Environmental Review Toolkit
Water, Wetlands, and Wildlife

 Previous Home Next    

2. STORMWATER QUALITY MONITORING OVERVIEW

This chapter introduces stormwater monitoring, provides insight into complexities specific to highway monitoring, and discusses approaches to developing stormwater monitoring programs to meet specific management goals.

2.1. Physical and Chemical Characteristics of Stormwater Runoff

In this guidance manual, the term "stormwater" refers to more than just storm driven surface runoff. Here the term is expanded to cover water and other substances that are transported through stormwater conveyance systems during, after, and between storm events. In addition to the runoff from rainfall or snowmelt, a typical stormwater sample may contain materials that were dumped, leaked, spilled, or otherwise discharged into the conveyance system. The sample may also contain materials that settled out in the system toward the end of previous storms and were flushed out by high flows during the current event being sampled. Stormwater also can include dry weather flows such as pavement washing, pavement cutting wash water, irrigation, or base flows.

Stormwater quality tends to be extremely variable (USEPA, 1983; Driscoll et al., 1990). The intensity (that is, volume or mass of precipitation per unit time) of rainfall often varies irregularly and dramatically. These variations in rainfall intensity affect runoff rate, pollutant washoff rate, in channel flow rate, pollutant transport, sediment deposition and re-suspension, channel scour, and numerous other phenomena that collectively determine the pollutant concentrations, pollutant forms, and stormwater flow rate observed at a given monitoring location at any moment. In addition, the transitory and unpredictable nature of many pollutant sources and release mechanisms (e.g., spills, leaks, dumping, construction activity, landscape irrigation runoff, vehicle washing runoff), and differences in the time interval between storm events also contribute to inter storm variability. As a result, pollutant concentrations and other stormwater characteristics at a given location should be expected to fluctuate greatly during a single storm runoff event and from event to event.

Numerous studies conducted during the late 1970s and early 1980s showed that a potentially significant source of pollution is stormwater runoff from urban and industrial areas (USEPA, 1983; Driscoll et al., 1990). As a result, federal, state and local regulations have been promulgated to address stormwater quality.

Increasingly, the impacts of hydrologic and hydraulic changes in watersheds are being recognized as significant contributors to receiving waters not meeting beneficial criteria. These impacts include stream channel changes (erosion, sedimentation) as well as water level fluctuations in wetlands.

2.2. Stormwater Quality Monitoring Challenges

The primary purpose of a monitoring program should be to obtain information necessary to make sound resource management decisions. For example, a typical stormwater monitoring program may be intended to identify pollution problem areas and determine which problem(s) are the most significant. Monitoring results would then be used to develop control strategies and prepare plans and budget estimates for addressing those problems.

The principal challenge facing developers of stormwater monitoring programs in selecting appropriate flow and water quality sampling equipment is the great variability in stormwater pollutant concentrations, both temporal and spatial. Stormwater quality at a given location varies greatly both between storms and during a single storm event; therefore, a small number of samples are not likely to provide a reliable indication of stormwater quality at a given site. As a result, collection of numerous samples is generally needed for accurate characterization of stormwater quality at a site. Stormwater quality also tends to be quite variable from place to place, and may need to be monitored at a number of strategically located stations to characterize stormwater quality over a larger area. Consequently, selecting monitoring equipment and procedures involves the need to make decisions that balance the cost of obtaining more extensive and accurate information (and the increased reliability it provides) against the cost of implementing less intensive and possibly misdirected and/or ineffective control programs based on sparse data.

Experience has shown that it is generally expensive and time consuming to collect enough stormwater samples to answer many of the common stormwater quality questions (e.g., What are the water quality trends at a given location? Is a given BMP effective?) with a high level of statistical confidence. An entire program budget could be devoted to a monitoring effort to achieve a high confidence level, but doing so would leave insufficient resources for pollution control. Conversely, a poorly designed monitoring program could lead to erroneous conclusions and poor management decisions, resulting in misdirected or wasted resources (e.g., staff time, funds, credibility, and political support). Therefore, before a monitoring program is begun, it is critical to clearly identify and prioritize decisions that must be made, determine the type and quality of information needed to support those decisions, and then compare this list of needs to the resources available for monitoring. If the available resources cannot support the scale of monitoring needed to provide the quality of information deemed necessary, then the following options should be considered:

  • Evaluate alternative means for acquiring the information needed to support management decisions. For example, sediment sampling and analysis may be a cost effective alternative to water column monitoring in some situations, especially those in which the focus is on long term impacts of erosion and sedimentation or tracing specific pollutants to their respective sources.
  • Consider a phased approach that addresses only a subset of the overall geographic area, or only the most important stormwater questions, to obtain useful results within resource limitations (e.g., funds, personnel, time).
  • Utilize available data from other locations to support decision-making.

The key question should be "Will the information provided from the monitoring program under consideration (and capable of implementation) significantly improve my ability to make sound management decisions?" If the answer is no, the monitoring program should be reexamined.

2.3. Complexities Specific to the Near-highway Environment

Numerous difficulties are encountered in monitoring stormwater runoff in the near-highway environment. Complications in designing and selecting methods for monitoring highway runoff fall into two primary categories: operational constraints and physical constraints.

2.3.1. Operational Constraints

Although health and safety are always of primary concern in establishing a stormwater monitoring program in any environment, the near-highway environment presents significant safety hazards not encountered in many other locations. The repercussions of improper selection of equipment and procedures may present a serious risk to monitoring personnel as well as the general public. Health and safety risks related to site selection and the implications for equipment selection are discussed in Sections 3.1.2.

2.3.2. Physical Constraints

Although many of the conditions found in monitoring highway runoff are also found in many urban settings, a few physical conditions are particularly relevant to near-highway monitoring sites. Factors found at near-highway sites that directly affect site and equipment selection include:

1. Predominance of small- to medium-sized watersheds for monitoring

Depending on the drainage system selected, watersheds for highway sections typically range in size from less than 0.5 acre to around 50 acres. Larger systems often cannot be verified (i.e., mapped and checked for illicit or municipal connections) in a cost-effective manner, particularly in urban areas, and may include runoff from adjacent land uses.

2. Short time of concentration and "peaky" flow

Small, highly impervious watersheds that have minimal times of concentration can be quite difficult to monitor due to "peaky" flows (i.e., flows rise and fall directly in response to a rainfall). The range of flows that needs to be measured accurately is large. Monitoring sites where large changes in flow rate occur in a relatively short period of time require particular attention during equipment selection, installation and use. Rapidly changing flow conditions can cause:

  • equipment with poor data density recording capabilities to miss brief periods of significant flow;
  • automatic samplers to collect samples from flows that have changed since the sampler trigger was initiated;
  • automatic samplers to collect no sample due to low flow conditions after the trigger has occurred; and
  • errors in flow measurement due to unsteady conditions or flows below the minimum that can be measured.

The frequency of flow measurements must be of the same magnitude as the time of concentration for the watershed if flows are to be estimated accurately. Time of concentration is a function of the rainfall intensity and watershed size, and it decreases as intensity increases. Data density must be adequate to ensure that high intensity, short duration events are recorded. For small watersheds, this means that flow rate data may need to be collected at less than five-minute intervals.

3. Possible large percent error in watershed parameters

Flow calculations and loading estimates can be greatly affected by the accuracy of the measurement or estimate of watershed parameters. Particularly in low gradient highways, it may be difficult to estimate or measure (to less than 10% error) watershed parameters such as tributary drainage area and percent imperviousness. For this reason, equipment that is capable of measuring a wide range of flow rates (often in excess of three orders of magnitude) without large errors (<20%) is optimal.

Watersheds that are smaller than 1 acre, have minimal longitudinal slopes, and are in high traffic areas are common in the near-highway environment. These factors may complicate watershed delineation. Shallow longitudinal slope often means that very small rises in the pavement surface due to uneven slabs, cracks, settling, minor modifications to the pavement surface, rutting, broken curbs, and dikes can have a major impact on drainage patterns, and thus on watershed area estimations. The inability to accurately assess watershed area may also be compounded by safety concerns related to surveying highways that are in use.

In addition, it has been suggested that runoff coefficients for the Rational Method for some highways may be on the order of 0.6 to 0.8, which is lower than expected for highly impervious systems. This may be related to losses due to infiltration through cracks and joints and conversion of a portion of runoff into aerosols and spray from high-speed traffic (Caltrans, 2000).

4. Downstream access issues

It may be difficult to obtain access to flows exclusively from a highway section due to direct connection to municipal systems. Connections between highway and municipal systems are quite common in urban areas. It is important to determine whether there are any possible connections that would affect the tributary area and water quality.

5. Steep pipe slopes and high velocities

Another factor frequently encountered in the near-highway environment is steep pipe slopes resulting in high flow velocities in both open conveyance systems and pipes. Steep gradient conveyance conditions may, for example, be present in fill sections and in areas where the grade of the road is steep. These conditions require special attention during equipment selection and possible modifications to the conveyance to facilitate accurate flow records and sampling.

6. Numerous outfalls

Many highway systems do not drain large watersheds to a single tributary drainage point as is typically observed in municipal systems. Connections to receiving waters or conveyance systems may occur at numerous locations, (e.g., there may be one outfall per inlet). This may limit the total area that can be effectively monitored with a small number of monitoring stations, complicate the site selection process, and decrease the applicability of results to similar watersheds.

7. Lack of a well-defined drainage system

Many highway sections may not have a well-defined drainage system or may include sections that do not specifically drain to a separate storm sewer system. This is the case where curbs and dikes are not in place or ditches or unlined channels serve as the primary conveyance system.

8. Right-of-way issues

Inter-governmental agreements may need to be established to enable monitoring efforts on federal, municipal, or county property because in many cases access to conveyances in the established right-of-way may not be possible.

9. Representativeness of highway sections

Depending on the goals of the monitoring program, selection of highway sections may require information beyond standard watershed parameters (e.g., average daily traffic volumes, number of lane miles, existence of specific features such as sound walls, or cut or fill construction). It may be difficult to extrapolate results from watershed monitoring studies completed on a small number of sites to a larger highway system without site-specific information to enhance the representativeness of the sites monitored.

2.4. Regulations as Drivers for Monitoring Programs

A number of regulatory drivers exist for implementation of stormwater monitoring programs including:

  • the Clean Water Act: total maximum daily load (TMDL) and National pollutant discharge elimination system (NPDES) Phase I and II;
  • the Endangered Species Act; and
  • state, county, and local regulations.

Details about each of these regulations can be obtained from the U.S. Environmental Protection Agency (USEPA), state, county, and local resources.

Descriptions of current federal laws, regulations, and proposed rules can be found on the USEPA laws and regulations home page:

http://www.epa.gov/lawsregs/

2.5. Highway Stormwater Monitoring Goals

The stormwater monitoring goals should have a specific effect on the scope of any monitoring effort. The following sections examine the most common objectives for stormwater monitoring in the near-highway environment including specific guidance on monitoring approaches and equipment to assist in meeting the goals. The following potential objectives are discussed:

  • monitoring to estimate pollution concentrations and loads;
  • monitoring to identify stormwater pollutant sources;
  • monitoring to characterize stormwater quality trends;
  • monitoring to evaluate BMP performance; and
  • monitoring to assess compliance with surface water quality criteria.

Each section can help to develop the combination of monitoring locations, frequency, parameters, and methods that are best suited to specific data needs and resources.

2.5.1. Monitoring to Estimate Pollutant Concentrations and Loads

This section provides guidance on stormwater monitoring to estimate pollutant concentrations and loads. The sections below discuss the key considerations associated with program development and implementation.

2.5.1.1. Determine Objectives And Scope

Information on stormwater pollutant concentrations and pollutant loads may be used for a variety of purposes including:

  • participation in a watershed-based monitoring program or permitting effort;
  • watershed management planning;
  • source pollution assessment;
  • other stormwater management needs;
  • assessment of parameters of concern; and
  • calculation of pollutant loading for TMDL development or compliance.

Developing specific, realistic approaches to achieve monitoring objectives is essential.

2.5.1.2. Develop Monitoring Plan

The following sections provide guidance on developing a specific approach to achieving monitoring objectives.

Select Monitoring Locations

The number of locations to be monitored depends on specific program objectives, regulatory requirements (if applicable), the size and complexity of the drainage watersheds and conveyance system, and the budget allocated to monitoring. In addition, the frequency of sampling at each location should be considered.

While some programs sample at a minimum number of sites mandated by a regulatory program, others monitor additional locations because the results help support critical decisions in the stormwater management planning process. If the highways in the study area are large and complex and differ in design and other factors expected to affect water quality, there may not be sufficient resources to conduct a monitoring program that will allow development of reliable estimates of pollutant concentrations and loads for every outfall. For this reason, most programs collect data at a few selected stations and extrapolate these data to develop estimates of water quality and pollutant loads for a larger area.

The first step in applying this approach is to review watershed characteristics and drainage system information. Based on this review, locations that are representative of highway types and watersheds in the monitoring area should be selected.

Some programs use stations that monitor relatively small catchments that have fairly homogeneous characteristics (grade, material type, etc). Data may then be extrapolated to represent catchments within the project area that are believed to have similar sources and pollutant-generating mechanisms. Other programs use stations that sample relatively large highway catchments representing a composite area of highway sections and types. These stations are typically located in streams or other stormwater conveyances toward the lower end of a highway watershed to collect samples that are indicative of runoff quality from the larger area.

When using the extrapolation approach, it should be noted that although many previous studies [e.g., USEPA, 1983: Driscoll et al., 1990] have identified correlations between stormwater quality and watershed characteristics (e.g., land use), the correlations were not very strong. In addition, there are several factors that can result in different runoff water quality from the same land use. These factors include differences in watershed size, seepage, slope, and time of concentration. Adjacent land use and activities (e.g., being near metal smelters) can impact dustfall and consequently runoff water quality. Thus, extrapolation of data to represent other watershed locations may not provide the most realistic basis for estimating cumulative pollutant loads from the drainage system. However, despite its limitations, extrapolation of results from a small watershed to a larger area is often the chosen alternative and may represent the only viable approach.

Select Sampling Frequency

Because of the variation in the concentration of pollutants observed between storms, even at a single sampling station there will generally be a need to monitor at least five storm events to obtain reasonably representative results. A statistical analysis may be conducted to estimate how many events need to be monitored to achieve various confidence levels.

To perform a power analysis, one will need to determine the magnitude of the change desired to be detected; the confidence level; and the statistical power, or probability of detecting a difference. As a starting point, the confidence level and power should be set at 95% and 80% respectively; under these conditions, there is a 5% chance that a significant change will be reported where none exists, and a 20% chance that a significant change will be missed. The power analysis is discussed in detail in Appendix A.

The power analysis often shows that many samples would be needed to discern a small (e.g., 25%) change. In such cases, a determination should be made as to whether overall objectives can be met without detecting small changes. If available resources prohibit the frequent monitoring of all locations, then reducing the number of locations or parameters may be necessary. It is recommended that statistical confidence in the results of the monitoring program be considered of higher importance than collecting information at a larger number of locations or obtaining detailed analytical results for a large number of water quality parameters.

Select Parameters and Analytical Methods

Monitoring studies requiring estimates of pollutant loads and concentrations typically include the following parameters.

Conventional Parameters

Total Suspended Solids (TSS)
Total Dissolved Solids (TDS)
Biochemical Oxygen Demand (BOD)
Chemical Oxygen Demand (COD)
Total Hardness
Fecal Coliform Bacteria
Oil and Grease
Total Petroleum Hydrocarbons (TPH)
pH
Temperature

Nutrients

Total Kjeldahl Nitrogen (TKN-N)
Ammonia Nitrogen (NH3-N)
Nitrate+Nitrite (NO3-N + NO2-N)
Total Phosphate
Ortho-Phosphate

Heavy Metals (Total and Dissolved)

Cadmium (Cd)
Copper (Cu)
Lead (Pb)
Zinc (Zn)

The above list may serve as a starting point for establishing a monitoring program.

If significant monitoring has already been done in the area, consideration should be given to deleting some or all parameters that were shown to be consistently below levels of concern. It may be necessary to monitor additional parameters, depending on the regulatory framework of the monitoring program. Once the parameters for monitoring have been selected, the analytical detection limits needed to meet data quality objectives must be determined. For example, when comparing runoff water quality to acute aquatic criteria, low metals detection levels often are required.

Select Monitoring Methods

Flow weighted composite sampling is often preferred for all parameters, except those that are likely to transform rapidly (fecal coliform or other human pathogens) or adsorb to sample containers (oil and grease). Grab sampling is required for these parameters. The flow weighted composite samples are typically collected during the entire period of discharge if this is possible. If a large number of monitoring locations (>5) are needed and flow-weighted composites are desired, use of automated monitoring methods is recommended. Because stormwater quality can vary dramatically during a storm event, a single grab sample will not provide a good basis for estimating pollutant concentrations or loads. Most monitoring programs will probably need to either analyze a series of grab samples collected at intervals throughout the storm, or analyze a single flow weighted composite sample collected throughout the storm. The latter alternative is generally far less expensive than the former.

It is possible to collect flow weighted composite samples using manual methods, but this is generally impractical if there are more than a few stations to monitor. Moreover, manual monitoring can be more costly than automated monitoring if the program encompasses more than a few storm events. For these reasons, many monitoring programs have found that using automated monitoring equipment and methods is more appropriate than manual monitoring. Details about selecting specific monitoring equipment are provided in Section 3.

Select Storm Criteria

The application requirements for NPDES permits that require monitoring specify that "representative" storms must be monitored. As defined in the regulations, a "representative" storm must yield at least 0.1 inch of precipitation; must be preceded by at least 72 hours with less than 0.1 inch of precipitation; and, if possible, the total precipitation and duration should be within 50% of the average or median storm event for the area. Programs that are not part of the NPDES permit application process or in fulfillment of an NPDES permit may have other requirements.

In general, it is desirable to monitor a broad range of storm conditions rather than just "representative" storms. For example, in the Pacific Northwest it is often difficult (and rare) to identify storms where there has been a 72-hour dry period prior to the storm.

2.5.2. Monitoring to Identify Stormwater Pollutant Sources

This section provides guidance on stormwater monitoring to identify sources of stormwater pollutants. The following sections discuss the key considerations associated with each phase of plan development and implementation.

2.5.2.1. Determine Objectives and Scope

One objective of some stormwater monitoring programs is to obtain information on stormwater pollutant sources. Monitoring for source identification is usually performed after monitoring at a downstream area of a watershed or catchment has shown strong evidence of a water quality problem. Thus, source identification monitoring is often the second phase in a two phase monitoring program; the first phase may have involved one or more of the following:

  • monitoring to estimate pollutant concentrations and loads (discussed in Section 2.5.1);
  • sediment monitoring;
  • dry weather inspections; and
  • biological monitoring.

The scope of the source identification monitoring is usually based on the results of initial monitoring to estimate concentrations and/or loads. If the initial monitoring finds significant pollutant concentrations and/or loads, follow up monitoring may be required to identify and prioritize the sources. On the other hand, if the initial monitoring did not reveal any significant stormwater pollution, there may be little reason to conduct source identification monitoring.

The results of the previous studies that triggered the source identification study should be carefully reviewed. Reviewing the literature for information on typical sources of the observed pollutants is also recommended. Land use data should be examined for the catchment area(s) where the pollution was observed to identify potential sources. A visual survey of the area should be conducted to identify any obvious sources of the observed pollution. The findings of source identification studies in other areas should be reviewed, and local conditions and initial observations should be discussed among peers and stormwater professionals. In some cases, the likely source(s) of a particular problem may be easily identified. Follow up monitoring can be used to confirm initial findings.

Source identification monitoring programs may be used for a variety of purposes including:

  • compliance with permits;
  • watershed management planning;
  • non point source assessment;
  • source control; and
  • illicit connection identification.

2.5.2.2. Develop Monitoring Plan

The following sections provide guidance on developing a specific approach to identify stormwater pollutant sources.

Select Monitoring Locations

The number of locations to be monitored depends on the size and complexity of the drainage basin(s), the number of pollutant sources present, and funds allocated to monitoring. In addition, the frequency of sampling at each location should be considered.

The typical approach for selecting monitoring locations is summarized below.

  • Review the results of previous monitoring to identify the locations that had pollutant concentrations near or above their respective water quality criteria. List the specific pollutants of concern associated with each of these monitoring locations. Review the published literature to identify the typical source(s) for each pollutant of concern. Review available information for each catchment area and identify any likely source areas for the observed pollutants.
  • Conduct wet- and dry-weather inspections of the stormwater conveyance system in each catchment area with an identified water quality problem.
  • Conduct stormwater monitoring. Select monitoring locations upstream and downstream of the likely potential source areas identified in Step 1. If limited resources and/or logistical constraints make it impractical to monitor every location, monitor upstream and downstream locations that "bracket" a number of sources. Alternatively, rank the potential source areas with regard to pollutant concentrations/loads observed in prior sampling, proximity to sensitive receiving waters, and other factors that may indicate their relative importance. Then select monitoring locations according to their priority.
  • Monitor at each location. Generally, at least three or four storms will need to be monitored. However, if major differences in upstream and down-stream water quality are found during the first storm or two, additional monitoring may not be necessary.
  • Review the analytical results. Compare each station to its adjacent upstream station. If the downstream station exhibits higher concentrations and/or loads, this may indicate that pollutants are entering the channel between the two stations.
  • Visually inspect the channel segment between the two stations. If there is an outfall in the segment, determine what activities occur within the catchment that drain to that outfall or the contributing watershed and determine whether they may account for the observed pollutants. If a tributary drainage channel enters the main channel within the segment, visually inspect the tributary channel for potential sources and monitoring locations. Continue this process until the probable source(s) of the observed pollutant(s) have been identified.

In general, choose monitoring sites that are as close as possible to the suspected sources. This approach will reduce the chance that pollutants released from the source will be masked or diluted during monitoring. Additionally, select locations where sampling and flow measurement can be conveniently and safely obtained. If it is not possible to monitor all sites, select those locations that align with the highest priority information needs.

Select Sampling Frequency

Source identification monitoring does not typically require long term monitoring of numerous storms. In general, monitoring should be halted once the key source(s) of the pollutant(s) of concern are identified. Once a key source has been identified, the focus should be on implementing control measures rather than continued monitoring.

Some stormwater pollutant sources may be major contributors in one storm and minor contributors in the next. Therefore, a potential source should not be ruled out based on the results from a single storm event. To reduce the chance of overlooking significant sources, monitor at least three or four storms that encompass a range of conditions and seasons. Try to monitor during the early portions of the first storm that occurs after a prolonged dry period, as this often represents the "worst case" scenario. If there is a major source within a given drainage area, it will most likely be identified while the flow rate is increasing during the storm's early stages or during high intensity periods of the storm.

Select Parameters and Analytical Methods

As noted above, source identification monitoring is typically conducted only after stormwater pollution has been identified through prior monitoring. Source identification monitoring usually focuses on those pollutants measured near or above their respective levels of concern (e.g., water quality criteria) in previous samples. It is not generally necessary to monitor parameters that appear consistently below their levels of concern, unless there is reason to believe that some easily monitored constituent is strongly correlated with a pollutant of concern that is difficult or costly to monitor.

Once the parameters for monitoring have been selected, the analytical detection limits needed to meet data quality objectives must be determined.

Select Monitoring Methods

Although use of either grab or composite sampling, and manual or automatic collection (or a combination of these methods) are appropriate techniques, this document specifically addresses selecting and using automated equipment aimed at collecting flow weighted composite samples. This section is intended to aid in the decision-making process during the program development stage.

Typically, the two basic choices for source identification monitoring are collecting a single grab sample and collecting a flow weighted composite sample. Another possible approach is to collect a series of grab samples at intervals during a storm and analyze them individually; however, this approach is seldom used because it is much more expensive than the typical approaches. Finally, a series of grab samples could be allocated and composited, but not flow- or time-averaged. The advantages and disadvantages of the single grab and flow weighted composite approaches are summarized below.

Single Grab Sample

A single grab sample collected during "first flush" or high intensity conditions, when pollutant concentrations are expected to be at their highest, can be used to identify stormwater pollution sources. This approach has several advantages when compared to flow weighted composite sampling.

  • It can be done using either manual or automated methods.
  • It is suitable for any parameter.
  • It is not necessary to sample the entire duration of the storm.
  • Continuous flow measurement is not required.
  • Generally, neither equipment installation or channel modifications are required, so it is easy to move to new locations as necessary to track down a source.

The single grab sample approach also has several disadvantages.

  • Peak pollutant concentrations do not always coincide with the "first flush," especially in areas subject to frequent, low intensity storms.
  • In large basins, peak pollutant concentrations may occur during peak flow, which could occur any time during the storm. This makes it difficult to collect a single grab sample from the expected "worst-case" portion of the storm event.
  • Depending on the location of the pollutant sources relative to the monitoring site, the runoff from a contaminated source area may not be present in the "first flush."

Flow Proportional Composite Sample

A flow proportional composite sample collected during the entire duration of a storm can be used to identify stormwater pollution sources. This approach has advantages when compared to the single grab sample approach.

  • It is less likely to omit a source due to stormwater quality changes during the storm.
  • The results provide a better indication of the relative importance of a source than does a single grab "snapshot."

Flow proportional composite sampling also has disadvantages.

  • Manual flow proportional composite sampling is generally impractical if there are more than a few stations to monitor.
  • Automated equipment is costly to buy and install and it requires frequent inspection and maintenance. Since source identification monitoring at any given location is usually a short term program, automated equipment may need to be moved from place to place to track a pollutant to its source. Some of the automated methods and equipment allow for a mobile "package station" to be set up for easier movement of equipment.

Some monitoring programs are set up to collect composite samples only during the initial portion (e.g., the first three hours) of a storm runoff event. This alternative is generally less labor-intensive than monitoring the full storm because it reduces equipment calibration time and the risk of unacceptable samples (due to overfilling or underfilling of bottles, capturing less than 60% of the storm, failing to collect enough material to conduct analytical tests, etc). However, this approach would not discern any elevated contaminant concentrations that could occur during the later stages of a storm as a result of increased rainfall intensity, and possibly from contributions from pervious areas and/or contributions from distant portions of the catchment area.

Select the best approach for the specific situation based on the advantages and disadvantages described. Most source identification programs rely on grab sampling because it is generally more cost effective and more flexible than flow proportional composite sampling. In order to increase the potential for grab sampling to detect sources, consider collecting multiple grab samples throughout the storm event and then compositing them on an equal basis.

Details about selecting specific monitoring equipment are provided in Section 3.

2.5.3. Monitoring to Characterize Stormwater Quality Trends

This section describes how to develop and implement a monitoring program to detect water quality trends or changes in pollutant levels over time. Trend analysis can be an important and powerful tool in demonstrating benefits of stormwater pollution control and/or the effects of increasing urbanization. This type of monitoring program may be appropriate for assessing the overall effectiveness of a stormwater pollution control program. The following sections discuss the key considerations associated with each phase of the monitoring plan development and implementation.

2.5.3.1. Determine Objectives and Scope

Readers of this section likely will have concluded (at least tentatively) that general monitoring objectives should include an evaluation of the effectiveness of stormwater pollution control programs through an analysis of long term trends in stormwater quality. Trend analysis may be an objective of a monitoring program to fulfill permit requirements, or it may be performed voluntarily to provide information to demonstrate program effectiveness and support planning decisions or permit negotiations. This type of monitoring is sometimes the only practical approach for assessing source control best management practices (BMPs) and many smaller structural BMPs in a watershed (e.g., input/output monitoring is impossible or difficult at best).

Even in cases where the minimum monitoring requirements do not include trend analysis, the potential merits of trend analysis should be reviewed to determine whether it might provide useful information. Specific monitoring requirements should also be reviewed to determine whether the information provided is sufficient to support management decisions. These requirements may or may not have been based on information relevant to site specific conditions. If a minimum monitoring program does not collect adequate data to allow statistically verifiable confidence in the results, there is little basis for any conclusions regarding the effectiveness, or lack thereof, of a stormwater quality management program.

Trend analysis generally requires monitoring a large number of storms in order to distinguish real changes in stormwater quality from natural variability or "noise." A power analysis should be performed using existing data in the watershed or from sites evaluated to be similar to help determine the monitoring frequency for your program. (Power analysis is described in detail in Appendix A.)

At this point specific, realistic monitoring objectives need to be developed. Section 2.5.3.2 should be reviewed to develop the combination of monitoring locations, frequency, parameters, and methods best suited to data needs and resources. While reviewing these sections, the following general process should be followed:

1. If enough data are available for a power analysis, perform the analysis (step 2); if not, plan to collect sufficient data to permit a power analysis.

2. Perform power analysis.

2a. Establish desired statistical confidence interval (typically 90 or 95% confidence in the mean value).

2b. Establish desired sensitivity.

3. Evaluate results of power analysis.

3a. Is a monitoring plan for trend analysis feasible based on recommendations of power analysis? If the answer is yes, go to Step 4.

3b. Are lower confidence limits or cruder sensitivity acceptable? If the answer is yes, adjust confidence limits and/or sensitivity of the power analysis and return to Step 2.

4. Does the existing program satisfy the sampling frequency and number suggested by the power analysis?

4a. If the answer is yes, develop a monitoring plan.

4b. If the answer is no, adjust the monitoring plan to fulfill sampling requirements suggested by the power analysis or change monitoring plan objectives.

Steps 1 through 3 involve power analysis, which is a statistical tool that can be used to ensure that the number of samples is sufficient to enable detection of a trend, with a specified level of statistical confidence. (Power analysis is described in detail in Appendix A.) It is used to determine the number of samples required achieving a desired level of statistical confidence. The desired level of confidence may be achieved by increasing the number of sample locations and/or the number of samples collected at each location depending on the sampling approach.

To reduce the cost of an extensive monitoring program, the sampling program could be limited to analysis of a few indicator parameters. Parameters such as turbidity and conductivity can be monitored continuously with automatic field probes at reasonable costs. These parameters are sometimes good indicators of pollutant levels (especially of metals). Whether these parameters are good indicators of the target parameters (chemicals of concern) can be confirmed by collecting samples for laboratory analysis during a number of storms and performing regression analyses. This approach has limited value, as it would likely be necessary to collect a large number of storms to establish the relationships, and then after BMP implementation the process may have to be repeated, as these relationships will change. A more straightforward approach would be to sample for a few constituents throughout the study (e.g., TSS, copper, zinc, and total phosphorous).

2.5.3.2. Develop Monitoring Plan

The following sections provide guidance on developing a specific approach to achieving monitoring objectives.

Select Monitoring Locations

Ideally, a trend analysis program should encompass all stormwater outfalls that leave the study area. However, staff and budget constraints may preclude monitoring every outfall at the optimum frequency; if the study area is large and complex, available staff and funding may be insufficient to monitor every catchment. If locations for a trend analysis are not specified in a permit, a sufficient number of locations should be chosen to adequately assess overall water quality.

Consider locations where:

  • the effectiveness of basin-wide programs can be assessed; or
  • the effectiveness of programs targeted to a specific basin or type of basin can be assessed.

Trend analysis generally requires monitoring a large number of storm events. Thus, if available resources are limited, only a few locations may be able to be monitored (the extrapolation approach described in Section 2.5.1.2 applies to trend analysis monitoring). The representativeness of one location may be established using the results of monitoring at several locations. For the purposes of statistical hypothesis testing, a control location must also be sampled unless pre-implementation monitoring of the catchment is conducted.

The number of required locations and samples may be balanced by reducing the number of parameters to be analyzed, or by identifying an indicator parameter that is amenable to continuous field monitoring. For trend analysis, the monitoring locations should be generally at stations located in the lower portion of a watershed that are fairly representative of the larger drainage basin. If previous data are available, select a monitoring location with relatively high pollutant loads. This will make any reductions due to management practices easier to detect by the statistical sampling design.

Select Sampling Frequency

A power analysis is particularly appropriate for determining the number of samples (i.e., storms) that will need to be monitored to detect a water quality trend. The power analysis may indicate that a large number of storms would need to be monitored to satisfy program objectives. If this is the case, the number of monitoring locations should be reduced. Alternately, using continuous monitoring for indicator parameters (such as turbidity or dissolved oxygen) might be considered; this approach may allow monitoring of a much larger number of storms than would be possible using traditional sampling and laboratory analysis.

The number of events to be sampled for trend analysis may be set by the power analysis, or may be determined by what is feasible, balancing the needs of compliance monitoring with those of trend monitoring. However, if collecting a sufficient number of samples is not possible, it may prove difficult to discern trends. Sampling for trend analysis need not be annual. A frequency convenient for the permittee (e.g., once every permit cycle or every five years) may be used to acquire the needed data. Thus, it may be more cost-effective to conduct an intensive sampling effort every five years rather than a less intense annual program.

The number of samples set by the power analysis accounts for the variability in the concentration of pollutants observed between storms, even at a single sampling station. It is recommended that this number of samples be collected over as long a period as necessary. An attempt should be made to collect samples during different times of the year to account for seasonal variations in pollutant concentrations.

Select Parameters and Analytical Methods

In most cases, a sampling program for the analysis of long term trends in water quality will be relatively narrow in scope, focusing on parameters that have been shown, through stormwater quality characterization studies, to occur in concentrations that impact water quality. Once the parameters for monitoring have been selected, the analytical detection limits needed to meet data quality objectives must be specified.

Select Monitoring Methods and Equipment

General considerations for the choice of monitoring methods and equipment are discussed in detail in Section 2.5.2.2. After considering the arguments for grab versus composite, and manual versus automated approaches to sample collection, choose a method or combination of methods consistent with program goals. In general, composite sampling is much superior to grab sampling for long term trend analysis, except for those parameters for which composite sampling is not appropriate (e.g., oil and grease, TPH, bacteria). Details about selecting specific monitoring equipment are provided in Section 3.

Select Storm Criteria

As discussed in Section 2.5.2.2, it is generally desirable to monitor a broad range of storm conditions, rather than just the USEPA defined "representative" storms. This is particularly applicable to trend analysis, since monitoring only the so called representative storms may introduce bias. If use of continuous monitoring of indicator parameters is feasible, data from all storms should be included in trend analyses.

2.5.4. Monitoring to Evaluate Individual Best Management Practice (BMP) Performance

Many studies have been conducted to assess the ability of stormwater treatment BMPs (e.g., wet ponds, grass swales, wetlands, sand filters, dry detention, etc.) to reduce pollutant concentrations and loadings in stormwater. However, these individual BMP evaluations have utilized a broad spectrum of methods and reporting procedures. These inconsistencies complicate, if not prohibit, comparisons of the findings of different studies. The studies have included the analysis of different constituents and different methods for data collection and analysis. These differences alone contribute significantly to the range of BMP effectiveness reported, which complicates assessment of other factors that may have contributed to the variation in performance. In addition, removal efficiencies are increasingly being questioned as an appropriate measure of performance since the removal. The efficiency appears to be mainly controlled by the influent concentration.

Typically, structural BMPs have well-defined boundaries and are generally relatively easy to monitor. Other types of BMPs, especially non-structural BMPs (e.g., street sweeping, catch basin cleaning, sewer cleaning, illicit discharge elimination), are more difficult to monitor; partly because they tend to be geographically interspersed with many pollutant sources and can be influenced by many factors that cannot be "controlled" in an experimental sense. Some non-structural BMPs, such as public education programs, oil recycling programs, and litter control programs are virtually impossible to monitor or at best can be evaluated using trend monitoring as described in Section 2.5.3.

This section provides guidance on monitoring well-defined structural BMPs. It is assumed that many stormwater quality management programs will want to consider the possibility of implementing some structural BMPs, but would be inclined to experiment with them on a pilot-scale by testing and demonstrating their performance, costs, and practical implications before committing to larger-scale implementation. Programs that already have structural BMPs in place may also want to test their performance for a variety of reasons.

2.5.4.1. Determine Objectives and Scope

Studies of BMP performance are usually conducted to obtain information regarding one or more of the following questions:

  • What degree of pollution control or effluent quality does the BMP provide under normal conditions?
  • How does this performance vary from pollutant to pollutant?
  • How does this normal performance vary with large or small storm events?
  • How does this normal performance vary with rainfall intensity?
  • How do design variables affect performance?
  • How does performance vary with different operational and/or maintenance approaches?
  • Does performance improve, decay, or remain stable over time?
  • How does this BMP's performance compare with the performance of other BMPs?

BMP performance monitoring has been prescribed by some permits, but often the wording of such requirements is vague. Local program-specific objectives are likely to be the soundest basis for planning a BMP monitoring study.

Nationally many stormwater programs need BMP performance data, and many are planning or conducting performance monitoring. The concept of sharing monitoring results is very appealing but could be seriously constrained if pre-planning to maximize the chances of yielding comparable/compatible monitoring approaches, analytical protocols, and data management is not implemented. Some of the guidance provided in this manual and referred to in literature citations is intended to stimulate the users of this manual to expand their thinking and look for ways to broaden their program's objectives to facilitate exchanges of more transferable data among programs.

As an example, in a review of the use of wetlands for stormwater pollution control (Strecker et al., 1992), a summary of the literature was prepared regarding the performance of wetland systems and the factors that are believed to affect pollutant removals. The studies reported in the reviewed literature were inconsistent with respect to the constituents analyzed and the methods used to gather and analyze data. Several pieces of information were improperly collected and recorded, which decreased the ability to evaluate the effectiveness of stormwater wetlands as BMPs. Furthermore, the lack of such basic information limits the transferability of the studies' findings into better design practices.

The technical literature has many reports of monitoring programs to evaluate BMP performance. Those that address some of the conceptual and strategic aspects of monitoring (e.g., Strecker, 1994; Urbonas, 1994) could be of particular value during this planning stage. In addition, USEPA and the American Society of Civil Engineer's Urban Water Resources Research Council have compiled a National Stormwater Best Management Practices Database (on the world wide web at www.bmpdatabase.org). The purpose of this effort is to develop a more useful set of data on the effectiveness of individual BMPs used to reduce pollutant discharges from urban development. A review of the protocols established for the database is useful in determining what and how information should be collected and can be used for improving information collected for local use.

It is also valuable to review the monitoring methods and findings of other reported programs, because they may contain concepts (or even data) that are transferable to your situation. In considering the use of data collected elsewhere, critical attention should be paid to differences that might lead to erroneous conclusions (e.g., weather, soil types, role of specific sources of pollutants). Particular care should be taken to avoid the types of errors that are often introduced by assuming (rather than determining) that certain pollutants are associated with certain sediment fractions. The association of pollutants with particular particle sizes is very important (in fact, this association is the reason that most BMPs are effective), but this association varies dramatically from place to place and must be determined based on careful local studies of relevant factors-not simply assumed from other studies. When using data from relatively early studies, it is important to consider the fact that the state of the art of analysis has advanced considerably in the past decade or so; for example, many data entries recorded as "non-detect" may no longer be relevant.

2.5.4.2. Develop Monitoring Plan

The following sections provide guidance on developing a specific approach to achieving monitoring objectives.

Select Monitoring Location

Care must be taken to locate flow measurement and sampling sites in places that are likely to yield good data over diverse operational conditions. For performance monitoring approaches that are intended to compare changes in pollutant loads (i.e., "loads in" versus "loads out" of the BMP), it is especially important to use accurate flow measurement methods and to site the points of measurement at locations that maximize the attainment of credible data. The added cost of a weir or flume may be justifiable because without it measurement errors could propagate through various aspects of the analysis.

Select Monitoring Frequency

The frequency with which monitoring should be performed will depend upon a program's specific objectives and the degree of accuracy needed. To address the latter, compare the cost of learning more (i.e., conducting more intensive monitoring) versus the cost implications of moving forward too far and implementing extensive controls before having learned enough to guide planning, stormwater management commitments, and/or negotiations with regulatory agencies. The cost of controlling unimportant pollutants and/or unimportant sources, or implementing ineffective BMPs, could easily exceed the cost of monitoring to learn more about actual BMPs, performance under the conditions that prevail in a system. Clearly, there is a need for balance. Endless studies should not be substituted for control actions.

In general however, many measurements (i.e., many samples collected during many events) are necessary to obtain enough data to be confident that actual BMP performance is observed and not just "noisy" data (e.g., variability artifacts caused by external factors, equipment and operator errors). Consequently, BMP effectiveness studies can be expensive and time-consuming.

Select Parameters and Analytical Methods

Under ideal circumstances, a given BMP will be targeted toward controlling a well-defined, locally important problem caused by a particular pollutant or combination of pollutants. When selecting parameters for performance monitoring, it follows that one would probably look for changes in concentrations (and/or loads) of the target pollutant(s), or would look within the BMP to examine accumulations of the target pollutant(s). In cases where it is known that there is a high degree of correlation between the concentration of the target pollutant(s) and some other parameter (e.g., fine particles, TSS, TOC), then it may be possible to use a less costly monitoring approach to track the substitute, or "proxy" parameter(s). Although this approach can introduce some uncertainty because it does not track the target pollutants, it is still worth considering. If the correlations are known to be strong and the cost differences pronounced, this strategy may provide a way to obtain much more data (i.e., more frequent observations during more storm events and/or at more locations). Such improvements in data quantity could more than offset the uncertainties introduced by imperfect correlations.

There are many precedents for using proxy parameters and indicators. For example, fecal coliform are bacteria often used as proxies for pathogens and as an indicator of fecal contamination. TOC and COD are sometimes used as proxies for BOD. Turbidity is commonly used as a proxy for suspended solids, which in turn, is sometimes used as a proxy for other pollutants of concern (e.g., metals, PAHs). It is important to remember that other factors could also account for observed changes in the proxy parameter relationship to other pollutants.

In many BMP monitoring programs, there are opportunities to obtain additional information at little or no incremental cost. Such information may turn out to be valuable to the overall stormwater program at some time in the future and/or to other programs.

Recommend Parameters

This section presents a recommended list of constituents for BMP monitoring. Strecker (1994), Urbonas (1994), and the ASCE Database website (www.bmpdatabase.org) provide more information on BMP monitoring parameters. The choice of which constituents to include as standard parameters is subjective. The following factors were considered in developing the recommended list of monitoring parameters:

  • The pollutant is one that has been identified as prevalent in typical urban stormwater at concentration levels that could cause water quality impairment (as identified by USEPA 1983 and recent Municipal NPDES data).
  • The analytical test is one that can be related back to potential water quality impairment.
  • Sampling methods for the pollutant are straightforward and reliable for a moderately careful investigator.
  • Analysis of the pollutant is economical on a widespread basis.
  • The pollutant is one that might be controlled through practical BMPs rather than elimination of the source. (e.g., treating to remove pesticides downstream instead of eliminating pesticide use in the right-of-way)

Although not all of the pollutants recommended here fully meet all of the factors listed above, the factors were considered in making the recommendations. When developing a list of parameters to monitor for a given BMP evaluation, it is important to consider the upstream land uses and activities. Table 2.1 presents a list of suggested standard parameters for assessing the effectiveness of BMPs. It assumes that flow-weighted composite samples will be collected using automated procedures; thus, the table does not include parameters not amenable to this type of sampling, such as fecal coliform. The parameters recommended in Table 2.1 are generally present and are of concern in typical near-highway runoff. The table includes a typical cost for each of the tests.

TABLE 2.1 RECOMMENDED STANDARD ANALYTICAL TESTS FOR URBAN STORMWATER BMP ASSESSMENTS
Lab Analyses
Detection Limit
Conventional  
TSS 1 mg/l
BOD5 3 mg/l
COD 1 mg/l
Total Hardness 25 mg/l
Nutrients  
TKN - N 0.3 mg/l
NH3 - N 0.3 mg/l
Total phosphorus -P 0.05 mg/l
Ortho-phosphate - P 0.05 mg/l
Nitrate + nitrite (NO3 + NO2) - N 0.1 mg/l
Total Metals  
Cd (cadmium) 0.2 µg/l
Pb (lead) 1 µg/l
Cu (copper) 1 µg/l
Zn (zinc) 1 µg/l
Dissolved Metals  
Cd (cadmium) 0.2 µg/l
Pb (lead) 1 µg/l
Cu (copper) 1 µg/l
Zn (zinc) 1 µg/l
Source: Strecker, 1994

The parameters listed in Table 2.1 represent the most basic arrangement of parameters. There may be appropriate applications where other parameters should be included. For a discussion of why some parameters were not included, see Strecker (1994).

Select Monitoring Methods and Equipment

BMP monitoring can be an especially useful application for some automated systems (e.g., continuous flow recorders, auto samplers, continuous monitoring probes) for the following reasons.

  • Automated systems can provide data covering virtually the entire volume of runoff that passes through the BMP (i.e., they are not likely to miss or leave out small events and the beginnings and ends of other events).
  • Automated systems are well suited to providing data sets that are useful (recognizing that performance evaluations are generally based on the differences between inlet and outlet concentration data sets, both of which are inherently noisy).
  • The information obtained from good performance monitoring programs can be so valuable (by protecting against inappropriate BMP applications) that the cost of using automated systems is often justifiable.

BMP monitoring can also be performed using manual methods. Such methods are usually preferred under the following circumstances.

  • Available resources for equipment purchase/installation (e.g., funds, personnel, time) are very constrained and/or there is not the political will to invest in a program, despite the inherent value of the resultant information.
  • The target pollutants are ones that do not lend themselves to automated sampling or analysis (e.g., oil and grease, volatile organic compounds, bacteria).
  • The physical setting of the BMP does not allow the use of automated systems.

Details about selecting specific monitoring equipment are provided in Section 3.

Select Storm Criteria

The establishment and application of appropriate storm selection criteria can be a challenging aspect of planning BMP monitoring programs. Ideally, data should be obtained from all phases of all storms for as long a study period as possible, for the following reasons:

  • It is desirable to know what the BMP does during periods of very low flow, normal flow, and very high flows. The performance of some BMPs varies dramatically with throughput rate. Some may even release pollutants that had been previously trapped.
  • Performance must be estimated on the basis of differences between relatively noisy data sets (i.e., inlet versus outlet data) and intensifies the value of large volumes of credible data (not just a few samples from portions of a few storms).
  • For some BMPs with significant wet storage and/or base flows it is important to characterize the water quality of dry weather flows as well. This is particularly important when the wet volume of the BMP is large relative to the storm event. The comparison of inflow to outflow during a storm event is not valid because the outflow may have had little or no relationship to the incoming storm. This mistake has been made often in past studies.

Despite the desire for extensive and high quality data, there is still a need to tailor program methods to be consistent with available resources.

2.5.5. Monitoring to Assess Compliance with Surface Water Quality Criteria

This section provides guidance on stormwater monitoring to assess compliance with surface water quality criteria for protection of human health and aquatic life. The section begins with an overview of surface water quality criteria. The section then shows how the general approach to stormwater quality monitoring can be applied to compliance monitoring. Sections 2.5.5.1 and 2.5.5.2 discuss the key considerations associated with each phase of program development and implementation.

In addition to surface water quality standards, stormwater discharges may affect compliance with standards for groundwater quality and/or marine sediment quality. However, stormwater monitoring is typically of limited value with regard to assessing compliance with groundwater and/or sediment quality standards. Compliance with the groundwater standards is generally assessed through groundwater monitoring (rather than stormwater monitoring) because stormwater quality is likely to change substantially while percolating through soils, and the extent of the change is very difficult to predict without a great deal of site-specific information. Similarly, compliance with sediment quality standards is generally assessed through sediment monitoring within receiving water bodies. This is because numerous storms would need to be monitored to develop useful estimates of total annual sediment loads, and the particulate portion of each sample would need to be divided into particle size fractions prior to chemical analysis to allow even a qualitative evaluation of potential sediment transport/deposition. For these reasons, this manual does not address stormwater monitoring to assess compliance with groundwater or sediment quality standards.

Overview of Water Quality Criteria

USEPA describes water quality criteria and their relationship to water quality standards in the following paragraphs:

"Water quality standards are laws or regulations that the States adopt to enhance and maintain water quality and to protect public health. Water quality standards provide the foundation for accomplishing the goals and objectives of the Clean Water Act. More specifically, water quality standards help to:

Restore and maintain the chemical, physical and biological integrity of the Nation's waters; and,

Where attainable, achieve water quality that promotes protection and propagation of fish, shellfish and wildlife and provide for recreation in and on the water. This goal is commonly known by the expression "fishable and swimmable"; and,

Prohibit the discharge of toxic pollutants in toxic amounts; and,

Eliminate the discharge of pollutants to navigable waters.

Water quality standards apply to surface waters of the United States, including rivers, streams, lakes, oceans, estuaries and wetlands. Water quality standards consist, at a minimum, of three elements: 1) the "designated beneficial use" or "uses" of a waterbody or segment of a waterbody; 2) the water quality "criteria" necessary to protect the uses of that particular waterbody; and 3) an antidegradation policy. Typical designated beneficial uses of waterbodies include public water supply, propagation of fish and wildlife, recreation, agricultural water use, industrial water use and navigation. Water quality criteria describe the quality of water that will support a given designated use. Under authority of section 304 of the Clean Water Act, USEPA publishes, on an advisory basis, water quality "criteria" that reflect available scientific information on the maximum acceptable concentration levels of specific chemicals in water that will protect aquatic life or human health.

These criteria are intended to provide protection for all surface waters on a national basis and may be used by the States for developing enforceable water quality criteria that protect the designated use as a part of their water quality standards. When properly selected criteria are met, they are expected to protect the designated use with a margin of safety. The antidegradation policy ensures that existing water quality is maintained and protected. States use criteria developed by USEPA under section 304 to adopt enforceable maximum acceptable concentration levels of a pollutant in ambient waters. The water quality criteria adopted into a State water quality standard may or may not be the same number published by USEPA under section 304. States have the discretion to adjust the section 304 criteria to reflect local environmental conditions and human exposure patterns or to derive a criterion from an independent methodology as long as it is scientifically defensible. Water quality criteria can also be expressed in either numeric form or narrative form by the States in their water quality standards. USEPA reviews and approves State water quality standards every three years. To date, virtually all States have narrative and numeric water quality standards that protect human health and aquatic life from exposure to some chemicals and conditions in the water, including toxic and bioaccumulative pollutants. However, few States have adopted numeric criteria for biological integrity, excessive nutrient enrichment, excessive sedimentation, wildlife protection or flow control" (USEPA, 1998).

Water quality standards may include bacteria, dissolved oxygen, temperature, pH, turbidity, and toxic organic and inorganic compounds in marine and freshwater bodies.

State Water Quality Standards (WQS) often are based on Federal Water Quality Criteria (WQC) for the protection of human health and aquatic life (40 CFR 131.36). However, Federal WQC may include additional compounds not listed in WQS.

Note that WQC are considered guidelines, whereas WQS constitute enforceable regulations. In this section, WQC is used to encompass both state standards and the Federal guidelines.

There are two general categories of water quality criteria: aquatic (or marine) criteria, and human health criteria. These are summarized below.

Criteria for the Protection of Aquatic/Marine Life

Criteria for the protection of aquatic and marine life were developed based on laboratory toxicity tests with representative organisms using test solutions spiked with pollutants to simulate exposure. In order to apply the results of these tests, USEPA has classified aquatic life standards as either "acute" or "chronic" based on the length of time the organisms are exposed to listed concentrations.

Criterion maximum concentrations (CMC acute) are intended to protect against short term exposure. Criterion continuous concentrations (CCC chronic) are designed to protect against long term exposure. In deriving the acute criteria, the laboratory organisms were exposed to pollutant concentrations for 24 to 48 hours. USEPA suggests one hour as the shortest exposure period that may cause acute effects and recommends the criteria be applied to one-hour average concentrations. That is, to protect against acute effects, the one hour average exposure should not exceed the acute criteria. USEPA derives chronic criteria from long-term tests that measure survival, growth, reproduction, or in some cases, bioconcentration. For chronic criteria, USEPA recommends the criteria be applied to an averaging period of four days. That is, the four day average exposure should not exceed the chronic criteria.

WQC for aquatic life were developed based on an allowable exceedance frequency of once every three years, based on the theory that an ecosystem is likely to recover from a brief water quality exceedance, provided it does not occur very often.

Human Health

Water quality standards for the protection of human health contain only a single concentration value and are intended to protect against long term (chronic) exposure. For carcinogenic compounds, a lifetime exposure over 70 years is generally used to calculate the criteria. For noncarcinogens, exposure periods are more chemical-specific and depend on the particular endpoint and toxic effect.

USEPA has defined two levels of protection for human health criteria. The first criteria were derived based on cumulative risks associated with drinking water and eating organisms that live in the water. The criteria for carcinogenic compounds is the calculated water-column concentration that would produce a one in a million (10 6) lifetime cancer risk if water were consumed and a given amount of organisms from that water was eaten every day. The second criteria are based on consumption of organisms alone (the water is not consumed). These standards apply to saltwater or other water that is not a drinking water source but does support a fishery that is used as food. The organism's only standard for carcinogenic compounds is the calculated water concentration that would produce a one in a million (10 6) lifetime cancer risk if a person were to consume a given amount of fish or shellfish from that water body (without drinking the water).

Application of Water Quality Criteria to Stormwater

The WQC are intended to protect the beneficial uses of streams, lakes, and other receiving water bodies. Most of the man made conveyances within a near-highway stormwater drainage system do not support these beneficial uses. Thus, monitoring to assess compliance with WQC is usually conducted in a receiving water body (rather than in the stormwater conveyance system that discharges into it) to provide a direct measure of whether the beneficial uses of the water body are impaired or in danger of becoming impaired.

Direct comparisons between stormwater quality and the WQC should be interpreted with caution because such comparisons do not account for mixing and dilution in the receiving waters or the effects of receiving water hardness levels on heavy metals. This is especially true when the stormwater discharge is very small relative to the receiving water body.

The variable nature of stormwater quality further complicates comparison to WQC. Stormwater quality varies both between storm events and during a storm event, so it is very difficult to extrapolate data from one storm to another or to generate statistically representative data for all types and combinations of storms.

In spite of the limitations mentioned above, comparisons between stormwater quality and WQC can provide valuable information for stormwater management. WQC can be used as screening criteria, or "benchmarks" for assessing stormwater quality problems and establishing management priorities. Direct comparisons with the WQC can over estimate the potential impact of the stormwater discharges on the receiving water bodies, because mixing and dilution are not taken into account. However, the relative frequency and magnitude of concentration exceedances within storm sewer systems higher than the WQC can help determine priorities for additional investigations and/or implementation of control measures. Frequently occurring, large exceedances are a clear indication that further investigation and control measures are warranted. Marginal or occasional exceedances are more typical and more difficult to interpret.

2.5.5.1. Determine Objectives and Scope

Readers of this section likely have concluded (at least tentatively) that it is necessary to compare stormwater quality to water quality criteria for protection of aquatic or marine life and human health. The results of a stormwater quality monitoring program to assess compliance with water quality standards or criteria may be used for a variety of purposes:

  • Compliance with watershed based permits
  • Compliance with an individual industrial permit
  • Determination of need for additional BMPs
  • Watershed management planning
  • Non point source assessment

Section 2.5.5 described the key issues associated with comparing stormwater quality data to WQC. In general, comparisons between stormwater quality and WQC should be used to assess stormwater quality problems and establish management priorities rather than to identify apparent water quality violations. The latter objective should be assessed through monitoring the receiving water body. This can provide a direct measure of the degree to which aquatic/marine life or human uses of the water body have been impaired.

As previously indicated, the WQC are based on specified periods of exposure (i.e., one hour or four day exposure for aquatic/marine life; lifetime consumption of water/fish by humans). Also, the standards allow for occasional exceedances (generally, once in three years). Therefore, a strict comparison with the WQS or WQC would require measuring stormwater quality at one hour increments, and determining whether any standards are likely to be exceeded more than once in three years.

Because most human health based criteria are based on lifetime exposures, direct comparisons with transient stormwater concentrations often may be inappropriate. Pollutant concentrations in water often decrease due to sedimentation, volatilization, biodegradation, and other attenuation processes during transport and storage prior to human consumption. Some fraction of the pollutants is likely to be removed if runoff is stored in a surface reservoir prior to consumption. Also, most surface drinking water supplies are treated prior to distribution. This treatment will likely remove a portion of pollutants that exceed criteria (typically PAHs and arsenic).

If comparisons are performed with criteria intended to protect humans who consume fish and shellfish, consider how stormwater quality compares to ambient concentrations during dry weather periods. In either case, the results of such comparisons should be used only as guide and not in a rigorous regulatory manner. Often, a more direct measure of the potential threat to human health is gained by measurement of pollutant concentrations in edible portion of the food organisms (tissue analysis) rather than through comparison with water quality standards.

Most dischargers do not have the resources (i.e., funds, personnel, time) to conduct the comprehensive monitoring that would be required to support a rigorous assessment of compliance with WQC. For these reasons, most stormwater quality studies focus on pollutants commonly encountered in stormwater and conduct limited sampling to determine whether unusually high concentrations are present at representative locations. Comparisons with WQC are then used to help identify problems and establish priorities for addressing them.

It is necessary to develop a specific, realistic approach to achieving monitoring objectives. Section 2.5.5.2 below should be reviewed to develop the combination of monitoring locations, frequency, parameters, and methods that are best suited to data needs and resources.

Compile and review the relevant existing information that is available for the area. A thorough review of the existing relevant information can help in selecting the most appropriate monitoring locations, parameters, and methods for a given situation. In particular, it is important to review any water quality monitoring data available for a stormwater system and the receiving water bodies to identify locations and pollutants of potential concern.

2.5.5.2. Develop Monitoring Plan

The following sections provide guidance on developing a specific approach to achieving monitoring objectives.

Select Monitoring Locations

The number of locations to be monitored depends on program objectives, permit requirements (if applicable), the size and complexity of drainage basin(s), and the resources allocated to monitoring. In addition, the frequency of sampling at each location should be considered.

In general, monitoring locations for WQC comparisons should be located in the main drainage channel just before it discharges into the receiving water body. Monitoring after mixing with the receiving water should not be conducted unless the data are to be used to support application for a mixing zone or a mixing zone has been granted, or the objective is to assess impacts on receiving waters. If possible, a few representative locations should be chosen rather than attempting to monitor all possible locations. Locations that drain directly to receiving waters that are known to be impaired should be selected first. Locations that drain directly to other receiving waters should be chosen second, and locations that drain to closed storm drain conduits should be chosen last.

Select Monitoring Frequency

The variable nature of stormwater makes determination of a representative exposure period difficult. Watershed or catchment specific characteristics often influence the duration of runoff, with runoff from large catchments typically lasting longer than runoff from smaller catchments for the same storm size. The site-specific nature of runoff event duration makes it difficult to determine which exposure period is appropriate for a given location.

It is not feasible or desirable to monitor every storm event to determine whether criteria are exceeded more than once every three years. Resources necessary for such monitoring may be better spent on implementing BMPs. Instead, it is recommended that monitoring programs attempt to sample a representative subset of storms that occur throughout the storm season. Storms that occur after a long dry period often contain higher concentrations of pollutants than similar storms that occur after a short dry period. If a limited number of storms is to be sampled, it is recommended that storms occurring after a long dry period be sampled in order to consider the "worst-case." If possible, it is recommended three to five storms per season be sampled. If "worst-case" data produce reproducible results showing compliance with water quality objectives, it may be possible to decrease the frequency of monitoring. Additionally, if data show consistent exceedances, it may be desirable to focus efforts on controlling the problem rather than additional monitoring.

Select Parameters and Analytical Methods

Entities conducting monitoring will need to select the parameters and analytical methods most appropriate to their specific situation. Table 2.2 lists the common urban runoff constituents that may be found at highway sites. This list may serve as a starting point for a monitoring program. If significant monitoring has already been completed in the area, consider deleting any parameters that were consistently below levels of concern, particularly if these previous monitoring efforts included highway sites.

TABLE 2.2 LABORATORY PERFORMANCE STANDARDS FOR COMPARISON OF TYPICAL URBAN STROMWATER RUNOFF CONSTITUENTS WITH FRESHWATER WATER QUALITY STANDARDS
Parameter
Units
Target Detection Limit
Conventional    
PH pH N/A
Turbidity mg/L 4
Total Suspended Solids mg/L 4
Total Hardness mg/L 5
Chloride mg/L 1
Bacteria    
Fecal Coliform MPN/100ml 2
Total Coliform MPN/100ml 2
Enterococci MPN/100ml 2
Metals-Total Recoverable    
Total Recoverable Digestion µg /L 0.2
Cadmium µg /L 1
Copper µg /L 1
Lead µg /L 5
Zinc    
Metals-Dissolved    
Filtration/Digestion µg /L 0.2
Cadmium µg /L 1
Copper µg /L 1
Lead µg /L 5
Zinc    
Organics    
Organophosphate Pesticides (scan) µg /L 0.05 - .2
Note:This list includes constituents found in typical urban stormwater runoff. Additional parameters may be needed to address site specific concerns.

Once the parameters for monitoring have been selected, it is necessary to specify the analytical detection limits needed to meet data quality objectives. Table 2.2 lists the recommended method detection limits for comparing stormwater samples to WQC.

It has long been recognized that different metal forms (species) show different levels of toxic effects. In general, the most toxic metal forms are ionic where the metal is present, dissolved in the free ionic form. Recognizing this fact, USEPA recently revised WQC for those metals for which criteria are based on toxicity tests to allow comparison with dissolved metal concentrations (40 CFR 131, May 4, 1995). Specifically, USEPA developed revised criteria for the following dissolved metals: arsenic, cadmium, chromium, copper, lead, mercury (acute only), nickel, silver, and zinc. Chronic criteria for dissolved mercury were not proposed because the criteria were developed based on mercury residuals in aquatic organisms (food chain effects) rather than based on toxicity. For comparisons with WQC, the dissolved metals fraction should be determined. If selenium or mercury is of concern, total concentrations should also be measured to enable comparison with criteria based on bioaccumulation by organisms.

The distribution of pollutants between the dissolved and particulate phases will depend on where in the system the sample is collected. Runoff collected in pipes where sediment is generally present at low to moderate concentrations will generally have a higher percentage of pollutants present in the dissolved form. Runoff collected in receiving waters will generally have a higher percentage of pollutants present in particulate form due to higher concentrations of suspended solids and therefore "receptor" sites to which pollutants can attach. It is difficult to determine how much of the dissolved pollutants found in storm system pipes will remain in the dissolved form when they are mixed with suspended sediments in receiving waters. As a result, it is difficult to determine the ecological significance of moderate levels of dissolved pollutants present within the conveyance system. In addition, hardness values for receiving waters are often different from stormwater. Hardness affects the bio-availability of heavy metals and this further complicates the ecological impact of dissolved heavy metals.

If loads to the receiving waters are of concern (for example, discharge to a lake known to be a water quality limited water body) it may be desirable to determine total recoverable metals in addition to dissolved metals to allow assessment of the relative load from different sources. Finally, total recoverable metals data together with dissolved metals can be used to assess the potential metals in sediment.

Select Monitoring Methods

The arguments for grab versus composite sampling, and manual versus automated approaches to the collection of samples, must be considered, then the combination of sampling that best fits program objectives and budget should be chosen.

Storm events affect stream flows for variable lengths of time depending on the storm duration, antecedent conditions and catchment characteristics. Runoff persists for a few hours or up to typically two days. This suggests runoff rarely persists long enough to be considered comparable to chronic exposure duration. Discrete sampling over the course of the storm event will provide concentration information that can be used to determine how long WQC were exceeded during the storm. Alternatively, discrete samples can be composited on a time weighted basis over time scales comparable to the acute and chronic WQC exposure periods (one hour and four days), respectively. However, the latter would likely include dry-weather flows since few storms last four days. For catchments that are relatively small (a few acres), one or more one hour composite samples should be collected during the first few hours of flow. This can be done by collecting and combining three or more grab samples.

Flow weighted composite sampling can be used for comparison with water quality objectives (for example if flow-weighted composites are collected to measure loads). However, it should be recognized that a flow weighted sample would contain more water from peak flows than from the initial part of the storm. Results from Santa Clara Valley Nonpoint Source Monitoring Program indicated that for a large watershed with significant suspended sediment concentrations (200 400 mg/L), the peak total metals concentration is generally 1.5 times the flow weighted composite concentrations (WCC, 1993). Results from monitoring a smaller highly impervious industrial catchment with lower suspended sediment concentrations were more variable and no conclusions could be drawn as to the relationship between flow composite concentrations and grab samples due to difficulties in grab sampling runoff that only occurred during precipitation.

Composite samples should not be collected for certain water quality measurements (e.g., PAHs, bacteria) due to sorption losses during compositing. Grab samples should be collected if analyses are to be conducted for these parameters.

Select Storm Criteria

Because the initial objective of the monitoring is to consider a to a search and make them consistent "worst-case" picture, it is desirable to select storms with the highest pollutant concentrations rather than a representative mix of storms. "Worst-case" conditions are likely to occur after long antecedent dry periods (72 hours to 14 days). Therefore, if feasible, storms should be selected with antecedent periods greater than 72 hours. Few relationships between storm volume and water quality have been observed. Lacking any basis for storm volume selection for "worst-case" conditions, and acknowledging that storm characteristics are highly dependent on climatic region, the following may be used as a starting point:

Rainfall Volume: 0.10 inch minimum No fixed maximum
Rainfall Duration: No fixed maximum or minimum
Typical Range: 6 to 24 hours
Antecedent Dry Period: 24 hours minimum
Inter-event Dry Period: 6 hours

If these criteria prove inappropriate, site-specific storm event criteria can be developed by analyzing long-term rainfall records using USEPA's SYNOP or another appropriate analytical program. This could include USEPA's SWMM model, which incorporates the features of SYNOP.

It should be noted that biasing the storm selection to the "worst-case" would not provide a representative sample of the population of all types of storm events. The resulting data should be used in screening mode and not to estimate statistically derived exceedance frequencies. The level of effort required to sample all representative types and combinations of storm conditions to generate reliable population statistics is beyond the resources of most agencies. For this reason a "worst-case" approach should be taken. Often permits require monitoring of "representative" storms that have been predefined. Operationally and practically, storm event criteria may need to be further defined beyond the regulatory definition. The use of a probability of rainfall above a certain magnitude, during a specific period, based on a quantitative precipitation forecast (QPF) serves as a good indication of when and how to mobilize for monitoring efforts. QPFs for a geographic area can be obtained from the National Weather Service and site specific information can be obtained from private weather consultants.

Back to top

Questions and feedback should be directed to Deirdre Remley (deirdre.remley@dot.gov, 202-366-0524).

spacer
HEP Home Planning Environment Real Estate

Federal Highway Administration | 1200 New Jersey Avenue, SE | Washington, DC 20590 | 202-366-4000