Back

**Reexamining Runoff Processes & Erosion in Urban Areas**
# Predicting the Magnitude of Erosion Downslope of Pipe Outlets

Oral

Stormwater (UWRRC)

Wednesday, May 24, 2023

3:18 PM – 3:36 PM PDT

Room: Del Fuego

Sarah E. Waickowski, PE (she/her/hers)

Environmental Engineer IV

Tetra Tech, Inc.William F. Hunt, III, PhD, PE, D.WRE

William Neal Reynolds Professor & Extension Specialist

North Carolina State University

Excess sediment discharged to streams is detrimental to channel stability, aquatic health, and water quality. Gullies are the result of concentrated flow, often from pipe outlets, forming headcuts that migrate towards the source of concentrated flow. Few resources exist that account for gully erosion or predict the magnitude of erosion for gullies that have or will occur downslope of pipe outlets. Sixty pipe outlets across North Carolina were assessed for downslope erosion. Data were collected immediately downslope of the pipe outlet, at the adjacent property’s furthermost boundary, and at the outfall or stream. Additional data were collected when there was a noticeable change in channel geometry and or stability. These data were used to develop regression equations and decision trees to predict the magnitude of erosion downslope of pipe outlets. The equations and trees were built using 85% of the data; the remaining 15% verified model performance through root mean square errors (RMSEs).The magnitude of erosion was quantified in terms of the ratio between cross-sectional dimensions at the top of bank and the respective bankfull dimensions as well as the estimated total volume of eroded soil per channel length. Key predictors for the equations and decision trees included the duration of runoff and peak discharge for the 1-yr and 10-yr, 24-hr storm events, percentage of hydrologic soil group (HSG) C and D soils downslope of the pipe outlet, and the composite curve number (CN) for the pipe outlet’s watershed. The RMSEs for the regression equations predicting the ratios for the cross-sectional area, width, and maximum depth were 10.20, 9.15, and 7.80, respectively. The RMSE for the regression equation predicting the volume of eroded soil was 0.0067. The RMSEs for the decision trees predicting the cross-sectional area, width, depth, and volume were 15.89, 8.45, 6.20, and 0.0017, respectively.