基于MLP-ANN神经网络的河流泥沙输移对气候变化的响应

Response for the Climate Changes to the Transportation of Sediment in Longchuanjiang River Based on MLP-ANN Model

运用MLP-ANN神经网络方法,对金沙江中游一级支流龙川江的河流泥沙输移状况与气候变化之间的响应关系进行了研究。结果表明:(1)龙川江河流泥沙输移对温度、降雨量、降雨强度和径流量等气候因子响应直接,而对降雨频率、蒸发量和湿度的改变没有明显响应;(2)当不考虑径流因子时,MLP-ANN模型对泥沙输移的拟合有一个月的时滞效应,而考虑径流因子时,则没有时滞效应;(3)在1993年以后,河流泥沙输移对气候变化的响应出现明显偏移,显示有其它因素对龙川江的河流泥沙输移产生了强烈干扰。

The MLP- ANN model was applied to research the response for the climate changes to the transportation of Sediment in Longchuanjiang River. The results showing： （ 1 ） The temperature, rainfall, intensity of storm and the water discharge have a directly relationship with the Sediment whereas the frequency of storm, evaporation and the humidity have no response explicitly ; （2） Without delay effect appears to the transportation of Sediment model while considering the water discharge, otherwise, A month delay effect appears ; （3） The response for the climate changes to the transportation of Sediment in Longchuanjiang River has a obvious deviation since1990, it shows that the other factors should disturb this response mechanism strongly except the climate.