Mathematical modelling of sediment transport and bed transients in multichannel river networks under conditions of unsteady flow.

Abstract

The fixed-bed, hydrodynamic and water quality model ONE-D has been successfully applied to virtually every major river system in Canada. In 1984 one of the supporting agencies, Environment Canada, expressed the desire to extend ONE-D for the simulation of sediment transport processes and river-bed dynamics. After performing a literature review it was found that currently available mobile-bed river models did not have the features that the basic ONE-D model had, particularly those required to model channel networks. Because of these numerical modelling limitations it was decided to proceed with this research effort. Sediment transport and bed dynamic routines were written in modular form and the resulting model (ONE-D-SED) became the mobile-bed version of ONE-D. ONE-D-SED was initially validated for hypothetical test cases. This initial application showed that the model reproduced analytical solutions for the 'linear' cases examined. In addition, these simulations provided documentation of the numerical properties of ONE-D-SED's finite difference schemes. Following these preliminary tests correction coefficients were generated for analytical solutions of bed transients of finite height. The subsequent application of ONE-D-SED to the Lower Fraser River was partly in response to the need expressed by Public Works Canada for accurate sediment budgets for certain reaches of this major river system. More importantly, part of this system is tidal, multi-channel network and hence is ideally suited for model validation purposes. Finally, superior sediment data were available for the study reach identified in this research. After analysis of Lower Fraser River sediment data, and calibration of the Ackers-White (A-W) sediment transport equation. ONE-D-SED was applied to the study reach at a "screening-level". Since the results of this application were encouraging, ONE-D-SED was then applied at a detailed level and simulated changes in bed elevation were compared to those observed. In addition to the work directly related to model development, application and validation work was also undertaken to address the issue of representing suspended sediment 'fractions' consisting of a range of grain sizes. This work was motivated by the findings of analyses of the suspended sediment load in the Lower Fraser River and by the necessity of characterizing the representative grain size for natural rivers where there is a large gradation in the bed material composition. This study revealed that the representative grain size of a fraction depends on certain local hydraulic conditions. In addition, correction factors for modelling suspended sediment concentrations were obtained when the geometric grain size of the fraction is used in lieu of the mathematically accurate representative size.