Effects of Watershed Subdivision on Peak Discharge in Rainfall-Runoff Modeling in the WinTR-20 Model

Abstract

The rainfall-runoff model, WinTR-20, uses subdivision to simulate runoff behavior for complex watersheds exhibiting heterogeneous conditions or storage. It has been shown by others that subdivision generally causes the predicted peak discharge to increase, though the underlying processes are often obscured by watershed complexity. This study instead focuses on a simplified, theoretical watershed, systematically comparing the unsubdivided watershed with a two-subbasin model in order to determine the most sensitive factors. Peak discharge sensitivity is evaluated with respect to (1) series subdivision with varying total area, (2) parallel subdivision with varying proportional area, (3) parallel subdivision with varying curve number, and (4) parallel subdivision with simultaneously varying area and curve numbers. Peak discharge is most sensitive to differences in curve number, which controls both the runoff volume and peak timing. Serial subdivision was found to produce a significant high peak discharge, regardless of relative area, while parallel subdivision produced a smaller and more variable effect, either increasing or decreasing peak flow based on the area ratio. Using these subdivision sensitivities, general guidelines are presented for the rational subdivision in rainfall-runoff modeling. For example, subdivision is recommended when subarea curve numbers differ by more than five and the relative sizes of subareas influence the effects of discretization.