基于神经网络的宽浅型湖泊水质反演研究

城市浅型湖泊治理是城市生态文明建设的重要组成部分。通过对黄石磁湖的IKONOS遥感影像进行预处理,建立了水质参数与卫星波段的多元线性回归模型、BP神经网络模型和RBF神经网络模型。通过比较不同模型的结果,运用可靠模型对整个湖体的COD、NH3-N、TN、TP指标进行反演。结果表明,神经网络模型对于磁湖水质指标的反演结果显著优于多元线性回归模型,其中BP神经网络模型对NH3-N、TP的模拟效果好,RBF神经网络模型对COD、TN的模拟效果较好。

Application of artificial neural network on water quality inversion in Cihu Lake

The management of urban shallow lakes plays an important role in the urban ecological civilization construction. Using the IKONOS remote sensing image, two artificial neural network models, based on the BP (Back Propagation) and RBF (Radical Basis Function), were set up to inverse the COD, NH3-N, TN and TP quality conditions of the Cihu Lake. The proposed models were also compared with the multivariate linear regression model. The results indicate that the model efficiency of the two ANN models are significantly higher than the multiple linear regression model. The BP model fits the observed data best in the simulation of the NH3-N, TP, while the RBF neural network shows advantages in the simulation of the COD and TN.