Surface-water quality and flow Modeling Interest Group
U.S. Geological Survey
30 Brown Road
Ithaca, NY 14850-1573
Internet: pzarriel@usgs.gov
Phone: (607) 266-0217 x3014
FAX: (607) 266-0521
This version of the fact sheet has all of the figures converted to thumbnails with links to the larger images. A version with all of the figures inline is also available; the download time will be longer, but the inline figures may be more convenient for viewing and printing.
Citation:
Zarriello, P.J., 1999, Watershed modeling approach to assessing the hydrologic
effects of future development in the Ninemile Creek Basin, Onondaga County,
New York: U.S. Geological Survey Fact Sheet FS 112-99, 8 p.
Background
Model Selection
Model Development and Calibration
Effects of Future Development
Effects of Detention Basins
Future Model Applications
Selected ReferencesThe Town of Camillus, a suburb of Syracuse, N.Y., like many suburban communities, has undergone recent growth and is expecting continued residential and commercial development. Concern over the hydrologic effects of future development has prompted efforts to assess the likelihood that (1) flooding of Ninemile Creek in parts of Camillus will increase as the amount of pervious surface area available for infiltration decreases, and (2) the use of stormwater detention basins to mitigate flooding could worsen flooding whenever the peak outflow from a basin coincides with the peak discharge in the receiving stream, thereby producing a larger peak discharge than would occur otherwise. In 1996, the U.S. Geological Survey, in cooperation with the Town of Camillus, developed a precipitation-runoff model representing a 41.7-mi2 part of the Ninemile Creek watershed (fig. 1) to assess the timing and magnitude of peak discharges that could result from future development and the use of stormwater-detention basins. This fact sheet describes the results of the model simulations.
Figure 1. Principal geographic features of Onondaga County, N.Y.,
and location of the Ninemile Creek watershed model area.
Calibration, a critical step in the development of any simulation model, is the process of comparing simulated streamflows with observed streamflows over a range of conditions, then adjusting the values assigned to the model variables until the discrepancies are minimized. Through the calibration process, model errors are calculated, and their effect on the simulation results are evaluated. Uncalibrated or poorly calibrated models can produce erroneous or misleading information and could result in poor planning or engineering decisions. The correlations between simulated and observed annual and monthly runoff for water years 1989-96 (October 1, 1988 through September 30, 1996) and between simulated and observed stormflow volumes and peak discharges for 30 non-winter storms in the Ninemile Creek watershed are depicted in figure 2.
Figure 2. Simulated runoff, volume, and discharge of Ninemile
Creek at Camillus, Onondaga County, N.Y., in relation to observed values,
water years 1988-96.
The annual maximum peak discharge in this area typically results from combined rainfall and snowmelt. For example, rain and snowmelt in January 1996 produced one of the highest recorded peak flows (2,530 ft3/s) on Ninemile Creek at Camillus (fig. 3B); this discharge occurs on average once every 15 years. This storm also produced the highest mean daily discharge recorded at the Camillus stream gaging station (1,690 ft3/s) since the station began operation in 1958. The simulated values for this 3-day storm (January 18-20, 1996) were within 3 percent of the observed peak discharge and within 10 percent of the runoff volume (fig. 3).
Figure 3. Winter runoff and snowpack buildup and melt at Camillus,
N.Y., 1995-96: (A) Simulated and observed snow-pack water equivalent. and
(B) Simulated and observed discharge of Ninemile Creek.
Runoff after future development will probably reflect the increase in open/residential and impervious lands. The most effective way to represent these factors in the model is uncertain because future drainage patterns are unknown, but comparing the simulated hydrologic effects of development as open/ residential land with the effects of development as impervious land provides an indication of the watershed's probable response to future development. Examples of the simulated watershed response to spring and summer storms under the two types of buildup are plotted in figure 4 and summarized in table 1.
Figure 4. Observed and predicted spring and summer stormflows of
Ninemile Creek at Camillus, N.Y. under present conditions and with future
development at 10-, 50-, and 100-percent buildup, represented in two ways:
A. As open/residential land, and B. As impervious land.
Table 1. Predicted increases in discharge in Ninemile Creek at
Camillus, N.Y., resulting from future development as open/residential land
and as impervious land.
The simulated effects of both types of development were most pronounced in the summer, when soil-water storage is low, and infiltration occurs readily in pervious areas. Precipitation during other times of the year, when soil-water storage is near capacity (because evapotranspiration is low), does not infiltrate and produces mostly surface flow; thus, the effects of increased impervious-surface area are smaller during these seasons than in the summer. Increases in peak discharge from large storms were also less than the increases resulting from small storms (fig. 5) because large storms deplete soil-water storage and produce mostly surface flow.
Figure 5. Simulated 1995-96 peak stormflow of Ninemile Creek at
Camillus, N.Y., based on future development at 10-, 50-, and 100-percent
buildup as: A. Open/residential land. B. Impervious land.
A comparison of log-Pearson Type-III probability curves (a method used to estimate the magnitude and frequency of floodflows) for simulations of development as impervious land indicated that the increase in peak discharge was about the same for storms of all magnitudes. Thus, the relative effect of increased urbanization diminishes with increasing peak discharge. Analyses of peak discharges for the 1989-96 water years indicate that, under 100-percent buildup as impervious land, stormflows that now occur on average once every 2 years will occur once every 1.5 years, and stormflows that now occur on average once every 5 years will occur once every 3.3 years. Future development is not expected to cause significant increases in flooding along Ninemile Creek during large-magnitude storms but is expected to increase the frequency and magnitude of flows resulting from small storms. These simulations also indicated a slight decrease in discharge of 3-day low flows.
Urbanization also affects three components of runoff-- baseflow (sustained ground-water flow), interflow (fast-responding ground-water flow), and surface flow, as indicated in figure 6 for simulations of the July 15, 1996 storm. Baseflow was only slightly lower in impervious-land simulations (0.116 in.) than in open/residential-land simulations (0.118 in.). This small difference is attributed to the relatively small area (about 7 percent of the watershed) that would be affected with 100-percent buildup as impervious land; the decrease in baseflow would probably be larger, however, in subbasins that have a relatively large percent increase in impervious area. Interflow was 74 percent smaller in impervious-land simulations (0.029 in.) than in open/residential-land simulations (0.111 in.), and surface runoff was about 165 percent greater in impervious-land simulations (0.114 in.) than in open/residential-land simulations (0.043 in). Accordingly, the peak discharges in impervious-land simulations were larger, and the response to precipitation more rapid, than in open/residential-land simulations. Flow component differences between the two types of simulations diminished as soil-water storage approached capacity.
Figure 6. Simulated runoff components of Ninemile Creek at Camillus,
N.Y. for the July 15, 1996 storm, for buildup represented as open/residential
land and as impervious land.
Model simulation of the effect of a basin serving a hypothetical 147-acre moderate-density development adjacent to Ninemile Creek at Camillus indicate that, the basin outflow would exceed the uncontrolled flow that would occur if the basin were absent (fig. 7) for a period after the peak discharge in Ninemile Creek. During this period, the difference between the detention-basin outflow and uncontrolled flow is less than 1 percent relative to flow in Ninemile Creek at Camillus. Simulation results for a hypothetical detention basin with half the original storage capacity indicated that the difference between basin outflow and uncontrolled flow would be about twice as large as with the full-size basin. Additionally, the maximum difference between basin outflow and uncontrolled runoff would occur closer to the time of peak discharge in Ninemile Creek. As a result, the smaller basin would contribute about 2 percent more flow to Ninemile Creek at Camillus near the time of the peak than would uncontrolled runoff from the development. Although the increases in peak discharge that would result from the coincidence of peak basin outflow with the peak flow in Ninemile Creek are small, the increases in peak discharge could be cumulative as new development increase and stormwater controls are added.
Figure 7. Predicted flows in Ninemile Creek watershed, Onondaga
County, N.Y., resulting from a 100-year, 24-hour storm at a hypothetical
147-acre residential development with and without a stormwater-detention
basin.
Many parts of the watershed that are expected to undergo development drain to streams tributary to Ninemile Creek; channel storage along these tributaries, particularly those with extensive wetlands, will decrease and delay peak discharge. Simulations of a detention basin that drains to a tributary, rather than directly to Ninemile Creek, indicated no increase in peak discharge at Camillus under any flow conditions. The peak outflow from the detention basin occurred long enough after the peak discharge in the tributary that the peak flows in Ninemile Creek for storms of all magnitudes were decreased.
Runoff from the simulated development, which drains to a tributary, would significantly increase the peak discharge in the tributary in the absence of a detention basin. The simulated peak discharge in a tributary for a 2-year storm was more than twice the peak discharge under current (undeveloped) conditions, and the simulated peak from a 100-year storm was about a third larger. Adding a detention basin decreased the peak discharge of the tributary to about the same as that under predevelopment conditions.
Moffa, P.E., Nix, S.J., and Dent, S.A., 1994, A watershed approach to CSO abatement, Onondaga County, New York: Syracuse, N.Y., Moffa and Associates Consulting Engineers, Unpublished Report for Onondaga Lake Management Council, 8 p.
Zarriello, P.J., 1999, A precipitation-runoff model for part of the Ninemile Creek watershed near Camillus, Onondaga County, New York: U.S. Geological Survey Water Resources Investigation Report 98-4201, 52 p.
Copies of this report can be purchased from: U.S. Geological Survey, Branch of Information Services, Box 25286, Federal Center, Denver, CO 80225-0286
Back to the SMIG Features Page
Home | Mailing List | Features | Conferences | Classes | Reading | Model Archives | Feedback