Predicting discharge capacity of vegetated compound channels: uncertainty and identifiability of 1D process-based models

Abstract

Abstract. Despite the development of advanced process-based methods for estimating the discharge capacity of vegetated river channels, most of the practical one-dimensional modeling is based on a relatively simple divided channel method (DCM) with the Manning's flow resistance formula. This study is motivated by the need to improve the reliability of modeling in practical applications while acknowledging the limitations on the availability of data on vegetation distributions and densities required by the process-based methods. We investigate whether the advanced methods can be applied to modeling vegetated compound channels by identifying the missing characteristics as parameters through the formulation of an inverse problem. We developed a new probabilistic approach for comparing six models of channel discharge capacity in respect of their uncertainty, with the model with the lowest uncertainty considered the most favorable. Calculations were performed for flume and field settings varying in floodplain vegetation submergence, density, and flexibility, and in hydraulic conditions. The output uncertainty, estimated on the basis of a quasi-Bayes approach, was analyzed for a varying number of observation points, demonstrating the significance of the parameter equifinality. The results showed that very reliable predictions with low uncertainties can be obtained for process-based methods with a large number of parameters. The equifinality affects the parameter identification but not the uncertainty of a model. The best performance for sparse, unsubmerged, rigid vegetation was obtained with the Mertens method and for dense, flexible vegetation with the generalized two-layer method combined with a description of the flexibility-induced reconfiguration. We found that the process-based methods are superior when applied for vegetative conditions they were developed for while the Manning based DCM seems to be the most flexible technique.