Based on the measured water level data after the impound of Three Gorges reservoir, the water level short-term prediction model of income flow of Chenglingji, Han river and Hukou is constructed by multiple regression method. The comparative of measured water level and predicted water level indicated that, the prediction of income flow is accord with the real flow. Meanwhile, according to statistical analysis of the water level and flow, and considering the total inflow and the jacking of branch inflow, the water level short-term prediction model for middle stream Yangtze River is set up separately. Then, by using multiple regression model, the multiple regression formula for water level prediction is constructed , to applied to the river reach where branch inflowed or river reach jacked by the downstream. Compared with the field observation data, the prediction results are quite precisely.
Hybrid Deep Learning Modeling for Water Level Prediction in Yangtze River