Evaluating the effects of parameterized cross section shapes and simplified routing with a coupled distributed hydrologic and hydraulic model

Abstract

► A distributed hydrologic and an unsteady hydraulic model are externally coupled. ► Scenarios based on different routing simplifications are examined. ► Consideration of both dynamic routing and floodplain form improved predictions. ► Kinematic and dynamic wave performed similarly for single power law cross section. ► Performance of hydrologic model and lateral inflows are crucial for coupling. With spatially distributed hydrologic models the need arises for determining the channel cross section shape for the entire stream network. In the absence of cross section data, assumed or parameterized cross section shapes are often used. The effects of parameterized cross sections are evaluated in this study by developing a modeling framework that externally couples a spatially distributed hydrologic model, HL-RDHM, with a one-dimensional unsteady hydraulic model, HEC-RAS. The evaluation emphasizes the effects of parameterized cross sections on simulated flows by focusing the analysis on the portion of the basin’s main stream reach where detailed cross section data and observed streamflows (at both ends of the reach) are available, and by developing and testing three cross section scenarios. The scenarios are designed to increase sequentially, in a stepwise fashion, the complexity of the parameterized cross section, starting with a single roughness parameter and channel power law cross section shape and then including additional power law or roughness parameters. This is done stepwise to help distinguish the effects associated with each parameterization, and decide the required level of cross section detail. The scenario simulations are evaluated using split sampling, changes in measures of performance and hydrograph agreement, hypothesis tests on Nash–Sutcliffe values, and overall predictive uncertainty. The coupling framework is applied to the Blue and Illinois River basins, in Oklahoma, US. Overall, we found that in these basins the coupling tends to improve predictions when dynamic wave routing and floodplain cross section geometry are considered concurrently. For this scenario, we found that on average typical measures of model performance may be improved and, based on a quantitative and qualitative assessment, uncertainty may be reduced. We also found that dynamic wave routing does not tend to perform better than kinematic wave routing for the most basic scenarios with a single power law cross section shape. Further, results indicate that the distributed hydrologic model performance at the main outlet and at the upstream boundary of the hydraulic model, and the relative contribution of lateral inflows, are key factors that need to be considered when deciding the applicability of the coupled framework to other basins. In the future, to effectively use resources, it will be beneficial to automate the coupling and accompany its application with a priori criteria for selecting those basins where benefits are most likely.