The City of Detroit Water and Sewerage Department is responsible for operation of 46 combined storm sewers, most of which have an overflow outlet to the Detroit River. In 1991, the Michigan Department of Natural Resources (MDNR) identified discharge from combined sewer overflows (CSOs) as a threat to the health of the Detroit River (Stage 1, Detroit River Remedial Action Plan (RAP), 1991). CSOs can result in the discharge of floatables, sediment, nutrients, organic material, oil and grease, metals, pathogens, and other pollutants in the receiving waters. Floatables and oil and grease cause short-term degradation of aesthetics, and elevated bacteria levels may cause the closing of beaches. Such impacts are typically of short duration and are the most likely to draw public attention. Longer-term impacts are related to the loads from one or many seasons of discharge, and include reductions in dissolved oxygen, sediment contamination, and eutrophication caused by nutrients, suspended solids, and biodegradable organic matter. Effects from heavy metals contamination may include long-term toxic effects.

Stage 2 of the Detroit River RAP identifies strategies to be undertaken to further identify, quantify, and begin correcting CSOs. An accurate assessment of the variability in pollutant loads among CSOs is one step in being better able to predict pollutant effects on receiving waters. In Detroit, Michigan, four CSOs discharging to the Detroit River were monitored to characterize storm-related water quantity and quality and to calculate their respective annual pollutant loads. A prior study was conducted on similar sites in an attempt to document the number of CSOs that were active in the City of Detroit, and the number of times that they overflowed each year (Giffels, Black, and Veatch, 1980). To accomplish these objectives, automated, unattended monitoring stations were designed and installed. These gages included monitors for stage, flow, and precipitation, as well as automated samplers for the collection of discharge samples.

METHODOLOGY

Four CSO sampling and flow measurement sites were established along the Detroit River in the City of Detroit. At each site it was necessary to monitor for stage or the presence of fluids; flow velocity and volume; and precipitation. All of this information needed to be stored in a readily accessible manner. To accomplish this data loggers were chosen that had the ability to be programmed to not only store but also manipulate data, and that could be accessed via modem. Because of the location of the sites, and uncertainty about the length of the monitoring program, cellular phones were chosen for the communication link. The City of Detroit was able to provide electrical power service to all sites; however, in order to guard against the loss of power to equipment during critical sampling events, all sites were backed up with 12 volt batteries to power the data loggers, samplers, and flow measuring devices. The cellular phones were protected with their own internal battery system.

Equipment at each site consisted of a refrigerated automatic sampler with four 10 liter glass jars with teflon lids, 0.95-centimeter (cm) Teflon\xa8 -lined sampler suction line, a datalogger programmed to activate the automatic sampler and store data, a modem and cellular phone for remote data retrieval, a tipping-bucket rain gage, and at three of the sites, upward looking, bottom mounted doppler flow meters to measure stage, velocity, and discharge. At the fourth site, a channel-mounted 4-transducer-array acoustic velocity meter (AVM) was installed to measure flow velocity under backwater conditions, and a stilling well with shaft encoder installed to measure stage. Flow measuring stations were located as near as practicable to the outlet of each CSO. Water-level, velocity, stage, discharge, and precipitation were measured continuously. At three of the sites, water samples were collected directly from enclosed sewer, whose diameters ranged from 1.22 to 4.19 meters (m). At the fourth site, an automatic sampler collected samples directly from an open channel.

Water quality samples at all sites were collected at discrete times during each storm event, based upon the event hydrograph. This allowed for unlimited 10 liter samples per storm event. Thus, estimates of the variability of pollutant concentrations during a single event can be made. In general, the routine for collecting samples at each site called for a sample to be collected at the detection of event inception. Event inception was indicated either by the presence of water in an otherwise dry overflow channel, or by positive flow values, sustained for a predetermined length of time, in an ordinarily stagnant channel. After the collection of the first sample, subsequent samples were collected based upon the performance of the event hydrograph at each site. In general, samples were desired at or near the peak of the hydrograph, and at later stages of hydrograph recession. To obtain samples at the peak of the hydrograph, an algorithm was developed that monitored the previously determined velocity or flow values, and assessed whether stage was rising or falling and whether flow was increasing or decreasing. If a decrease/fall was detected after a sustained period of increase/rise, sampling was initiated for the peak-of-the-hydrograph sample. Recession samples were generally manually triggered by personnel who by this time had arrived on site to process the samples. In order to assure that samples were not missed, a timed sample routine was also included in case personnel could not access a site in time to manually trigger a sample.

Water-quality samples were stored in glass jars in a refrigerated sampler. Each glass jar was prewashed, rinsed with deionized water (DIW), washed with acid, rinsed with DIW, rinsed with methanol, and triple rinsed with DIW. The glass jars were capped with Teflon\xa8 lined caps. Upon collection of a sample, field personnel removed the jars and transported them to a central sample processing site, where they were split using 20 liter churn splitters.

The samples were split into separate bottles for the analysis of constituents listed in table 1. The processed samples were then preserved and delivered to the Detroit Water and Sewerage Laboratory for analysis of conventional field parameters, including pH and dissolved oxygen, and biological and inorganic pollutants. Additional processed samples were sent by overnight carrier to the USGS National Water Quality Laboratory for the analysis of organic pollutants.

Quality assurance and quality control samples were submitted in numbers equal to about 20 percent of all sampled events for which analyses were made. Selection of chemical constituents and properties of water quality, and analytical protocols, including minimum detection levels, holding times, and quality assurance, were as required by EPA's final storm-water regulations.

RESULTS

Samples were first collected in August, 1994, at Conner Drain, and were subsequently collected at all sites between December 1994 and December 1995. Table 2 lists the number of samples collected at each site, along with the number of distinct events sampled at each site. Laboratory results are not yet complete for some samples collected later in 1995, but all laboratory results are expected to be returned by March, 1996.

It was intended that early events be oversampled in order to determine the proper sample frequency, and to determine which later events to sample, and which to bypass. Because of the logistics of operating 4 sites concurrently, in the end all sites were sampled for each event, rather than sampling only selected events. Rosa Parks/Twelfth Street Drain at Jefferson Avenue was not sampled until June, 1995, because there were no measurable events at this site until that time; conversely, Schroeder Drain at W. Jefferson Avenue was sampled more frequently than anticipated because of its proximity to the waste water treatment plant. When the plant was at capacity, and water was diverted through CSOs to the river, Schroeder Drain was often the first site to be impacted, and the last to get relief when the treatment plant was back at or under capacity.

In the early phases of the project, Conner Drain and Fischer Drain at Burns Avenue flowed during most significant precipitation events, and their flow could be somewhat anticipated. During the latter half of the project, however, both of these sites flowed less frequently and with less regularity. It is the opinion of City personnel involved in the project that changes made in the operation of the treatment plant and by operators controlling the flow in the sewerage system, both intended to improve the efficiency of the treatment system, may have been responsible for this shift in the number of CSOs at both sites (Oral comm, J. MacDonald, R. Meah, and K. Prybys, 1995, 1996).

In order to rate the AVM and calculate discharges at Conner Drain, it was necessary to make physical discharge measurements using standard USGS methods. At the inception of many events a measuring crew was assembled and travelled to the gage site. However, in all but one instance the event duration was not sufficient to allow completion of a discharge measurement. One measurement was made in November, 1995, near the end of an event. This measurement has been processed and is being used to calculate preliminary discharge data for Conner Drain. Any events that occur in early 1996 will be used to make additional discharge measurements to improve the rating at this site.

Conventional field parameters, including pH and dissolved oxygen were measured, and samples for the analysis of biological, inorganic and organic pollutants were collected. Analyses were made for conventional pollutants and metals listed in table 1 by the City of Detroit Water and Sewerage Department Analytical Laboratory; analyses for organic pollutants listed in table 1 were made by the USGS National Water Quality Laboratory. Preliminary data analysis indicates that there may be, in some instances, discharge-related trends for some pollutants.

DISCUSSION

A relationship between CSO discharges and pollutant concentrations has been hypothesized for each site. At this time, such relationships have not been confirmed; however, some interesting relationships between different pollutants have been established. A plot of the hydrograph for an event at Schroeder Drain (figure 1) shows total suspended solids (tss), chromium, copper, and cobalt plotted against time and in relation to flow indicates that during the beginning part of an event there is an increase in the concentration of chromium in the system, but that tss concentrations remain relatively constant. Further analysis of relationships such as these is being conducted, and should lead us to a better understanding of the relationship between CSO and pollutant loadings to receiving waters.

Figure 2 shows the relationship between total suspended solids and total kjeldahl nitrogen (TKN) at all 4 sites; figure 3 shows the relationship between tss and chromium for the same sites. In general, as tss increase, concentrations of TKN and chromium also increase. The relationships vary by both site and pollutant of interest, but the general trend is an increase in one pollutant associated with another. For three of the sites, these relationships are typical of what would be expected.

However, at Schroeder Drain this relationship is inverted; as tss increases, both TKN and chromium concentrations decrease. This indicates that there is another process at work in the transport system that is modifying the relationship of some pollutants with respect to suspended solids. Metals such as chromium are typically sorbed to particles, hence it is expected that the concentrations of chromium would mimic those of suspended solids. Further analysis of additional events will be needed to confirm this unexpected relationship, although, as shown in figures 2 and 3, this relationship appears to be consistent throughout the period of study.

CONCLUSIONS

The principal results expected of this investigation were the development of a relationship between discharge at CSOs and pollutant loads to receiving waters. Early results presented here indicate that the instrumentation package designed for this project, and the field approach used, were correctly implemented to verify these expectations. Although relationships vary between sites, at individual sites from event to event, and from pollutant to pollutant, and are dependent upon event duration, time between events, and drainage basin characteristics, it appears that our results will allow us to determine a pollutant loading model for each of the CSO sites studied.

An important difference between Schroeder Drain, the most downstream CSO site studied, and the other three CSO sites which are upstream of Schroeder Drain, indicate that processes may be active within the sewerage transport system that have the potential to either mitigate or modify pollutant loads as combined storm sewerage is moved through the system.

ACKNOWLEDGEMENTS

This study was conducted as part of a cooperative agreement between the United State Geological Survey, the City of Detroit Water and Sewerage Department, Michigan Department of Environmental Quality, United States Environmental Protection Agency, Wayne State University, and Southeast Michigan Council of Governments. Successful completion of this project would not have been possible without the dedicated efforts of Ranu Meah, Ken Prybys, and John MacDonald of the Detroit Water and Sewerage Department, Waste Water Collection Group, who were typically the first to respond to events.

Use of trade names in this report is for identification purposes only and does not constitute and endorsement by the U.S. Geological Survey.

REFERENCES

Giffels, Black and Veatch, 1980, Quantity and quality of combined sewer overflows Volume II Report, CS-806, final facilities plan, interim report, City of Detroit Water and Sewage Department.

Michigan Department of Natural Resources, 1991, Draft Michigan State-Wide Combined Sewer Overflow Permitting Strategy, Surface Water Quality Division, Municipal Permits Unit, Lansing, MI, 48909.