Submitted to: Journal of Irrigation and Drainage Engineering
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 9/29/2010
Publication Date: N/A
Citation: Interpretive Summary: City landscapes contain some area that can absorb rain (like a lawn or garden) interspersed with some area that cannot (like a driveway or parking lot). This mixture creates complications in predicting flooding and designing storm drainage systems to manage the right amount of water. In making these calculations, engineers use a numeric value (called a curve number) to describe how much water is expected to run off of an area. Engineers look up the appropriate curve number in a table that depends on how much of the area is impervious to water, and characteristics of the soil that exists in between. In this study, we examined whether the published curve numbers for city areas are accurate enough to use in computer models to predict timing and amount of runoff water so that storm sewers can be properly designed to handle storm runoff. We used artificial rainfall on boxes of soil and/or pavement (4-m slope length, 0.6-m wide) and measured the runoff to calculate our own curve numbers. Our measurements showed that the published curve numbers cause computer models to underestimate runoff, which could lead to unexpected flooding. The curve numbers we developed create a more accurate picture of storm runoff in the city. The impact of this research is that city developers and engineers can use our method to more accurately predict how much storm runoff will be produced, and therefore how big the storm sewers and retention ponds need to be to protect people and homes from frequent flooding.
Technical Abstract: Urban drainages are mosaics of pervious and impervious surfaces, and prediction of runoff hydrology with a lumped modeling approach using the NRCS curve number may be appropriate. However, the prognostic capability of such a lumped approach is complicated by routing and connectivity amongst infiltrative surfaces and their impervious counterparts. In this study we investigate the application of empirical curve numbers derived from laboratory runoff simulations to reproduce the runoff response of composite impervious and pervious areas. We applied a 4.3-cm storm (20, 30, 40 mm hr-1 rainfall rates for 0.8, 0.4, and 0.4 hours, respectively) to generate runoff from 0.6 m2 boxes (impervious or pervious-soil) that were connected together in series (0, 25% impervious) to produce different arrangements of impervious surfaces having different connectivity to the outlet (disconnected, connected), under two different antecedent moisture conditions for pervious areas. Curve numbers were calculated from observed cumulative rainfall and correspondent runoff data and compared with published values for similar land use or impervious area conditions. We then used USEPA SWMM 5.0 to generate runoff hydrographs from experimental curve numbers and qualitatively compared these with observed hydrographs. Experimental curve numbers ranged from a minimum of 93 for 0% imperviousness (bare soil) to a maximum of 97 for 25% impervious cover that is connected to the outlet. Experimental curve numbers were overall higher than values recommended in NRCS guidance. When NRCS curve numbers were used to simulate hydrographs in USEPA SWMM, runoff quantities were underestimated compared to observed hydrographs. Simulation of runoff in SWMM 5 based on experimental curve numbers also underestimated hydrographs, but to a lesser degree. Therefore, this experimental approach shows promising results for predicting runoff hydrographs. Researchers are encouraged to use this methodology to generate experimentally-derived curve numbers for other urban land uses.