基于BP神经网络降雨径流预测的方法探讨

Methods of Rainfall and Runoff Prediction Based on BP Neural Network

本文通过ArcGIS水文分析进行基础数据探索,结合BP神经网络进行降水径流预测,并对比分析了不同空间扩展估算方法中对面雨量进行空间扩展估算结果与水文站实测值对比精度,同时考虑水文分析中下垫面植被覆盖度,选用归一化植被指数对降水径流预测结果的影响,综合分析神经网络进行降水径流预测时不同下垫面计算结果精度,得到以下结论:1)采用空间扩展估算时反距离权重法获得的面雨量精度较高;2)基于面雨量进行BP神经网络降雨径流量预测时考虑NDVI指数可大幅提高预测精度;为降水径流预测及降水致洪分析提供科学依据及理论基础。

In this paper,the basic data was explored by ArcGIS hydrological analysis,meanwhile,combining BP neural network to predict precipitation and runoff,and the comparison accuracy between the estimation results of opposite rainfall in different spatial expansion estimation methods and the measured values of hydrological stations was analyzed.At the same time,the vegetation coverage of underlying surface in hydrological analysis was considered.The influence of normalized difference vegetation index(NDVI)on precipitation and runoff prediction results was used to comprehensively analyze the accuracy of different underlying surface calculation results when precipitation and runoff was predicted by neural network.The following conclusions were drawn:1)The accuracy of surface rainfall obtained by inverse distance weight method in spatial expansion estimation is higher;2)When BP neural network is used to forecast rainfall runoff based on areal rainfall,the prediction accuracy can be greatly improved by considering NDVI index;it provides scientific and theoretical basis for precipitation runoff prediction and precipitation flood analysis.