Evaluation of overland flow models using laboratory catchment data III. Comparison of conceptual models

Open Access
Publisher Website
Google Scholar
Abstract
In the previous paper of this series (Bell et al., 1989), data from 96 experiments on a marine-plywood plane were used to investigate the performance of numerical models of overland flow based on the kinematic wave approximation. These models incorporated laminar and turbulent friction relationships, based on Darcy-Weisbach, Manning and Colebrook-White formulations, and four parameters were required to account for the effect of raindrop impact on effective roughness. In this paper a simpler kinematic formulation based on the Manning turbulent flow relationship (with two parameter values to allow for raindrop impact) is tested, and although physically less realistic, is shown to give comparable fits to the experimental data and to be more readily amenable to automatic optimization. The conceptualization of overland flow is extended further, to consider linear and non-linear reservoir cascades. The simplified kinematic wave model is shown to be superior in fit, in consistency of optimized model parameters and in split-sample tests, and has the advantage of physically interpretable parameters, but this improvement in performance is achieved at significantly greater computational cost.