Inversion of geochemical compositions of basalts based on field measured spectra
-
摘要: 岩石地球化学成分对岩石分类、成因及演化研究有重要地质意义.利用遥感手段反演岩石地球化学成分是一个较新的课题,也是遥感相关应用研究的难点.以甘肃省柳园镇南部二叠纪玄武岩带为研究目标,在进行系统采样、光谱分析、地球化学测试的基础上,采用偏最小二乘回归(partial least squares regression,PLSR)对拥有2 150个波段的实测波谱数据及相应的6种主量矿物数据进行反演建模.首先选择有效的预处理方法对目标数据集进行优化,再利用k折交叉检验方法获得最小均方根误差下的最适主成分个数.运算结果表明,PLSR模型具有较好的稳定性和精度,在利用遥感数据进行岩石地球化学成分反演方面有很好的应用前景.Abstract: Geochemical compositions have significant implications for rock classification,identification of the petrogenesis and evolution of the rocks.The utilization of remote sensing method to estimate the geochemical compositions of the rocks is a new subject,and is also a difficult point in remote sensing related researches due to its relatively immature applications.In this study,he Permian basalts were chosen as the study object.Based on systematical sampling,spectral analysis and geochemical test,the authors constructed a mathematical model between field measured spectra data (2 150 bands) and available data of six representative major elements by using partial least squares regression (PLSR).It is essential to initially choose proper preprocessing method to optimize the spectra data,and then search for the optimal number of principal components withminimum root-mean-square error through k-fold cross-validation.The results show that the PLSR model yields higher stability and precision,and plays a significant role in applications of geochemical composition inversion using remote sensing data.
-
-
计量
- 文章访问数: 320
- PDF下载数: 13
- 施引文献: 0
下载: