Time-series prediction of a river stage during typhoons or storms is essential for flood control or flood disaster prevention. Data-driven models using machine learning (ML) techniques have become an attractive and effective approach to modeling and analyzing river stage dynamics. However, relatively new ML techniques, such as the light gradient boosting machine regression (LGBMR), have rarely been applied to predict the river stage in a tidal river. In this study, data-driven ML models were developed under a multistep-ahead prediction framework and evaluated for river stage modeling. Four ML techniques, namely support vector regression (SVR), random forest regression (RFR), multilayer perceptron regression (MLPR), and LGBMR, were employed to establish data-driven ML models with Bayesian optimization. The models were applied to simulate river stage hydrographs of the tidal reach of the Lan-Yang River Basin in Northeastern Taiwan. Historical measurements of rainfall, river stages, and tidal levels were collected from 2004 to 2017 and used for training and validation of the four models. Four scenarios were used to investigate the effect of the combinations of input variables on river stage predictions. The results indicated that (1) the tidal level at a previous stage significantly affected the prediction results; (2) the LGBMR model achieves more favorable prediction performance than the SVR, RFR, and MLPR models; and (3) the LGBMR model could efficiently and accurately predict the 1–6-h river stage in the tidal river. This study provides an extensive and insightful comparison of four data-driven ML models for river stage forecasting that can be helpful for model selection and flood mitigation.

Physically based numerical models can predict scour depth at embankments located in bend reaches. However, methodologies for utilizing these numerical models to assess the risk of reinforced concrete embankment failure are rarely investigated. Therefore, a new assessment methodology is proposed to predict the riverbank failure caused by bend scour. The methodology is primarily based on a bend scour simulation model that integrates a one-dimensional (1D) hydraulic model, a two-dimensional (2D) hydrodynamic finite-volume model, and an empirical equation of bend scour prediction. The model was first applied to the Shuiwei Embankment located in a river bend reach of Da-An River in Taiwan and verified against results from the 1D hydraulic model and field data. The model was then used to simulate 2D flow field and the temporal evolution of bend scour depth under different return period flood events to examine the relationships between river discharge, water level, shear stress, and bend scour depth. The influence of shear stress on the stability of toe protections was also investigated. The field data (from two events) and numerical solutions (four scenarios) were assessed to conceive two empirical equations for predicting shear stress and bend scour depth. A new assessment methodology was proposed using these two equations to predict the risk of river embankment failure during flood periods. The proposed methodology can be easily applied in other disaster-prone regions to mitigate the effects of disasters caused by bend scouring.

No abstract