By Teresa Zumwald

After several significant storms in York County, VA - including Hurricanes Dennis and Floyd in 1999 and a summer 2000 storm that dumped 8 in. of rain in six hours - the county knew it was time to update the Moores Creek drainage study conducted more than a decade ago.

Complaints of roadway and neighborhood flooding in the Moores Creek area were mounting, according to Anna Drake, manager of environmental programs and an engineer in the county's Department of Environmental and Development Services. "When you can canoe down the street, that's an issue," she remarks. "Citizens had good reason to want improvements for drainage."

But Drake and others knew that short-term fixes here or maintenance improvements there wouldn't get to the root causes of drainage problems in Moores Creek. "Sure, we could do something to lower the water surface elevation here, but how did that impact someone else?" she says. "We wanted to come up with the most environmentally sound fixes to improve drainage in the entire watershed."

Defining and Solving Drainage Problems in a Complex System
For Drake, a holistic approach meant developing a computer model that truly represented field conditions so solutions could alleviate flooding along major watercourses in the drainage area. The model also had to reflect all the changes and improvements that had occurred since the previous drainage study in 1988. For example, a development boom through the 1990s had produced at least eight major subdivisions, new churches, a high school, and a fire station in the 960-ac. Moores Creek watershed, which includes a mix of woodlands, wetlands, neighborhoods, schools, churches, and farms. Drake says that while stormwater management controls were constructed in conjunction with all this development, no downstream improvements had been made, and the natural drainage ditch was never adapted to handle the additional volume of stormwater.

For example, new retention and detention facilities, designed independently from other drainage systems in the area, did not consider the effect on the entire watershed, which included drainage from existing subdivisions. This independent development resulted in a net increase in stormwater runoff; discharge from each new subdivision was at a reduced peak flow - but not at a reduced volume. Thus, this combination produced a large, cumulative increase in stormwater runoff downstream. As development continued, the existing system became inadequate for handling the increased runoff.

Drake knew drainage solutions would not be straightforward because the Moores Creek watershed, located in a relatively flat coastal area, is complex, thus making modeling challenging. For example, the flat topography makes it difficult to determine exactly where floodwaters spread or the extent of flooding once stormwater exceeds channel banks. "We also knew that drainage solutions would be complicated by the outfall, which is tidally influenced," Drake explains.

The drainage system includes a series of open channels, roadway culverts, and stormwater management ponds. The system encompasses large neighborhoods, and portions drain into ponds, some of which are interconnected. A long, natural channel draining several subdivisions travels through a large wetland area.

"A lot of water gets stored in the wetland area, and that's OK, but the channel was never meant to carry this much runoff," Drake says. "The water in the wetlands gets so high it spreads out over a large area - acres and acres - affecting areas outside the watershed. If we enlarged the channel, we could end up draining the wetlands."

Some past roadway drainage improvements, which have drainage pipes crossing underneath roadways in several locations, have been inadequate, producing upstream flooding, Drake notes. The drainage area also features ditches - now undersized - located along homes built in the 1930s. Together all these conditions in Moores Creek presented a modeling challenge.

"We didn't want a model that just worked on paper," Drake says. "We wanted it validated by considering citizen input and [using] flow monitoring."

Soliciting Citizen Input
In 2001, York County hired Woolpert LLP to complete the Moores Creek drainage study, which included six major tasks:

1. Analyzing existing data and performing a field survey to collect topographic data, channel information, and attribute data, such as pipe sizes and invert elevations
2. Performing a condition assessment of drainage channels and roadway culverts
3. Completing a wetland delineation
4. Collecting rainfall data and conducting flow monitoring
5. Developing and calibrating hydrologic and hydraulic models for analyzing existing conditions and various drainage improvement options
6. Making recommendations to the county

Drake and Joe Sisler, the county's manager of engineering, defined four objectives for the project:

Objective 1: Eliminate flooding during a 10-year storm event so stormwater will be contained in the system, stay within channel banks, and not overtop roadways. Roadways should be visible and remain passable during a 10-year storm event.

Objective 2: Prevent erosion of all channels 12 in. or greater in depth from a two-year storm event.

Objective 3: Provide conveyance under the watershed's two major thoroughfares for a 25-year storm event.

Objective 4: Provide a solution that will help ensure that garages and homes are not flooded during a 100-year storm event.

The first priority was getting an overall feel for drainage and flooding problems and compiling general trends. Although citizens had called the county over the years to voice flooding complaints, the county did not have a consistent, automated method for logging complaint locations and details. Woolpert needed a cost-effective way to get residents' input about drainage problems. Precisely where had residents experienced flooding in the past? And during which storms?

The county considered hosting a public meeting, but Woolpert suggested a more cost-effective option that would require less time and fewer staff and perhaps produce better results: Why not solicit citizen input about drainage problems and flooding concerns through a brief questionnaire supplementing the county's Permission to Enter Property form? The county was preparing to send the form to obtain residents' permission for surveyors to enter properties as needed to collect survey data for the drainage study.

"We wanted to move along quickly, and this solution cost us virtually nothing because we had to send the permission mailing anyway," Drake relates. Her office mailed 450 Permission to Enter Property forms, and of those returned, 160 included a response to the question about historical flooding information. Residents sent handwritten responses, sketches, typed letters, and color photographs of neighborhood and personal property flooding; one resident sent a video along with a photo album showing backyard flooding. Still others appeared in person at county offices to discuss their flooding concerns. "Even those who wrote ‘I have no flooding problems' gave us great information for the study," Drake remarks.

Some went into great detail on flooding locations, dates, and extent of damage:

• During strong rains, the large ditch beside my house becomes full of water. This ditch serves the whole street plus houses in the development…. The water flows quite rapidly. Children could drown if they get caught in the ditch. I was wondering if a covered culvert could be put in this ditch to help solve this problem. This has been going on for 14-plus years, as long as we have lived here.
• During Hurricane Floyd (fall '99), our property … was completely surrounded by stormwater that had backed up from the drainage pond behind our property. The water was approximately one inch from overflowing the side door to the garage, on the west side of the property. Water in the street in front of the property was mid-thigh on a person of average height (5¢ 10?) and much deeper where [our street] intersects [another street]. [Our street] was not passable by car, and the only means of escape would have been by wading.
• During the rainstorm associated with Hurricane Floyd of September 1999, stormwater backed up 18? in my garage, and the crawl space was flooded up to the bottom of floor joists.
• During September 1999, when Hurricane Floyd came through, water from the concrete ditch in the back of my yard came within 10 feet of my house. It flooded my wooden shed, and there was a "watermark" on the side of the shed that indicated approximately 1_ feet. My riding lawn mower had to be taken in for repairs due to water damage. There were two 8? x 8' railroad ties that were behind my shed; they floated to about 15 feet from my house.

Many residents reported drainage problems related to Hurricane Floyd and described how flooding affected their homes (e.g., "Waters entered the garage and family room, which is on a slab"). Those living close to a lake reported flooding of neighborhood roadways and yards during storms in the late 1990s. Residents identified minor drainage problems (e.g., standing water in ditches) that typically occurred after specified periods of rain, and they reported maintenance issues that needed attention.

Jack Wall, a project engineer at Woolpert, reviewed and ranked all 160 responses as severe, moderate, or none. Severity rankings were color-coded and plotted in a geographic information system (GIS) parcel map, called the historical flooding map of citizen complaints, so Wall could determine where the worst flooding problems were occurring and focus on those areas.

Wall was surprised at the locations and severity of some flooding problems reported. "Someone can say there is flooding, that the roadside ditches are full, but when you have pictures of a whole neighborhood pretty much underwater, that really brings it to light," he says.

The citizen input was useful not only in identifying the location and extent of drainage problems but also in validating the results of the hydraulic model, which Woolpert used to make recommendations for improvements. "The residents' written observations let us verify that what we were predicting was actually occurring in the field," Wall says.

Drake says the citizen input redirected some of the county's initial thoughts about where flooding problems were likely occurring. "We expected more people around the lakes to be complaining," she notes. "Although there were some, the citizen input changed our focus to residents downstream from the lakes. We focused our attention on the outfalls."

By reviewing photos supplied by residents, Woolpert could determine whether a certain roadway flooded during a particular storm event. "So when we modeled a similar size or type of rainfall event, we knew the model results would have to show roadway flooding," explains Shelly Frie, P.E., manager of the project for Woolpert. "We checked to see if the historical flooding map of citizen complaints correlated with our model showing flooding in the same locations. That helped us determine if our model was correct."

Woolpert could even calculate water levels on some photos. "That would tell us the approximate water elevation level for a particular storm so we could see whether it matched what we had in the model," Frie says. Elevations that didn't match were addressed during model calibration. "We knew we were either overestimating or underestimating the amount of flooding in the model," she says.

The citizen input was valuable because it provided data on portions of the system too small to be included in the model and where no flow monitoring data were available. While the majority of recommendations were based on modeling results, Woolpert made two recommendations based solely on citizen input; these included a spillway at an existing pond and a new ditch. Woolpert recommended constructing the ditch after a resident complained about backyard flooding. When the neighborhood was built, an earthen embankment on a wooded parcel was constructed to prevent upstream runoff from entering the subdivision. During large rain events, however, water was breaching the berm and flooding residents' yards. Because the properties have little grade, and the storm drain system was not designed to accommodate the excess runoff, backyard flooding resulted. "If the resident hadn't reported the problem, we would have had no idea [it] existed, and we would not have been able to make recommendations to solve it," says Frie.

This approach to citizen input also helped the county begin associating flooding complaints with addresses - information that forms the basis of the county's new flooding-complaint database. "Now when people call in, we log the complaint in the database," Drake reports.

Without the citizen input, it would have been much more difficult to determine the location and extent of flooding using the hydraulic model alone. What's more, Wall notes, "I don't think we would have been able to address as many of the complaints in the recommendations. We really felt that if residents took time to answer our questions, we had to have a plan for fixing the problems."

Conducting Flow Monitoring
A GIS was used to help determine initial model parameters, such as size of drainage subareas, SCS curve numbers, and time of concentration. The Army Corps of Engineers's Hydraulic Engineering Center Hydrologic Modeling System (HEC-HMS) was used to model the watershed hydrology and compute peak runoff flows for the two-, 10-, 25-, and 100-year storm events. The River Analysis System (HEC-RAS) was used to model the channels to determine the hydraulic grade line and flooding areas.

Drake knew flow monitoring was crucial for hydrologic and hydraulic model calibration because county information, such as flooding data and high watermarks, was not available for validating modeling results. Because the flat topography made it difficult to predict where stormwater was flowing, the flow monitoring provided some answers. "If you spend a lot of money on modeling but don't know how close you are to what's happening in the field, how can you have any confidence in your model?" Wall asks. "How can it answer your questions?"

Woolpert installed flow meters in two strategic locations where large areas of the watershed could be monitored. Each location was monitored for flow depth and flow rate. Two rain gauges at one location were installed, one for redundancy. The initial flow monitoring was conducted for 75 days (July through September 2001); only one significant rainfall event occurred.

"Because we had unusually dry conditions and didn't have many storm events to compare our modeling data against, we didn't get a good correlation with data in our model, and we weren't comfortable with the huge parameter adjustments that would have been required," notes Frie. For example, the flow meters recorded 6 ft.3/sec. (cfs) at one station; the noncalibrated model, however, predicted a flow of 89 cfs. At another station, the observed volume of water was 670,000 ft.3, but the predicted volume was 5.5 million ft.3

"The anticipated flow rates compared to the actual, observed flow rates were so disparate that the model could barely be calibrated within reason," Drake says.

John Hudgins, director of the department, and Brian Woodward, chief of utilities/stormwater, agreed that additional flow monitoring should be performed in late winter/early spring 2002, when wet-weather conditions were more likely and better data could be generated. Four large storm events were used for calibration after this second round of flow monitoring. Flow values were entered into the hydrologic model, HEC-HMS; parameters were adjusted so the hydrographs more closely matched what was observed in the field for all four storm events.

Woolpert used calibrated parameters in the hydrologic model to predict the runoff for two-, 10-, 25-, and 100-year storms; those values were used in the hydraulic model to predict the water surface elevations in channels and at culverts. Recommendations were based on the 10-year storm event.

"We got flows that could be compared easily with the model, so we felt comfortable with the calibrations that were done," Drake says. For example, one observed peak flow was 17.5 cfs while the calibrated peak flow was 18 cfs. "The adjustments were all supportable."

The county faced an additional challenge during the second round of flow monitoring: Meter sensors began getting covered with sediment and weren't recording accurately. The county and the Virginia Department of Transportation had to clean meters frequently; even Drake spent time in a ditch shoveling sediment out of pipes until another alternative was proposed: raising the monitors slightly to avoid sediment accumulation. This was accounted for in the model.

In the end, Woolpert analyzed 150 cross-sections, 16 culverts, and more than 12,000 lin. ft. of channel in the hydraulic model and determined three general causes of flooding:

• The lack of capacity in the existing drainage system
• Overflow from one drainage subarea to the next
• The lack of a drainage system altogether in some areas

Drake says flow monitoring was key to gaining confidence in the hydrologic model results, which varied considerably from the initial prediction of peak flows using standard methods for selecting hydrologic parameters. From the historical flooding information and hydraulic analysis, Woolpert determined that six areas in the watershed had drainage problems related to excessive flooding. The model was used to analyze various alternatives and solutions.

"Because the drainage system is so complex, and many elements are interconnected, there were many options," Frie says. "We looked at how each solution affected the rest of the drainage system, and if it had a positive or negative effect. If we widen or deepen the channel or fix the downstream culvert, does that solve the problem?"

Proposed recommendations included increasing channel and roadway-culvert capacity, adding drainage ditches, and adding an overflow structure at a stormwater pond. "We always referred back to the historical flooding map of citizen complaints to make sure we stayed focused on citizen concerns," Wall notes.

Lessons Learned

A York County resident sent these photos, plus notes, revealing flooding at Yorktown Road. The hydraulic model also predicted flooding here.

In many municipalities, decisions about capital improvements are based on citizen complaints—the squeaky wheel gets the grease. The more complaints in an area, the more likely that area will be considered for capital improvements. But rarely do municipalities receive such valuable feedback from residents, to the extent that York County did, for such a limited investment of time and money. "If a local government is conducting a drainage improvement study, there is great value in actively soliciting input from residents who live in the area," Drake says.

Frie observes that because citizen input can affect the solutions suggested, it's important to obtain citizen input up-front. It might even be useful to send additional requests for flooding information to residents outside the area being surveyed—even to residents in areas not near a drainageway or where flooding isn't expected, she says.

Drake cautions that local governments requesting quantitative, objective reports shouldn't be surprised if they get subjective responses. "Some people thought they had terrible flooding if they had water standing in their backyard. But that is certainly not considered terrible flooding when others have cars that are underwater. We asked residents to state in inches how much flooding they'd experienced, but we still got some subjective responses."

Drake says the citizen input was so valuable, it was like having additional flow monitoring. "If a person says, ‘I have 4 inches of water in my garage,' you can shoot an elevation on that garage, correlate it to a particular storm event, and then test it against the model," she explains.

Frie says local governments should be prepared to do additional flow monitoring if Mother Nature doesn't cooperate with wet-weather events. Flow meters should be checked and cleaned routinely so sediment doesn't accumulate. "Don't underestimate the level of effort it takes to clean out pipes and keep gauges free from debris," she stresses. "Some counties may not be prepared for the work required or may not be sure which governmental entity is responsible for keeping pipes clean."

If an area includes wetlands, Drake recommends completing and certifying a wetland delineation long before the first recommendation is proposed. "Otherwise recommendations may not be practical from a permitting standpoint," she says.

Next Steps

Woolpert plotted 160 citizen responses, based on color-coded severity rankings, in a GIS parcel map, which made it easier to determine where the worst flooding problems were occurring in Moores Creek. For a larger picture click here.

Now that the modeling is complete and recommendations have been proposed, York County will use Woolpert's condition assessment of open channels and roadway culverts to prioritize maintenance efforts and systematically solve some immediate problems.

"Sometimes your first thought is to clean the ditch adjacent to the property owner who's complaining," Drake says. "But by knowing the condition of your entire system, you might discover that maintenance dollars would be better spent cleaning ditches farther away. Sometimes it's not so intuitive where to clean."

Drake says that when construction is ready to begin, the citizen input forms will be reviewed prior to public meetings so county staff can anticipate questions. "I think we will get better citizen buy-in, especially because we've used their data. The higher our rate of confidence, the more we can support the model."

Drake explains that the flow monitoring will help the county Department of Environmental and Development Services sell the recommendations to county decision-makers, who can feel confident they will be spending money on improvements that are truly needed. Without the flow monitoring, Frie notes, the model would have shown - incorrectly - a great deal more flooding.

"You could end up making unnecessary improvements," says Frie. "I've known municipalities that feel more secure in making such improvements - even if the historical results say a roadway didn't flood during the last storm event and the model doesn't show flooding. You're more cautious if you don't have any data to back up your model."

Guest author Teresa Zumwald is based in Ohio.

SW - January/February 2003

Home + About + Subscribe + News + Calendar + Glossary
Talk + Images + Advertise + Contact Us + Search + Register + Services

Erosion Control Magazine | MSW Management Magazine
Grading & Excavation Contractor | ForesterPress | Forester Media

© 2000 - 2003 FORESTER MEDIA, INC.