Enhancing water level prediction of the Yangtze River based on RGL-stacking model

The accurate prediction of water level along the Yangtze River main channel is a critical issue in water resource management. However, due to the diversity within the Yangtze River basin, such as seasonal variations, differences in rainfall distribution, and variations in topography, a single model often struggles to address these complex challenges. This study aims to enhance the accuracy of water level prediction in the Yangtze River basin through the construction of a deep learning stacked model for water level forecasting research. The research analyzes the characteristics of water level data from four important stations in the middle and lower reaches of the Yangtze River. By comparing the predictive results of different deep learning models, it is found that the RGL (RNN-GRU-LSTM) stacked model outperforms individual model in water level prediction. It exhibits superior adaptability and accuracy for stations with varying water level patterns. These research findings are expected to provide new methods and practical experiences for addressing hydrological issues and offer reliable data support for decision-making processes.