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.

Zhang C. K., Yang Y. Z., Tao, J. F., Chen, Y. P., Yao, P., Su. M., 2013. Suspended sediment fluxes in the radial sand ridge field of South Yellow Sea. The radial sand ridge at the South Yellow Sea is one of unique geomorphic units in the world. It is famous of its radial current pattern and high suspended sediment concentration, forming abundant tidal flat resources which could be reclaimed for the land use. In order to make a scientific plan for the land reclamation, it is necessary to quantitatively understand the distribution of suspended sediment fluxes in this field. In this study, a 2D tidal current and suspended sediment transport model is developed and validated by the hydrologic survey data measured in 2006. The suspended sediment fluxes are calculated by the period-averaged multification of the simulated tidal current and sediment concentration in two consecutive tidal periods. The numerical results show that, in either winter or summer season, there has a positive sediment flux comes f...

Chen, Y. P., Xie, D. M., Zhang, C. K, Qian, X. S., 2013. Estimation of long-term wave statistics in the East China Sea. This study used a state-of-the-art approach to estimate the long-term wave statistics in the East China Sea. The wave climate in the past 60 years (1950–2009) was simulated by a third generation wave model driven by the NCEP/NCAR reanalysis wind data. Based on the statistical analysis of the numerical results, the mean wave fields and the extreme wave fields with 100-year and 50-year return period on the whole computational domain were obtained. The spatial variations of the wave fields show that the statistical waves decrease from south to north and from sea to coast in general, which can be associated with the local topography of the East China Sea. The seasonal variations show that the mean waves in the winter and autumn are slightly larger than those in the spring and summer, while the extreme waves in the summer are much larger than those in the other seasons, which is main...

Influenced by river discharge, the tidal properties of estuarine tides can be more complex than those of oceanic tides, which makes the tidal prediction less accurate when using a classical tidal harmonic analysis approach, such as the T_TIDE model. Although the nonstationary tidal harmonic analysis model NS_TIDE can improve the accuracy for the analysis of tides in a river-dominated estuary, it becomes less satisfactory when applying the NS_TIDE model to a mesotidal estuary like the Yangtze estuary. Through the error source analysis, it is found that the main errors originate from the low frequency of tidal fluctuation. The NS_TIDE model is then modified by replacing the stage model with the frequency-expanded tidal–fluvial model so that more subtidal constituents, especially the “atmospheric tides,” can be taken into account. The results show that the residuals from tidal harmonic analysis are significantly reduced by using the modified NS_TIDE model, with the yearly root-mean-square-error values being only 0.04–0.06 m for the Yangtze estuarine tides.