Development of a Hydrologic Model for an Urban Headwater Stream: The Influence of Pervious Surface Properties on Runoff Dynamics

A hydrologic model was developed for the Mullins Creek (MC) catchment located on the University of Arkansas campus in Fayetteville, Arkansas. The MC catchment is a small, dynamic urban stream system with a range of land use/land cover (LULC), an extensive and well-developed stormwater drainage network, and extensive urbanization (over 90% developed, and almost 50% impervious surface area (ISA)). Selected datasets provided information on the stormwater drainage network, the physical attributes of the catchment and receiving waterway (i.e. drainage area, slope, etc.), infiltration potential of soil map units, LULC, and percent ISA. These datasets were analyzed to provide input parameters and develop a Storm Water Management Model (SWMM) for the MC catchment. To more accurately characterize the infiltration potential of the pervious areas of the catchment, infiltration testing was performed in the field to compare infiltration in paired highly maintained and minimally maintained sites in the same soil map unit in or near the catchment. As a component of the modeling process, a sensitivity analysis was run on two sets of parameters: percent ISA and Horton infiltration rates. The sensitivity of the model output to changes in percent ISA and changes in infiltration rates varied based on storm event size. The MC SWMM model was most sensitive to changes in percent impervious surface for small storm events (25 - 50 mm), while the model was most sensitive to changes in infiltration rates for moderate to large storm events (≥150 mm). In an effort to test the utility of developing a hydrologic model of this type for other small, urban drainage systems, the uncalibrated model was validated using selected storm events. Observed storm hydrograph data were downloaded from a USGS gaging station located at the outlet of the catchment. Overall, the MC model performed well for the selected storm events, although it performed best for the storms that occurred closest to when the LULC data were collected. The uncalibrated model outputs were most accurate when observed soil infiltration data were incorporated into the model, and when the percent ISA in the model was set equal to total impervious area