基于BP神经网络的盐湖矿物离子含量高光谱反演

发布时间：2018-06-25 05:28

本文选题：矿物离子含量 + 遥感反演　；参考：《国土资源遥感》2016年02期

【摘要】：高光谱遥感数据能够提供比多光谱遥感数据更为丰富的光谱信息,从而更精确地刻画地物的光谱特征。在水体遥感原理基础上,采用自适应波段选择(adaptive band selection,ABS)方法对HJ-1A卫星高光谱数据的波段相关性和信息量进行分析,结合BP神经网络技术确定最优波段组合并构建盐湖矿物离子含量的反演模型,对柴达木盆地西台吉乃尔湖的K+,Mg2+,Na+,Cl-和SO2-4离子含量进行定量反演,获得盐湖矿物离子含量的空间分布情况。研究结果表明,BP神经网络反演模型的盐湖矿物离子含量反演精度在85%以上,反演得到的矿物离子含量的分布情况与实地调查结果基本一致。因此,利用高光谱数据和BP神经网络可以对盐湖矿物资源进行大范围动态监测,为盐湖资源的合理开发和高效利用提供科学依据。
[Abstract]:Hyperspectral remote sensing data can provide more spectral information than multi-spectral remote sensing data, thus more accurate description of the spectral characteristics of ground objects. Based on the principle of water remote sensing, the band correlation and information content of HJ-1A satellite hyperspectral data were analyzed by adaptive band selection (adaptive band selection method. Combined with BP neural network technology to determine the optimal band combination and to construct the inversion model of mineral ion content in salt lake, the quantitative inversion of K _ (mg _ 2) Na ~ (2 +) Cl- and so _ (2-4) ion content in Lake Taijinaire, Qaidam Basin, was carried out. The spatial distribution of mineral ion content in salt lake was obtained. The results show that the inversion accuracy of salt lake mineral ion content of BP neural network inversion model is more than 85%, and the distribution of mineral ion content obtained by inversion is basically consistent with the result of field investigation. Therefore, hyperspectral data and BP neural network can be used to monitor the mineral resources of salt lake on a large scale and provide scientific basis for rational exploitation and efficient utilization of salt lake resources.
【作者单位】：中国农业大学信息与电气工程学院;北京信息科技大学经济与管理学院;
【基金】：国家科技支撑计划项目“循环经济试验区产业集群科技服务集成平台研发与应用”(编号:2012BAH10F01)资助
【分类号】：TP751;TP183

【相似文献】