.5 acres with the balance remaining woodland, this would represent: [(6.59 Post-developed) (2.61 Pre-developed)] x 1 ft/12 x 2.7 acres = 0.9 Acre-Feet (AF), then taking 62% of 0.9 AF results in 0.55 AF. Given the site limits and ignoring any modifications to the basin design for water quality mitigation (forebays, etc), the basin would be about 0.5 acre in plan and 3 feet deep, with 1 foot of net storage. A smaller basin would be correspondingly deeper to provide the same storage. 10
TABLE 1. Storm Frequency in the Northern Piedmont for 24-hour Duration Rainfall Storm Frequency (years) Rainfall (inches) 1 2.6 2 3.3 5 4.2 10 5.0 25 5.7 50 6.3 100 7.3 Figure 9. Rainfall Event Distribution in Piedmont Watersheds (Cahill, 1997) 11
Figure 10a. Site Design for Hydrologic Model: 3-Acre Parcel with Building (1 acre), Pavement (1.2 acres) and Detention Basin (0.5 acres). Figure 10b. Site Design for Hydrologic Model: 3-Acre Parcel with Building (1 acre), Pavement (1.2 acres) and Remnant Woodland (0.8 acres) 12
TABLE 2. Site Characteristics Parcel 3.0 acres, Existing cover woodland Proposed impervious cover 2.2 acres (2.7 incl. Basin), Soil - HSG B, Ia = 1.45 inches Pre-development Post-development HSG Existing Cover CN Area (acres) Developed Cover Area CN B Woodland 58 3 Building 1 98 Pavement 1.2 98 Basin 0.5 98 Woodland 0.3 58 Weighted CN 94 TABLE 3. Cover Complex Analysis for Detention Basin Design 3-acre wooded parcel 1 acre building, 1.2 acre pavement, 0.5 acre basin Pre-development conditions Post-development Conditions HSG soil Group B Wooded 90% impervious CN = 58 CN = 94 S = (1000/CN) - 10 = (1000/58) - 10 = 7.24 inches S = 0.638 Ia = 0.2S = 1.45 inches Ia = 0.128 inches Total runoff (Q) = (P -Ia) 2 / (P-Ia) + S Time of Concentration (Tc) = 0.2 hrs (12 min) Tc = 0.1 hrs. Peak discharge rate = Qp = Qu x A(sm) x Q (inches) Drainage area = 3 acres = 3/640 = 0.00469 sm Rainfall (P) Storm Total Ia/P Ratio Unit peak Peak Rate Total Total Ia/P Ratio unit peak Peak Rate Total Frequency Runoff (Q) discharge discharge Runoff Runoff (Q) discharge discharge Runoff inches/24 hrs (yrs) inches csm/in cfs Volume inches csm/in cfs Volume Qu (AF) off scale (AF) 3.3 2 0.38 0.439 500 0.88 0.09 2.64 0.039 1000 12.39 0.66 4.1 5 0.71 0.354 650 2.16 0.18 3.42 0.031 1000 16.05 0.86 5.2 10 1.28 0.279 710 4.26 0.32 4.51 0.025 1000 21.13 1.13 6.1 25 1.82 0.238 720 6.14 0.45 5.40 0.021 1000 25.31 1.35 6.6 50 2.14 0.220 740 7.43 0.54 5.89 0.019 1000 27.63 1.47 7.3 100 2.61 0.199 750 9.20 0.65 6.59 0.018 1000 30.89 1.65 Thus the resultant stormwater management system would consist of a detention basin about 0.5 acres in size, positioned in what had been the remaining woodland in the rear of the parcel, with a release structure discharging to some element of surface drainage in the woodland. There are many measures that can be added to soften this impact, such as adding a smaller pre-detention structure that collects the initial flush of runoff (usually estimated at 0.1 inches to 0.5 inches), and which is assumed to contain the greater portion of NPS pollutants. This small basin must be hydraulically disconnected from the larger basin (flowing in a parallel configuration) in order to 13
assure that accumulated or deposited pollutants will not be scoured during subsequent runoff events, and must be maintained. The basin itself could also be designed with permanent storage and planted with wetland vegetation, to provide some pollutant removal and possible uptake. In order for this mechanism to be effective, the detention times for storage would need to be significantly increased, resulting in a larger basin (but not a deeper basin). A variety of structural measures could also be installed in the flow pathway and conveyance piping, including devices to collect sediment and debris, or even filter the stormwater. For this example, these measures are not considered. Infiltration Design For the same building program, it is proposed that the stormwater management system include a 1-acre sub-surface storage and infiltration bed, situated beneath the proposed parking lot as shown in Figure 10b. In order to reduce impermeable surfaces, facilitate inflow to this bed, and eliminate some of the storm sewer infrastructure, a porous pavement will be used in most of the parking area, although the entrance driveway will remain as an impervious pavement. The basic design of a porous pavement with a groundwater recharge bed (Cahill, et al, 1988) is illustrated in Figure 11. The system could also be designed with standard pavement utilizing inlets to convey runoff into the underlying infiltration bed. By elimination of a detention basin, a larger portion of the original wooded parcel remains in natural cover and continues to infiltrate quite efficiently. Thus it is essential that the type of stormwater management system be selected prior to any site plan development, so that disturbance can be held to a minimum. The rain that falls on the rooftop will be conveyed to the storage/infiltration bed beneath the parking lot, with traps placed in the lines for collection of any detritus from the gutters or roof. The rain that falls on the porous pavement will drain quickly through the surface (or run to the inlets) and enter the same stone storage bed. The combined inflow (Figure 12, shown in units of acre-inches) will be distributed over the bed bottom and soak in, at the rate of soil infiltration. As the storm begins, the rate of inflow to the bed will be well within the soil infiltration rate, and so there will be very little volume in storage. As the inflow rate increases with our standard design storm (S- curve, see Figure 4), the rate of inflow will exceed the bed bottom exfiltration (Figures 13 and 14), and storage will begin to build within the bed. During the peak of the storm (hour 11 to 12), the available storage capacity may be reached, and under extreme storm events an overflow discharge may occur from the overflow control box, depending on the outlet invert. Normal design standards require that the overflow invert be set approximately 8 inches (or more) below the finished pavement surface. In all designs, the release or overflow hydrograph will not exceed the rate of runoff from the original undeveloped site, thus satisfying the general criteria of peak rate attenuation, while also maintaining groundwater recharge and greatly reducing NPS pollutant transport. Table 4 presents the detailed run of the Infiltration Model for the well-drained B soil condition (2 inches per hour), and Table 5 summarizes the bed operation with a moderate to poorly drained C soil (0.5 inches per hour). The Definitions and Parameters section and the labels explain the values and assumptions made in the Model, but the best way to explain the changes in values is with a set of graphics that show how the various elements of the bed function during rainfall. 14
Figure 11. Typical Section of Porous Pavement with Groundwater Recharge Bed 18 Cumulative Inflow to Bed (acre-inches) 16 14 12 10 8 6 4 2 2 YR 100 YR 0 0 4 8 12 16 20 24 Time (Hrs) Figure 12. Rainfall Inflow to Recharge Bed over 24-hour Period 15
8 7 Cumulative Volume (ac-in) 6 5 4 3 2 Inflow Infiltration 1 0 0 4 8 12 16 20 24 Time (hrs) Figure 13. Cumulative Inflow and Bed Infiltration over Time 2 year storm with Moderate to Poorly Drained Soils 18 16 Cumulative Volume (ac-in) 14 12 10 8 6 4 2 Inflow Infiltration 0 0 4 8 12 16 20 24 Time (hrs) Figure 14. Cumulative Inflow and Bed Infiltration over Time 100 year storm with Moderate to Poorly Drained Soils 16
17
18
The best visualization of this infiltration system is to begin with the conventional detention model. If one envisions the conventional detention basin as a bathtub filling with runoff, the infiltration model is a tub with holes in the bottom, so that two distinct outlets are possible under various flow conditions. The major difference is that for the detention model, the structure fills as the storm runoff inflow occurs, with overflow to discharge at a rate no greater than the predevelopment runoff rate of the 100-year frequency rainfall. For the infiltration model, the increasing inflow is continuously offset at the basin bottom outflow by soil infiltration (or bed exfiltration), so that volume storage does not occur until an excess inflow begins (inflow exceeds infiltration), and then surface release does not occur until both infiltration and storage capacity are exceeded (Figure 16). An infiltration bed designed for the 2-year frequency volume will also mitigate the peak rate of the 100-year rainfall, and for most designs will be far less. This Model uses the same general format as applied for detention designs, with infiltration equation and storage function modifications included, and the variable demand on bed volume storage expressed in terms of bed depth. The final design bed depth is usually not selected until the initial Infiltration Model run is complete, unless site constraints limit this dimension. Most systems use a depth of stone storage of at least 24 inches, or a total bed of about 36 inches below finished grade. This bed depth is frequently selected independent of storage volume requirements, so that the soil mantle is below local frost penetration depths, a concern in cold winter locations. In terms of volume criteria, bed depth can usually be designed for 18 inches or less, but is driven by the amount of external impervious surfaces (rooftops and roadways) that are conveyed to the bed. The ratio of impervious surfaces to available recharge bed bottom is frequently 4:1 or more, rather than the 2.2:1 ratio illustrated here. Some intense development plans have resulted in ratios of 7:1 or more, where good soil and open space is limited. In these designs the bed will usually have some overflow or release during the major rainfall events, but in no case are they allowed to exceed the peak rate of runoff during pre-development (Figure 17). Where soil conditions are less suitable, the fact is that impervious cover impacts are less severe, since the net change in hydrologic response is less. This may seem illogical, but if we build in locations that are presently poorly drained, then the impact of our development produces less of an increase in runoff. The other negative effects, however, such as riparian encroachment and more direct opportunity for NPS pollutants to enter surface waters, greatly increase as we attempt to build in low lying, wet locations. 19
Depth of Water in Stone Bed (in) 20 16 12 8 4 0 0 4 8 12 16 20 24 Time (hr) 2-year 100-year Figure 15. Storage Depth in the Stone Bed Assuming 40% Void Space Moderate to Poorly Drained Soils Depth of Water in Stone Bed (in) 20 16 12 8 4 0 0 4 8 12 16 20 24 Time (hr) 2-year 100-year Figure 16. Storage Depth in the Stone Bed Assuming 40% Void Space Well Drained Soils 20
Figure 17. Comparison of Detention vs. Infiltration Design Systems CONCLUSION Infiltration technology will evolve as the primary method of stormwater management in the near future, as increased stormwater volume and NPS pollutant transport are recognized as critical design criteria. Where infiltration systems are simply not feasible, as may be the case in highly disturbed and compacted (or contaminated) soils, or where development is situated in low lying, wet soils, other technologies will need to be applied. These include stormwater storage as a part of the developed structure, such as in the form of green roof storage systems, or sub-surface vaults and chambers. Vegetated roof systems offer the opportunity to accomplish significant evapotranspiration and so are consistent with the concept of sustaining the hydrologic cycle, while the storage chambers (frequently considered as irrigation storage) can be expensive but functional additions to the site plan. In any case, the current efforts to build a better detention basin (wet basins, pre-treatment units, etc.) will not be adequate for future stormwater management, and a significant reconstruction and retrofitting program will gradually evolve in most urbanized watersheds. 21
REFERENCES Cahill Associates, 1997. A Model Program to Balance Water Resources and Land Development in the French and Pickering Creeks Watershed, Chester County, Pennsylvania. Green Valleys Assn., Pottstown, PA, Jan., 1997 Cahill, et al, 1991. GIS Analysis of Nonpoint Source Pollution in the New Jersey Coastal Zone. T. H. Cahill, M. Adams, C. L. Smith, and J. S. McGuire, Cahill Assoc. and S. Whitney and S. Halsey, NJDEP, Div. Of Coastal Resources, Natl. Conf. On Integrated Water Information Mgmt., Atlantic City, NJ, Aug. 8, 1991 Cahill, et al, 1988. The Use of Porous Pavement for Groundwater Recharge in Stormwater Management Systems. T. H. Cahill, M. Adams, and W. R. Horner, Floodplain/Stormwater Mgmt. Symposium, PA State Univ., State College, PA, Oct. 1988. Cahill, T. H., P. Imperato and P. K. Nebel, 1976. Magnitude and Sources of Non-Point Pollution in the Maumee River Basin, 19 th Conf. On Great Lakes Research, Univ. of Guelph, Kitchner, Ont., May, 1976. Cahill, T. H., J. Adams and D. Backer, 1975. The Importance of Diffuse Pollutants in River Chemistry, 19 th Conf. On Great Lakes Research, Univ. of Guelph, Kitchner, Ont., May, 1976. Cahill, T. H., P. Imperato and T. H. Hammer, 1975. Historical Trends in Water Quality in the Brandywine Basin. Tech. Paper No. 1, Tri-County Conservancy of the Brandywine, Chadds Ford, PA Maidment, D. R., 1992. Handbook of Hydrology, McGraw-Hill Inc., ISBN 0-07-039732-5 U.S. Department of Agriculture, Soil Conservation Service, 1986. Urban Hydrology for Small Watersheds. Technical Release 55. USEPA, 1993. Guidance Specifying Management Measures for Sources of Nonpoint Pollution in Coastal Waters, Sect. 6217(g) CZM Amdmt. Manual, US Environmental Protection Agency, Office of Water, Wash., DC, 840-B-92-002. 22