摘要
随着人类活动的影响,重金属污染逐渐成为土壤和环境研究的重点。采用遥感技术可以克服传统重金属监测过程中的缺点,做到快速、高效地反映重金属空间分布。本文以克拉玛依市某区域为研究区,利用SVC HR-768光谱仪和Landsat8影像对41块土壤样品进行地物光谱和波段反射率的获取,采用相关性分析和偏最小二乘回归的原理,建立地物光谱与Landsat8数据的土壤铅含量反演模型。结果表明,基于一阶微分变换的地物光谱能更好地反映光谱与重金属铅含量的相关性,建立的模型为最优预测模型;通过波段比值和波段差值方式建立的基于Landsat8波段反射率的反演模型精度较好,能粗略预测土壤重金属铅的含量,并且基于Landsat8影像反演的土壤铅含量空间分布符合土壤样点实测值的空间分布,为今后土壤环境监测土壤重金属含量提供数据支撑。
With the influence of human activities,heavy metal pollution has gradually become the focus of soil and environmental research.The application of remote sensing technology could overcome the shortcomings in the process of traditional monitoring of heavy metals and reflect the spatial distribution of heavy metals quickly and efficiently.In this study,the SVC HR-768 spectrometer and Landsat8 images were used to obtain hyperspectral data and band reflectance of 41 soil samples in Karamay City.By using the principle of correlation analysis and partial least-squares regression,the inversion model of soil lead content based on hyperspectral and Landsat8 data was established.The results showed that the first order differential transform could better reflect the correlation between the spectrum and the heavy metal lead content,and it was the best prediction model.The retrieval model based on Landsat8 spectral reflectance established by band ratio and band difference had better accuracy,and could roughly predict the content of heavy metal lead in soil,moreover,the spatial distribution of soil lead content based on Landsat8 model was consistent with the measured spatial distribution of soil sample points,and provided data support for monitoring heavy metal content in soil environment in the future.
作者
马磊
颜安
Ma Lei;Yan An(College of Grassland and Environment Sciences,Xinjiang Agricultural University,Urumqi 830052,China;Xinjiang Key Lab of Soil and Plant Ecological Processes,Urumqi 830052,China)
出处
《山东农业科学》
2019年第12期120-126,共7页
Shandong Agricultural Sciences
基金
新疆自治区农业技术推广与服务项目
新疆自治区青年博士科技人才培养项目(QN2016BS0705)
关键词
高光谱
Landsat8
铅
相关性分析
偏最小二乘法
土壤环境监测
Hyperspectral
Landsat8
Lead
Correlation analysis
Partial least squares regression
Soil environment monitoring
