An Overview of Machine Learning Techniques for Sediment Prediction †

Abstract

Most hydrological and water resources researchers prioritize the development of an accurate sediment prediction model. Several conventional techniques have failed to accurately predict suspended sediment. Because of the intricacy, non-stationarity, and nonlinearity of sediment movement behavior in streams and rivers, many techniques fall short. Over the last several years, there have been meaningful theoretical improvements in the understanding of machine learning approaches, vis a vis strategy for the implementation of their processes and uses of the technique for practical and hydrological issues. To produce the desired output, machine learning models and other algorithms have been employed to predict complicated nonlinear connections and patterns of huge input parameters. This paper examines a few key works of the literature on sediment transport prediction while focusing on a variety of machine learning applications. Sediment transport models aided by machine learning have attracted a growing number of researchers in recent years. As a result, they must gain in-depth knowledge of their theory and modeling methodologies. Furthermore, this chapter includes an overview of the machine learning technique and other developing hybrid models that have produced promising outcomes. This overview also includes various examples of successful machine learning applications in sediment prediction.