Nebraska LTAP
Date of this Version
5-2025
Document Type
Article
Citation
Maughan, T.A., Hotchkiss, R.H., and Hales, R. (2025). "Drainage Area Limitations of Single Watershed, Peak Flow Estimates from NRCS Methods". NDOT Research Report SPR-FY24(033).
Abstract
Most state Department of Transportation roadway design sections predict peak flow for culvert design using, amongst other approaches, Natural Resources Conservation Service (NRCS) TR-20 technology. Even though this technology is more than 50 years old, there are no clear guidelines for how large a single watershed drainage area may be while remaining appropriate for predicting peak discharge with this method. Our objective was to identify the drainage area where TR-20 peak flow predictions significantly deviate from flow frequency predictions. We developed flow frequency estimates for 130 small-area stream gage sites in rural Nebraska and compared the calculated return period discharges with those from TR-20 using both the segmental and lag equation approaches for estimating the time of concentration. Additionally, we compared available regression predictions to both flow frequency and TR-20 estimates.
We found that there are no significant differences between peak discharges calculated using the TR-20 lag method and segmental method for estimating the time of concentration. If TR-20 continues to be used in the future, we recommend using the segmental approach to be more consistent with commonly accepted practice. Results did show, however, that predictions are consistently higher than those from stream gage estimates and become worse for drainage areas larger than fifteen square miles. The regression equations developed for small drainage areas (perhaps uniquely available for Nebraska) perform better than the TR- 20 estimates. As a first step to further investigate the performance of the TR-20 equations, we made peak flow estimates assuming drier soil conditions that effectively reduce the runoff curve number for each watershed. Results again showed poor agreement, but instead of being consistently high, were consistently low. We therefore discourage using an uncalibrated TR-20 model to calculate peak flow for culvert design for any size drainage area in Nebraska. If peak discharge estimates are required for changing land use conditions, we recommend a TR-20 model be calibrated to the regression model results for present conditions, thus allowing the simulation of changed land use conditions easily done with TR-20.
Next generation hydrologic approaches such as the National Water Model and GEOGloWS currently lack the resolution required to simulate peak flows from smaller watersheds. Tests showed universally low estimates compared to gage estimates of return period discharges.