基于CNN-LSTM的中小河流水位预报模型研究OA

In order to solve the problem of flood water level forecasting for small and medium-sized rivers in areas without flow data,this study proposes a flood forecasting model that integrates a Convolutional Neural Network(CNN)and a Long Short-Term Memory network(LSTM).By integrating the spatiotemporal distribution characteristics of the rainfall sequence within the river basin and the temporal evolution law of the water level,the nonlinear mapping relationship between rainfall and water level is found,and a rainfall-water level flood forecasting model is constructed.An experiment was conducted with the Bengkan River,a tributary of the Longjiang River,as the study area.The results show that,in the absence of flow data,the model can achieve good results in short-term and imminent water level forecasting.The results show that,in the absence of flow data,the model can achieve good results in short-term and imminent water level forecasting.It maintains a high prediction accuracy(determination coefficient of 0.82)within a 6-hour forecast horizon.This study provides a new solution idea for flood water level forecasting work and can further improve the accuracy of water level forecasting in data-scarce areas.