Due to the strong nonlinear interaction with river discharge, tides in estuaries are characterised as nonstationary and their mechanisms are yet to be fully understood. It remains highly challenging to accurately predict estuarine water levels. Machine learning methods, which offer a unique ability to simulate the unknown relationships between variables, have been increasingly used in a large number of research areas. This study applies the LightGBM model to predicting the water levels along the lower reach of the Columbia River. The model inputs consist of the discharges from two upstream rivers (Columbia and Willamette Rivers) and the tide characteristics, including the tide range at the estuary mouth (Astoria) and tide constituents. The model is optimized with the selected parameters. The results show that the LightGBM model can achieve high prediction accuracy, with the root-mean-square-error values of water level being reduced to 0.14 m and the correlation coefficient and skill score being in the ranges of 0.975–0.987 and 0.941–0.972, respectively, which are statistically better than those obtained from physics-based models such as the nonstationary tidal harmonic analysis model (NS_TIDE). The importance of subtide constituents in interacting with the river discharge in the estuary is clearly revealed from the model results.

High-quality wind field data are key to improving the accuracy of storm surge simulations in coastal and estuarine water. These data are also of great significance in studying the dynamic processes in coastal areas and safeguarding human engineering structures. A directional correction method for ECMWF reanalysis wind data was proposed in this paper based on the correlation with the measured wind speed and direction. The results show that the accuracies of wind speed and direction were improved after being modified by the correction method proposed in this paper. The modified wind data were applied to drive the storm surge model of the Yangtze Estuary for typhoon events, which resulted in a significant improvement to the accuracy of hindcasted water levels. The error of the hindcasted highest water levels was reduced by 16–19 cm.

According to the demand of large-scale simulation, a numerical simulation model for vegetated flows was developed based on two-dimensional unsteady flow equations. According to the experimental flume results, the influence of aquatic plants on water flow was investigated. The vertical distribution characteristics of the drag coefficient induced by flexible vegetation at different positions of vegetation patches were obtained. The drag coefficients obtained from the experiments were substituted into the model to verify the model's reliability and explore the flow characteristics of open channels with vegetation patches. The results show that: 1) the simulation results of the model are in good agreement with the measured data, which can well simulate the water flow process under the influence of vegetation patches, especially for the far areas with weak threedimensional effect; 2) The water-blocking effect of the vegetation patch reduces the flow velocity inside the patch and increases the flow velocity in the non-vegetated area on the side, forming a longitudinal shear effect at the edge. The proposed model can support the dynamic analysis of vegetation in rivers, lakes, and estuarine wetland systems.