基于高分遥感影像的总磷浓度反演研究

Total Phosphorus Concentration Inversion Research Based on High-resolution Remote Sensing Images

以定量遥感反演水质参数为目的,以天津市海河下游段为研究区,利用总磷实测水质数据和同期GF-2PMS 2遥感影像数据,建立两者的偏最小二乘回归模型、单隐含层神经网络模型、双隐含层神经网络模型及粒子群算法优化的双隐含层神经网络模型(DP-BPNN模型)。通过决定系数、平均绝对误差、均方根误差进行精度检验,选出研究区水体适用的总磷浓度的反演模型。结果表明:与偏最小二乘模型精度对比,所建总磷反演模型精度提高了48%。

In this paper,the inversion model for total phosphorus concentration was established based on the measured water quality data and GF-2 satellite data in the Haihe River Basin,Tianjin.The optimal inversion model for total phosphorus concentration was determined by comparing the partial least squares regression model,the single hidden layer neural network model,the double hidden layer neural network model and the dual neural network model optimized by the particle swarm algorithm.Through the determination coefficient,mean absolute error,and root mean square error,the accuracy was verified,and the inversion model of total phosphorus concentration applicable to the water body in the study area was selected.The results show that the accuracy of DP-BPNN of total phosphorus concentrations is improved by 48%as compared with that of the partial least squares model.