An often-overlooked nonpoint source of stormwater pollution is roadside ditches. They can collect a wide range of pollutants from fecal coliform bacteria to petroleum hydrocarbons to heavy metals from roads, parking lots, and construction sites and transfer them to watersheds. The King County, WA, Department of Transportation Roads Maintenance Section is experimenting with stormwater treatment best management practices in these ditches and needs to continuously monitor flows through the ditches to determine which best management practices (BMPs) are most effective.
Roadside ditches have dynamic impacts on stormwater pollution levels; in that respect, they are no different from many nonpoint pollution sources. And as with many nonpoint sources, they warrant monitoring and/or sampling so that stormwater management authorities can keep up with rapidly changing conditions and keep emerging problems from developing into catastrophic ones. The example from Washington is just one of how stormwater management authorities are successfully meeting this challenge through the use of intelligent data-collection systems and data analysis.
The King County Roads Maintenance Section’s focus on the ditches originated from a study funded through the Washington State Department of Ecology’s Stormwater Management Implementation Grant Program, and that will last through 2011. The department is working on adapting low-impact-development (LID) concepts for use within existing county rights of way for two purposes: either treating stormwater or attenuating storm flows in roadside ditches. Jim Crawford, King County engineer, points out that the LID concepts are being used so that the county does not have to buy additional rights of way to construct stormwater ponds, an expensive proposition. All four projects are testing a modified rock check dam design that includes a “treatment cell”–an area within the dam lined with a porous erosion control fabric and filled with a water-quality treatment media.
“There has been a lot of focus on urban stormwater runoff through piped stormwater systems, and we didn’t really think people were looking at roadside systems,” says Crawford. “Our road maintenance department tends to have more semirural to rural roads, so we have more ditches, and we have a lot of water in them without people really looking at them very much. We looked around and really didn’t see anybody else in the country doing this. There’s a lot of focus using the kind of things we’re doing for construction runoff, but we’re working in typical roadside, semirural, one-house-per-acre land types to more rural older neighborhoods, where you have ditches right in people’s front yards, as opposed to pipe stormwater systems.”
Since late summer 2009, the Roads Maintenance Section has been evaluating BMP effectiveness by comparing conditions upstream and downstream of the BMP installations at the four project sites–the first ones developed for the study. The field monitoring program includes continually monitoring flow using a primary flow measurement device, i.e., a flume or weir. Flow at all of the sites is measured using data-logging equipment that records water level and calculates flow based on the geometry of the device. Extra-large 60-degree V-shaped trapezoidal flumes have been installed directly in the ditches at three of the study sites; weir inserts in existing culverts are being used at the fourth site. The weir inserts are equipped with tubing that is connected directly to a bubble meter. No telemetry is installed on any of the equipment, although the department has access to real-time, online rainfall data from the county. Department staff makes frequent site visits to verify the operation of the equipment, set up or demobilize sampling equipment, program the autosamplers, and download data from the data loggers to a laptop computer.
The two treatment projects are in a ditch along the east side of 148th Avenue S.E. just south of Southeast 102nd Street, and on the north shoulder of Southeast 136th Street between 169th Avenue S.E. and 170th Avenue S.E. The treatment cells on these projects are filled with compost treatment media mixed with washed gravel and wrapped in filter fabric and secured inside a rock check dam; the compost meets Washington State Department of Transportation specifications for use in stormwater treatment projects. The department is also collecting flow-weighted composite samples during stormflow events using autosamplers controlled by bubble meters.
The autosamplers are manually set to trigger based on a rise in water level, and they collect samples based on the forecasted rainfall and the staff’s understanding of the response of each ditch to the rainfall. The monitoring protocol requires documenting hand measurements of water levels during frequent site visits to verify the levels recorded by instrumentation. Level measurement devices include self-logging bubble meters and pressure transducers monitored by separate data loggers. The water level in the flumes is measured in stilling wells attached to the flumes.
The section is also collecting grab samples for fecal coliform bacteria and total petroleum hydrocarbons and taking readings of dissolved oxygen (DO), pH, conductivity, turbidity, and temperature using data sondes. Flow composite samples are collected for total suspended solids, total Kjeldahl nitrogen, nitrate-nitrite, orthophosphate phosphorus, dissolved metals, and polycyclic aromatic hydrocarbons. Additionally, the section is continuously measuring water temperature using Onset HOBO Pro v2 water temperature data loggers deployed in the ditches.
At the 148th Street site, three treatment cells are installed in a section of ditch approximately 100 feet in length. The ditch collects stormwater from numerous rural residential properties and the paved roadway and drains to May Creek. The 136th Street site ditch is dominated by residential properties with a density of four houses per acre and drains to a catch basin across from 169th Avenue N.E., that discharges to a small tributary stream.
The two flow-control BMP projects, which use treatment cells filled with sand, could potentially be widely used by jurisdictions trying to minimize the impacts of stormwater runoff on surface water bodies and groundwater aquifers. They include a section of the southwest shoulder of Southeast Petrovitsky Road and the west shoulder of 276th Avenue S.E. between Southeast 208th Street and Southeast 216th Street.
The Petrovitsky Road project is surrounded by low-density rural residential properties, and the City of Seattle’s Lake Youngs watershed is located nearby. The ditch collects runoff from adjacent properties and the paved roadway and drains southeast to Shady Lake, which outfalls to Honey Creek and Patterson Lake and eventually enters the Lower Cedar River. Five BMPs are secured inside a rock check dam, along a section of roadside ditch approximately 287 feet in length.
The 276th Ave S.E. project is characterized by rural residential properties and collects runoff from adjacent properties and the paved roadway. The ditch drains to a stream that crosses 276th Avenue S.E. just north of S.E. 208th Street and flows west toward Issaquah Creek. Ten BMPs have been installed along a section of roadside ditch approximately 340 feet in length.
Crawford reports that, as of June 2010, the section was wrapping up the first phase of data collection and evaluation and planning to install more BMP installations. The weather was very dry when the equipment was installed, he adds, and results were scheduled to be compiled after 12 storms. The key aspect of the study involves evaluations of the equipment used, e.g., whether bubble meters or pressure transducers work best for the BMPs.
Crawford reports that the idea behind using the treatment cells is to determine whether a small amount of BMP surface area makes an impact on stormwater quality. The county treats sheet flows off of highways with compost-amended soils, but treating stormwater runoff from ditches is a different story because most county roads have small shoulders. “We’re limited with the amount of surface area of treatment media we can add,” points out Crawford. “Those are some of the things we’re finding out–maybe we need a formula for how much water we’re expecting, how much treatment we can expect, so we’re trying to figure out more ways to put these BMPs in with our design.”
Once the department identifies efficacious and cost-effective BMPs, the next step will be “to evaluate how much maintenance they would need,” says Crawford. “Would we need to remove silt from them? Would we need to replace the media every year? Basically, maintenance costs for anything are huge, so there will be a weighting of maintenance costs and installation, in effect.”
Rob Fritz, supervising ecologist for King County, notes that stormwater flow conditions were not optimal during the first nine months of the study. “It’s been a challenge this year with all of the water we’ve had to deal with in the ditches as far as seeing any significant impact of our BMPs,” he says. “We’re waiting for low flows.”
“A big focus for us was to get as good flow measurements as we thought we could get, and we use different equipment to do that,” adds Crawford. “We’ve got a primary device in place–a flume or a weir–and we have a meter to measure water level.” Crawford also adds that the bubble meters are effective for accurately measuring slight changes in water levels.
The section is building a database using the BMP data for the purpose of making a decision based on a cost-benefit analysis. The monitoring system records the background conditions of water coming into the BMPs, as well as the condition of the water downstream from the BMP. Under this setup, comparisons can be made between upstream and downstream water quality.
Crawford says that the section has learned a lot about roadside ditches through the study. “The biggest piece of this was understanding the resource we’re trying to study,” he says. “The ditches are very different from pipe systems. We went into this thinking that a ditch is a ditch, and we can put something in it and test it. But as we looked at sites to pick, there was a difference between a site we could put a BMP in and a site we could actually study. We’re using these flumes, and they act as a BMP themselves.
“One aspect was trying to figure out the BMP design, something we could put in a ditch that would withstand the flows we’d get and that would have a positive effect. Another thing was figuring out how the ditches actually function–there are all sorts of issues from right of way and land ownership. We’re working in people’s front yards, and we’ve been incredibly lucky. We have some rain gauges out in the open and some other equipment that we’re unable to protect. But it’s been going great so far.”
More Realistic View of Pollution Levels
In a high-growth area such as Charlotte, NC, stormwater conditions can change rapidly during a major storm event with a large number of construction sites in the area. Mecklenburg County grew by more than 20% between 1990 and 2000, and the Land Use and Environmental Services Agency has increasingly dealt with the consequences of this growth since then. Olivia Hutchins Edwards, senior environmental specialist with Mecklenburg County, recalls that before the county invested in a real-time stormwater monitoring system, the county’s reporting methods failed to provide an accurate picture of conditions.
“We typically relied on doing grab sampling, where we would physically go out and collect a water sample and have it analyzed,” says Edwards, adding that sediment is a major pollutant. “What we found is that, if we suspected that we had a pollution source, we had to time the collection just right–a lot of pollution is intermittent.” In 1999, the county used a federal grant to explore continuous monitoring and eventually purchased a system with a dedicated computer. However, “we found that we couldn’t get real-time data” from the county’s numerous miles of streams and lake shoreline, she recalls. “We could collect data 24/7, but, say every two weeks, we would get a download of our data so that we could isolate a problem; we weren’t immediately notified of when an issue was happening. We saw a couple of instances where, if we had caught the pollution discharge immediately, the environmental impacts would have been minimal. We’ve had fish kills as a result, just devastation to the ecology within the stream. Pollution’s bad, but if you catch it pretty quickly, you can minimize those impacts. That was a motivator for us to try to go to something that was more sophisticated and more real-time-based.”
The physical aspects of collecting data also took time away from enforcement activities, Edwards says. “The older methods of tracking pollution sources were a lot more labor-intensive,” she notes. “Countless hours had to be spent just trying to isolate and track down data. As an agency, we have multiple roles that we serve, so we couldn’t give as much attention to detecting pollution problems because we have a whole slew of other problems that have to be addressed, too.”
In April 2005, the county began deploying continuously sampling YSI 6820 multiparameter sondes. Thirty-seven of these devices are linked by YSI’s EcoNet Web-enabled monitoring and control system to form a Continuous Monitoring and Alert Notification Network (CMANN). The sondes measure pH, conductivity, temperature, DO concentration, and turbidity. An automatic sensor wiping system cleans the turbidity probe regularly, reducing the impact of fouling.
Each station logs onto EcoNet every 15 minutes and uploads its latest readings via a wireless phone connection onto a Web site maintained by YSI. The data are displayed for either private or public access. Elevated pollutant readings trigger alarm messages, and upgrades can be made remotely.
“Now it’s such an expedited process as far as being able to isolate and detect problems,” says Edwards. “We can spend more time on it, because less legwork frees up more time for enforcement.” When state pollution thresholds are exceeded, Edwards receives an e-mail message. Two duplicate out-of-tolerance readings are confirmed, and then staff makes a site visit to investigate. After a major rain event, the county might be alerted to a turbidity reading over allowable nephelometric turbidity units; if the reading persists, it might indicate a silt fence blowout, Edwards points out.
Perhaps most importantly, the establishment of the CMANN has allowed the county to establish a water-quality baseline. The county uses the baseline and CMANN data to present to the public a monthly water-quality index. “Our old method of calculating the index was based solely upon snapshot grab samples,” says Edwards. “It made our data look really good. Typically, we were taking our samples when there hadn’t been any rain–typically, it’s during rain events that the majority of pollution is inputted into the stream system. Since we have this enormous network of automated units, we now use that data to calculate our index, and it’s been amazing. It’s more representative of what our water quality is really like. It did make our numbers a little worse, but it’s more true to what we do see out there.
“Since these devices are out there 24/7, over seasons, it’s really neat to see how the water chemistry is affected by seasonal changes and the elements,” concludes Edwards. “It’s really given us a better picture of water quality from a long-term perspective.”
Complex Interactions Identified
Another proponent of the use of sondes and continuous monitoring telemetry is Dayton, OH-based Woolpert Inc. which, among its services, provides stormwater management from 20 offices across the country. James Riddle, a project manager in Woolpert’s Columbia, SC, office, reports that he is currently using the YSI sondes and has used EcoNet and other remote telemetry options for several clients’ National Pollutant Discharge Elimination System (NPDES) MS4 permit compliance.
“I’m a big proponent of [the sondes],” says Riddle. “The fact that they give you continuous data in surface water allows you to see things that you otherwise wouldn’t have any idea are going on. We used to use automatic samplers exclusively–I’m not sure automatic is the best way to describe them–to supplement manual grab sampling, and they’re really challenging to use, quite frankly. They require a lot of attention and an awful lot of maintenance, but you’re still getting a snapshot look of sample data at some preprogrammed interval or under set conditions. There is the collection and the setup, the processing time, waiting on the lab; you have restrictions on how much sample volume you can take and how many samples will fit in the samplers.”
Noting that automatic samplers were the only monitoring technology available until several years ago, “the sondes have just totally changed the way we do this and interpret the results,” adds Riddle. “We can make real-time decisions and not only get data to look for trends during a storm or during ambient conditions, but we can even look for illicit discharges, things that shouldn’t be there, so that we can potentially track down a source. If you’ve got a discharge that shouldn’t be there, one of the different analytes–whether it’s a drop in pH or a drop in dissolved oxygen, an increase in turbidity, a drop in conductivity, whatever the case may be–it allows us to see it in real time, whereas before we never could. We’ve got so many more data points of value. We can see not only storm event trends and dry-weather trends but also things that are going on over a monthly or even daily basis, like photosynthesis and respiration and how that impacts dissolved oxygen and temperature–these really complex interactions.”
Riddle does not claim that the use of these new tools reduces upfront costs, maintenance requirements, or staffing requirements. The investment is well worth his ability to do his job better, though. “We get so much more for each dollar than we would have gotten otherwise,” he says. “It’s what you get out of the technology that far surpasses what we would otherwise use.”
An advantage of monitoring with remote telemetry is viewing stormwater monitoring data from any computer with Internet access, Riddle points out. Additionally, he is experimenting with setting alarms for pollutant levels that are above a preset threshold. One alarm might indicate that other pollutants are trending upward at the same time, he says.
Handling Tricky Testing for Detergents
Stormwater quality standards for the San Diego Basin–one of nine regions in California required to adopt a Water Quality Control Plan, or Basin Plan–have gotten stricter since a 2001 NPDES permit was issued, the first since 1990. The new permit contained comprehensive changes to stormwater standards and regulations, and led to the integration of various City of San Diego divisions and, ultimately, the creation of the city’s Storm Water Department.
Stepped-up enforcement efforts to keep detergents out of stormwater runoff were another result: For example, the city soon required pressure-wash operators to recapture their washwater. Then, a 2007 permit added methylene blue active substances (MBAS) to the list of analytes to be tested as part of the routine field screening portion of the Dry Weather Monitoring Program. According to Courtenay White, biologist in the Pollution Prevention Division of the City of San Diego Storm Water Department, other major sources of detergents in urban runoff are household and industrial/commercial cleaning products–which can find their way into stormwater following an old-fashioned driveway car wash.
Despite the fact that the department encourages people to wash their vehicles over a permeable surface or collect their runoff and dispose of it to the sewer system, the practice continues, and MBAS are detected during testing in manholes, outlets, and catch basins. The department uses new CHEMetrics detergents test kits, which utilize a double-point ampoule to deliver solvent and a visual format as well an instrumental read finish. The kits are designed to use less solvent than other surfactants kits and minimize exposure to the solvents. Anionic detergents, some of the most prominent methylene blue active substances, react with methylene blue to form a blue complex that is extracted into an immiscible organic solvent. The intensity of the blue color is directly related to the concentration of MBAS in the sample. Test results are expressed in ppm (mg/L) linear alkylbenzene sulfonate.
White notes that the department uses glass ampoule test kits for all of its field screening. Most of these use a vacu-vial, which, when snapped into the sample cup, draws in the sample/reagent mixture and is read by a light-emitting instrument called a spectrophotometer. Detergent testing is different from the other tests in several ways. First, a double-tipped glass ampoule containing the reagent that mixes with the sample gets snapped at both ends using a special tool. After mixing, the liquid is left to separate in suspension, and the heavier layer is extracted into a test tube. This liquid-filled vial is inserted into the meter. Second, the detergents test kit requires the use of a single analyte meter to read the liquid in the glass vial. It is the only test the department conducts that requires the use of a dedicated instrument to read the results. All of the other CHEMetrics tests conducted make use of a multi-analyte photometer that is programmable for the analyte being tested. Finally, because the reagent contained in the double-tipped ampoule consists primarily of chloroform, all of the waste from the detergents test must be disposed of as hazardous waste.
Monitoring Dissolved Oxygen Trends in Albuquerque
In some cases, aquatic life is threatened by stormwater pollution and monitoring DO–not pollution levels–is the best way to ensure that water in lakes and streams is life-sustaining. Such has been the case in and around the Rio Grande near Albuquerque, NM, where DO is continuously monitored using sensors that provide comprehensive DO data.
In the 1960s, the city built a network of concrete stormwater channels to allow stormwater to find its way to the river because natural levees along the banks otherwise prevented it, explains Todd Kelly, Albuquerque-based hydrologist with the US Geological Survey (USGS). The North Diversion Channel within the network takes stormwater flows from a 90-square-mile area in the northeast part of the city and diverts it to a large “mixing area”–a manmade earthen channel located adjacent to the river. It was in this mixing area that a large fish kill occurred in June 2004, concurrent with DO readings of less than 1 mg/L.
Kelly recalls that the use of handheld DO meters indicated low dissolved oxygen content, probably caused by high biochemical oxygen demand and/or chemical oxygen demand in the stormwater runoff. Eventually, the city will have to modify its stormwater management infrastructure in some way, he says. The USGS’s continuous DO monitoring has illustrated to the city the need to make modifications. Four second-generation In-Situ Rugged Dissolved Oxygen (RDO) optical sensors were deployed in and around the Rio Grande. One was deployed in the mixing area in fall 2008. Three are located in the river itself. In fall 2009, one was deployed upstream of the mixing area to determine what the river is bringing into it. In spring 2010, another was installed farther upstream, near the town of Bernalillo, and another located downstream of most of the city. A fifth probe is used for field checks. In addition to DO, specific conductance, turbidity, and temperature are monitored using other probes.
The RDO sensors measure DO using a “dynamic luminescence quenching” principle. The sensor is equipped with a lens, blue LED and filter, red LED and filter, and a photodetector or photodiode. When the blue LED emits light, it causes the lumiphore molecules embedded in the gas-permeable sensing foil to emit red photons. The “phase,” i.e., delay, of the returned signal compared with the excitation signal is measured. As a result, the measurement is based on the “lifetime” rather than the “intensity” of the luminescence. The concept behind this operating principle is that, although measuring luminescence intensity is easier to implement in optical DO sensors, the data generated are less robust data than with lifetime of luminescence, a physical constant. The development of optical DO technology has been devised to increase the ability to accurately monitor DO levels over long periods of time.
The sensors provide readings every five minutes, providing the USGS with more trend data than was available from its previous periodic site visits. The existing monitoring network has no telemetry; rather, USGS personnel make site visits at least once a month and upload the data into a laptop computer. The sensors are also checked for fouling, although they are designed to be cleaned and redeployed without the need for recalibration.
“EPA was made aware of the fish kill, so this is a high-profile study,” says Kelly. “We needed reliable data, and we didn’t want to be going out there every week. The timing is very critical, and that’s why we needed DO readings every five minutes and why we needed them to be very reliable. The membrane-type DO meters would clog so quickly that we’d be out there every week trying to keep them clean.”
