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:
Hyper-spectral remote sensing data can provide more spectral information and describe the spectral signature of salt lake more accurately than multi -spectral remote sensing data. Based on the theory of remote sensing on water, the authors analyzed the band correlation and information of HJ-1A satellite hyper-spectrum image by using adaptive band selection( ABS) method. Combined with BP neural network techniques, the authors determined the optimal band combination, established the retrieval models for mineral ions salinity of salt lake, quantitatively determined the salinities of K+, Mg2+, Na+, Cl-, SO2-4 ions of west Taijinar Salt Lake in Qaidam Basin, and acquired the spatial distribution siuation of mineral ions salinity. The results show that the forecast accuracy of BP neural network models are exclusively higher than 85%, the spatial distribution of mineral ions content of salt lake is consistent with the result of field survey. The research confirms that the correlation of BP neural network and domestic hyper-spectral remote sensing data can be used to monitor the mineral resource of salt lake dynamically, thus providing the scientific foundation for the reasonable development and efficient utilization.
下载: