Estimation of total sediment load in rivers using artificial intelligence methods

Correct estimation of sediment volume carried by a river is very important for many water resources projects. Sediment concentration is generally determined from direct measurements, or estimated from sediment transport equations that require detailed information about the flow and sediment characteristics. However, there is often a large discrepancy between these models and observations. The complexity of sediment transport processes presents an opportunity for the application of alternate methods. As a fairly recent computing tool, relevance vector machines (RVMs) are gaining popularity in the fields of artificial intelligence methods.This dissertation presents two scenarios. The first scenario is to develop artificial intelligence methods for estimation of total sediment concentrations. The resulting artificial intelligence models are then trained and tested on a large data set and the performance of the approaches are compared with more conventional transport formulae. The second scenario is to obtain a unified approach to estimate total sediment transport, with a focus on elucidating the differences in the empirical predictions of laboratory and field data. RVM based probabilistic models were developed using laboratory data, and their performances were tested against field data and with conventional prediction methods. For total sediment transport, the RVM model trained only on laboratory data yielded results for field conditions that are better or at least comparable with existing methods. The findings of this study suggest that the main phenomenon governing fluvial process in flumes and rivers are closely related, and that the choice of dimensionless input variables should be in the same range for both laboratory and field data for successful extrapolation from flumes to rivers.