论文部分内容阅读
针对碳储量回归预测模型存在共线性和精度较低的问题,利用森林资源二类调查数据和SPOT5影像数据对北京市延庆县的杨树林进行碳储量反演研究。先对选取的10个指标进行主成分分析,在此基础上采用径向基函数(RBF)神经网络方法构建碳储量反演模型,用预留测试样本验证,并与实测值进行比较。研究结果表明:SPOT5数据和二类数据可以很好地结合起来用于森林地上碳储量反演研究;PCA-RBF神经网络森林碳储量遥感反演模型拟合精度为99.90%,平均预测精度达到96.71%,预估效果较理想;模型训练完成后,可以应用于延庆县森林地上碳储量反演。
In order to solve the problem of collinearity and accuracy of the regression model of carbon reserves, the inversion of carbon stocks of poplar forests in Yanqing County of Beijing was studied by using the second type of forest resources survey data and SPOT5 image data. Firstly, the principal component analysis of the selected 10 indicators is carried out. Based on this, the RBF neural network method is used to build a carbon storage inversion model, which is verified by reserved test samples and compared with the measured values. The results show that the SPOT5 data and the second-class data can be well combined for the retrieval of forest carbon stocks on the ground. The fitting accuracy of forest carbon storage remote sensing model with PCA-RBF neural network is 99.90% and the average prediction accuracy is 96.71 %, The predicted effect is better; after the model training is completed, it can be applied to the inversion of the above-ground forest carbon stocks in Yanqing County.