干旱区生态环境知识资源中心

The Shibanjing ophiolite complex, located in the Mingshui-Shibanjing-Xiaohuangshan ophiolite belt of Beishan collage, is one of the most important components in the Beishan orogenic belt, Inner Mongolia, China. It comprises tectonic slices of ophiolite rocks including mylonitic gabbros, meta-ultramafic rocks, metabasalts, and elastic rock in a matrix of turbidites, and the melange contains blocks of schists, gneiss, and marble. In this study, multispectral data Sentinel-2A and ASTER were employed to discriminate lithological units in the unvegetated arid region. Several image enhancement algorithms, including Band Ratio (BR), Principal Components Analysis (PCA), Minimum Noise Fraction (MNF), and Support Vector Machine (SVM), were applied to the data for lithological mapping. A specialized band ratio, (Sentinel-2A band 3+ASTER band 9)/(Sentinel-2A band 12+ASTER band 8), combining ASTER and Sentinel-2A data was designed to distinguish serpentine minerals in the ophiolite complex. Color composites of principal components (PCs) of ASTER and Sentinel-2A data were utilized to differentiate rock units within the ophiolite complex. It was found that the color composites of PCs of Sentinel-2A have higher color contrast and saturation than ASTER, enabling to better visually interpret the ophiolite rock units and provide a more detailed geological map of the study area. The SVM classification results of the original and MNF versions of the ASTER data indicate that the image processing methods such as MNF could highlight specific rock units and improve classification accuracy. In addition, with the comparison between MNF images for lithological classification of Sentinel-2A and ASTER data, it was demonstrated that Sentinel-2A data outperformed ASTER in lithological mapping in the Shibanjing ophiolite complex. (C) 2018 COSPAR. Published by Elsevier Ltd. All rights reserved.