Hydrodynamic Separation of Particulate Matter Transported by Source Area Runoff

Hydrodynamic separation is a preliminary unit operation frequently utilized in wastewater and more recently in storm water for separation of coarse particulate matter (PM) and gross solids. In order to examine the behavior and separation mechanisms of a screened hydrodynamic separator (HS) not influenced by scour, this study examined the event-based performance of an empty-bed (clean sump and volute) HS for PM fractions transported in eight runoff events from a 1,088?m2 paved source area urban watershed. Influent particle-size distributions (PSDs) {d50?m from 270?to?2,202?μm} and HS particle separation efficiency (PSE) (from 38 to 70% of mass) exhibited variations influenced by hydrology and previous loadings. When examined as PM size fractions, results demonstrate separation of the sediment fraction (>75?μm) ranging from 76 to 94% while for settleable and suspended (1–25?μm) fractions, the PSE was variable and significantly lower; from 3 to 57% and 2 to 43%, respectively. Results demonstrate a correlation between higher influent PM concentration and coarser PSDs, illustrating why higher PM concentrations promote higher PSEs; and why HS performance must be specified at a PM concentration, PSD, and flow rate. Results demonstrate that HS behavior is influenced by influent PSDs coupled with flow rate. Hydrodynamic separation is effective for high-rate gross solids control. However, current HS designs require incorporation of hydrologic control, methods of frequent sludge zone management before scour, and stored runoff management to control interevent redox conditions.     

Experimental and Field Studies of Type I Settling for Particulate Matter Transported by Urban Runoff

Settling velocity is an important constitutive parameter of particulate matter (PM) transported by runoff. Settling velocity is either explicitly or implicitly utilized when designing or modeling unit operations, and in situ or watershed controls for urban rainfall-runoff. Utilizing two common settling devices, a settling column and an Imhoff cone, settling velocities of discrete noncolloidal particles in source area urban rainfall-runoff were measured. A comparison of settling models applicable to discrete (Type I) PM settling was developed. Models were compared to measured results across the noncohesive silt- and sand-size PM gradation from 2 to 2,000?μm, utilizing measured particle-size distributions (PSDs) and specific gravity. Results indicate that Newton’s Law can reproduce measured settling velocity when measured inputs of PM diameter, specific gravity, and temperature are utilized. Alternative models to Newton’s Law (in the Stokesian regime) did not improve agreement with measured settling velocities determined using PSDs from laser diffraction. Settling velocity distributions using Newton’s Law were applied for two limiting classes of storm events loading a screened hydrodynamic separator (HS) at an urban watershed. Results indicate that for a low flow and high flow event, Newton’s Law and a simple ideal overflow model of the HS could reproduce PM separation and the PSD of eluted PM (2 to ～ 250?μm) within 17% of measured results on a gravimetric basis.     

Modeling Volumetric Clarifying Filtration of Particulate Matter for Transient Rainfall-Runoff Loadings

Volumetric clarification incorporating filtration with engineered granular media is increasingly used as a viable combined unit operation for separation of rainfall-runoff particulate matter (PM). Such combined unit operations are typically operated at the catchment-level for rainfall-runoff clarification of transient loadings, in contrast to centralized watershed or sewershed regional treatment. Using computational fluid dynamics (CFD), this study models the PM separation by a volumetric clarifying filter (VCF) subject to unsteady event-based hydrologic, hydraulic, mass, and particle size distribution (PSD) loadings. Modeled and measured physical model results indicate that the VCF is capable of PM load reductions and effluent concentrations at or below 30 mg/L. These results, with PM measured as suspended sediment concentration (SSC) represent reductions ranging from 83 to 97% on an event basis. CFD model results predict effluent PM mass loads. Modeled effluent median particle diameter (d50m), as an index for the filtered effluent PSD, reproduces the d50?m from the VCF physical model on an event basis. Filter head loss is examined as a function of flow rate. Despite geometric asymmetry of the multiple radial cartridge configuration tested, hydraulic loading for each individual cartridge is relatively uniform.     

Transport and Distribution of Particulate Matter Phosphorus Fractions in Rainfall-Runoff from Roadway Source Areas

During runoff transport, phosphorus (P) partitions between dissolved and particulate matter (PM) phases. PM-based P distributes across the particle-size distribution (PSD). This study investigates the transport and distribution for P and PM in runoff from a fully paved highway watershed in Baton Rouge, La. Eight events with discrete manual runoff sampling are studied. PSDs are modeled with a cumulative gamma distribution and PM-based P distributions are modeled with a Freundlich-type power law. P and PM fractions examined are dissolved, suspended, settleable, and sediment. Measured mass transport of these fractions is modeled based on flow-limited (zero-order) or mass-limited (first-order) delivery. Results demonstrate that transport of each fraction can be represented by these limiting categories, but fractions illustrate differing elution rates during the same event. Event-based signatures for PM or P are controlled by the fraction that dominates the transported mass. Even for small source area catchments such as roadways without complex flow patterns, where first-order transport should dominate, transport of P and PM fractions is not consistently first-order; exceptions are mainly dissolved and suspended fractions. A water quality volume (WQV) for 25 mm of runoff resulted in 100% capture for all fractions of seven events and significant bypass for all fractions of a single event with a 1-year return frequency. By contrast, a WQV of 5 mm of runoff resulted in significant bypass for most fractions for seven events and 100% capture for the single event of the lowest runoff volume.     

Volumetric Clarifying Filtration of Urban Source Area Rainfall Runoff

Volumetric clarification is a common storm-water unit operation for hydrologic attenuation that couples particulate matter (PM) separation. Recent volumetric clarification can also include integrated filtration. This study examines the unsteady hydraulic and head loss response of a volumetric clarifying filter (VCF) system to urban source area hydrologic loadings in Baton Rouge, La for 19 fully captured events. The rainfall-runoff response of the 1,088?m2 paved watershed is examined as a direct VCF loading. Watershed responses yielded two classes of behavior; high volume events with an equilibrium volumetric runoff coefficient from 0.6–0.8 while low volume events were 0.4–0.6. Runoff PM as suspended sediment concentration (SSC) yielded coarse heterodisperse influent particle-size distributions (PSDs); transformed to finer and more monodisperse PSDs after treatment. While event-mean head loss is less than 25 mm, instantaneous values up to 200 mm were dependent on instantaneous flow to the filters. Without backwashing, filter ripening head loss is small due to the coarse uniform filter media and radial filter configuration, with a loss of 2% porosity across the series of 19 events. Despite filter ripening an Ergun model was capable of predicting head loss across the entire flow rate range. Head loss and flow frequency distributions were exponential. Results indicate that a volumetric clarifier, filter geometry, and engineered media combination are capable of reducing effluent SSC to <30?mg/L through serial mechanisms of sedimentation followed by filtration.     

Adsorption Kinetics for Urban Rainfall-Runoff Metals by Composite Oxide-Coated Polymeric Media

A manganese oxide-coated polymeric media (MOPM) utilized in sorptive filtration systems as a rainfall-runoff or snowmelt unit operation and process media was characterized using scanning electronic microscopy and adsorption kinetics were studied using a flow-through batch reactor. Results indicated the MOPM adsorption kinetics can be described as a fast adsorption reaction occurring within 30?min followed by a slower reaction that continued from 5?to?15?h, as a function of initial pH and sorbent dosages. A potential driving kinetic model was developed based on an elementary second-order rate law. Modeled results were compared to experimental data using this model and a series of comparative kinetic models. Manganese oxide surface morphology and the ability of a parabolic diffusion model to predict the adsorption kinetics of MOPM suggest diffusion-controlled adsorption for divalent heavy metals on MOPM. Based on a goodness of fit test, the potential driving model best represented the experimental data. Using the potential driving model, it was found that rate constants increased with increasing solution pH, but were independent of sorbent dosages. Results indicated that metal ions with the highest adsorption affinity had the highest rate constants. Observed porosity, the excellent fit of the potential driving model, and breaks in Elovich model plots all suggest a complex adsorption mechanism. Results suggest MOPM can be an effective media for rainfall-runoff and snowmelt metal adsorption.     

Attenuation of Roadway-Derived Heavy Metals by Wood Chips

Wood chips were evaluated for their ability to attenuate heavy metals in roadway runoff. Column experiments with controlled synthetic runoff composition and flow rate were used to assess effects of flow rate (intercepted sheetflow from a 3-m wide roadway section), runoff salt concentration, wood exposure to alternating wetting and drying cycles, wood aging, competition among dissolved heavy metals, and removal of particle-associated heavy metals. Overall, wood chips damped the “pulse” of copper in the synthetic runoff such that the effluent was characterized by lower concentrations (3–25% of input) over longer periods of time, but with little retention of the total copper mass. The most effective treatment was wood chips aged up to 9 months. Increased aging and chip water content reduced effluent concentrations, relative to no treatment. Flow rate had no effect on effluent concentrations. The presence of salt (>2?mS/cm) or dissolved lead (500?μg/L) in the runoff caused greater copper effluent concentrations than the no treatment case. Removal of suspended particles (and associated contaminants) was greater than 85% with an estimated capacity of 0.16?g/gwood. Field evaluation with concentrated flow to a gutter containing a wood chip treatment showed little effect on total or dissolved copper and zinc runoff concentrations and indicated that wood chips may be a source of contaminants in subsequent storm events. Applications of wood chips to treat roadway runoff would not provide a significant decrease in total maximum daily load contributions (e.g., kg/d); however, there may be some scenarios for which wood chip treatments to decrease peak storm water concentrations of dissolved heavy metals in sheetflow runoff is desirable.     

Distribution of Particulate-Bound Metals for Source Area Snow in the Lake Tahoe Watershed

The porous matrix and long residence time of snow in traffic corridors can result in the accretion of particulate matter (PM) and metals. This temporary repository contributes PM-based and dissolved metals to surrounding environs during snowmelt. This study focused on distribution and settling of PM-based metals (Al, As, Cd, Cr, Cu, Fe, Pb, and Zn) in snow. Snow was sampled from six sites during four winter seasons. PM-based metals are examined herein as a function of PM granulometry, specifically particle-size distributions (PSDs) and PM surface area. Cumulative metal mass distributions across each PSD are modeled as gamma functions. Results indicate Al (15 g/kg) and Fe (4.2 g/kg) are the highest PM-based concentrations; Cd (0.18 mg/kg) and As (4.9 mg/kg) are the lowest. The PM size (d50?m) associated with the median metal mass ranges from 179 to 542?μm. The constitutive gamma results are integrated with Hazen’s settling algorithm to model PM-bound metal separation for a sedimentation basin at a local snow storage site. Flows are modeled in the storm water management model (SWMM) from snowmelt and historical rainfall time series. Results indicate that Type I sedimentation is capable of separating the sediment fraction (>75?μm) and majority of metal mass. While the basin is effective at separation of coarser PM-based metals, additional practices such as pavement and drainage appurtenance cleaning, as well as adsorptive-filtration can further manage suspended PM and metals, as well as dissolved metals.     

Kinematic Wave Model of Urban Pavement Rainfall-Runoff Subject to Traffic Loadings

Rainfall-runoff quantity and quality relationships are impacted by both the built environment in particular “impervious” paved surfaces and anthropogenic activities such as traffic. Through the capture, analysis, and modeling of eleven discrete rainfall-runoff events, the impacts of the paved urban surface and traffic were examined with respect to the temporal delivery of storm water runoff quantity. A kinematic wave model accurately captured the significant aspects of typical urban runoff events such as time to peak, total volume of flow and peak discharge from a 300-m2 paved surface subject to traffic. Abstractions associated with traffic, represented as the volume-based runoff coefficient, were estimated based on the relationship between runoff and vehicular traffic. It was found that for high intensity storms, with less than 10 vehicles/L of runoff volume (VRV), the runoff coefficient asymptotically approached a maximum value between 0.6 and 0.9. For low intensity storms, with more than 10 vehicles/L (as VRV), the runoff coefficient asymptotically approached a lower maximum value between 0.2 and 0.4. The kinematic wave theory also gave predictions of the time of concentration that were more accurate than other, more common methods currently in use including those by the FAA and the Soil Conservation Service. Prediction of the rainfall-runoff process impacted by the built environment and traffic permits determination of urban pavement hydrographs to determine the unsteady loadings of in situ treatment strategies under a variety of storm conditions. Such unsteady loadings are necessary inputs for selection, design and analyses of in situ storm water unit operations and processes that are developing for the control of both urban runoff quantity and quality.     

Corrosion Deterioration in Consumer Battery Litter

Recent studies have shown that the annual rates of consumer battery litter on urban pavements can be as high as 215 batteries per hectare of pavement and 0.4 batteries per meter of curb. As littered batteries deteriorate, they release components (Ag, Ba, Cd, Cr, Cu, Hg, Li, Mn, Na, Ni, Pb, Ti, and Zn) that can be significant sources of storm-water contamination. Results of ambient environmental and laboratory-accelerated corrosion studies are presented to quantify the mechanisms that yield chemical deterioration of littered batteries. The analysis concentrates on AA size alkaline and zinc chloride/zinc carbon (ZnCl/ZnC) cells since these are the most commonly littered batteries. Results indicate that littered batteries exposed to ambient well-drained environmental conditions for more than two months will develop surface corrosion over most of their surface but fewer than 10% will be ruptured by corrosion within the first 6 months. This agrees well with field observations. Exposure under poorly drained conditions yields more rapid deterioration but it requires exposure to more aggressive conditions such as those produced by road salts to reproduce the high degree of deterioration observed in some battery litter.     

Zinc Pollution Potential of Consumer Battery Litter

Consumer batteries in urban litter can be significant sources of storm-water pollution. Recent studies have reported annual urban battery litter rates as high as 215 batteries per hectare of pavement and 0.4 batteries per meter of street curb. On average, 75% of these batteries are leaking or have already discharged their internal reactants. However, many battery sizes, brands, and power chemistries are littered and this diversity makes it difficult to quantify their cumulative pollution potential. The amount of zinc released from AA size alkaline and zinc chloride/zinc carbon (ZnCl/ZnC) batteries which account for approximately 90% of urban battery litter is examined. Results are presented for batch rupture release extractions of 52 alkaline battery products yielding zinc releases of 20–40 mg/L and 57 ZnCl/ZnC products yielding zinc releases of 400–1,400 mg/L. Results of continuous flow column extractions are also presented to gauge releases following initial battery rupture. Disassembly analyses are used to bound the total zinc release potential of common battery products. Results indicate that ZnCl/ZnC batteries release more zinc when they are first ruptured, but if deterioration is complete, alkaline batteries can release approximately 25% more zinc. Therefore, the relative importance of these two classes of batteries depends on site-specific factors such as the proportion of each in litter, battery deterioration rates, and the length of time that battery litter remains unremediated by maintenance.     

Process Modeling of Storm-Water Flow in a Bioretention Cell

A two-dimensional variable saturated flow model was developed to simulate subsurface flow in bioretention facilities employing the Richards’ equation. Variable hydrologic performances of bioretention are evaluated using the underdrain outflow hydrographs, outflow volumes for 10 storms with various duration and depth, and flow duration curves for 25 different storms. The effects of some important design parameters and elements are tested, including media type, surrounding soils, initial water content, ratio of drainage area to bioretention surface area, and ratio of cell length to width. Model results indicate that the outflow volume via underdrain is less than the inflow; the flow peak is significantly reduced and delayed. Underdrain outflow volume from loamy sand media (with larger Ks) is larger than that from sandy clay loam media. The saturated hydraulic conductivity, storage capacity, and exfiltration into surrounding soils contribute to the hydrologic performance of a bioretention cell. Initial media storage capacity is affected by the hydraulic properties of media soils, initial water content, and bioretention surface area. The exfiltration volume is determined by the surrounding soil type and exfiltration area, dominated by flow through the bottom of the media.     

Volumetric Filtration of Rainfall Runoff. I: Event-Based Separation of Particulate Matter

As a unit operation, filtration generally requires flow equalization and primary clarification. This study examines the separation of runoff particulate matter (PM) in a volumetric clarifying filter (VCF). The VCF is a detention/retention vault integrating filtration after sedimentation. A paved source area (1,088?m2) directly loaded the vault (4.2?m3) with five radial filters (4?m2 of filtration surface area). PM separation was examined for 19 runoff events through monitoring of influent and effluent granulometric fractions. During the monitoring phase no maintenance was conducted and subsequent to the 19 events a measured material balance of the sedimentation vault and the radial cartridge filters generated a 94% recovery of PM. During 5 months of monitoring and PM mass, suspended sediment concentration (SSC) was reduced from 334 to 34 mg/L (90% reduction) with effluent sediment (>75?μm) of <3?mg/L, settleable of 14 mg/L, suspended PM (<25?μm) of 17 mg/L as event mean concentrations; with turbidity reduced from 96 to 23 NTU (76% reduction). Based on separate PM recovery from the vault and filters, 77% of the PM separation was sedimentation in the vault and 23% as filtration. Captured particle-size distributions are heterodisperse with a d50?m of 300 μm in the vault and a d50?m that ranged from 34 to 63 μm with filter depth. Filter forensics indicated PM capture was nonuniform, with the bottom and middle most heavily loaded by PM as compared to the upper third of the filter. While paired testing of automatic and manual sampling produced similar median effluent SSC, automatic sampling significantly misrepresented the median influent PM as SSC (p ? α = 0.05).     

Comparison of Sorptive Filter Media for Treatment of Metals in Runoff

Rainfall runoff and snowmelt impacted by anthropogenic activities can transport significant loads of metals. Ecological concerns and recent regulatory guidance have spurred development of unit operations such as ex situ sorptive filters and engineered media infiltration systems with the intent of including sorption mechanisms for metals as compared to conventional filter media. Applications of sorptive media for rainfall or snow unit operations include infiltration systems, sorptive clarifiers, separation systems, and deformable, cartridge, or tubular filters. Column breakthrough experiments were conducted for selected sorptive filter media and compared to conventional filter media. Comparing plain sand, granular activated carbon, and cementitious media to oxide coated/admixture media, manganese oxide coated media (MOCM) had the best overall operational behavior with 10% breakthrough bed volumes (Vb), breakthrough capacity (X/Mb), and exhaustion capacity (X/Mexh) two times higher than those of iron oxide coated sand (IOCS). As the empty bed contact time (EBCT) for MOCM increased from 0.5?to?1.1?min; the values of Vb, X/Mb, and X/M increased by a factor of 2. Compared to metal breakthrough for uncoated sand or polymeric media, manganese oxide polymeric media (MOPM) as well as bench scale partial exfiltration reactor media (combining uncoated cementitious media and oxide coated media) provided significant capacities for Pb, Cu, Cd, and Zn. Removal mechanisms for MOCM include adsorption, surface complexation, ion exchange, and filtration, accounting for MOCM’s high capacity. Although uncoated cementitious media also had a significant capacity for metals through precipitation and filtration, breakthrough instability of metal precipitates and high effluent pH can limit application in monomedium applications.     

Integrated Storm-Water Management for Watershed Sustainability

One aspect of integrated watershed management evaluates the impact of development on the local hydrologic cycle and, in particular, drinking water, wastewater, and storm-water infrastructure. Sustainable storm-water management focuses on selecting storm-water controls based on an understanding of the problems in local receiving waters that result from runoff discharges. For example, long-term problems associated with accumulations of pollutants in water bodies include sedimentation in conveyance systems and receiving waters, nuisance algal growths, inedible fish, undrinkable water, and shifts to less sensitive aquatic organisms. Short-term problems associated with high pollutant concentrations or frequent high flows (event-related) include swimming beach closures, water quality violations, property damage from increased flooding, and habitat destruction. A wide variety of individual storm-water controls usually must be combined to form a comprehensive wet weather management strategy. Unfortunately, combinations of controls are difficult to analyze. This will require new modeling techniques that can effectively evaluate a wide variety of control practices and land uses, while at the same time ensure that the flood-control objectives also are met. The results of these new models and novel techniques used for storm-water control then can be incorporated into an evaluation of the urban water cycle for a specific service area to determine whether storm-water controls can provide additional benefits such as reduction of potable water use and reduction of sanitary sewer overflow events.     

Water Quality Improvement through Reductions of Pollutant Loads Using Bioretention

As an increasingly adopted storm-water best management practice (BMP) to remedy hydrology and water quality impairment from urban development, bioretention facilities need rigorous investigation to quantify performance benefits and to allow design improvements. This study examines water quality improvements [total arsenic, total cadmium, chloride, total chromium, total and dissolved copper, E. coli, fecal coliform, lead, mercury, nitrogen species, oil and grease, phosphorus, total organic carbon (TOC), total suspended solids, and total zinc] via monitoring for a 15-month period at two bioretention cells in Maryland. Both bioretention cells effectively removed suspended solids, lead, and zinc from runoff through concentration reduction. Runoff volume reduction promotes pollutant mass removal and links BMP water quality benefits with hydrologic performance. From a load perspective (kg/ha?year), all but TOC at one cell showed pollutant reduction. Bioretention effluents exhibited good water quality for all significant pollutants except for nitrate, copper, and phosphorus in one cell, the latter two of which may be attributed to media organic matter dissolution. Copper dissolved/particulate analyses showed that significant changes in copper speciation behavior result from transport through the bioretention media.     

Pollutant Mass Flushing Characterization of Highway Stormwater Runoff from an Ultra-Urban Area

Water quality of highway stormwater runoff from an ultra-urban area was characterized by determining the event mean concentration (EMC) for several pollutants and by evaluating pollutant flushing. Thirty-two storm events were monitored between June 2002 and October 2003. Mean EMCs in mg/L were 0.035, 0.11, 0.22, 1.18, 420, 3.4, 0.14, 1.0, and 0.56 for Cd, Cu, Pb, Zn, total suspended solids (TSS), total Kjeldahl nitrogen (TKN), NO2–N, NO3–N, and TP. First flush as defined by flushing of 50% of the total pollutant mass load in the first 25% of the event runoff volume occurred in 33% of the storm events for NO2?, 27% for TP, 22% for NO3? and TKN, 21% for Cu, 17% for TSS, 14% for Zn, and 13% for Pb. Median values for the mass flushed in the first 25% of runoff volume were greater than the mass flushed in any 25% portion beyond the first for all pollutants. The mass in later 25% volume portions were greater than in the first 25% volume in at least 17% of the events for all pollutants, indicating that a significant amount of the pollutant load can be contained in later portions of the runoff volume. Nonetheless, management of the first 1.3?mm (1/2?in.) of runoff was able to capture 81–86% of the total pollutant mass.     

Evaluation and Optimization of Bioretention Media for Treatment of Urban Storm Water Runoff

Bioretention is a relatively new urban storm water best management practice. The objective of this study is to provide insight on media characteristics that control bioretention water management behavior. Eighteen bioretention columns and six existing bioretention facilities were evaluated employing synthetic runoff. In columns, the runoff infiltration rate through different media mixtures ranged from 0.28 to 8.15?cm/min at a fixed 15 cm head. For pollutant removals, the results showed excellent removal for oil/grease (>96%). Total lead removal (from 66 to >98%) decreased when the total suspended solids level in the effluent increased (removed from 29 to >96%). The removal efficiency of total phosphorus ranged widely (4–99%), apparently due to preferential flow patterns, and both nitrate and ammonium were moderate to poorly removed, with removals ranging from 1 to 43% and from 2 to 49%, respectively. Two more on-site experiments were conducted during a rainfall event to compare with laboratory investigation. For bioretention design, two media design profiles are proposed; >96%?TSS, >96%?O/G, >98%?lead, >70%?TP, >9%?nitrate, and >20%?ammonium removals are expected with these designs     

Sampling Issues in Urban Runoff Monitoring Programs: Composite versus Grab

Storm-water monitoring generally uses flow-weighted automatic composite samplers to collect a representative sample of an entire storm event. Automatic samplers are convenient but unfortunately they can be expensive, especially for temporary sampling needs or for short-term research projects. An alternative method is to use a series of grab samples. This paper examines the accuracy of event mean concentrations (EMCs) and mass first flush ratios calculated from a finite number of grab samples, and compares them to results from flow-weighted automatic samples. Both sampling techniques were evaluated using data collected from a three-year investigation of three highway sites. A large number of grab samples is needed to approach the accuracy and precision of flow weighted composite samples, and 30 grab samples per storm event generally estimated the EMCs within 20% average error. To detect a first flush, it is necessary to take even more grab samples or to adjust the timing of the sample collection toward the beginning of the storm. The superiority of automatic sampling for estimating EMCs for constituents compatible with automatic sampling is demonstrated.     

Prediction of Nutrient Concentrations in Urban Storm Water

Excessive quantities of nutrients in urban storm-water runoff can lead to problems such as eutrophication in receiving water bodies. Accurate process based models are difficult to construct due to the vast array of complex phenomena affecting nutrient concentrations. Furthermore, it is often impossible to successfully apply process based models to catchments with limited or no sampling. This has created the need for simple models capable of predicting nutrient concentrations at unmonitored catchments. In this study, simple statistical models were constructed to predict six different types of nutrients present in urban storm-water runoff: ammonia (NH3), nitrogen oxides (NOx), total Kjeldahl nitrogen, total nitrogen, dissolved phosphorus, and total phosphorus. Models were constructed using data from the United States, collected as a part of the Nationwide Urban Stormwater Program more than two decades ago. Comparison between the models revealed that regression models were generally more applicable than the simple estimates of mean concentration from homogeneous subsets, separated based upon land use or the metropolitan area. Regression models were generally more accurate and provided valuable insight into the most important processes influencing nutrient concentrations in urban storm-water runoff.