基于LSTM网络鄱阳湖抚河流域径流模拟研究

水文预报及其径流变化趋势预测能够为防汛工作提供辅助决策，是水库调度兴利的重要手段.与传统分布式水文模型相比，利用长短期记忆网络(LSTM)建立降雨径流预报模型具有简单可行和精度较高的优点.该文以鄱阳湖抚河流域为研究对象，采用抚河流域的降雨和径流数据分别作为模型驱动数据和标签数据，通过LSTM网络实现抚河流域的径流模拟工作.结果表明：在使用气象站数据建立的日尺度径流模拟模型中，模拟结果与实测值相关性均达到0.9以上，偏差在±5%以内，模型表现非常好；在使用TRMM数据建立的月尺度模型中，整体模拟结果与实测值相关性在0.9以上，整体偏差在±5%以内，模型表现优秀.

The research of stream flow simulation using Long and Short Term Memory (LSTM) network in Fuhe River Basin of Poyang Lake

Studies of the changing trend of runoff through hydrological prediction can provide auxiliary decision-making for flood control work， which is also an important method for reservoir regulation. Compared with the traditional SWAT model， the runoff simulation model based on LSTM network is more practical and accurate. Focusing on the Fuhe river basin of Poyang Lake， we use rainfall and runoff data collected from Fuhe river basin as model driving data and label data respectively， and achieve the runoff simulation through LSTM network. The results show as follows. In the daily runoff prediction using data observed from meteorological stations， the correlation between measured and simulated runoff is above 0.9 and Pbias is within ±5%， indicating that the model performs well. In the monthly runoff prediction using TRMM dataset， the overall correlation between measured and simulated runoff is above 0.9 and Pbias is within ±5%， illustrating the excellent performance of the model.