A Comprehensive Approach to Urban Stormwater Impact Assessment

Betsy Johnson, Kimberly Yandora, and Scott Bryant, P.E.

Greensboro Storm Water Services

401 Patton Avenue, Greensboro, NC 27406



Stormwater monitoring has been conducted in Greensboro, North Carolina, since April 1994 as required by the City’s National Pollutant Discharge Elimination System (NPDES) municipal stormwater permit. Runoff sampling is conducted quarterly at storm sewer outfall pipes from seven sites, which generally represent the City’s land uses. Over thirty water quality parameters are tested in grab and composite storm event samples. As part of the permit requirement, results are evaluated and pollutant loading estimates are calculated to determine pollutant specific annual and seasonal loads for each site and the entire city. To complement and expand upon this data, City staff also monitored ambient stream water quality under baseflow conditions, evaluated wet weather runoff for acute toxicity, and collected benthic macroinvertebrates for biological assessment in order to characterize stormwater impacts.

The results of the stormwater runoff sampling and load estimates are sufficient to characterize the discharges of varying land uses in the City. However, these data do not provide relevant information on the impact to the receiving waters. Since there are currently no end-of-pipe regulatory limits on stormwater nor wet weather criteria for streams, a tiered approach to impact assessment is needed. Although the runoff data show significant pollutant loads, the baseflow stream sampling and toxicity testing results indicated no major problems. The benthic data, however, show a stressed aquatic community. These studies point to a need for further monitoring including wet weather stream sampling and sediment sampling and/or instream toxicity testing. The data collected to date indicate that the City’s urban stream system is stressed due to stormwater runoff quantity and quality impacts.


Stormwater monitoring has been conducted in Greensboro, North Carolina, since April 1994 as required by the NPDES municipal stormwater permit conditions. The only stormwater monitoring explicitly required by the NPDES permits is the storm event sampling. Municipalities larger than 100,000 persons (NPDES Phase I stormwater regulations) sample runoff from five to ten representative drainage areas to characterize runoff for their locale. However, these data do not provide relevant information on the impact to the receiving waters. In order to characterize impacts to the receiving waters, a more comprehensive study and approach is needed. In Greensboro, the Storm Water Services Division is examining many facets of its stream and aquatic environment to monitor and quantify the impacts of stormwater runoff so that a watershed-based management program can be developed to mitigate those impacts.


In the first five years of its NPDES stormwater permit, Greensboro’s staff developed a monitoring program both to meet the permit requirements and to make initial assessments of stormwater impacts to the City’s stream resources. The staff sampled storm event runoff, stream quality, and the biological community, assessed habitat, and collected samples for toxicity tests. This approach has shown that stormwater impacts are diverse and a range of control measures within the watershed must be used for urban water quality improvement.

Storm Event Runoff Monitoring

Storm event sampling characterizes the quality of stormwater runoff from urban nonpoint sources feeding the municipal storm sewer system during wet weather. In general, the quality and quantity of stormwater runoff depends on the land use type and impervious surface area. In developed urban areas such as Greensboro, there is significant impervious area in which water will not infiltrate and which causes runoff to increase over predevelopment flows. Industrial, commercial, and residential activities bring pollutants into contact with stormwater runoff, which are then carried into receiving streams and lakes. Wet weather sampling enables the City to determine pollutant concentrations and loadings from urban nonpoint sources and provides data to assist in the development and location of effective best management practices for minimizing urban pollutant discharges to area receiving waters.

Wet weather monitoring in Greensboro currently focuses on characterization of discharges from various land use types. Staff performs quarterly monitoring of representative storms (0.1 to 0.8 inches of rainfall within 3 hours) to measure the quality of runoff from 7 sites (see Table 1). Each site represents a different land use type. These data provide a baseline for comparison with special study areas and provide data for estimates of city-wide pollutant concentrations and loads.

Each wet weather site is sampled by a team of two monitoring technicians who are on 24-hour call for storm events. Samples are taken manually from each storm sewer pipe outfall and delivered to a contract lab for analysis. Two types of samples are taken during a storm event: first a grab sample to measure the "first flush" of stormwater runoff and then a time-weighted composite sample is taken over a three-hour period to measure average runoff quality. Field parameters including water temperature, pH, conductivity, and turbidity are measured in addition to site specific rainfall and flow depths. Samples are analyzed for an extensive list of parameters per the NPDES requirements (see Table 2).

Ambient Stream Monitoring

To augment the storm event runoff monitoring and to establish a general baseline for determining impacts to the receiving streams, an ambient instream monitoring program was developed. Since July 1996, staff has monitored seven stream sites on a monthly basis (see Table 3). (North and South Buffalo Creeks and their tributaries represent the major stream systems in the most urbanized areas of Greensboro.) Similar to the storm event sampling process, samples are collected by a team of two monitoring technicians and delivered to a contract lab for analysis. Field parameters including flow level, water temperature, pH, conductivity, and turbidity are also measured. Conventional parameters are sampled monthly while metals are added on a quarterly basis (see Table 4). Samples are taken on the second Tuesday of each month, generally under baseflow conditions.

This stream monitoring provides water quality data and information on the health of the stream during dry weather. The data will provide a reference point for determining stormwater impact under wet weather conditions.

Toxicity Testing

To estimate the toxic impact of stormwater runoff on the receiving stream, toxicity testing of runoff was conducted. Due to the sporadic nature of stormwater runoff events, the State of North Carolina recommended acute toxicity testing. During the winter quarter of 1997 (January - March 1997) grab samples were collected during the first flush of storm events at each of the seven land use characterization outfalls. Samples were delivered to a contract lab where a 48-hour test was run on each sample using fathead minnows (Pimephales promelas). Sample dilutions (0, 12.5, 25, 50, 75, and 100 percent) were fabricated and mortality of minnows measured over a 48-hour period.

Biological Monitoring - Benthic Macroinvertebrates

During the summer of 1997, staff sampled the aquatic benthic invertebrate communities at 31 stream sites across Greensboro to further assess the impacts from urban stormwater. Sites were selected to complement wet weather sampling sites, ambient stream sampling sites, and to provide a comprehensive coverage of the city including streams draining to the city’s drinking water supply reservoir. Samples were taken using a kicknet in riffle area habitats. Subsamples of 100 organisms were taken from each sample and sent to a contractor for taxonomic identification.

Habitat Assessment

Along with the biological assessment of the city’s streams, habitat assessments were also conducted. Urban streams typically become degraded once contributory drainage impervious area exceeds approximately ten percent (Schueler, 1995). In addition, prior to 1994 and implementation of the municipal NPDES stormwater permit, the City dredged some local streams as part of its routine drainage maintenance. These factors have contributed to degrade the aquatic habitat and impact the biological community. With an improved understanding of the local impacts of past stream channel maintenance practices, the City no longer performs routine dredging of streams.

Two assessment methods have been used to evaluate stream channel conditions and habitat. The first was a qualitative survey of a subsample of city streams to assess stream channel stability, vegetation, and maintenance practices. A quantitative survey was done in conjunction with the biological monitoring study using the format provided by EPA’s Rapid Bioassessment Protocol. This assessment created a numerical score for localized sites including substrate characterization, channel sinuosity, channel alteration, streambank vegetation and stability, and riparian vegetation.

Results and Discussion

Urban Stormwater Runoff

The characterization of receiving streams as impaired varies depending on the amount of impervious area and land uses in the drainage basin. Sites with little impervious area and restricted land uses, undeveloped or low-density residential sites have few problems. Age and quality of the drainage system and maintenance of the land uses appear to play a role in runoff quality. Sites with large quantities of pavement and roof area generate significantly more runoff and more pollutants. On these sites, there is greater area for pollutants to accumulate between rain events.

Pollutants of concern identified in the City of Greensboro’s sampling program include:

A significant finding of the runoff sampling has been that stormwater contains pollutants similar to wastewater discharges to streams and in some cases with concentrations that are higher. Also, the findings from Greensboro’s storm event sampling program are generally comparable to the national NURP study.

Elevated levels of suspended and dissolved solids and turbidity are found in nearly all samples except the undeveloped site (Country Park). The TSS data average about 200 mg/l during the first flush of runoff but drop to below 80 mg/l in the composite samples. For comparison, wastewater discharges must meet a limit of 30 mg/l, but in Greensboro are usually less than 10 mg/l due to stringent treatment requirements. Turbidity frequently exceeds the state standard of 50 NTU in the first flush of stormwater runoff.

Fecal coliform and fecal streptococcus have been present at all sampling locations. Exceedances of the state standard for fecal coliform (200 colonies/100 ml) have occurred at all sites. The only site that is occasionally within the standard is the undeveloped site. Most fecal data is far above the required levels for wastewater discharges (200 colonies /100 ml). Some sites have exhibited exceptionally high values, which may indicate leaks from the sanitary sewer lines to the storm sewer system.

Although the dissolved oxygen (DO) of runoff itself does not appear to be of concern, the BOD and COD levels are generally much higher in urban runoff than domestic wastewater discharges. This is especially true locally in Greensboro, where due to the low flow streams and high temperatures, wastewater discharges are required to meet very stringent BOD limits of less than 5 - 10 mg/l. Though wastewater discharges are continuous and stormwater discharges are sporadic, the BOD in urban stormwater runoff is exerted over a longer period of time than in many other wastewaters (Field and Pitt, 1990). The long-term BOD of some storm runoff may be much higher than that of domestic wastewater; and sediments may store BOD which become resuspended and move the area of DO deficit further downstream.

Greensboro is located at the headwaters of the Cape Fear River Basin in an area that has been designated by the State as Nutrient Sensitive Waters (NSW). This stream classification requires wastewater discharges to meet a phosphorus limit of 2 mg/l to protect Jordan Lake, a major regional water supply and flood control reservoir, from more frequent algal blooms. The runoff data indicates that nutrient loading from urban sources is less severe than from wastewater discharges. However, the nutrient levels (phosphorus and nitrogen) from the Athena (90% impervious) first flush runoff are greater than or equal to the required wastewater discharge limits. The nutrient loading levels are approximately 50% lower in the composite sampling. A more important consideration may be the total nutrient loading from nonpoint sources as compared to the point sources.

Metals have been detected at all sites. Higher concentrations are present at sites with greater amounts of impervious area and with more industrial land uses. Copper, lead, and zinc exceed the state standards (7 m g/l, 25 m g/l, and 50 m g/l respectively) at all sites except the low density residential and undeveloped sites. Other studies (e.g., Moran, 1998) indicate that significant sources of these metals, particularly copper, include automotive wear and tear (including automotive brake pads) and roof runoff.

Annual Pollutant Loading Estimates

Annual pollutant loading estimates have been calculated using the "Simple Method" (MWCOG, 1987), where: pollutant load (lbs/yr) = [ runoff (acre-ft/yr) * event mean concentration of pollutant (mg/l) * conversion factor ]. An effective annual rainfall of 38.3 inches per year (based on the average for Greensboro) was used to estimate runoff for each sampling site. At the end of the sampling year, the actual annual rainfall is converted to effective annual rainfall and the load estimates are revised accordingly. The estimates attached as Table 5 are annual pollutant loading averages based on analysis of local storm event sampling results between June 1995 and June 1997. Table 5 also provides the loadings based on pollutant event mean concentrations from the EPA NURP study for comparison to the local findings. The estimates are comparable, but the local pollutant concentrations and loadings are generally lower than findings and estimates based on the national NURP study.

Ambient Stream Monitoring

Monitoring results to date indicate that during dry weather the water quality of the city streams is only slightly impaired. There are no significant instream problems during dry weather except in areas with dry weather discharges. One site is located in an industrial corridor which has both stormwater runoff problems and dry weather discharges. Fecal coliform levels consistently exceed state standards. A master plan is currently under development for the largest of the industrial sites, which will address containment and treatment of stormwater runoff and dry weather discharges.

Water quality is consistently uniform during baseflow conditions. The most noticeable changes during or immediately following wet weather is the sediment load from upstream construction areas. Turbidity levels exceed the state standard when there is soil loss from construction sites. There are also significant differences between sediment loads in North and South Buffalo Creeks. South Buffalo carries a much higher load, which can be largely attributed to increased construction activities in the South Buffalo Creek watershed.

The instream baseflow water quality data contrasts with wet weather data, which indicates that significant levels of pollutants are entering the streams via stormwater runoff. In July 1998, baseflow monitoring will be reduced to quarterly and wet weather monitoring will be conducted twice during the year. Stream monitoring will also be added at six United States Geological Survey (USGS) flow-gaged stream sites in order to correlate water quality and flow information for future watershed modeling and master planning efforts.

Acute Toxicity Tests

The results of the acute toxicity tests indicated that the first flush samples were not toxic according to the standards of the test. No acute toxicity (48-hour, fathead minnow) was found at any site. The LC50 was greater than 100% at all sites. The LC50 is a measure of the strength of a sample in which 50% of the population is found to die after a 48-hour exposure (Standard Methods, 1994). For the samples obtained in the city, even full strength or the 100% dilution did not cause a 50% mortality. Very little mortality was seen at the storm sites. Merritt (75% impervious, commercial site) had limited mortality of 35% at the full strength sample. The cause of this mortality is unclear. The water quality data for Merritt were generally better than the other sampled sites. However, conductivity was higher and dissolved oxygen was lowest of all the samples. This is consistent with a study in Kentucky that found that mortality in the bioassays of stormwater runoff was most affected by low DO concentrations in the runoff (Marsh, 1993).

The tests indicate that the first flushes of stormwater runoff were not acutely toxic. However, reviews of the data show exceedances of water quality standards for copper, lead and zinc as well as detectable levels of other metals. Therefore it is likely that stormwater may have a chronic impact. Acute toxicity testing was selected because storm events are short-term events. Yet, these events occur frequently with an average of four significant events per month. Some of the effects of stormwater discharges are associated with organic and toxic pollutant accumulations over a long time and are not associated with individual runoff events (Field and Pitt, 1990). The true impact on the stream then can become a chronic impact. Chronic toxicity testing could determine whether the stormwater runoff itself has a toxic impact.

Other studies suggest that the toxic impact from stormwater runoff is manifested in the interface between the sediment and water column. Pollutants bound to suspended particles in the water accumulate in the bottom sediment and are readily available to aquatic organisms or may be resuspended during storms. Sampling by Wisconsin DNR found that petroleum byproducts and heavy metals were present in bottom sediment (Masterson, 1994). Another researcher, Dr. G.J. Pesecreda of North Carolina, has conducted instream toxicity tests using the Stonefly (Pteronarcys dorsata) to determine their response to urban runoff and wastewater in comparison to a control site. He found that mortality in sites receiving urban runoff experienced greater mortality than an instream reference site. In addition, there was no significant difference in mortality to the test organisms exposed above and below a wastewater plant when both were exposed to urban stormwater runoff (Pesecreda, 1997). Sediment sampling and sediment toxicity tests may be needed to evaluate the relationships of pollutant transport and storage in sediments to water quality impairment as measured by the biological community.

Biological Community

Results from benthic macroinvertebrate samples indicated that biotic communities are relatively tolerant to periodic storm events and are more severely impacted by degradation in habitat and continuous water quality problems. Thirteen of the 31 sites sampled rated "good-fair" using the North Carolina Biotic Index (NCBI). This "good-fair" rating was supported by similar ratings from taxa richness and EPT abundance values. Only two sites had "poor" biotic communities. These two sites were located in the North Buffalo Creek basin and had no Ephemeroptera, Tricoptera, and Plecoptera (EPT) and high percentages of Chironomids. Both of these sites receive runoff from old industrial and commercial areas and also had degraded habitat.

Overall, South Buffalo Creek indicated less impaired biotic communities and water quality than North Buffalo. An in-stream site (Big Tree) actually indicated "excellent" NCBI and had a high habitat score. Even the highly industrialized Gillespie site had "fair" NCBI rating and 15 EPT species. However, water quality and stream degradation was observed as the South Buffalo Creek traversed downstream through the city.

As expected, water quality and biotic communities in the water supply watersheds were less impaired than North and South Buffalo Creeks due to better habitat and less urban runoff. In general, the water supply watershed areas are less developed than the North and South Buffalo basins. However, Bryan Park, which was selected as a reference site, received only a "good-fair" biotic rating but had a high habitat score. Although little urban development is present in the Bryan Park drainage area, agricultural activities in the area are believed to have contributed to the stream degradation.

Macroinvertebrate samples are useful to determine short-term changes in habitat and water quality changes. They would be useful to determine the effects of stream restoration or changes in riparian buffers as well as the effects of new construction or point source discharges. Habitat, water chemistry and biotic communities interact to form the aquatic ecosystem; therefore, there is a need to study and identify all three to manage aquatic resources properly.

Aquatic Habitat

The assessment of stream channels to evaluate maintenance practices and channel stability indicated that the past City maintenance practices in combination with high velocity and increased volume of storm flows resulted in stream channels that are unstable and highly erodible. The historical practice of maintaining stream channels as a drainage network included routine dredging to remove sandbars and widen channels, and routine mowing up to and including the stream banks using a boom mower. These practices have disturbed habitat within the stream and removed most vegetative cover. In 1994, the routine dredging ceased. While some mowing has continued, the City continues to evaluate its vegetative maintenance practices to develop an optimum balance between the environment and other public concerns related to stream systems. Wildlife studies conducted by the Audubon Society in Greensboro compared stretches of the same stream, North Buffalo Creek, with and without vegetative cover. Their findings indicate that there is a more diverse songbird community where there is more cover. This study also found that radio-tagged turtles would not enter areas without vegetative cover (Audubon, 1998). As noted, the City continues to evaluate its maintenance practices to provide a better balance between flood routing, aquatic habitat, and related stream system issues.

Habitat assessment was conducted at 31 locations throughout the city. All locations indicated some impact from human development. Some sites, especially in parks and the water supply watershed, had very good riparian buffer zones, channel flow and pool variability but lacked streambank stabilization and had high sediment deposition. Conversely, other sites no longer have meanders, canopy cover, or epifaunal substrate cover but have stabilized banks and low sediment deposition. However, all sites were able to support aquatic communities. Schueler states that stream degradation occurs at 10-20% impervious (Schueler, 1995). However, stream alteration takes place any time the watershed is disturbed. Habitat should always be evaluated when assessing water quality. Future monitoring is planned to evaluate restoration of riparian vegetative zones along stream reaches that had previously been mowed and/or dredged.


The monitoring of stormwater runoff required by the NPDES permits is not sufficient to determine the overall impact of stormwater on the City’s receiving waters. And, without this determination, the City cannot develop a complete strategy for improving water quality in the city streams. A more comprehensive approach to urban stormwater monitoring and determining the impacts of urban stormwater runoff upon receiving waters includes:

While meeting its NPDES requirements, the City of Greensboro has begun implementation of a comprehensive program for water quality monitoring and stormwater impact assessment. In 1998, a network of USGS stream gages will be added to allow continuous tracking of rainfall, streamflow, and provide early warning of potential flooding. The city’s monitoring staff will sample the streams at the USGS sites at least six times per year during variable flow regimes in order to develop a database for model calibration.

A watershed-based stormwater runoff and stream model will be developed during 1998 and 1999 to predict both water quality and quantity impacts including a determination of the impacts of future land use changes. With this tool, the City can develop watershed-based strategies for reducing the impacts of future urban development, as well as mitigating impacts of existing development and other factors within the watershed. Comprehensive and proactive watershed management, including improved water quality, is the goal of the City of Greensboro’s stormwater program.


Audubon Society, T. Gilbert Pearson, February 1998, StreamGreen Streamlife Study.

Field, R. and R.E. Pitt, 1990, "Urban Storm-Induced Discharge Impacts," Water Environment & Technology 2(8): 64-67.

Marsh, J.M. Assessment of Nonpoint Source Pollution in Stormwater Runoff in Louisville, Kentucky, USA. Bulletin of Environmental Contamination and Toxicology.

Masterson, J. P., and Bannerman, R.T. 1994. Impacts of Stormwater Runoff on Urban Streams in Milwaukee County, Wisconsin. National Symposium on Water Quality, AWRA.

Moran, Kelly D., 1998, "Copper, Brake Pads, & Water Quality: Can a National Voluntary Partnership Improve Water Quality?", Proceedings of the Water Environment Federation Specialty Conference, Watershed Management: Moving From Theory to Implementation, Denver, CO.

Pesecreda, G.J. 1997. Response of the Stonefly Pteronarcys dorsata in Enclosures from an Urban North Carolina Stream, Bulletin of Environmental Contamination and Toxicology.

Schueler, T.R., December 1995, Environmental Land Planning Series: Site Planning for Urban Stream Protection, Center for Watershed Protection.

Standard Methods for the Examination of Water and Wastewater, 1992.

Metropolitan Washington Council of Governments (MWCOG), July 1987, Schueler, T.R., Controlling Urban Runoff: A Practical Manual for Planning and Designing Urban BMPs.

Table 1. Wet Weather Monitoring Sites in the City of Greensboro

Site Drainage Area Land Use % Impervious
Athena Court 21 acres Commercial – Heavy 90
Country Park 20 acres Open 2
Husbands St. 13 acres Industrial 74
Merritt Dr. 23 acres Commercial – Light 75
Randleman Rd. 26 acres Residential – High 50
Union St. 33 acres Mixed 75
Willoughby Blvd. 13 acres Residential – Low 20






Table 2. Sampled Parameters

Ammonia Nitrogen (NH3) Total Kjeldahl Nitrogen (TKN) Selenium*
BOD, 5-day (BOD5) Chlorides Silver *
Chemical Oxygen Demand (COD) Antimony * Thallium *
Cyanide, Total * Arsenic * Zinc, Total
Fecal Coliform, MF * Beryllium * EPA 624 *
Fecal Streptococcus – Tube * Cadmium EPA 625 *
Nitrate + Nitrite, Nitrogen Chromium EPA 608 *
Phosphorus, Total Dissolved (TDP) Copper, Total  
Phosphorus, Total (TP) Lead  
Solids, Total Dissolved (TDS) Mercury, Total *  
Solids, Total Suspended (TSS) Nickel  
* These parameters apply only to the grab sample. Other parameters apply to both grab and composite samples.


Table 3. Stream Monitoring Sites

Site Location and Stream Order Land Use
South Buffalo Creek at Big Tree Way
(2nd order stream)
High density residential
South Buffalo Creek at Hillsdale Park
(3rd order stream)
Tributary to Mile Run Creek at Gillespie Golf Course
(2nd order stream)
South Buffalo Creek at McConnell Creek
(4th order stream)
Agricultural, low density residential
North Buffalo Creek at City Arboretum
(3rd order stream)
North Buffalo Creek at Lake Daniel Park
(4th order stream)
Tributary to Richland Creek at Battleground Park
(1st order stream)



Table 4. Sampled Parameters in Ambient Stream Monitoring Program

Monthly Parameters Quarterly Parameters
Dissolved Oxygen Ammonia Nitrogen Aluminum Silver
Temperature Nitrate/Nitrite Arsenic Zinc
PH TKN Cadmium Nickel
Turbidity TP Chromium  
Conductivity TDP Copper
Chlorides TSS and TDS Iron
BOD5 Fecal Coliform Lead
COD Fecal Streptococcus Mercury



Table 5. Comparison of Annual Pollutant Loadings Based on NURP and Local Sampling Data


Estimated annual loadings based on NURP data (lbs / yr)

Estimated annual loadings based on local sampling data (lbs / yr)